WORLD BANK DISCUSSION PAPER NO. 365 Work in progress WD P365 for public discussion In IIovatiOnS ill I7lICltlh Care "i iILicI i g P/'o,(f , /i,/Is (/,X/ It //,/ IP'k (Clfi'T/'I ''' ( U ¼! I. /(L //! /(y Recent World Bank Discussion Papers No. 296 Reforming the Energy Sector in Transition Economies: Selected Experience and Lessons. Dale Gray No. 297 Assessing Sector Instititions: Lessons of Experiencefrom Zambia's Education Sector. Rogerio F. Pinto and Angelous J. Mrope No. 298 Uganda's AIDS Crisis: Its Implications for Development. Jill Armstrong No. 299 Towards a Payments System Lawfor Developing and Transition Economies. Raj Bhala No. 300 Africa Can Compete! Export Opportunities and Challengesfor Garments and Home Prodiucts in the Eluropean Market. Tyler Biggs, Margaret Miller, Caroline Otto, and Gerald Tyler No. 301 Review and Outtlookfor the World Oil Market. Shane S. Streifel No. 302 The Broad Sector Approach to Investment Lending: Sector Investment Programs. Peter Harrold and Associates No. 303 Instituitional Adjustment and Adjuisting to Institutions. Robert Klitgaard No. 304 Putting Instituitional Economics to Work: From Participation to Governance. Robert Picciotto No. 305 Pakistan's Putblic Agricutltutral Enterprises: Inefficiencies, Market Distortions, and Proposals for Reform. Rashid Faruqee, Ridwan Ali, and Yusuf Choudhry No. 306 Grameen Bank: Performance and Stability. Shahidur R. Khandker, Baqui Khalily, and Zahed Khan No. 307 The Urtguay Rouind and the Developing Economies. Edited by Will Martin and L. Alan Winters No. 308 Bank Governance Contracts: Establishing Goals and Accountability in Bank Restructuring. Richard P. Roulier No. 309 Public and Private Secondary Education in Developing Coluntries: A Comparative Stuldy. Emmanuel Jimenez and Marlaine E. Lockheed with contributions by Donald Cox, Eduardo Luna, Vicente Paqueo, M. L. de Vera, and Nongnuch Wattanawaha No. 310 Practical Lessonsfor Africafrom East Asia in Indutstrial and Trade Policies. Peter Harrold, Malathi Jayawickrama, and Deepak Bhattasali No. 311 The Impact of the Urugutay Round on Africa. Peter Harrold No. 312 Procurement and Disbursement Manuialfor Projects with Comnmuznity Participation. Gita Gopal No. 313 Harnessing Informationfor Development: A Proposalfor a World Bank Group Strategy. Eduardo Talero and Philip Gaudette No. 314 Colombia's Pension Reform: Fiscal and Macroeconomic Effects. Klaus Schmidt-Hebbel No. 315 Land Quiality Indicators. Christian Pieri, Julian Dumanski, Ann Hamblin, and Anthony Young No. 316 Suzstainability of a Government Targeted Credit Program: Evidencefrom Bangladesh. Shahidur R. Khandker, Zahed Khan, and Baqui Khalily No. 317 Selected Social Safety Net Programs in the Philippines: Targeting, Cost-Effectiveness, and Optionsfor Reform. Kalanidhi Subbarao, Akhter U. Ahmed, and Tesfaye Teklu No. 318 Private Sector Development Duiring Transition: The Visegrad Couintries. Michael S. Borish and Michel Noel No. 319 Eduication Achievements and School Efficiency in Rural Bangladesh. Shahidur R. Khandker No. 320 Houisehold and Intraholusehold Impacts of the Grameen Bank and Similar Targeted Credit Programs in Bangladesh. Mark M. Pitt and Shahidur R. Khandker No. 321 Clearance and Settlement Systemsfor Secufrities: Critical Design Choices in Emerging Market Economies. Jeff Stehm No. 322 Selecting Development Projectsfor the World Bank. Jean Baneth No. 323 Evalutating Public Spending: A Frameworkfor Putblic Expenditutre Reviews. Sanjay Pradhan No. 324 The Bangladesh Rural Advancement Committee's Credit Programs: Performance and Suistainability. Shahidur R. Khand- ker and Baqui Khalily No. 325 Instituitional and Entrepreneutrial Leadership in the Brazilian Science and Technology Sector:Setting a New Agenda . Edited by Lauritz Holm-Nielsen, Michael Crawford, and Alcyone Saliba No. 326 The East Asian Miracle and Information Technology: Strategic Management of Technological Learning. Nagy Hanna, Sandor Boyson, and Shakuntala Gunaratne No. 327 Agricultzural Reform in Ruissia: A Viewfrom the Farm Level. Karen Brooks, Elmira Krylatykh, Zvi Lerman, Aleksandr Petrikov, and Vasilii Uzun No. 328 Insuiring Sovereign Debt Against Default. David F. Babbel No. 329 Managing Transboutndanj Stocks of Small Pelagic Fish: Problems and Options. Max Aguiero and Exequiel Gonzalez No. 330 China: Issutes and Options in Greenhoutse Gas Emissions Control. Edited by Todd M. Johnson, Junfeng Li, Zhongxiao Jiang, and Robert P. Taylor (Continuied on the inside back cover) WORLD BANK DISCUSSION PAPER NO. 365 Innovations in Health Care Financing Proceedings of a World Bank Conference, March 10-11, 1997 Edited by George J. Schieber Tkt World Bank Washington, D.C. Copyright © 1997 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 July 1997 Discussion Papers present results of country analysis or research that are circulated to encourage discussion and comment within the development community. To present these results with the least possible delay, the typescript of this paper has not been prepared in accordance with the procedures appropriate to formal printed texts, and the World Bank accepts no responsibility 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 whatsoever 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 pernission 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 portions for classroom use is granted through the Copyright Clearance Center, Inc., Suite 910,222 Rosewood Drive, Danvers, Massachusetts 01923, U.S.A. ISSN: 0259-210X George J. Schieber is health sector leader in the World Bank's Middle East and North Africa Region. Library of Congress Cataloging-in-Publication Data Innovations in health care financing: proceedings of a World Bank conference, March 10-11, 1997 / edited by George Schieber. p. cm. - (World Bank discussion paper; ISSN 0259-210X; 365) ISBN 0-8213-3964-8 1. Medical care-Developing countries-Finance-Congresses I. Schieber, George. II. World Bank Conference on Innovations in Health Care Financing (1997: Washington, D.C.) m. Series: World Bank discussion papers; 365. RA410.55.D48I55 1997 338.4'33621'091724-dc2l 97-17183 CIP Contents Foreword v Acknowledgments vi Abstract vii A Curmudgeon's Guide to Financing Health Care in Developing Countries 1 George Schieber and Akiko Maeda Government Financing of Health Care 39 Bengt J5nsson and Philip Musgrove From Beveridge to Bismarck: Health Finance in the Russian Federation 65 Igor Sheiman Private Insurance: Principles and Practice 77 Deborah . Chollet and Maureen Lewis Private Health Insurance in Egypt 115 Nadwa Rafeh Strategies for Pricing Publicly Provided Health Services 127 Paul J. Gertler and Jeffrey S. Hammer Cost Recovery Strategies in Sub-Saharan Africa 155 Joseph Wang'ombe Rural Risk-Sharing Strategies 163 Andrew Creese and Sara Bennett Rural Health Care Financing in Thailand 183 Sirilaksana Khoman 'iii Market-Based Reform of U.S. Health Care Financing and Delivery: Managed Care and Managed Competition 195 Alain C. Enthoven Managed Care and Managed Competition in Latin America and the Caribbean 215 Andre Cezar Medici, Juan Luis Londonio, Oswaldo Coelho, and Helen Saxenian Medical Savings Accounts for Developing Countries 233 Len M. Nichols, Nicholas Prescott, and Kai Hong Phua Medical Savings Accounts and Health Care Financing in Singapore 247 Kai Hong Phua INNOVATIONS IN HEALTH CARE FINANCING iv Foreword F inancing health care is a critical concern for rich than 70 countries to consider the broad range of issues relat- and poor countries alike, as health care systems ing to the financing of health care systems in low- and mid- account for 9 percent of global production. dle-income countries. Both the conceptual and operational Developing countries face particularly serious challenges as policy contexts for introducing changes in health care financ- they attempt to improve the well-being of their populations, ing were explored. Traditional public and private financing achieve economic development objectives, and integrate approaches were addressed together with more recent meth- themselves with the global economy. Health care financing ods such as medical savings accounts and managed compe- is a particular concern for these countries, which account tition. Issues of particular relevance to low-income countries, for 84 percent of the world's population and 93 percent of including user charges and infonnal rural risk-pooling schemes, its disease burden but only 18 percent of its income and 11 were also discussed. Case studies from each developing region percent of its health expenditures. Imbalances between were used to highlight the various approaches. spending and the disease burden will be exacerbated as a It is hoped that this volume will help countries develop result of the changing composition of illness toward non- effective health care financing policies, and stimulate fur- communicable diseases and injuries, which by 2020 will ther policy dialogue and research on this critically impor- account for almost 80 percent of these countries' disease tant social and economic issue. burdens, compared with just over 50 percent now. These diseases are more expensive to treat and harder to prevent David de Ferranti than the infectious diseases that were previously the lead- Director and Head ing causes of illness and death. Human Development Network This volume contains the thirteen papers presented at the World Bank's Conference on Innovations in Health Care Richard G. A. Feachem Financing, held inWashington, D.C., on March 10-1 1,1997. Director The conference brought together 400 participants from more Health, Nutrition, and Population v Abstract T his volume contains thirteen papers prepared for the sizing underlying characteristics of insurance markets as World Bank Conference on Innovations in Health well as the need for regulation. Nadwa Rafeh presents an Care Financing, held in Washington, D.C., on March example of the evolution of private health insurance in Egypt. 10-11, 1997. Together these papers provide conceptual and Two revenue-raising methods of particular importance practical policy prescriptions for financing health care systems to developing countries are user charges and informal rural in developing countries. Health care financingis a serious con- risk-pooling arrangements. Paul J. Gertler and Jeffrey S. cem forthese countries: they contain 84 percent of the world's Hammer analyze user charges and review their effects on population and 93 percent of its disease burden but account revenue raising, equity, and efficiency from conceptual, forjust 18 percent of its income and 11 percent of global health empirical, and country perspectives. Joseph Wang'ombe spending. Moreover, developing countries have the capacity summarizes Sub-Saharan Africa's experience with user to raise less than 60 percent of the revenues raised by indus- charges. Andrew Creese and Sara Bennett discuss the con- trial countries. With 5 percent of their gross domestic prod- ceptual underpinnings of and real-world experiences with ucts devoted to health spending, split almost evenly between informal rural risk-sharing arrangements. Sirilaksana public and private sources of spending, raising and managing Khoman presents an example of one of these arrangements, health sector revenues is a major challenge. The papers in the health card used in Thailand. this volume provide an overview of the health care financing How revenues are managed has important implications for issues that are most relevant for developing countries and pre- the efficiency with which such revenues are used. By provid- sents case studies illustrating their experiences with various ing individuals with strong incentives to manage health care revenue generation and management techniques. funds, managed competition and medical savings accounts The overviewpaperbyGeorge SchieberandAkikoMaeda are two recent innovations that have important demand-side describes demographic, epidemiological, service delivery, and effects. Alain C. Enthoven describes the basis for managed health expenditure patterns for the world's industrial and competition, discusses how managed care is a logical con- developing countries, analyzes from both conceptual per- comitant to managed competition, and analyzes the applica- spectives and real-world experiences the range of public and bility of these methods to developing countries. Andre Cezar private revenue-raising instruments, and provides general Medici, Juan Luis Londoflo, Oswaldo Coelho, and Helen perspectives for health financing reforms in each of the world's Saxenian describe the experiences with managed competition six developing regions. Bengt Jonsson and Philip Musgrove and managed care in Latin America and the Caribbean. Len analyze the issues and experiences with government financ- Nichols, Nicholas Prescott, and Kai Hong Phua discuss the ing of health services in both industrial and developing coun- conceptual and operational bases for medical savings accounts, tries. These issues are highlighted in Igor Sheiman's case study describe real-world experiences with these accounts, and ana- of recent health insurance reform in the Russian Federation. lyze the necessary conditions for implementing these accounts DeborahJ. Chollet and Maureen Lewis discuss private health in developing countries. Phua also provides an in-depth analy- insurance as a mechanism for financing health services, empha- sis of Singapore's experience with medical savings accounts. vi Acknowledgments T he papers presented in this volume were commis- the papers and helped ensure their relevance for develop- sioned for the World Bank Conference on ing countries. The efforts of Xavier Coll, Edward Elmendorf, Innovations in Health Care Financing, held in Theresa Ho, Eva Jarawan, Maureen Lewis, Sandy Washington, D.C., on March 10-11, 1997. Special thanks Lieberman, Jo Martins, and Nicholas Prescott are greatly are due to the Bank's conference cosponsors: the appreciated. Commonwealth Fund, the U.S. Agency for Health Care Special thanks are due to the World Bank staff mem- Policy and Research, the U.S. Agency for International bers who were peer reviewers for the papers: Shanta Development's Bureau for Europe and NewlyIndependent Devarajan, David Dunlop, Charles Griffin, John States, the U.S. Health Care Financing Administration, the Langenbrunner, Chris Lovelace, William McGreevey, World Bank's Economic Development Institute, and the Alexander Preker, and Jacques van der Gaag. The Alpha World Health Organization. Center helped the Bank organize the conference, and their The conference was initiated by Armeane Choksi, for- logistic and substantive support in meeting deadlines and mer vice president for Human Resource Development and ensuring the quality of the papers is greatly appreciated, Operations Policy; David de Ferranti, director and head, with special thanks to Amy Bernstein and Deborah Chollet. Human Development Network; and Richard Feachem, Jillian Cohen of the World Bank's Human Development director, Health, Nutrition, and Population. Their support Department coordinated the peer review process. Finally, was instrumental in ensuring high-quality, relevant papers. the papers were edited by Paul Holtz and laid out by Glenn Many Bank staff contributed to the thematic content of McGrath, both with American Writing Corporation. vii A Curmudgeon's Guide to Financing Health Care in Developing Countries George Schieber and Akiko Maeda U nderstanding how countries finance their health priate financing arrangements for developing countries- care systems is of critical importance for indus- recognizing their underlying economic and institutional trial and developing countries alike. The methods structures. used to mobilize the resources that support basic public Conditions in developing countries often preclude use health programs, provide access to basic health services, of the financing and management arrangements used in and configure health service delivery systems affect peo- industrial countries. Thus this paper also draws attention ple's health status-as well as every aspect of a country's to the costs associated with generating public revenues- social, economic, and political well-being (box 1). Moreover, costs that usually far exceed the revenue that is raised. These health care systems account for 9 percent of global pro- costs, along with their distribution across income groups, duction and a significant portion of global employment. are often overlooked in discussions of health care financ- Health care systems also affect imports and, in some coun- ing. Yet inefficiencies and inequities in generating revenues tries, exports. often compound inefficiencies and inequities in allocating Decisions on the methods used to raise revenue for health expenditures. Such issues are of crucial importance for devel- care systems have important consequences for equity across oping countries, where low,income levels limit the scope for income groups, the amounts of revenue raised, and the raising revenue. losses in consumer welfare and production generated by Given its emphasis on financing, this paper focuses on different revenue-raising techniques. Thus public and pri- economic and administrative issues. But noneconomic con- vate programs to finance and deliver health care affect siderations, particularly each country's social, political, government budgets, macroeconomic stability, employment, and cultural environment, are of critical importance. An imports, exports, and international competitiveness. attempt is made to capture these elements in the policy This paper focuses on how governments raise revenues and technical discussions below, particularly in the discus- to finance their health care systems as well as on recent inno- sions on taxation and on the public-private mix of financ- vations for public and private management of these rev- ing, as well as in the summaries of the health care systems enues (including managed competition, medical savings in each region. Still, since these discussions are fairly gen- accounts, private insurance, and community risk-pooling eral, they may underemphasize the importance of noneco- schemes). It also discusses the rationales for public and nomic factors in health policymaking. private finance, assesses the criteria that should be used to Although this paper focuses on sources of financing, such evaluate different revenue sources, and identifies appro- discussions cannot take place without considering the demo- George Schieber is health sector leader in the Middle East and North Africa Region at the World Bank. Akiko Maeda is health econ- omist at the World Bank. The data presented in this paper are based on a revised version of the World Bank Health Data Base; as such, some of the statistics differ from those in the conference version of this paper. The authors are grateful to Deborah Chollet, Shanta Devarajan, Nicole Klingen, Maureen Lewis, Chris Lovelace, Bill McGreevey, Len Nichols, Alex Preker, Nicholas Prescott, Gail Richardson, and Jacques van der Gaag for helpful comments. 1 tries given their institutional characteristics. Section six clis- BOX I cusses options for private financing, with an emphasis on Goals of a health care system private health insurance markets and their implications for * Improving a population's health status and promoting social government regulation. Finally, section seven provides con- well-being cluding observations on health care reform debates in dif- * Ensuring equity and access to care ferent parts of the developing world. * Ensuring microeconomic and macroeconomic efficiency in the use of resources * Enhancing clinical effectiveness Global Overview * Improving quality of care and consumer satisfaction * Assuring the system's long-run financial sustainability Although this paper's focus is on the sources of health care financing, this section summarizes current trends in such financing, including its relation to service delivery outputs graphic, epidemiological, and service delivery characteristics and health outcomes by region. A comparison of economic of different countries. The next section summarizes demo- indicators, health outcomes, and health services across graphic and epidemiological conditions in developing regions. regions and income groups is shown in table 1. The third section analyzes regional health spending in terms In 1994 global spending on health totaled $2,330 billion, of income levels, total spending, and public and private shares or about 9 percent of global income (figure 1). Of this, high- of that spending. The fourth section discusses the main issues income countries (per capita income above $8,500) concerning the public-private nix of spending, the rationales accounted for just over $2,000 billion-89 percent of the for public and private financing, and the advantages and mar- total health expenditure. The populations of these coun- ket failures associated with financing health care through tries, however, accounted for just 16 percent of the global insurance mechanisms. The fifth section providesanoverview population (figure 2). The extreme disparity between the of the different sources for public financing of health care, amount of resources low- and middle-income countries and evaluates these sources in terms of economic efficiency, equity, high-income countries devote to health care reflects the and administrative feasibility, and discusses which public widely varying capacities of these country groups to provide financing sources are most appropriate for developing coun- health services. TABLE I Economic and health indicators by region and income group, circa 1994 Economic indicators Health outcomes Health services Per capita Per capita GDP growth, Under-five Adult mortality, Hospital GDP 1996-2005 mortality ages 15-60 Physicians per beds per Region/income group (1994 US$) (percent) (percent)a (percent)a 1,000 people 1,000 people East Asia and the Pacific 1,214 6.8 5.3 17.9 0.3 1.63 Europe and Central Asia 1,792 3.7 3.5 20.3 3.4 7.14 Latin America and the Caribbean 3,138 2.2 4.7 14.8 1.0 1.45 Middle East and North Africa 2,699 0.4 7.2 19.4 0.9 1.51 South Asia 440 3.7 10.6 23.5 0.2 0.53 Sub-Saharan Afrca 776 0.9 15.7 39.7 0.1 1.35 Low income 396 - 10.4 - - 0.87 Middle income 2,707 - 5.3 - - 2.12 Low and middle income 1,774 3.7 8.8 21.4 0.7 1.05 High income 18,611 2.4 0.9 9.7 2.5 6.29 Note: Regional figures are country-weighted averages. Income groups are based on 1994 GDP per capita: low income is $725 or less, middle income is $726-8,500, and high income is $8,501 or more. a. Based on current life tables. Source: World Bank 1996a and 1997; World Bank data. INNOVATIONS IN HEALTH CARE FINANCING 2 The gap between rich and poor nations is even more dra- ing from some form of noncommunicable disease. Over the matic when the distribution of the global disease burden is next three decades developing countries will undergo a considered. Of the estimated 1.4 trillion disability-adjusted major demographic and epidemiological transition, with sig- life-years (DALYs) lost in 1990, industrial countries nificant increases in the burden of injuries and noncom- accounted for just 7 percent (figure 3). Of these, 81 per- municable diseases (figure 4). These diseases are more cent were attributable to noncommunicable diseases. expensive to treat and harder to prevent. This transition will Developing countries, which accounted for 93 percent of reorient demand for health services and increase pressures the global disease burden, had a rather different disease for new investment in health care. profile. Except for countries in Europe and Central Asia, What are the prospects for narrowing the disparities which have demographic and epidemiological profiles sim- between rich and poor nations? Some perspective on this ilar to those in industrial nations, nearly half of the DALYs question can be gained by comparing the two groups' health lost in developing countries were caused by communicable service capacities and prospects for economic growth. diseases, mainly among children. Industrial countries have three times as many physicians per Aging populations and the rising incidence of noncom- capita and six times as many inpatient beds per capita as municable diseases will continue to drive up the cost of developing countries (see table 1). To close the resource patient care. In industrial countries a large portion of health gap, developing countries will have to make sizable invest- spending is used for a small percentage of patients in the ments in health services and increase spending at rates faster final years of their lives. Most of these patients are suffer- than those of high-income countries. FIGURE I FIGURE 2 Global distribution of health spending, 1994 Global population and income distribution, 1994 Global distribution Population distribution by region Latin America and the Caribbean middle-income ad High income South~~~ ~outre coutae 11% G6o 89% Europe and Central Asia Sub Sahar.an 18% Af X ./_ _ca Total global health I East / 10% expenditure: $2,330 billio Midd le Eas.n the Pacific and North Afr 3ca 43~~~~~~~~~~~~~% Low- and middle-income countries South Asia 8% ~~~~~~~~~~~~~~~~Global distribution of GDP European Central ZAiak Latin Amnerica M E W,\and the Caribbean 43% Sub-Saharan Afr'ica ~ Vihicm 6 \ countrie Total health expenditure: Middle East and North Afnca $250 billionl 9%Note: South Asia includes India and East Asia and the Pacific includes China. Source: World Bank data. Source: Wodd Bank data. A CURMUDGEON'S GUIDE TO FINANCING HEALTII CARE IN DEVELOPING COUNTRIES 3 FIGURE 3 FIGURE 4 Global disease burden, 1990 Changing burden of disease pattem in developing (disability-adjusted life-years lost) countries, 1990 and 2020 OECD countries 1990 Comunicable Total: 99 million DALYA, (7% of global total) Total: 1,280 million DAY Developing countries 2020 Total: 1,280 million DALYs (93% of global total) Total: 1,292 million DALYs Source: WHO i 996b. Source: WHO I 996b. Such an adjustment maybe achievable in East Asia, where zational arrangements in deterrmining the efficiency, qual- per capita GDP is projected to grow by 6.8 percent a year ity, and equity of health delivery systems (OECD 1995). over the next decade. But in other regions-especially the Although there have been many studies of health deliv- Middle East and North Africa and Sub-Saharan Africa- ery systems, less information is available on the inefficien- annual economic growth rates will be less than 1 percent cies and inequities associated with different health financing (see table 1). These projections have particularly serious systems. At least one study suggests that systems that rely consequences for Sub-Saharan Africa, where the base of on social security financing might be more costly to admin- health infrastructure is already quite weak. ister than systems that rely on general revenue sources Strengthening health service capacities will require (Poullier 1992). In developing countries the lack of infor- expanding facilities and personnel as well as improving the mation on access to and distribution of services, utilization quality of services. Countries at similar income levels show rates (inpatient admission rates, physician visits per capita, considerable variation in the performance of their health and so on), and quality measures have limited cross-coun- systems-variation that can be partly ascribed to differences try comparisons of what a dollar's worth of health expen- in the equity, efficiency, and quality of health services. For diture buys in terms of effective health services. example, the average number of hospital beds and physi- cians per capita is higher in Europe and Central Asia than Health Care Financing and Spending in high-income countries (see table 1). Yet overutilization Patterns and inappropriate clinical interventions raise questions about how effectively these resources are being used. Recent stud- Policymakers face the perpetual challenge of raising suffi- ies of OECD countries point to the importance of organi- cient revenue for the health sector in an equitable and effi- INNOVATIONS IN HEALTII CARE FINANCING 4 cient way. Although most countries recognize that health occurred only sporadically during the 1970s and 1980s care is a right for all citizens-as embodied in the World (World Bank 1987). The first sustained and replicable effort Health Organization's (WHO) goal of "Health for All in to develop such an information base for the twenty-four the Year 2000"-there are no clear guidelines on how this member countries of the Organization for Economic objective translates into health service delivery, and whether Cooperation and Development (OECD) began in 1977, such services are affordable. and the information is now updated each year (OECD 1977, What resources can a country with a per capita income 1985, and 1993). The experiences of the OECD countries of $400 expect to raise for its health sector, and what kind led to the development of a system of national health of services can it provide for its citizens? The World Bank's accounts, which only began to be used in developing coun- World Development Report 1993: Investing in Health offered tries in the past two or three years.1 World Development a normative response to that question using the concept of Report 1993 was the first comprehensive effort to system- a minimal package of care and services. In principle such a atically develop expenditure information for all developing package would cut the number of lost DALYs relative to countries (World Bank 1993). available resources in countries at different income levels Despite these efforts, health expenditure data for devel- (World Bank 1993). Although this approach provides one oping countries are often incomplete or unavailable, espe- set of objective criteria for rationing health care services, cially for private spending. Definitions of health spending decisions about health spending cannot be isolated from a vary by country, and disaggregation of health spending country's social, political, and economic characteristics. beyond the broad categories of public and private is even Moreover, factors beyond the control of policymakers often more problematic. As a result developing countries lack affect spending. Understanding the interaction of these fac- the basic information and tools needed to assess how health tors is essential to designing health policy. system resources are being raised and used. Without such Policymakers must have some way to evaluate the per- information it is extremely difficult for policymakers to formance of their country's health systems against those of understand the effects of their policies and to determine other countries or regions at comparable income levels. One which decisions are likely to ensure equity in financing and approach is to divide the performance of health financing increase returns on the resources devoted to the health mnechanisms into three broad categories (figure 5). The first sector. category is concerned with how efficiently and equitably Moreover, without such information it is difficult to gauge revenues are raised, and what effect they have on the size the effectiveness of past investments or to evaluate current and distribution of the resources available to the health sec- investments. Indeed, the importance of these data and the tor. The second category involves evaluating how efficiently and equitably resources are used to provide health ser- Different ways of measuring performance vices. The third category relates to the effects health expen- in health financing ditures have on health outcomes. This last measure is tied In_healthfinancin to intersectoral factors such as education, water and sani- Revenue Health services Health status generation output and outcomes tation, and women's status, since health services are just Key issues: Key issues: Key issues: Efriciency, stability, and Eficiency in service output Intersectoral determinants one factor among many that determine a population's health sustainability in revenue per dollar invested; access; of health inccuding medical status. This paper focuses on issues related to the first group generation; effective- quality; patient satisfaction care, household behavior, ness in risk pooling and and patient choice. women's status and educa- of measures. redistribution. tion, water and sanitation, International comparisons of health financing are diff- Examples of instruments: environment, nutrition, aExomples of instruments: Unit cost of effective and lifestyle changes. cult, partly because of the lack of reliable data. Efforts to Burden of taxation, services by different distribution of public facilities, payment systems, Examples of instruments: compile comprehensive cross-country data on health expen- subsidies, extent of organizational and manage- Mortality and morbidy rate ditures date back to Brian Abel-Smith's work in the 1960s insurance coverage ment stnrctures. reduction (increase in life (formal and informal). expectancy, reduction in for the World Health Organization (see Abel-Smith 1967). infant mortality, life years Efforts to update that information for developing countries saved per dollar). A CURMUDGEON'S GUIDE TO FINANCING HEALTEI CARE IN DEVELOPING COUNTRIES 5 difficulty in compiling them strongly suggest the need for collected, about 37 percent of the private health expendi- direct, systematic, and regular collection of health system data ture data were imputed from a regression model, compared from all developingcountries, possiblyby adapting the national with 17 percent for public health expenditures. health account approach used by OECD countries. Priority In this paperpublic expenditures on health refer to funds should be given to developing an accounting system that is from government budgets, compulsory (that is, publicly affordable and easy to use in the developing country context. mandated) health insurance funds (social security schemes, The World Bank and other intemational organizations could mutual funds, sickness funds), and external loans and grants.2 support this effort with financial and technical assistance. Private expenditures refer to direct household expenditures, The World Bank has attempted to update global health including out-of-pocket payments for services, expenditures expenditure data to 1994 using the latest available sources through private health insurance plans, direct payments (actual data range from 1990 to 1995). Data have been for health services by firms and corporations, and charita- collected on total, private, and public health expenditures, ble contributions. and limited information has been obtained on sources of financing (social health insurance, donors) and types of ser- Regional health expenditure patterns vices (hospitals, pharnaceuticals). The following discussion is based on preliminary analyses of this database. Although Average per capita health expenditures range from $16 in efforts have been made to ensure comparability across coun- low-income countries to $1,827 in OECD countries-a tries, the analysis should be interpreted with caution. Data hundredfold difference (table 2). OECD countries also were obtained from various sources (public expenditure spend more on health as a percentage of GDP Low-income reviews, government budgets, household surveys, World countries spend about 4 percent of GDP on health; OECD Bank reports, sector reports), so definitions and collection countries spend more than 8 percent. methods varied. Moreover, data on private health expen- Among regions, South Asia spends the least on health diture are even more prone to measurement errors because as a percentage of GDP-less than even Sub-Saharan Africa. of a lack of reliable information on households and private However, a significant portion of health costs in Sub-Saharan enterprises. Of the 202 economies for which data were countries are financed by external sources (see below). Other TABLE 2 Per capita GDP and health expenditures by region and income group, circa 1994 Public health Per capita expenditure Per capita GDP health expenditure Health expenditure as a share of total as percentage health expenditure Region/income group PPP$ US$ PPP$ US$ of GDP (percent) EastAsiaandthePacific 4,554 1,214 158 38 4.1 52 Europe and Central Asia 3,847 1,792 346 154 7.2 72 Latin American and the Caribbean 5,729 3,138 367 200 6.1 49 Middle East and North Africa 7,181 2,699 353 116 5.2 50 South Asiab 1,887 440 65 12 3.7 39 Sub-Saharan Africa 2,070 776 III 38 4.0 54 Low income 1,565 396 71 16 4.3 47 Middle income 5,790 2,707 364 168 5.3 57 High income 20,615 18,611 1,521 1,468 6.9 67 OECDc 21,169 22,498 1,777 1,827 8.3 76 Note: Regional fgLures are country-weighted averages. International dollars (PPP$) are local currencies converted to U.S. dollars through the use of purchasing power parities ('exchange rates' that adjust for cost differences across countries). Income groups are based on 1994 GDP per capita: low income is $725 or less, middle income is $726-8,500, and high income is $8,501 or more. a. Indudes China. b. Indudes India. c. Exdudes Hungary, Mexico, and Turkey. Source: World Bank data. INNOVATIONS IN HEALTH CARE FINANCING 6 regional differences cannot be explained by differences in FIGURE 6 per capita income alone. For example, although countries Public share of health expenditures and per capita in Europe and Central Asia have the third-highest income GDP, various countries, circa 1994 level among low- and middle-income countries, they spend Public share (percent) more than 7-percent of GDP on health, the highest among '00 this group of countries. Per capita incomes and the public share of health care It* . 1 costs tend to rise together, indicating an expanding gov- 60 ernment role in health care financing as countries develop economically (figure 6). In OECD countries the public 40 sector accounts for, on average, more than 75 percent of total health spending. Developing countries show consid- 20 e erable variation in the public share of health spending. ___ This heterogeneity underscores the diversity of approaches 100 1.000 o0,000 100.000 to health care financing in developing countries and reflects Per capita GDP, US$ (log scale) these countries' historical, political, and economic struc- Source: Wodd Bank data. tures. For example, the large public share of health spend- ing in Europe and Central Asia is a legacy of highly centralized Asia, for example, social insurance funds finance a significant government structures under socialism. portion of health expenditures. In 1994 social insurance costs in the fifteen (of thirty-four) Latin American and Caribbean Composition of public expenditures countries that reported such information ranged from 7 to 60 percent of total health expenditures (PAHO 1996). In As noted, data on public health spending have been drawn Europe and Central Asia social insurance funds accounted from government budgets, compulsory health insurance for 20 to 80 percent of health expenditures. But in the Middle funds, and external loans and grants. Finding a consistent East and North Africa just seven (of nineteen) countries definition of compulsory health insurance funds is some- reported having compulsory insurance schemes, which what problematic; thus the data on this component of health financed between 9 and 37 percent of health expenditures. expenditures should be interpreted with caution. Still, some (Most high-income oil-exporting Gulf countries finance health interesting trends emerge. services directly from government budgets.) Public insurance schemes play a limited role in most low- External assistance continues to be an important source income countries, where public health expenditures usually of financing for health services in low-income countries, espe- come directly from government budgets. For example, in ciallyin Sub-SaharanAfrica and SouthAsia (exdudingIndia). the four Sub-Saharan African countries that reported on Since the mid-1980s, however, the share of bilateral official public health insurance, the schemes accounted for 1-9 per- development assistance (ODA) allocated to health has been cent of total health expenditure. In India they accounted declining, while ODA from multilateral sources-including for just over 1 percent. Yet in China the Government the World Bank-has been increasing (Michaud and Murray Insurance System accounted for nearly 30 percent of health 1996, p. 230). The tightening of ODA resources makes it expenditures in 1993. hard to predict how ODA to the health sector will evolve Social insurance schemes play a larger role in middle- over the next decade. For many middle-income countries income countries. Again, though, the pattern that emerges external assistance may fall, with the emphasis of aid shift- is fairly diverse. Some countries continue to rely on-govem- ing from financial assistance to technical assistance. Yet ment budgets to finance public health systems, while others low-income countries, especiallythose in Sub-SaharanAfrica, are moving toward payroll tax-based insurance schemes. In will likely continue to depend on external assistance for at Latin America and the Caribbean and Europe and Central least the next decade, and possibly longer (table 3). A CURMUDGEON'S GUIDE TO FINANCING HEALTH CARE IN DEVELOPING COUNTRIES 7 Composition of private expenditures provide benefits or externalities that extend beyond the individual) and personal healt services (which benefit only Data on private health spending are even harder to obtain the individual; see Musgrove 1996). Funds can be raised than data on public spending. Most of the data on private (or derived) through taxes, mandates, private health insur- spending are estimates drawn from household expendi- ance, direct private out-of-pocket payments (including user ture surveys. These data usually are not disaggregated into charges for publicly provided services), grant assistance, different forms of payment, such as direct fees for service, charitable contributions, and domestic or foreign borrow- insurance premiums or other forms of prepayment, and ing. They can be managed by government (ministries of cost-sharing payments. And only a few countries report health) or quasi-goverrment agencies (social security orga- expenditures from private insurance schemes and direct nizations, sickness funds), for-profit or nonprofit private corporate spending on health services, which often account entities (private insurers, purchasing cooperatives, employ- for a significant portion of health expenditures, especially in low-income countries. These categories should be included TABLE 3 in future data collection efforts. External assistance for health costs by region, circa 1990 Income elasticities (percent) Extemal assistance Income elasticities provide a useful measure of how dif- as a share of per capita ferences in countries' income levels translate into differ- Region health expenditure ences in health expenditures (table 4 and figure 7). The East Asia and the Pacific (exduding China) 3.7 global elasticity is estimated at 1.13. Thus for every 10 per- Latin American and the Caribbean 3.6 cent difference in per capita income there is a 11.3 percent Middle East and North Africa 1.5 difference in per capita health expenditures-that is, coun- South Asia (excluding India) 13.1 India 0.7 tries with higher incomes tend to devote a larger share of Sub-Saharan Africa 16.4 those incomes to health expenditures. The income elastic- Note: Regional data are country-weighted averages. Not enough data were avaibble ity for the public component of health expenditures is 1.21; for Europe and Centurlriat 9p9r4pWord Bank data. for private spending it is 1.02. This pattern suggests that public health spending is more responsive to income dif- TABLE 4 ferences than is private health spending, and is consistent Income elasticities for total, public, and private with the fact that high-income countries have larger public health care spending, circa 1994 shares of total health expenditures. Income Number of Adjusted Income elasticities for countries by income level are Spending category elasticity (il) observations R2 shown in table 5. Income elasticities for per capita health Total health expenditure 1.13 122 0.94 expenditures relative to per capita GDP are highest for high- Public 1.21 162 0.91 income countries (1.47), followedby middle-income (1.19) Private 1.02 126 0.85 Note: Dependent variable is per capita heakh expenditure (US$). and low-income (1.00) countries. This pattern is also con- Source: World Bank data. sistent with the fact that higher-income countries devote a larger shre of resurces tothe healt sector.TABLE 5 larger share of resources to the health sector. Income elasticities by income group, circa 1994 The Public-Private Financing Mix Income Number of Adjusted Income group elasticity (il) observations R2 Healh care systems are financed by many sources, public Low income 1.00 31 0.34 Middle income 1.19 57 0.82 and private. These funds are managed by public and pri- High income 1.47 34 0.64 vate entities and spent on bothpublic health services (which Source: World Bank data. INNOVATIONS IN HEALTH CARE FINANCING 8 ers, unions), or consumers (out-of pocket payments, med- FIGURE 7 ical savings accounts).3 Funds are then used to purchase Per capita health spending and GDP, various publicly or privately provided health services (figure 8). countries, circa 1994 The basic issues relating to the appropriateness of pub- Per capita health expenditure lic or private sources of finance are predicated on govern- (log scale) ments' allocational, distributional, stabilization, and economic 0,000 goals and on the policies that are used to correct for mar- ket failures and externalities in the financing, consumption, 1.000 and provision of health services. Particularly relevant are insurance market failures and instabilities, which may pre- 100 clude people from obtaining the benefits of collective risk reduction through efficient insurance provision. 10 Several other potential market failures affect the health sector (see Hsiao 1995; Musgrove 1996; Jonsson and Musgrove in this volume). One important market failure 100 1P000 10(000 100,000 involves extemalities in consumption, whereby the collec- tive benefits from consumption of health services are greater Source: World Bank data. than the individual benefits. Market failure also can pro- vide a rationale for public financing because of the effects Although the taxonomy of services described above is ill health has on income redistribution, income levels, and used in this analysis, other categorizations of health services poverty. Other areas of market failure that have implica- are relevant for discussions of risk pooling and private and tions for public and private financing deal with informa- public responsibilities. In particular, health services can be tion gaps and asymmetries, interdependence between supply classified as preventive, curative for unexpected health prob- and demand, and supply-side market failures.4 lems, curative for chronic predictable health problems, and curative for lifestyle-induced health problems. Once these Health services with collective benefits services have been defined, societies can then choose appro- priate direct subsidies and cross-subsidies (through risk pool- Certain health services-vector control, clean air and water, ing) to finance the costs and determine individual sanitation systems, environmental health, medical research, responsibilities for lifestyle choices-substance abuse, promis- most health promotion and education activities-are purely cuity, obesity-that adversely affect health (Nichols 1996). public goods. That is, no individual can be excluded from World Development Report 1993 outlines how govern- the benefit, and consumption by one individual does not ments can invest scarce public funds in cost-effective basic reduce the amount available to others. Other goods, known public health services. For $12-22 per capita developing as merit goods, benefit individuals as well as communities. countries will get the best return on public health spend- Immunizations and treatments for contagious diseases are ing by investing in a basic package of public health and examples. Left to their own devices, most people will spend essential clinical services, including immunizations, school- on public and merit goods only up to the point at which based health services, programs to reduce alcohol and their private (marginal) benefit equals the private cost, and tobacco consumption, family planning services, tuberculo- society as a whole will underconsume these services.5 Thus, sis control, control of sexually transmitted diseases, and care in order to ensure an appropriate collective consumption for childhood illnesses such as acute respiratory infections, level, such services should be publicly financed (or subsi- diarrheal diseases, measles, malaria, and acute malnutrition dized). Around the world, most of the health services that (World Bank 1993). In many of the poorest countries, how- are consumed are personal health services, for which ben- ever, governments have not mobilized sufficient resources efits accrue largely or exclusively to individuals. for such a package of care, and in some cases both gov- A CURMUDGEON'S GUIDE TO FINANCING HEALTII CARE IN DEVELOPING COUNTRIES 9 ernments and individuals have chosen to spend funds on is no universal prescription on the appropriate mix of pub- other, less cost-effective services (World Bank 1994). lic and private financing or the appropriate role of the state Although World Development Report 1993 does not pro- in financing personal health care services (other than those vide definitive answers on how to maximize the returns to for the poor and other vulnerable groups, which are justi- public and private expenditures above this threshold, many fied on equity and income redistribution grounds), there is developing countries could obtain a better return on their considerable scope for developing countries to improve their public health investments by providing this basic package health investments and outcomes. of public health and clinical services. Many of these coun- tries have invested significant public resources on state-of- Redistribution the-art curative services, which are often available only to the ruling elite, the politically connected, or those able to In most countries the bulk of health expenditures are for buy such services from the private sector. Although there personal (individual) health services. Although in principle FIGURE 8 Sources, management, and provision of health care financing Revenue Managed Provided source by by ~~~~~~Taxes § .~~~~~~~~ - - - - - - - - - - - - - - -o _ Government _. i ~~~~~~~~~~~~~~~~~~~~~~agencies ~~~~User charges --- _ __ . Public Public - ~~~~~~~~~~~~~~~~~~~~~~~~~~~~sources l 0 . ~~~~~~~~~Socia] insurance/ l l _ ~~~~~~~~~~sickness funds Mnates "L%. Private hearPnvate organizabons/ Gnsurance 1X.. insurers S | ~~~Lc,ans s 1ll l 0 | | = | ~~~~~~~~~~~~~ ~~Employers ll l Private P vateNNOVATIOS IN E C ' NANCINt | Charitable } a . l l W | ~~~~~~~~~~~~Individuals l . =8ffi=Ri2=.tt*'.