WT__ 2 WT 4 WORLD BANK TECHNICAL PAPER NUMBER 274 March I q95 Health Expenditures in Latin America Ramesh Govindaraj, Christopher J. L. Murray, and Gnanaraj Chellaraj fO IOLO iMA~WN UFCUI TT P n O ADTEUEEN T RECENT WORLD BANK TECHNICAL PAPERS No. 191 Frederiksen, Water Resources Institutiotts: Some Principles and Practices No. 192 McMillan, Painter, and Scudder, Settlement and Development in the River Blindness Control Zone No. 193 Braatz, Conserving Biological Diversity: A Strategyfor Protected Areas in the Asia-Pacific Region No. 194 Saint, Universities in Africa: Strategiesfor Stabilization and Revitalization No. 195 Ochs and Bishay, Drainage Guidelines No. 196 Mabogunje, Perspective on Urban Land and Land Management Policies in Sub-Saharan Africa No. 197 Zymelman, editor, Assessing Engineering Education in Sub-Saharan Africa No. 198 Teerink and Nakashima, W'Vater Allocation, Rights, and Pricing: Examplesfrom Japan and the United States No. 199 Hussi, Murphy, Lindberg, and Brenneman, The Development of Cooperatives and Other Rural Organizations: The Role of the World Bank No. 200 McMillan, Nana, and Savadogo, Settlement and Development in the River Blindness Control Zone: Case Study of Burkina Faso No. 201 Van Tuijl, Improving Water Use in Agriculture: Experiences in the Middle East and North Africa No. 202 Vergara, The Materials Revolution: What Does It Mean for Developing Asia? No. 203 Cleaver, A Strategy to Develop Agriculture in Sub-Saharan Africa and a Focusfor the World Bank No. 204 Barghouti, Cromwell, and Pritchard, editors, Agricultural Technologies for Market-Led Development Opportunities in the 1990s No. 205 Xie, Kuffner, and Le Moigne, Using Water Efficiently: Technological Options No. 206 The World Bank/FAO/UNIDO/lndustry Fertilizer Working Group, World and Regional Supply and Demand Balancesfor Nitrogen, Phosphate, and Potash, 1991/92-1997/98 No. 207 Narayan, Participatory Evaluation: Tools for Managing Change in Water and Sanitation No. 208 Bindlish and Evenson, Evaluation of the Performance of T&V Extension in Kenya No. 209 Keith, Property Tax: A Practical Manual for Anglophone Africa No. 210 Bradley and McNamara, editors, Living with Trees: Policies for Forestry Management in Zimbabwe No. 211 Wiebers, Integrated Pest Management and Pesticide Regulationl in Developing Asia No. 212 Frederiksen, Berkoff, and Barber, Water Resources Management in Asia, Volume I: Main Report No. 213 Srivastava and Jaffee, Best Practices for Moving Seed Technology: New Approaches to Doing Business No. 214 Bonfiglioli, Agro-pastoralism in Chad as a Strategyfor Survival: An Essay on the Relationship between Anthropology and Statistics No. 215 Umali, Irrigation-Induced Salinity: A Growing Problem for Development and the Environment No. 216 Carr, Improving Cash Crops in Africa: Factors Inifluencing the Productivity of Cotton, Coffee, and Tea Grown by Smallholders No. 217 Antholt, Getting Readyfor the Twenty-First Century: Technical Change and Institutional Modernization in Agriculture No. 218 Mohan, editor, Bibliography of Publications: Technical Department, Africa Region, July 1987 to December 1992 No. 219 Cercone, Alcohol-Related Problems as an Obstacle to the Development of Human Capital: Issues and Policy Options No. 220 Kingsley, Ferguson, Bower, and Dice, Managing Urban Environmtienltal Quality in Asia No. 221 Srivastava, Tamboli, English, Lal, and Stewart, Conserving Soil Moisture and Fertility in the Warm Seasonally Dry Tropics No. 222 Selvaratnam, Innovations in Higher Education: Singapore at the Competitive Edge No. 223 Piotrow, Treiman, Rimon, Yun, and Lozare, Strategies for Family Planning Promotion No. 224 Midgley, Urban Transport in Asia: An Operational Agendafor the 1990s No. 225 Dia, A Governance Approach to Civil Service Reform in Sub-Saharan Africa No. 226 Bindlish, Evenson, and Gbetibouo, Evaluation of T&V-Based Extension in Burkina Faso No. 227 Cook, editor, Involuntary Resettlement in Africa: Selected Papersfrom a Conference on Environment and Settlement Issues in Africa (List continues on the inside back cover) WORLD BANK TECHNICAL PAPER NUMBER 274 Health Expenditures in Latin America Ramesh Govindaraj, Christopher J. L. Murray, Gnanaraj Chellaraj The World Bank Washington, D.C. Copyright (© 1995 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 First printing March 1995 Technical Papers are published to communicate the results of the Bank's work to the development com- mnLMity with the least possible delay. The typescript of this paper therefore has not been prepared in accor- dance with the procedures appropriate to formal printed texts, and the World Bank accepts no responsibili- ty for errors. Some sources cited in this paper may be informal documents that are not readily available. The findings, interpretations, and conclusions expressed in this paper are entirely those of the author(s) 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. The World Bank does not guarantee the accuracy of the data included in this publication and accepts no responsibility whatso- ever for any consequence of their use. The boundaries, colors, denominations, and other information shown on any map in this volume do not imply on the part of the World Bank Group any judgment on the legal status of any territory or the endorsement or acceptance of such boundaries. The material in this publication is copyrighted. Requests for permission to reproduce portions of it should be sent to the Office of the Publisher at the address shown in the copyright notice above. The World Bank encourages dissemination of its work and will normally give permission promptly and, when the reproduction is for noncommercial purposes, without asking a fee. Permission to copy por- tions for classroom use is granted through the Copyright Clearance Center, Inc., Suite 910, 222 Rosewood Drive, Danvers, Massachusetts 01923, U.S.A. TIhe complete backlist of publications from the World Bank is shown in the annual Index of Publications, which contains an alphabetical title list (with full ordering information) and indexes of sub- jects, authors, and countries and regions. The latest edition is available free of charge from the Distribution Unit, Office of the Publisher, The World Bank, 1818 H Street, N.W., Washington, D.C. 20433, U.S.A., or from Publications, The World Bank, 66, avenue d'1ena, 75116 Paris, France. ISSN: 0253-7494 Ramesh Govindaraj is a research associate and Christopher J. L. Murray is an associate professor, both at the School of Public Health at Harvard University. Gnanaraj Chellaraj is a consultant in Health Economics to the Middle East and North Africa Country Department II at the World Bank. Library of Congress Cataloging-in-Publication Data Govindaraj, Ramesh, 1961- Health expenditures in Latin America / Ramesh Govindaraj, Christopher J. L. Murray, Gnanaraj Chellaraj. p. cm. - (World Bank technical paper, ISSN 0253-7494; no. 274) Includes bibliographical reference (p. ). ISBN 0-8213-3142-X 1. Medical care, Cost of-Latin America. 2. Medical care, Cost of-Latin America-Statistics. I. Murray, Christopher J. L. II. Chellaraj, Gnanaraj, 1958- . III. Title. IV. Series. RA410.55.L29G68 1995 388.4'33621'098-dc2O 95-3945 CIP - 111 - Contents Foreword ....................................... iv Acknowledgments ................................. v Chapter I. Introduction ............................ 1 Chapter II. Past Studies of National Health Expenditures ..... . . . . 3 Chapter III. Definitions, Methods, and Materials ..... . . . . . . . . . 5 A. Definitions ............. ... 5 B. Methods ................................ . 6 1. Currency Conversions ...... . . . . . . . . . . . . . . 6 2. Disaggregation of Government Expenditures ..... . . 6 C. Materials .......... . .. . .. .. . .. . .. . .. . .. . . . 8 Chapter IV. Estimating Out-Of-Sample .................. . . 11 A. Estimating Public Sector Expenditure .12 B. Estimating Private Health Expenditure .13 Chapter V. Results .15 A. Regional Health Expenditures .15 B. Time Trends .16 C. Disaggregated Data on Government Health Expenditures . . .. 18 Chapter VI. Discussion and Conclusions .21 A. Tracking Health Expenditures .21 B. Income and Other Determinants of Health Expenditure .... . 21 C. How Do We Expect Health Expenditures to Change with Income Per Capita ......... .. . .. . . .. . . .. . . 23 D. What Does Health Expenditure Buy? ....... . . . . . . . . . . 24 Bibliography ............... . .................... . 27 Appendices . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 31 - iv - Foreword The preparation of the World Bank's World Development Report 1993: Investing in Health (WDR) included a substantial effort to assemble intemationally comparable statistics on a broad range of health and health system indicators. Much of this effort involved compiling data from existing sources. It also involved substantial analytical efforts to improve the quality and comparability of data (e.g. for trends in under-5 mortality rates and for levels of health expenditures). Appendix A of the WDR contains the resulting statistical tables. In particular, that Appendix reports the first estimates that have been assembled on global public and private health expenditures. This Discussion Paper presents an expanded and updated version of the WDR estimates of 1990 health expenditures for the countries of Latin America and the Caribbean (LAC). While the population cut-off for inclusion in the WDR was 3 million, these tables include almost all countries in the region. This update of the WDR expenditure assessments results in a substantial upward revision in the estimated percentage of GNP spent on health in Latin America in 1990. Thus, while the WDR estimated that 4% of GNP was spent on health in the countries of LAC, these revisions suggest that the figure was around 6%. Development of policies to contain cost escalation becomes correspondingly more salient. Preparation of this document was enormously aided by collaboration with staff and consultants of the Pan-American Health Organization (PAHO). If the document proves of value, we expect to update it every few years, ideally in continued collaboration with PAHO and the Inter-American Development Bank. Sri-Ram Aiyer Director Technical Department Latin America and the Caribbean September 1994 - v Acknowledgments The authors gratefully acknowledge the contribution of colleagues at Harvard University, The World Bank (particularly the WDR 1993 team), and various U.N. agencies, whose input and assistance made this study possible. Special thanks are due to participants from the World Bank, PAHO, and IDB (in particular Philip Musgrove, Emesto Castagnino, Rubdn Sudrez, Cdsar Vieira and Pamela Henderson) at a series of meetings on Latin-American health expenditures in Washington, D.C. Chapter I. Introduction Measurement of health expenditures is important from several perspectives. The satisfactory realization of the objectives of health planners - effective planning and management of health programs; intra-sectoral and intersectoral priority-setting and resource allocation; assessment of the distributional impact of health programs; planning an optimal public-private mix in the provision and financing of health services; and undertaking research into the determinants of changes in health status - is dependent, to a significant extent, on the availability of reliable, up-to-date data on health spending. The absence of standardized and systematic national health expenditure estimates, thus, is often a stumbling block in undertaking economic analyses of the health sector at the intra-national and cross-national levels. As health care is financed by a multitude of sources in most countries, tracking overall health spending, without a sustained and coordinated effort, is often a difficult task. In Latin American countries, many of which have sizable health spending through various government, parastatal, and social security institutions, this task is particularly complicated. The purpose of this study, commissioned by the Latin America and Caribbean Technical Department of the World Bank, is to document, in detail, the levels and trends in health spending in the Latin American countries. As such, it represents an extension of a study on health expenditures undertaken by the authors as background for the World Development Report (WDR) 1993, which was the first comprehensive and systematic effort to compare global health spending (Murray, Govindaraj, and Chellaraj, 1993). We have attempted, as part of this study, to estimate health spending in many smaller Latin American countries, that were excluded from the WDR background paper (see Appendix 1 for list of countries included in database). In addition, we present a more detailed breakdown of spending by ministries of health in these countries. The objectives of this study are five-fold: a) To access existing information on national public, private and social insurance health expenditures from governments, international agencies and ad-hoc studies; b) To explore the relationships between national health expenditures and important socio-economic variables, using econometric models, for out- of sample prediction and analysis of health expenditures; c) To use the available information on health expenditures and the predictive equations derived from (b) to ascertain the level of national -2- health expenditures in every country in Latin America for 1990; d) Where possible, to analyze patterns of expenditure disaggregated by activity, type, and source of finance; and e) To carry out analysis on time series data on health expenditures (in the decade preceding 1990) for countries where such data are available. This study has benefited substantially from interactions and several rounds of discussion with the team from PAHO, responsible for estimating health spending in the Latin American and Caribbean region for the 1994 annual report on "Health Conditions in the Americas", and the Inter-American Development Bank . Indeed, these efforts should be seen as a potentially continuing collaborative relationship between these institutions in monitoring on a regular basis the spending on health in Latin America. It should be noted, however, that this study differs, in important respects, both from the study on health expenditures for the World Development Report (WDR), 1993, and from the PAHO study. Thus, the estimates in this study represent a more up-to-date and comprehensive assessment of spending in this region compared