=. . . Out-of-pocket|lll expenses |I I insurrc . . .. . . .... INNOVATIONS IN HEALTH CARP FINANCING - ~ ~~~~~ 10 these services can be financed privately, almost al societies of ilness and medical expenses, information asymmetries view access to health care as a basic human right. Thus between insurers and consumers, adverse selection, and governments often provide access to health services for peo- moral hazard. To understand how these factors can cause ple who cannot afford them. Moreover, many governments market failure, it is essential to understand the conceptual fund health insurance for vulnerable groups or provide basis for insurance.8 personal health care because of the direct link between ill- ness, earning potential, quality of life, and poverty, as well Rationales for insurance. Most people prefer to avoid as the random nature of ilness. facing risks that result in substantial economic losses. Thus Yet equity considerations can generate inefficiency and they are risk averse and, given the opportunity, will avoid, market failure in private insurance markets and impose minimize, or shift risks to others. They are willing to pay a significant costs on government. Put another way, equity relatively small certain price to avoid a relatively large unpre- concerns often reflect judgments about tradeoffs between dictable loss. The degree of risk faced by an individual (or desired redistribution and the distorted incentives that often insurer) depends on the accuracy with which the probabil- accompany redistribution (Pauly 1996). If consumers choose ity of the adverse event occurring can be predicted. The not to spend any money on health-related goods and ser- greater the uncertainty, the higher the risk. vices, knowing that the state will pay for their medical costs Insurance reduces risks by improving the predictabil- on equity grounds, risk pooling through private markets is ity of the adverse event through the pooling of a large num- undermined, and the state may be left funding the bulk of ber of similar risks. From a financial perspective, insurance society's health risks-with deleterious fiscal and redis- is an arrangement that redistributes the costs of unex- tributive implications for both the government and the pri- pected losses (Dorfman 1982, p. 5). From a legal per- vate sector. Moreover, such actions reduce the potential spective, insurance is a contract in which the third party gains from risk pooling and could result in excessive taxa- agrees to compensate the subscriber for specific costs tion so that government can finance medical expenses (see incurred when a specific loss occurs (Hall 1994, p. 6). below). Thus societies must weigh the welfare costs of Insurance has two components: the expected loss (total individuals being unable to privately purchase "needed" loss incurred times the probability of the loss occurring) health services against the butdens of publicly financing and the risk premium and load factors (amount the sub- such needs. scriber is willing to pay to avoid the expected loss and, from the insurer's perspective, costs of marketing, prof- Market failures in health insurance its, and administration). Insurance lowers risks in the aggregate and makes them Insurance is prepayment for services that will be paid for by more predictable because pooling a large number of simi- a (public or private) third party (the insurer) should a pre- lar events increases the predictability of the event.9 Thus defined event occur.6 Insurance is a substitute for (or in some the larger is the insurance pool for a particular risk, the cases a complement to) direct out-of-pocket payment for greater is the likelihood of correctly assessing the proba- such services. As discussed below, insurance reduces risks bility of the loss occurring. Risks that are unpredictable for by pooling them. Whether public or private, insurance affects the individual become predictable for the group and can the distribution of health care financing among households be estimated accordingly. Since there are large economies and can also affect the delivery of health services.7 of scale in terms of both administration (load factor) and The potential market failures that arise from the insta- improved accuracy of loss prediction, a risk becomes eas- bilities inherent in insurance markets provide another poten- ier to insure as the insurance pool for that risk grows. tial rationale for government financing of personal health Conversely, risk premiums and load factors increase as the services (as well as for government regulation of private pool shrinks. For example, private health insurance premi- insurance). Much of this instability occurs because of indi- ums in the United States are much higher for small groups vidual aversion to risk, uncertainty about the random nature and individuals than they are for large employment groups. A CURMUDGEON'S GUIDE TO FINANCING HEALTII CARE IN DEVELOPING COUNTRIES 11 Under certain circumstances the load factors maybe so high adverse selection and moral hazard. Adverse selection, that insurance markets are not viable. also known as biased selection, occurs because of an infor- Successful insurance systems share several characteris- mation asymmetry that arises when insurance subscribers tics: have better information about their individual risks than • Most individuals are risk averse and prefer to substitute the insurer. As a result there is a higher than average prob- a small outlay (the insurance premium) for a large uncer- ability of the adverse event occurring, since people are more tain loss. 10 likely to purchase insurance that is offered at an actuarially * By pooling a large number of similar risks, insurers are fair price for the entire community. Thus higher-risk indi- able to predict losses accurately and charge a premium viduals pay an average premium that is well below what an appropriate to that risk. actuarially appropriate rate for their risk group would be * Individuals pool their potential losses and pay a relatively (that is, insureds are not charged a rate for transferring their small premium for the right to collect individual losses exposure to loss that fairly reflects the cost of the transfer; from the pool. see Dorfman 1982, p. 24). Such actions can destabilize * The premium is small relative to the potential loss voluntary insurance markets, since healthier individuals will (Dorfman 1982, pp. 22-23). eventually drop out as premiums rise, creating yet higher Several other supply and demand factors affect the via- premiums and further healthy dropouts. This phenome- bility of insurance markets. The main factors affecting non is referred to as the premium death spiral. demand for insurance are the size and predictability of There are severalways to dealwith adverse selection. Since risk. People will insure against large unpredictable losses adverse selection occurs largely in voluntary private insur- (Pauly 1986). People will not insure against losses that are ance markets, one solution is to create a mandatory public certain to occur, since in that case pooling does not pro- insurance system. By requiring everyone to join, the adverse vide any advantage. Similarly, while in a perfectly rational selection problem is elimninated. But so too is a great deal of world most individuals should not insure against small losses consumer choice, which results in a welfare loss to society. (since they can self-insure against such events and avoid the Private insurers deal with adverse selection in three ways: risk premium), many individuals want to be insured against by obtaining information about the underlying medical risks small predictable losses as well as large ones.11 of individual subscribers, .by not covering some of these An important distinction regarding health care financ- underlyingrisks, andbyselling insurance products thatpre- ing and the appropriate roles of the public and private sec- dude selection on the basis of risk. Insurers use a variety tors follows from these observations. In its purest form of methods to obtain information about the underlying med- insurance isa mechanism that reduces risks by pooling them. ical risks of individuals so that they can set appropriate Insurance can also be used as a financing mechanism through premiums or not cover these risks. These methods, some- which public or private entities collect premiums (includ- times classified under the general rubric of risk selection (as ing administrative costs) to cover highly probable or com- well as risk rating and medical underwriting), include requir- pletely predictable losses. Where losses are completely ing medical examinations, examining previous medical predictable, insurance provides no collective benefit in the claims, establishing waiting periods, excluding preexisting form of risk reduction. Still, there may be equity reasons medical conditions, not guaranteeing renewability of the for a government to cover services in this manner, and insurance policy, and refusing to insure the individual. individuals may demand that such benefits be included in Marketing insurance to individuals or groups formed for basic insurance packages. The inclusion of such benefits, a purpose other than to obtain health insurance also miti- however, undermines insurance markets (Hall 1994, p. 25; gates the risk of adverse selection. Marketing to employee Chollet and Lewis in this volume). groups has this characteristic because individuals join the group for employment, not health insurance.12 Employee P'roblemsin insurancemarkets. Two supply-side aspects of groups also tend to be healthier and are generally large insurance create major problems for insurance markets: enough to effectively pool risks. INNOVATIONS IN HEALTEI CARE FINANCING 12 A second supply-side instability in insurance markets kets that provide informed consumers with a range of choice occurs as a result of moral hazard. Moral hazard is the ten- of insurance policies can maximize social welfare. But risk dency for insurance to increase the probability of the occur- pooling for individuals and small groups, adverse selection, rence of the risk being insured against (Arnott and Stiglitz risk selection, and moral hazard can create economic ineffi- 1988). (In more general terms, moral hazard is the risk ciency and inequity in insurance markets that justify govern- that individuals will change their behavior because of the ment intervention in public finance, provision, and organization existence of a contract.) Because insurance lowers the cost of insurance, and public regulation of private insurance (Aaron of service to the consumer at the time of use, individuals 1991, pp. 11-19; Arnott and Stiglitz 1990). In any event the tend to consume more health services than they would in costs of government intervention must be weighed against the absence of insurance. Similarly, insurance may cause these inherent private market inefficiencies. 15 individuals to use less preventive services or take fewer Still, it is clear from the experience of the United States- precautions to avoid accidents or deterioration in their the one major industrial country to rely on voluntary (tax- health. Moreover, since the costs of excess use are spread subsidized) private insurance-that there are significant over all other purchasers of insurance, individuals have lit- problems with risk pooling, access to insurance, and the tle financial incentive to restrain their use. costs of health insurance and health services. Many of Moral hazard occurs for all types of insurance-public or these problems occur as a result of the rating factors used private, voluntaryor compulsory. Thebenefit packages of most to set premiums as well as the risk selection methods used insurance programs are designed to deal with moral hazard. by insurers to prevent adverse selection.16 Rating factors Features that mitigate moral hazard in health insurance include include health status, age, sex, industry or occupation group, cost sharing (deductibles, copayments, coinsurance), physi- group size, and geographic location. Rating and risk selec- cal (for example, maximum of twenty mental health visits) tion often discriminate against small employer groups and and financial limits on benefits (for example, $10,000 in cov- individuals as well as vulnerable populations such as the erage for pharmaceuticals), and total expenditure limits on handicapped and the elderly They result in significant mar- policies (for example, lifetime insurance coverage limited to ket segmentation and premium differentiation that under- $500,000). Frequent renewability is also used to deter moral mine effective risk pooling. hazard since premiums can be adjusted to account for the Since the purpose of insurance is to pool risks, some ana- actual and projected experience of the group. lysts argue that the most effective way to do so is through Another innovation to deal with moral hazard is the uti- community rating, inwhich (except for adjustments for fam- lization management practiced by managed care organiza- ily status, geography, and benefits design) everyone in the tions (Enthoven in this volume). In their most complete community pays the same average premium (American form managed care organizations integrate the functions Academy of Actuaries 1994).17 But community rating in of managing financing and providing care. As a result, voluntary insurance markets increases the likelihood of through various forms of utilization management by med- adverse selection and creates "death spiral" instability, with ical professionals employed directly or under contract to the healthier individuals opting out. The arguments for com- managed care organization, subscribers have far less dis- munity rating are that risk.pooling is the purpose of insur- cretion to demand services. 3 Managed care principles can ance, that younger healthier individuals who opt out may be adopted by both public and private financing entities. end up needing medical care that will ultimately be pro- It is not clear whether these insurance market instabilities vided at public expense, and that these same individuals at justify compulsory public financing (or provision of publidy a later point in their life cycle will end up benefiting from sponsored insurance) for all individuals. 4Assuming that pub- this arrangement. Unlike the United States, most industrial lic health services and redistributive concerns are dealt with countries have dealt with these problems through public separately, the question becomes one of whether regulated financing and provision of health insurance or health ser- private insurance markets can operate efficiently. On eco- vices. Some U.S. reformers (including President Clinton) nomic welfare grounds there is little doubt that efficient mar- have argued that public finance and compulsory coverage A CURMUDGEON'S GUIDE TO FINANCING HEALTI-I CARE IN DEVELOPING COUNTRIES 13 for all is the only way to deal with this problem. Opponents In addition, direct purchase does not result in the gains from of public finance have proposed dealing with it through reg- risk pooling obtained through insurance. ulation of private insurance markets, particularly by regu- lating rating factors and medical underwriting practices and Implications for the public-private mix through public subsidies to or taxes on the insurance indus- try to subsidize the premiums of high-risk groups (Helms, These concems lead to several prescriptions for public and Gauthier, and Campion 1992). private financing of health services: * Public health services should be financed publicly. Market failure and direct consumer purchase of * Personal health services that have collective benefits health services should be publicly subsidized. * Personal health services that have no collective benefits Direct consumer purchase of health services-whether basic can be publicly or privately financed. services, supplementation of insurance benefit packages, or * Personal health services for vulnerable populations are higher-quality services than are offered under an insurance generally financed publicly on equity grounds. program or by the public sector-are an important source of * Insurance reduces overall risks by pooling them and is health care revenues and expenditures in all systems, but espe- thus a preferred method for fnancing health services. ciaily in low-income countries. It can be argued that insur- * Insurance for personal health services can be publicly or ance, whether publicly orprivately financed, is the best vehicle privately financed. for financing health services, since by pooling risks, overall * Instabilities in insurance markets necessitate government risks are reduced. But since no society predudes its citizens regulation and under certain circumstances public from buying legal goods and services and since in many low- financing. income countries governments cannot raise sufficient rev- * Market failures create inefficiencies relating to individ- enues to finance personal health care services for their ual out-of-pocket purchase of health services. populations, it is important to consider whether market fail- * In some cases government regulation and provision of ures result in inefficient consumption and provisionwhen indi- information may be a viable alternative to public financ- viduals purchase services directly on an out-of-pocket basis. ing of personal health services. Several market failures affect the direct consumer pur- chase of personal health services. Some of these failures, Public Financing Sources such as entry barriers and decreasing costs of production (which provide a rationale for public provision or public As noted, health care systems can be financed by a variety utility-type regulation), are onthe supply side. Others, such of public and private sources, and these funds can be spent as consumers lacking information about what services they on many types of public and personal health services. Certain need, about what works in medicine, and about the prices public health services (such as immunizations) are quasi- and quality of competing medical care providers, are infor- public goods and should be publicly financed or subsidized. mational. Because of such information gaps, the care provider Public fnancing of personal health services is justified on often acts as the consumer's agent-yet the financial moti- redistributive and equity grounds. vations of the provider may not be entirely consonant with Two other potential rationales for public financing of per- maximizing the welfare of the patient. As a result no indus- sonal health services for the nonpoor as well as the poor, trial country relies exclusively on free markets to produce either directly or through publicly financed health insurance and allocate health care (Aaron 1991, p. 8). Moreover, direct systems, result from market failures in private health insur- purchase of services cannot contribute to equity objectives ance markets that preclude effective risk pooling and in (that is, it is a pure benefit approach) and, as noted, pri- the consumption and provision of personal health services. vate purchase without government subsidies will result in Public "insurance" systems can be of two general types: less than optimal consumption of certain health services. national health service approaches (Beveridge model) and INNOVATIONS IN HEALTH CARE FINANCING 14 TABLE 6 Government revenues by region and income group (percentage of GDP) Standard Range Number of Region/income group Mean Median deviation Minimum Maximum observations East Asia and the Pacific 23 20 9 8 37 11 EuropeandCentralAsia 39 41 12 12 60 13 Latin Amerca and the Caribbean 24 25 8 8 42 2 1 Middle East and North Arica 32 31 10 12 48 10 South Asia 27 20 15 10 47 6 Sub-Saharan Africa 26 22 15 11 63 20 Developing countries 28 26 12 8 63 76 Industrial countries 45 44 8 31 62 21 Low income 20 19 9 8 44 22 Middle income 31 30 12 8 63 54 High income 42 44 11 12 62 24 Note: Data are forthe latest available year between 1990 and 1995. Incudes central, state, and lcal govemment revenues. Also includes taxand nontax revenues, as well as grant assistance. Source: IMF, Govemment Finance Statistics, various issues. social insurance funds (Bismarck model; see OECD 1992). Government revenues In both models collective risk pooling is achieved through compulsory taxation. Total government revenues (central, regional, and local) as There are, however, two general distinctions. First, a percentage of GDP vary significantly by region and by national health services tend to be financed from a mix of income level (table 6). The relationship between per capita general taxes and other public revenue sources, while social income and governments' ability to raise revenue is shown insurance funds tend to be financed with earmarked pay- in figure 9. Several patterns are apparent: roll taxes.18 Second, because national health services are * Revenue-raising capacity increases with income. financed from the general budget, they are subject to annual * Relative to their GDPs, low-income countries and regions budget processes. Social insurance funds tend to be more can raise less than half the revenues that high-income independent of such annual political machinations. There countries can raise. is no one "right" approach for developing countries to use on this issue, and various mixes of public and private financ- FIGURE 9 ing are possible (seeJ6nsson and Musgrove in this volume). Government revenues and per capita GDP, various Although much of the discussion of health financing c focuses on the benefits of health expenditures and on what Revenues (percentage of GDP) services should be purchased, cost and equity issues are also involved in raising revenues to finance public expenditures. 80 These costs make it of critical importance for governments to evaluate the benefits of public expenditures against the 60 0 * costs (both economic and equity) of raising these revenues., 4 o * As stated in World Bank (1991), "incremental changes in 40 O O the level of taxation should reflect, among other things, 20 44° the benefits derived from incremental changes in the pub- *o o lic expenditure program and the relative costs of financing o 100 1,000 10,000 100,000 it by means of taxation or non-tax revenues" (p. 18). Thus Per capita GDP (US$) policymakers must consider both the sources and uses of funds as they consider health care financing reforms. Source: IMF Government Finance Statistics. A CURMUDGEON'S GUIDE TO FINANCING HEALTH CARE IN DEVELOPING COUNTRIES 15 * As a group developing countries raise less than two-thirds since both regions have significantly higher per capita incomes of the revenues raised in industrial countries. than Sub-Saharan Africa and South Asia, yet their ratios of Regional variations also suggest several interesting pat- revenue to GDP are similar. This similarity may reflect the tems. First, the high ratio of revenue to GDP in the Middle disparate mix of low- and middle-income countries in both East and North Africa likely reflects oil and gas revenues. regions, incomplete revenue data for some countries,19 inef- Second, high revenue generation in Europe and Central Asia fective tax administration, or societal preferences for indi- likely reflects the legacy of and continued reliance on the vidual rather than government responsibility. centralized revenue-raising systems created under socialism, Still, the basic conclusion on developing countries' rev- in which taxes are established by the national government enue-raising capability is clear: it is significantly less than in and each region receives a percentage of that base, and in industrial countries, and the poorer is the country, the less which high payroll taxes are used to finance social pro- is the capacity. This limited capacity has important impli- grams. Third, the results for East Asia and the Pacific and cations for developing countries' ability to finance health Latin America and the Caribbean are somewhat anomalous, and other public services. The different sources of public FIGURE 10 Taxation classification g Income (direct) taxes | Property (direct) taxes Company Personal taxesl taxes Corpoatiand ancome Cogapin Gift Inhentance Death Wealth taxtandsax Poll tax gains taxes taxes duty tax Ad valorem Source: AJlan 1971. INNOVATIONS IN HEALTH CARE FINANCING 16 revenues, the criteria by which they should be evaluated, the financing models for industrial countries may be less the institutional realities of developing countries that cir- relevant to developing countries. The recent literature on cumscribe the use of various revenue-raising modalities, taxation theory and public finance in developing countries and the policy implications for financing health systems in provides important insights into this issue. developing countries are discussed below. Governments have many options for raising (or mobi- lizing through compulsion) revenues: direct and indirect Sources of public revenues taxes, user charges, mandates, grant assistance, and bor- rowing.20 Direct taxes are taxes on individuals, households, The sources of health care financing in developing coun- and firms and include personal income taxes, corporate tries have not received much in-depth treatment in the lit- profits taxes, payroll taxes, social security taxes, property erature. This is unfortunate because the cost, equity, and taxes, and wealth taxes (figure 10). Indirect taxes are taxes administrative implications of raising revenues in develop- on transactions and commodities and include general sales ing countries are different than in industrial countries. Thus taxes, value added taxes, excise taxes, turnover taxes, import Consumpton taes I I l l I Sales taxes |Factor taxes Ttrnwer Sales added Purd se inlds. ditre Stamp and roil Mre estae kactr| IDrnl l ~~~~~~~~~~~~~I I I I _____ ____ ____ ____ ____ ___l l l l l _ _ _ _I__ _ ______ _____ _____ ______ _____ _ l I l I l- - - -- - + ~ ~~~ l- II I I I I~~~~~~~~~~~~~~~ Specik l A CURMUDGEON'S GUIDE TO FINANCING HEALTH CARE IN DEVELOPING COUNTRIES 17 duties, and export taxes. Expenditures can also be financed incidence of a tax (that is, the entity legally responsible for through user charges and taxes on state-owned enterprises. the tax) and the economic incidence of the tax (that is, the Another way to provide services through public interven- change in the distribution of private real income resulting tion is with employer and individual mandates. Charitable from the tax or the entity who ultimately bears the burden contributions (discussed below under private financing) can of the tax). Impact and economic incidence differ through also augment government revenues. Grant assistance in the a process known as tax shifting. Economic incidence is the form of grants to governments (or nongovernment entities) more appropriate measure for assessing the overall effects is another potential source of health revenues. Borrowing of alternative taxes (Rosen 1995, pp. 273-77). from domestic or foreign sources can also be used to finance The basic criteria for evaluating alternative financing public health spending. In evaluating the altemative meth- modalities are: ods used to finance health care systems in developing coun- * Econornic efficiency: the tax system should not interfere tries, the economic, equity, political, and administrative with the efficient allocation of resources. effects of these methods must be assessed. * Equity (fairness): the tax system ought to be fair in its All countries use a combination of these methods to treatment of different individuals. finance their health systems. As discussed below, there are * Administrative simplicity: the tax system ought to be easy significant differences between the methods used in indus- and inexpensive to administer.21 trial and developing countries. Some OECD countries Tax policy often involves tradeoffs among these crite- (Canada, Sweden, the United Kingdom) finance their health ria.22 Each is discussed in turn. systems through general revenues. Others (France, Germany) rely on social security (payroll) taxes. Payroll taxes are also Economic efficiency. Whenever the government uses tax- used in many transition economies and in Latin America. ation to extract resources from the private sector, there is Other countries use multiple methods of financing. The an economic cost that in most cases exceeds the amount of U.S. Medicare program, a national public program for the resources extracted. This additional cost is referred to as elderly and the disabled, is financed with social security the excess burden, efficiency cost, or deadweight loss of taxes, general revenues, and premium payments from taxation. In other words, excess burden results from the enrollees. User charges are a common feature of most pub- production and consumption inefficiencies associated with licly financed programs, especially for goods (like pharma- taxes that distort the decisions made by firms and house- ceuticals) and services where most countries require holds. This excess burden occurs because virtually all taxes consumers to share the costs of covered services (OECD create distortions (substitution effects) in economic deci- 1992 and 1994). President Clinton's proposed reform of sionmaking that reduce incomes and production by more the U.S. health care system was based on a mandate for all than the amount of revenue transferred to the govern- employers to finance coverage for their employees (Zelman ment. Taxes generally distort the decisions firms make about 1994). A variant of this approach proposed in previous production, trade, and investment, and the decisions house- unsuccessful U.S. reform efforts is the "play or pay" holds make about consumption and savings, both currently approach, in which employers would either provide employ- and over time. For example, one study of the U.S. tax sys- ees with health insurance at a level determined by the gov- tem found that a 1 percent increase in all tax rates would ernment or pay an additional tax that would be used to generate efficiency costs of 17-56 percent over the amount finance coverage for employees (Moffet 1993, p. 2). of revenue raised. A study of Sweden's tax system found efficiency costs to be 70-130 percent of the amounts of rev- Taxation enue raised (World Bank 1991, p. 3;Jbnsson and Musgrove in this volume). Taxation is the main source of revenue in nearly every coun- Only taxes that do not change behavior do not result in try. In assessing the equity effects of different taxes, a dis- an excess burden. Such taxes are generally limited to lump- tinction must be made between the impact or statutory sum and poll taxes (that is, a flat tax on each individual), INNOVATIONS IN HEALTEI CARE FINANCING 18 which are hard to administer in developing countries and important criterion for evaluating various revenue-raising perform poorly on equity grounds. Taxes on entities or trans- measures. actions in which price responsiveness (that is, elasticity) is limited or nonexistent (that is, the quantities demanded or Administrative simplicity. Running a tax system imposes supplied are not affected by the change in price resulting significant costs on both government authorities and tax- from the tax) also have lower excess burdens. A tax's excess paying units. But perhaps more important are the signifi- burden is also related exponentially to the tax rate-the cant equity, efficiency, and revenue implications of an higher is the tax rate, the higher is the excess burden.23 Thus inefficiently administered system. Administrative and com- taxes with low rates on entities or commodities with little pliance costs depend on the types of personnel and equip- price responsiveness have lower excess burdens or efficiency ment needed, the types of records kept and information costs to the economy than taxes with high rates on entities needed, the complexity of the tax system (for example, or commodities with considerable price responsiveness. special provisions and exemptions), and rate differentiation Excess burdens are minimized when the ratio of the tax across individuals and categories of income. An inefficiently rates is equal to the ratio of the relative price responsive- administered tax system can generate high administrative ness of the commodities.24 costs, cause losses of tax revenues (through tax avoidance and evasion), increase excess burdens of taxation, and fos- Equity. A second criterion for evaluating taxes is the ter inequity. It is becoming increasingly recognized that suc- system's fairness or equity. This aspect of revenue raising cessful changes in tax policy require an effective tax receives significant attention in the literature on health administration. care financing (van Doorslaer and Wagstaff 1995; van In developing countries, where institutional capacity is Doorslaer, Wagstaff, and Rutten 1993; WHO 1996a). Equity often weak and tax system design overly complex, tax admin- has two dimension: horizontal equity and vertical equity. istration and tax reform are closely related. Simplicity has Horizontal equity means treating taxpayers with the same been advocated as the fundamental principal for effective amount of income equally, irrespective of the source of the taxation. This suggests: income. Vertical equity means treating taxpayers with dif- - Eliminating unproductive taxes. ferent incomes differently or, more specifically, distribut- * Keeping differential tax rates and provisions to a ing the tax burden among taxpayers on the basis of ability minimum. to pay. Defining income and ability to pay is not straight- Drafting clearly and communicating effectively the pro- forward. Although horizontal equityis often consistent with visions of tax laws (Khalilzadeh-Shirazi and Shah 1991, efficient taxes that are simple to administer, such taxes often p. xx). violate vertical equity considerations (see below). These requirements for administrative efficiency have Using the tax system to redistribute income from the rich important implications for economic efficiency and equity. to the poor is difficult because a sizable portion of rich In fact, in designing optimal tax systems, all three criteria people's incomes comes from capital that is internationally must be considered and explicit tradeoffs made. mobile, because much of their income is entrepreneurial The literature defines an optimal tax structure as one and thus is hard to measure and tax, and because the rich that maximizes a country's social welfare by making the are politically powerful (Newbery and Stern 1987, p. 187). explicit tradeoffs between equity and efficiency that best More generally, there has been increasing acceptance of reflect the country's attitudes toward achieving these goals.25 the notion that in developing countries the tax system is an As such there are no universal guidelines for designing an extremelypoor instrument for redistributing income. Income optimal tax system; rather, countries must make explicit redistribution can be better achieved by using the revenues choices among competing efficiency and equity objectives. from efficient taxes to provide subsidies to the poor (World In principle, if all individuals were identical, nondistorting Bank forthcoming). Nevertheless, the economic incidence lump-sum taxes could be imposed, horizontal and vertical of taxes on the distribution of income is one of the most equity would be preserved, and excess burdens would be A CURMUDGEON'S GUIDE TO FINANCING HEALTH CARE IN DEVELOPING COUNTRIES 19 avoided. But since individuals are not identical (and indeed, of the economy (Musgrave 1987, p. 244). Industrial coun- information on their differences can be only imperfectly tries have far more flexibility in designing optimal tax sys- obtained by tax authorities), lump-sum taxes result in fur- tems because, relative to developing countries, they have ther income inequality. Since such taxes are undesirable on higher income levels, more equal income distributions, more equity grounds (and possibly on administrative grounds), "tax handles," less open economies, and more efficient tax governments must use distorting taxes. Thus tax systems administrations. In fact, the institutional features that char- reflect the different attitudes that countries have about acterize developing countries severely limit their ability to equity and efficiency objectives as well as their basic polit- implement the full range of instruments that could be used ical structures and administrative capabilities (Stiglitz 1988, to execute optimal tax policies (box 2). As countries' incomes pp. 479-80; Newbery and Stern 1987, p. 167). rise, these impediments to effective taxation disappear. Institutional structure of taxation in developing countries. Taxation practices in industrial and developing countries. A country's ability to raise revenues, or its tax capacity, As noted, industrial countries raise almost twice as much depends on its per capita income level, the distribution of revenue relative to GDP as developing countries. Tax capac- income, the availability of "tax handles," and the openness ity grows with income, and urbanization facilitates tax col- lection and increases demand for public services. In BOx 2 attempting to deal with the efficiency and equity tradeoffs Institutional features of developing countries inherent in optimal tax theory, industrial countries rely on that affect their tax capacity income taxes, both personal (one of the best instruments for redistribution) and corporate, as their principal sources * Most of the population is seif-employed in subsistence small- of revenue, followed by indirect taxes. Developing coun- scale agriculture, where much of the income is in kind, trans- tries rely on trade (especially import) taxes, indirect taxes, actions are hard to trace, and high rates of illiteracy and poor and to a lesser extent corporate income taxes. Developing accounting and record-keeping limit the use of personal countries are far less reliant on personal income taxes because income or profits taxes. these are not an easily accessible "tax handle" for the rea- * In urban areas there is a large informal sector of small and sons stated above. In fact, in many developing countries transient firms, and even individuals employed in the formal manufacturing sector work in small firms, personal income taxes are essentially wage taxes on gov- * Large firms tend to be government enterprises or extractive ernment employees and employees of large corporations. industries that are often owned by foreigners. In many cases these corporations are foreign owned and * Agricultural products and mineral resources face unstable and engaged in the exportation of mineral resources. As such unpredictable world prices. they generate large profits and are convenient and visible * The dualism of a modern urban sector and a traditional rural entities to tax. Personal income taxes become more impor- sector, and the market segmentation it creates, distorts com- tntias coutrs' incom es gecorpore income moit an,ao akt,icraigtxbres tant as countries' incomes grow, while corporate income modity and labor markets, increasing tax burdens., * High levels of income inequality tend to result in higher tax taxes follow a bell-shaped curve, initially rising in impor- rates, greater tax avoidance, and higher efficiency losses. tance and then falling. * Trade distortions-import tariffs, quotas, export taxes, dif- Domestic taxes on goods and services (such as excise ferential exchange rates, foreign exchange ratoning-abound, and sales taxes) are an important source of revenue in all resulting in resource misallocations and inequity. countries. In developing countries such taxes are the most * The influence of state-owned enterprises, coupled with nonoptimal user charge structures, often results in ineficient ta nt so rceo rene aftererad taxes, wici public investment decisions. taxes playing a larger role than general sales taxes (which * Tax administration capacity is limited. are often imposed with multiple rates and strongly resem- ble excise taxes). Excise taxes on alcohol, cigarettes, and Source: World Bank 1991 and forthcoming, Newbery and Stern 1987; 1 p Khalilzadeh-Shirazi and Shah 1991. petroleum products are particularly important, accounting for almost three-quarters of excise tax revenues. There is INNOVATIONS IN HEALTH CARE FINANCING 20 no relationship between these taxes and growth in a coun- and equity and efficiency concerns can be addressed by try's income. broadening the tax base, keeping tax rates low, limiting Trade taxes, especially import duties, are the most impor- tax rate differentiation, and not discriminating by the tant source of revenue in developing countries. Import duties source of production (foreign or domestic). The best fall sharply as countries' incomes increase (as do export instrument for this is a single-rate value added tax (VAT) duties, which are far less important). that exempts items that are a significant component of Social security taxes are a minor source of revenues in expenditures by the poor. developing countries compared with industrial countries, Equity would also be fostered by introducing luxury where they are several times as important. Moreover, social and excise taxes with just three or four rates on income- security taxes become increasingly important as incomes elastic goods that are not distinguished by source of grow. The basic institutional weaknesses of developing coun- production (foreign or domestic). tries described above preclude widespread use of social Given that the supply of land is inelastic and owners tend security taxes. Sociopolitical factors often lead to their use, to be wealthy, land taxes would score high on both effi- however, as evidenced by the importance of these taxes in ciency and equity grounds. Such taxes generally are not Latin America and the Caribbean and Europe and Central politically feasible, however. Asia. Both industrial and developing countries often ear- Corporate income taxes, especially on mineral extrac- mark such taxes to support health insurance systems. tion, are an important and accessible "tax handle" and Wealth and property taxes account for a small percent- may score high on equity and efficiency grounds, espe- age of tax revenues in industrial and developing countries, cially if the country is a major supplier of these resources. although they increase in importance as income increases But for corporate income taxes more generally, tax rates (Tanzi 1987, pp. 205-36; World Bank 1991, p. 16). that exceed international standards would likely lead to capital flight, retarding economic growrth. Since capital Implicationsfor tax revenue in developing countries. In pur- is more mobile internationally than labor, capital should suing optimal tax policies, industrial countries tend to bal- be taxed less heavily than labor. Corporate income taxes ance equity and efficiency considerations by using broadly with a single statutory rate should be considered. based progressive income taxes and (more or less) propor- * Personal income taxes would likely fall only on govern- tional payroll, corporate profits, and general consumption ment employees and employees of large firms, discour- (sales, value added) taxes with little tax rate differentiation aging saving and limiting their potential for raising and relatively low tax rates. Inequality in the distribution revenue. Moreover, given low formal employment, taxes of income is often handled with cash transfers and subsi- on labor (such as social security taxes) are unlikely to dized services for the poor. Because the supply of labor is raise substantial revenue and may generate excess bur- relatively inelastic, the bases for different commodity taxes dens (depending on the supply elasticity of labor and are broad, and tax rates are relatively low (with little rate whether such taxes encourage inappropriate capital sub- differentiation), these types of taxes generally do not result stitution). Such distortions in production could retard in the large excess burdens found in the tax systems of devel- economic growth. But since personal income taxes score oping countries-where reliance on trade taxes, commod- high on equity grounds, consideration should be given ity taxes, and corporate income taxes with widely to a single-rate personal income tax combined with a differentiated and high tax rates, market segmentation, and generous personal exemption in lieu of the more com- limited administrative capacity result in significant excess mon systems of sharply progressive rates and a variety burdens and inequity. of deductions and credits. Given the institutional weaknesses (and social welfare * Import taxes result in larger excess burdens than gen- functions) in developing countries, which revenue-raising eral consumption taxes. However, they are easy to col- mechanisms best approximate optimal tax policies? lect at the point of entry and to the extent that imports * Commodity taxes are an important source of revenue, are highly income-elastic (that is, responsive) luxury A CURMUDGEON'S GUIDE TO FINANCING HEALTH CARE IN DEVELOPING COUNTRIES 21 goods, such taxes may have favorable equity impacts. Yet cially in the absence of developed financial markets and gov- if imports have a high capital content, import taxes may emient regulation. Similarly, firms may not have the exper- generate significant economic distortions. Countries must tise to self-insure or to purchase insurance. Nevertheless, as carefully weigh these tradeoffs as well as the potential the formal sector, financial markets, and administrative capac- long-run economic development effects of these taxes ity develop, using tax subsidies to encourage employer-based (World Bank 1991, pp. 57-58; Newbery and Stern 1987; insurance may be an effective way to encourage provision of Thirsk forthcoming). insurance that is largely privately financed. There are, however, significant drawbacks to using tax Tax subsidies for purchase of health insurance and medical subsidies in this manner. First, there is a potentially large services. Another tax-based mechanism that warrants fur- revenue loss to the government. Second, depending on how ther discussion, given its widespread use in many coun- the tax subsidy is structured, it may be an open-ended loss tries, is tax subsidization for the purchase of private insurance of revenues based solely on the decisions of private employ- and medical expenses. These subsidies are generally in the ers. Third, the benefit incidence may be highly regressive form of deductions or credits against personal or corporate (with high-wage firms and individuals getting the most ben- income taxes and, in the case of employer-provided health efit). Fourth, such subsidies mute the cost consciousness insurance, are provided by not counting the employer's con- of employers (in terms of the comprehensiveness of the tribution to health insurance premiums (or medical services) insurance that they purchase) and employees (since they as income to the employee. are paying less than the full cost of the services used, by the Subsidizing employers to purchase health insurance (or amount of the subsidy). In other words, such subsidies will to self-insure by paying for or directly providing services) increase health care costs and may encourage coverage of has several advantages. First, when private firms provide health services of marginal benefit. Fifth, employer-financed the insurance or service, the expenditures are not a gov- insurance will result in lower cash wages (see CBO 1994; emmnent budget expense. Second, as discussed above, Hall 1994; Hoff 1996; Arnett 1996). employer groups are a logical "community" to insure because In summary, developing countries need to replace nar- (for large employers) they provide a large enough group to row, distorting tax bases that have high and widely differ- effectively pool risks, employers can be more informed pur- entiated rates and numerous loopholes with broader tax chasers than individuals, and adverse selection tends to be bases that generate higher revenues at lower rates and do mitigated by the fact that employees join the group for not discriminate against the various sources and uses of employment purposes. Moreover, if employers also cover income. Doing so would result in the simultaneous achieve- the employee's dependents (as in the more advanced devel- ment of efficiency, administrative simplicity, and horizon- oping countries), large segments of the population will have tal equity (Thirsk forthcoming). Moreover, effective tax privately provided health insurance. reforms can only take place if they are accompanied by Developing countries also must consider issues of tax admin- reforms in tax administration. istration and industrial structure. If the bulk of the popula- Countries also need to consider using sources of public tion works in subsistence agriculture and the formal urban revenue other than taxes to finance their health systems, manufacturing sector is composed of many small transient including user charges, employer and individual mandates, firms, a significant portion of the population is unlikely to be grant assistance, and borrowing. These financing sources can covered by employer-based insurance. Moreover, if much of be evaluated using the same criteria-economic efficiency, the modem urban sector is working at or near the minimum equity, and administrative simplicity-described above. wage, requiring employers to provide health coverage will increase unemployment. Although tax subsidies through the User charges corporate income tax are generally easy to administer, there isastrongpossibilityoffraudincountrieswithweaktaxadmin- In a traditional public finance context, individuals pay user istration.Thesupplyof insurance alsomaybe aproblem, espe- charges (based on the principle of cost recovery) for a pub- INNOVATIONS IN HEALTH CARE FINANCING 22 lidy provided good or service. Such charges should be used increase taxes. Household surveys in a number of coun- whenever a publicly produced good or service has benefits tries indicate that people are willing to pay for services that that can be assigned to an individual. Taxes are more appro- they deem of benefit to them, making user charges a less priate to finance public goods with collective (rather than coercive way than taxation to raise revenues to finance pub- individual) benefits, to compensate for market failures, and lic services. Moreover, efficiency, equity, and social welfare to achieve distributional objectives (World Bank 1988). can be improved if the additional revenues generated are User charges should be evaluated using the same crite- used to provide public services with the highest social return. ria used to evaluate taxes (economic efficiency, equity, and User charges for publicly financed or provided goods administrative simplicity). Although user charges for pub- and services can take a variety of forms. If the service is pub- licly provided services are a source of government revenue, licly provided health insurance, individuals may be required from a national healh accounts perspective they are tech- to pay premiums or share costs (for example, through nically private expenditures for publicly provided services. deductibles, coinsurance, and copayments). While the health In this context they are no different from direct private insurance is provided publicly, the health services could be out-of-pocket payments for nongovernment services, dis- provided through a public or private institution. Under the cussed below. But since they have implications for public more traditional concept of user charges, individuals are financing and government policy, user charges for publicly required to pay a fee for publicly provided services in a pub- provided services are analyzed in this section. lic facility The different types of user charges relating to User-charges embody the benefit approach of taxation- public health insurance (for example, premiums) add some that individuals who benefit from a service should pay for complexity to the traditional analysis of user charges. Still, it. User charges can be used when public services have largely the basic arguments and analysis are the same as they are private benefits (most personal health services fulfill this for less complex publicly provided services. criteria, although basic public health services do not). User The premiums and cost sharing paid for a publicly financed charges differ in one important aspect from taxes: if set insurance package are sources of public revenue, but they appropriately (that is, so that the private marginal benefit also influence the allocation (use) of resources. Premiums equals the marginal cost of providing the service), user have the virtue of spreading costs (in a true insurance sense charges can generate revenues without any efficiency loss of spreading risks) over the entire population covered by (World Bank 1991, p. 20). Yet user charges set in this way the program. Cost sharing essentially taxes individuals atthe would not be particularly equitable, since the poor and non- point of service use, affects far fewer individuals, and may poor would pay the same rate for services. This outcome be viewed as a tax on the sick. But because it counteracts can be offset by exempting the poor or by basing user charges the moral hazard inherent in health insurance, cost sharing on ability to pay. As in the case of optimal taxes, however, is an important tool for achieving allocative efficiency. For this will result in some efficiency loss. example, the U.S. Medicare program covers hospital and In practice, user charges are often set below marginal inpatient services using social security taxes, but outpatient costs, resulting in excess consumption, overproduction, and services are covered by a voluntary medical insurance pro- inefficiency. As a result some analysts have argued that gram (Medicare SupplementaryMedical Insurance) inwhich user charges should be set at levels above the marginal cost individuals pay a highly subsidized premium equal to just of producing the service-for income-elastic goods (that is, 25 percent of the actuarial costs of this insurance (the rest services for which consumption increases disproportion- of the costs are covered by general government revenues). ately relative to increases in income), and that the profits Cost sharing, both deductibles and coinsurance, is also be used to subsidize services for the poor (Ray 1975). As required at the point of service use to encourage efficient in the previous case, there is a tradeoff between excess use of services under both programs. burden and equity. In designing premiums and cost-sharing arrangements, User charges are an important source of revenue in cases several important equity, efficiency, and administrative con- where it is politically unfeasible to reduce spending or cerns emerge. To ensure equity in most cases the poor A CURMUDGEON'S GUIDE TO FINANCING HEALTH CARE IN DEVELOPING COUNTRIES 23 must be exempt from such charges. To ensure efficiency, insurance experiment conducted by the Rand Corporation copayments must be set high enough to discourage frivo- found that well-designed cost sharing can provide additional lous use, but not so high as to discourage the use of needed revenues and increase efficiency without discouraging nec- or cost-effective services. In fact, some countries have dif- essary service utilization. The evidence from Canada and ferential cost sharing depending on the type of service, so Europe regarding the equity and efficiency effects of user that preventive services generally do not have any cost charges is more mixed, however. Gertler and Hammer (in sharing (and indeed may be provided directly by the gov- this volume) summarize much of the evidence and discuss erninent as a pure public good outside the health insur- the links between user charges and social insurance. ance system), while less important or more discretionary Evidence on user charges in developing countries is exten- services may require high levels of cost sharing. sive and varied. The findings indicate: Another aspect of premiums and cost sharing relates to * A strong utilization response to user charges, with dis- the need to exempt individuals with catastrophic medical proportionately larger reductions in utilization on the expenses. In industrial and some developing countries 10 part of the poor. percent of the population may incur 70 percent of health - Exemptions from user charges often benefit high-income care costs. Exempting these people on equity grounds results groups. in a substantial loss of potential revenues. This adjustment * User charges need to be considered in the context of also violates the efficiency rule that user charges should be time costs and managerial and administrative capacity set equal to marginal costs (a rule more easily applied to * Although user charges can be used to supplement pub- more traditional publicly provided services, such as water lic revenues for curative care at the facility level, total and electricity). revenue generation has not met expectations (being on Administrative issues are also important. If the costs of the order of 5 percent of total revenues). collection exceed the user charge (unless there are signifi- * Effects on health status and efficiency are less clear, cant allocational savings), revenues will actually decline. although there is some evidence of adverse effects on Another administrative compliance issue concerns who gets health outcomes. to keep the additional revenue. In a number of developing * People appear more willing to pay for tangible benefits countries, unless the health facility or provider gets to keep (such as drugs). some or all of the additional revenue, they have few incen- * User charges for public services have elicited higher prices tives to collect the cost sharing. Furthermore, as long as for private services, resulting in declines in total utiliza- the facilities and providers use these revenues for quality tion (see Gerder and Hammer in this volume; Wang'ombe improvements or service enhancements, individuals will in this volume; Gertler and van der Gaag 1990; Creese continue to use these facilities. and Kutzin 1995). Another issue concerns policymakers relying on user In other words, user charges can increase efficiency and charges to avoid considering altemative financing approaches ensure equity if the systems are designed carefully and that might generate larger revenues more easily and equi- generate tangible benefits for individuals. But time costs, tably. For example, social insurance systems might have a administrative capacity, and impacts on private pricing poli- much larger impact and longer-lasting ability to finance cies must be analyzed carefully. Moreover, while revenue underfunded public systems and provide the government generation can have important effects at the facility level, with an opportunity to promote efficiency by creating incen- overall revenue generation may be lower than original tives for better service delivery (Lewis 1993). expectations. Many industrial and developing countries have extensive experience with various types of cost sharing-both formal Mandates user charges and direct out-of-pocket payments for private services-including controlled experiments in the United When fiscal constraints are tight and market failures pre- States and China. In the United States a $70 million health dude public provision of a particular benefit-such as health INNOVATIONS IN HEALTII CARE FINANCING 24 insurance-government mandates to require provision of wages, and impose lower direct costs on businesses. But such benefits, either by employers or by individuals, can such mandates maybe even more regressive than mandates help government achieve its policy goals.26 Employer and on employers; they provide incentives for employers who individual mandates can be evaluated using the same cri- currently provide health insurance coverage to drop it and, teria used to evaluate alternative revenue-raising sources. unless accompanied by subsidies to the poor to offset the In fact, under certain conditions mandates are more effi- equity problem, impose high marginal tax rates and notch cient than taxes, though they tend to be less equitable. effects if subsidies are phased out as an individual's income Employer mandates can be more efficient than public increases. Moreover, other issues unique to health insur- provision financed by inefficient taxes in several ways. The ance markets arise, such as the higher premiums charged costs of the mandated benefit will be treated by the employer for individual policies relative to group policies (Committee as an additional labor cost and will be shifted back on to on Energy and Commerce, House of Representatives and employees in the form of lower wages, discretionary bene- Special Committee on Aging, U.S. Senate 1989). fits, employment, or future wage increases. If employees Mandates can be an efficient means of correcting for the increase their labor supply to offset the costs of the bene- failure of private markets to provide health insurance with- fit, the increase in labor supply will reduce the deadweight out increasing public expenditures and introducing the loss. If the increase in labor supply fully offsets the costs of distortive taxes needed to finance such increased expendi- the benefit (that is, there is full valuation of the benefit by tures. But equity considerations and the effects on work- employees), then there is no excess burden and the employer ers and firms must be carefully considered in light of possible mandate is a more efficient means of providing the bene- labor market rigidities that can result in adverse employ- fit then taxation (Cutler and Madrian 1996). In effect, man- ment and production effects. And mandate enforcement, dates are a benefit tax. a difficult task in industrial countries, would be even harder There may, however, be institutional constraints that pre- in developing countries. Enforcement mechanisms such as clude this employment response and wage or discretionary withholding welfare checks, drivers licenses, or school enroll- benefit adjustment from happening, including antidis- ment would not be effective incentives in developing coun- crirnination rules, workplace rules, union rules, minimum tries because most developing countries have no such wage laws, and so on. Thus even if everyone in the employ- programs or lack the administrative mechanisms needed ment group fully values the benefit, these rigidities will cause for enforcement. efficiency losses since the full wage adjustment will be inhib- ited. Firms unable to reduce compensation would close, Grant assistance and those in noncompetitive markets would shift costs to consumers. High-cost (elderly) and low-wage workers could Grants from foreign donors (borrowing is discussed below) lose their jobs. In fact, in certain cases these efficiency are a major source of health care financing and of total costs can exceed those that would occur under a payroll expenditures in low- and some middle-income countries. tax falling on all workers to finance the benefit (Gruber In 1994 official development assistance (ODA) was more 1994). than $47 billion, or about 1 percent of developing country The main problem with mandates relates to equity, since GDp.27 In Africa (excluding South Africa) donor assis- the valuation of the benefit and the subsequent wage adjust- tance accounts for an average of almost 20 percent of health ments generally bear no relationship to ability to pay. spending, and in several countries for more than 50 per- Mandates are a benefit tax, not an ability to pay tax. Indeed, cent (World Bank 1993). These revenues cannot be read- in a world of rigidities high-cost and elderly workers would ily classified as public or private. If they are tied to public likely suffer the most, making the mandate a regressive activities, the revenues essentially augment government approach to financing health insurance benefits. efforts. If they go to nongovernmental organizations (NGOs) Mandates can also be imposed on individuals. Mandates or directly to individuals, they augment private financing. on individuals do not have any adverse effects on jobs or Donors are concerned about whether such assistance A CURMUDGEON'S GUIDE TO FINANCING HEALTII CARE IN DEVELOPING COUNTRIES 25 increases net expenditures in the targeted sectors or sim- investments, overall growth and future well-being may be ply substitutes for government spending that may then be reduced. More generally, public investments financed by used for unproductive sectors, such as military spending domestic borrowing do not contribute to capital formation (Feyzioglu, Swaroop, and Zhu 1996). if they simply divert these funds from private investment Donors typically finance large shares of both capital opportunities. In either case the loan will have to be paid and recurrent expenditures. Donors should both coordi- back with interest out of future production.28 nate their efforts and get their priorities right, especially One essential difference between domestic and foreign since such activities have important implications for pub- borrowing is that foreign borrowing does not require that lic and private financing and for current and future expen- current consumption be reduced. Moreover, while the net ditures and revenue needs (World Bank 1993, p. 167). gain to future generations is lower than it would have been Donors should target their assistance to high-priority areas, had government spending been financed out of current and with better coordination could avoid fragmentation and tax revenues, the cutbacks in future consumption needed conflicting impacts from their assistance. to pay off and service the loan come out of a higher level Grant assistance is subject to the budgetary situations of-income (assuming wise investment choices) because and political agendas of donors. As a result, while grant assis- foreign borrowing does not impose any immediate oppor- tance is an important financing source in the short run for tunity costs (Musgrave and Musgrave 1976). many countries to purchase essential health services and Borrowing, like other potential forms of revenue raising, develop critical infrastructure capacity, it is not a reliable requires careful consideration of intertemporal tradeoffs. long-run source of financing. Thus it cannot be relied on Borrowing imposes a burden on future generations and so to ensure long-term financial sustainability. must be carefully considered, in the context of economic * growth and development objectives. Given the poor bal- Borrowing ance of payments situations in many low-income countries as well as higher-priority development needs, borrowing for Borrqwing, like grant assistance, can be either a public or social programs, except where such borrowing is on a con- private financing mechanism. When the government is the cessional basis, is likely to be quite limited. Moreover, con- borrower, it is a public revenue source. When a private entity cessional borrowing from itnternational organizations (such is the borrower, it is a private revenue source. Funds can as the World Bank) generally has policy conditions attached. be borrowed from either domestic or foreign sources. Foreign sources of health financing include international develop- Public financing implications for developing ment organizations, bilateral donor assistance agencies, pri- countries vate commercial institutions, and foreign medical suppliers. Such lending may include a significant grant (donor) ele- Governments can use their revenue-raising and regulatory ment or may be at full market rates. The essential feature powers to publicly finance or to require private financing of borrowing is that the funds eventually must be repaid. of health insurance and health services. Yet the government's Borrowing, in effect, imposes a burden on future generations. ability to raise revenues and the private sector's ability to Domestic and foreign financing have different implica- complywithgovernment mandates and user charge require- tions. Domestic borrowing has a clear opportunity cost- ments are inversely related to country incomes. Economic namely, the opportunity cost of the alternative uses of those bases, institutional structures, and administrative capacities funds. If, in the absence of borrowing, those funds would stronglyinfluence public revenue-raisingpotential andinstru- have been used for domestic consumption, while the bor- ments. In developing countries several issues merit partic- rowed funds are used for capital investment, then economic ular attention: growth and the future well-being may improve. But if the * Taxes and other public revenue sources must be evalu- borrowed funds are used for low-priority or inefficient recur- ated in terms of economic efficiency, equity, and admin- rent health expenditures rather than productive capital istrative simplicity. INNOVATIONS IN HEALTH CARE FINANCING 26 * The inherently agrarian/rural nature of much economic Private Financing Sources activity, the small and transient nature of urban formal sector firms, and the openness of their economies gen- This section discusses private sources of revenue for financ- erally preclude heavy reliance on the more efficient and ing health care, focusing on private insurance, out-of-pocket progressive income taxes used in most industrial countries. payments for direct purchase of medical services, and char- * Limited taxable capacity and inefficient administration itable contributions. Borrowing and grant assistance, which limit revenue generation potential. can be treated as either a public or private revenue source * Broadly based taxes with limited rate differentiation and depending on the entity receiving the funds, were discussed few exemptions, deductions, and exclusions applied to in the previous section. commodities, as well as entities that display little respon- siveness to the tax, are preferable for economic efficiency Private health insurance and equity * A broadly based value added tax with limited rate dif- The nonpoor may have several reasons to prefer using pri- ferentiation, exemptions for the poor, and nondiscrim- vate health insurance to finance personal health services. ination between domestically produced and imported First, by pooling risks, overall risks are reduced. Second, goods ranks high in terms of equity, economic efficiency, consumer sovereignty in choosing an insurance package that and administrative feasibility. best fits that individual's preferences maximizes welfare. * Public financing of personal health services pools risks, Third, by relying on private markets rather than government eliminates adverse selection, and ensures equity but coercion, the benefits of risk pooling and consumer choice reduces choice and imposes excess burdens. can be achieved while the efficiency costs of taxation are * Since developing countries can only generate about two- avoided. Fourth, private insurers can negotiate with providers thirds the revenue generated in industrial countries (rel- over cost and quality more effectively than individual con- ative to GDP), in the absence of regulated private sumers. insurance markets developing countries could improve On the other hand, adverse selection by individuals and welfare by organizing these markets so that people who risk selection by insurers may preclude effective and equi- are able to pay can purchase health insurance at an actu- table risk pooling. Moreover, large numbers of insurers arially fair price. and inefficient insurance operations may result in high admin- * User charges can be used to enhance public revenues istrative costs (such as marketing costs), and individual insur- where the services in question have individual (rather ers may be too small to effectively negotiate with medical than collective) benefits. But such charges need to be care providers over prices. If this latter situation is the carefully structured to balance equity, efficiency, and rev- case, insurers will only be able to compete on their load enue generation objectives. factor, which tends to be a small percentage of the total Developing countries will continue to rely on public premium (in the United States, for example, the average and private revenues to finance their health care systems. load factor for large group policies is about 10 percent; The instruments used will affect economic efficiency, the Enthoven in this volume). distribution of income, and the revenues generated. Limited A number of regulations and innovative approaches to administrative capacity and infrastructure hinder imple- risk pooling have been implemented to dealwith these prob- mentation of effective policies. Dealing with these imped- lems. Examples include the insurance reforms undertaken iments to fiscal and health financing reform should be a by many U.S. states, managed competition as a mecha- priority; otherwise the effects of economically and politi- nism for organizing insurance markets, and informal (gen- cally "correct" policy prescriptions will be marginal. erally rural) risk-pooling schemes being implemented in Moreover, implicitly underlying these technical and admin- developing countries. istrative concerns is the need for a stable and sustainable Each U.S. state regulates private health insurance com- political environment. panies.29 Such regulations generally apply to solvency, mar- A CURMUDGEON'S GUIDE TO FINANCING HEALTH CARE IN DEVELOPING COUNTRIES 27 keting, information disclosure, premium levels, reinsurance, requirements are designed to offset favorable risk selec- and so on (box 3). Although such regulations are needed to tion by insurers (guaranteed issue, renewability, continuity correct forthe market failures inherent in insurance markets, of coverage), redress informational gaps for consumers they also raise the costs of private health insurance policies (information disclosure, standardized benefits), and pro- (because of, for example, mandated benefits; GAO 1996). mote community rating (rating band restrictions). To ensure better functioning of state insurance mar- kets, forty-four U.S. states have recently augmented their Managed competition. Managed competition is a recent insurance regulations, and some states (such as New York) innovation that organizes public and private insurance mar- now require pure community rating (box 4). These new kets, fosters competition among insurers, and addresses BOX 3 U.S. states and insurance regulation Market conduct requirements Plan benefit coverage and description States review and approve insurance policies to ensure that they are not vague or misleading and meet state requirements (such as mandatory benefit provisions). Small-group reforms Most states require insurers selling to small employers to accept and renew employ- ees who want health insurance coverage, establish short waiting periods for preexist- ing conditions, and require portability of coverage when an individual changes jobs or insurers. Consumer protections and complaints States monitor insurers' actions to ensure that they are not engaging in unfair busi- ness practices or otherwise taking advantage of consumers. They also assist con- sumers by investigating complaints, answering questions. and conducting educational programs. Financiol requirements Licensing States license insurance companies and the agents who sell insurance to ensure that companies are financially sound and reputable and that agents are qualified. Financial solvency States set standards for and monitor the financial operations of insurers to determine whether they have adequate reserves to pay policyholders' claims. States restrict how insurers invest their funds. Rate reviews States review and approve rates or require actuarial certification to ensure that rates are reasonable for consumers and sufficient to maintain the solvency of insurance companies. Some states regulate insurer rating practices in the small-group market to determine the factors insurers can use in setting premiums. Tax requirements Premium taxes Nearly all states assess premium taxes on insurers. Guaranty funds States require insurers to finance guaranty funds that provide financial protection to enrollees who have outstanding medical claims in case of insurer insolvency. High-risk pools Some states require insurers to finance losses in high-risk pools that provide health coverage for individuals who had otherwise been denied coverage because of a med- ical condition. Source: GAO 1996, pp. 5-6. INNOVATIONS IN HEALTH CARE FINANCING 28 several aspects of insurance market failure. The basic goal vidual savings accounts from which individuals pay for health of managed competition is to organize insurance markets care, coupled with a backup financing mechanism (along so that individuals can make informed choices about their the lines of a catastrophic insurance policy). The source of purchase of insurance while reducing moral hazard by mak- the savings account and the backup financing can be pub- ing them financially responsible for the consequences of lic or private. Like direct out-of-pocket payments (includ- their actions. Moreover, many of the inappropriate risk selec- ing user charges) and managed competition, medical savings tion practices of private insurers can be mitigated by orga- accounts provide consumers with strong incentives to be nizing insurance through purchasing cooperatives subject cost conscious. Such accounts also preserve freedom of to rules such as standard benefit packages, coordinated choice of medical care provider and offset the moral haz- open enrollment periods, guaranteed issue, guaranteed ard implicit in insurance arrangements. In addition, med- renewability, limits on the number of rating bands, infor- ical savings accounts may help achieve basic development mation disclosure on plan performance, and marketing objectives by encouraging domestic savings. through the purchasing cooperative (not the insurance plan). There are at least two basic models of medical savings By eliminating (or precluding) tax subsidies for the pur- accounts: the system used in Singapore, which supplements chase of insurance, subscribers are more cost-conscious, other publicly funded health programs, and the models and moral hazard is reduced. Purchasing cooperatives-are used in the United States, which cover a much broader essentially a supermarket in which informed and cost-con- range of services. Both models are backed by a public or scious consumers compare prices and services before buy- private insurance mechanism that covers catastrophic costs ing health insurance. This approach also creates strong above some threshold deductible if the savings account is incentives for insurers to be efficient purchasers of services exhausted. To encourage cost consciousness, this deductible from medical care providers (Enthoven in this volume). must be significant-perhaps 5-10 percent of family For managed competition to work effectively and equi- income. tably, the revenues that finance the system (from whatever One of the biggest advantages of a medical savings source) should be pooled in a fund that distributes them account is that individuals have a strong incentive to be pru- to the insurers who chose to participate (through the pur- dent consumers of medical services, since they can use any chasing cooperative) on a risk-adjusted capitation basis. unspent funds in a variety of ways (Nichols, Prescott, and Risk-adjusted capitation payments are essential to ensure Phua in this volume). In other words, the benefits from that insurers will enroll individuals with greater health risks. being a prudent consumer accrue to the individual-not Developing operational risk adjustment tools is far from simple (Enthoven in this volume). The California Public BOX 4 Employees Health Plan, the U.S. Government's Federal Recent U.S. reforms in state insurance Employees Health Benefits Plan, health plans in the U.S. regulation states of Connecticut, Florida, Iowa, Kentucky, Minnesota, * Guaranteed issue (everybody must be offered a policy). and Washington, U.S. President Clinton's failed health care * Guaranteed renewability except for fraud, nonpayment of reform initiative, the now-suspended Dutch health reform, premiums, and the like. and the recently established Russian health insurance sys- C continuity of coverage provisions, including insurer limita- tem all rely on managed competition organized by a pub- tons on preexisting condition exclusions and waiting periods. lic entit with participation by pu*hc or private insurers andRestrictions limiting the rating factors that can be used to lrovicenti witheparticipation byis polubl rpan n determine premium rates (for example, limits on the num- ber of rating bands based on age or limits on premium dif- ferentials across groups). Medical savings accounts. Medical savings accounts also * Full disclosure of information. provide incentives for individuals to be effective purchasers * Standardized plan benefits. of health services (see Nichols, Prescott, and Phua in this Source: Institute for Health Policy Solutions 1995. volume). Simply put, medical savings accounts are indi- I A CURMUDGEON'S GUIDE TO FINANCING HEALTH CARE IN DEVELOPING COUNTRIES 29 the government, employer, or private insurance company ical evidence regarding direct out-of-pocket payments were (or its subscribers, collectively). Unspent savings can be discussed above.) used for nonmedical consumption, passed on as a cash trans- Much of the discussion about health insurance focuses fer at the time of death to the deceased's heirs, or rolled on how such insurance can cover most basic personal health over from year to year and used for future medical expenses services needs. But insurance policies are just as relevant (so that in a sense individuals are pooling noncatastrophic in other contexts-such as covering dread diseases, pro- risks over their life cycle). Although there have been few viding supplemental benefits to individuals covered under rigorous evaluations of the experiences of U.S. corporations a public system, and covering long-term care and social that have switched to medical savings accounts, recently services that are not covered by standard health insurance passed U.S. health insurance reforms provide tax subsidies policies or public systems. These are important issues in to individuals establishing such accounts, making it likely middle- and high-income countries. Although a detailed dis- that their use will increase. cussion of these issues is beyond the scope of this paper, such insurance needs to be evaluated using the same crite- Informal risk-pooling arrangements. Informal, generally ria (economic efficiency, equity, and administrative sim- rural, and usually voluntary risk-pooling schemes are one plicity) defined earlier. variant of the health insurance model common in devel- Furthermore, it is important to analyze the effects these oping countries, particularly-in Asia and Africa (see Creese supplemental policies have on the insurance policies or and Bennett in this volume). These schemes generally systems that they supplement. For example, if a supple- encourage prepayment of individual premiums into an iden- mental policy covers the cost-sharing requirements of the tifiable fund, provide some notion of entitlement to bene- primary policy, then individuals will use more of the basic fits, and work with a defined set of service providers. The services covered by the primary policy Thus several coun- schemes are usually organized by and often linked to gov- tries' insurance regulations forbid private insurers from emnment providers, although in some cases they are run by filling in for cost sharing in public programs. Other issues community health committees. Zaire's Community Health of concern include high administrative costs and deceptive Insurance Scheme, Thailand's Health Card, and the United marketing practices, which are especially problematic when Nations Children's Fund's (UNICEF) Bamako Initiative policies are marketed to the elderly, who tend to overin- (in more than thirty countries) are examples of such arrange- sure (Chollet and Lewis in this volume). ments (see La Forgia and Griffin 1993; Khoman in this Much of the above discussion has focused on the con- volume; Tibouti). Creese and Bennett (in this volume) ceptual and operational bases for health insurance and describe the underlying conceptual bases for these schemes on policy prescriptions for industrial countries. Developing and experiences in developing countries. countries, however, face additional challenges. It is diffi- cult to provide private insurance in the absence of well- Out-of-pocket payments for health services developed financial markets. Operating private insurance is complex, requiring effective management information As noted above, consumer out-of-pocket payments for health and accounting systems as well as mechanisms to deter services are private expenditures. But such expenditures fraud and abuse. Reinsurance mechanisms and consumer can take a variety of forms, including direct purchase of protections are needed. Private health insurance needs to private services, direct purchase of publicly provided ser- be regulated by government, yet such regulations are com- vices (for example, public user charges), and cost sharing plex and may be beyond the capacity of governments in for publicly or privately financed services. When such pay- many developing countries. There are also potentially high ments are in the form of user charges for publicly provided administrative costs associated with insurance, potentially services, they are a public revenue source. When they are diverting money from health services into health admin- direct payments for privately provided services, they are a istration. All these factors must be given careful private revenue source. (The conceptual issues and empir- consideration. INNOVATIONS IN HEALTH CARE FINANCING 30 Charitable contributions * Public regulation of private health insurance is essential to ensure effective risk pooling, affordability, solvency, Because charitable contributions can be directed- to pub- informed choice, and continuity of coverage. lic or private institutions, they can be a public or private In the absence of a private insurance market, govern- revenue source. In industrial countries such contribu- ments in developitg countries may want to consider tions are sometimes encouraged by the tax system (that establishing a publicly organized (or quasi-public) but is, such contributions are tax deductible), but they are gen- privately financed insurance system. erally a small portion of total health spending. Such con- * Governments need to provide the regulatory framework tributions are generally beyond the control of policymakers, to ensure, quality, efficient pricing, and relevant infor- and cannot be relied on as a stable and long-term source mation concerning both health insurance and health ser- of financing. vice provision and consumption. Moreover, there are virtually no data on the importance Reliance on private financing is a necessity in many of this source of financing for developing countries. low-income countries: As a result governments' abilities Charitable contributions can be domestic or foreign. Foreign to deal with equity and redistribution objectives are some- charitable contributions are the same as foreign grant assis- what limited. Still, for people who are able to pay, gov- tance (discussed above) that has few or no policy condi- emient regulation of insurance and public provision of tionalities attached. There are no efficiency or equity costs privately financed health insurance can result in more in obtaining these funds. Domestic charitable contributions equitable and efficient insurance markets and in gains do have an opportunity cost-the alernative domestic uses from risk pooling. Moreover, effective oversight of the of the funds raised by the charity. If such contributions are entire delivery system (public and private) can result in encouraged by the tax system, there will be efficiency and higher quality; greater efficiency, and better value for perhaps equity costs to the economy. money for both publicly and privately financed services (insurance and out-of-pocket). The difficult questions Private financing implications for developing are: Given the limited institutional capabilities of many countries developing countries, is effective regulation of the insur- ance industry and private providers possible? Can effec- Although private finance cannot address equity and public tive insurance markets be developed given the state of health issues, it is generally essential given governments' these countries' financial markets? And can administra- limited ability to marshal sufficient resources to finance tive costs be kept within reasonable limits? These impor- (directly or through transfers) most personal health services. tant questions need much more attention from the Yet given the market failures affecting both the demand and countries themselves as well as from development agen- supply sides of the market for health services, efficiency in cies and donor organizations. health services consumption and provision is unlikely to be achieved through consumer sovereignty and a laissez-faire Regional Perspectives on Health Financing approach to the health services market. This is especially Reforrn true in developing countries, where the bulk of health spend- ing, especially in the poorest countries, is private. Putting This section summarizes the main challenges confronting aside issues of equity, if heavy reliance on private spending each region's health care financing and delivery systems and is necessary for financial reasons or desirable on efficiency proposes essential reforms. These observations are based grounds, governments should implement regulations that on the epidemiological and economic information presented offset health market failures. earlier as well as on the characteristics of the countries and Several government interventions in the private market health systems in each region. Many of the health financ- for health services can improve the efficiency and equity of ing challenges facing developing regions can also be ana- private financing: lyzed from an income perspective (box 5). A CURMUDGEON'S GUIDE TO FINANCING HEALTH CARE IN DEVELOPING COUNTRIES 31 Europe and Central Asia countries in the region. As a result health outcomes have declined, and in some countries diseases such as diphtheria The formerly socialist economies of Europe and Central and tuberculosis (both under control before the transition) Asia face major challenges as they move away from centrally are on the rise. Adult health is also a serious problem. planned health care systems toward systems based on social Overnutrition, substance abuse (alcohol, tobacco), and vio- insurance models (see World Bank 1996b; Goldstein and lence are major public health concerns. Although under cen- others 1996; and Klugman and Schieber 1996). The region tral planning Europe and Central Asia managed to achieve contains about 9 percent of the world's population and universal access in all countries-the only region to do so- accounts for 3 percent of its income. Among developing these systems relied on large numbers of low-quality inputs. regions Europe and Central Asia has the highest public share Moreover, central planning provided few incentives for of health spending-more than 70 percent. Moreover, health efficiency. And clinical standards are dated, resulting in systems account for 7.2 percent of GDP, or $154 per capita, inefficient and ineffective treatments for many conditions. the highest regional per capita expenditure after Latin The challenge for the region is to maintain universal America and the Caribbean. access, increase efficiency, reduce adult mortality, and Yet although Europe and Central Asia is within the mid- improve quality. Over-resourced delivery systems need to dle-income range, with a per capita GDP of about $1,800, be reconfigured. Clinical practices must be updated. Health the transition to market-based economies and the breakup promotion and disease prevention programs should be imple- of the Soviet Union have brought economic crisis to many mented. Sustainable and efficient financing arrangements BOX 5 Trends in health care financing at different income levels Low-income countries (per capita health expenditure of less that rely on general revenues. There is usually rapid growth in pri- than $80, with health sectors accounting for 3-4 percent of GDP). vate spending as modern private health services expand; private Govemment budgets finance most public health care expenditures, insurance becomes more common, although it (as well as private but are limited by narrow tax bases and weak collection capacity. providers) remain largely unregulated. Expansion in formal employ- Although public spending is sometimes supplemented by user ment, capital markets, and financial sectors, improved institutional fees, these account for a small percentage of public revenues. contexts for formal insurance, and urbanization and other social External assistance continues to be a significant source of revenue, changes help households and enterprises make the transition from especially in Sub-Saharan Africa, South Asia, and the Pacific. Apart informal to formal risk-pooling mechanisms. But countries in this fromthegovemmentbudget, optionsforformal insuranceschemes group still fall short of universal coverage. Rural-urban disparities (public or private) are limited by the small size ofthe formal employ- and the persistence of poverty contribute to these problems, and ment sector, limited savings, underdeveloped financial sectors, the multiplicityoffinancingsources, oftenwithoverlappingorincon- and weak institutions. Instead, households rely on informal arrange- sistent policies, adds to inefficiencies and inequities. Key issues: ments (extended families, traditional community support systems, Developing consistent social insurance schemes, expanding cover- NGOs, charitable organizations, rural cooperatives) to provide pro- age to rural and informal sectors, and regulating the private sector. tection in the event of catastrophic illnesses. Key issues: Developing informal risk-pooling mechanisms (such as NGO or community High-income countries (per capita health expenditure of more schemes) to expand coverage to the poor and making better use than $400, with health sectors accounting for 6-15 percent of of extemal assistance. GDP). Except for the United States, all industrial countries have achieved universal coverage, largely through public financing (whether Middle-income countries (per capita health expenditure of publicly managed or publicly mandated). Also with the exception $80-400, with health sectors accounting for 5-6 percent of GDP). of the United States, private insurance is used mainly to supple- Middle-income countries rely on more financing sources than low- ment the core services covered by public financing. The Republic income countries, including social security schemes for civil ser- of Korea and a number of newly industrialized countries also have vants and other groups of formally employed workers. Although attained, or are close to attaining, universal coverage. Key issues: some countries preferto expand coverage using social security sys- Containing costs, dealingwithaging populations, and ensuring qual- tems, others are moving toward national health service models ity of service and patient satisfaction. INNOVATIONS IN HEALTH CARE FINANCING 32 must be put in place, and the adverse economic conse- their financing systems to increase access, generally by quences of high payroll taxes to support social programs expanding social insurance systems and by encouraging pri- (including health insurance funds) must be mitigated. Service vate financing. In some countries these efforts are leading provision should be separated from finance, and incentive- to two-tiered systems of care. Fiscal sustainability is also a based provider payment systems should be introduced. The concern, especially since health and other social insurance ongoing transition from public finance and provision to pub- funds (such as pensions) are often commingled without spe- lic and private finance and provision must be carefully mon- cific earmarks. A number of countries have introduced itored. With real per capita GDP projected to grow at 3.7 broadlybased reforms to address these issues. With real per percent a year over the next ten years, the need for new gov- capita GDP projected to grow at 2.2 percent a year over ernance structures and human development, rather than the next decade, the fourth highest of the six developing financial constraints, may be the biggest challenge to effec- regions, many slow-growth countries in the region will likely tive reforms. face significant financial constraints in improving access and quality Latin America and the Caribbean Middle East and North Africa Health systems in Latin America and the Caribbean are moving toward universal access, with mixed public and The Middle East and North Africa region is extremely diverse private provision and financing (see Burki 1995; World Bank in terms of economic development, political orientation, 1996c; and Medici and others in this volume). The region, and social conditions (see World Bank 1996d). The region which contains 8 percent of the world's population and contains 5 percent of the world's population and accounts accounts for 6 percent of its income, has a per capita income for 2 percent of its income. At $2,700, the region's average of some $3,100-the highest among the six developing per capita income is second only to that in Latin America regions. Health systems account for an average of 6 percent and the Caribbean. The public sector accounts for half of of GDP, or $200 per capita, the highest per capita expen- health expenditures, the fourth-highest share among the six diture of any developing region. developing regions. Health expenditures account for 5.2 At 49 percent, Latin America and the Caribbean ranks percent of GDP, or about $120 per capita, the third high- fifth among developing regions in terms of the public sec- est among developing regions. tor's share of health spending. Health care is financed by Non-oil and oil-producing countries as well as socialist, ministries of health, with social insurance funds playing a nonsocialist, and statist health systems have very different prominent role, although out-of-pocket payments and pri- approaches. There is considerable political instability in vate insurance are also major sources of financing. Most the region as a result of interregional rivalries. High adult countries in the region need to work harder to ensure coor- illiteracy (particularly among women) and poor women's dination between ministries of health and social insurance health indicators are important concerns, especially in the funds on the financing and provision of services, and achieve lower-income countries. Most countries in the region face a balance in the financing and delivery of preventive and dual disease burdens-noncommunicable and communi- curative services. Nutrition, reproductive health, and com- cable. Given the region's high levels of fertility and popu- municable diseases are still major problems in some of the lation growth, reproductive health continues to be a major poorer countries. Still, injuries and noncommunicable dis- issue in many countries. The public-private mix in financ- eases account for most of the burden of illness-and receive ing and provision is important in the lower- and middle- most of the attention of policymakers-in nearly every coun- income countries, while the wealthier oil-exporting countries try. In the larger countries there are significant urban-rural tend to have national health service systems. Social insur- differentials in access and health outcomes. ance is not widespread and private health insurance plays Universal access, quality, and efficiency are major con- a small, albeit increasing, role in some countries. The qual- cerns in most countries. Many countries are reforming ity of care in the public sector is often a problem. A CURMUDGEON'S GUIDE TO FINANCING HEALTH CARE IN DEVELOPING COUNTRIES 33 Like the region, countries' health care reform agendas regional and local levels are weak. Traditional medicine is are diverse, though there are commonalties. The poorer a significant part of the health system. Clinical skills are countries still face the unfinished public health agenda of often limited. Most countries face dual disease burdens communicable diseases, with a focus on women's health (communicable and noncommunicable) but have been slow and education. All countries are experiencing substantial to react. Access, universal coverage, efficiency, quality, and increases in their burden of noncommunicable disease. the public-private mix are major concerns in most countries. Efficiency and quality are problems in most countries. Needed reforms in the region span virtually every major Unfettered and uncoordinated growth in private health care health services issue, including the unfinished public health systems is increasing costs and resulting in two-tiered sys- agenda, access, quality, costs, financing, and the public-private tems of care. Appropriately dealing with the public-private mix. Rebuilding the shattered health systems in Southeast mix in financing and delivery should be a priority. More Asia, completing the unfinished health agenda in the poorer equitable risk pooling and financing of public systems is also countries and rural areas, and implementing insurance important in most of the region's lower- and middle-income reforms and effective risk pooling are important issues for countries. With the end of the oil boom in the mid-1980s the region. Given the emergingprivate market and the empha- and the globalization of the world economy, many Middle sis placed on individual and family responsibility for health Eastern and North African countries are facing serious care in many countries, particular attention needs to be economic constraints. With real annual growth in per capita paid to the public-private mix in delivery and financing and GDP projected at just 0.4 percent over the next decade, to user charges and cost recovery. Given the region's prag- the lowest of the six developing regions, increasing effi- matic governments and the high rates of annual growth in ciency in financing and delivery will be essential to prevent real per capita GDP projected over the next decade-aver- deterioration in access and quality. aging 6.8 percent, the highest among developing regions- there are major opportunities for basic health reforms. It will East Asia and the Pacific be particularly important to ensure that these systems absorb the lessons learned in other countries (such as in the Republic The economies in East Asia and the Pacific are extremely of Korea, with its serious cost-containment problem) as their diverse, accounting for some of the world's largest (China, incomes rise and their systems are reformed. Ensuring uni- Indonesia, Philippines) and smallest economies. The region versal access with an appropriate public-private mix in financ- contains 30 percent of the world's population and accounts ing and delivery in an expanding economy creates major risks for 4 percent of its income; per capita income, at about for cost escalation as well as for two-tiered systems of care. $1,200, is the fourth highest among developing regions. The Pooling risks through private and social insurance, medical public sector accounts for just over half of health expendi- savings accounts, and other approaches to national health tures, the third-highest share among developing regions. services will be essential as access imnproves. Health expenditures account for 4.1 percent of GDP, or almost $40 per capita, making it (along with Sub-Saharan South Asia Africa) the second-lowest health expenditure region. The region has experienced rapid and sustained eco- South Asia also contains some of the world's largest coun- nomic growth for manyyears, with most countries approach- tries (Bangladesh, India, Pakistan), as well as some of the ing or having achieved middle-income status and&market poorest. But South Asia differs from East Asia and the Pacific friendliness. There is confusion about the government's role in several important ways, including its much lower income in the health sector, however, and largely unregulated pri- level and lower projected economnic growth. South Asia con- vate financing and delivery have grown rapidly, often dupli- tains 22 percent of the world's population and accounts cating public efforts. Many public health services are poorly for 2 percent of its income. At $440, per capita income is funded and inefficiently delivered. Decentralization is the lowest among developing regions. The public sector needed, but administrative and managerial capacity at accounts for just 39 percent of all health expenditures, the INNOVATIONS IN HEALTH CARE FINANCING 34 lowest public share among developing regions. Health expen- Sub-Saharan Africa ditures account for 3.7 percent of GDP, or $12 per capita, also the lowest among developing regions. Sub-Saharan African countries are generally small and het- Except in India, external assistance is an important com- erogeneous, a result of tribalism and migration (World Bank ponent of health spending, accounting for more than 10 1994). The region contains 10 percent of the world's popu- percent of expenditures. The region contains the largest lation and accounts for 1 percent of its income. With a per number of people living in poverty among developing regions, capita income of about $775, the region has the second low- and faces a high burden of disease and undernutrition asso- est income among developing regions. The public sector ciated with poverty. Women's education and roles in soci- accounts for 54 percent of health expenditures, the second- ety are important issues that affect their health. There is highest share among developing regions. Health expendi- significant inefficiency in health care systems, both sys- tures account for 4 percent of GDP or $38 per capita, ranking temwide and in individual facilities. Public health services the region (along with East Asia and the Pacific) as the sec- are often poorly targeted in terms of easing the burden of ond lowest in per capita health expenditures after South Asia. disease and ensuring cost-effectiveness. Formal insurance, The poverty of individuals and the financial solvency of both social and private, is extremely limited, although some governments are major concerns, The small size of most household expenditures are collectivized through informal countries makes it difficult to achieve economies of scale arrangements. As in East Asia and the Pacific, decentral- in the provision of certain services. External assistance is a ization is an important issue, but administrative and man- major component of health system support in the region, agement capacity is weak. Quality is poor in both the public accounting for more than 15 percent of health spending on and private sectors, and traditional medicine is an impor- average and more than 50 percent in some countries. Formal tant component of the health care system. Significant seg- public and private insurance is limited, and risk pooling gen- ments of the population lack coverage and access. erally takes the form of extended families and informal rural Reforms in these countries should focus on basic pub- risk pooling (such as Bamako-type initiatives). Basic pub- lic health, nutrition, and women's health and education. lic health interventions for disease control and prevention, Targeted disease programs are still needed. Given limited including the spread of HIV/AIDS, continue to be the resources, reforms should focus on the poor, on financing region's main concern. The communicable disease burden and delivering services efficiently, and on financial sustain- is the highest in the world and will continue to be for at ability. Moreover, the private sector's sizable role in financ- least twenty-five years despite a shift toward noncommu- ing and delivery makes the public-private mix an important nicable diseases. Affording and delivering basic public health issue. Given the size of many of these countries, decen- and nutrition services are major challenges. There is little tralization is essential, and intergovemmental fiscal rela- coordination between the public and financially dominant tions must be addressed. Management capacity at all levels private sector. Service quality is poor, and the knowledge of government must be improved, and quality in both the base is problematic. Over the next ten years real per capita public and private sectors needs to be assured. Training for GDP is expected to increase by just 0.9 percent a year, the health care workers and managers is needed. Providing the lowest growth rate among developing regions after the regulatory and policy framework for effective risk pooling Middle East and North Africa. for people who can afford private insurance would also Given this poor economic prognosis and low levels of enhance social welfare and increase the efficiency of pri- income and health spending, reform in Africa must be viewed vate financing. Despite myriad basic health problems and over the long term-say, twenty years. External assistance a low baseline expenditure level, projected real per capita will continue to be essential to reform. Although sectorwide GDP growth of 3.7 percent a year over the next ten years reforms are needed, efforts should focus on improving tar- will allow countries in the region to deal more effectively geting, ensuring cost-effective delivery of basic public health with the poverty and health system challenges they face. services, and promoting women's education and health. Effective management and policymaking will be key, Improving physical and financial access to care is a neces- A CURMUDGEON'S GUIDE TO FINANCING HEALTII CARE IN DEVELOPING COUNTRIES 35 sary concomitant to such efforts. Governance capacity needs insurance the preferred vehicles for financing personal to be strengthened, and coordination between the public health services. and private sectors needs to improve. Although much of * Given these advantages, their limited revenue-raising the policy focus needs to be on communicable diseases capabilities, and the importance of private financing, (including HIV/AIDS), countries should start addressing the governments in developing countries should increase impending noncommunicable disease burden through their institutional capacity to ensure the availability of focused prevention and health promotion programs. Better efficiently run and privately financed health insurance risk sharing for personal health services through Bamako- to supplement government efforts. type initiatives and through publicly (where affordable) and * Informal risk-pooling schemes, often voluntary and spon- privately financed insurance arrangements should be a priority. sored by local governments, appear to be viable mech- anisms for pooling risks in poor rural areas. Conclusion * Recent innovations in managing health revenues-includ- ing managed competition and medical savings accounts- Although this paper has focused on health care financing may help promote the efficient use of resources and can in terms of public and private revenue sources, national poli- offset some of the problems inherent in insurance mar- cies must also be concerned with the management and kets in countries with sufficient administrative capacity uses of such funds. Still, strictly from the financing side, a and developed financial markets. number of health policy prescriptions have emerged that * Priority needs to be given to collecting information on apply to all countries and regions: public and private sources of health care revenues and * Revenue-raising efforts involve tradeoffs between equity expenditures in all regions. National health accounts and efficiency. are essential for effective policymaking. * Administrative capacity is an essential component of rev- enue-raising efforts. Notes X Since raising revenue imposes significant economic costs, governments must maximize the returns on the uses of 1. For example, the Pan American Health Organization (PAHO) such funds. is currently updating health expenditure data for countries in Latin , A gove.ment's abitytoraisereveueincreasssigni America and the Caribbean. The authors are grateful to Ruben Suarez, PAHO, for sharing the early results of this effort. icantly as income increases. 