to the WDR (see Appendix 18 and 19 for a description and explanation of the differences between the estimates of this study and WDR 1993). The study is also not strictly comparable with the PAHO study, as the definitions of the various categories of the health sector; the methods used (e.g., the predictive regression equations are based on a global, rather than a regional analysis); and the sources of information (e.g., on expenditures and of the independent variables used in the regressions), are somewhat different. These differences have persisted, despite attempts made to resolve them. The differences are, however, not dramatic, and it is hoped that future collaborations will help in explicating them. The paper is structured as follows: Section II provides a brief review of previous literature on health expenditures; Section III describes the definitions, methods and materials used in this study; Section IV discusses the regressions undertaken for the prediction of public and private sector health expenditures; Section V summarizes the key results of the study; and Section VI discusses the results and the main conclusions. Details of the definitions and methods, and of the results, are provided in the appendices. - 3- Chapter HI. Past Studies of National Health Expenditures For a detailed review of previous studies on national and international health expenditures, readers are referred to the WDR 1993 background paper (Murray, et al., 1993b). To provide the context for this study, however, four themes from past studies, that we have alluded to elsewhere (Murray, Govindaraj and Musgrove, 1994), should be highlighted. First, the information on health expenditure has evolved considerably in the past three decades in industrialized countries but not in developing countries. Abel-Smith (1963, 1967) was the first to try to standardize cross-national data by defining the constituent components of health services, listing the main sources of finance, and laying down a standard classification of expenditures which he applied to several industrialized countries. His efforts were followed by a series of comparative studies, including the development of an standardized, annual database on OECD health expenditures. The development of health expenditure data for developing countries has been less successful. WHO, PAHO, USAID and the Sandoz Institute for Health and Socioeconomic Studies have attempted to improve information by promoting household surveys and publishing manuals for estimating national health expenditures. Despite these efforts, most estimates of national health expenditure come from ad hoc studies or development agency missions to countries, often conducted over brief periods of time. Consequently, the unpublished literature from agencies such as the World Bank remains an important but difficult to obtain source of expenditure estimates for developing countries. Regional reviews drawing largely on these sources have been prepared for Asia (Griffin, 1992), Africa (Vogel, 1989), and Latin America (McGreevy, 1992). Second, many cross-sectional studies have explored the determinants of national health expenditure, particularly in OECD countries. Taken together, these studies show that income per capita explains most of the variance in health expenditure per capita; Newhouse (1977), for example, found that 90% of the variance in OECD health expenditure was explained by GDP per capita. Some studies report that other variables such as reimbursement methods, institutional variables and the inpatient/outpatient mix can explain some of the variance in health expenditure. Nevertheless, the strongest factor in nearly all studies, including those few which examine developing countries, has been income per capita. Most studies have also found that health expenditure has an income elasticity greater than one. On this basis, Newhouse (1987) concludes that health expenditure in OECD countries must be purchasing caring (which is more of a luxury) rather than curing (which seems to be more of a necessity). However, Parker, Maguire and Yule (1987) take issue with the empirical observation that health expenditure has an income elasticity greater %I - 4 - than one and challenge the interpretation of health care as a luxury good. Third, most studies at the household level in developed countries do not show a greater than unitary elasticity for health expenditure with respect to income. The discrepancy between the relations at the national level and at the household level has been attributed in Canada to non-price rationing, so that consumers do not buy health care to the point of zero marginal utility when the price is zero to the consumer (Culyer, 1988). Relaxing such non-price rationing with increasing aggregate income leads to a much greater income elasticity (even exceeding unity as empirical studies have shown) at the macro level than at the level of the household (Gerdtham, et al., 1991). As we have argued elsewhere (Murray, et al., 1994), the implication that high-income consumers are more rationed than those with lower incomes seems far less persuasive than the explanation that social insurance plays a greater role in financing large health care expenditures with increase in household income. This claim seems to find some substantiation in a health sector study on Mexico, currently being undertaken by the World Bank (Chellaraj, 1994). The results of this study show that, in Mexico, the total share of out-of-pocket health expenditures declines significantly, among the rich more than the poor, with increases in insurance coverage through government social security institutes and private insurers. Finally, few studies on either OECD countries or developing countries have examined public health expenditures and private health expenditures and their determinants separately. Musgrove's study (1983), using household survey data from six Latin American countries, is a noteworthy exception. In these countries, private care had a higher income elasticity than public sector health expenditures, suggesting that private care is a luxury relative to public care and that consumption shifts from public to private, ceteris paribus, as household income rises. This may be partly attributed to differences in real or perceived quality which make private and public health care only imperfect substitutes. The finding that higher income shifts expenditure to the private sector is not generally observed at the aggregate level, when countries outside Latin America are also studied. - 5 - Chapter IH. Defmitions, Methods. and Materials A. Defrintions Assessment of health spending requires a consistent definition of expenditure and agreement on how to group spending by different agents and express its value in internationally comparable terms. Further, defining health expenditure requires defining health, the set of health promoting activities, and the subset of activities that promote health to be included in the health sector. These and other related issues are discussed by us in Murray, et al., 1993b, and, therefore, not reiterated here. We present below, for the convenience of the reader, some key definitions relevant to this paper. The operational definition of health expenditures used in this study, along the lines of the previous study, includes all expenditures incurred on preventive and curative health services for individuals, and on population- based public health programs, as well as some programs with a direct impact on health status (e.g. family planning programs, nutrition programs, and health education but not other kinds of education). Programs that only indirectly affect health, such as relief and food programs, and environmental programs related to water and sanitation, were excluded. We had hoped to be able to estimate health expenditure according to who pays for it and also who provides it. Categorizing health expenditure according to both provision and financing of services (represented as the two axes of the matrix) by the government, parastatal agencies and the private sector defines a 3X3 matrix. In this matrix, typically, data were available for the total financing provided by each of the three sub-sectors. The breakdown of government financing for services provided by the government itself, by parastatal agencies and by the private sector was also often available. However, data were rarely found for the other cells in the matrix. The study was therefore restricted to the financing of health services by the various sectors. Domestic expenditures for each country are classified as government, parastatal, or private sector spending (see Appendix 2). Total health expenditures comprise these expenditures and external assistance. Government health expenditure is defined as spending on health by the government at various administrative levels or by institutions wholly controlled by the government. Parastatal expenditures consist of the health components of social security and social insurance programs, and the expenditures on health by other parastatal agencies. It should be noted here that, in some rare instances (e.g. in the case of the Instuto Assist. Medica Collectivo [IAMCs] - mutual fund agencies in Uruguay), the decision to classify an organization as parastatal or private was difficult. In such cases, decisions had to be made, somewhat arbitrarily, based on the extent of the organization's funding through -6- private or public sources. Public expenditures are defined as the sum of government and parastatal expenditures (and external aid), to permit comparisons with the OECD countries where expenditures on health-related social insurance and social security programs are not distinguished from government expenditure. Private sector health expenditure refers to spending by all non-governmental entities, including individuals, households, private corporations and non-profit organizations. Private expenditures are the sum of private institutional and individual expenditures (including both direct or out-pocket costs and purchase of insurance by institutions and individuals or households) B. Methods 1. Currency Conversions The base year for the study is 1990. For countries with estimates prior to 1990 but no data for 1990, it was assumed that spending on health as a share of GDP was the same in 1990 as in the year of the most recent estimate. Estimates of expenditure in 1990 local currency have been converted into US dollars (US$) using 1990 official exchange rates. The results were also calculated in "International dollars" (I$) using purchasing-power parity (PPP) ratios from the World Bank's modification of the United Nations International Comparisons Program (ICP Phase V, 1985). In calculating expenditures in international dollars, external assistance, which is primarily paid in US. dollars or other hard currency, was assumed to fund only tradable goods, so it was not corrected for purchasing power parity. Purchasing power parity ratios calculated specifically for the health sector would be preferable to those based on total GDP, since GDP PPP rates measure "non-health consumption foregone" rather than the more appropriate "health care output". However, while a complete set of GDP purchasing power parity ratios for all the countries of the world have been estimated by the World Bank, as noted above, health-sector PPPs are available only for the countries for which these figures were estimated in ICP Phase IV and V (ICP project, 1980 and 1985). Therefore, only the GDP PPPs have been used in estimating national health expenditures in international dollar terms. 2. Disaggregation of Government Expenditures An attempt was initially made to be comprehensive in the categorization of the government health expenditures, i.e. attempts were made to track allocations to all major government health programs. Given the extreme paucity of disaggregated data and other limitations that made cross-country comparability difficult, the number of functional categories had to be - 7 - restricted. Besides the fact that many countries do not provide in their budgets a breakdown of expenditures by individual programs, the categorization of expenditures often varies from one country to the other. In fact, definitions are often inconsistent even for those expenditure categories, e.g. secondary care, where one might expect congruity across countries and over time. Therefore, only those categories of expenditures were retained in this study that lent themselves to a reasonably consistent definition across countries. As part of the study, govemment health expenditures for each country that had a reasonable level of disaggregation were classified: a. by activity i) MCH/Family Planning services - All govemment expenditures on matemal and child health and population/family planning programs; ii) Primary Health Care services - Services (preventive and curative) rendered at the first level of health care delivery, e.g. at rural and urban dispensaries, health centers, polyclinics or outpatient departments of hospitals (excluding MCH/FP services); iii) Secondary services - Services rendered by primary and secondary level hospital inpatient departments and tertiary care centers; iv) Administration - Administrative expenditures at the central and local levels; v) Other - All other programmatic activities and services, such as immunization, vertical disease control programs, vector control programs, nutrition programs, health education programs, etc., and b. by type i) Capital expenses - All investments involved in the creation of physical capital, either at initial set-up or during augmentation of health programs and services, such as buildings, machinery, other equipment, vehicles, etc. ii) Recurrent expenses - Periodic expenditures involved in running the programs and services, such as 1. Salaries - Includes personal compensation in the form of salaries, wages, and other allowances, 2. Drugs and supplies - includes pharmaceutical and drug supplies, and other hospital and clinic supplies needed for treating patients, and 3. Others - All other recurrent expenditures, such as transport and travel, including pr diem payments (but not reimbursement of expenses), personnel training, maintenance, utilities, and other miscellaneous expenditures. - 8 - C. Materials Considerable effort was invested in obtaining data on public, parastatal and private health expenditure directly from governments, supplemented with reports and data from the World Health Organization, PAHO, the