2. There are some definitional problems in the case of publicly * The structural characteristics of labor and industry in mandated but privately managed funds. Although the national low-income countries limit the instruments available to health accounts for OECD countries categorize these types of these governments to raise significant revenues. funds under the public domain, in many developing (mainly mid- * For reasons of equity and economic efficiency, broadly dle-income) countries that distinction is not always clear. For exam- based taxes with few loopholes should be the preferred ple, private insurance in Chile is a privately managed fund that is tax approach in developing countries. part of a publicly mandated health financing system, and is des- tax approach in developing countnes. ignated as being private in the Pan American Health Organization's * User charges that are properly designed and focused on regional database (Suarez 1997). services with tangible benefits to consumers and service 3. For an evaluation framework of health financing reforms see facilities can lead to increased financing and better ser- WHO 1995. vices, although user charges are not equitable unless they 4. Other market failures that may justify government involve- are applied to income-elastic goods and services. ment, such as entry barriers and decreasing production costs, are * Public health services should be publicly financed or sub- beyond the scope of this discussion; see Jonsson and Musgrove sidized. in this volume and Hsiao 1995. 5. In the case of certain purely public goods from which no * The advantages of health insurance as a means of pool- individual can be excluded from the benefit (vector control, national ing risks and the random nature and potentially high defense), individuals maybe unwillingto purchase any of the good, costs of treating many illnesses make public and private since they cannot be excluded from consuming it. INNOVATIONS IN HEALTH CARE FINANCING 36 6. In this context insurance simply means pooling risks through ance funds. Moreover, user charges are also prevalent under both a public or private entity that is the recipient of a premium, tax, arrangements. or other financial contribution. A national health service pools 19. The International Monetary Fund's (IMF) Government risks through collective contributions to finance health services Finance Statistics often do not include revenue data for regional even though an insurance fund is not specifically established. In and local governments. Although we obtained such data from other addition, a national health service directly controls the uses of such sources where they were absent for large countries, it is possible funds through direct provision of services, although in many devel- that missing data for some small countries biases the ratio down- oping countries social insurance funds also have their own deliv- ward. This would be a more serious problem in Latin America ery systems. In most OECD countries with social insurance funds, and the Caribbean given the region's higher income and govern- such funds purchase services from public and private providers ments' greater ability to raise revenues. that are not owned by the fund. Similarly, many of the OECD 20. Grant assistance and borrowing can be considered private countries with national health service systems are promoting sources of financing when the entity receiving the aid or loan is a facility autonomy and separation of finance from provision. private entity (nongovernmental organizations, private providers, 7. For an in-depth discussion of private health insurance see private citizens, and so on). Health services can also be financed Hall 1994 and Chollet and Lewis in this volume. For an analysis by cutting other public expenditures. Moreover, another method of insurance market failures and potential remedies see Hsiao of financing, not addressed in this paper, is to increase efficiency 1995, pp. 130-34. in the consumption and production of services. Scarce public tax 8. For a theoretical treatment of insurance see Laffont 1989, dollars should be used to purchase services that provide the max- chapter 8. imum social benefit, not just in terms of health status but also in 9. This occurs because pooling a large number of similar risks terms of economic efficiency in the production and consumption reduces the variance in the probability distribution of the adverse of such services. event occurring. The reduction in variance occurs through the law 21. Two other criteria for evaluating taxes are flexibility: the tax of large numbers, which states that the variance for a given risk is system should be able to respond easily (in some cases automatically) reduced the more times the risk occurs (see Hall 1994, p. 7). to changed economic circumstances; and political responsibility: 10. People payinsurance premiums both to be relieved of uncer- the tax system should be designed so that individuals can ascertain tainty about the loss and to be compensated should the loss occur. what they are paying so that the political system can more accurately Thus there is a value to the individual even if the loss does not reflect the preferences of individuals (see Stiglitz 1988, p. 390). occur (Dorfman 1982, p. 5). 22. Revenue generation is also sometimes included as a sepa- 11. For a risk to be insurable it must be important, accidental, rate criterion. The amount of revenues that will be raised depends and calculable, with definite losses that are not catastrophic rela- on the tax base, tax rates, exclusions, deductions, exemptions, tive to the size of the pool (Bickelhaupt 1983, pp. 13-14). tax avoidance, and tax evasion. The more recent taxation litera- 12. If individuals do choose their employers for insurance ture takes the position that since there are efficiency costs to rais- purposes, then there is clearly an adverse selection problem within ing revenues, for a given level of expenditures governments should employee groups as well (CBO 1994). choose the sets of taxes, rates, and so on that maximize that 13. Indemnity insurance (in which the individual or provider is country's social welfare objectives in terms of economic efficiency, reimbursed on a fee-for-service basis by the insurer for the medical equity, and administrative feasibility. expenses incurred) produces more moral hazard thanmanaged care, 23. In a partial equilibrium setting (that is, where the tax on but also has weaker incentives to risk select (Newhouse 1996). this segment of the economy will not affect any other segment), 14. Arrow (1963) argues that "competitive insurance markets the excess burden (B) can be defined using the following formula: will yield optimum allocation when the events insured are not con- trolled by individual behavior" and "non-market controls, whether B =1/2*p*Q*T2 internalized as moral principle or externally imposed, are to some 1/TI + l/e extent essential for efficiency" (pp. 537-38). where P is price, Q is quantity, T is the tax rate, Tl is the elasticity 15. See Newhouse (1996) for an interesting conceptual and of demand, and E is the elasticity of supply (Rosen 1995, p. 314). policy-relevant discussion of tradeoffs between production effi- 24. Assuming that there are two commodities, X and Y, that ciency and selection by health plans and providers. are taxed at rates Tx and Ty, that the elasticities of supply are 16. Some of these problems have been exacerbated by tax sub- infinite, and that lTx and TIy represent the elasticities of demand sidies for employer-provided health insurance; see Hall 1994. for commodities X and Y, the excess burden is minimized when: 17. Community rating differs from experience rating, in which premiums are based on the experience of a specific, generally more Tx/Ty = TIx/TIy homogeneous, group (such as an employment group). 18. In practice the situation is more complicated because in See Rosen 1995, p. 332. many countries general revenues are used to subsidize social insur- A CURMUDGEON'S GUIDE TO FINANCING HEALTH CARE IN DEVELOPING COUNTRIES 37 25. There are alsolikelytobe tradeoffs between equityand growth, CBO (Congressional Budget Office). 1994. The Tax Treatment of especiallyregarding taxes on capital income. Such taxes maybe prob- Employment-Based' Insurance. Washington, D.C.: U.S. lematic since capital is highly mobile in the global economy Government Printing Office. 26. For a conceptual and policy-oriented debate on mandates Chirba-Martin, MaiyAnn, andTroyen Brennan. 1994. "The Critical in the United States see HealthAffairs 2 (spring) 1994, pp. 7-107. Role of ERISAin State Health Reform."HealthAffairs2 (spring): 27. Official development assistance is defined as grants and 142-56. loans made on concessional terms (that is, having a grant element Committee on Energy and Commerce, House of Representatives of at least 25 percent). See World Bank 1996a, pp. 44-45, and and Special Committee onAging, U.S. Senate. 1989. Costs and Feyzioglu, Swaroop, and Zhu 1996, p. 1. Effects of Extending Health Insurance Coverage. Washington, 28. Some countries have attempted to finance such borrow- D.C.: U.S. Government Printing Office. ing out of the most destructive tax of all, inflation. See Musgrave Creese, Andrew, and Joseph Kutzin. 1995. "Lessons From Cost- and Musgrave 1976, chapters 33-34. Recovery in Health." Discussion Paper 2. World Health 29. Each of the fifty U.S. states is responsible for regulating Organization, Geneva. WHO/SHS/NHP/95.5. private health insurance. However, employer-provided health insur- Cutler, David, and Brigitte Madrian. 1996. "Labor Market Responses ance is often exempt from state regulation because of a federal to Rising Health Insurance Costs." NBERWorking Paper 5225. preemption in the Employee Retirement Insurance Security Act National Bureau of Economic Research, Cambridge, Mass. of 1974 (ERISA), which provides for federal oversight of certain Dorfman, Mark. 1982. Introduction to Insurance. Englewood Cliffs, employer-provided benefits, including health insurance. Such dual N.J.: Prentice Hall. responsibility has precluded states from effectively dealing with a Feyzioglu, Tarhan, Vinaya Swaroop, and Min Zhu. 1996. "Foreign large element of the market segmentation problem. See Chirba- Aid's Impact on Public Spending." Policy Research Working Martin and Brennan (1994). Paper 1610. World Bank, Washington, D.C. GAO (U.S. General Accounting Office). 1996. "Health Insurance Regulation, Varying State Requirements Affect Cost of References Insurance." GAO/HEHS-96-161. Washington, D.C. Gertler, Paul, and Jacques van der Gaag. 1990. The Willingness to Aaron, Henry. 1991. Serious and Unstable Condition: Financing Payfor Medical Care. Washington, D.C.: World Bank. America's Health Care. Washington, D.C.: The Brookings Goldstein, Ellen, Alexander Preker, Olusoji Adeyi, and Gnanaraj Institution. Chellaraj. 1996. Trends in Health Status, Services, and Finance: Abel-Smith, Brian. 1967. An International Study of Health The Transition in Central and Eastern Europe. World Bank Expenditures and Its Relevance for Health Planning. Geneva: Technical Paper 341. Washington, D.C. World Health Organization. Gruber, Jonathan. 1994. "The Incidence of Mandated Maternity Allan, Charles. 1971. The Theory of Taxation. Middlesex, England: Benefits." The American Economic Review (June): 622-41. Penguin. Hall, Mark. 1994. Reforming Private Health Insurance. Washington, American Academy of Actuaries. 1994. An Analysis of Mandated D.C.: American Enterprise Institute. Community Rating. Washington, D.C. Helms, David, Anne Gauthier, and Dan Campion. 1992. "Mending Arnett, Grace-Marie. 1996. "Health Care: The Time Bomb Is the Flaws in the Small Group Market." Health Affairs (sum- Ticking." Paper presented at the Galen Institute Conference mer): 7-27. on A Fresh Approach to Health Care Reform, March 25, Hoff,John. 1996. "Improvingthe System for Delivering Subsidies: Washington, D.C. Cap or Scrap the Exclusion?" Paper presented at the Galen Arnott, Richard, and Joseph Stiglitz. 1988. "The Basic Analytics Institute Conference on A Fresh Approach to Health Care of Moral Hazard." NBER Working Paper 2484. National Reform, March 25, Washington, D.C. Bureau of Economic Research, Cambridge, Mass. Hsiao, William. 1995. "Abnormal Economics in the Health Sector." .1990. "The Welfare Economics of Moral Hazard." NBER Health Policy 32: 125-39. Working Paper 3316. National Bureau of Economic Research, IMF (International Monetary Fund). Various issues. Government Cambridge, Mass. Finance Statistics. Washington, D.C. Arrow, Kenneth. 1963. "Uncertainty and the Welfare Economics Institute for Health Policy Solutions. 1995. State Experiences With of Health Care." American Economic Review 53: 537-38. Community Rating and Related Reforms. Washington, D.C. Bickelhaupt, David. 1983. General Insurance. Homewood, Il.: Khalilzadeh-Shirazi,Javad, andAnwar Shah, eds. 1991. TaxPolicy R.D. Irwin. in Developing Countries. Washington, D.C.: World Bank. Burki, Shahid. 1995. "The Health Gap and the Challenges for Klugman, Jeni, and George Schieber. 1996. A Survey of Health Reform in Latin America and the Caribbean." World Bank, Reform in Central Asia. World Bank Technical Paper 344. Latin American and the Caribbean Region, Washington, D.C. Washington, D.C. INNOVATIONS IN HEALTH CARE FINANCING 38 La Forgia, Gerard, and Charles Griffin. 1993. Extending Coverage on A Fresh Approach to Health Care Reform, March 25, and Benefits of Social Financing Systems in Developing Countries, Washington, D.C. Washington, D.C.: The Urban Institute. Poullier, Jean-Pierre. 1992. "Administrative Costs in Selected Laffont, Jean-Jacques. 1989. The Economics of Uncertainty and Industrial Countries." Health CareFinancingReview (summer): Information. Cambridge, Mass.: MIT Press. 167-72. Lewis, Maureen. 1993. "User Fees in Public Hospitals: Comparison Ray, Anandarup. 1975. "Cost Recovery Policies in Public Sector of Three Country Case Studies." Economic Development and Projects." Policy Paper 206. World Bank, Washington, D.C. Cultural Change 41 (April): 513-32. Rosen, Harvey. 1995. Public Finance. 4th ed. Chicago: Irwin. Michaud, Catherine, and Christopher Murray. 1994. 'Aid Flows Saltman, Richard, and Joseph Figueras. 1996. European Health to the Health Sector in Developing Countries." Background Care Reforms: Analysis of Current Strategies. Copenhagen: World paper prepared for World Development Report 1993. World Health Organization. Bank, Washington, D.C. Stiglitz, Joseph. 1988. Economics of the Public Sector. 2nd ed. - 1996. 'Annex 5." In Report of the Ad Hoc Committee of New York: WW. Norton. WHO: Investing in Health Research and Development. Geneva: Suarez, Ruben. 1997. Personal communication. Pan American World Health Organization. Health Organization, Washington, D.C. Moffet, Robert. 1993. "Overdosing on Management: Reforming Tanzi, Vito. 1987. "Quantitative Characteristics of the Tax Systems the Health Care System through Managed Competition." of Developing Countries." In David Newbery and Nicholas Heritage Lectures 441. Heritage Foundation, Washington, D.C. Stern, eds., The Theory of Taxation for Developing Countries. Musgrave, Richard. 1987. "Tax Reform in Developing Countries." New York: Oxford University Press. In David Newbery,and Nicholas Stem, eds., The Theory of Taxation Thirsk, Wayne. Forthcoming. "The Substance and Process of Tax for Developing Countries. New York: Oxford University Press. Reform in Nine Developing Countries." In World Bank, Musgrave, Richard, and Peggy Musgrave. 1976. Public Finance in Reforming Taxes in Developing Countries. Washington, D.C. Theory and Practice. New York: McGraw-Hill. Tibouti, Abdelmajid. "The Bamako Initiative: Experience to Date Musgrove, Philip. 1996. Public and Private Roles in'Health. World with Community Financing in Poor Rural Areas." United Bank Discussion Paper 339. Washington, D.C. Nations Children's Fund, New York. Newbery, David, and Nicholas Stern, eds. 1987. The Theory of Van Doorslaer, Eddy, and Adam Wagstaff. 1995. "Equity in the Taxationfor Developing Countries. New York: Oxford University Finance and Delivery of Health Care in Europe." Papers pro- Press. duced for the European Union's Biomed Programme. Erasmus Newhouse, Joseph. 1996. "Reimbursing Health Plans and Health University, Rotterdam. Providers: Efficiency in Production Versus Selection."Journal Van Doorslaer, Eddy, Adam Wagstaff, and F.F.H. Rutten. 1993. of Economic Literature 34 (September): 1236-63. Equity in the Finance and Delivery of Health Care:An International Nichols, Len. 1996. Personal communication. Urban Institute, Perspective. Oxford: Oxford University Press. Washington, D.C. WHO (World Health Organization). 1995. "Health Financing OECD (Organization for Economic Cooperation: and Reform: A Framework for Evaluation." Revised Working Development). 1977. Public Expenditures on Health. Paris. Document WHO/SHS/NHP/96.2. Division of Analysis, 1985. Measuring Health Care, 1960-83. Paris. Research, and Assessment, Geneva. 1992. The Reform of Health Care: A Comparative Analysis . 1996a. Equity in Health and Health Care: A WHO/SIDA of Seven OECD Countries. Paris. Initiative. Geneva. - 1993. OECD Health Care Systems: Facts and Trends, . 1996b. Investing in Health Research and Development. 1960-91. Paris. Geneva. - 1994. The Reform of Health Care Systems: A Review of World Bank. 1987. Financing Health Services in Developing Countries: Seventeen OECD Countries. Paris. An Agenda for Reform. Washington, D.C. 1995. New Directions in Health Care Policy. Health Policy . 1988. World Development Report 1988: Opportunities and Studies 7. Paris. Risks in Managing the World Economy. New York: Oxford PAHO (Pan American Health Organization). 1996. NationalHealth University Press. Expenditure and Financing in Latin America and the Caribbean: . 1991. Lessons of Tax-Reform. Washington, D.C. Challengesfor the 1990s. Washington, D.C. - 1993. World Development Report 1993: Investing in Health. Pauly, Mark. 1986. "Taxation, Health Insurance, and MarketFailure New York: Oxford University Press. in the Medical Economy" Journal of Economic Literature 24 .. 1994. Better Health in Africa: Experience and Lessons (June): 639-40. Learned. A Development in Practice book. Washington, D.C. 1996. 'An Efficient and Equitable Approach to Health . 1996a. Global Economic Prospects and the Developing Reform." Paper presented at the Galen Institute Conference Countries 1996. Washington, D.C. A CURMUDGEON'S GUIDE TO FINANCING HEALTH CARE IN DEVELOPING COUNTRIES 39 . 1996b. 'AHealth SectorAssistance StrategyfortheEurope Population, and Nutrition Sector Assistance Strategy." Middle and Central Asia Region." Europe and Central Asia Region, East and North Africa Region, Washington, D.C. Washington, D.C. . 1997. World DevelopmentIndicators 1997. Washington, D.C. - 1996c. "Investing in Human Development in the Latin . Forthcoming. Reforming Taxes in Developing Countries. America and Caribbean Region." Latin America and the Washington, D.C. Caribbean Region, Washington, D.C. Zelman, Walter. 1994. "The Rationale Behind the Clinton Health 1996d. "Middle East and North Africa Region Health, Reform Plan." Health Affairs 1 (spring): 9-29. INNOVATIONS IN HEALTH CARE FINANCING 40 Government Financing of Health Care Bengt Jonsson and Philip Musgrove JIn a market economy prices serve three functions: they Relations between Finance and Provision guide the allocation of resources, they ration scarce of Health Care goods and services, and they finance the payment of compensation to the factors of production. When the mar- There is no necessary connection between the way that ket provides these functions, the question of how to pay health care is paid for and the way that it is delivered; in for a particular good or service does not arise: consumers particular, public finance does not imply public provision. buy the product in amounts that are determined by their The arguments for and against public provision of health wants and capacity to pay, and producers deliver those services are very different from those related to public and amounts, with prices equilibrating demand and supply. private financing (J6nsson 1996). It is common to distin- There are several reasons direct payment by consumers guish three relations between funders and providers of health for health care is inefficient and unequitable, creating the care: the reimbursement, contract, and integrated approaches need for government interventions and for alternative mech- (OECD 1995). These relations are largely independent of anisms to allocate resources and financing. This paper con- specific taxes and other sources of funds. siders the financing mechanism-that is, the role of Under the reimbursement approach, providers receive government financing of health care. Other issues related retroactive payments for services supplied. These payments to government intervention, such as the optimal allocation may be billed directly to insurers or to patients, who may of resources in the provision of health care, are discussed be partly or entirely reimbursed by insurers. The reim- only if they directly relate to government financing. bursement approach, often coupled with fee-for-service pay- In considering the alternatives for government financ- ment arrangements, can be found in systems with multiple ing of health care, we focus on three questions: Should private and public insurers and multiple (usually private) government pay for health care? Does it matter which of suppliers, as in the United States. In low- and middle-income two general models-social insurance and direct financing countries it is rare for the reimbursement model to be com- from general revenues-is used for public finance? Which bined with public finance. Chile is an exception, with part taxes should be used to finance health care, and do the types of government financing reimbursing private providers and levels of taxes matter for coverage, benefits, and expen- retroactively. ditures? Nearly all the available empirical evidence relevant The contract approach involves an agreement between to these issues is for OECD countries; thus the conclu- third-party payers (insurers) and health care providers aimed sions do not necessarily apply to developing countries. at greater control over total funding and its distribution. Wherever possible, the situation in developing countries is This approach tends to be found in social insurance systems discussed separately. with predominantly private (nonprofit) providers. Bengt J6nsson is professor of health economics at the Stockholm School of Economics. Philip Musgrove is principal economist in the Human Development Department at the World Bank. 41 Prospective budgets are combined with per diem, case mix The role of government differs between these two orga- (diagnostic related group, or DRG), and fee-for-service pay- nizational models. In the first type the government must ments. A variant of this system is used in Brazil, where raise most of the necessary funds, and then contractwith the budgets are set by the state or municipality and providers providers. In the second type the government must regu- are paid under a DRG tariff (Lewis 1994). Preferred provider late competition between insurers, or third-party payers, and organizations in the United States also use the contractual distribute the public subsidies needed to guarantee univer- approach. sal access to a certain level of health care. Determining the In integrated health systems the same agency controls criteria and mechanisms for the distribution of risk-adjusted both the funding and the provision of health services. Medical subsidies is a major problem (Newhouse 1994; van Vliet and personnel are generally paid salaries, and budgets are the van de Ven 1992). Such subsidies are usually differentiated main instrument for allocating resources. Integrated pub- by the consumer's age and sex and by region, taking into lic systems are used in the Nordic countries and until recently account the income of the population served (which affects were the model in the United Kingdom, and are the com- the demand for services) and the costs of providing care. mon organizational form for ministries of health in devel- This approach has been used in Chile for primary health care oping countries. In many such countries the integrated (which is a municipal responsibility) and is being imple- approach is also used for social security systems, which have mented in Argentina and Colombia. their own hospitals and clinics, although there are often also Insurance is central to any discussion of health care contractual relations with private providers. Health main- finance. And while there are markets for many kinds of tenance organizations (HMOs) in the United States are insurance, health care insurance is peculiar because of the examples of integrated private systems. nature of the asset being protected-human health rather Most health care systems include elements of all three than nonhuman capital (Musgrove 1996). The introduction systems, just as most have a mix of models for public finance. of insurance for health care, whether voluntary and private There also have been significant changes over time. Italy or publicly financed, has consequences not only for the and Spain's public health care systems have moved from a distribution of payments for health care, but also for the contract approach to an integrated system, while those of allocation of resources to and within the health care system. New Zealand and the Uriited Kingdom have moved from By introducing a third party that collects revenue and pays an integrated system to a contract approach. providers, health care insurance changes the relation between There maybe a trend toward two types of relation between consumers and providers of health care (unless, as in the funders and providers (J6nsson 1996; van de Ven, Schut, integrated model, insurance and provision are combined and Rutten 1994). The first type involves a (near) public in a single agency; figure 1). The crucial importance of insur- monopoly in health care funding, through taxes or com- pulsory social insurance contributions, and competitive con- FIGURE I tracts with private and public providers. Thus financing and Economic relations in the fiance and delivery provision may be separated, in what is sometimes referred of health care to as a purchaser-provider split. The second type is an inte- Money (direct payments) grated model with competition between different integrated Consumers of care Providers of care systems (HMOs). (patients) Health services (docors, hospitals) In the first type consumers usually have no (or limited) choice of insurer, but do have a choice of provider. In the Insurance Claims second type there is a choice of insurer, but once this choice coverage is made the consumer is tied to the providers linked with that insurer. No health care system in the world offers a freeMoney Money choice of both insurer and provider to everyone in the pes (government) budgets) population. INNOVATIONS IN HEALTH CARE FINANCING 42 ance and the problems peculiar to it provide one of the ratio- intervention raises the questions mentioned earlier-whether nales for public financing of care, and raise most of the issues to establish public insurance, how to organize it, which discussed here about how best to pay for it. sources to use to finance it, and with what decisions and consequences for coverage, benefits, and spending. Should the Government Pay for Health Care? Financing of individuol (personal) health core expendi- tures One major reason for public financing of health care is the provision of public goods, such as programs for medical There are three ways to finance individual health care: pri- research, health promotion, vector control, and food and vate individual payments, private collective payments, and water safety. Such public goods provide benefits that are public finance. shared by many people, regardless of whether they pay for them. Thus entirely private markets would yield an ineffi- Private individual payments. These payments are also cient allocation of resources, and government financing called out-of-pocket costs. The problem with direct indi- (or some other nonmarket alternative) is needed to opti- vidual payments for health care is that medical expenses are mize allocation. The situation is essentially the same for sometimes so high that even people with higher than aver- goods with externalities (such as vaccinations), which can age incomes cannot afford them. This problem is aggra- be produced and consumed privately but whose conse- vated by the fact that increased health care costs may coincide quences-good or bad-affect other consumers regard- with reduced income due to the health problems. In such less of whether they choose to consume those goods. For a situation personal savings may be inadequate and oppor- example, immunization of part of the population also pro- tunities to borrow for investments in health are limited. tects the unimmunized. Moreover, people with the lowest incomes, who often need However, most health care interventions produce pri- care the most, will be excluded from much care if it is vate goods, with benefits limited to individual consumers. financed by direct payments. Arguments favoring public finance of purely private per- This problem does not preclude direct individual pay- sonal health care expenditures differ from those for public ments for health care-which usually account for 10-50 goods, for two reasons. The first is the need to finance health percent of all payments-but it does call for protection care for the poor-that is, people who cannot afford what against high costs. That means paying for health care col- society considers an adequate amount of health care either lectively, and sharing the financial risk. (There are no gen- out of pocket or by buying insurance. The second affects eral reasons for government to finance small amounts of the entire population, and derives from imperfections in medical expenditures. The social benefits are small-all but insurance markets that prevent them from providing an effi- the very poor can afford some medical expenditure-and cient and equitable allocation of health care resources. the social costs may be high.) The relative importance of these two domains-subsi- dies for the poor and insurance for people who can help Private collective payments (insurance). Because many finance it-depends on a country's level and distribution illnesses occur rarely and seemingly at random, health care of income, and explains much of the difference in health expenditures are uncertain as well as possibly high. By pool- care outcomes between rich and poor countries. Many ing a large number of people, insurance reduces the vari- governments have become the main insurer for personal ability of their incomes net of medical expenditure. Health health care, particularly in high-income countries, and this expenditures may be highly variable for a given member of involvement becomes the quantitatively most important the pool, but average outlays can be predicted fairly well. reason for governments to raise money for health care. Thus insurance reduces financial risk for consumers who Because of the complexity of the insurance market, and are risk averse (that is, who have a diminishing marginal the fact that both efficiency and equity are involved, such utility of wealth or income) and lowers health risks since GOVERNMENT FINANCING OF HEALTII CARE 43 care is more accessible. Financial risk is usually not elimi- and Zeckhauser 1971; and Mirrlees 1971). The optimal nated because coinsurance and deductibles are used to make insurance contract is a second-best, nonlinear solution the insured person share the costs (see Chollet and Lewis with a mix of risk spreading and incentives such as a mod- and Creese andBennettinthisvolume).Apolicymayrequire erately high deductible and a diminishing coinsurance rate that the insured pay the first $200 of healh care costs out (Blomqvist forthcoming). An alternative is to include 'bonus of pocket each year (deductible) and then pay 20 percent options" (Zweifel 1992) or rebates in the event that the of all charges (coinsurance). insured does not submit any claims during the year, with This cost sharing is one way to control moral hazard- the rebate increasing in subsequent years without a claim the increased use of services and reduced precaution in tak- until a maximum is reached. This approach provides first- ing care of one's health that results when risk pooling leads dollar coverage but still provides incentives to reduce moral to reduced marginal costs for services. Moral hazard can hazard. Few insurance contracts have these features, but manifest itself in two ways, one static and the other dynamic. many include suboptimal provisions such as an annual ceil- People with health insurance tend to see doctors more often ing on copayments. Integrating insurance with service pro- and to use costly treatments even if the benefits are small vision is another alternative, and removes the incentive for (Pauly 1968; Zeckhauser 1970). Doctors also may change providers t6 overtreat since they then bear the financial risk. their behavior, particularly in fee-for-service systems. Since costs are not borne by the patient, it is easier for doctors to Public finance: government as insurer Moral hazard is a suggest more expensive treatments. The dynamic effect of problem in any insurance system, but adverse selection moral hazard is the incentives it creates to introduce new and the attempts of insurers to counter it by excluding poten- medical technology for which there would be no market in tial consumers and adjusting premiums are peculiar to pri- the absence of insurance (Weisbrod 1991). Both problems vate insurance. This is perhaps the main argument in favor derive from the inability of the insurer to monitor service of public insurance, which can more easily be made uni- providers and the insured. versal and in effect force everyone to share the risks. Public Insurance firms incur costs for doing business such as insurance is also often justified by some related problems- processing claims and marketing. These are called loading of free riders, of excluded population groups, and of col- costs, and they generally make competitive private insurance lective risks that are largely independent of individual risks. more costly to administer than uniform public insurance. In a voluntary insurance system people can choose not Many of these costs arise because insurance companies have to insure. This is not a problem if the uninsured can be an incentive to exclude high-risk consumers or to at least ignored when they need medical care but cannot pay for it. identify them so that they can be charged more, but have If they are taken care of anyway-that is, allowed to "ride trouble identifying which risk class people belong to. In free"-the incentive to have insurance is reduced. It is dif- the short run this situation of asymmetric information- ficult to judge how important this problem is. In Switzerland consumers who know their risks better than the insurer most families have insurance despite the fact that it is not does-may benefit high-risk people who, if they know they compulsory in all cantons. Cultural tradition probably plays are likely to need medical care, will be eager to buy insur- a large role; and if most people make an effort to take care ance. This tendency of the highest risks wanting the most of themselves, there is room for generosity to those who do insurance is called adverse selection. If insurance companies not. It is often suggested that private insurance be made compensate for it by raising premiums, some low-risk per- compulsory in order to avoid the free-rider problem. But sons may decide not to buy insurance. This can lead to a doing so would raise another problem-which sanction to vicious circle in which only high-risk people remain. use for people who do not comply-and in any case is infea- There are several ways to reduce moral hazard and adverse sible for poor populations. selection, although there is no complete solution for com- Exclusion is the opposite of the free-rider problem. People petitive insurers in a situation where information is asym- may want insurance but cannot buy it because of low income metric and imperfect (Pauly 1974; Zeckhauser 1970; Spence or high risk. One solution may be to give them a voucher INNOVATIONS IN HEALTEI CARE FINANCING 44 (subsidy) so they can buy private insurance. The practical and in particular how public finance for health care affects problems of calculating and administering such subsidies the private market. may be considerable, making public insurance simpler- again, particularly in countries with large poor populations. Public finance: government subsidizes but does not insure. The elderly, with high risks for illness and related expendi- It is also possible for governments to finance health care tures, pose a particular problem, and some countries have without acting as insurers, by subsidizing private insurance created public insurance just for them. The need to pre- through the tax system. Employers often pay a significant pare for health expenses in old age can be partly solved portion of workers'health insurance premiums: in the United through a funded system in which each insured pays into a States, for example, about 80 percent of the premiums for fund that covers future needs (in contrast to a pay-as-you- private health insurance are paid by employers (Phelps 1986). go system, where each person's contribution goes toward If employers are allowed to deduct these costs from the the current expenses of all members). income on which they pay corporate taxes but employees This approach also has its problems. Knowledge about are not taxed on the value of the premium-that is, the the future incidence and prevalence of illness and poten- cost of the insurance is not treated as income to either the tial treatments is limited. Thus it is difficult, if not impos- company or the worker-then the insurance is partly financed sible, to calculate the premiums that 20-year-olds should by a subsidy or "tax expenditure" equivalent to the tax that pay for health care that they will receive in fifty years. One the government does not collect. The same situation occurs solution to such collective risks is for the government step if individuals' private health insurance premiums are tax in as a re-insurer. The problem with life-long insurance can deductible, or if employers pay directly for health care be seen in countries where private insurance funds go bank- (self-insurance). This kind of public subsidy is not used rupt when their members become older, and have to be much outside the United States, and there are few estimates merged into funds with younger members. (This problem of its cost. But in Brazil in the early 1980s it appears to also affects pay-as-you-go public insurance, even in middle- have accounted for $1 billion in health care spending, about income countries, as the population ages.) An alternative a quarter of what the government spent (Lewis 1994). is for voluntary insurance to be restricted to a certain age Paying for health care with employment-related tax deduc- group, for example below 65. The governrment must then tions can solve some of the problems of private insurance, finance care for people over that age. But with increasing in that adverse selection is limited by contracting in large life expectancy and the concentration of costly illness at groups rather than one person at a time, lowering adminis- advanced ages, this approach means that the government trative costs. However, this approach introduces two other will end up paying for the bulk of health care. problems. First, employees may not recognize that they are It is important to distinguish between actuarially fair paying for their insurance, at least partly, through lower wages. insurance, as provided through risk pooling, and govern- This leads to higher than optimal insurance coverage (overin- ment social insurance programs. Actuarially fair insurance surance) and thus higher health care expenditures. The con- is provided through markets in which buyers voluntarily pay sequences for total health care expenditures can be substantial. for protection against infrequent high medical expenditures Phelps (1986) estimates that employer group health insur- whose probabilities can be statistically determined, with pre- ance premiums in the United States would be about 45 per- miums adjusted accordingly (see Chollet and Lewis in this cent lower if the tax subsidy were not in effect, even though volume). Sodal insurance programs are provided by gov- marginal tax rates are only 25-35 percent. There are also wel- ernment, often involve an income transfer between popu- fare losses due to employment choices and wage levels, which lation groups for reasons unrelated to health, have a defined are affected bythe subsidized excess insurance (Feldman and set of eligibility rules, and are partly or wholly financed Dowd 1991; Feldstein 1973; Manning and others 1987). through taxes or compulsory insurance premiums that need Second, the size of the subsidy increases with the marginal not be actuarially fair. These differences raise questions tax rate; if taxes are progressive, the higher subsidies go to about the best combination of private and public insurance, people with higher incomes, which is inequitable. GOVERNMENT FINANCING OF HEALTH CARE 45 Total health expenditures and govemment shares or discretionary income. On this basis public efforts to finance health care in poor countries would probably look Total and public financing of health care in different regions larger than in OECD countries, making issues of organi- are shown in table 1. The public share of health care expen- zation and sources of revenue just as important, if not ditures is at least 50 percent in every region except Asia. more so. The share is highest in rich countries, which also have the highest total expenditures. Private financing dominates in Which Model for Public Finance? low-income countries, and direct out-of-pocket payments are more important than private insurance as a source of As noted earlier, there is a basic distinction between tax- revenue. based (directly financed) and insurance-based public health A similar, though less clear, picture emerges for OECD care systems. In tax-based systems general revenue taxes countries (table 2). The countries with the lowest incomes are the main source of finance, and the government usu- also have the lowest shares of public finance. The United ally acts as the main provider of health care. Insurance-based States is an exception, having the lowest share of public systems are financed mainly through payroll taxes, up to a finance of all OECD countries. By contrast, the govern- ceiling onwages at which point the marginal tax rate becomes ment provides nearly all healh care resources in Iceland, zero. The number of insurance agencies varies by country, Norway, Sweden, and the United Kingdom. from one in most Latin American systems, to several in Since annual public spending on health typically reaches Europe, to more than 200 in Argentina. Directly financed $1,000-2,000 per capita in rich countries, the questions systems also provide the protection that characterizes insur- of how to organize and pay for care are highly relevant. In ance, but the insurance is implicit, and individuals need low-income countries the amounts at stake are much not be explicitly affiliated to receive benefits. smaller, both absolutely-$10 per capita per year in many There is also considerable variation in the connection African countries, and less than $100 in most of Asia and between finance and provision. Social security institutes Latin America-and as a share of income, but it is harder operate their own clinics and hospitals in most Latin to raise revenue and a larger share of the population is American countries, but contracting with private providers too poor to afford any significant amount of private health is the norm in Argentina, Brazil, and most of Europe. In care. A better way to compare rich and poor countries may most cases these institutes oversee nonprofit institutions be to estimate subsistence income and compare total and and independent physicians. Different systems use differ- public health spending with the remaining nonsubsistence ent methods to reimburse providers for services; some- TABLE I Global health care expenditures by region, 1990 Total health Health Public health Share of expenditure expenditure as expenditure as Share of GNP Per capita health world population (billions of percentage of percentage of spent on health expenditure Region (percent) U.S. dollars) world total regional total (percent) (U.S. dollars) OECD countries 15 1,483 87 60 9.2 1,860 Transition economies of Europe 7 49 3 71 3.6 142 Developing countries 78 170 10 50 4.7 41 Labn America and the Caribbean 8 47 3 60 4.0 105 Middle East and North Africa 10 39 2 58 4.1 77 Other Asia and islands 13 42 2 39 4.5 61 India 16 18 1 22 6.0 21 China 22 13 1 59 3.5 11 Sub-Saharan Africa 10 12 1 55 4.5 24 Worid 100 1,702 100 60 8.0 329 Source: World Bank 1993. INNOVATIONS IN HEALTH CARE FINANCING 46 TABLE 2 In practice, the distinction between the two models is Public and per capita health care spending, OECD not so sharp as this discussion suggests. A tax-based sys- countries tem can obtain part of its financing from the social insur- Public financing Total health ance system (as in Sweden), and part of the services can be as percentage spending GDP provided by private contractors (for example, general prac- Country of total per capitae per capitaa titioners in Denmark and the United Kingdom). Australia 69 1,606 18,970 Moreover, social insurance systems often receive com- Austia 63 1,965 20,216 Belgium 88 1,653 20,184 plementary financing from general revenues (as in Canada 72 2,010 20,608 Switzerland), and the insurance system can be supplemented Denmark 83 1,362 20,784 with other services del paid operated the Finland 75 1,357 16,362 o1rectl for and by France 78 1,866 19,169 government (for example, the Veterans Administration in Germany 74 1,869 19,720 the United States). In extreme cases such as Brazil and Iceland 84 1,577 19,402 Ireland 76 1,201 15,202 Costa Rica, the distinction between the two systems ceases Italy 71 1,561 18,698 to exist because coverage is extended to (nearly) all non- JMexico 58 14395 27,484 contributors, and wage taxes become just one of the sources Netherlands 78 1,641 18,570 of revenue. This extension of coverage, whichwas large and New Zealand 77 1,226 16,424 rapid in both countries during the 1980s, effectively merges Norway 95 1,604 21,980 Portugal 56 938 12,313 social insurance with the existing tax-financed system. Spain 79 1,005 13,572 Changes in other countries have also blurred the dis- Sweden 83 1,348 17,435 Switzerland .- 72 2,294 23,961 tinction. Tax-based systems have introduced a split between Turkey 58 223 5,271 purchaser and provider, as in the United Kingdom, and United Kingdom 84 1,211 17,560 insurance-based systems have introduced global budgets United States 44 3,516 24,629 in order to control costs, as in France and Germany. In the Note: Data are for 1994 or latest available year. a. Expressed in purchasing power parity dollars. Czech Republic a purely tax-based system has been sup- Source: OECD Health Database 1996. -plemented with social insurance funds that are supposed times it is fee for service, but contracts with negotiated to compete for clients, with a general fund acting as the prices are more common. Another difference between the insurer of last resort. But the use of general revenue taxes two models is that while tax-based systems are always admin- has not disappeared, because the government now buys istered directly by the financing agency (usually a ministry insurance for people who do not contribute through what of health), the administration of social insurance is some- is effectively a wage tax. Because this change was introduced times taken care of by independent government-regulated together with a move from salaried providers to largely fee- bodies (such as Germany's Krankenkassen and France's for-service medicine with inadequate regulation, one of mutuelles). the results has been an explosion of costs. That outcome, Dependence on wage taxes means that social insur- however, is not specific to a social insurance or a tax-based ance is paid for by workers (and possibly employers) rather system. The same logic, of competing but regulated insur- than the general public. Tax-based or direct systems, by ers financed partly by a uniform wage tax and partly by contrast, are in principle universal. From a strictly financ- subsidies for the poor and the unemployed from general ing perspective, the same problem has to be solved whether taxation, characterizes current reforms in Argentina and the financing mechanism is taxes or compulsory insur- Colombia. ance contributions: the state obtains involuntary payments Even when the two models remain distinct, any country from members of society, and these payments are unre- can use both at once. This is the typical Latin American lated to individual benefits. In terms of coverage, effi- model, with both a ministry of health and a social security ciency, and equity, however, the two systems can differ institute, where in principle the institute serves only its substantially. contributors but the ministry offers services to the entire GOVERNMENT FINANCING OF HEALTH CARE 47 population. In practice the users of ministry services are Public spending pottems under the two models determined by the quality of services and by the existence of private insurance, as well as by costs to consumers, and As noted, the public share of health spending rises and the the same people may draw on a variety of providers and private share falls as country income grows, and (except sources of finance-paying out of pocket for drugs and inex- for the United States) rich countries are more alike than pensive consultations, using public facilities for hospital- poor countries in this respect. The situation is much less ization, and even covering some services with private homogeneous when the comparison is between countries insurance. This complexity also characterizes South Africa, with tax-based and social insurance systems. And low-income where the tax-based system finances care for the poor major- countries are more similar than middle- and high-income ity, social insurance covers middle- and high-income work- countries in shares of spending on the two models. ers, and there is also private insurance. Providers in this The share of pubic health care spending that is financed system tend to work with only one financing system and so directly or by taxes (that is, not through social insurance) are almost as sharply segregated as the sources of funding. in low- and middle-income countries for which estimates Such overlapping patterns make it difficult to describe "cov- are available is shown in table 3. In many countries all erage" consistently, and the coexistence of two or more public spending is tax-based and social insurance does not financing systems makes it almost impossible to judge which exist; nearly all these countries are quite poor. This is a (if any) of them works best. natural consequence of a small formal employment sec- TABLE 3 Public health care financing in various countries, circa 1990 (percent) Share of public health spending Public financing Total health financed directly as percentage Income spending as Region/country or by taxes of total per capitaa percentage of GDP Sub-Saharan Africa Cameroon 68 38 2,400 3 Ethiopia 100 61 370 4 Kenya 56 63 1,350 4 Madagascar 76 50 710 3 Malawi 62 58 800 5 Mozambique 100 75 600 6 Nigeria 100 44 1,360 3 Sierre Leone 52 71 800 2 Tanzania 100 68 570 5 Zimbabwe 65 52 2,160 6 Asia Bangladesh 100 44 1,160 3 Bhutan 100 620 2 China 0 60 1,680 4 India 21 22 1,150 6 Indonesia 67 35 2,730 2 Korea, Rep. of 12 41 8,32.0 7 Malaysia 100 43 7,400 3 Myanmar 100 3 Nepal 44 49 1,130 5 Pakistan 100 53 1,970 3 Papua New Guinea 100 64 1,830 4 Philippines 80 50 2,440 2 Sri Lanka 83 49 2,650 4 Thailand 90 22 5,270 5 INNOVATIONS IN HEALTH CARE FINANCING 48 tor, which makes it difficult to raise significant revenue ers, social insurance coverage is lower but those covered from wages and salaries. Even where such taxes can be have higher than average incomes, and tax-based finance collected, they are often pooled with other tax revenues may cover as much or more of the population but cost rather than used separately to finance social insurance. very little per person. In general, shares of finance are unre- (Social insurance schemes limited to civil servants are a lated to shares of coverage or utilization of services; on more common exception, and these exist even in some poor the contrary, when both systems operate in a country, social countries.) There are few countries in which all public health insurance usually spends more per person. This outcome spending goes through social insurance; even in countries simply reflects the fact that formal employment pays wages that rely heavily on insurance there is usually some direct, well above the average income, at least in low-income tax-based expenditure-if only for those public goods countries. that cannot be associated with individuals and therefore At middle and high incomes there is great variation in cannot be insured. how public spending is divided. In some countries one model The countries that rely most on social insurance-Bolivia, or the other dominates, while in others substantial public China, the Czech Republic, Denmark, Israel, the Republic resources flow through both the direct and the social insur- of Korea, Mexico, the Netherlands, Sweden-are a het- ance channel. Both systems are used in Australia, Latin erogeneous group. In some, social insurance dominates America, some countries in Eastern Europe and the Middle because almost everyone can pay wage taxes and it is pub- East, and the United States. There are two reasons to believe lic policy to use that model to finance health care. In oth- that this is an inefficient way to organize government finance TABLE 3 Public health care financing in various countries, circa 1990 (continued) (percent) Share of public health spending Public financing Total health financed directly as percentage Income spending as Region/country or by taxes of total per capital percentage of GDP Eastern Europe. Middle East, and North Africa Czech Republic 5 85 6,280 6 Hurgary 84 83 6,080 6 Turkey 58 38 4,840 4 Tunisia 63 67 4,690 5 Latin Amnenco ond the Caribbean Argentina 45 61 5,120 10 Bolivia 17 29 2,170 6 Brazil 50 43 5,240 6 Colombia 43 57 5,460 5 Costa Rica 19 82 5,100 9 Dominican Republic 77 34 3,080 6 Ecuador 62 63 4,140 4 El Salvador 32 30 2,110 6 Guatemala 61 33 3,180 5 Jamaica 100 35 3,670 9 Mexico 15 56 7,170 5 Nicaragua 100 62 2,550 8 Panama 41 60 4,910 9 Paraguay 31 25 3,420 4 Peru 59 34 3,130 3 Uruguay 81 76 6,670 8 Venezuela 75 47 8,120 4 a. Expressed in purchasing power party dollars. Source: Musgrove 1996. GOVERNMENT FINANCING OF HEALTH CARE 49 of health care, and that it is advantageous to have a single At least in high-income countries, coverage by publicly government program. First, it is easier to control the total financed care is unrelated to whether one or both models flow of resources if there is only one channel (Reinhardt are used or in what proportions, because coverage is usu- 1992). Second, having one channel reduces the risk of ally almost universal. Among OECD countries the only suboptimization through incomplete coordination between exceptions are Mexico (which has much lower income than the different systems; for example, Sweden's combination the other members), the Netherlands (where high-income of an open-ended reimbursement system for prescription people do not pay the contributions but buy private insur- drugs and global budgets for hospital care promotes a trans- ance), and the United States (which has two substantial fer of costs from inpatient to outpatient care. To avoid public programs, one of each type, limited to the elderly such problems, governments must take a comprehensive and the poor). Coverage is harder to estimate in poorer view of their role in financing health care. countries, and there may be some relation between bene- The share of public financing in total financing differs ficiaries of public finance and which model or combination among different kinds of health expenditures. The shares of models is followed; data are insufficient to support any of drugs and hospital care in total health spending in OECD conclusion on this point. countries are shown in table 4. Such detailed data are scarce in developing countries, but the public share in hospital Features and failings of the two systems financing is generally high and the public share in paying for drugs is rather low. The lower is the share of public expen- The great variation among countries, and even within them, diture in the total, the more it is likely to be concentrated strongly suggests that neither the tax-based nor the social on hospital care, independent of whether the financing is insurance model is a systematically superior way of paying tax-based or social insurance. A large share of out-of-pocket publidy for health care. It is easy to list the theoretical virtues spending typically goes for drugs, particularly where pri- of either system: for example, in principle direct, tax-based vate insurance is nonexistent or covers only a small share finance is easier to extend to everyone, and social insur- of the population. Differences in reimbursement for dif- ance makes people (at least those with formal employment) ferent health care services can encourage the use of ser- contribute proportionally to their ability to pay. But these vices-such as hospital care-that are not necessarily the potential advantages are inconclusive without an examina- most cost-effective. tion of the specific taxes used to finance direct payment, TABLE 4 or the coverage and benefits under one scheme or the other. Public expenditures for drugs and hospital care, And the theoretical benefits of one model are lost when selected OECD countries, 1983 and 1993 the two forms of financing are used together, which is com- (percentage of total spent on drugs and hospital care) mon. This is why there is no discernible pattern to public Drugs Hospital care health care financing in high-income countries. The greater Country 1983 1993 1983 1993 homogeneity among low-income countries that depend on direct finance results from the difficult task of raising suf- Austria 29 63 42 36 Belgium 54 60 66 68 ficient revenue from wages and salaries to finance care for Canada 7 27 88 86 any but a small share of the population. Denmark 44 49 100 100 One complexity of mixed systems is that there are many France n.a. 62 n.a. 9 1 Germany 68 61 85 85 partial public health insurance programs in various coun- iceland 64 69 100 100 tries that are limited to different groups or different treat- Italy 72 49 85 85 Netherlands 62 94 84 84 ments. Such programs illustrate several issues in the choice New Zealand 80 66 95 n.a. of public financing mechanisms and other characteristics of Sweden 72 69 na, n.a. public payment for health care. The best-known subsystems Unted Kingdom 65 63 n.a. n.a. United States 7 12 53 57 are Medicare and Medicaid in the United States. Medicare Source: OECD Health Database 1995. is a uniform federal program that provides compulsory INNOVATIONS IN HEALTH CARE FINANCING 50 hospital insurance to the elderly (part A) as well as optional incentive to save and an incentive to dispose of wealth supplementary medical coverage to which nearly all elderly through gifts. Medicaid covers nursing home costs, which subscribe (part B). Medicaid is a program operated by the are not covered by Medicare. People who need nursing states, with matching federal dollars but different criteria home care, and who do not have private funds or insurance, and benefits from one state to another, that finances health must shift to Medicaid for coverage. care for poor households that already receive benefits from either or both of two welfare programs unrelated to health. Provider reimbursement. Medicare and Medicaid were Some other countries also link preferential health insurance enacted within a system of private insurance, and initially for the poor to an existing non-health welfare program, to copied the way private insurance companies reimbursed avoid the need for another targeting mechanism (Grosh providers. However, unexpected large cost increases have 1992). led to changes in the way providers are paid, by making them share part of the financial risk through diagnostic Contributions. Medicare part A is financed through a 1.45 related group-based payment for hospitalization and through percent addition to each worker's social security tax for both capitation (HMO enrollment). Proposals have been made employers and employees.Thus it is clearly social insurance, to introduce global budgets to control doctors' fees. In differing from most such programs in other countries only Germany such budgets have been used since 1976 for hos- in its limitation to the elderly. Part B is voluntary and financed pital care, ambulatory care, and drugs, and in the United by premiums, just as with private insurance, except that Kingdom the system of indicative drug budgets is based on about 75 percent of the premium is subsidized with gen- the same idea. eral taxation-which makes it a directly financed system. As this summary indicates, non-universal public health Users pay a deductible and coinsurance, which in part finance may respond to important needs and attempt to A increases with increased expenditures. This is contrary combine equity with control over expenditures as part of to the recommendation derived from the theory of optimal overall efforts to resolve how govemment finances health insurance, of a large deductible and decreasing copayments care. In every country the elderly and the poor are likely to to provide protection against catastrophic financial risk be the neediest groups, although often in different senses rather than "first-dollar" coverage (Arrow 1963). In part B of "need"-the elderlyneed more care, independent of their copayments are constant, which also leaves consumers at financial status, and the poor need more help paying for severe financial risk. This lack of catastrophic coverage has care, independent of their health status. It is partly the desire created a market for "Medi-gap" private insurance, so the to satisfy different kinds of need that makes the choice of system operates with all three forms of collective financ- revenue sources and financing mechanisms so important ing, which is inefficient (Blomqvist and Johansson 1996). and complicated. Medicaid is financed by state general taxation with match- ing federal grants. It has no deductibles or proportional Do the Type and Level of Taxes Matter? coinsurance, but some nominal copayments. There is cat- astrophic protection, but no incentives to control moral haz- The two main issues concerning the combination of taxes ard or costs. (and compulsory social insurance contributions) used to finance health care are equity and efficiency. For equity, Benefits. Eligibility for Medicare depends on age or, for the important distinction is between financing and the use some non-elderly, on health status (such as permanent dis- of services. For efficiency, there is similarly a difference ability). In addition to beneficiaries of the two welfare pro- between how revenue is raised and howit is used. The choice grams, Medicaid enrollment can be extended to other of taxes may create inefficiency outside the health system, medically needy groups. Thus eligibility depends on income through welfare losses associated with obtaining the funds; and wealth. As a result patients may become eligible for or inside the system, through the effects of taxes on the Medicaid by spending down their assets, and so have a dis- behavior of providers and consumers. GOVERNMENT FINANCING OF HEALTH CARE 51 General principles on health care and fiscal policy tion to their ability, which in practice largely means in pro- portion to their income. To the extent that the need for Since health care is just one of many uses for public rev- and use of health care are not highly correlatedwith income, enue, it is important to consider whether public finance the- this principle favors subsidies from the rich to the poor, ory offers any principles about how to raise and spend money and probably corresponds to the most widely supported for health. This question is applicable whether health is paid notion of how to finance health care (van Doorslaer, Wagstaff, for from the same budget as all other uses, or there are and Rutten 1993). The notion of financing according to sub-budgets defined by specific taxes that are earmarked ability to pay can be interpreted in terms of both vertical or assigned to specific expenditures. equity-individuals with a greater ability to pay ought to pay more regardless of their health-and horizontal equity- The benefit principle. The benefit principle states that, individuals with the same ability to pay ought to pay the unless taxation is explicitly used for redistribution, the cost same regardless of their health. Because income differences of any public service should be borne by those who use it can be large, vertical equity and progressivity in health care (Musgrave 1959). This principle is similar to what happens financing have been studied more thoroughly. The pro- in a market-people pay in proportion to their consump- gressivity of a health care financing system refers to the tion-and can be applied to all individual or private goods extent to which people's payments for health care rise as a in health care. It is harder to apply to public goods, for which share of their income when their income rises. individual consumption is poorly defined and in any case Regardless of how it is financed, any insurance system difficult to measure. If it is assumed that everybody con- transfers resources from the healthy to the sick, but it does sumes the same amount, then people should pay accord- not necessarily make the rich subsidize the poor-depend- ing to their marginal valuation of the public good. However, ing on the sources of revenue and patterns of health care it is highly questionable whether even public health activi- utilization, the opposite can happen. It is difficult to design ties and emergency services can be considered public goods health finance systems, partly because both types of subsi- of which everybody consumes the same amount. If con- dies usually seem desirable for equity reasons but they sumers are heterogeneous, everybody should theoretically need not be consistent. And neither is consistent with the pay an individual tax. The information needed for such an benefit principle (paying for what one consumes). approach simply is not available at any cost, however, so the benefit principle cannot be followed exactly where pub - Earmarking, or linking specific revenues to specific expen- lic goods are concerned. ditures. For the most part there is no reason to relate a spe- Since the incidence of disease is to a large extent ran- cific tax to a specific area of public expenditure. It is more dom and everyone is at risk of needing health care, the rational to decide the optimal level of government spend- benefit principle suggests that the tax base for healh care ing and to decide how best to raise the necessary revenue financing should be as broad as possible to ensure that no without any expenditure being tied to any source (Musgrave one escapes contributing. This is an argument against financ- 1959). There is nothing unique to health about the choice ing health care with, say, tobacco or alcohol taxes, which of financing mechanism: the same considerations apply to have a rather narrow base. The fact that there may be exter- all areas of government spending. The choice of the tax nalities associated with behaviors such as drinking or smok- base-which taxes or contributions, at what levels and in ing is an argument for taxes to correct the behavior, but what proportions-is a problem of public finance, not health these taxes maybe used to finance other government expen- economics. ditures. However, it is common for social insurance systems to be financed primarily by earmarked taxes on wages and The ability-to-pay principle. At the opposite extreme from salaries, where the contribution is thought of as paying the the benefit principle (paying in proportion to one's use of insurance premium even though the premium differs accord- a service) is the notion that people should pay in propor- ing to income (rather than representing a market price). INNOVATIONS IN HEALTH CARE FINANCING 52 Moreover, in recent years there has been renewed interest States in 1980, 5 percent of the population accounted for in earmarked (hypothecated) taxes for health care (Jones 55 percent of health care expenditures (table 5). In a sim- and Duncan 1995). This interest has been sparked by con- ilar study for France in 1980-81, 5 percent of the popula- cerns about underfunding of health systems, as a result of tion consumed two-thirds of health care expenditures, while the general public finance problems that most governments 25 percent had no health care consumption at all (OECD experience. The immediate questions are whether ear- 1987). A Swedish study on the consumption of prescrip- marking in fact makes more funds available, and whether tion drugs found that 3.8 percent of consumers accounted it has other effects on the equity and efficiency of health for half of all costs, and that one-third of the population care spending. had no costs for prescription drugs. These large differences in per capita expenditures are Equity aspects of publicly financed health core related to age and the risk of mortality. In Sweden, for exam- ple, annual per capita health care expenditures are closely Defining equity is not easy Applied research on equity in linked to whether a patient died during the year or sur- the delivery of health care often focuses on horizontal vived (table 6). This trend is indicative of two health spend- equity-that is, whether people with equal needs (in terms ing characteristics in high-income countries: the tendency of morbidity) are treated the same (in terms of utilization) to consume more health care with advancing age (at least regardless of income (Wagstaff and van Doorslaer 1993). after infancy) and the tendency to have very large expen- A major argument for public financing of health care is to ditures in the final year of life. guarantee everyone, including people who are too poor to In retrospect this distribution of spending looks extremely buy private health insurance or to pay out-of-pocket for inefficient, because some people die despite having con- health care, access to a minimum package of health ser- sumed large amounts of care. But it is uncertainty about vices. People with the lowest incomes and wealth are gen- outcomes, not simply wastefulness, that leads to this result. erally in the worst health and thus have a greater need for Whether the distribution is equitable is hard to say, but the services. One reason is that both health and income are increasing expenditure with age does reflect increasing need often low among the elderly. Thus equity aspects are cen- as health deteriorates. In poor countries neither tendency tral to discussions of health care financing for individual is likely to be so marked, because fewer resources are avail- services. able and people are more likely to die without receiving sub- The equity argument may be less important for financ- stantial care. ing of public goods. Interventions to improve public health, The distribution of health care expenditures is neither such as sanitation and vaccination programs, generally ben- evidence of inequity nor a measure of service benefits. If it efit the entire population. Depending on the program, the were possible to examine consumption over a lifetime instead benefits may be more or less important for some groups. It of within a single year, the variation between people would can be argued that the poor generally gain the most, since people with more money can compensate by substituting TABLE S private for public expenditures. On the other hand, health States, 9 70c 1977 a and 1980 promotion activities may be of particular benefit for peo- (percentage of total expenditures) ple with more education and greater capacity to benefit from such activities. Such groups are often the first to change Population expense group 1970 1977 1980 their behavior to improve their health, as with reduced smok- Top I percent 26 27 29 ing and increased physical exercise. Top 2 percent 50 38 39 Top 10 percent 66 70 70 Distribution of health care expenditures in the population. Top 30 percent 88 90 90 Top 50 percent 96 97 96 The distribution of health care expenditures in a given year Bottom 50 percent 4 3 4 is highly concentrated among a few consumers. In the United Source: Berk, Monheit. and Hagan 1988. GOVERNMENT FINANCING OF HEALTH CARE 53 be much smaller, at least so long as they were of the same and makes them more dependent on public subsidies. age. What we really need to know to answer questions about Although such subsidies are seldom enough to equalize equity and efficiency is the lifetime consumption of health health care consumption across income levels, they can care and the resulting health status at different ages, and transfer substantial amounts of real income to the lower- how this differs according to income and other factors. But income deciles when coverage by public programs is high. improvement in health status over time and rapid techno- Van Doorslaer and Wagstaff (1993) used two methods logical change in health care make such comparisons mean- to measure income-related inequity in ten (mainly European) ingless for people of different ages. countries. The first method involved ranking individuals by income and comparing the cumulative health care expen- Income-related equity in the use of services. Another way diture (standardized for differences in morbidity) across of looking at the distribution of health care expenditures is income groups. The second method controlled for mor- to look at consumption in different income brackets. A study bidity by using regression analysis to test for significant of the United Kingdom and the United States found that income effects on healh care received. They found that about one-tenth of health care spending occurs in each income-related inequity exists in most countries, and that income decile in both countries (table 7). This is rather it usually arises from the effect of income on the amount of surprising given the differences in health care financing and care received by people who use at least some health care. provision between the two countries: both income distrib- In fact, the effect of income on the probability of seeking ution and access to health insurance coverage are less equal care was significant in only one country (Denmark). This in the United States. result almost surely does not apply to developing coun- In other words, in rich countries health care consump- tries, where income is a major determinant of whether peo- tion is almost independent of income, thanks to near-uni- ple obtain care-not only because of the price of care, but versal coverage of insurance and especially of public because poor people often live far from the nearest doctor insurance. In low- and middle-income countries with a less or clinic (particularly in rural areas), and travel time and even distribution of income, the lowest deciles are unable cost are major deterrents to seeking care (Gertler and van to afford much health care, and insurance coverage is typ- der Gaag 1990). ically much less complete. For both reasons there are usu- Data on the distribution of health care expenditures in ally sizable differences in health care spending between high- the population are important to understanding the oppor- and low-income eamers. The poor are naturally more sen- tunities and limitations for financing health care through sitive to prices than the rich, which greatly limits their use direct payments, and for assessing the need for government of full-cost private services (Gertler and van der Gaag 1990) TABLE 7 TABLE 6 Distribution of health care expenditures Annual health care expenditures per capita in (consumption) by income decile, United Kingdom different age and mortality groups, Sweden, 1994 and United States (kronor) (percent) Died during Income decile United Kingdom United States Age group Survivors the year I 9.1 8.9 2 11.0 10.3 0-4 9,900 410,000 3 11.2 8.8 5-14 4,000 133,000 4 10.4 9.6 15-44 7,000 186,000 1 0.3 9.5 45-64 9,300 174,000 6 9.5 9.2 65-74 16,000 141,000 6 9.5 9.3 75-84 22,000 93,000 8 8.8 10.8 85 and above 23,000 61,000 9.7 11.7 Total 9,200 106,000 10 10.4 11.9 Source: SOU data. Source: Gottschalk, WoHfe, and Haveman 1989. INNOVATIONS IN HEALTH CARE FINANCING 54 support to health insurance, to compensate for differences tions; such contributions are negligible in Denmark, Ireland, in age-related needs and in incomes. They are less infor- Portugal, and Switzerland. The balance between direct mative about the appropriate taxes to use to finance pub- and indirect taxation is less varied. lic health care, although the fact that expenditures usually The distinction among direct taxes, indirect taxes, and rise with age carries some important implications for how social insurance contributions is important, but further social insurance is financed, particularly as to whether it is distinctions are possible. Direct taxes can be applied at the funded or pay-as-you-go. Systems with pay-as-you-go financ- central, regional, and local levels. Indirect taxes can be ing are vulnerable to changes in age composition as the pop- general, such as a value added tax (VAT), or on particular ulation ages; funded systems escape that that problem but goods, such as an excise tax. Social insurance contribu- are vulnerable to cost-raising technical change in medi- tions can be paid by employers or employees and be pro- cine. To have a better idea of which taxes are most appro- portional or have an upper limit (and thus be regressive). priate, we need to examine the taxes that are used and who Different taxes and their design have different consequences pays them. and must be assessed against the objectives of health care arnd fiscal policy. Distribution of income and tax payments: progressivity of Kakwani progressivity indexes for thirteen countries are financing. The way health care is financed affects people in shown in table 9. Direct taxes are the most progressive, a number of ways, and the distributional consequences are and indirect taxes are regressive, in all thirteen countries, difficult to assess. An increase in direct payments may reduce and are especially regressive in Spain and the United the use of health care, which probably has a greater effect Kingdom. Social insurance-financingis regressive in Germany on the poor, who are sick more often. Higher direct pay- and the Netherlands (which explains why total public finance ments also affect the distribution of disposable income, and is slightlyregressive in those countries) but is otherwise pro- thus other types of consumption. Depending on how tax- gressive, although much less so than direct taxes. Private payers are affected, increased public financing can also have direct (out-of-pocket) payments are strongly regressive, as several consequences. is to be expected since medical care needs are largely inde- In the absence of consensus on how much more the pendent of ability to pay. Because it spreads risk across better-off should pay than the worse-off, health care sys- income groups, private insurance is less regressive or even tems can be judged by the progressivity of the taxes used progressive, so total private payments are more equitably to finance them. A variety of indexes have been pro- distributed than those paid out of pocket. posed to measure progressivity (Lambert 1989), the most common of which is Kakwani's index (1977). This index Mix of public funding in OECD health care systems measures the extent to which a tax system departs from (percentage of total public funds) proportionality. The cumulative proportion of the popu- lation, ranked according to pretax income, is plotted Public financing against the cumulative proportion of tax payments to as a share obtain the tax concentration curve. A zero index means Direct Indirect Social Total of total the tax is exactly proportional to income; positive values County taxes taxes insurance public (percent) indicate progressivity (the rich pay a larger share of taxes Denmark 58 42 0 100 83 France' 0 3 97 100 78 than their share of income) and negative values, regres- Ireland 38 52 10 100 76 sivity. Italy 25 28 47 100 71 Detailed data on the mix of funding sources are avai- Pbrtugal 239 64 91 100 576 able for only a small number of countries (table 8). There Spain 10 8 82 100 79 are large differences among countries in the composition Switzerland 78 18 4 100 72 Unted Kingdom 44 36 20 100 84 of the taxes used to finance health care. France, the United States 52 15 33 100 44 Netherlands, and Spain rely on social insurance contribu- Source: van Doorslaer, Wagstaf and Rutten 1993. GOVERNMENT FINANCING OF HEALTH CARE 55 TABLE 9 Progressivity of components of health care financing (Kakwani indexes), selected countries Direct Indirect General Social Total Private Direct Total Total Country, year taxes taxes taxes insurance public insurance payments private payments Denmark, 1987 0.062 -0.113 0.038 0.000 0.038 0.031 -0.265 -0.241 -0.003 Finland, 1990 0.128 -0.097 0.056 0.090 0.060 0.000 -0.246 -0.246 0.017 France, 1989 0.000 0.000 0.000 0.094 0.094 -0.186 -0.228 -0.218 0.012 Germany, 1988 0.251 -0.092 0.112 -0.081 -0.040 0.093 -0.103 -0.022 -0.037 Ireland, 1987 0.267 n.a. n.a. 0.126 n.a. -0.021 -0.147 -0.096 n.a. Italy, 1991 0.161 -0.112 0.038 0.112 0.075 0.177 -0.077 -0.057 0.045 Netherlands, 1992 0.200 -0.089 0.071 -0.129 -0.100 0.083 -0.038 0.043 -0.070 Portugal, 1990 0.218 -0.035 0.060 0.185 0.072 0.137 -0.242 -0.229 -0.045 Spain, 1990 0.214 -0.152 0.048 0.050 0.049 -0.012 -0.212 -0.190 -0.003 Sweden, 1990 0.053 -0.083 0.036 0.010 0.030 -0.240 0.027 0.029 Switzerland, 1992 0.172 -0.072 0.131 0.038 0.113 -0.270 -0.403 -0.319 -0.165 United Kingdom, 1992 0.284 -0.152 0.046 0.187 0.079 0.077 -0.223 -0.092 0.052 United States, 1987 0.192 -0.065 0.124 0.019 0.090 -0.175 -0.461 -0.288 -0.131 n.a. is not available. Source: Wagstaff, van Doorslaer, and others 1996. Overall, the way health care is paid for in these coun- Efficiency aspects of publicly financed health care tries is nearly proportional to incomes. Since consumption of care is relatively independent of income, there is usually As with equity, efficiency is a concern for how revenues are a substantial net transfer from the rich to the poor when raised and how they are used. Put another way, the type benefits are compared with payments. The exception is and level of taxation used to finance health care may have Switzerland, where a large share of care is financed privately effects both inside and outside the sector. and regressively. Otherwise, the results do not seem to depend much on the exact combination of private finance Excess burden of taxation. Since people prefer not to pay and taxes. taxes, and since they usually have choices about employ- These findings probably do not all carry over to develop- ment and consumption that affect how much tax they pay, ing countries. Out-of-pocket spending in these countries is taxes affect economic behavior and hence the allocation of undoubtedlyregressive, as inhigh-income countries-in fact, resources. This impact is referred to as the excess burden of it is generally more so, since it may be the only form of taxation. The implication is that public financing comes at spending for the very poor, who do not have access to pub- a price. All taxes other than a lump-sum tax are associated lic subsidies. Private insurance spending is highly progres- with a welfare loss. Thus there is a tradeoff between effi- sive because only the rich buy it, directly or through their ciency-which generally calls for a small tax burden-and employers. Indirect taxes, which the poor do not escape, are equity-which calls for progressive subsidies and thus for slightly regressive. Direct taxes are not used much in devel- a larger share of income taken in taxes. oping countries because of the ease of evading and difficulty The optimal tax structure is one that maximizes society's of collecting them; thus they may be less progressive than in welfare, where the balance between deadweight loss and industrial countries. Social security payments can be regres- equity reflects attitudes toward the competing goals of sive or progressive, since both coverage and the incomes to efficiency and equity. It is possible to estimate only the dead- which they apply vary greatly between countries. Thus wage weight loss; there is no theoretical basis for deciding how taxes are probably progressive when coverage is low (10-20 much a particular improvement in equity is worth paying percent), even though the rich do not pay them, because only for. Although the theory of optimal taxation offers a few high-income workers are covered. They become less pro- simple insights, in practice there maybe disagreements about gressive or even regressive as coverage is extended, although values and about the empirical question of what the trade- even then the very poor do not pay them. offs are. Thus it is impossible to make recommendations INNOVATIONS IN HEALTH CARE FINANCING 56 about the best mechanisms for financing health care from of mixed systems in which the public sector provides some theory alone. It is also important to consider the problem type of compulsory and universal plan but the private sec- of second best-that is, the design of government policies tor is allowed to offer complementary coverage. Such an in situation where there are important distortions that can- arrangement might allow public funds to be concentrated not be removed. on the poor without having to raise more resources gener- A study for Sweden concluded that the excess burden ally, but there is no consensus about the overall effect on is lower for a payroll tax or value added tax than for an efficiency. This is true even when the argument over the income tax, because an income tax is easier to evade introduction of government catastrophic insurance in a (Hansson 1984). Raising 1 Swedish krona (SEK) through system of private insurance (and how such a mixed system a payroll or value added tax costs SEK 2.30 if the revenue would compare with a purely private competitive insur- is used for transfers and SEK 1.70 if the revenue is used ance market or an optimally designed government monop- for public consumption, making the excess burden SEK oly plan) is limited to the issue of moral hazard and excess 1.30 for transfers and SEK 0.70 for public consumption. consumption of health care (Besley 1989; Selden 1993; For an income tax the excess burden is SEK 3.00 for trans- Blomqvist and Johansson 1996). fers and SEK 2.30 for public consumption. These are much higher estimates than earlier ones for the United States, Effects on overallfunding. Government finance may influ- which suggested that a tax on labor income, instead of a ence total spending on health care in different ways. The lump-sum tax, would cost only 2.5 percent of revenue raised traditional argument has been that the introduction of in deadweight loss (Harberger 1964). They are, however, public insurance, as with private insurance, will increase close to recent estimates on the deadweight loss of the spending because of moral hazard-particularly when gov- income tax in the United States, which include the effects ernment finances care through tax subsidies on health insur- of tax avoidance through changes in the form of compen- ance premiums and stimulates overinsurance. The optimal sation (such as employer-paid health insurance or housing expenditure on health would include insurance, but the that is not be counted as income). Feldstein (1995) esti- costs of that insurance should be recognized by the people mates that a proportional rise in all personal income tax paying for it, and they should make rational decisions about rates involves a deadweight loss of $2 per incremental $1 how much of it to buy. in revenue. Another argument is that in health care systems funded Comparable estimates of the cost of raising tax revenue through direct taxation, there is a risk that expenditures are scarce or nonexistent in developing countries, but the may be too low. This could happen because public goods possibilities for evading taxes and changing economic behav- and externalities are undervalued by the public, or because ior to escape taxation are at least as great as in industrial people are unwilling to pay the optimal level of taxes if countries. One particularly important escape is self-employ- they do not think the resources are being used to buy iden- ment in the autonomous or informal sector, where taxes tifiable insurance for themselves. This possibility has revived on income are nearly impossible to collect. The difficulties discussion of earmarked taxes, which dropped from the of raising revenue explain both why governments pay for a mainstream of public finance many years ago. A number of smaller share of health care in most developing countries proposals have recently been put forward, particularly in and why they rely on indirect taxes (sales and excise taxes), the United Kingdom, for introducing a specific tax to finance import and export duties, and social security contribu- health care (Jones and Duncan 1995). tions. These difficulties also explain why, when a social insur- One reason for these proposals is surveys showing that ance scheme exists, it is rare for coverage to be universal, people favor increases in the scale of public financing (and and why there is often a struggle over using those resources provision) of health care. At the same time, some countries to subsidize health care for noncontributors. have cut public spending. Thus there seems to be little cor- Because of the problems of financing adequate public relation between public expenditures and the preferences insurance, increasing attention has been paid to the design of voters. However, surveys may ask questions in such a way GOVERNMENT FINANCING OF HEALTH CARE 57 that the respondents do not think of a real tradeoff between usually earmarked for health. There is little evidence that different types of spending. If the questions were formu- this approach makes more resources available for health, lated correctly, the discrepancy would be much smaller since central budgets can adjust the contributions from other (Eckerlund and others 1995). taxes. A second reason earmarked taxes might be attractive is From a theoretical perspective there are strong argu- that people would accept tax increases for health, but not ments against using differential commodity taxes for financ- for other areas. Earmarking provides greater transparency ing. First, if there is a well-designed income tax, differential and responsiveness to the preferences of voters and thus commodity taxation is likely to add little, if anything, to the would help ensure that the resources are used for health ability to redistribute income. Of course, where income and nothing else. However, it is difficult to test the hypoth- taxes are poorly designed and hard to collect, there may be esis that people are more willing to pay taxes for health ser- a gain from excise taxes; this argument applies to develop- vices if they know the proceeds are earmarked for that ing countries. Second, differential taxation may be admin- purpose. Does earmarking make the government more istratively complex. Third, such taxes can be used to responsive to public preferences, or are the benefits real- discriminate against certain groups. The arguments against ized through behavioral responses by taxpayers (who con- differential commodity taxes are relevant for distortive taxes sume more of the taxed goods, declare their incomes more only honestly, and so on)? For corrective taxes, such as a tax on products with neg- An earmarked tax will not by itself determine the opti- ative externalities, the argument is different. However, the mal level of spending. In fact, it will have no effect on total theory of corrective taxes does not tell us anything about health care spending if revenues are insufficient to pay for how the revenue from these taxes should be used. One con- everything, and other tax contributions are lowered to off- sequence of corrective taxes is that if the behavior is affected set the earmarking. A designated tax is usually sufficient to in the desired way, the revenue for health services will be pay for a health care program only in the case of social insur- smaller. And the more people smoke and drink, the more ance. Even so, these contributions may not be strictly ear- money will be available for health care. Thus governments marked, since they often finance both health care and transfer may be ambivalent about reducing the behavior that brings payments such as pension or unemployment insurance. in revenue, especially if any gains from reduced health expen- Using the same taxes to finance health care and pensions diture will only materialize in the future. And health gains has caused severe problems in Latin America because pen- do not necessarily mean reduced lifetime expenditure on sions usually get priority, causing severe financial crises for health care. health spending when revenues are inadequate for both (McGreevey 1990). Earmarking a tax for two uses is of lit- Effects on the allocation of resources within the health care tle use unless the proportions are specified. sector. It is often advocated, or hoped, that a change in how A third reason is that in some health care systems taxes health care is financed will increase efficiency in the allo- can be described as earmarked for health, although the cation of health care resources. Some of the proposals for correspondence is not necessarily one-to-one. Sweden's earmarked taxes can be interpreted in that way. But in the- county councils levy a regional income tax that is used mainly ory there is no specific link between the way money is raised to finance health care. But the councils also spend the money and the way it is spent. It is possible to combine different on other areas, and part of their revenue comes as grants mechanisms for allocating resources in the health care sys- from the central government. In such cases where the tax tem and different ways of providing government funds. is almost completely earmarked, offsetting adjustments in Moreover, the level of spending is not in any significant other revenue sources are likely to be small, and the link way determined by the level or composition of public finance between sources and uses is effective. Earmarked taxes (OECD 1987, annex A). This observation does not rule out are common in Latin America, especially "sin taxes" on alco- that changes in health care financing mechanisms can be hol and tobacco, and part of the proceeds of lotteries are an important element of health care reform in a particular INNOVATIONS IN HEALTH CARE FINANCING 58 country at a particular point in time. However, general and the sources of revenue; and examine some of the prob- conclusions about the relation between efficiency and financ- lems affecting developing countries. ing mechanisms cannot be supported by theory or empiri- cal evidence. The current situotion An exception to this generalization occurs when spe- cific sources or amounts of revenue are allocated not to Public finance is the main source of revenue for health overall health care spending but to specific programs or cat- care systems in most parts of the world. In high-income egories of spending, and the earmarking is not offset by countries public finance accounts for about 75 percent of changes in other sources of funds. This approach affects the the total. The main exception is the United States, where allocation of resources between individual medical care public finance covers only about 50 percent of health care (which is highly visible and therefore politically attractive) financing. Taking into account the high health care expen- and public goods whose benefits may not be perceived or ditures in the United States, however, the share of public appreciated by voters; it is an extension of the argument finance for health care relative to GDP is similar to that in that direct financing may be suboptimal for such goods. It other industrial countries. In developing countries public is also an example of the issues raised by public choice the- finance is less important, sometimes as little as 20 percent, ory (Buchanan 1963), which takes self-interest into account. and out-of-pocket payments are a larger share of the total. It does not follow that intrasectoral earmarking will be The composition of public finance varies considerably, optimal, but it may improve on the political allocation of particularly in high-income countries. France, Germany, and resources that would otherwise result (particularlyinadecen- the Netherlands rely mainly on social insurance, while tralized system, where one political level would choose dif- Canada, the Nordic countries, and the United Kingdom ferently than another). Colombia's current health system rely on general taxation. (The United States is again unusual, reform is an example of trying to use earmarking to improve in that it operates both kinds of publicly financed systems.) the overall allocation of resources while leaving decisions Social insurance is rare in developing countries, and it is about individual care to providers and insurers. common to find 100 percent direct financinginmanyAfrican and Asian countries. At intermediate incomes-as in most Conclusions and Recommendations Latin American and Caribbean as well as some Asian coun- tries-the shares are variable, with some countries using Direct out-of-pocket payments can solve part of the financ- both models and some relying primarily on social insurance. ing problem in health care, and are the natural way to pay Except for a few countries (Switzerland, the United for inexpensive goods and services. But health care can be States), private health insurance is a minor source of finance. so costly as to make direct payment infeasible, making risk In developing countries this is because most of the popu- sharing necessary. Private, competitive, voluntary insurance lation cannot afford private insurance; in industrial coun- exists for this reason and can finance a large share of health tries it is because the state has assumed most of the insurance care. Yet such insurance is unaffordable by the poor, dis- function. Private insurance usually complements public criminates against those most in need, and can be expen- insurance, and a country may finance health care in differ- sive to administer. For all these reasons, there remains a role ent ways according to the income, employment, age, and for government in health care financing. location of the insured and according to the setvices cov- The issues relating to how much and what form of gov- ered and their cost. emnment finance are best affect both equity and efficiency and include poverty, high-risk groups, and the difficulties Ro/e of government financing of predicting future needs and costs. We end with a brief summary of how health care is financed in the world today; Government health care financing serves several purposes. consider what conclusions can be drawn about the proper One is to ensure the provision of public goods. Most pub- role of government, the appropriate model for public finance, lic financing, however, is for private goods in the form of GOVERNMENT FINANCING OF HEALTH CARE 59 individual health care, particularly when they are costly. The just as they have little or no private insurance-is not evi- only reason for governments to finance inexpensive health dence to the contrary. It simply reflects the difficulty of care is to make available to the poor what the nonpoor can financing insurance for a mostly poor population, and espe- buy out of pocket. cially the problem of collecting the taxes on labor that usu- More generally, the main argument for government inter- ally finance social insurance. Explicit social insurance is vention is the desire to achieve universal access to health harder to organize and depends on more favorable eco- care. This argument is based both on efficiency (arising from nomic conditions, but that does not make it better or worse. market failures in the provision of private insurance) and For many countries the question is not which model to equity considerations (the financing of health care should follow so much as how they should interact when both be according to ability to pay, and the distribution of ser- already exist-social insurance for urban, formal wage labor vices according to need). The need to provide public goods and civil servants, and direct finance for the poor and infor- and to alleviate poverty are important reasons for govern- mal workers. Should contributors to social insurance help ments to pay for health care in developing countries; it is subsidize the tax-financed beneficiaries, who usually have public insurance for high-cost care that accounts for the lower incomes? Should different taxes be pooled so that higher share of public spending in industrial countries. social insurance can be more extensive and not depend only Variations in the general pattern of financing reflect differ- on labor taxes? And how should either model relate to pri- ences in the resources available to governments, the func- vate insurance? Several countries have experimented with tioning of markets, people's values, and the responsiveness hybrid models that distinguish contributions from subsi- of the political system to public preferences. There is little dies, in an attempt to meet the two government roles of pro- consensus or empirical evidence as to the "right" level of tecting the poor and guaranteeing and regulating insurance financing for health care, in total or by the state. for those who can afford it. Conclusions about how well any Government financing can be combined with many dif- system works cannot be derived from theory or general- ferent models for delivering health care, by private as well izations but depend on the specific arrangements for financ- as public providers. In particular, public finance does not ing, coverage, and benefits. imply public provision. As with private insurance, exten- sive government finance can create moral hazard and inef- Sources of revenue ficiency in the allocation of resources to health care as well as within the health care sector. However, there is no sim- According to the benefit principle, assuming that the pur- ple relation between the source of financing and efficiency pose of government intervention is to achieve universal cov- in health care provision and delivery. erage, the tax base should be as broad as possible. Specific taxes (such as excise duties) that only part of the popula- Models for govemment finonce tion will pay should not be used. Such taxes may be justi- fied to curb the use of products with negative externalities, The fact that some high-income countries follow the directly but there is no reason the revenues should be devoted to or tax-financed model while others rely on social insur- health care. ance, and that in many countries both systems operate at One way to achieve a broad tax base is to create a fund once, suggests that neither model is better than the other. that receives revenues from several taxes. This is what gen- There is no empirical evidence that the theoretical virtues eral revenue financing does, but the fund could combine taxes of either model (universality and a less distorting tax base in different proportions specifically to increase the stability in tax-based systems, transparency and ease of revenue of revenues and protect against cyclical variation. These out- collection in social insurance) lead to systematically better comes would be useful because the need for health care, health, lower costs, or any other sign of superiority. The particularly for public subsidies, is likely to be countercycli- fact that low- and middle-income countries tend to use only cal, increasing under adverse economic conditions. It is also direct finance and to have little or no social insurance- important that the revenue base grow at least as fast as the INNOVATIONS IN HEALTII CARE FINANCING 60 general economy, given the pressure to increase health expen- public finance, since they do not create any distortions. Such ditures as income grows. Income, value added, and sales taxes taxes are difficult in practice, however, and choosing among are good candidates. Indirect taxes, employer contributions different tax alternatives in terms of excess burden is diffi- to social insurance, corporation taxes, and other sources make cult. The fact that the burden associated with any tax it less clear to the public that there is a link between what increases with the marginal tax rate reinforces the argument they pay and the services they receive, which is what the for a broad base of revenues that may include many dif- benefit principle is about. Particularly with employer contri- ferent taxes at lower rates rather than a few taxes at higher butions there is a risk for misunderstanding about the inci- rates. Unfortunately, there are few estimates of marginal or dence of taxes, and consequently a tendency to overinsure. total deadweight losses due to particular taxes in develop- To achieve equity, government financing should be based ing countries, and the ranking of different taxes in indus- on ability to pay, which makes a case for proportional or trial countries may not apply to conditions of much greater progressive income taxes. Taxes on labor income are sec- poverty, informal employment, and easier tax evasion. ond-best in this respect, since capital is not taxed and the The deadweight losses associated with taxes limit the contribution is usually proportional to earnings only up to amount that can be raised for health care spending with- some ceiling. An income tax also has the advantage of being out making an economy worse off. If people are willing to transparent. The empirical evidence is that direct taxes are pay more for health care but not for public spending in generally more progressive than social insurance contribu- general, then earmarked taxes are an attractive alternative tions, which are more progressive than indirect taxes. In this for achieving a higher and more optimal allocation of respect, the same taxes that can be recommended for cre- resources to and between different health care services. ating a broad base for funding are also those most likely to Earmarked taxes are often used to pay part of government collect from people in proportion to their ability to pay. health care costs, but the arguments for such taxes are weak, Excise taxes on luxury consumption are an exception: they and there is no evidence that they lead to more or better are progressive but narrowly based. expenditure because the amounts raised are usually smali Government finance by any combination of taxes and and allocations from other taxes can be reduced to offset contributions is almost certain to be more progressive than them. Social insurance contributions are usually the only out-of-pocket payments. This finding implies that govern- earmarked taxes capable of paying for a health system. ments should finance as much health care spending as effi- Of course, another broad-based tax could be earmarked ciency considerations permit in developing countries, where for health care, and if the revenues were large enough and a large share of finance is direct payments by consumers. the tax were used to buy insurance with subsidies to the Public finance may be less progressive than private insur- poor, it might be possible both to guarantee adequate financ- ance financing, but that is not an argument in favor of pri- ing and to fulfill equity and efficiency criteria. For this pur- vate insurance if the progressivity is simply due to the fact pose, as well as for a general ranking of taxes with respect that only the rich buy private coverage. Public subsidies for to their distorting effects, a value added tax is more effi- insurance purchase by the poor (for example, through vouch- cient than a payroll tax (since it does not tax labor while ers) can in principle respect ability to pay while giving peo- ignoring capital), and relative to an income tax encourages ple a choice of insurers. However, subsidies through the saving and discourages consumption. A value added tax is tax system-deducting the price of insurance from income also harder to evade and, when taken in conjunction with for tax purposes-are regressive and should be avoided. the health benefit it provides, is clearly progressive (Fuchs They are also inefficient, stimulating the purchase of insur- 1996). ance without control over costs, coverage, or benefits. From an efficiency perspective, taxes should minimize Structure of insurance distortions to economic activity and encourage an optimal overall level of spending and an efficient provision of health In addition to deciding how much insurance to pay for and care. Theoretically, lump-sum taxes are the ideal source of for whom, and with which taxes, governments can also deter- GOVERNMENT FINANCING OF HEALTH CARE 61 mine how to share costs with patients and what economic resources should be spent on private curative care, with incentives to build into insurance. The same principles of some mix of consideration for costs (to provide catastrophic optimal insurance apply to public financing and private cov- protection), effectiveness (to ensure real health gains), and erage. Thus, for example, deductibles and coinsurance should response to needs perceived by the public (World Bank 1987 be introduced so that the insurance protects against the and 1993; Musgrove 1996). For very poor countries it is highest financial risks rather than leaving patients unpro- possible to design a basic package for which government tected after some limit. Public insurance generally does finance would be justified; for less poor countries the choice not respect these principles, and user charges are far from of what services to finance becomes more complex. optimal. Especially in developing countries, this situation What implications does an appropriate strategy have reflects the difficulty of discriminating according to ability for public finance in developing countries?The general prin- to pay as well as the information requirements for efficient ciples discussed above all apply, but three questions acquire protection. particular urgency when governments' capacities to finance Another conclusion is that social insurance that is lirn- a reasonable level of universal care are limited: ited to special groups (such as the elderly or the poor), dis- * How can insurance coverage be increased without incur- eases, or treatments leaves part of the population without ring perverse subsidies or overinsurance? coverage and increases the risk for suboptimization. Basing * How should users be charged for services that are pub- eligibility on income or wealth also creates incentives for licly subsidized (see Gertler and Hammer in this vol- inefficient behavior in the intertemporal allocation of funds. ume)? It is probably better to establish a single insurance cover- * How can governments ensure that benefits are concen- age and then reduce or waive payments by the poor. trated on the poor? If there is little alternative to charging users of govern- Problems of developing countries ment facilities and private providers that public finance sub- sidizes, then differential fees to protect the poor are crucial. Although public and private levels of spending on health In countries where high-income consumers carry private care differ substantially, the problems related to public spend- insurance but use public facilities for free, collecting the ing are similar in economies at different income levels. In cost of services is urgent for both equity and efficiency. most countries general budget constraints make it difficult And the taxes used to finance what is not charged to users to increase public spending on health care, whether financed need to be as progressive and free of distortions as eco- by debt, taxes, or reallocation from other sectors. High- nomic conditions allow. Tax expenditures in the form of income countries have nearly all slowed the rate of increase tax exemptions are even less advisable than in richer coun- in health spending in recent years, sometimes stabilizing tries. spending as a share of GNP In very low-income countries, Perhaps the most difficult question is how to extend however, expenditures are still so low that even if funds were catastrophic insurance coverage to people who cannot afford spent as cost-effectively as possible, they would meet only adequate, unsubsidized private insurance. The government the most critical health needs. And many middle-income can encourage the development of an efficient insurance countries face simultaneously rising expenditures and unmet market by imposing appropriate regulation on private insur- health care demands (World Bank 1993). ance, which often operates with little control. To avoid What can be done? More attention has been paid to what subsidizing individual health care for the better-off, the gov- developing countries should buy with their public health ernment should avoid direct and indirect subsidies for pri- care resources than to how those resources should be raised. vate health insurance. And to avoid cost escalation, These countries should concentrate spending on services compulsory insurance plans should include a deductible and that benefit society as a whole, particularly cost-effective coinsurance up to a ceiling. When insurance is subsidized public goods such as immunization, sanitation, health edu- for the poor, both the deductible and the coinsurance pay- cation, and control of vector-borne disease. Remaining ments must be lower, or the insurance cannot be used INNOVATIONS IN HEALTH CARE FINANCING 62 when it is most needed. The problem is the same as for Feldstein, Martin. 