World Bank, the International Labor Organization, Regional Development Banks (i.e., the IDB and CDB), and the United Nations Statistics Division as well as the published literature. The collection includes material from numerous reports, articles and budgets, much of which is not published. Comparability across data sources was a significant problem for all three subsectors. For several countries there is a wide divergence in the quoted expenditure figures for the same year across data sources, and over fairly short periods of time (which may be explained by radical changes in the levels of spending from one year to the next for some countries, but seem very unlikely for others). Discussions with the country officers at the World Bank or with people familiar with those countries led to a choice of which estimate was most plausible (see Appendix 3 for sources of chosen expenditure estimates). The order of selection of sources of government health expenditures was, in general, government budgets, followed, in descending order, by World Bank studies, GFS data, and ad-hoc studies. There were some exceptions to this general algorithm. For example, for some Latin American countries, where hyper inflation and changes in currency rendered sensible calculations of health expenditures impossible, data from one of the other sources were used, even if the data were not the most recent available. Similarly, if, for example, the World Bank studies had captured local expenditures which neither the budget nor the GFS were successful in capturing, the expenditure figures from the World Bank study were chosen over the budget. The primary source of information on social security programs was the ILO, with some augmentation through World Bank and ad-hoc studies on social security programs. The order in which a selection was made between the various sources of information on private expenditures was adjusted household survey estimates (the adjustment procedure is discussed below), World Bank studies, national accounts data, and ad-hoc studies - the latter two, where no other source was available. Information on government health expenditures was available, for the years ranging from 1982 and 1990, for a total of 34 countries, i.e., almost all the Latin American countries with the exception of some of the very small islands. These islands have been left out of the dataset since accurate information was not available for them on even the determinants that have been used to predict health expenditures. Their exclusion, however, is unlikely to make a significant difference in the overall regional expenditures, given the small levels of health expenditures in these countries. The information for the majority of the countries on which studies on government expenditures were - 9 - available (often from multiple sources for some countries) was post-1986. Our final data set (which includes the sources that in our assessment offered the most detailed and, possibly, the most accurate estimates of government health expenditures), consisted of information on fifteen countries from Government Budgets, for ten other countries from the Government Financial Statistics (GFS) published by the International Monetary Fund, and for an additional seven countries from World Bank country studies. Information on parastatal spending was similarly available for most of the major Latin American countries. For some Caribbean islands (e.g. Antigua and Barbuda, St. Vincents, Suriname, etc.), however, data on health expenditures through the various social security institutes were not available. In these cases, our estimates are probably the lower bound of health spending in the public sector, although it may be that health-related social security in these countries is not substantial. In our data set, information on eighteen countries was obtained from ILO studies, for an additional eight countries from World Bank health sector studies, and for a further two countries from ad-hoc sources. Even using multiple sources, reasonable data on private sector spending were available for only nineteen countries for the period 1975 to 1991. In our final data-set, data on seven countries were compiled from household surveys, for one country (Costa Rica) from a WHO country study, for two countries from the National Accounts surveys, and for an additional nine countries from World Bank studies (some of which have used data from household surveys). As we have pointed out elsewhere (Murray, et al., 1993b), though, even when these assessments were based on surveys--either institutional or at the household level--many estimates were suspect. Household surveys, although widely acknowledged to provide the most reliable assessment of private spending on health, often exhibited systematic sampling and non-sampling bias. For several reasons, including non- representative sampling, many household surveys in developing countries may overestimate per capita private consumption. However, private health expenditures as a share of total private expenditure may not be biased if the income elasticity across households is close to one (and any bias in the data is independent of income). Household survey results were therefore adjusted by applying the percentage of household spending on health from these surveys to total private consumption numbers from the national accounts, to estimate private sector financing. Since figures were not available for several countries, however, we had to predict the numbers for these countries based on a regression equation. It is important to re-emphasize here that the data-set used for private health regressions in this study differs, to some extent, from the one used in the WDR 1993 study, in that estimates for certain countries have been - 10 - updated, and, further, only those estimates which included household and private institutional spending have been used in the new set of regressions. Also, the criteria for selecting estimates for countries was far more stringent, in that estimates that could not be verified were excluded. - 11 - Chapter IV. Estimating Out-Of-Sample One of the objectives of this study is to estimate total health expenditures in 1990 for every country in Latin America. Although public sector data were available for all the major countries, for fifteen countries in Latin America, as noted, there was no information on private sector expenditures. Predictive equations were, therefore, developed to estimate, out of sample, the private sector expenditures, as needed (see Appendix 5). These predictive equations are based on a sample of countries from the whole world (the data set, as noted, is an updated version of the one used for the study on health expenditures for the WDR 1993), rather than only the countries of the Latin America region, since the larger sample size was expected to provide better results. A dummy variable for Latin America was introduced to capture any regional differences in health expenditures from the overall global spending pattern. In constructing the private sector predictive equation, it was assumed that private sector expenditure is a function of public sector expenditure, but not vice-versa. This hypothesis is grounded in the belief that while most governments are probably largely unaware of the magnitude of the private health sector, or at least do not take it into account in determining public budgets (and, therefore, the public sector expenditure may not be a function of private sector expenditures), the health services that people are willing to buy for themselves, in contrast, may depend on what the public sector is already financing. Hence, in examining the determinants of public sector expenditures, we have not included private sector expenditures as one of the predictors. We have argued that the private sector is sensitive to the size of government financing of health services, and that, therefore, public sector expenditures needs to be used as an independent variable in the private sector equation. There are however two reasons why observed private spending cannot simply be regressed on observed public expenditure. First, the global private sector estimates span 16 years from 1974 to 1990. Estimates of public sector expenditure are not always available for the same years. Second, if private sector expenditure is a function of GDP per capita, other socio- economic variables, and public sector health expenditure, while public health expenditure is also a function of GDP per capita, the parameter estimates from OLS regression will be biased. To deal with both problems, we had to first develop public sector regressions (see Appendix 4) to predict public sector expenditure in the same year as the private sector expenditure estimate, effectively creating an instrumental variable for public sector health expenditure (the public and the private sector regressions were also used in analyzing the determinants of such - 12 - expenditures). The independent variables, GDP per capita and government consumption as a share of GDP, were also taken from the same year as the private sector estimate in generating the instrumental variable. We have assumed, in effect, that the functional relationship between the share of GDP spent by the public sector on health and GDP per capita and government consumption has not changed over the last 16 years. A discussion on the regressions for public sector health expenditures is now presented. A. Estimating Public Sector Expenditure We examined the relation between public sector health expenditure and GDP per capita, government consumption as percent GDP, private consumption as percent GDP, life expectancy at birth, the infant mortality rate, percent urban population, average years of schooling completed, literacy (which was not correlated with the average schooling variable), hospital beds per capita, and regional dummy variables. In addition to these variables, we added a dummy variable for former British colonies which gained independence after World War II and another for former French colonies, on the assumption that colonial history might play a significant role in explaining the variance in public health expenditures. Regressions were estimated in both US dollars and International dollars; in each case the dependent variable, public sector health spending, was measured both per capita and as a percent of GDP. The independent variables were derived primarily from sources at the World Bank, with some augmentation from the IMF (Government and International Financial Statistics) and other UN agencies. For the per-capita specification, univariate tests with the different independent variables showed closer association with the logarithm of expenditure than with expenditure itself. Strong univariate relations were observed, among others, for public sector expenditures per capita as a function of GDP per capita in US and International dollar terms (RA2 of 0.91 and 0.85, respectively), and of health status indicators such as infant mortality and life expectancy at birth. However, as argued in the WDR 1993 background paper, close relations between public sector expenditure denominated in per capita terms and income per capita are not so impressive as one might assume, since even randomly generated expenditure shares can suggest a close fit between per capita expenditure and per capita income. A more exacting test of the relation between public health expenditure and income as well as other independent variables that are highly collinear with income is to examine public health expenditure as a share of GDP, which is the specification used in the regressions. We tested the most general model first, using all the independent variables. Non-significant independent variables were dropped until the most parsimonious form was generated. Groups of independent variables were F- - 13 - tested, and retained if the F-test was significant. Four parsimonious regressions were estimated for the share of GDP: linear forms with independent variables in US$ and in I$, and double-log forms with independent variables in US$ and in I$. For prediction, we chose the form with the highest adjusted R2. This equation: Public Health Expenditure as % GDP = 0.02 + 8.54E-7 GDP per Capita + 0.09 Government Consumption as % GDP - 0.03 Dummy for MEC - 0.02 Dummy for OAI - 0.02 Dummy for LAC - 0.03 Dummy for SSA shows public expenditure on health as a share of GDP to be a linear function of GDP per capita in I$, government consumption as a percent of GDP, and dummy variables for MEC, OAI and SSA (which are indistinguishable from one another) and LAC (All coefficients are non-zero at the 0.01 confidence level). The adjusted RA2 was 0.79. Higher income was associated with a higher share of income spent on health; the elasticity from the double-log form was 1.43 (1.34 in US$). Governments that consumed a larger share of GDP in total also had higher expenditure on health. The significant dummy variables indicate greater regional differences in share of GDP spent on health than can be explained by income per capita alone. However, the infant mortality rate and life expectancy at birth were not related to public sector health expenditure. Thus the equation says nothing about the causal relations between expenditure and health status (we will return to this question in the discussion section). B. Estimating Private Health Expenditure Based on the same arguments as the public sector regressions, private sector health expenditures as a percent of GDP (rather than private health expenditures per capita) was analyzed as the dependent variable. As before, regressions were run using US dollar and PPP-adjusted incomes. All independent variables were from the same year as the private expenditure estimate, for each country. The steps used in obtaining the parsimonious models were, essentially, similar to the public sector regressions. In addition to the variables included in the public sector regressions, however, we added a dummy variable (opecdum) for countries that have revenues from oil exports that made up a greater than 50 percent share of their GDP, since such revenues provide an accessible source of financing for public sector social programs, including health. The expectation was that a well financed public health system would decrease the level of private spending and, therefore, that the coefficient on this dummy variable in the private sector regression would be negative. - 14 - As for the public sector expenditures, four parsimonious