1973. "The Welfare Loss of Excess Health user fees generally. Insurance." Journal of Political Economy 81: 251-80. Concentrating spending on certain services, as suggested . 1995. "Tax Avoidance and the Deadweight Loss of the Cbonentratinghift spmendingurces ono Income Tax." NBER Working Paper 5055. National Bureau above, will shift some resources from the rich to the poor, of Economic Research, Cambridge, Mass. because the poor suffersomewhat more from diseases with Fuchs, Victor R. 1996. "Economics, Values, and Health Care cost-effective preventive or treatment measures. But the Reform." American Economic Review 86: 1-24. best way to ensure equity in government finance is through Gertler, Paul, and Jacques van der Gaag. 1990. The Wi'llingness to the financing mechanisms themselves-the combination of Pay for Medical Care: Evidence from Two Developing Countries. taxes, contributions, and fees that pay for services that the Baltimore, Md.: Johns Hopkins University Press for the World taxles , Bank. whole population may need. Gottschalk, P., B. Wolfe, and R. Haveman. 1989. "Health Care Improving public financing of health care will not solve Financing in the U.S., U.K. and the Netherlands: Distributional all the problems of the healh sector. Political decisions are Consequences." In A. Chiacone and K Messere, eds., Changes needed to ensure that funds are not only raised as efficiently in Revenue Structures. Detroit: Wayne State University and equitably as possible, but also are used for the poor Press/International Institute of Public Finance. and for services with public benefits. Nor are the suggested Grosh, Margaret. 1992. Administering Targeted Social Programs in Latin America: From Platitudes to Practice. A World Bank reforms simple to ilmntnRegional and Sectoral Study. Washington, D.C.: World Bank. to require much more information about insurance bene- Hansson I. 1984. "Marginal Cost of Public Funds for Different ficiaries, patients, providers, and services than governments TaxInstruments and Government Expenditures." Scandinavian are accustomed to dealing with. Current reforms in several Journal of Economics 86: 115-30. mniddle-income countries show just how important and dif- Harberger, Arnold C. 1964. "Taxation, Resource Allocation, and ficult this process is. Still, there is scope for improving how Welfare." In J. Due, ed., The Role of Direct and Indirect Taxes public resources forhealthcareareobtainedandin the Federal System. Princeton, NJ.: Princeton University public resources for health care are obtained and used, press and the potential benefits are very large. Jones, Andrew, and Alan Duncan. 1995. 'Hypothecated Health Taxes-An Evaluation of Recent Proposals." Office of Health References Economics, United Kingdom. J6nsson, Bengt. 1989. "What Can Americans Learn from Arrow, KennethJ. 1963. "Uncertainty and the Welfare Economics Europeans?" Health Care Financing Review (annual supple- of Medical Care." American Economic Review 53(5): 537-38. ment): 79-93. Berk, M.L., AC. Monheit, and M.M. Hagan. 1988. "How the U.S. -. 1996. "Comprende la r6forme des systemes de sant6 Spent Its Health Care Dollar: 192940."HealthAffairs 7(4): 46-60. (Making sense of health care reform)." In OECD, La Re'forme Besley, Timothy. 1989. "Publicly Provided Disaster Insurance for des Systemes de Sante'-La Volonte' de Changement Etudes de Health and the Control of Moral Hazard." Journal of Public Politique de Sante (OECD Health Policy Studies) 8. Paris. Economics 39: 141-56. J6nsson, Bengt, and U. Gerdtham. 1995. "Cost Sharing for Blomqvist, A. Forthcoming. "Optimal Non-linear Health Pharmaceuticals: The Swedish Reimbursement System." In N. Insurance." Journal of Health Economics. Mattison, ed., Sharing the Costs of Health: A Multi-country Blomqvist, A., and P-O Johansson. 1996. "Economic Efficiency Perspective. Basel: Pharmaceutical Partners for Better and Mixed Public/Private Insurance." Working paper 10. Healthcare. Stockholm School of Economics, Sweden. Kakwani, N.C. 1977. "Measurement of Tax Progressivity: An Buchanan, J.M. 1963. "The Economics of Earmarked Taxes." International Comparison." Economic Journal 87: 71-80. Journal of Political Economy 71: 457-69. Lambert, L. 1989. "Horizontal Equity: Measures in Search of a Eckerlund, I., M. Johannesson, and P-OJohansson, M. Tambour, Principle." National Tax Journal 42: 139-54. and N. Zethraeus. 1995. "Value for Money? A Contingent Lewis, Maureen. 1994. The Organization, Delivery and Financing Valuation Study of the Optimal Size of the Swedish Health of Health Care in Brazil: Agenda for the 90s." World Bank, Care Budget." Health Policy 34: 135-43. Latin America and the Caribbean Department 1, Human Feldman, R., andB. Dowd. 1991. "A New Estimate of the Welfare Resources Division, Washington, D.C. Loss of Excess Insurance." American Economic Review 81: Manning, WG., Joseph Newhouse, N. Duan, and others. 1987. 297-301. "Health Insurance and the Demand for Medical Care: Evidence GOVERNMENT FINANCING oF HEALTH CARE 63 from a Randomized Experiment."American Economic Review Spence, M., and Richard Zeckhauser. 1971. "Insurance, 77: 251-73. Information and IndividualAction." American Economic Review McGreevey, William. 1990. Social Security in Latin America: Issues Papers and Proceedings 61: 380-87. and Options for the Bank. World Bank Discussion Paper 110. van de Ven, W.P, FT. Schut, and Frans Rutten. 1994. "Forming Washington, D.C. and Reformiing the Market for Third Party Purchasing of Health Mesa-Lago, Carmnelo. 1992.Health Careforthe Poor in LatinAmerica Care." Social Science and Medicine 39: 1405-12. and the Caribbean. Washington, D.C.: Pan-American Health van Doorslaer, Eddy, and Adam Wagstaff. 1993. "Equity in the Organization and Inter-American Foundation. Finance of Health Care." In Eddy van Doorslaer, Adam Mirrlees, James A. 1971. "An Exploration in the Theory of Wagstaff, and Frans Rutten, eds., Equity in the Finance and Optimum Income Taxation." Review of Economic Studies 38: Delivery of Health Care: An International Perspective. Oxford: 175-208. Oxford University Press. Musgrave, Richard A. 1959. The Theory of Public Finance. New van Doorslaer, Eddy, Adam Wagstaff, and Frans Rutten. 1993. York: McGraw-Hill. Equity in the Finance and Delivery ofHealth Care: An Intemational Musgrove, Philip. 1996. Public and Private Roles in Ilealtb Care. Perspective. Oxford: Oxford University Press. World Bank Discussion Paper 339. Washington, D.C. Van Vliet, R. C., and W. P van de Ven. 1992. "Toward a Capitation Newhouse, J. P 1994. "Patients at Risk: Health Reform and Risk Formula for Competing Health Insurers: An EmpiricalAnalysis." Adjustment." Health Affairs 13(1): 132-46. Social Science and Medicine 34(9): 1035-48. O'Donnell, Propper, and Upward. 1993. "United Kingdom." In Wagstaff, Adam, and Eddy van Doorslaer. 1993. "Equity in the Eddy van Doorslaer, Adam Wagstaff, and Frans Rutten, eds., Delivery of Health Care: Methods and Findings." In Eddy Equity and Financein the Delivery ofHealth Care: An International van Doorslaer, Adam Wagstaff, and Frans Rutten, eds., Equity Perspective. Oxford: Oxford University Press. in the Finance and Delivery of Health Care: An International OECD (Organization for Economic Cooperation and Perspective. Oxford: Oxford University Press. Development). 1987. Financing and Delivering Health Care: A Wagstaff, Adam, Eddy van Doorslaer, and others. 1996. Comparative Analysis of the OECD Countries. Paris: OECD. "Preliminary Results on the Redistributive Effect of Health - . 1995. "New Directions in Health Policy." OECD Policy Care Financing in Twelve Countries." Paper presented at the Studies 7. Paris. Equity Project workshop, April, Berne. V Various years. Health Database. Paris. Weisbrod, Burton A. 1991. 'The Health Care Quadrilemma: An Pauly, Mark V 1968. "The Economics of Moral Hazard."American Essay on Technological Change, Insurance, Quality of Care, and Economic Review 58(3): 531-37. Cost Containment. "JournalofEconomicLiterature 29(2): 523-52. 1974. "Overinsurance and Public Provision of Insurance: World Bank. 1987. Financing Health Setvicesin Developing Countries: The Roles of Moral Hazard and Adverse Selection." Quarterly AnAgendaforReform. AWorld Bank Policy Study. Washington, Journal of Economics 88: 44-62. D.C. Phelps, C. E. 1986. "Large-Scale Tax Reform: The Example of . 1993. World Development Report 1993: Investing in Health. Employer-Paid Health Insurance Premiums." University of New York: Oxford University Press. Rochester Working Paper 35. Rochester, N.Y . 1995. "Brazil: A PovertyAssessment." Report 14323-BR. - 1992. Health Economics. New York: Harper Collins. Washington, D.C. Reinhardt, Uwe E. 1992. "Organizing the Health-Care Sector: Zeckhauser, Richard. 1970. "Medical Insurance: A Case Study of Regulatory versus Market Approaches. " Princeton University, the Tradeoff between Risk Spreading and Appropriate Woodrow Wilson School, Princeton, NJ. Incentives." Journal of Economic Theory 2: 10-26. Selden, T M. 1993. "Should the Government Provide Catastrophic Zweifel, P 1992. Bonus Options in Health Insurance. Norwell, Mass.: Insurance?" Journal of Public Economics 51: 241-47. Iluwer Academic Publishers. INNOVATIONS IN HEALTH CARE FINANCING 64 From Beveridge to Bismarck: Health Finance in the Russian Federation Igor Sheiman F or decades the Soviet health sector developed using The conceptual basis of the reform is regulated compe- the Beveridge model-that is, a tax-financed and tition, which combines market incentives and regulation in highly organized system with an emphasis on uni- the purchase and provision of health care. The Health versal access to comprehensive care. In the early 1990s a Insurance Act, passed in June 1991 and amended in early countrywide reform of the health care system began, the 1993, is the legal basis for health sector reform. The Act man- core of which was a transition to a payroll tax-based manda- dates universal coverage for all citizens, including a com- tory health insurance approach known as the Bismarck prehensive package of medical benefits defined by the basic model. This shift is not limited to the method of raising (national) and territorial programs of mandatory health insur- health revenue, and entails profound changes in the way ance. The system is financed by an earmarked payroll tax the system is managed and financed. This new approach is and general budget revenue. Employers make income-based a reaction to the negative features of the former command contributions to the newly created territorial mandatory and control system, the most important of which were chronic health insurance funds (3.2 percent of payroll) and the fed- underfunding, government dominance of health care finance eral mandatory health insurance fund (0.4 percent) to cover and provision, top-down noncontractual resource alloca- their employees. Local governments make contributions to tion, no consumer choice, and input-based funding of health the territorial funds for the nonworking population and care providers. These characteristics led to considerable directly finance a number of health programs and providers. inefficiencies and irrational structures in health care provi- The system is highly decentralized. Each of the sion. Moreover, comprehensive and universal coverage were Federation's eighty-eight regions (oblasts), with populations undermined by inadequate funding and inefficient utiliza- ranging from 500,000 to several million residents, is respon- tion of health resources. sible for its mandatory health insurance system. This The main goals of reform are to: approach reflects the country's general move toward polit- * Raise additional funds ical and economic decentralization. * Increase the intemal and allocative efficiency of health Each territorial fund pools the premiums and allocates care provision by moving toward a system based on con- them to insurers based on a weighted capitation formula. Some tractual relationships between health care consumers and oblasts have developed a pluralistic system of health care providers using new payment incentives purchasing with a few competing insurers; others use a one- * Enhance the quality of care and ensure consumer pro- purchaser model acting through the local branches of the ter- tection through third-party payers ritorial fund. Since community rating restricts risk selection, * Maintain social solidarity and equity while increasing competition among insurers is focused on increasing market consumer choice. share. Insurers, acting as third-party purchasers of health care, Igor Sheiman is an assistant professor at the Moscow Medical Academy and senior health economist at Kaiser Permanente International in Moscow. 65 contracthealth providers andpaythem accordingto the method (1996, p. 20) use 1990 as a benchmark for Russia and Central determined by each oblast's mandatory health insurance reg- Asia. Here 1992 is used-the last year before reform and ulation. Rates are set by multilateral tariff agreement with the a starting point for big structural changes in the Russian involvement of the oblast health commnittees, mandatory health economy. insurance fund, and medical associations (for details on the Considerable changes in health spending occurred dur- design see Sheiman 1994 and Klugman and Schieber 1996). ing 1992-95, with two opposing trends (table 1). During Three and a half years of reform implementation pro- 1992-94 the share of public health spending in the econ- vide some indicative outcomes that are the subject of hot omy increased substantially, from2.6to 4.1 percent of GDP debate in Russia. Evaluations of the reform range from deep In 1994 real public health expenditures were 24 percent frustration to high enthusiasm and often depend on the higher than in 1992. This increase is overstated, however, vested interests of the evaluators. because the GDP deflator seems to understate growth in The international community is increasingly interested the prices of health goods and services, particularly phar- in the process and outcomes of Russia's transition to manda- maceuticals. Still, even when a more reasonable price defla- tory health insurance. Although recent reports from the tor for medical goods and services is used, real health World Bank, the U.S. Agency for International Development spending in 1994 was 20 percent higher than in 1992. (USAID), and other donors evaluate health reforms in a Such growth was a rare positive trend in a declining econ- number of countries in Central and Eastern Europe and omy, however. Mandatory healh insurance started at a time Central Asia (Ensor 1993; Klugman and Schieber 1996; of relatively good funding and high hopes. But by 1995 the Langenbrunner and others 1996; Goldstein and others opposite trend had started, and is still under way. Health 1996), their multicountry approach makes generalizations spending has fallen in both absolute and relative terms. The difficult and sometimes misses essential characteristics of share of health spending in GDP dropped to 3.3 percent health reform in individual countries. in 1995 and, according to preliminary estimates, to 3.1 This paper reviews recent developments in the Russian percent in the first nine months of 1996. Real health expen- health sector and appraises their impact using traditional ditures fell to pre-reform levels. Thus the drop in health criteria. Thus the emphasis is on the reform's successes spending was more substantial than was the decline in the and failures in securing additional funding, achieving higher country's overall economy. As a result the overall trend for equity, providing efficient and quality care, and overcom- the entire period is positive in terms of the relative weight ing structural distortions. The main problems of the reform of public health expenditures but negative in terms of the and potential ways of resolving them are then discussed. absolute volume of real health expenditures (with a 10 The paper concludes with health policy implications for percent drop). transition economies. Expectations for mandatory health insurance were based on two assumptions. First, that local governments, after being The Reform's Impact on Health Finance liberated from centralized resource allocation, would reori- ent their budgets toward health and other social services. Current debates on health care reform focus on how to raise Second, that economic decline would not last long. Both additional revenues to strengthen the health sector. This assumptions were wrong. Regional and local governments section assesses the new revenue-raising model's effects on health revenues and identifies the relative contribution TABLE I of different sources of finance. Public health spending in the Russian Federation, 1992-95 Public health expenditure trends 1992 1993 1994 1995 Public spending/GDP 2.60 3.77 4.08 3.29 It is important to choose the right benchmark when esti- Real health expenditure index 100 135 124 90 mating trends in health expenditures. Klugman and Schieber Source: Estimate based on Korchagin 1996 and Shishkin 1996. INNOVATIONS IN HEALTH CARE FINANCING 66 face increasing fiscal difficulties and often cannot afford their Netherlands, and 78 percent in Germany. Even in Belgium, new health care responsibilities. They also tend to support where state subsidies are more important, the payroll tax is agriculture, new road construction, and other more tangi- still the main source of revenue-58 percent. ble and visible services and projects. The economy has been The reason the payroll tax accounts for such a small declining for six years, and only in recent months have there percentage of public health revenue in Russia is that, accord- been signs of improvement. As a result much less money ing to the legislation, workers are not supposed to contribute has been infused into the health sector than was anticipated. to mandatory health insurance. Thus, contrary to declara- Still, the decline in real public health expenditures should tions about the transition to a mandatory health insurance be seen in the context of changes in other sectors. A 10 model, the health system is still financed mainly by taxes. percent drop in real spending is somewhat tolerable given Moreover, local governments are reluctant to allocate that education has lost 27 percent of its funding and cul- resources to mandatory health insurance funds. Their con- ture has lost 31 percent. In relative terms the health sector tributions cover just 24 percent of mandatory health insur- is doing better than other social services, all of which have ance enrollees, although they are supposed to pay for 56 been fuhded for decades using the residual method-that percent of the population. In twenty-three regions local gov- is, after priority allocations to defense, agriculture, and other emments make no contributions for nonworking popula- major sectors have been made. The new payroll tax pro- tions (Federal Fund of Mandatory Health Insurance 1995; tected the health sector from a more dramatic fall. To be Kravchenko 1996)-a clear violation of the legislation, which more precise, substantially decreasing public resources have states that such contributions are mandatory and much be been redistributed toward the health sector with very little equal to per capita contributions for employees. The tran- value added for health care providers and their patients. sition to mandatory health insurance has given local gov- Comparisons with other transition economies support ernments an excuse to underfund the health system, with the view that the new payroll tax has helped stabilize the the share of health expenditure coming from local budgets health care system. For example, during 1990-94 all Central dropping from 18 percent in 1993 to 12 percent in 1996. Asian countries experienced a much more substantial drop in health care funding, in both absolute and relative terms, Total health expenditure structure than did Russia (Klugman and Schieber 1996, pp. 17, 20). Similarly, the shift to national health insurance systems has One of the goals of health reform in Russia was to enable had a positive impact on public health funding in Croatia, citizens with high living standards to purchase health care the Czech Republic, Hungary, and the Slovak Republic using voluntary health insurance and direct out-of-pocket (Goldstein and others 1996, p. 34). payments. This goal can be achieved by explicitly specify- ing the package of medical benefits under mandatory health Public health expenditure structure insurance-that is, by determining a clear-cut border between public and private finance. The relative contribution of the two sources of public health funding is shown in table 2. The contribution of the pay- TABLE 2 roll tax is increasing, but not to the extent that might have Public health revenue in the Russian Federation been expected. In 1995 it provided just 26 percent of pub- by source, 1992-95 lic health revenue; general budget revenue (mostly from (percent) local budgets) is still the main source of health finance. Source 1992 1993 1994 1995 This outcome is contrary to the design of health revenue General budget revenue 100.0 88.9 76.5 73.9 under the Bismarck model, in which a payroll tax is sup- Federal budget 17.8 9.5 9.1 6.3 posed to be the main source of revenue, with the govern- Regional and local budgets 82.2 79.4 67.4 67.6 Payroll tax - 11.1 23.5 26.1 ment providing small subsidies. A payroll tax provides 90 Total 100.0 100.0 100.0 100.0 percent of public health revenue in France and the Source: Estmate based on Korchagin 1996. FROM BEVERIDGE TO BISMARCK: HEALTH FINANCE IN THE RUSSIAN FEDERATION 67 I I The main source of private spending-direct payments The private sector's role in total health expenditure has by enterprises that run their own medical facilities-has been declining (table 3). The health system still lacks an fallen dramatically (by half over the past four years) as the effective mechanism to raise private money. The main rea- result of the economic crisis. This dedine in direct health son is that public commitments to the coverage, eligibility, provision by employers has not been compensated by the and comprehensiveness of health care under mandatory growth of voluntary health insurance. Moreover, direct social health insurance are too declarative and are not based on services from employers have decreased substantially an actuarial approach. The lack of specification for the pack- More than 500 private insurance companies underwrite age of medical benefits under mandatory health insurance voluntary insurance plans that provide enrollees with addi- also limits the development of formal private finance. tional services and access to the best hospitals and poly- In summary, the expectations of additional funding have clinics. But coverage is limited to 3-5 percent of health not been met. The payroll tax has brought insignificant care expenditures. Voluntary plans are popular in Moscow, growth to health sector resources. There has been no shift St. Petersburg, and other big industrial centers with well- in the formal public-private mix of health finance. Private equipped teaching hospitals and high living standards, but finance and provision are not structured as a component are less developed in the rest of the country. Most medical of national health policy, and have been developing in an facilities are not involved in these plans because insurers unregulated, chaotic manner. prefer to work with the medical elite. The role of formal private out-of-pocket health fund- Impact on Equity ing is negligible-no more than 1.5 percent of total health spending. Most private money comes to the health system The transition to mandatory health insurance raises the issue through the black market as under-the-table payments, par- of its effect on the equity of health care provision for dif- ticularly in big hospitals in urban areas. Patients are increas- ferent groups and on the redistribution of income. A pre- ingly charged for drugs, materials, and surgery. In addition, vailing attitude in the literature is that the general revenue most pharmaceuticals for outpatient use are purchased out approach ensures a higher degree of income redistribution of pocket, although some groups are exempted from direct and social solidarity than do social insurance models. Two payment. Thus private spending is substantially understated arguments are usually presented. First, an income tax is because it is impossible to measure informal private (out- more progressive than a payroll tax. Second, the Beveridge of-pocket) payments for items that are not fully covered model tends to pool risks more equitably than the Bismarck by national health finance statistics. There has been no model, which has trouble pooling risks among multiple sick- attempt to conduct a large-scale household expenditure ness funds (Ensor 1993; Klugman and Schieber 1996; WHO survey to assess the scale of out-of-pocket health 1996 ). expenditures. However, the effect a health care financing model has on equity is more dependent on the design of the system, TABLE 3 particularly the way resources are allocated, than on the Total health revenue in the Russian Federation by mode of fund raising. The Beveridge and Bismarck mod- source, 1992-95moeofudriigThBeeiganBsacko- (percent) els may or may not be designed to pool risks by equalizing financial resources across groups and areas. Health insur- Source 1992 1993 1994 1995 ance systems in Germany, the Netherlands, and elsewhere Public 73.5 84.5 88.3 83.5 used risk-adjusted capitation formulas to ensure equitable Tovtealb 1200 100.0 100.0 16050 allocation of resources (Van Vliet and Van de Ven 1992; Files Total revenue/GDP 3.7 4.5 4.8 3.9 and Murray 1995). Yet some countries with the Beveridge a. Sum of general budget revenue (federal, regional, and local) and mandatory health model have not made much progress in equalzing resources, insurance contnbutions of employers. and for seventy years the Russian health care system (using b. Sum of consumer health expenditure and direct health expenditure of enterprises. Source: Estimate based on Korchagin 1996. the Beveridge model) failed to achieve equitable health care INNOVATIONS IN HEALTH CARE FINANCING 68 provision for different groups. The elite of communist soci- thus forcing jobs into the black market or increasing the ety had access to high-quality medical facilities, employees number of unemployed. Although it is unrealistic to expect of big enterprises enjoyed high standards of health care in an increase in the tax burden, tax revenue sources could be the facilities owned by their enterprises, and high-income restructured, with a shift from a general to an earmarked groups had better access to the best medical facilities. health tax. The effect a transition to a payroll tax has on income The negative effect reform has had on equity in health redistribution also depends on a country's structure of tax- care provision is evident from data on the geographic dis- ation. If the earmarked tax replaces a highly progressive tribution of health funds. There has been a sharp turn from income tax, the impact on equity is clearly negative. But highly centralized, top-down resource allocation to exces- the replacement of highly regressive indirect taxes may sive decentralization of health finance and management. increase social solidarity. In Russia the structure of tax rev- The Ministry of Health does not subsidize the oblasts. enue is skewed toward indirect taxes. The value added tax Moreover, the subsidies provided to the oblasts by the provides 44.1 percenr of consolidated budget revenue and Ministry of Finance are not earmarked for health, and are excise taxes provide 19.8 percent, while the profit tax sup- usually allocated by oblast governments according to tra- plies 18.3 percent and personal income taxes just 2.6 per- ditional budget priorities. Given the great differences cent (Izvestia 1996). Given this taxation structure, the new between regions' industrial bases and taxation revenues, payroll taxes may have a neutral or even slightly progres- the gap between rich and poor regions is growing. Per capita sive effect. health expenditures range from 720,000 rubles in Moscow In addition, it is easier to collect payroll taxes than profit to 470,000 rubles in Far East oblasts to 130,000 in Northem taxesintransition economieswithweak tax collection capac- Caucuses oblasts (Kravchenko 1996, p. 54). ity, because it is harder for entrepreneurs to hide payroll To some extent these inequities are offset by the federal than profits. In Russia mandatory health insurance tax col- mandatory health insurance fund, which provides small sub- lection provides about 90 percent of expected revenue sidies to seventy of the eighty-eight oblasts. But these (Federal Fund of Mandatory Health Insurance 1995), while resources are insufficient to fund the equalization program. the rate of income tax collection is so low that in late 1996 At the oblast level, territorial mandatory health insur- a presidential decree on emergency measures to collect taxes ance funds provide some equalization of health funding was required. However, the amount of payroll taxes actu- across areas using capitation formulas to allocate resources ally collected is reduced by a number of factors inherent to to insurance carriers. However, the equalization capacity transition economies, such as nonpayment of formal wages of these funds is limited because they control less than and underreporting of actual wages, particularly in small one-third of health revenue. The remaining funding is allo- businesses. cated by local governments, with little or no risk pooling One feature unique to Russia is that the health payroll between local areas. Few oblasts collect enough taxes in tax has not increased the tax burden of employers. Instead, the mandatory health insurance fund to implement the it replaced a portion of the old-age payroll tax when it became equalization program. clear in 1993 that the pension fund had surplus funds. The lack of a uniform oblast funding policy and man- Although a basic package of medical benefits was estimated agement strategy has limited access to health care in rural to require 7.2 percent of payroll, the health sector receives areas and small towns. Major city governments have no only 3.6 percent; the rest is covered by general budget allo- interest in providing specialty care to residents from areas cations for health. The share of social costs in payroll is 41 where such capacity does not exist. They make their bud- percent-lower than in some Western European coun- gets with no regard to the needs of outside residents within tries, but. high for a transition economy that is starving for the oblast. investment resources. High labor costs have narrowed Growing inequity in the Russian health system cannot resources for investment and growth. Moreover, this bur- be attributed to the transition to mandatory health insur- den discourages employers from creating additional jobs, ance. It is the result of the excessive decentralization and FROM BEVERIDGE TO BISMARCK: HEALTH FINANCE IN THE RUSSIAN FEDERATION 69 fragmentation of health systems that started before imple- ing services, and increasing their workload-efforts that mentation of mandatory health insurance. Federal and ter- are essential to fulfilling contractual obligations, ensuring a ritorial mandatory health insurance funds are looking for surplus, and increasing the salary of medical personnel. ways to equalize financial resources, but they cannot offset The problem is that the scope of contracting is still rel- the isolationist strategies adopted by local governments. atively narrow, with big differences across regions. For exam- ple, in Samara oblast around 80 percent of health Mandatory Health Insurance, Contracting expenditures are contracted; in most oblasts no more than Arrangements, and the Efficiency and 30 percent are. Many health authorities are reluctant to sur- Quality of Care render control over resources and tend to allocate finan- cial resources on a noncontractual basis, making contracting The recent innovations in the Russian health sector reflect inconsistent. Contracting's effect on efficiency and quality the shift to contractual interactions between health fund- can be evaluated by examining new payment methods, the ing authorities and health care providers. Acting as third- new role of management information, mechanisms for qual- party purchasers of health care, insurers change the ity control and consumer protection, and administrative performance of providers. Although contractual arrange- costs. ments are possible under the Beveridge model, the scope for contracting is higher under the Bismarck model. First, New payment methods an independent insurer can interact with providers (or groups of providers) only through a contractual mechanism. Second, Contracting has encouraged the development of perfor- a third-party payer is more flexible in its purchasing policy, mance-related payment methods. Of hospitals working since it has no obligations to state-owned medical facilities under mandatory health insurance, 53 percent are paid by and may reject the services of inefficient providers. insurers based on rates for each inpatient diagnosis. Another In theory, the division between finance and provision is 7 percent are paid a flat rate according to the average rate clear-cut. Moreover, contracting has the potential to improve by specialty. These two methods are used for 6.2 million health sector performance by decentralizing management, cases-64 percent of inpatient cases. This approach has cre- improving health care planning and management, and ated incentives to increase the occupancy of hospitals and increasing local choice of health providers. In practice, the reduce the average length of stay. Length of stay in these effects of contracting depend on many preconditions, the hospitals is lower for eighteen cost categories than in hos- most important of which are the role of purchasers in encour- pitals that are still paid using traditional methods aging competition among providers, the design of contracts (Langenbrunner and others 1996, p. 164). and methods of payment, the adequacy and stability of fund- In outpatient care, 19.4 percent of polyclinics are paid ing, and the skills to manage the contracts (Savas and according to a capitation method, 14.0 percent for each Sheiman forthcoming). episode of outpatient care, 12.2 percent on a fee-for-ser- In Russia contractual relationships are a growing part of vice basis, and 10.0 percent using some combination of these the health system. The main outcome of this transition has (Sheiman, Shevski, and Zelkovitch 1996). Polyclinics work- been increased operational autonomy of providers. Providers ing under capitation tend to shift to provision of primary are still owned by the state but are now self-governing enti- care and preventive services, while fee-for-service payments ties that can sell their services to different purchasers. encourage them to increase the number of services. Fee- Hospitals and polyclinics can keep surplus funds, hire and for-service payments were used in Moscow for three years, fire medical personnel, reduce bed capacity, deploy new units resulted in overutilization of some services, and gave way to (like outpatient clinics), and determine employees' paywith- capitation. Still, the negative experience was helpful for both out authorization of the health administration. After seven insurers and polyclinics. Fee-for-service payments encour- decades of the Soviet health system, health managers now age the development of information systems that are used have an intefest in coll&ting management information, pric- to monitor payment and management. Moreover, the shift INNOVATIONS IN HEALTH CARE FINANCING 70 from input- to output-based indicators is helping to reduce not changed. In their reports, federal policyymakers still praise unnecessary bed capacity and to create outpatient clinics themselves for "maintaining and strengthening the network and other alternatives to costly inpatient care. of medical facilities." According to a recent statement by The effect of performance-related payment methods the Ministry of Health, an additional 34,500 hospital beds on allocative efficiency and the structure of health care is were put in place during 1993-95 (Tzaregorodtsev 1996). much more controversial. Decades of bureaucratic control There is evidence that some purchasers are interested over health care systems created substantial distortions in in modern cost containment mechanisms. For example, in the structure of health care provision that are unknown to Kemerovo oblast rates are regulated in order to avoid inap- Western countries. Estimates of inappropriate inpatient propriate admissions and strengthen primary care. As a result cases range from 20 to 35 percent. Inpatient care stays aver- the share of inpatient care spending in total spending age 3.7 days; in the United Kingdom the average is 2.0 days, dropped from 64 to 60 percent over the past three years. in the United States it is 1.2 days, and in the most successful To reduce excess capacity, Kemerovo's health committee managed care settings it is 0.3 days (OECD 1993; and mandatory health insurance insurers introduced uti- Kongstvedt 1993). lization management procedures, with an emphasis on reduc- The main cause of long stays is the excessive number of ing the use of hospital resources. Excess hospital capacity physicians and excess capacity in medical facilities. Such was identified and three hospitals were closed within a few capacity is nearly impossible to maintain in a financial cri- months. In Samara oblast the financial scheme of "poly- sis. Mandatory health insurance has not yet contributed to clinic as fund holder" stimulates primary care providers to the reduction of excess capacity and dismissal of unneeded assume the main burden of health care (Galkin 1995). The personnel. The failure to lower excess capacity can be attrib- scheme has helped lower the share of inpatient care expen- uted to a lack of skills in planning and utilization manage- ditures. Other regions are experimenting with different ment. The prevailing mode of interaction between health approaches to managed care in order to overcome struc- authorities and health providers is still input-based alloca- tural distortions. tion of financial resources. Insurers, as purchasers of care, The USAID-sponsored Zdrav reform program, with contract providers predominantly on a cost-per-case basis Abt Associates, Inc. as a major contractor, helped develop without properly analyzing utilization and planningvolumes managed care principles and techniques in six oblasts of care. Contracting is noncompetitive and lacks cost con- and several cites during 1994-96. Currently, Kaiser tainment mechanisms such as global budgeting, analysis of Permanente International and Boston University, as new appropriateness of inpatient cases, and so on. Cost and contractors, are implementing a new program focusing volume contracts, which imply planning forvolumes of care on the same mechanisms. Efforts to disseminate outcomes and linking that information to available resources, are not of the Zdrav reform program have started, as has involve- used. As a result limited resources are spread across too ment at the federal level through the Duma and the Ministry many medical facilities. of Health. This conservative policy is aggravated by the peculiari- Thus Russia has followed an approach used in many ties of rate setting. The prevailing approach is to exclude countries: startingwith relatively open-ended performance- utilities and some other fixed costs when calculating rates related payment methods and then replacing them with of payment. These expenditures are directly paid by local more advanced managed care approaches and cost con- governments. In other words, most providers do not pay tainment mechanisms. anything for heating and electricity. Thus they are not inter- ested in closing facilities, even if there is excess capacity. New role of management information The main reason for the conservative policy is a deeply rooted strategy to build the health system by establishing One of the most beneficial outcomes of mandatory health new facilities rather than to improve the performance of insurance has been a growing demand for management existing facilities. Despite financial crisis, this approach has information. When financial mechanisms changed, invest- FROM BEVERIDGE TO BISMARCK: HEALTH FINANCE IN THE RUSSIAN FEDERATION 71 ment in clinical and financial information systems increased. criticized for their emphasis on imposing penalties rather Nearly all territorial mandatory health insurance funds and than ensuring quality. There is growing interest in devel- insurers have established information systems. Health care oping quality assurance systems based on continuous qual- purchasing is increasingly based on data on utilization and ity improvement models. There is also interest in integrating costs across medical facilities, specialties, patient groups, such systems with payment mechanisms. and even diagnostic groups. Data on cross-boundary flows of patients are also collected. Administrative costs Physicians, health purchasers, and decisionmakers are increasingly using computer systems to communicate, mon- Most oblasts use a pluralistic model of purchasing with a itor, educate, acquire data, keep records, check bills, store number of insurers. Mandatoryhealth insurance funds spend information, analyze data, and support decisions. The sys- 2.6 percent of collected premiums on administration. tems add basic value to patient and managed care, provider- Independent insurers spend 3.9 percent, for a total of 6.5 payer contracting, and financial management. percent. Such levels are comparable to those in Westem countries (Poullier 1992). In addition, mandatory health Quality control and consumer protection insurance funds use (temporarily) free resources for bank- ing operations, which have high financial returns given gal- Mandatory health insurance funds and oblast health com- loping inflation and high interest rates (the investments mittees have developed medical-economic standards that made by mandatory health insurance funds are protected specify requirements for the process and outcome of health by regulation). Federal mandatory health insurance fund care for each diagnosis. Insurers have established units that managers claim that the revenue from these operations review the cases and impose sanctions on providers that exceeds administrative costs (Kravchenko 1996). This is violate standards. The units also identify cases of delayed only partly true, because premiums might have been used admission to hospitals and penalize polycinics for the delay. for financial operations not only by purchasers but also by Insurers also scrutinize inpatient cases where the length of providers (presumably with a smaller return due to an stay is substantially shorter than the norm specified by the absence of banking skills). standard. The scope of this work is growing. For example, It is hard to draw any generalizations about the size of in Kemerovo oblast 8 percent of inpatient cases are sub- administrative costs. These costs should clearly be taken ject to quality control by insurers. About 70 percent of into account when analyzing the effect mandatory health collected penalties are returned to medical facilities to sup- insurance has had on the performance of the health care port quality improvements (Kemerovo Department of system. It is also clear that Russia cannot afford 545 pri- Health 1995). vate insurers with luxurious offices and well-paid employ- Insurers are increasingly acting as champions of patients' ees. In general, multiple funds have higher administrative interests. They have set up special units that are respon- costs (Poullier 1992). sible for settling patients' claims (including court cases), Intense debates are underway on this issue. Amendments monitoring patient satisfaction, and recommending qual- to the Health Insurance Act have been submitted to the ity improvements. Independent expertise brings discipline Duma that would replace private insurers with mandatory into the system, making physicians improve their perfor- health insurance funds as major purchasers. An alternative mance. approach, taken in Moscow, is to decrease the number of The preoccupation with uniform requirements and norms insurers by merging small entities. Moscow's network of is one drawback of these innovations. In some cases these twenty-five companies is being merged to form eight larger requirements make physicians too defensive of their clini- ones that will share the mandatory health insurance mar- cal practice and impede innovations. Moreover, it is hard ket on a cartel basis. This approach will likely decrease for insurers to ensure compliance with standards when ser- administrative costs, but at the cost of less competition vices are severely underfunded. These efforts are also being among insurers. INNOVATIONS IN HEALTII CARE FINANCING 72 Implementation Issues governments pay for inpatient care), items of expenditure (utilities and equipment are paid for by the government, Obstacles to successful reform while other items are paid for by insurers), population groups, and specialties. These distinctions impede planning and man- Health reform in the Russian Federation faces many obsta- agement of care focused on more cost-effective arrange- cles. Some are objective, others are the result of poor design ments. In addition, providers act under different economic in the finance and management of the system. regimes: contractual (with insurers) and noncontractual (with local governments). This approach limits the application of Economic and political instability. Economic instability performance-related methods and so reduces the efficiency aggravates health system underfunding, making it difficult of the contractual part of provider performance. to introduce performance-related payment methods. Providers are responding to the new payment methods, Little or no marketpressure. The transition to health insur- but incentives in the informal market are much stronger- ance has introduced elements of competition, but mostly simple under-the table payments are easier than sophisti- on the purchasing rather than the provision side of the cated formal payment arrangements. market. Independent insurers compete vigorously to expand The unstable political situation also impedes radical the mandatory health insurance market. The main tools of changes in the system. Local policymakers may recognize competition are contracts with the best-equipped medical the need to close some facilities and fire some personnel, facilities, sound consumer protection procedures, and effi- but they are reluctant to do so because elections take place cient quality control. But little has been done to encourage twice a year. competition among providers. In primary care there is no altemative to large state-owned polyclinics; In the hospital Excessive public commitments. Given the current economic sector there are elements of competitive purchasing by insur- situation, the government should no longer try to deliver free ers, but they are neutralized by the policy of local govern- care to all groups. Excessive state guarantees distort eco- ments and health comrittees to support hospitals regardless nomic relationships in health care because unbalanced of their performance. mandatory health insurance programs hinder full-scale con- The private sector is the only area with substantial com- tracting between purchasers and providers. In addition, petition. Hospitals, polyclinics, and private health insurance free care is becoming more elusive and public discontent is companies are competing for patients and subscribers. growing. Having declared health care entirely free of charge, Voluntary health insurance plans are offering additional the government is losing its capacity to provide it to the services. groups that need it the most. The health bureaucracy tends to refer to the "constitutional rights" of citizens for free Lack of management capacity. Despite recent progress in care rather than look for better ways to attract private money. building capacity, purchasers still lack effective management skills. Contracting requires skills that were not needed under Poorly specified roles and responsibilities of the government direct public provision, including identifying cost-effective and mandatory health insurance funds. The establishment of medical interventions, planning volumes of care, and nego- mandatory health insurance funds and insurance carriers has tiating, evaluating, and monitoring providers' performance. divided the roles and responsibilities for health finance and These skills are especially needed at the middle and bot- provision, creating misunderstandings and conflicts between tom levels of the system, where capacity is extremely weak. old and new actors in the health care system. The system is fragrnented both vertically and horizontally. Local govem- The proposed strategy ments and mandatory health insurance funds in most oblasts separate their responsibilities for funding health across health The following points are drawn from "Health Finance and system subsectors (insurers maypay for outpatient care, local Management Reform Strategy in the Russian Federation," FROM BEVERIDGE TO BISMARCK: HEALTH FINANCE IN THE RUSSIAN FEDERATION 73 written in 1996 by a group of independent experts under the putting nonacute cases on a waiting list. People who want USAID Zdrav reform project, with Boston University as the to reduce their waiting times should pay for part or all of main contractor (Starodulov, Sheiman, and Zelkovitch 1996). medical service costs (mostly through the voluntary health insurance system). Reforming state commitments. State obligations to pro- vide free care