regressions were estimated for private expenditures as a share of GDP: linear forms with independent variables in US$ and in 1$, and double-log forms with independent variables in US$ and in I$. Using the updated version of our private sector dataset, we obtained a reasonably good relationship between private health expenditures as a share of GDP and the independent variables in our dataset - a result different from our previous study (Murray, et al., 1993b). What is striking, however, is that there is still no relationship at the aggregate national level between income and private health expenditures. As we had shown in our previous study (WDR 1993 background paper), the global income elasticity of private health expenditures is 1.03 (i.e., indistinguishable from unity). In other words, the share of GDP privately spent on health is nearly constant over the range of GDP per capita. This result is further confirmed in this analysis. Also, notably, public sector health expenditure was not significant in any of these regressions. The dummy variables for colonial history, meant to capture potential institutional effects, were also not significant. The equation used in predicting private health expenditures was, as for public expenditures, the one with the highest adjusted R2. The equation Private Health Spending as % GDP = -3.2 - 3.7 Government consumption as % GDP - 0.4 Human Capital Stock - 0.7 Dummy for OPEC + 0.07 Beds per capita shows private expenditure on health as a share of GDP to be a function of government consumption as a percent of GDP, the human capital stock (schooling) variable, beds per capita, and the dummy variable for the oil producing countries (All coefficients are non-zero at the 0.01 confidence level). The adjusted RA2 was 0.29. As expected, governments that consumed a larger share of GDP in total also had lower private expenditure on health. The significant (negative) coefficient on the dummy variable for the oil producing states suggests that, as anticipated, these countries have a lower level of private health spending than would be otherwise predicted. Countries with a higher bed per capita, on average, seem to have higher private health expenditures. Also, countries with, on average, a higher number of years of schooling (humcap) have a lower level of private expenditures. - 15 - Chapter V. Results The regional spending on health care in Latin America and the Caribbean in 1990 was estimated by combining observed values with those predicted by the regressions for the public and private subsectors. These regressions estimates were used for 15 countries in the private sector, but they account for only about one percent of estimated total expenditure in Latin America, because the great bulk of spending occurs in countries for which data were available and it was not necessary to predict values from the equations. Estimates of public, private and total health expenditure are provided in Appendices 9-11 for every country: estimates derived from the regression analysis are in bold face. A. Regional Health Expenditures (A Comparative Perspective) Global spending on health care in 1990 was a little over U.S.$ 1.7 trillion, which constituted a little more than 8 percent of global GDP (see Appendix 6). Of this, almost 86 percent of spending occurred in the demographically developed countries of the world (i.e. the established market economies and the former socialist economies of Eastern Europe). Spending in the Latin American and Caribbean region constitutes just more than 4 percent of global spending and approximately 36 percent of spending in the demographically developing regions of the world. As a share of GDP, health spending in the LAC region is about 6.3 percent, which is significantly greater than the spending in China (3.5%), the Middle-Eastern Crescent (3.6%), Sub- Saharan Africa (4.2%), Asia (4.5%), and somewhat more than the spending in India (6%). In general, even as a share of GDP, spending in the developing regions is significantly lower than in the market economies, which averaged almost 9.3% in 1990. When expenditures are corrected for purchasing parity, global spending amounted to over 1.9 trillion International dollars (see Appendix 7). Adjustment for purchasing parity increases expenditures in developing regions substantially (422 versus 193 billion dollars), without significantly changing the health spending in the developed countries. Health spending in the Latin American region in international dollars is about 6 percent of global health expenditure, and about 28 percent of the total spending on health in the developing regions. While globally, the public sector (in US or International dollars) accounts for 60 percent of the total health spending with private sector financing constituting the other 40 percent, in the LAC region the public sector share of total health spending (despite the sizable social security spending) is only 49 percent (see Appendices 8 and 9). Thus, spending by both the public - 16 - sector and the private sector is about 3.1 percent of GDP. This is in keeping with the trend in developing regions for the private sector to play a significantly greater role (inversely correlated with aggregate income, as noted by us in the WDR 1993 background paper) in the financing of health care. External aid plays a very small role in total health financing in the LAC region, accounting for just less than 1 percent of the total health expenditure (see Appendix 9). Aid, however, is a significant component of total health expenditures for certain countries (see Appendix 10) in the Latin American region (e.g. Guyana ), although the dependence on aid is nowhere as significant as it is in Sub-Saharan Africa, where aid flows as a percent of total regional health expenditures comprise between 9-10 percent, and, for some individual countries is greater than 50 percent of total expenditures. Per capita total health spending in 1990 in the LAC region (see Appendices 6, 7, and 12) was about $162 (280 International dollars), which was significantly greater than spending in any other developing region of the world, and even the former socialist economies (which spent, on average, US$142 or I$ 240 per capita). For example, China and India spent only about $11 and $20, respectively on health (although the amount does go up when converted to International dollars, i.e., I$72 and I$61, respectively). Given the high income levels and the high share of GDP spent on health in the market economies, however, their spending of $1869 (I$ 1793) per capita dwarfs spending in the other regions of the world (see Appendices 6 and 7). B. Time Trends The trends in government health expenditures, where such data is available, vary among the countries of Latin America and the Caribbean (see Appendix 13). It should be noted that these trends represent comparisons of only the government health expenditures, except for a few countries (e.g. Colombia, Paraguay, and Peru) where, in the absence of disaggregated information for previous years, the estimate for 1990 maintains comparability by including parastatal expenditures. There are at least five discernible patterns of health expenditure during the period 1980-90. These are, respectively, a rising, falling, or constant trend; or a pattern of spending where expenditures grew, initially, followed by a drop, or vice versa. Thus, for example, countries, such as Peru, Paraguay, and Venezuela, experienced steady declines in health expenditures as a proportion of GDP. Others, such as Ecuador, Jamaica, and Mexico, have had consistently increasing expenditures. Yet others, such as El Salvador, and St.Vincents, have had a relatively constant level of expenditures over the ten year period. Expenditures as a proportion of GDP, in countries like Argentina, the Dominican Republic, and Uruguay, declined initially (e.g. from 1.57% of the GDP in the Dominican Republic in 1980 to 1.48% in 1983), before climbing back (e.g. to about 1.64% of the GDP in the Dominican Republic) in 1990. - 17 - However, other countries, such as Barbados, Belize, and Trinidad, experienced declines in health expenditures as a proportion of GDP by 1990, following a rise in the 1980s. It is important, however, to emphasize a few caveats vis-A-vis these trends. First, these estimates of government expenditures were obtained from different sources, and, therefore, may not be strictly comparable. Second, the period between 1980-90 may or may not be representative of the general long- term trend in these countries. Third, even if these general patterns are accurate, different countries have different absolute levels of expenditures, and, thus, it is difficult to make any generalizations about countries which fall into any of the above five categories. Furthermore, changes in government health expenditures might have been adequately substituted by corresponding changes in social security spending. Unfortunately, this is difficult to ascertain from the available information. Finally, it is not entirely clear that these periods of fluctuations in health spending in these countries always correspond to any obvious socio-economic or political events, although, for some countries, the debt crisis and structural adjustment, obviously, have played a role. Despite these caveats, certain observations can be made on the patterns of government health expenditures. It is striking that the Caribbean countries, with one or two minor exceptions, experienced no major declines in health expenditures as a proportion of GDP. It is tempting to make an association between this observation, and the fact that, with the exception of Jamaica, these countries did not experience serious economic problems unlike the Latin American countries. It is also possible that these countries, unlike the Latin American countries, have made a deliberate attempt to keep their levels of health expenditures at a certain level, in spite of other problems. Thus it is, for example, that, despite economic problems, health expenditures as a proportion of GDP actually increased in Jamaica as a result of the emphasis given to the sector by successive Jamaican governments. It should be noted, though, that a few countries in Latin America too have experienced steady increases in health expenditures despite economic problems, e.g. Nicaragua and Panama. It is possible that the nature of the political regime in these countries is an explanatory factor. For example, in Nicaragua, health expenditures as a proportion of GDP rose from 3.2% in 1980 to 4.9% in 1990, as a result of the emphasis given to the health sector by the then-Nicaraguan govemment. A similar situation existed in Panama during most of the 1980s and the health expenditure as a proportion of GDP in 1990 stood at about 5.2%. - 18 - C. Disaggregated Data on Government Health Expenditures (1990) Appendices 9-11 show total health expenditures broken down into public sector, private sector and external assistance for each country in Latin America. Figures are provided in US dollars, International dollars, and percent of GDP terms. Total health expenditures as a share of GDP ranges from 9.7 percent in Guyana and 9.6 percent in Argentina, to 3.7 percent in Ecuador and 3.1 per cent in Peru. In terms of per capita total health spending in US$, the range is from almost $600 in the Bahamas and the Cayman Islands, to less than $50 in countries like Bolivia, Ecuador, Guatemala, and Haiti. Even adjusting for purchasing power, a number of countries still spend less than I$ 100 per capita (e.g. Bolivia, Honduras, Peru and Haiti). We present below the results of government health expenditures disaggregated into various categories (see Appendix 14). It should be noted that the sample sizes differ for each expenditure category, and, further, that the countries represented in each sample are different. Local and provincial expenditure on health as a proportion of total government health expenditures varies widely across countries. For example, in Colombia, over 90 percent of the expenditures (excluding transfers from the central government) are at the provincial and local levels, with the central government having a negligible role in the financing of health care services. In many countries, e.g. Argentina, public sector spending on health at the local and provincial levels constitutes a significant majority of all public sector health expenditures. On the other hand, in countries such as Chile, local expenditures are less than 10 percent of total public sector health expenditures. It should be emphasized that it is very likely that significant expenditures at the local level are often missed in assessments of public health expenditures, due to the difficulty in tracking them. Thus, estimates of local expenditures for several countries where such data is collected are likely to be conservative. Given the observed wide variability in the share of local expenditures, however, it seems meaningless to calculate any average figure for such expenditures, that might be applied to other countries where local expenditures might be expected but for which no information exists. Other ministries (i.e., other than the Ministries of Health) also spend a significant amount in providing health services. These expenditures include spending such as by Ministries of Defense for providing medical care to the armed forces, or by Education Ministries for providing school health services (but not social security health expenditures). It is probable that spending by other ministries for health care services is missed in many countries, and, further, even in countries with such information, the estimates are likely to be conservative. Unfortunately, however, there is as wide a variation in these expenditures, across countries, as there is in local health spending. Thus, it is - 19 - very difficult, without personal knowledge of a country, to predict the level of these investments in any particular case. Capital expenditures comprise, on average, about 15-20 percent of total government health expenditures in our sample of 16 countries. 7 countries had capital expenditures of less than 5 percent; 5 countries had spending between 5-15 percent; while 4 countries had capital expenditures of greater than 15 percent of total government health expenditures. While there is variation across countries, in the majority of countries in our sample (i.e., 12/16), capital expenditures were less than 15 percent of total health expenses. This observed pattern, however, give little indication of the level of capital investments across countries and over time, as capital investments vary dramatically between countries and from year to year. Expenditures on Primary Health Care Services (PHC) constitute, on average, about 18 percent of recurrent expenditures of the government, according to government budgets. In our sample of 12 countries, PHC expenditures of 10-20 percent of recurrent expenses were the most represented. 2 countries had PHC expenditures of less than 5 percent; 3 countries had expenditures on PHC between 5-10 percent; 3 countries had expenditures on PHC between 10-15 percent; while 4 countries had expenditures on PHC greater than 15 percent - of which only 2 countries had a greater than 35 percent share for expenditures on PHC services, none of them major countries in Latin America. While this is a small sample of countries, the data does indicate that, despite a decade and a half of rhetoric, the majority of health expenditures are not for primary health services. To an extent, this may reflect the higher levels of investment required in establishing and operating facilities for secondary and tertiary level services. Salaries, in our sample of 17 countries, constituted, on average, about 55 percent of total recurrent expenditures. This feature is very consistent with estimates from other countries, regardless of income and expenditure levels. Further, this pattern is stable over time. Most countries in our sample had salaries ranging between 50-75 percent of recurrent expenditures. Of the 17 countries, 12 countries had expenditures on salaries of between 40 and 70 percent; 2 countries had expenditures on salaries as a proportion of recurrent expenditures of greater than 70 percent; while 3 countries had salaries constituting between 10-40 percent of total recurrent expenditures. The portion of recurrent expenditures constituted by drugs and supplies was about 19 percent, in our sample of 17 countries. Unlike salaries, however, there is a wide variation in spending on drugs and supplies across countries. Of the 17 countries, 1 had expenditures on drugs of under 10 percent; 11 had expenditures between 10 and 20 percent; 3 had expenditures of between 20 and 30 percent; and 2 countries had drug expenditures as a proportion of recurrent expenditures of greater than 30 percent. - 20 - In this sample, there is no clear relationship between the income of the country and expenditure on drugs and supplies. A rich country is just as likely to spend a small (or large) proportion of its recurrent expenditures on drugs as a poor country. The claim that poor countries are more likely to spend a significant proportion of their recurrent budgets on procuring drugs does not hold in this admittedly small sample. This may indicate a tendency in most countries to have consumers buy drugs from private sources, rather than provide them through the government health services. Of course, if such a tendency were uniform, there might still be a relationship; the lack of a relation suggests that the differences across countries are random. - 21 - Chapter VI. Discussion and Conclusions A. Tracking Health Expenditures Our study of health expenditures in Latin America has demonstrated the significant gaps in the knowledge of health expenditures, particularly (but not exclusively) in the private sector. Although information on countries of this region are somewhat better than in some other developing regions, there is still much work to be done in this area. As noted in our earlier study (Murray, et al., 1993b), the World Health Organization (and PAHO), the World Bank, and the regional Banks (Inter-American Development Bank and the Caribbean Development Bank) have not, to date, devoted sufficient resources to maintaining a database on national health expenditures. If costly ad hoc studies are to be avoided in the future, data collection needs to be more systematized, and immediate attention needs to be directed towards this issue. As noted, measurement of private sector expenditures is particularly inadequate in the developing world. Even for those countries with detailed ad- hoc studies, the data are subject to doubt. One way forward would be the development of national health accounting akin to the OECD health expenditure database. However, the majority of developing countries probably cannot institute such information systems in the near future. Rapid assessment techniques therefore need to be developed and implemented, in conjunction with an international database on government expenditures, to fill the information gap in the short-term. The efforts of the International Monetary Fund and the ILO - that already maintain a database of government and parastatal expenditures, respectively - are steps in the right direction. This work could be supplemented by the other multilateral agencies, at little extra cost, so as to enable the creation of a coordinated and comprehensive database on health expenditures. It is, thus, important that the momentum generated by the collaboration between the World Bank, PAHO, and the Inter-American Development Bank on this Harvard study be sustained in the future. B. Income and Other Determinants of Health Expenditure The data reviewed in this study suggest that private health spending relative to GDP is unrelated to income. In other words, the income elasticity of private health expenditures is indistinguishable from unity. This is an important observation, since it suggests that the relationship of public and private health expenditures with income is essentially different. While, as previous studies have also found, public health expenditures as a share of GDP rises (at an even faster rate) with increasing income, no such correlation is - 22 - seen in the case of private spending. Previous studies that have studied health expenditure as a homogenous entity (by running a regression for total as opposed to individual public and private health expenditures) have, for obvious reasons, failed to make this differentiation between private and public sector expenditures. Furthermore, our analysis demonstrates that private expenditure as a share of GDP is unrelated to health status indicators, geographical region, or public health expenditure (see Appendix 5). It is particularly surprising to find no association with public expenditure, since it was expected, a priori, to provide an alternative to private expenditures. Apparently public and private spending are not simply substitutes, because they finance services which differ in kind, or quality, or in utilization by different population groups. As discussed above, private expenditures are a function of the size of government, the average years of schooling, the number of hospital beds per capita, and the dummy variable for the oil-producing countries. The association of the education variable with private health expenditures is reassuring since it conforms to our hypothesis that education influences people's understanding of their health needs and their demand for health care. The negative and significant coefficient on the dummy variable is also interesting. This suggests, as hypothesized, that oil resources represent an easy source of revenues that is used to fund publicly financed health services, resulting in declining private health expenditure. It should be emphasized that the proportion of variance in private expenditures explained by the identified determinants is comparatively small (although the results are much more significant than our previous study). It may well be that private health expenditure are determined in part by historical, cultural and institutional factors not captured in this analysis. Furthermore, errors or mis-specifications in the data may reduce the statistical significance of the variables tested. It is conceivable that further refinement of the dataset (through more carefully carried out private health expenditure surveys) might improve the results. We have shown that, in contrast to private spending, public health expenditure has an income elasticity substantially greater than one. Moreover, income has a dominant impact on public health expenditures as a share of GDP, and explains a large proportion (albeit not the entire amount) of the variance in the dependent variable (see Appendix 4). We have examined the income elasticities of the public and private sectors. What, however, is the income elasticity of total health expenditure? Is total health expenditure a luxury good? It is important to note that total health expenditure also includes external assistance that flows primarily to low income countries (Michaud and Murray, 1993). For all developing countries with observed data (not derived from our - 23 - estimating equations), a double-log regression of total health expenditure per capita against income per capita gives an elasticity of 1.003, which is indistinguishable from unity. In other words, the share of GDP spent on health does not increase with income. As noted above, however, average total health expenditure in EME is substantially higher than in all other regions, so a regression including these countries shows health care to be a luxury good. Compared to the pattern in poorer countries, high health expenditure in EME is not accounted for simply by higher average income. C. How Do We Expect Health Expenditures to Change with Income Per Capita? More income means more resources with which to deal with health problems. We have argued elsewhere (Murray, et al., 1994), however, that there may be two separate factors involved in the "health problems" which generates demand for health care: observed or objective health status and perceived or subjective health status. Numerous interview surveys in poor developing countries have recorded higher rates of self-reported morbidity and disability in rich than in poor households (Murray, et al., 1993a). We had argued that such counter-intuitive patterns of reported morbidity may be at least partly explained by changing expectations of health status. If expectations of good health increase faster than actual health status--because people have more access to health care, or because more education makes them understand more about health--then perception of ill health may increase with income. The result will be increasing expenditure which is only loosely related to objective health problems. Moreover, if, as we suspect, perceived health status is more of a luxury whereas treatment for objective health problems is more of a necessity, then the elasticity of the combined tendencies to spend might increase with income per capita as "health status" comes to be more a matter of subjective perception. However, as a population ages and develops chronic health problems that are costly to treat, even objective health status may generate pressure, with increasing income, to spend an increasing share of income on health care. Furthermore, the relation between total health expenditure and income per capita will be affected by the effect of the expanding role for the public sector in financing health care, that makes public spending respond to perceived health status and the demand for health services from the population and not only to objective needs. Any understanding of what accounts for health expenditure and how it is related to health status that goes beyond the superficial, therefore, will have to disentangle these effects. In order to illustrate some of these issues, and examine the income elasticities at different levels of aggregate income, we undertook separate regressions between income and (public and private) health expenditures, for countries with low (<$635), middle ($635-$7199), and high income. The - 24 - income elasticity for public sector expenditures were, respectively, 1.03, 1.46 and 2.04 for the three groups of countries. Interestingly, private health expenditures show the exact opposite trend. The elasticities for the three groups were 1.26, 0.95, and 0.66, respectively. A similar set of results are seen when the regressions are run for the different regions, ordered according to income. These results, while not presented here in the interest of brevity, are available with the authors for the interested reader. D. What Does Health Expenditure Buy? The relations studied here raise the perennial question about what health expenditures actually purchase - in particular whether they buy improved objective health ("curing") or something more subjective ("caring"), or whether they are largely wasted through inefficiency in the production of services and the choice of which services to provide. Using the improved estimates of national health expenditures including external assistance provided in this study, we can examine some relations between total health expenditures and measures of health status for Latin America. For example, life expectancy at birth can be analyzed in comparison to health expenditures per capita. Appendices 15 and 16 shows life expectancy at birth plotted against total health expenditures per capita and as a share of GDP. The per capita plot demonstrates the rise of life expectancy (at a decreasing rate) with increase in total health expenditures, while the plot of health expenditures as a share of GDP seems essentially random. Another such analysis is shown in Appendix 17. Over the last decade, there has been considerable attention directed to those countries, such as Sri Lanka, China, Costa Rica, Cyprus, and Cuba that have "good health at low cost" (Halstead et al., 1985). Analyses of good health at low cost are usually based on good health at low income, as a comparable set of total health expenditures have not been available before. Using the relationship for life expectancy and total health expenditures as a share of GDP, we can identify those countries in Latin America with higher or lower than expected life expectancy for their total health expenditures. Accordingly, GDP per capita and a Human Capital variable summarizing schooling levels were used to predict both observed total national health expenditure (as a share of GDP) and life expectancy at birth, for 58 countries worldwide. Once again the global relationships were used because of the larger sample size. The equations were: Total Health Expenditure as % GDP = -0.0485 + 0.0119 Natural Log GDP per Capita - 0.0055 Natural Log Human Capital and Life Expectancy at Birth (Years) = 41.98 + 3.120 Natural Log GDP per Capita + 5.316 Natural