should be balanced with available resources. Shzfting excess capacity to theprivatesector, especiallyforinpa- Declarations guaranteeing comprehensive health care must tient care in big cities. A precondition for implementation of give way to an actuarial approach that develops a package this approach is an efficient purchasing policy based on the of medical benefits under the basic mandatory health insur- analyses and planning capacity of both short- and long-term ance program. Major changes are needed to implement providers. This would require restructuring the hospital sec- the new approach. tor merging some facilities and changing their case mix. The strategy also proposes restructuring big polyclinics into free- Improving management of inpatient care, with an emphasis standing physician practices to increase competition between on avoiding inappropriate admissions and tightening require- physicians and phase out the least competent ones. ments on length of stay. This will require better planning of Some of these proposals have become part of the cur- volumes of inpatient care as the main component of the rent health policy agenda at the federal level. The Duma is purchasing policies of insurers and local governments. A reviewing proposed amendments to the Health Insurance package of medical benefits should be based on realistic Act providing for a new procedure to determine the volume estimates of utilization and identification of excess capac- of state guarantees. The basic mandatory health insurance ity. According to preliminary estimates, a 12-15 percent program for 1997 has been drafted by a group of experts drop in the number of bed-days (which can be achieved in based on the concepts of resource management and actu- 1997) would ease the most dramatic shortages of funding, arial approaches. As mentioned above, Kemerovo oblast is and thereby partly balance the basic program of manda- experimenting with planning health care volumes and devel- tory health insurance. oping more cost-effective alternatives. Adopting a new procedurefor establishing the basic manda- Integrating the system vertically and horizontally. The goal tory health insuranceprogram. The minimum social standard, of this effort is to centralize management and finance respon- set by the USAID program, must be approved by the sibilities sufficiently to integrate the oblast health care sys- Government and the Duma together with the premium size tem and equalize resources across local areas. Two and budget allocations required to support this standard. mechanisms are proposed. First, establishing an earmarked An analogous procedure of determining the state's duties federal fund for resource equalization based on mandatory should be accepted at the oblast level. The volume of state health insurance contributions as well as federal budget rev- budget allocations to implement the oblasts' mandatory health enues, with a clear-cut equalization formula and close coor- insurance program should be determined by subtracting dination between the federal mandatory health insurance employer premiums from the approved program cost. fund and Ministries of Health and Finance. Second, at the oblast level, centralizing the mandatory health insurance When estimating costs for the territorial mandatory health contributions of local governments and allocating most insurance programs, determining the level of mismatching with health revenue on a capitation basis. available resources and using the data to develop cost sharing. In addition, the multisource model of funding care To minimize negative social implications, it is appropriate providers through mandatory health insurance programs, to set limits on primary care and to provide free medical health committees, and local governments (by types of ser- services to people who cannot afford even minimum con- vice, budget items, and population groups) should be scrapped tributions. Specific forms of cost sharing have been pro- in favor of a one-channel model. All cash inflows to the pub- posed. There is also the option of rationing utilization by lic health system should be pooled at a level higher than INNOVATIONS IN HEALTH CARE FINANCING 74 individual care providers. Specifically, management of 70 to has induced positive changes that would been impossible 80 percent of the funds should be carried out by a single fund- under the former "frozen" system, with its emphasis on com- ing party-that is, competing insurers. Merging cash flows mand and control methods. Yet the experimentation of the will allow competing insurers to build up a reasonable care late 1980s, with health committees as purchasers of care, provider contracting model, work out efficient incentives, has not been supported by the health bureaucracy because improve the care delivery structure, and control utilization. of a feature inherent to the owners of medical facilities- a tendency to control providers directly and allocate resources Shifting to payroll tax as the main source of health revenue. on a noncontractual basis. Independent insurers as pur- With mandatory health insurance funds controlling a neg- chasers of care have started the transition from ahighlyinte- ligible portion of funding, it is hardly possible to claim that grated to a contractual mode of relationships with providers. a transition from the Beveridge to Bismarck model is tak- The reform has helped stabilize funding during a period ing place. Rather, a transitional model is emerging that has of economic crisis, encouraged a transition from input- to no analog in the world. Eventually it should be transformed output-based payment methods, introduced tougher qual- into either a tax-financed or a social insurance model. ity control and elements of consumer protection, and cre- The existing fund-raising scheme should be modified to ated new incentives to collect information and use it for emphasize earmarked taxes. Their share of total health decisionmaking. However, resource allocation is becoming revenues should increase substantially from the current 26 less equitable, the system is disintegrating vertically and hor- percent to at least 50 to 60 percent. To achieve this goal, izontally, and structural distortions in the health system the payroll tax should be increased, although the overall have not been addressed. Thus reform has not fully met orig- tax burden should be lowered (for example, by lowering inal expectations, but it has had some promising outcomes. income taxes). This implies a transition to the fund-raising The failures of reform can be traced to an unfavorable scheme used under the Bismarck model, with employees economic and political environment and to mistakes in the obtaining insurance for themselves and their dependents. design of the reform. The biggest mistake is the isolation Several arguments favor this shift: of mandatory health insurance from the overall health sys- * The health sector remains one of the last budget prior- tem. Because this approach makes the system less coher- ities. An earmarked payroll tax can be a reliable and ent, it impedes implementation of the strategy for phasing predictable source of health revenue, even in a declin- out excess capacity and overcoming structural distortions ing economy. in health care provision. • Less dependence on general budget revenues will allow The way health funds are raised cannot automatically local governments, health committees, and mandatory increase resources and change the performance of the health health insurance funds to clarify their roles and respon- system. Rather, the design of relationships between pur- sibilities. Collecting most revenues, mandatory health chasers and providers determine actual changes in the sys- insurance funds will serve as the main funding party, tem. At the core of the reform is a shift from an integrated responsible for the integrated purchasing policy to a contractual model. * A shift to a payroll tax as the main source of funding will The lessons of the Russian reform for other transition result in vertical integration of the health system, since economies can be summarized as follows: under current legislation the centralization of financial * Moving toward a payroll tax-based model may raise addi- resources can be achieved only through mandatory health tional funding, provided the economy is not declining insurance channels. too steeply and too long. But the change in the funding method cannot be seen as the main area of reform. It is Conclusion much more important to ensure the operational auton- omy of providers, to increase their support for health The transition to mandatory health insurance has had a con- policy objectives through performance-related payment troversial effect on health sector performance. Clearly, it methods and managed care mechanisms. FROM BEVERIDGE TO BISMARCK: HEALTII FINANCE IN THE RUSSIAN FEDERATION 75 * If the mandatory health insurance model is used as the Files, A., and M. Murray. 1995. "German Risk Structure basis of health reform, mechanisms should ensure that Compensation: Enhancing Equity and Effectiveness." Inquiry this model prevails and can be easily implemented. To 32 (fall): 300-9. Galkin, R. 1995. "Mi ne zhaluemsa na zhizn." Meditsinskaya Gazeta that end, payroll taxes should replace general taxes as 15(3): 2. the main source of health funds, supplemented by state Goldstein, Ellen, Alexander Preker, Olusoji Adeyi, and Gnanaraj subsidies (with little or no additional tax burden for Chellaraj. 1996. Trends in Health Status, Services, and Finance: employers); mandatory health insurance carriers should The Transition in Central and Eastern Europe. World Bank control most health revenue, with the government directly Technical Paper 341. Washington, D.C. responsible for a small portion of health care funding Izvestia. 4 December 1996, p. 4. and provision; the roles and responsiblities of manda Kemerovo Department of Health. 1995. QualityArsuranceofHealth and provision; the roles and responsibilities of manda- CaeKmro. ' ~~~~~~~~~~~~Care. Kemerovo. tory health insurance funds and the government should Klugman, jeni, and George Schieber. 1996. A Survey of Health be clearly specified; and coordination mechanisms should Reform in Central Asia. World Bank Technical Paper 344. be developed. One predominant model of raising and Washington, D.C. allocating funds may be the best way to avoid vertical Kongstvedt, P 1993. The Managed Health Care Handbook. 2nd ed. and horizontal disintegration of the health system. Washington, D.C. Korchagin, V 1996. "Sostoyanie finansirovaniya zdravoohrane- *Contracting should be the prevailing model of health ni"EcoiaZdaohnea6:0 nia.>} Economica Zdravoohranenia 6: 10. care finance, planning, and management. The share of Kravchenko, N. 1996. 'Analiz sostoyania finansirovania zdravoohra- revenue allocated directly by the government should be nenia v 1995." Economica Zdravoohranenia 4: 14-20. limited to clearly specified expenditures or services. Langenbrunner,J., A. Wouter, T. Makarova, and K. Quinn. 1996. Contracts should be designed in the context of health "Hospital Payment Policies and Reforms: Issues and Options policy objectives, the most important of which is to in Russia." The Journal of Health Administration Education 2 eliminate the structural disproportion and excess capac- (spring): 164. OECD (Organization for Economic Cooperation and Development). ity of inherited systems. 1993. Health Systems: Facts and Trends, 1960-1991. Vol. 1. Paris. * Regardless of the fund-raising model, purchasers should Poullier,J-P 1992. "Administrative Costs in Selected Industrialized encourage competition among providers by allowing Countries." Health Care Financing Review (summer): 14-20. more consumer choice, using competitive procedures Savas, S., and Igor Sheimnan. Forthcoming. "Contracting Models for contracting, and dismantling local monopolies wher- and Provider Competition in Europe. " In European Health Care Reforms. Geneva: World Health Organization. ever possible. In addition, purchasers should ensure an Reom.Gnv:WrdHatOgnito. ever posibe.nddtio,urcasrssholdensre Sheiman, Igor. 1994. "Forming the System of Health Insurance open selection of providers in order to fund them for inThe Russian Federation." Social Science and Medicine 39(10): the real value of their services, and carefully monitor and 1425-32. evaluate providers' performance. Sheiman, Igor, V Shevski, and R. Zelkovitch. 1996. "Methods of * Under the mandatory health insurance model it is crit- Payment for Outpatient Care." The Journal of Health ical to base state commitments on an actuarial approach Administration Education 2(spring): 182. Shishkin, S. 1996. "Ne samie bednie sredi bednih." Meditsinski rather than on declarations about free care for all. 'This vsnc2:3 vestnic 22: 3. requires planning of volumes of care, improving man- Starodulov, Vladirnir, Igor Sheiman, and Roman Zelkovitch. 1996. agement of health care, and making bold political deci- "Conceptia Reform upravlenia I finansirovaniazdravoohrane- sions about cost sharing or rationing of health care. nia Rossii." Meditsinski Vestnic 18. Tzaregorodtsev, A. 1996. "Ob itogah deatelnosti zdravoohranenia References Rossiskoi Federatsii, 1993-1995." Econiomica Zdravoohranenia 8: 5-6. Ensor, T. 1993. "Health System Reform in Former Socialist Van Vliet, R., andWVandeVen. 1992. "Toward a Budget Formula Countries of Europe. " International Journal of Health Planning for Competing Health Insurers." Social Science and Medicine and Management 8: 169-87. 34: 25-40. Federal Fund of Mandatory Health Insurance. 1995. "MHI in WHO (World Health Organization). 1996. Regional Health Care the Russian Federation: Figures and Facts." Moscow. Reforms: Analysis of Current Strategies. Copenhagen. INNOVATIONS IN HEALTII CARE FINANCING 76 Private Insurance: Principles and Practice Deborah J. Chollet and Maureen Lewis P rivate health insurance is a growing phenomenon ance, but such systems can be difficult to design and to in much of the world. Fueled by rising incomes and administer. Second, insurers may want to deny coverage to growing dissatisfaction with publicly financed (and people who are sick, and to limit coverage for high-cost con- often publicly delivered) health care services, private insur- ditions or services. By refusing people who are sick or by ance coverage typically begins among large company man- deterring them from seeking coverage, private insurance agers in white-collar industries, and in multinational can contribute to higher average costs in public financing companies with large numbers of employees. However, programs that enroll larger populations and serve as the informal insurance arrangements exist even in the poorest insurer of last resort. The selection of low-cost patients countries, reflecting a universal desire for financial protec- into private insurance can cause serious problems for the tion and, in some countries, access to better-quality health public financing system (or the health care delivery system care than is offered in the public system. that it finances) if it cannot readily adjust to higher average A private insurance market can offer a number of advan- costs-even when total costs decline. Governments can tages over a purely public system of health care financing. require private insurers to accept sick people, restrict how It can allow governments to develop and maintain smaller insurers price coverage, and require insurance plans to cover and targeted systems of health care financing to serve peo- various types of high-cost health care. However, insurers ple who do not have access to private insurance. Private may be unwilling to enter or remain in markets with such insurance can help health care providers (private and pub- requirements. lic) rebuild infrastructure and amortize needed investment Other problems with private insurance derive from the when government payments for health care are inadequate. complexity of insurance contracts. Consumers typically do In the best cases private insurance encourages health care not understand many aspects of insurance contracts. Because systems-both the financing and the delivery of care-to consumers usually are unable to detect which insurers are innovate and to become more efficient, and offers a point unscrupulous or financially unsound, the market can attract of reference for improving the quality and efficiency of these insurers. Consequently, insurance regulators play an care in the public system. Typically encumbered by politics essential role in stabilizing a competitive private insurance and bureaucracy, governments may find it difficult to inno- system. However, insurance regulation is more art than vate without a private market to lead the way. science. It should encourage innovations that improve effi- Despite these advantages, private insurance presents a ciency and service, but it must require financial integrity number of problems. First, health insurance can be unaf- and discourage practices that threaten the stability and effec- fordable for low-income people. In principle, governments tiveness of the health insurance system-and, in turn, the can develop subsidies to help low-income people buy insur- health care system that it supports. Deborah J. Chollet is associate director at the Alpha Center in Washington, D.C. Maureen Lewis is principal economist in the Human and Social Development Group of the Latin America and the Caribbean Regional Office at the World Bank. The authors are grateful to Jeffrey Hammer, Gerard La Forgia, Jack Langenbruner, and Len Nichols for helpful comments. 77 This paper has three sections. The first section reviews The second section addresses the extent of private insur- private health insurance principles-what private insurance ance coverage in developing countries, summarizes selected is, how it works, and why insurance practices tend to evolve countries' experience with private insurance, and describes in particular ways. It then discusses the role of government emerging regulation in these countries. regulation-specifically, how regulation can stabilize and The final section offers conclusions and a number of guide the performance of private insurance markets-draw- lessons for developing effective health insurance regulation. ing on examples from OECD countries (especially the By melding theory, practice, and experience, we hope to United States, which relies the most heavily on voluntary provide a context for evaluating the role of private health competitive private health insurance) and developing insurance and for designing effective systems of health insur- counuties. ance regulation in all countries. Principles of Private Health Insurance Markets and Regulation Like all forms of insurance, health insurance is a system of will tend to drive up the quality and drive down the price protection against financial loss. In a health insurance sys- of available insurance. But different prices are often asso- tem a group of individuals agree to pay certain sums for a ciated with a range of plans that provide different bene- guarantee that they will be compensated for costs related fits and may be difficult to compare. People with low to the use of specific kinds of health care. Formal health incomes are unlikely to be able to afford any insurance at insurance contracts typically stipulate that covered health all, much less insurance that offers them access to com- care services must be medically necessary and provided by prehensive health care. appropriately trained health care professionals. Because By comparison, in a public financing system consumers health providers generally prefer to see patients who are typically do not choose among health plans. Moreover, pub- insured rather than risk nonpayment for care, in many coun- lic financing may or may not allow patients to choose among tries having health insurance (private or public) is equated hospitals, physicians, and other service providers. A public with having access to health care. system typically is financed in large part through taxes that The premise of health insurance is simple: individual are unrelated to the use of health care but that maybe related health care needs can be unpredictable and costly, but rel- to ability to pay. Thus differences in the quality of coverage atively few people need health care at any particular time. may not be a problem. Instead, problems with public financ- Thus, by pooling the risk of large health care expenditures ing usually derive from the absence of competition and, over many people, health insurance can make necessary therefore, the absence of incentives for public systems to health care affordable to all. respond to consumers. Public systems typically have little Private health insurance systems differ from public incentive to continuously improve quality and customer ser- financing systems in several ways. Most important, a pri- vice and also contain costs. vate insurance plan typically competes for customers, either During the past two decades managed care plans have with a public system or with other private health insurance emerged in a number of countries. Managed care plans com- plans. Thus consumers may choose among plans that have bine the financing and delivery of health care in the same different features (more or less financial protection, dif- contract, offering enrollees both insurance protection and ferent access to physicians and hospitals, a greater empha- a prescribed network of health care providers. In industrial sis on customer service and satisfaction, and so on). Typically, countries managed care plans have been offered as a lower- these plans also have different prices, and consumers have cost alternative to fnancial insurance plans that do not con- to decide if a more desirable insurance plan is worth its strain participants' choice of provider. But as enrollment in higher price. In principle, competition among health plans managed care plans has grown, some plans are also striv- INNOVATIONS IN HEALTH CARE FINANCING 78 ing to be known as a better-quality alternative to the frag- healthy and high-income people buy private insurance mented, fee-for-service system of health care delivery. Unless and others rely on the public system. otherwise indicated, in this paper health insurance includes Supplemental coverage for services not covered by a uni- both financial insurance plans (which pay for covered ser- versalpublic insuranceprogram. For example, in the United vices from any qualified provider) and managed care plans Kingdom, where the public insurance program is popu- (which pay only for covered services that are delivered by lar and provides comprehensive coverage, people may providers who are under contract to the plan). buy private insurance to finance care from specialists in private practice, "jumping the queue" for specialty care Roles of Private Insurance in the public program. Similarly, in Brazil no one may withdraw from the public system, but some people buy Most countries have a private health insurance sector. In private insurance to get more timely or higher-quality general, private insurance tends to emerge when the pub- care in the public system. In Australia private insurance lic financing system is perceived as financing lower-quality pays only for hospital care, either in private facilities care (usually also restricting patients' choice of provider) (which offer patients a choice among physicians) or in or covers only some types of health care. Reflecting the public facilities. In the United States enrollees in diverse reasons that a private insurance sector would emerge, Medicare, the social insurance program for the elderly the role of private insurance varies widely among countries and the disabled, can buy private supplemental cover- that allow or encourage it. In general, these roles are of three age to pay for the public plan's extensive deductibles and types: coinsurance amounts, and to pay for major items (such * Coverageforpeople who are ineligibleforpublic insurance. as prescription drugs) that are not covered by the pub- For example, in the United States private insurance is lic plan. About one-third of retirees have a private considered the main source of coverage, while public Medicare supplement insurance plan. insurance is intended to cover groups whom the pri- Annex tables 1 and 2 provide additional detail about the vate insurance market is likely to fail-the elderly (in alternative roles of private insurance in selected countries. Medicare) and people who are unable to work (chil- In many countries a large portion of health care spend- dren, the elderly, and the disabled) and poor (in ing is financed privately, either through insurance or out of Medicaid). People who are ineligible for public insur- pocket. But rarely does private insurance finance most health ance do not always buy private insurance. Some rely on care use. In the United States, where private health insur- public hospitals (funded by local government) for care. ance is unusually well developed (but purchase is volun- In communities that do not have a public hospital, peo- tary), private insurance financed only about 37 percent of ple without insurance may be unable to obtain routine all personal health care spending in 1994; 59 percent was care. publicly financed. The Republic of Korea is unique in that * Coverage for people who withdraw from a universal public private insurance is mandatory, and it finances most health insurance program. For example, in Germany individuals care. maywithdraw from the national payroll tax-financed sys- Even in industrial countries where public health care tem of sickness funds, which offer coverage to all resi- financing is universal, private insurance may still finance a dent workers, their families, and retirees. People who significant share of health care. For example, in the United withdraw are not required to buy private insurance, but Kingdom private insurance financed 14 percent of health theyusuallydo so. Few people withdraw from the national care in 1990. Similarly, in Canada (where public health public insurance system, however, since they can never care financing is universal and private insurance is prohib- reenter it. In Chile insurance is compulsory but individ- ited from covering publicly insured health care services) a uals can choose between buying regulated private insur- sizable minority of people buy private insurance to finance ance or relying on publicly financed and delivered health services that are not covered by provincial programs. care. This approach has created a two-tier system in which Information about the shares of health expenditures financed PRIVATE INSURANCE: PRINCIPLES AND PRACTICE 79 TABLE I Private and pubic expenditures for personal health care services, selected countries (percentage of total expenditures) Private health care expenditures Public health care expenditures Issued or Out-of- Social insurance Country prepaid pocket Total programs Other Total Argentina - 23 - 36 22 58 Brazil, 1995 _ - - - - 75 Canada 20 - - 75 - - Ecuador - 63 - 17 14 31 Egypt - - - 9 30 39 France 21 - - 75 - - Gen-nany. 1985a 7 7 14 69 12 81 India, 1990_91b 3 75 78 - - 21 Jamaica 9 - - 35 - - Jordan, 1994 - - 53 39 8 47 Kenya, 1994 - - - - 43 Niger 14 - 67 - - Nigeria 44 - - 45 - - Peru, 1995 - 28 - 36 30 69 South Africa, 1993-94c 37 14 55 - - 45 Tanzania 14 - - 68 - - Thailand, 1992 - 74 74 2 24 26 Tunisia 25 - - 67 - - Uganda I5 - - 47 - - United Kingdomd 13 - 14 85 2 87 United States, 1994e 32 3 37 30 29 59 Uruguay 14 - - 76 - - Note: Unless otherwise indicated, source did not provide a reference year. In all cases data are the most recent available. a. Excludes 4.3 percent of expenditures financed from other sources. b. Excludes 0.8 percent of expenditures financed by external donors. c. Excludes 4 percent of private spending categorized as industrial health expenditures. d. Excludes I percent of private spending not allocatable to categories. c. Excludes 3.4 percent of private spending from other sources. Social insurance figures include Medicare Part A and Part B. Source. Lewis and Medici 1995; Musgrove 1996; Reinhardt 1995; Wodd Bank 1995 and 1996; Collins and others 1996; Abel-Smith 1995; Levit and others 1996; Femandez 1997; TA1 1997; Nittayaramphong and Tangcharoensathien 1994; Rafeh in this volume. by private insurance, out-of-pocket expenditures, and pub- ticular ways-usually trying to package coverage for unin- lic insurance programs in selected countries is summarized surable risks together with coverage for insurable risks. in table 1. The kinds of situations in which some or all health care may be uninsurable are described below. Concept of Insurable Risk Nonrandom health care risk Risk is defined in terms of both the probability and the mag- nitude of potential health care expenditures. A high-risk sit- Possibly the main reason that health care would be unin- uation may entail a high probability of expenditure (regardless surable is if it were nonrandom. For example, during a war of how great the expenditure may be), a high magnitude of or civil conflict health care risks are systemic: the likeli- expenditure (regardless of the probability), or both. hood that any person will need health care is highly corre- In general, health care for any illness or condition that lated with the likelihood that many others will need care as occurs randomly among a population is insurable. But in a well. Similarly, in communities where serious, communica- number of high-risk circumstances health care maybe unin- ble health problems have reached epidemic proportions (for surable. In these circumstances insurers will be unwilLing example, in communities with a high incidence of AIDS), to offer coverage, or will design insurance contracts in par- much health care may be uninsurable. In these communi- INNOVATIONS IN HEALTH CARE FINANCING 80 ties insurers may be unwilling to insure much of the popu- oped insurance sector. If regulation permits, insurers will lation, or they may refuse health care for the specific injuries shun people with chronic health problems, people who are or illnesses that are most likely (for example, those due to terminally ill, or people living or working in circumstances war or civil conflict). that suggest a high risk of illness or injury. Even if private health insurance is available to such people, it may be unaf- High-probability health core services fordable.2 Although affordable private health insurance may emerge for relatively high-risk populations (such as the Even when the incidence of illness or injury is random, some elderly), it is likely to be available only to supplement exten- health care services may be uninsurable if the probability sive coverage from a public insurance program. that people will use those services is very high. The reason such services may be uninsurable relates to how insurance Dynamics of Private Insurance Systems prices are determined. Specifically, the price of an insur- ance plan that would cover high-probability losses may equal In private, voluntary health insurance systems, people can or exceed the cost to consumers of remaining uninsured, choose whether to buy health insurance. In a competitive even if they could afford to buy coverage. I When this is the system they can also choose which health insurance plan to case, private insurance for those services may not emerge. buy. In many countries some workers "buy" health insur- Instead, insurers may offer insurance products that specif- ance through their employers, taking insurance in lieu of ically exclude coverage for high-use services or for services higher wages. In this case the employer is the direct buyer that, when covered, would attract enrollment by high-use of the health insurance plan for a group. Alternatively, con- patients. In the United States mental health care is one sumers may buy health insurance directly, either as indi- example of such a service; most private insurance plans viduals or as afamily-much astheywouldbuyanyproduct. strictly limit coverage for mental health care or care related As with most products, buyers will tend to choose an insur- to substance abuse. ance plan that has a lower price if its essential features are acceptable. Very low-cost health care services Insurers can lower the price of a health insurance con- tract in four ways: Similarly, very small health care expenditures may be unin- * By trying to insure only low-risk people, denying cover- surable, whether they are likely or not. For very small losses, age to people who are sick, or excluding coverage for the administrative costs of insurance may exceed consumers' some conditions. demand to be protected from the associated risk. This does * By offering less coverage, limiting the scope or extent of not mean that private insurance would not cover such expen- covered services. ditures, but it probably would not cover only such expendi- * By discouraging excessive use of covered health care tures. Instead, insurers would package coverage for very services. small expenditures with coverage for more costly, less likely, * By reducing the administrative costs of the plan. and therefore insurable services (such as hospitalizations). Each of these methods can create immense problems for some consumers. Consumers may be unable to buy ade- Uninsurable individuals or groups quate insurance (or any insurance), especially if they are sick, and they may find that customer service under their Finally, health care that is insurable for some people may plan (for example, timely and accurate payment of claims) be uninsurable for others. Specifically, insurers are likely to is poor. But each method offers an economic advantage to view people as uninsurable if they are likely to need exten- consumers who are healthy. Because healthy consumers sive and costly health care. This is the main reason that pri- are unlikely to need much health care, they are unlikely to vate insurance (when it is voluntary) does not finance most use their health plan extensively if at all. Thus private insur- health care spending, even in countries with a well-devel- ance can offer them relatively low-cost financial protection. PRIVATE INSURANCE: PRINCIPLES AND PRACTICE 81 Insuring low-risk people: insurance underwriting and pricing in different classes (or tiers) of risk. The practice of pricing insurance based on enrollee health status (or various indi- Consumers seeking insurance are always more knowl- cators of medical risk) is called tiered rating. Tiered rating edgeable than insurers about their health status and about is a natural, stable result of competitive insurance markets: the likelihood that they will need health care.3 Moreover, tiered rates simply reflect differences between the risks con- consumers who have or anticipate health problems are more tained in different insurance pools.8 However, critics of likely to seek insurance than are healthy consumers.4 These tiered rating view it as splitting up risk unnecessarily. They facts dictate a great deal of how insurance contracts are sold. argue that tiered rating makes health insurance unafford- Enrollment by people with greater health care needs than able to people who have health problems or even to peo- the insurer anticipated when setting the price of insurance ple who are in a demographic group that might suggest is called adverse selection. Adverse selection can destabilize higher medical expenses. They argue that a single rate an insurance pool5 and even cause it to fail.6 Thus insurers class, several rate classes reflecting broad geographic dif- have developed techniques to avoid or reduce adverse selec- ferences in the cost of care (pure community rating), or broad tion. Possibly the most important of these is underwriting. rate classes based on demographic factors but not reflect- Insurance underwriting is the practice of evaluating indi- ing individual differences .in health status (modified com- vidual health status and either rejecting potential buyers munity rating) would make health insurance more affordable who are deemed to pose excessively high risk or placing to high-risk people by forcing other members of the pool them in plans with other people who represent approxi- to subsidize them. But since low-risk consumers tend to mately the same risk. Insurers are inclined to underwrite in prefer low-cost insurance products over subsidizing peo- order to avoid adverse selection, but competitive insur- ple with predictably higher health care costs, community ance markets also tend toreward insurance plans that exclude rating does not naturally occur in insurance markets. or isolate people with extensive health care needs. That is, (Underwriting and community rating are discussed further in a competitive market consumers search for the lowest- in a later section on insurance regulation.) priced plan that provides themwith the coverage theywant. Although some insurers re-underwrite enrollees at the Insurance plans that are able to exclude high-risk partici- time of renewal, they are more likely to rely on a pricing pants are likely to be less costly and more comprehensive strategy which assumes that customers who are renewing than plans that insure everyone and try to control cost in coverage are likely to have more (and more costly) health other ways. Thus consumers who are searching for the low- care needs than new customers in the same plan. The prac- est-priced insurance plan are likely to prefer an insurance tice of charging more for renewal than for first-issue cov- plan that excludes people who are more costly than they erage in the same plan is called durational rating. are.7 Durational rating assumes that the claims experience of Insurers that underwrite coverage typically require appli- any risk pool will worsen over time-a phenomenon that cants to disclose their medical history (allowing the insurer in fact is usual in insurance pools. Some participants who to review their medical records) and may require that appli- were healthy at the start of the contract become sick or cants undergo a physical examination by an approved physi- injured. Women become pregnant and require maternity cian. They may require applicants to present such "evidence and obstetric care. Participants who were unfamiliar with of insurability" at the time the contract is first issued, and the provisions of their health insurance plan at the start of again each time the contract is renewed. Thus, while insur- the contract (and therefore hesitant to seek care) begin to ance underwriting enables insurers to price insurance plans leam what services are covered and what they must pay more accurately, many consumers find it to be personally out of pocket. In plans that restrict choice of provider, par- intrusive and offensive. ticipants must select a doctor within the constraints of the Insurers that are able to identify high-risk consumers plan; they will hesitate to use care for minor health prob- may nevertheless be willing to sell coverage to many of them lems if they have not yet taken the time to make their selec- if they are able to price insurance differently to consumers tion. For all these reasons, insurance claims rise gradually INNOVATIONS IN HEALTH CARE FINANCING 82 over the course of an insurance contract. Thus insurers are bers of a social organization. Many insurers require that inclined to offer a low price to attract new participants, but the sponsoring organization pay a significant share (at least they will raise the price at renewal to reflect the growth in half) of the cost of coverage for group members to ensure average medical losses as the insurance pool ages. that even the lowest-risk group member would find enrolling As with tiered rating, durational rating has its critics, who advantageous. Also, insurers may require that a minimum argue that it is evidence of a noncompetitive insurance mar- percentage of the group (for example, 80 percent) enroll ket. They liken it to sirnple price discrimination, noting that in the plan regardless of the sponsoring organization's con- insurers raise prices when people become sick and (in mar- tribution, further reducing the chance that adverse selec- kets where insurers underwrite) when no other insurer will tion will occur within the group. sell them an insurance plan. Moreover, they argue that insur- Group underwriting reduces the insurer's need to bear ers use durational rating to "chum" their business: by rais- the cost of careful, individual underwriting. Instead, the ing the price of insurance at renewal, durational rating insurer can look at the broad demographics and circum- encourages people to shop for new coverage and to change stances of the group and decide whether it represents an insurers frequently. When they change insurers, they are under- insurable risk. However, in highly competitive markets written again as new business. People who have health prob- (where competition has driven insurers to set prices very lems maybe deniedcoverage altogether, orthey maybe denied low) insurers may attempt to underwrite within the group- coverage for the care they are most likely to need. People who denying coverage to some group members based on their are healthy are able to find new insurance at a lower price. health status. When within-group underwriting occurs, it In either case durational rating will encourage consumers typically is in insurance markets for small-group coverage. to sort themselves into different insurance plans, separat- For reasons related to how large-group coverage is negoti- ing high-risk consumers from low-risk consumers. By iso- ated and priced, it is rare in the large-group market. lating people into relatively homogeneous risk pools, each of these practices-underwriting, tiered rating, and dura- Preexisting condition exclusions. To deter people from tional rating-tends to reduce cross-subsidies among peo- seeking insurance after they become sick, insurance con- ple who are insured. In many countries this outcome is tracts typically exclude coverage for conditions that existed socially unacceptable; and the more perfectly insurers are (or that could have been known to exist) when the insur- able to achieve this result, the more unacceptable it is. ance contract started. Preexisting condition clauses typi- Nevertheless, it is an economically efficient result: it min- cally stipulate a "look back" period to deem medical imizes involuntary transfers among individuals. Regulating conditions as preexisting (for example, medical conditions insurers to "correct" this result (that is, to force more het- that were manifest or could have been known to exist six erogeneous risk pools) increases excess burden by artifi- months before the start of the insurance contract). For such cially raising prices to low-risk consumers. conditions the contract will stipulate an exclusion period In addition to underwriting and pricing, insurers have (for example, six to twelve months into the contract period) developed two other techniques to reduce or to avoid adverse during which any care related to a preexisting condition is selection in their health insurance plans: marketing to groups uninsured, but care related to other conditions is insured. rather than to individuals and excluding coverage for pre- Preexisting condition exdusions are particularly prob- existing conditions. lematic when participants with ongoing health problems try to change insurance plans, or when they lose and try to Group coverage. To limit the amount of adverse selec- regain coverage. When health insurance is provided through tion that can occur in a health insurance pool, many insur- an employer, workers may change insurance plans or lose ers prefer to insure people who have grouped themselves coverage altogether when they change jobs. But the propen- for reasons other than the purchase of insurance. Such sity of consumers to buy insurance only when they are sure groups typically include employees of a particular firm, mem- to need health care is so great-and the potential for adverse bers of a professional or trade association, or even mem- selection to destroy a health insurance pool is so signifi- PRIVATE INSURANCE: PRINCIPLES AND PRACTICE 83 cant-that in markets like the United States insurers are more care and to providers' inclination to offer more care almost never willing to write an insurance contract that does when it is insured. 12 Managed care plans typically establish not exclude or delay coverage for preexisting conditions. networks of primary care physicians from which enrollees choose, as well as a network of specialists under contract. Limiting the scope or extent of covered services The plans cover expenditures for specialty care only if the primary care physician refers the enrollee to a specialist in Insurers may be willing to insure relatively high-risk people the network. Managed care plans typically attempt to estab- if they can deny coverage for specific high-cost illnesses or lish a lower-cost "culture" of practice among their primary procedures. Coverage exclusions for care related to specific care physicians, emphasizing greater use of preventive care high-cost illnesses (for example, hemophilia or diabetes) and less aggressive treatment for some conditions.'3 Thus, or for specific procedures (for example, organ transplants) while the features described below are widely used by finan- reduce the amount of insurance available to some or all buy- cial insurance plans, managed care plans use them much ers in the market. These exclusions differ from preexisting less often and much less extensively condition exclusions in that they apply equally to all buy- ers of that particular plan, regardless of health care status Deductibles. An insurance plan deductible requires the at the beginning of the insurance contract. Moreover, such insured consumer to pay all charges for covered services out exclusions are permanent: they exist for the life of the con- of pocket until the total cost reaches the deductible amount. tract.9 U.S. insurers have used coverage exclusions for ser- After that, the insurance plan begins to pay Insurance vices such as obstetric care to develop low-priced plans plans also may impose separate deductibles for specific types that predictably attract low-risk buyers; in this case a dis- of services (for example, hospital care or prescription drugs), proportionately large number of young men. either to deter unnecessary use or to avoid the cost of admin- istering very small claims. Discouraging excessive use of covered services To administer a deductible, the insurance plan must keep an accounting of each enrollee's expenditures for covered Consumers tend to use more health care services when they services (and adjudicate which expenditures qualify for reim- are insured than when they are not, even when their health bursement from the plan), even though the plan does not status does not change. This tendency is called moral haz- issue payment. Although this process has been greatly facil- ard, a term that at one time implied a judgment about con- itated by the growing use and sophistication of computer sumer dishonesty but no longer carries that connotation. technology, deductibles nevertheless require careful record- Moral hazard partly reflects efficient consumer behav- keeping and, therefore, administrative cost for the plan. ior: people use more health care when insurance reduces This cost is worthwhile if the deductible effectively deters the price that they must pay for each service. To a fully unnecessary use. However, a number of studies have found insured consumer, the cost of using an additional health that patients may not distinguish appropriately between nec- care service is zero; thus they are very likely to use it.10 But essary and unnecessary health care. Instead, they may since consumers typically rely on the medical judgement of respond to deductibles and other means of cost sharing by health care providers, much moral hazard behavior reflects delaying necessary care, potentially raising the ultimate cost the propensity of providers to deliver more services when of care by seeking care too late. Still, in the short term (that insurance makes them affordable for their patients. " I Most is, within the duration of an insurance contract) deductibles insurance plans incorporate cost-sharing features that are are very effective in reducing the use of covered health designed to deter excessive use of medical care by making care and, therefore, in reducing plan cost (Lohr and others consumers pay some of the cost of insured health care. 1986). In some industrial countries (and especially in the United Deductibles are rare in managed care plans, in part States) managed care has emerged in part to control moral because they are administratively infeasible for most of the hazard-as it relates both to consumers' propensity to use care that the plan insures. That is, because the health care INNOVATIONS IN HEALTIH CARE FINANCING 84 providers who contract with managed care plans charac- physical therapy or a second routine examination during a teristically do not charge a separate fee for each service, calendar year). Unlike coinsurance, copayments impose most plans do not have an obvious way to compare the value no significant administrative cost; they require only basic of delivered care against a deductible. Also, the concept of record-keeping by health care providers (who typically col- a significant deductible is incompatible with managed care lect the copayment). Because copayments are administra- plans' emphasis on preventive care and low-cost care within tively simple, managed care plans are more likely to use a network of primary care providers. them than any other form of cost sharing to manage con- sumer demand. Coinsurance. Coinsurance is the share of costs for cov- ered services that the consumer is required to pay out of Internalandexternallimitson coverage. Most group insur- pocket. Typical coinsurance levels for covered services range ance plans and all individual insurance plans place a limit from 10 to 20 percent. Like a deductible, coinsurance is on the cumulative amount that the plan will pay for cov- intended to reduce moral hazard by imposing some of the ered services. An extemal limit on coverage is a limit on cost of care on the insured consumer. Coinsurance may the total value of coverage under the plan, usually over the apply uniformly to all covered services, or only to some- full period of the contract (for example, total plan payments for example, only to inpatient hospital care. Separate, higher per year). An internal limit on coverage is a limit on the total coinsurance amounts may apply to types of care for which value of coverage for a particular service covered by the underwriting is particularly difficult and high use can pose plan. Commonly, insurers will place internal plan limits on a serious cost problem for the plan-for example, outpa- coverage for mental health care, but may also limit cover- tient mental health care. age for inpatient care. Internal limits may be denominated Insurance contracts that use coinsurance typically also as units of currency, or as a maximum number of inpatient set an out-of-pocket limit above which the coinsurance days or outpatient visits. Many plans place one or more amount drops to zero. For example, consumers may be internal limits on coverage as well as an external limit. required to pay a $10 deductible, plus 20 percent of any Plan limits serve a number of purposes for insurers. expenditure for covered services, until the sum of these Internal limits may deter people who anticipate needing out-of-pocket payments reaches a specified amount. At substantial amounts of a particular type of care from seek- that point the insurance plan will pay 100 percent of cov- ered charges up to the plan's external limit, if any. The FIGURE I enrollee's cost of using health care in such a plan is depicted Out-of-pocket health care expenditures for an in figure 1. uninsured patient and for an insured patient in a conventional health insurance plan As with deductibles, coinsurance provisions create admin- . . . . ,, ~~~~~~~~~Patient cost of care istrative expense: they require insurers to keep a careful / accounting of accumulating health care expenditures for for uninsured patient covered services, and to adjudicate each claim. For the same reasons that they are unlikely to impose deductibles, man- aged care organizations are unlikely to use coinsurance to deter moral hazard. / Out-of-~pocket cost for: / msured~i patfient, net of, Copayments. A copayment is a small fee that an insur- i ance plan requires patients to pay each time they obtain a covered health care service. Like coinsurance, copayments applies are intended to encourage patients to exercise judgment about their need for care. Copayments typically apply tO Deductible Out-of-pocket Extera plan to ~~~~~ ~ ~~limit rna,drnum specific services that are discretionary in nature (such as Use of care PRIVATE INSURANCE: PRINCIPLES AND PRACTICE 85 ing health insurance, since the plan would not cover all the marketing to a few large groups of enrollees can be much care that they would need. Thus insurers are able to limit lower than the cost of marketing to many small groups or the cost of failing to underwrite accurately. Plan limits also individuals. Finally, client-initiated turnover in the insurer's limit the amount of moral hazard that the plan will sustain. business (and thus the administrative cost of the contract) With respect to mental health care services, moral hazard can be less. Employee turnover in large firms is lower than is the most significant reason that insurance plans typically in small firms (so fewer workers enter and leave the plan place an internal limit on coverage. Although such services during the contract period), and large firms are less likely maybe curative, whether mental health is restored is largely to go out of business than are small firms. 14 a subjective judgement by the patient or the provider- The economic advantages to insurers of writing group neither of whom is directly responsible for paying most or coverage generally diminish with the size of the group. 