Log Human Capital. Estimates of expenditures derived from the regressions reported earlier - 25 - were not used in this exercise, which was limited chiefly by the availability of estimates for private health spending and the human capital variable. (Human capital was just significant at the 0.05 confidence level in explaining health expenditure; otherwise all variables were significant at the 0.01 level. In a similar analysis for 73 countries in WDR, 1993, human capital did not contribute significantly to explaining health spending.) The values of expenditure and life expectancy predicted from these equations were then compared to the observed values, and the differences or residuals plotted in the figure. The result shows for each country whether it spends more or less than might be expected, given its income and education level, and whether its population lives longer or less than might be anticipated. Although income, education and health expenditure are not the only factors influencing life expectancy, this comparison indicates, roughly, whether health expenditure in a given country is buying increased life to the same degree as in other countries with similar resources and human capital. Points in the upper right (e.g. Costa Rica, Chile, Colombia, etc.) and lower left quadrants (i.e., Peru) correspond to countries showing a systematic relation between more health expenditure and longer life: that is, spending on health appears to be buying more years of life. Countries in the upper left quadrant (e.g. Paraguay, Venezuela, etc.) achieve gains in life expectancy without spending so much-- their health expenditure appears to translate more effectively into improved health status. The data do not indicate, of course, whether this occurs for reasons directly related to how resources are spent in health care, or because the population takes better care of its health through diet and other habits and therefore needs less medical care to achieve the same result. Countries shown in the lower right quadrant (i.e. Bolivia) are in the opposite situation, with shorter than expected life despite spending more on health care than would be expected on the basis of income and schooling (see Appendix 17). Similar relations could be explored using other indicators of health status such as child mortality. The most interesting comparison would relate health expenditure to the total burden of disease in a country, including the effects of disability as well as premature mortality, as measured in DALYs (Murray, 1994). We cannot provide an analysis parallel to that above, however, for two reasons. First, the disease burden has so far been estimated only for the eight regions of the world and for a few individual countries. Secondly, even the regional estimates now available describe only the burden of disease remaining as a result of everything that has been done, including expenditure on health care, to improve health. In fact, comparison should really be made to the reduction in disease burden which can be attributed to health spending, which implies comparison to the situation that would exist in the absence of that expenditure. Such information, unfortunately, is not currently available. - 27 - Bibliography Abel-Smith, B. Paying for health services (Public health papers. no. 17), Geneva, WHO, 1963. Abel-Smith, B. An international study of health expenditure (Public health papers. no. 32), Geneva, WHO 1967. Anderson, 0. Health care: can there be equity? the United States. Sweden. and England. New York, John Wiley and Sons, 1972. Chellaraj, G. Private out-of-pocket expenditures and utilization in Mexico. Forthcoming LAC region monograph, The World Bank. Washington. DC, 1994. Culyer, A. Health expenditures in Canada: myth and reality: past and future (Canadian Tax Paper No. 82). Toronto, Canadian Tax Foundation, 1988. Culyer, A. .Cost supplement in Europe. Health care financing review. 1989 annual supplement: 21-32 (1989). Fraser, R.D. An international study of health and general systems of financing health care. International journal of health services. 3(3):369-397 (1973). Gerdtham U.-G. et al. Econometric analysis of health care expenditures: a cross-section study of the OECD countries. Center for Medical Technology, University of Linkoping, Sweden, 1988. Gerdtham, U.-G. & Jonsson, B., Conversion factor instability in international comparisons of health care expenditures. Joumal of health economics. 10: 227-234 (1991). Gerdtham, U.-G. & Jonsson, B. Price and quantity in international comparisons of health care expenditure. Applied Economics, 23: 1519-1528 (1991). Gerdtham, U.-G. et al. An economic analysis of health care expenditure: a cross-section study of the OECD countries. Joumal of health economics, 11: 1163-84 (1992). Gertler, P. and van der Gaag, J. The willingness to pay for medical care: evidence from two developing countries. Baltimore, World Bank, 1990. Getzen, T.E. Macro forecasting of national health expenditures. Advances in - 28 - Health Economics and Health Services Research, 1127-48 (1990). Glaser, W.A. Paying the Hospital. San Francisco, Jossey-Bass Publishers, 1987. Griffin, C.C. Health care in Asia: comparative study of cost and financing. Washington, World Bank, 1992. Halstead, S.B., Walsh, J.A., Warren, K.S. Good Health at Low Cost. Rockefeller Foundation Conference Report. New York, 1985. International comparison project. World comparisons of purchasing power and real product for 1980 (ICP phase iv). New York, United Nations, 1986. International Monetary Fund. Government financial statistics yearbook, Washington, IMF, 1991. Jonsson, B. What can Americans learn from Europeans? Health care financing review. 1989 annual supplement: 79-110 (1989). Kleiman, E. The determinants of national outlay on health. In: Perlman, M., ed., The economics of health and medical care. London, Macmillan, 1974. Leu, R. E. The public-private mix and international health care costs. In: Culyer, A. and Jonsson, B., eds., Public and private heath services. Oxford, Basil Blackwell, 1986. McGreevy, W. Social security in Latin America: issues and options for the World Bank, (World Bank discussion paper 110). Washington, World Bank, 1992. Michaud, C. & Murray, C.J.L. External assistance to the health sector in developing countries: a detailed analysis 1972-90. Bulletin of the WHO, June, 1994. Murray, C.J.L. et al. Adult mortality: levels, patterns, and trends. In: Feachem, R. et al., eds., Adult health in the developing world , Washington, World Bank, 1993. Murray, C.J.L., Govindaraj, R., Chellaraj, G.. Global domestic expenditures on health. Background Paper for WDR, 1993. Washington, World Bank, 1993. Murray, C.J.L., Govindaraj, R., Musgrove, P. Global expenditures in health. Bulletin of the WHO, June, 1994. - 29 - Murray, C.J.L. Quantifying the burden of disease: the conceptual basis for disability adjusted life years. Bulletin of the WHO, June, 1994. Musgrove, P. Family health spending in Latin America. Journal of health economics. 2: 2245-2257 (1983). Newhouse 1977. Medical care expenditure: a cross-national survey, Joumal of human resources, 12:115-125 (1977). Newhouse 1987 J. P. Newhouse. Cross national differences in health spending -what do they mean? Joumal of health economics, 6:159-162 (1987). OECD. Measuring health care 1960-1983: expenditure. costs, performance. (OECD social policy studies no. 2!. OECD, 1985. OECD. Health care expenditure and other data: an international compendium from the OECD. Health care financing review. 1989 annual supplement: 111- 194 (1989). Pan American Health Organization. Health Conditions in the Americas. PAHO, Washington, D.C., 1994 forthcoming. Parkin, D. et al. Aggregate health care expenditures and national income, Joumal of health economics. 6: 109-127 (1987). Pfaff, M. Differences in health care spending across countries: statistical evidence. Joumal of health politics, policy and law. 15(1):1-65 (1990). Poullier, J.P. Health data file: overview and methodology. Health care financing review. 1989 annual supplement: 111-118 (1989). Vogel, R.J. Trends in health expenditures and revenue sources in Sub-Saharan Africa. Washington, World Bank, 1989. Vogel, R.J. Financing health care in Sub-Saharan Africa: a 2olicy study. Phoenix, AZ, University of Arizona, 1992. White, K.L. Intemational comparisons of medical care. Scientific American, 233 (8): 17-25 (1975). World Bank. World development report 1993: Investing in health. New York, Oxford University Press for the World Bank, 1993. Appendices -33 - Appendix 1 Countries in the Latin America and Caribbean region Anguilla* Antigua and Barbuda Argentina Aruba* Bahamas Barbados Belize Bolivia Brazil British Virgin Islands Cayman Islands Chile Colombia Costa Rica Cuba* Dominica Dominican Rep. Ecuador El Salvador Falkland Islands* French Guiana* Grenada Guadeloupe* Guatemala Guyana Haiti Honduras Jamaica Martinique* Mexico Montserrat* Netherlands Antilles* Nicaragua Panama Paraguay Peru Puerto Rico* St. Kitts and Nevis St. Lucia St. Vincent Suriname Trindad and Tobago Turks and Caicos* Uruguay Venezuela Virgin Islands* * indicates countries not included in study dataset -34 - Appendix 2 Definition of Subsectors * Government health expenditure = Spending on health by the government at various administrative levels or by institutions wholly controlled by the government. * Parastatal expenditures = Health components of social security and social insurance programs, and the expenditures on health of other parastatal agencies. * Public expenditures = Sum of government and parastatal expenditures. * Private sector health expenditure = Spending by all non-governmental entities, including individuals, households, private corporations and non-profit organizations, i.e. the sum of private institutional and individual expenditures (including both direct or out-pocket costs and purchase of insurance by institutions and individuals or househords). I Total health expenditures = Government + Parastatal + Private + foreign aid. nscaain Lxl)enuirurcs in Larin America - sources ot E;xpenditure JEstimates Chosen Country Government Expenditures Parastatal Expenditures Private Expenditures Year Source Year Source Year Source Latin America &Caribbean Antigua and Barbuda 1990 Budget N.A. 1990 Predicted Argenlina 1988 GFS 1990 Cetrangolo 1985 WB Bahamas 1986 GFS 1986 ILO 1990 Predicted Barbados 1989 Budget 1986 ILO 1990 Predicted Belize 1983 Budget 1986 ILO 1980 H.Hold Survey Bolivia 1990 GFS 1985 ILO 1990 H.Hold Survey Brazil 1989 WB 1986 WB 1987 H.Hold Survey British Virgin Islands 1988 WB N.A. 1990 Predicted Ca% man lslands 1988 WB N.A. 1991 H.Hold Survey Chile 1988 Budget 1988 WB 1992 WB Colombia 1986 GFS 1980 WB 1985 WB Costa Rica 1988 WB 1988 WB 1986 WHO Dominica 1985 Budget 1986 ILO 1990 Predicted Dominican Rep. 1989 GFS 1986 ILO 1984 WB Ecuador 1990 GFS 1986 WB 1984 WB El Salvador 1987 WB 1986 ILO 1975 Nat.Accts. Grenada 1985 Budget 1986 ILO 1990 Predicted Guatemala 1989 GFS 1986 ILO 1990 Predicted Guvana 1983 Budget 1986 ILO 1990 Predicted Haiti 1986 Budget 1986 ILO 1985 WB Honduras 1986 Budget 1986 ILO 1990 Nat.Accts. Jamaica 1993 WB 1993 WB 1993 WB Mexico 1990 WB 1990 WB 1989 H.Hold Survey Nicaragua 1986 WB 1986 ILO 1990 Predicted Panama 1982 Budget 1986 ILO 1986 H.Hold Survey Paraguay 1990 Budget 1990 WB 1990 Predicted Peru 1984 Budget 1990 Nat.SS.Rep. 1990 H.Hold Survey St. Kitts and Nevis 1987 GFS 1986 ILO 1990 Predicted g St. Lucia 1986 Budget 1985 ILO 1990 Predicted 0L St. Vincent 1988 GFS N.A. 1990 Predicted x Suriname 1988 Budget N.A. 1990 Predicted Trindad and Tobago 1988 Budget 1986 ILO 1990 Predicted Uruguay 1992 WB 1992 WB 1992 WB Venezuela 1986 GFS 1986 ILO 1992 WB Note: N.A. implies N.ot Available - 36 - Appendix 4 Public Sector Regression Equation . Reg pubgdp gdpcap asia govcon afr lac mec Source I SS df MS Number of obs - 106 ---------+------------------------------ F( 6, 99) - 70.09 Model I .034143298 6 .00569055 Prob > F = 0.0000 Residual I .008037183 99 .000081184 R-square - 0.8095 ---------+------------------------------ Adj R-square - 0.7979 Total I .042180481 105 .000401719 Root MSE - .00901 pubgdp I Coef. Std. Err. t P>Itl [95% Conf. Interval] gdpcapus 1 8.54e-07 1.56e-07 5.475 0.000 5.45e-07 1.16e-06 asiadum I -.0238509 .0035661 -6.688 0.000 -.0309268 -.016775 goveon I .0927046 .0141226 6.564 0.000 .0646822 .120727 afrdum I -.0248389 .0033358 -7.446 0.000 -.0314579 -.0182199 lacdum I -.0191125 .0033532 -5.700 0.000 -.0257659 -.0124591 mcdum I -.027473 .0034568 -7.947 0.000 -.0343321 -.0206139 cons I .0210799 .0035832 5.883 0.000 .0139701 .0281897 pubgdp: Public health expenditures as a share of GDP gdpcap: GDP per capita asia: Dummy variable for OAI govoon: Governmnent consumption as a share of GDP afr: Dummy variable for africa lac: Dummy variable for Latin America ec: Dummy variable for NEC - 37 - Appendix 5 Private Sector Regression Equation * Reg priperx govper bedcap humcap opec Source j Ss df MS Number of obs - 46 ---------------------------------------- F( 4, 41) - 5.65 Model I 4.09952455 4 1.02488114 Prob > F s 0.0010 Residual I 7.4323142 41 .181275956 R-square = 0.3555 ---------+------------------------------ Adj R-square - 0.2926 Total I 11.5318388 45 .256263083 Root MSE = .42577 --------------------------------------------------------------------__-------_ priperx I Coef. Std. Err. t P>jtl (95% Conf. Interval] ---------+…___________________________________________________________________ govper I -3.6894 1.260262 -2.927 0.006 -6.234551 -1.144249 bedcap I .0725247 .02609 2.780 0.008 .0198349 .1252146 humcap I -.3718042 .1061984 -3.501 0.001 -.5862761 -.1573323 opecdum I -.674999 .3132419 -2.155 0.037 -1.307604 -.0423942 -cons I -3.183827 .2268098 -14.037 0.000 -3.641879 -2.725775 ---------------------------------------------------------__------------------_ priperx: Private health expenditure as a share of GDP govper: Government consumption as a share of GDP bedcap: Hospital beds per capita humcap: Human capital stock variable opecdum: Dummy variable for oil-producing states Regional Total Healtli Expenditures in 1990 United States D)ollars I '990 'I) Ibic 199(1 Ivriv atc I 99"( Aid I k iws 'I o1aI 'I otal l ikallth I K xpenditllrcs 1'99() ( liI) IlcVallth F\PCIil1UTeS I Iiilih ENpcnditures ior I Icaltlh I-ealhl Expcnditurces lcr (Capita tE{I(ill)N (NijilionM 190 UISSS) (N1ilIlio, 190( IJSs) 'milliiin l() )t ISS) Nlillion i(9)9) I NS ) (Million 1990 USS) - .",) As% I P 1')JiSS) Lstablishicd Markct Icolfnonlics 15,974.547 905,998 577.287 0 1,483,285 9.29% 1,869 Nliddlc IKasIcrn (Ccswiti 1,248,990 25,414 18,887 330 44,631 3.57% 88 I-otriorl\' Socialist Iconolliics 1,380,409 34,864 14.250 0 49,114 3.56% 142 ol' Liir. India 291,561 3,499 13,703 286 17,488 6.00% 20 (lina 365,557 7,494 5,248 77 12.819 3.51 % 11 00 Other Asia aid Islands 817.304 13,972 22.303 542 36,817 4.50% 53 Stub-Saliaran Africa 275,580 5,102 5,432 1,072 11.607 4.21% 22 |Latin Anicrica and tht (icarihhcan 1,109,135 34,104 34,598 542 69,243 6.24% 162 TO( AI 21,463.083 1,030,447 691,708 2.848 1.725,003 8.04% 354 x 0s Regionial Total Health Expenditures in 1990 International Dollars 1990 Public 1990 Private 1990 Aid Flows Total Total llealth ExpenditLres 1990 GDP Healti Expenditures Healil Expenditures for Health Healtil Expenditures Pler Capita REGION (Million 1990( lt (Million 1990 Sy (Million 1990 IS) (Million 1990 IS) (Million 1990 IS As% (3DP 1990 1$ Establisied Market Economics 15,202,504 864,110 565,850 0 1,429,961 9.3% 1,793 Middle Eastern Cresceit 2,091,124 41,722 34,076 330 76,128 3.6% 167 Formerly Socialist Economies 2,214,726 58,849 22,950 0 81,799 3.6% 240 of' Europe India 878,687 10,544 41,298 286 52,128 6.0% 61 China 2,346,464 48,103 33,685 77 81,864 3.5% 72 ()herAsiaand Islands 1.777,383 24,131 42,936 542 67,608 4.5% 111 Sub-Saharan Africa 669,148 11,278 12,466 1,072 24,858 4.2% 50 I,alin Amierica and the Carihbean 1,990,934 60,308 58,533 542 119,382 6.2% 280 tD x Regional Total Health Expenditures as Percent of Regional Total Health Expenditures Public Health Exp. Private Health Exp. Aid Flows as % as % Total as % Total Total Health REGION Health Expenditures Health Expenditures Expenditures Established Market Economies 61% 39% Middle Eastem Crescent 56% 43% 1% Fomnerly Socialist Economies 71% 29% of Europe India 20% 78% 2% China 58% 41% 1% Othier Asia and Islands 38% 61% 1% Sub-Saharan Africa 44% 47% 9% [Latin America and the Caribbean 49% 50% 1% Health Expenditures in Latin America 1990 Total Health Expenditure Health Expenditures as a percentage of GDP Development Assistance for Health (ofricial exchant.e rate dollars) Aid Flows Aid Flows Region and Economy Millions Per Capita Total Public Sector Private Sector Total Aid Flows Aid Flows as % total (1990US$) (1990US$) 1990 1 99( 199() (Mill. 1990US$) per capita Health Exp. 1 99( 1990 1990 Latin America &Caribbean 69,243 162 *6.2% 3.1% 3.10/%) 542 1.3 0.8% Antigna and Barbuda 21 262 5.0% 2.9% 2.1% 0.7 8.7 3.3% Argentina 10.0)7) 312 9.6% 5.9% 3.7% 10.9 0.3 0.1% Bahamas 15O 580 4.8% 2.6% 2.2% 0.0 0.0 0.0% Barbados 13( 400 6.2% 3.3% 2.9% 1.6 6.1 1.5% Beli7e 22 118 5.8% 3.4% 2.4% 2.4 12.8 10.8% IBolivia 247 34 5.5% 2.4% 3.1% 36.8 5.1 14.9% Brazil 3 3 445 222 6.4% 2.8% 3.6% 84.1 0.6 0.3% lBritishi Virein Islands 6 536 4.7% 2.7% 2.0% 0.0 0.0 0.0% ( ayntial Islanlds I i 701 4.0% 2.0% 2.0% 0.0 0.0 0.0% ( hilc 2,017 153 7.3% 3.4% 3.9% 9 7 0.7 0.5% (,ololnhia 2.'. 16 65 5.1% 3.0% 2.2% 26.0 0.8 1.2% (tosta Rica 523 186 9.2% 7.6% 1.6% 4.4 1.6 0.8% 4 D)onilic:j 14 195 8.2% 6.4% 1.8% 2.0 27.9 14.3% l)(1i1illican Relp 417 59 5.7% 2.1% 3.6% 10.9 1.5 2.6% ILcuaLclhr 40)2 39 3.7% 2.6% 1.1% 30 7 3.0 7.6% 1-1 Salvatdor ThO 58 5.9% 2.6% 3.3% 44 1 8.5 14.7% (irena.a I1' 133 5.9% 4.3% 1.6% 04 4.7 3.5% (itlat1iilalfa -,2 37 5.0% 2.1% 2.9% 31.5 3.4 9.2% (iuv;lia 31 39 9.6% 8.8% 0.9% 14.7 18.4 47.5% Ial il 173 27 7.0% 3.2% 3.8% 32.8 5.1 19.0% llu lLd.iras . 1 65 5.7% 2.9% 2.8% 20 3 4.0 6.1% Jamaica 37' 154 9.4% 3.6% 5.8% 19.0 7.8 5.1% Mexico t1(l46 155 5.5% 3.1% 2.4% 65.3 0.8 0.5% Nicaralua 1t) 31 7.9% 6.7% 1.2% 26.6 6.9 22.4% I'aniama 417 173 8.7% 5.5% 3.2% 14.7 6.1 3.5% ParaLIuav 213 49 3.9% 1.2% 2.7% 10.3 2.4 4.8% Peru l.28(0 59 3.1% 1.1% 2.0% 29.3 1.4 2.3% St. Kitts and Nevis i(I 239 6.8% 4.3% 2.4% 1.2 29.9 12.5% > St. Lucia 27 179 7.6% 5.6% 2.0% 0.4 2.4 1.3% t Si. Vincenit 12 110 6.1% 4.0% 2.1% 0.3 2.7 2.5% Suriname 5q 133 4.1% 1.2% 2.9% 1.7 3.8 2.9% X TIrindadandlobago 235 190 4.8% 2.9% 1.9% 1.4 1.1 0.6% X Uruguay 07) 219 8.3% 6.4% 1.9% 5.3 1.7 0.8% s Venezuela 2,0(9) 102 4.2% 2.0% 2.2% 2.5 0.1 0.1% N( ) tI-: oItd f)ce de I "es pr'nt I i Ie ( aIue HIaIth Exwntditu..s In LOtin Anie"k Iin 199 US Dollars Total Health Country Population 1990 CDP 1990 Pubblc Heath Fx%piditures 1990 Pnw Health E dines_ 1990 Aid Flows 1990 Total Health Expenditures Public Private Aid Flows Expenditure 1000) (1990 USS) Expeiditue (1990 USS) Expenditure (1990U) Expen,diure (1990USS) Expendbture (1990 US$) as % tobla as% total as % total PerCapita As An.oDP As %1G3P As %GDP As q6GDP (1990 US$) Latin America &Cairbbean 426.893 1,109,134,962,133 307% 34,103,174,850 3 12°o 34,597,889,249 005% 541,603040 624% 69.242.667,139 4925% 49 97% 078% 162 Aiirigt .and Ba1b1id. 79 418.703.692 2 69% 11.258,519 2.10% 8.792.778 0 16% 684,500 4 95% 20,735.796 54 30% 42 40% 3.30% 262 Argeirilna 32.293 105.437.652 631 5 85% 6 166 794 735 3 70% 3 901.193.147 0 01% 10.890.0W0 9 56% 10 078.877,883 61 19% 38 71% 0 11% 312 R.ahamas 258 3100000000 263% 81 394.905 2.20% 68,200000 0 483% 149594905 5441% 4559% 000% 580 Barbados 257 1.647.300.000 3 24% 53.454.061 2.90% 47.771.790 0 10% 1.570.350 6 24% 102.796.111 52 00% 46 47% 1 53% 400 Blhze 188 382,450 048 275% 10.525.769 241% 9202,979 063% 2,398.200 579% 22126,948 4757% 41 59% 10 84% 118 llolin-a 7 171 4.478,220,499 1 60% 71 724.438 310% 138 824 835 0.82% 36 770 440 5 52./ 247 319 714 29 00% 5613% 14.87% 34 P-razl lS0 348 521 607.637.091 2 76% 14 374 721 003 3 64% 18 986 517,990 0 02% 84 131 000 6 41% 33,445 369,993 42 98% 56 77% 0 25% 222 Bnuihli b poii lol,iiids 12 137.800 000 2 67% 3.679 260 2.00% 2 756.000 0 4 67% 6 435 260 57 17% 42 83% 0 00% 536 a-an- Istand 22 382.500,000 2 03% 7 774 167 2 00% 7 650,000 0 4 03% 15 424 167 50 40% 49 60% 0 00% 701 Clhd, 13 173 27.790 737,347 3 32% 923 011 422 3 90% 1 083 638.757 0 03% 9,681 500 7 26% 2,016,531.678 45 77% 53 75% 0 48% 153 Colombia 32 345 41,123.004 586 2 91% 1 197.137 600 2 17% 892 369,200 0 06% 26,049,300 5 14% 2115 556 100 56 59% 42 18% 1 23% 55 C-ia R-ca 2 807 5 702.391 276 7 50% 427 894 324 1 60% 91238 260 0 080/ 4 366 400 918% 523 498 985 81 74% 17 43% 0 83% 186 D.iUUUca 72 171 555 556 5 23% 8 974 339 1.80% 3 088 000 1 17% 2 005 500 8 20% 14 067.839 63 79% 21 95% 14 26% 195 Douiiomiinaii Rop 7 074 7 304 319 730 1 96% 142 980 955 3 60% 262 955.510 0 15% 10 890.400 5 71% 416 826 866 34 30% 63 09% 2 61% 59 Ecu^ador l10 284 1O 875.528,435 2 31% 251 437.697 1 10% 120 1 1S 206 0 28% 30 668 180 3 70m/ 402.221.083 62 51% 29 86% 7 62% 39 FltS.ador 5.213 5112813.382 1 74% 89174887 326% 166589319 086% 44096462 586% 299860668 2974% 5556% 1471% 58 Oire-ada 91 203.703.751 4 10% 8 346 506 1.63% 3 320.371 0 21% 423.250 5 94% 12 090.127 69 04% 27 46% 3 50% 133 Guaiammuoolo 9.197 6 787.540 280 1 64% 111 093,016 2.93% 198 874 930 0 46% 31 543 950 5 03% 341 511.897 32 53% 58 23% 9 24% 37 Ilmnitad 798 319,924,463 4 22% 13 493 361 0.85% 2.719,358 4 58% 14 659,000 9 65% 30 871.719 43 71% 8 81% 47.48% 39 Hamt. 6 472 2,467100.058 1 84% 45 304 209 3 83% 94 539.023 1 33% 32 752,650 7 00% 172 595 882 26 25% 54 77% 18 98% 27 Vl,nd..ra. 5.105 5830.000128 257% 149959335 2 76% 160908.004 035% 20265 200 568% 331,132,538 4529% 4859% 6 12% 65 Jaimiaica 2.420 3,967.956.668 3 10% 123,006 657 5 80% 230 141,487 0 48% 18 955,050 9 38% 372 103 193 33.06% 61 85% 5.09% 154 \tnSmnm l84 154 237.748 337.662 3 10% 7 370 198 468 2 36% 5 610 860 769 0 03% 65 288.000 5 49% 13 046 347,236 56 49% 43 01% 0 50% 155 Nicaragula 3 853 1 500,000000 4 90% 73 500.000 1.24% 18 600 000 1 77% 26 592 780 7 91% 118 692 780 61 92% 15 67% 22 40% 31 Piano 2.418 4 815.797 760 5 18% 249 361 876 3 18% 153.142.369 0 31% 14.715 300 8 66% 417 219 545 59 77% 36 71% 3 53% 173 Pmag-iar 4 314 5.477 268,680 0 98% 53 410 689 2.72% 148 981 708 0 19% 10 263 950 3 88% 212 656.348 25 12% 70 06% 4 83% 49 Porn 21 663 40 838 444 372 1 06% 433 641 140 2 00% 816 768 887 0 070/ 29 260 298 3 13% 1 279 670 325 3 89/% 63 83% 2 29% 59 Si K1l1o mid .\.. 40 141.555 591 3 48% 4 925 388 2.44% 3 453.956 0 85% 1 196 850 6 76% 9 576 194 51 43% 36 07% 12 50% 239 SI lucia 150 352 500.000 5 53% 19 484 673 2.00% 7 OS0 000 0 10% 353 000 7 63% 26 887.673 72 47% 26 22% 1 31% 179 Si \ 107 191.185 150 3 90% 7 449 915 2.10% 4 014 888 015% 291 250 6 15% 11 756 054 63 37% 34 15% 2 48% 110 Sii--u-on 447 1 438 994,449 1 09% 15,697 152 2.92% 42 018 638 0 12% 1.704.680 4 13% 59 420.470 26 42% 70 71% 2 87% 133 Tmidad aid Toobugo 1 236 4.890,823.499 283% 138 617.702 1.94% 94,881 976 003% 1412000 480% 234 911 678 59 01% 4039% 060% 190 lr-unla) 3.094 8217.979,465 632% 519 376 302 1 88% 154.498.014 006% 5,257,250 826% 679 131 566 7648% 2275% 077% 219 Veizsi"eld 19738 48273235.887 1 96% 944370380 220% 1 062011 190 001% 2466350 4 16% 2008847919 4701% 5287% 012% 102 NOTE Bot1tan1 donins prrduri,d cmld c it it (D 0~ X< C li-:l8h E,p-ndit.-e In Latin A-1r-c In 1990 HIm-ri.n.lll D.11-r CounU ~~~~~~Pqupdto _ 1- °° DP _ - I NJPbbc He,ahh Fxpend,t, 1 es _ 90ou P-z1, Hculff Ex,ptndIturs____ Ics. 0 AJd Fl.-. 1°°OTmJo H-alRh Expend it-, ToUd HealGh E,p-admnu Per Capda mm H0 (1°O [,,IM Intol,Hl ~as GDP (1-o nlwl'lS) as-.DP -(1°O IPI'lS) u°(iDP (I Q fl IrS) 5° GDP 1-lToOlm;lS) (1 990ULSS) L~UAb.lsn,- 426.8-3 1 990 933SOS5875 60 307 761,860 58,533,474,557 541 603 040 119 382 154,957 280 162 Anr,"a and Elrb,, 79 418 7038692 11,258,519 2 69% 8,792 778 2.10°/ 684 500 0 16Y% 20 735.796 4 95% 262 262 Argelilula ~~~~~32 293 141.247,140 000 8 261 205,534 S 65% S 226 144 180 3 70% 10 890 000 0 01% 13 498 239 714 9 56% 418 312 BhdlA-m 258 3 100 000 000 81 394 905 2 63% 66 200 000 2 20%/ 0 149 594 905 4 83% 560 580 8Rb d.A.s 257 2 6334 150 000 85 476,850 3 24% 76 390,350 2 90% 1 570 350 0 10% 163 437 SS0 6 24% 636 400 B:h7e 166 699 129 243 19 241,397 2 75% 16 823,300 2 41% 2 398 200 0 63% 38,462 897 S 79% 205 118 Robena 7 171 1 3 677 160 000 219 057 559 1 50% 423 992 580 3 10% 36 770 440 0 82% 679 820.579 S 52% 95 34 Br,zI 150 348 694.700 160 GOO 19 144 890 278 2 76% 25 287 085 824 3 64% 64 131 000 0 02% 44 516 107 102 6 41% 296 222 8n.l10,V.gul. 1,1-& 1 2 137 800 000 3 679,260 2 67% 2 756 000 2.00°b 0 6 435 260 467% 536 536 Cdysildn IsI,.sid, ~~~22 382 S00 000 7 774 167 2 03% 7 650 000 2 00% 0 15 424 167 4 03% 701 701 <~~~luJ. ~~~~~13 173 78 906 270,000 2 620 707,309 3 132% 3 077 344,530 3 90S% 9 681 S00 0 03% 5 707 733 339 7 26% 433 153 t.]1.walil,u ~~~~32 345 158 624 160000 4617 730 930 2 91% 3 442 144 706 2 17% 26 049 300 0 06% 8 065924,936 514% 250 65 6..l-RG 2 807 74 140 350.000 1061 059 337 7 50% 226.245 600 1 60% 4 366 400 0 08% 1 291,671 337 9 18% 460 186 D-."nlCd 72 310 125.844 16 223 167 S 23% S 582 265 1 80°/ 2 OOS S00 1 17% 23,810 933 8 20% 331 195 E>.Iulci - Kp 7 074 20 076 000 000 392 984 667 1% 96% 722 736 000 3 60% 10 APO 400 0 15% 1 126 611 067 S 71% 159 59 E-A,.ld. 10 264 38 430 810 000 8838 504 353 2 31% 4 24 450 609 1 1 0Ss 30 668 16QC 0 28% 1 343,623 142 3 70% 131 39 1.1 SuAd.l, S 213 9 558 640.000 166 716 557 1 74% 311 446 401 3 26% 44 096 462 0 86% 522 259 420 S 86% 100 58 Cr-'Iadu 91 389 377 967 15.954.274 4 10% 6 346 861 1.63%/ 423 250 0 21% 22 724 385 S 94% 250 133 , M^-iAa 9,197 25 843 570 000 422,986,830 1 64% 757,216 601 2 93% 31 543 950 0 46% 1 211,747 381 5 03% 132 37 ,uI!-IIn ~~~ ~~~~798 1 639 760 000 69 159,681 422% 13 937 960 0 85%/ 14 659000O 4 58% 97,756.641 9 65% 123 39 1 jd10 ~~~~~~~~6,472 6 447 870 DOO 118,404,460 1 84°6 247 081 722 3 83% 32 752 650 1 33% 3-98 238 832 7 00% 62 27 Hm,.-iua 5.105 8 066,660 000 207.490 728 2 57% 222.639.816 2 76% 20 265 200 0 35% 450,395,744 S 68% 88 65 J u-dCa 2,420 7 122 400 000 220 794 400 3 10% 413 099 200 S 80% 16 955 OS0 0 48% 652 848 650 9 38% 270 154 xle'ico 84 154 515 640 360 000 15 984 851 160 3 10% 12 169 1 12.496 2 38% 65 288 000 0 03% 28 219,251 656 5 49% 335 155 ulLArlpla ~~~~~~3 853 7 354 900 000 360 390 100 4 90% 91 200 760 1 24% 26 592 780 1 77% 478 183 640 7 91% 124 31 p|d- 2 418 9 768 720 000 SOS 824 OSS S 18% 310,645 296 3168% 1 4715 300 0 31% B31 184 651 8 66% 344 173 INr.lelluE ~~~~~~4,314 12 916 540 000 1 25 953 527 0 98% 351 329 886 2 72% lO 263 950 0 19% 487,547 365 31 88% 113 49 P-nl ~~~~~~~~21 6363 56 892 000 OO0 604 105 080 1 06% 1 137 840 000 2 00% 29 260 298 0 07% 1 771 205 378 3 13% 82 59 S, K.11n.,]d o.40 141 SSS 591 4 925 388 3 48% 3 453 956 2"4% 1 196 650 0 85% 9 576 194 6 77% 239 239 9 I ~~~ ~ ~~~~~~150 7 76 733 807 42 934 480 5 53% 15 534 676 2 00% 113S3000 0 10% 58 822 156 76-3% 392 179 %' .l-nil 107 498 3,33 422 19 418 568 3 90% 10 465 002 2 10% 2 5 0 15% 30 174 820 815 282 110 simulSlilic 447 1 747 080 000 ~~~~~~~~~~~19 057 877 1 09% 51 01 4 736 2 92°%170 8 0 12% 71 777 293 4 1 3% 16113 I mld.d.,rMdT.A,-, 1 236 10 619 490 000 30981 889 2 83% 206 01 8 106 1 94% 1 41 2000 0 03% 508 411 995 4 80% 411 190 I ni-uo 3 094 18 192 720 000 1 149 779 904 63 -2% 342 023 136 1 88% 5 25,7 250 (I 0f% 1 497 060 290 8 26% 484 219 % -m171el' 19,738 130 251 000 000 2 548 103 190 1 96% 2 865 522 000 2 20% 2 466 350 001% 54186,091 540 4 16% 274 102 '-t, RArdt-c uld-11,e pr-dMctd v.Jlu M,der I[LIAILA 5 L)P fig,-c ::I, s..be E- b$ lgip-e -v1 -bud .Iild -or .. .....e.l.les of tdbl- >>- - 44 - Appendix 12 Health Expendituires in Latin America Country Total Health Expenditure Per Capita (1990 Int'l $) (1990 US$) Latin America 280 162 &Caribbean Antigua and Barbuda 262 262 Argentina 418 312 Bahamas 580 580 Barbados 636 400 Belize 205 118 Bolivia 95 34 Brazil 296 222 British Virgin Islands 536 536 Cayman Islands 701 701 Chile 433 153 Colombia 250 65 Costa Rica 460 186 Domin1ica 331 195 Dominican Rep. 159 59 Ecu]ador 131 39 El Salvador 100 58 Grenada 250 133 Guatemala 132 37 Guyana 123 39 Haiti 62 27 Honduras 88 65 Jamaica 270 154 Mexico 335 155 Nicaragua 124 31 Panama 344 173 Paraguay 113 49 Peru 82 59 St. Kitts and Nevis 239 239 St. Lucia 392 179 St. Vincent 282 110 Suriname 161 133 Trindad and Tobago 411 190 Uruguay 484 219 Venezuela 274 102 - 45 - Appendix 13 TIME TRENDS - GOVERNMENT HEALTH EXPENDITURE AS % GDP 1980 1983 1987 1990 United States 3.91% 4.41% 4.68% 5.24% Latin America &Caribbean Anguilla - 2.85% - Antigua and Barbuda - - - 2.69% Argentina 1.51% 1.46% 1.43% 2.53% Bahamas 3.08% 3.21% 2.96% 2.62% Barbados 3.36% 3.12% 3.88% 3.24% Belize 2.26% 2.75% 2.39% 2.71% Bolivia 2.70% 2.20% - 1.60% Brazil 1.29% 1.68% 1.33% 1.57% British Virgin Islands - - 3.07% 2.67% Cayman Islands - - - 2.03% Chile 2.56% 2.92% 2.27% 2.21% Colombia 2.16% 2.19% 2.00% 2.91% Dominica - 5.64% 4.60% 4.95% Domninican Rep. 1.57% 1.48% 1.49% 1.64% Ecuador - 1.08% 1.43% 1.59% El Salvador 1.29% 1.29% 1.24% 1.24% Grenada - - - 4.10% Guatemala 1.90% 1.10% 1.10% 1.16% Guyana 3.34% 4.45% 2.81% 4.08% Haiti 1.35% 1.80% - 1.84% Honduras - 3.07% 2.39% 1.89% Jamaica 1.82% 2.78% - 2.78% Mexico 2.10% 2.10% 2.30% 3.10% Nicaragua 3.20% 4.90% 4.80% 4.90% Paraguay - 2.06% 1.12% 0.98% Peru 2.90% - 2.90% 1.06% St. Kitts and Nevis - 4.30% 3.48% St. Lucia - - - 5.53% St. Vincent 3.74% 3.83% 4.00% 3.90% Suriname - - 1.91% 1.09% Trindad and Tobago - 2.95% 4.47% 2.83% Uruguay 1.07% 0.85% 0.99% 1.28% Venezuela 2.28% 2.04% 1.55% 1.15% Government Health Expenditures, by expenditure type - Latin America & Caribbean Country _ _ Percentage of recurrent health expenditures: Capital Expenditures PHC Salaries Drugs and As percentage of total Other Supplies health expenditures Anguilla 1.0% 53.5% 17.0% 3.5% Antigua - 74.5% 11.5% 0.2% Bahamas 76.9% 16.1% 7.1% Barbados - 61.6% 23.1% 3.2% Belize 8.0% 58.0% 15.6% 13.2% Chile - 60.2% 27.1% 4.3% Costa Rica - 65.2% 18.3% 0.3% Dominica 31.1% 67.9% 13.4% 12.6% Dominican Republic 3.5% 35.6% 10.4% 23.2% Ecuador 12.8% - - 5.8% Guatemala 16.5% 41.0% 28.8% 44.6% Guyuna 42.1% 56.8% 19.7% 26.7% Honduras 8.6% 39.2% 37.8% 35.3% Jamaica 12.0% 40.6% 12.1% 3.2% St. Kitt's and Nevis 58.3% 11.5% 37.2% St. Vincent - 65.3% 10.4% 13.4% Trinidad and Tobago 7.1% 69.3% 6.8% 2.0% TurksandCaicos 11.9% 49.7% 10.1% 5 [Plot o~flif~e expectancy at birth and health expenditure per capit 80 ___ 75 [ C(M BAR ANT JAM PAN URU CHI STLTRI ARG 70 VEiV GRE MEX STK ~~~~COL CL >BELSUR m PSE%/ AR > ECU BRA 2 65 NIC HON GUY a ~~GUAEL x G PER a) W60 HAI BOL C 50 0 100 200 300 400 500 Total health expenditure per capita lPlot of life expectancy at birth and health expenditure as percentage of GDPI 80 75 BAR DOM COS ANT URU PAN JAM TRI CHI STL ARG ] - 70 .- VEN STV MEX GRE STK t ~~~~~~~~~~COL _ ~~~~~~SUR BEL PAR DMR ECU BRA 65 HON NIC GUY BOLS PER ~~~GUAEL 60 K HAI 55 BOL 50 - - _ l _ _._ _- 2% 4% 6% 8% 10% Total health expenditure as percentage of GDP Life expectancies and health expenditures in selected countries: deviations from estimates based on GDP and schooling Deviation from predicted life expectancy at birth (years) 15 12 Costa Rica 9 ~~~~~~~~~~~~~~~~~~Argentina ° El Salvador A Paraguay Colombia 0 OChile 0Panama 6 Venezuela C) CX) OHonduras E uador 0 Guatemala 3 B razil 0Haiti 0 -3 0 Peru -6 0 Bolivia -9 D I_I_I_> -2 0 2 4 6 Deviation from predicted percentage of GDP spent on health Health Expenditures in Latin America - Comparison of WDR 1993 and Present Study Health Expenditures as a percentage of GDP Region and Economy Total Public Sector Private Sector Present Study WDR 1993 Present Study WDR 1993 Present Study WDR 1993 Latin America &Caribbean 6.2% 4.0% 3.1% 2.4% 3.1% 1.6% Argentina 9.6% 4.2% 5.9% 2.5% 3.7% 1.7% Barbados 6.2% 5.0% 3.3% 3.3% 2.9% 1.7% Belize 5.8% 5.8% 3.4% 3.4% 2.4% 2.4% Bolivia 5.5% 4.0% 2.4% 2.4% 3.1% 1.6% Brazil 6.4% 4.2% 2.8% 2.8% 3.6% 1.4% , Cayman Islands 4.0% 3.8% 2.0% 2.0% 2.0% 1.8% 0 Chile 7.3% 4.7% 3.4% 3.4% 3.9% 1.4% Colombia 5.1% 4.0% 3.0% 1.8% 2.2% 2.2% Dominican Rep. 5.7% 3.7% 2.1% 2.1% 3.6% 1.6% Ecuador 3.7% 4.1% 2.6% 2.6% 1.1% 1.6% El Salvador 5.9% 5.9% 2.6% 2.6% 3.3% 3.3% Guatemala 5.0% 3.7% 2.1% 2.1% 2.9% 1.6% Haiti 7.0% 7.0% 3.2% 3.2% 3.8% 3.8% Honduras 5.7% 4.5% 2.9% 2.9% 2.8% 1.6% Jamaica 9.4% 5.1% 3.6% 3.4% 5.8% 1.7% Mexico 5.5% 3.2% 3.1% 1.6% 2.4% 1.6% Nicaragua 7.9% 8.6% 6.7% 6.7% 1.2% 1.9% Panama 8.7% 7.1% 5.5% 5.5% 3.2% 1.6% Paraguay 3.9% 2.8% 1.2% 1.2% 2.7% 1.6% 3 Peru 3.1% 3.2% 1.1% 1.9% 2.0% 1.3% Uruguay 8.3% 4.6% 6.4% 2.5% 1.9% 2.1% Venezuela 4.2% 3.6% 2.0% 2.0% 2.2% 1.6% t NOTE: Boldface denotes predictcd value -51 - Appendix 19 Explanations for Changes in Expenditure Estimates from WDR 1993 Country and Suib-Sector Reasoii for Change Government Argentina Addition of local expenditures (previously unavailable data) Colombia Addition of local expenditures (previously unavailable data) Jamaica Availability of more recent and comprehensive WB study Mexico Availability of more recent and comprehensive WB study Uruguay Availability of more recent and comprehensive WB study Parastatal Jamaica Availability of more recent social security estimate from WB Mexico Availability of more recent social security estimate from nWB Peru Availability of more recent national social security estimate Private Argentina Availability of more recent and comprehensive WB study Barbados New prediction Bolivia Availability of more recent, national household survey Brazil Availability of more recent, national household survey Cayman Islands Availability of more recent, national household survey Chile Availability of more recent and comprehensive WB study Dominican Republic Availability of inore recent, national household survey Ecuador Availability of more recent and comprehensive W'B study Guatemala New prediction Honduras Availability of Inore recent, national household survey Jamaica Availability of more recent and comprehensive WB study Mexico Availability of inore recent, national household survey Nicaragua New prediction Panama Availability of more recent, national household survey Paraguay New prediction Peru Availability of more recent, national household survey Uruguay Availability of more recent and comprehensive V1B study Venezuela New prediction Distributors of World Bank Publications ARGENTINA The Middle East Obmerver KOREA,&REPUBLICOF SOUTH AFRICA, BOTSWANA Cario HirLch, SRL 41, Sherf Stneet Psn Xmsa Book Corpmation For anli 11s,, Galria Guemes Cair PD. Box 101, Kwangwhanwn 0x6nDd UnivesIty Press Florida 165, 4th Floor-Ofc 453/465 Seoul Southen Afria 1333 Bueno Aires FINLAND PD. Box 1141 Akateeminen Ktrjaauppa KoreanStockBookCentre Cape Town 8010 Oficna del Libro Internaconal PD. Box 128 PD. Box 34 Alberii40 SF-001O HeisinkilO Yeoeido ForsacFritiinorrrs 1082 Buenos Aires Seoul Intrnatonal Subscriptlon ServIce FRANCE PD. Bo 41095 AUSTRALIA, PAPUA NEW GUINEA, World Bank Publications MALAYSIA Craighall FilI, SOLOMON ISLANDS, 66, avenue d'ltn Unvestyof Malaya Cooperative johnnesburg 2024 VANUATU, AND WESTERN SAMOA 75116 Paris Bookshop, Lirrted DA. Information Services PD. Box 1127, jalan Pantal Barn SPAIN 648 Whitehorse Road GERMANY 5970D Kua Laumpur Mundi-Prenna Libros, SA. Mitcham3132 UNO-Veriag Castello 37 Victoria PoppeLadorfer Allee SS MEXICO 2800 Madrid 53115 Bonn INIOTEC AUSTRIA Apartado Poal 22460 Libreria Intemacional AEDOS Gerold and Co. GHANA 14060 Tlipan Mexico DF Conetl deCent, 391 Graben 31 Greenwich Mag. and Books 08009 Barcelona A-1011 Wien Rivera Beach Hotle NETHERLANDS PO Box 01198 De Lindeboom/InOr-PubUlkaties SRI LANKA AND TH E MA LDIVES BANGLADESH Osu-Accra PD. Box2il2 LkeHou seBookshop Micro Industries Development 7480AEIHaklergen PD. Box 244 Assistance Society OM DAS) GREECE i0, SirChittampalam A. H-ouxse, Road 16 Papasoiriou SA. NEWZEALAND GardinerMawatha Dhanmnndi R/Area 35, Stoumanra Sr. EBS0 NZ Ld. Codonbo2 Dhaka 1209 10682Athens PrivateMiLlBag99914 New Market SWEDEN BELGIUM HONG KONG, MACAO Auckland Frizes Fackbokaforetaget Jean De Lannoy Asia 2000 Ltd. Regeringpgatan 12 Box 16356 Av. du Roi202 4648WyndharnSreet NIGERLA S-10647SockhoLm Ii0 Brussels Winning Centre University Pre Limited 7th Floor Three Crowns Building Jericho Wennergren-Williarns AB BRAZIL Central Hong Kong Prtvate Mall Bg S09S P. . Box 1305 Publicacon Ternicas Intemationals Ltda. Sedan S-171 25 Solna Rua Pel.otoGomrde. 209 HUNGARY 014119 Sao Paulo, SP Foundation for Market Economy NORWAY SWITZERLAND Donbovari Ut 17-19 Narvesen Information Center Librairle Payct CANADA H-1117 Budapest Book Department Cae potale 3212 te Ditfuseur PD. Box 6125 EAterstad CH 1002 Lausanne ISIABoul de Mortagne INDIA N-06020 slo6 Boucherville, Qu&bec Allied Publishers Private Ltd. Van Dierman Editions Techniques. ADECO J4B SE6 751 Mount Road PAKISTAN P.O. Box 465 Madras - 600) 12 Mirzs Book Agency Cli 1211 Geneva 16 Renout Publishing Co. 65, Shahrah-e-QuaId-e-Azmrn 1294 Algoma Rod INDONESIA PD. Box No. 729 TANZANIA Ottawa, Ontario Pt. Indira Limited Lahore 540D0 Oxford University Pess K1B3W8 Jalan nBrobudur 20 MaklabaStreet PO Box lI'I PERU P.O. Box5299 CHINA Jakarta10320 Editorial DsrroIloSA DaresSalaam China Financial & Economic Apartado3824 Pubifshing House IRAN LUnm I THAILAND 8, Da FoSi Dongjie Kowkab PubLishers Central Departnent Store Beijing P.O. Box 19575-511 PHItPPINES 306 Silom Road Tehran International BookCenter Bangkok COLOMBIA Suite 1703,Cttylhnd 10 Infoenlace Ltda. IRELAND CondominiumcTower I rRINIDAD&TOBAGO Apartado Aereo 34270 Government Supplies Agency AyaLa Avenue, IHV. deLa Systemrtcs Studies Unit Bogota D.E 4-5 Harcourt Road Costa ExtenJion 89 Watts Street Dublin2 Maka tlMetroManila Curepe COTE D'IVOIRE Trinidad, West Indies Centre d'Edition et de Diffusion ISRAEL POLAND Atricainer (CEDA) Yczxrwt Literature Ltd. International Publahing Service UGANDA 04 B.P. 541 P.O. Box 56055 Ul. PLekna 31/37 Gustro Ltd. Abidjan 04 Plateau Tel Aviv 61560 00-677 Waraws lIt Floor, Room 4 GeogladisCharnter P.O. Box 9997 CYPRUS ROY. International PORTUGAL Pot (69) Kampala Center of Applied Resesrch PDOB. 13056 Lvsri Portugal CyprusCollege Tel Aviv 61130 Rua DoCsrrno7D-74 UNITED KINGDOM 6, Diogenes Street Engomi 1200 Libn MicroinfoLtd. P.O. Box2006 ITALY PD. Box 3 Nicosia Lrsa Corrrmtwlonaria Sansoni SPA SAUDI ARABA QATAR Alton Flampshire GU34 2PG Vi UDucs Di Calabria1/1 Jrir Book Store England CZECH REPUBLIC Caseils Postale 552 PD. Box 3196 NationalInformatlonCenter 50125Firenze RiyadhlI471 ZAMBIA P.O. Box 668 UniversIty of ZambLa Hookshop CS-I 1357 Prague I JAMAICA SLOVAK REPUBUC Great East Road Campus Ian Randle Publishers Ltd. Slovart G.TG Lid. PD. Box 32379 DENMARK 206 Old Hope Road K"upinbka4 Lu[sita SamfundsLitteratur KLngston 6 PD. Box 152 Rosenoerns Alit S1 8S299 Bratislava5 ZIMBAiWE DK-1970 Frederiksberg C JAPAN Longnan Zlntabwe Pavt.) Ltd. Eastern Book Service SINGAPORE, TAIWAN, Tourie Road, Ardbennie DOMINICAN REPUBLIC Hongo 3-Chora, Buhkyo-ku 113 MYANMARBRUNEI PO. Bx ST 125 Editora Taller, C. porA. Tokyo GowerAa UPadficPtekltd. Southerlon Reataurad6n e Iabella Cat6lica 309 Golden Wheel Building H-are ApartadodeCorreus 2190Z-1 KENYA 41, Kalang Pudding, 030 Santo Domingo Afdra Book Service CEA.0 Ltd. Singapore 1334 Quarn Hose, Mfangano Street EGYPT, ARAB REPUBLIC OF P.O. Box 45245 Al Ahram Nairobi Al Calaa Street Cairn RECENT WORLD BANK TECHNICAL PAPERS (continued) No. 228 Webster and Charap, The Emergence of Private Sector Manufacturing in St. Petersburg: A Survey of Firms No. 229 Webster, The Emergence of Private Sector Manufacturing in Hungary: A Survey of Firms No. 230 Webster and Swanson, The Emergence of Private Sector Manufacturing in the Former Czech and Slovak Federal Republic: A Survey of Firms No. 231 Eisa, Barghouti, Gillham, and Al-Saffy, Cotton Production Prospectsfor the Decade to 2005: A Global Overview No. 232 Creightney, Transport and Economic Performance: A Survey of Developing Countries No. 233 Frederiksen, Berkoff, and Barber, Principles and Practicesfor Dealing with Water Resources Issues No. 234 Archondo-Callao and Faiz, Estimating Vehicle Operating Costs No. 235 Claessens, Risk Management in Developing Countries No. 236 Bennett and Goldberg, Providing Enterprise Development and Financial Services to Women: A Decade of Bank Experience in Asia No. 237 Webster, The Emergence of Private Sector Manufacturing in Poland: A Survey of Firms No. 238 Heath, Land Rights in Cote d'lvoire: Survey and Prospectsfor Project Intervention No. 239 Kirmani and Rangeley, International Inland Waters: Conceptsfor a More Active World Bank Role No. 240 Ahmed, Renewable Energy Technologies: A Review of the Status and Costs of Selected Technologies No. 241 Webster, Newly Privatized Russian Enterprises No. 242 Barnes, Openshaw, Smith, and van der Plas, What Makes People Cook with Improved Biomass Stoves?: A Comparative International Review of Stove Programs No. 243 Menke and Fazzari, Improving Electric Power Utility Efficiency: Issues and Recommendations No. 244 Liebenthal, Mathur, and Wade, Solar Energy: Lessonsfrom the Pacific Island Experience No. 245 Klein, External Debt Management: An Introduction No. 246 Plusquellec, Burt, and Wolter, Modern Water Control in Irrigation: Concepts, Issues, and Applications No. 247 Ameur, Agricultural Extension: A Step beyond the Next Step No. 248 Malhotra, Koenig, and Sinsukprasert, A Survey of Asia's Energy Prices No. 249 Le Moigne, Easter, Ochs, and Giltner, Water Policy and Water Markets: Selected Papers and Proceedingsfrom the World Bank's Annual Irrigation and Drainage Seminar, Annapolis, Maryland, December 8-10, 1992 No. 250 Rangeley, Thiam, Andersen, and Lyle, International River Basin Organizations in Sub-Saharan Africa No. 251 Sharma, Rietbergen, Heimo, and Patel, A Strategyfor the Forest Sector in Sub-Saharan Africa No. 252 The World Bank/FAO/UNIDO/Industry Fertilizer Working Group, World and Regional Supply and Demanid Balancesfor Nitrogen, Phosphate, and Potash, 1992/93-1998/99 No. 253 Jensen and Malter, A Global Review of Protected Agriculture No. 254 Frischtak, Governance Capacity and Economic Reform in Developing Countries No. 255 Mohan, editor, Bibliography of Publications: Technical Department, Africa Region, July 1987 to April 1994 No. 256 Campbell, Design and Operation of Smallholder Irrigation in South Asia No. 257 Malhotra, Sinsukprasert, and Eglington, The Performance of Asia's Energy Sector No. 258 Willy De Geyndt, Managing the Quality of Health Care in Developing Countries No. 259 Chaudry, Reid, and Malik, editors, Civil Service Reform in Latin America and the Caribbean: Proceedings of a Conference No. 260 Humphrey, Payment Systems: Principles, Practice, and Improvements No. 261 Lynch, Provisionfor Children with Special Educational Needs in the Asia Region No. 262 Lee and Bobadilla, Health Statisticsfor the Americas No. 263 LeMoigne, Subramanian, Xie, and Giltner, editors, A Guide to the Formulation of Water Resources Strategy No. 264 Miller and Jones, Organic and Compost-Based Growing Media for Tree Seedling Nurseries The World Bank Headquarters European Office Tokyo Office 1818 H Street, N.W. 66, avenue d'1ena Kokusai Building a Washington, D.C. 20433, U.S.A. 75116 Paris, France 1-1, Marunouchi 3-chome Chiyoda-ku, Tokyo 100, Japan Telephone: (202) 477-1234 Telephone: (1) 40.69.30.00 Facsimile: (202) 477-6391 Facsimile: (1) 40.69.30.66 Telephone: (3) 3214-5001 Telex: MCI 64145 WORLDBANK Telex: 640651 Facsimile: (3) 3214-3657 MCI 248423 WORLDBANK Telex: 26838 Cable Address: INTBAFRAD WASHINGTONDC Cover design by Walton Rosenquist ISBN 0-8213-3142-X 13142 POP 100 0-8213-3142-X HEALTH EXPENDITURES AMER 4 00000016412 $6 .95