15 any of the cost of care. Finally, plan limits reduce the amount For very small groups the administrative cost of a group and cost of reinsurance that an insurer needs, either as a insurance plan approaches that for individual coverage. U.S. matter of prudent business practice or to comply with insurers report that the marketing and administrative costs regulation. of individual and small-group coverage can make coverage Plan limits have very different effects on consumers 40 percent more costly per enrollee than the same cover- and health care providers. Limits on coverage may leave a age for a large group. Moreover, because many insurers do significant "tail" of uninsured expenditures, exposing con- not want to undertake the considerable cost of aggressive sumers to financial risk and the risk of needing to termi- underwriting, some of the largest insurers (those that can nate care because they are unable to pay. For providers, achieve significant economies of scale for administrative professional ethics and concern for the patient may con- costs such as claims processing) will not cover individuals flict with the plan terminating payment for care. Even in or small groups. The absence of large insurers may further cases where plan limits are very high (for example, in explain the high administrative costs in these markets. employer group insurance plans in the United States most The greater administrative cost of insurance for indi- external plan limits are $1,000,000 or more), patients who viduals and small groups in the United States can make cov- reach these limits may be in-hospital, and the hospital may erage extremely costly and, for many, unaffordable. Lacking be ethically unable to terminate care. While these cases are any legal requirement that everyone be insured, many indi- rare, when they occur the hospital generally continues care viduals and small groups without access to group coverage without payment. are uninsured. This experience is not universal, however. In the Czech Republic quasi-private insurance plans that Reducing administrative costs compete with the large, central government plan enroll groups and individuals alike, and find feasible the statu- While the practices and features of insurance plans described tory 7 percent limit on the plans' margins over medical losses. above effective in avoiding or reducing adverse selection In Australia, where private insurance supplements public and moral hazard, some can raise a plan's administrative coverage but covers only inpatient care, private insurers costs. For example, while careful underwriting can be an are not particularly concerned about adverse selection or effective way to avoid adverse selection, it can also add sig- the high administrative cost of marketing to individuals, and nificant administrative cost. As a result insurers have devel- are not inclined to foster a group market. oped a number of practices that are effective alternatives To reduce the cost of underwriting coverage for groups to careful underwriting. or individuals, insurers may practice a form of underwrit- Possibly the most important of these practices is group ing commonly called redlining: denying coverage to broad underwriting-that is, accepting or rejecting entire groups classes of groups or individuals without actually consider- instead of screening individual health status. Moreover, writ- ing their insurability. For example, if insurers deem residents ing coverage for large groups may offer other administra- of a particular geographic area-such as a low-income urban tive-cost advantages for insurers. For example, the cost of area-as high risk, they might routinely deny coverage to INNOVATIONS IN HEALTII CARE FINANCING 86 anyone who lives or works in that area. Insurers might also the market. Such insurers prey on unsophisticated con- redline workers in particular industries or occupations if sumers who find many aspects of insurance contracts dif- they are likely to be exposed to hazardous materials or suf- ficult to understand. fer a job-related injury.'6 Finally, various plan design features that curb the high Financialstandardsforentryandoperation. The most effec- use of health care under the plan may also yield adminis- tive and efficient way to minimize the chance that people trative efficiencies. For example, a plan deductible allows will buy insurance from financially insecure insurers is to the insurer to avoid fully processing and paying small claims. bar such insurers from entering the market. Thus govern- The net effect on administrative costs is likely to be small, ments may require that insurers be licensed.18 As a condi- however, since financial insurance plans must adjudicate tion of receiving a license, insurers may be required to even small claims in order to accumulate claims against the meet minimum standards for financial soundness and to deductible. demonstrate past or intended ethical business practices. Managed care plans that pay their primary care doctors As evidence of financial soundness, governments may a salary or a fixed fee per patient (instead of fee-for-ser- require that an insurer meet minimum capital and surplus vice) can vastly reduce the amount of paperwork involved requirements. A capital requirement establishes a minimum in sending and paying bils and thus can substantially reduce level of financial assets for insurers seeking to enter (and administrative costs. But because well-run managed care remain in) the market. Governments may require that insur- plans require active management of a large network of ers hold these assets in highly secure investments. 19 health care providers, the net cost advantage of these plans In addition, governments that regulate private insurance typically relates to lower rates of hospital admission among typicaly establish a surplus requirement: a minimum level of managed care participants, not to lower administrative financial assets that an insurer must hold relative to its esti- costs.17 mated liabilities (principaly, its estimated medical losses). Both capital and surplus requirements may be set by law, and they Government Regulation of Insurers may vary by class of business. For example, insurers that write both personal health insurance and personal life insur- Government regulation of insurance typically has three goals: ance may be required to meet separate minimum capital and maintaining a stable insurance market, protecting consumers, surplus requirement for each class of business.20,21 The gov- and maximizing consumer participation in the private mar- ernment also may set the surplus standard higher for insur- ket. Various practices related to each of these goals are ers that write a larger amount of coverage. By scaling the described below. surplus standard to the insurer's volume of business, the gov- ernment can encourage insurers to enter the market and reduce Stabilizing the insurance market: standards for insurer the likelihood of a major market disruption from the insol- entry and exit vency of a large insurer. Finally, governments may require that insurers entering the market bring a higher amount of Regulation to stabilize insurance markets includes setting initial surplus, which they may then deplete (so long as it financial standards for market entry and ongoing opera- exceeds the minimum surplus that they must hold on an ongo- tions, ethical standards for market entry, and conditions ing basis) in financing their initial operations. for insurer exit from the market. Entry or exit standards Successful solvency regulation entails intensive peri- that are set too high will allow fewer insurers into the mar- odic review of each insurer's fiscal condition. Typically, insur- ket and reduce competition among insurers that do enter. ance regulators are given broad statutory authority to audit In turn, insurance prices are likely to be higher in a less com- and investigate insurance companies. U.S. insurance regu- petitive market, and buyers will have less choice among lators even have statutory authority to assume control of insurance products. But standards that are set too low can domestic insurers in financial difficulty, with the intent of allow financially unsound or unscrupulous insurers into "rehabilitating" the insurer-modifying the insurer's man- PRIVATE INSURANCE: PRINCIPLES AND PRACTICE 87 agement and financial practices to improve its prospects for based on concerns that the insolvency of one could con- remaining solvent.22 tribute to the insolvency of the other. But insurers may form In developing countries establishing solvency standards a holding company to affiliate with an unrelated enterprise.23 for insurers entails careful review and standardization insur- Still, any affiliated enterprise can destabilize an insurer, and ers' accounting and actuarial practices. If regulatory stan- insolvency problems have resulted from holding company dards for capital and reserves are to be meaningful, insurers affiliations (U.S. House of Representatives 1990 and 1994). must use a conmmon accounting convention to measure the Especially in countries with a newly emerging private insur- value of assets and liabilities. The government must estab- ance sector, governments should consider regulatory safe- lish the basis for evaluating assets and health insurance lia- guards in order to prevent such situations.24 bilities, and require all licensed insurers to report on that Finally, many countries have some arrangement to guar- basis. The risk of financial distortions (and the opportunity antee insurance benefits if an insurer becomes insolvent. for unsound financial practice) by insurers rises when In some countries the government operates the guaranty accounting principles are undear and industrypractice varies. system; in others a government-authorized private agency However, even with common accounting practices, antic- or association performs that role, but all insurers are required ipating health insurance liabilities in developing countries to participate as a condition of licensing.25 Government- can be extremely difficult. The introduction of private insur- authorized private guaranty funds may be financed by an ance may cause health care prices and real spending for assessment on all health insurers (usually at the time of the health care to accelerate sharply. Thus the integrity of the insolvency). These assessments may be limited by law (for plan's actuarial estimates rely fundamentally on how well example, to 2 percent of gross premium volume per year). the plan is able to control the cost of covered services. Government-run guarantee funds also maybe financed from Such controls are generally more reliable in managed care sales taxes on insurance premiums. If the guaranty liability plans, especially if they pay providers a fixed fee per patient exceeds what is collectible within the annual assessment (called capitation) for most or all care covered by the plan. limit, participating insurers may be reassessed each year. Because plans without meaningful expenditure con- Thus guarantee-fund payments to policyholders may stretch trols may be unable to anticipate liability with sufficient over a number of years in the case of a large insurer insol- accuracy, regulators must carefully consider whether to let vency. Insolvency guarantee funds may pay policyholders them enter the market. If the insurer's premium levels are only a portion of each claim (for example, 60 percent), and likely to be insufficient, it is at great risk of insolvency- may leave policyholders responsible for much of the cost even when it holds the required reserves. Especially in newly of care that would have been paid by the insurer had it developing markets, regulators should review the adequacy remained solvent. of insurance premiums as vigilantly as they review compli- ance with financial standards. Nonfinancialstandardsfor entry and operation. Regulation Governments can also prohibit specific organizational governing the ethical practice or intent of insurers may forms of insurance that are deemed financially unstable take any of several forms. Some governments require that (even though they may purport to broaden enrollment in officers, board members, and incorporators meet residency private insurance plans). These may include multiple- or citizenship requirements. Some countries require that employer trusts, fraternal organizations, or membership certain minimum ownership rights be held by nationals. associations. Since not all regulators agree on which orga- Governments also may investigate the experience or char- nizational forms are inherently unstable, some jurisdic- acter of individuals seeking licensing.26 They may require tions may allow organizational forms that others prohibit. that insurers domiciled in another jurisdiction or country In the United States and many other industrial countries demonstrate that they are lawfully organized and licensed insurers are prohibited from directly engaging in any busi- in their home jurisdiction. They may require that insurers ness that is not reasonably related to insurance. For exam- seeking a license submit a proposed business operation plan, ple, insurers may be prohibited from affiliating with banks, assessing the economic soundness of that plan as part of INNOVATIONS IN HEALTH CARE FINANCING 88 the process of admitting the insurer into the market. Finally, actuarial opinion certifying the adequacy of their reserves states may prohibit some organizational forms of health relative to the nature of the risk they have assumed. (All plans if their practices are deemed too difficult to monitor U.S. states now require that licensed insurers submit an and regulate. For example, at least one U.S. state (Minnesota) actuarial opinion each year.) Regulators also may require will not license managed care plans that are operated as that insurers submit to a periodic financial audit by an for-profit enterprises. independent certified public accountant, and that the results be disclosed.27 The costs of these professional certifications Conditionsfor insurer exit. Constraints on market exit can should be paid by the insurer. be as important as restrictions on market entry. Allowing insurers to haphazardly enter and leave the market increases Consumer protection market instability and erodes consumer confidence in pri- vate insurance systems. Ideally, insurance regulators should Consumer protection regulation typically is of two types: examine an insurer's commitment to staying in the market regulation that governs the language and marketing of insur- as a condition of admitting them to the market. But such ance contracts and regulation that governs the relationship an appraisal can be difficult, and regulators may instead rely between insurance plans and health care providers. Given on establishing conditions for exit. the complexity of even relatively simnple insurance contracts, Exit rules for insurers may take several forms, all ofwhich government regulation of the language of insurance con- are designed to minimize market disruption as a result of tracts can be extremely important. Governments can require insurer exit. At a minimum, departing insurers should be that insurers explain coverage using common terms and that required to give reasonable notice to policyholders and to specific features of health insurance plans be explained in submit a plan demonstrating how claims and other obliga- all insurance contracts. Such features might include the tions will be satisfied. They also may be required to pay a plan's deductibles, coinsurance provisions, copayment processing fee to the regulatory agency to cover the cost of amounts, out-of-pocket limits, and internal and external overseeing an orderly exit (Skipper 1992). limits on coverage. Plans that contract with provider groups (physician group practices or specific hospitals, for exam- Reporting requirements, examinations, andlprofessional over- ple) and restrict or modify coverage according to policy- sight. Reporting requirements constitute the core of insurer holders' use of those providers should also be required to surveillance. Governments should require all licensed insur- fully disclose the nature and details of those limitations. To ers to submit full financial reports annually, and may require enforce the use of clear language and honest marketing abbreviated financial statements quarterly. At a minimum, practices, insurance regulators should require insurers to these reports should allow the regulatory agency to under- submit all marketing and enrollee materials, and they should stand whether the insurer is operating within the financial be prepared to review these materials in a timely manner. limits dictated by law or by prudent business practice. In practice, much of the consumer protection provided In most developed insurance markets governments con- by regulatory agencies occurs only after a problem arises duct periodic onsite examinations of insurers' financial and in response to consumer concerns and complaints. records. For example, all U.S. states require onsite exami- Prompt and vigorous response to consumer grievances can nation of domestic insurers, typically every three to five be an effective and efficient way to augment review of insur- years. Regulators also might conduct a targeted examina- ers' contract language and marketing materials. This is espe- tion if they suspect that an insurer is in financial difficulty cially true in large or fast-growing insurance markets, where (Skipper 1992). the paperwork burden of meticulous document review can Governments also may enlist the accounting and actu- overwhelrn a small regulatory agency arial professions to discourage inappropriate insurer behav- Regulation that govems the relationship between health ior and to reveal it if it occurs. For example, insurance plans and providers is sometimes also considered consumer regulators may require insurers to submit an independent (or patient) protection. Typically, such regulation seeks to PRIVATE INSURANCE: PRINCIPLES AND PRACTICE 89 preserve the professional autonomy of health care plans to cover specific types of services or health care providers-especially if they participate in managed care providers. Each type of regulation is discussed below. contracts, which may discourage providers from recom- mending high-cost treatments or from referring patients to Guaranteed issue and renewal. Guaranteed issue regula- specialists for care. tion requires insurers to accept all applicants for coverage, Two main forms of regulation have emerged affecting regardless of their health status. By contrast, jurisdictionrs the relationships between health care providers and man- that require guaranteed renewal (and by inference, not guar- aged care plans. "Any willing provider" regulations pro- anteed issue) allow insurers to underwrite when they first hibit insurers from "locking out" physicians or other health issue coverage, but prohibit them from underwriting when care providers who are willing to accept the plan's payment the insurance contract is renewed. Thirty-six U.S. states levels, practice guidelines, and reporting requirements for require guaranteed issue of some or all insurance products participating providers. Conversely, antitrust regulation may in the small-group market; thirteen require guaranteed issue prohibit health insurance plans from "lockingin" providers- in the individual health insurance market.29 that is, requiring that providers who contract with them Guaranteed issue regulation may require insurers to not accept payment (or patients) from any other insurance accept applicants into any health insurance plan that they plan. offer, or into only one or two selected plans. By itself guar- Health care providers generally view regulation of either anteed issue regulation does not address how insurers may type as essential to maintaining their professional inde- set prices for these plans. pendence and, therefore, protecting the best interests of their patients. Health insurance plans, however, view these Community rating. Sixteen U.S. states require insurers regulations as limiting their ability to establish a professional to use some form of community rating-to charge one price culture of conservative (low-cost), high-quality medical prac- for all plan participants within broad geographic or demo- tice among providers who contract with them. graphic groups, without regard to health status. Community There is no simple resolution of these competing per- rating regulation prohibits insurers from using the tiered or spectives. But many analysts believe that the development durational rating techniques described earlier. Alternatively, of generally accepted measures of health care quality would some states require insurers to price within specified "rate reduce the need for such patient protection regulations. bands"-that is, they allow insurers to price coverage accord- They contend that quality measures would allow both reg- ing to health status, but limit the variation in rates around ulators and enrollees to monitor the quality of care in com- the median rate that insurers charge for the same plan. peting managed care plans and in financial healthinsurance Insurance rate regulation is intended to subsidize plan plans. They believe that health plans would begin to com- participants who are more likely to need health care. Thus, pete on quality (as well as on price)-reducing the need to to achieve a public policy goal (more affordable coverage regulate relationships between health insurance plans and for people with ongoing health problems), rate regulation providers as long as the quality of health care financed by creates some economic inefficiency by forcing involuntary the plan remains acceptable.28 cross-subsidies among plan participants. Experience suggests that guaranteed issue regulation nei- Improving the fairness of private insurance ther destabilizes health insurance markets nor causes a sig- nificant increase in the average price of health insurance. In the United States a growing body of state regulation is However, actuarial research indicates that it does raise the emerging to reconcile insurer practices with social percep- price of coverage for some low-risk consumers (American tions of fairness. These regulations generally are of three Academy of Actuaries 1993). In states where insurers are types: those prohibiting or restricting medical underwrit- required to guarantee issue and to community rate health ing, those prohibiting or restricting insurers from setting insurance products, some small insurers have left the mar- prices based on health status, and those requiring insurance ket, and some insurers have complained that plan costs have INNOVATIONS IN HEALTH CARE FINANCING 90 risen because younger and healthier enrollees have left their approved. The federal government also has set a target loss plans (Chollet and Paul 1994). However, it is unknown ratio for insurance products that supplement Medicare cov- whether participants who exited these plans became unin- erage for the elderly and the disabled, although it leaves rate sured or simply changed plans. review to the states. In newly emerging private insurance markets the transi- tion effects of guaranteed issue and rate regulation might be Mandated benefits. Health care providers can be the great- greater than in well-established markets. Well-established mar- est opponents of insurance plans that limit the scope of cov- kets may have one or two relatively large insurance compa- ered benefits as a way to offer low prices. Health care nies that anchor the market and that absorb much of the providers who favor specific benefit mandates usually make market's high risk. In most U.S. markets Blue Cross and two arguments: that without a mandate the insurance mar- Blue Shield plans play this role. Many "Blues" plans have peri- ket is evicting patients who need a specific type of care and odic open enrollment (once or twice a year, for three to six who cannot afford to pay for it out of pocket; and that weeks) during which they guarantee issue. (In some states their services represent lower-cost alternative to other types periodic open enrollment is a condition of the Blues' non- of care. The second argument is usually very difficult to profit, tax-exempt status.) Moreover, the Blues historically prove conclusively. have community rated their products (again, sometimes as a In every U.S. state various types of health care providers condition of their nonprofit status), although many no longer (psychiatrists and psychologists, chiropractors, physician do so. In markets that have one or two large insurers anchor- assistants and nurse practitioners, marriage counselors, ing the market and accepting all applicants (at least periodi- homeopathic medical practitioners, and even faith heal- cally), the large insurermay alreadyhold much of the market's ers) have sought and won legislation requiring that insur- high risk, and smaller insurers may develop new marketing ance plans cover their services. Their success has led states practices that in effect preserve their underwriting advan- to regulate the content of health insurance plans extensively, tage.30 Thus strict guaranteed issue and community rating reg- encouraged employers to self-insure (by federal law, the ulation may have few early effects in well-developed markets. states cannot regulate self-insured plans; Jensen, Cotter, But in developing markets that have only a few small and Morrisey 1995), and increased the cost of health insur- insurers writing coverage, such regulation may have a chill- ance (Jensen and Morrisey 1990; Jensen 1993). Most state ing effect on market development. Health insurers in devel- legislators have come to understand the economic ineffi- oping markets are likely to experience immnediate entry of ciency that these regulations can create, and the rate of high-cost participants, to have a smaller base over which to new enactment has markedly diminished.32 spread risk, and to have no ready reinsurance market to Some mandated benefits are not so clearly inefficient, finance the high risk that regulation requires them to accept.31 however. Instead, they may correct market failures that arise from consumers' mnisinformation about the coverage that Raterevieworapproval. Governmentsthatdonotrequire their plan provides. For example, every U.S. state requires insurers to community rate their health insurance products health insurers to cover newborns immediately under the may still undertake rate review. They may even require that parents' family health insurance plan. This requirement pro- insurers receive government approval for their rate levels hibits insurance plans from deeming any congenital health and increases. Where rate review and approval are required, problem of the newbom as a preexisting condition and, there- insurance regulators typically examine the reasonableness fore, denying coverage for care that is often life-saving and of insurers' loss ratios: that is, the ratio of payments for med- usually very costly. If few parents anticipate that their pol- ical care to premium income. In effect, approval of the icy would not cover such care when they purchase family insurer's rates is approval of their loss ratio. insurance, such regulation may correct for consumers' imper- Anumber of U.S. states have established target loss ratios fect information. However, it also may raise the price of for health insurers; eight states require insurers to meet or coverage, and probably encourages insurers to develop health exceed a minimum loss ratio in order to have any rate increase plans that exclude coverage of obstetric care altogether. PRIVATE INSURANCE: PRINCIPLES AND PRACTICE 91 Private Insurance in Developing Countries The first section of the paper has provided the context for common practice. The existence of mature insurance mar- reviewing the coverage and circumstances of private insur- kets has helped strengthen the health insurance industry ance in low- and middle-income countries. Although the and support its growth. theory and practice of insurance have been honed in OECD Traditionally, much of Asia has relied on out-of-pocket countries (particularly the United States), non-OECD coun- expenditures to finance health care, and even public pro- tries have drawn on that experience, experimented with dif- grams tend to require copayments. Private health insurance ferent approaches, and had distinct experiences with private is negligible in these countries (see annex table 1). Indeed, insurance. This section emphasizes the practice and expe- China has no private insurance. The one outlier is the rience that has evolved from these efforts. Republic of Korea, which has the most privately financed The extent of private insurance in any country is related health care system in the world, with 94 percent of the to a number of factors. Among the most important are per- population covered by compulsory private health insurance. sonal income, maturity of financial markets, extent of pri- India, with its 3.3 percent coverage and 1 billion people, vate health care services, cultural factors, and government has the most individuals covered (33 million). In all these policy. These characteristics also determine how insurance systems, copayments are attached to the use of most ser- evolves and its pace of growth. Annex table 2 summarizes vices. At the other extreme, Bangladesh and Pakistan are the characteristics and coverage of private insurance in thirty- seeing the beginnings of an insurance industry, with cur- four non-OECD countries, as well as in Mexico and Turkey. rent enrollments in the thousands. The table includes countries with relatively strong insurance Although private health care is common in the Middle markets, as well as some with emerging markets. Countries East, private insurance is just being established. Lebanon's with negligible markets, like Bangladesh and Ghana, are emergence from years of civil war make it an exception, not included despite available data. but its private sector-oriented health system will likely In Africa private insurance coverage is highly variable, lead to rapid growth in insurance-type financing. With an ranging from South Africa (with a well-established market established private sector and rising incomes, demand for and 16 percent coverage) to countries like Angola (with no private insurance is likely to increase in other Middle Eastern measurable market). Outside of C6te d'Ivoire, Kenya, countries, where it already has a significant foothold. Egypt Nigeria, and Zimbabwe private insurance is negligible, is a prime example (see Rafeh in this volume). although the potential for growth exists throughout the con- Eastern Europe and Turkey have limited private insur- tinent. The Commonwealth countries in the region are the ance coverage. In Eastern Europe low coverage can be most likely to have an established private insurance mar- ascribed to comprehensive social insurance, and in Turkey ket. Although these markets have a foothold in Tanzania to high inflation, traditional distrust of life insurance, and and Uganda, they are imited. Higher-income countries with cultural factors (Fuenzalida-Puelma 1996). The region's large employers have pioneered and established private potential for private insurance is strong given rising incomes insurance, but small, poor communities also have chosen and growth in private providers. Moreover, some Eastern to share risks. European countries perceive private insurance as a source Private insurance is extensive in Latin America and the of funds to recapitalize their health care delivery system. Caribbean, where a long tradition of private providers and In some countries community groups, rural cooperatives, payers offers a solid base for expansion. Moreover, the and mission hospitals have created networks of informal proliferation of social insurance through social security insti- insurance funds to finance catastrophic care or health ser- tutes and employer-based sickness funds throughout the vices for members. Nonprofit organizations like the Grameen region has made reliance on insurance to finance health care Bank in Bangladesh have expanded into health insur- INNOVATIONS IN HEALTII CARE FINANCING 92 ance-type arrangements, and cooperatives and other com- everything else. In Turkey cancers, heart conditions, and munity endeavors have led to private arrangements for even high blood pressure are outside the package of most financing health care. Government has taken a role in Zaire benefit plans, and in Brazil plans typically exclude infec- and Guinea-Bissau, providing physical infrastructure and tious diseases, chronic conditions, mental illness, kidney often seed capital to establish community insurance funds dialysis, and AIDS (Fuenzalida-Puelma 1996; Lewis and that are usually designed, managed, and operated by com- Medici 1995). munities to pay for health care services for a defined pop- Argentina's sickness funds have highly differentiated ulation (TAI 1997; Shaw and Griffin 1995; La Forgia and benefit plans because benefits are a function of earnings Griffin 1993).. (since the revenue base is drawn from a percentage of wages). Private health insurance coverage data give an idea of Thus high-wage industries provide insurance that covers a the extent of health insurance in developing countries. But full range of surgeries, psychiatric services, cancer treat- the nature of those systems-how they operate, how they ments, and dental care, while low-wage industry insurance are regulated, and how well they meet the objectives dis- funds cover only limited care, strictly ration access to costly cussed in the first section of this paper-vary significantly. care, and force members to rely on public facilities (World Indeed, the orderliness conveyed in the first part of the Bank and IDB 1997). paper is not always apparent in these countries. The fol- Uruguay's sickness funds, the Instituciones de Asistencia lowing sections discuss the characteristics of these coun- M&dica Colectiva, have a legally mandated benefit package tries' health insurance systems, emphasizing the differences and exclude services that are covered by the reinsurance relative to U.S. approaches; impediments to development fund (see below on reinsurance). Private insurance is often of a private insurance industry in different countries; and purchased for specific types of benefits: emergency, med- the role of health insurance regulation, and its scope and ical, surgical, diagnosis, or hospitalization. There are no limitations, in developing countries. requirements for private insurance benefits. In Korea expensive high-technology medical services Health Insurance in Developing Countries (CAT scanning, magnetic resonance imaging, PET scan- ning, some chemotherapy) are excluded from private insur- As in OECD countries, private health insurance is rarely ance plans, and hospitalization coverage-is limited to 180 the main source of health care financingindevelopingcoun- days a year (210 days for the elderly; Yang 1996). In addi- tries. Instead, most private insurance is supplementary (an tion, the quality and scope of benefits varies by income level. outgrowth of social insurance that covers services not This pattern is also apparent in Chile's private insurance included by the social insurance or "primary insurance" ben- companies-higher-income industries receive a more gen- efit plan) or additional to services financed under an exist- erous package of services because their contributions are ing public system. Characteristics of insurance are very higher (World Bank forthcoming). different from those in industrial countries. Indeed, even Benefit plans are often regulated by government to ensure the rationale for insurance in non-OECD countries is dif- a basic level of coverage under private insurance plans (as ferent, with basic costs covered but rare, high-cost proce- in Uruguay for social insurance and much of Europe for dures excluded. both social and private insurance). In countries where gov- ernment does not play a role in defining primary insurance Benefit plons coverage, benefits are less likely to include low-risk, high- cost events, and high-risk individuals are less likely to obtain Private health insurance plans in developing countries typ- private coverage. ically exclude costly or chronic diseases and have low ben- In the absence of government requirements on the scope efit ceilings (in terms of number of inpatient days or total of benefits, private insurers typically cover only low-risk expenditures). In many cases these limits translate into insur- groups, leaving government to finance catastrophic care ance plans that do not cover catastrophic care but cover through supplementary insurance and to provide services PRIVATE INSURANCE: PRINCIPLES AND PRACTICE 93 to the chronically ill as the insurer of last resort. Although tem but require that citizens have health insurance cover- market segmentation may bolster the insurance industry age-which is why both countries have a robust private and may also maximize the number of people who are insured insurance industry. The costs of this private coverage are voluntarily, it has social consequences that many countries borne by consumers, reducing overall government find unacceptable. Hence the role of health insurance expenditures. regulation. In Argentina and Uruguay users are not satisfied with the cost and quality of services provided under the sick- Government policy and health insurance ness funds, which is putting pressure on governments to make changes and spurring the purchase of insurance and Most governments have not taken a strong position regard- services from the private sector. It is not clear, however, ing private health insurance. And since insurance regula- whether these developments are reducing the government's tion is almost nonexistent in most countries, policymakers health care financing burden. Indeed, in Uruguay costly are not always aware of the extent or nature of private financ- high-technology services remain the financial responsibil- ing. Some policies, however, encourage private insurance. ity of the government. Among the most important is not having a policy against private insurance, or at least allowing companies to exper- Reinsurance iment with different ways of financing health care for employees. Reinsurance allows insurance companies to spread the risks Policies that permit people to opt out of social insur- that they assume from policyholders. Because spreading ance (although usually with some nominal payment to the risks allows health insurers to include rare events in the government) are also important, as in Egypt, Singapore, package of covered services, reinsurance is in effect insur- and Uruguay. But opting out may have negative effects on ance for insurance companies. Reinsurance can take the the risk pool that remains under social insurance. The young form of sharing risks among affiliated insurers, sharing and the healthy are the most likely to opt out, leaving lower- high risks among a large number of participating compa- income and less healthy individuals in the social insurance nies, and transferring risks to another insurer, among other system and raising the average cost of services in the pub- options. Some governments require sharing among licensed licly sponsored system. Even where there are costs to opt- insurers, but this is rare outside OECD countries (Mclsaac ing out-as in Egypt, where 1 percent of earnings must and Babbel 1995). continue to be paid to the government (Rafeh in this vol- The ability to spread risk across enrollees is often rein- ume)-these tend to be nominal, and do not discourage forced by reinsurance that ensures insurance company sol- people from buying alternative coverage. vency in the event of high, unforeseen risk. Indeed, the Developing country governments often allow opting amount of risk that an insurer can absorb increases with out in order to share the cost of health service delivery with the availability of reinsurance; otherwise companies tend to the private sector, provide citizens with choice of payers and be small and vulnerable to unforeseen high risks. Reinsurance providers, and reduce the overall burden on public provi- coverage is rare in countries where capital markets are weak, sion. More recently, governments have seen private insur- and unavailable where actuarial data are poor. In Turkey ance as the possible financial base for recapitalizing a decayed reinsurance companies consider health insurance a high- health delivery infrastructure; this is most relevant in Eastern risk business for many of these reasons. Limited benefit Europe. Some countries have responded to demands for packages are one symptom of a lack of reinsurance, and choice by allowing competition among approved private most developing countries implicitly rely on government as providers under a regulated or mandated (social) insur- the insurer of last resort. In Brazil and Egypt the govern- ance system (Brazil, Chile, Korea, Philippines, Uruguay) ment finances high-technology tertiary services through sup- or by contracting out specific services to private plans. Chile plementary insurance, and covers people who lose private and Korea rely extensively on opting out of the public sys- coverage. INNOVATIONS IN HEALTEI CARE FINANCING 94 Sickness funds in Argentina and Uruguay have public for limiting or not investing in private insurance (TAI 1997). reinsurance systems-that is, the government taxes and. But in some instances these difficulties have had little effect manages reinsurance funds. Argentina does this through on the growth or profitability of private health insurance the Fund for Highly Complex [Procedures] (Fondo de Alta markets. Argentina and Brazil, for example, weathered years Complejidad), which theoretically finances rare and costly of high inflation accompanied by growth in private insurance- treatments but in reality finances overflow from the sick- suggesting other mitigating factors for investors. ness funds (World Bank and IDB 1997). Uruguay's National Government policy toward the sector also inhibits invest- Fund for Resources (Fondo Nacional de Recursos) is a form ments. Indeed, lack of transparency regarding laws and reg- of social insurance that finances sophisticated medical pro- ulation and extensive bureaucratic hurdles have limited cedures at specialized medical centers located in the sick- investment in health insurance in countries such as the Czech ness fund facilities. Mandatory contributions from workers Republic, Egypt, and Indonesia. Inconsistency in financial finance, among other procedures, cardiovascular surgery, and health polices discourages private activity, and inTurkey renal dialysis, and transplants. The fund functions as de has contributed to slow growth and limited products in the facto reinsurance for the sickness funds, whose risk pools insurance market (Fuenzalida-Puelma 1996; Rafeh in this are heterogeneous. volume; TAI 1997; World Bank 1992). Policies that subsidize or make capital accessible for infra- Impediments to Health Insurance structure expansion, often for designated areas or under- Development served groups (as in Brazil, Colombia, and Jordan), have resulted in a large number of private health care facilities Impediments to the establishment and development of (World Bank 1994 and 1996). Such facilities encourage private health insurance stem from rigidities in the market, development of a private health insurance industry. inmmature capital markets, and direct and indirect govern- Finally, limited administrative and management capac- ment policies regarding insurance, capital markets, and ity of companies or intermediaries have shrunk private insur- health sector activities. ance in Kenya and Tanzania. In both countries high Health insurance is often a by-product of other forms incidences of fraud among consumers and providers have of insurance-as in Egypt, Jamaica, and Turkey-and often resulted in a shift from third-party payers to self-financed serves as a loss leader for other insurance products. plans among large companies. Self-regulation clearly is not Inadequate actuarial data raises risks for investors and viable in these circumstances (TAI 1997). Similar lapses restricts reinsurance to catastrophic losses. Access to and have occurred in other countries, particularlywhere the pri- the cost of capital and reinsurance have posed problems vate insurance market is small (as it is in Kenya and Tanzania). for private insurance growth in C6te d'Ivoire, Hungary, and Similar problems arise in countries like the Dominican Jamaica (Fuenzalida-Puelma 1996; TAI 1997; Lewis 1988). Republic, where private insurance is evolving based on small In Costa Rica, where private insurance is illegal, health insur- employers or individuals who purchase health insurance and ance exists but is arranged by employers in creative ways obtain care from small clinics and individual practitioners through contracts and other payment mechanisms. There under fee-for-service arrangements. Because public over- are no insurance companies. sight is minimal, transactions costs are high and.abuses are Agovernment's role in fosteringorimpedingprivate health- difficult and costly to monitor, the insurance market is inher- related investment is a key element in the growth of private ently unstable. health care financing and service delivery. Moreover, private Some governments have made concerted efforts to financing requires private service delivery to be viable, and improve the climate for private investments in insurance private service delivery requires private financing. Import con- and insurance-provider arrangements. South Africa has done trols Jordan), high tariffs (Nigeria), foreign ownership restric- so through deregulation, Uganda and the Philippines through tions (Brazil, Philippines), inflation (Peru, Turkey), and political tax reform, and Sri Lanka through a policy encouraging uncertainty (C6te d'Ivoire) are commonly cited as reasons private insurance and reducing tariffs on medical imports. PRIVATE INSURANCE: PRINCIPLES AND PRACTICE 95 The ability to repatriate profits in Eastern Europe has pro- Regulation of private insurance is most commonly an duced a favorable investment climate despite the current extension of insurance law, and the institutions that over- government-dominated health financing system. The see health insurance are often part of or affiliated with absence of such impediments make other difficulties more (generic) insurance regulation. This is the case in Brazil, manageable for investors (TAI 1997). South Africa, and Turkey, among others. Insurance regula- tion oversees financial viability, reserves, reinsurance, and Health Insurance Regulation exit from the industry. Even where those functions are com- petently executed for the non-health insurance industry, The problems discussed earlier-adverse selection (com- as in South Africa, their record in regulating and enforcing panies avoiding high-risk enrollees), moral hazard (increased rules for the health insurance industry is poor. consumption by consumers who are covered by insurance), Regulation is generally weak and uneven in much of the imperfect information (consumers who do not understand developing world. This shortcoming is partly attributable to the market in which they are purchasing health insurance), an inadequate regulatory framework, but also to weak insti- and high administrative costs-are highly relevant to devel- tutional capacity. With a few exceptions (Colombia, Hungary) oping countries. This section summarizes available infor- there are almost no comprehensive regulations for health mation on the regulation of private health insurance in insurance. Even Korea, with its heavy reliance on private developing countries, and the experiences to date. insurance, has almost no regulations. Moreover, even when The incentives of private insurers are often incompati- there are regulations on the books, enforcement is often ble with the social objectives of affordable universal cov- limited or ineffective. erage that many governments have pledged to ensure. This is why few countries have been able to build their health Regulotory institutions in developing countries care systems exclusively on private insurance financing. These two sets of objectives have become so intertwined The body regulating private payers varies by country. It can that regulation of insurers has evolved not only to ensure be central, state, or even local. What is crucial is that reg- that the insurance market is competitive and financially ulators be independent of the industry and have the author- sound, but also that it meets social objectives of access, ade- ity to influence insurer behavior to prevent abuse and damage quate benefits, and consumer responsiveness. This dis- to the industry and consumers. tinction is important. Chile, Korea, South Africa, and the United States can rely on private insurance because the Argentina. The Argentine regulatory agency, Adminis- social contract does not guarantee equal access to the same traci6n NacionaldelSegurodela Saluid (ANSSAL), is respon- set of services. Efficiency and reliance on market forces sible for overseeing the financial adequacy of sickness funds are bigger priorities, and the private insurance market is (Obras Sociales). The agency is largely political, with close seen as the best way of achieving those objectives. ties to organized labor and with inadequate and inappro- Insurance regulation is essential in societies that endorse priately trained staff. These features have undermined effec- the concept of equal access to payers of health care, to ensure tive regulation and enforcement of rules, since political that companies cannot exclude high-risk individuals or costly imperatives prevail in the face of financial difficulties. Current preexisting conditions. Moreover, the cost increases fre- reforms aim to improve the scope, benefits, and solvency quently associated with largely private payers (as in Brazil, of the better-run Obras Sociales. Doing so, however, may South Africa, and the United States) cannot be contained exacerbate the regulatory agency's weaknesses, since it will effectively without incentives for controlling the cost and require it to play a larger and more complex role. volume of care. In short, governments can play a key role in promoting cost containment through insurance regula- Brazil. Private payers have burgeoned in Brazil in the past tion. But this function is secondary to fairness, efficiency, decade, and the variety of financing mechanisms has grown and financial accountability functions. commensurately. The Superintendencia de Seguros Privados, INNOVATIONS IN HEALTH CARE FINANCING 96 the Sao Paulo state insurance regulatory body, oversees Uruguay. Although its system is based on sickness funds, only indemnity insurance, which represents just 4 percent Uruguay takes a different approach to regulation, and has of all private financing. Almost no control or oversight is pro- a relatively comprehensive set of regulations. In 1981 the vided in the establishment, operation, and medical and finan- government made a legal distinction between for-profit pay- cial performance of the prepaid group practices, medical ers and the nonprofit sickness funds (called Instituciones de cooperatives, and company health plans (self-insured com- Asistencia Medica Colectiva, or IAMCs), and regulations panies) that together with insurance finance health services. for the two groups differ. For-profit insurance plans are accredited by the Ministry Colombia. Colombia's new health care reform law includes of Public Health, which requires them to meet certain a set of decrees establishing a health regulatory institution standards, defines norms for their establishment and oper- (Supeznntendencia de Salud) that is meant to take on many ation, and requires them to provide it with basic informa- of the functions outlined in the first section of this paper. tion on a regular basis. The Ministry of Economy and Finance The regulatory structure was modeled on those of OECD regulates financial aspects of the IAMCs. countries and is designed to address a range of issues that traditionally are not included in middle-income countries' Financial oversight health insurance regulations. The Supenintendencia is intended to ensure adequate structure, performance, and outcomes Monitoring the financial practices and solvency of private by accrediting private prepaid plans (to be financed through health insurers is probably the best understood and most vouchers), establishing basic quality standards and han- widely addressed issue in regulation. However, experiences dling consumer complaints, and ensuring the financial sol- and policy vary considerably across developing countries. vency of insurers bymonitoring reserves, liquidity and capital. In Chile regulation of financial solvency has been the sole responsibility of the Superintendencia de ISAPREs, the Hungary. All private insurance products in Hungary come regulatory body of private insurers, or ISAPREs (Instituciones under the aegis of the State Supervisory Authority of Insurance. de Salud Previsional). The superintendency monitors finan- Originally set up to oversee all insurance, the authority has cial reserves and ensures that the ISAPREs remain finan- been expanded to include for-profit health insurance com- cially viable. panies. The law was revised in 1993 to conform to European As the first such arrangement in LatinAmerica, the ISAPREs Union standards. Nonprofit insurance is regulated by the law were established for individual rather than group policies. As governing voluntary mutual insurance funds. Limited regu- such, they have engaged in aggressive "cream skimming" latory opportunities have left these institutions largely untested, (attracting low-risk, low-cost populations into private health but the necessary structure is in place to deal with a poten- plans). Because they are not required to accept any enrollees, tial market (Fuenzalida-Puelma 1996). the ISAPREs have refused to insure high-risk workers or the elderly. People with preexisting conditions and chronic dis- Korea. Korea's compulsory national health insurance law eases, as well as the elderly, have been forced to rely on the of 1989 expanded private health insurance based on fee- public health system. The lack of regulation has led to classic for-service payments, with the government setting fee sched- problems of adverse selection in private insurance. ules in consultation with other players. Further government Financial oversight is the most developed and system- regulations have not been developed. atic element of Colombia's new regulations. Insurance reg- ulations require that insurers be visited every three months, Turkey. Recent insurance reform in Turkey has included and an information system has been designed to allow track- establishment of the Department of Life and Health ing of financial flows and productivity. Insurers can invest Insurance in 1995 and the creation of an insurance super- in other enterprises, but the health investments must be a vision board to regulate the insurance industry. Health insur- legally autonomous entity (that is, a subsidiary company) ance oversight is almost nonexistent. with separate investment interests (Londofio 1997). The PRIVATE INSURANCE: PRINCIPLES AND PRACTICE 97 scope of the new regulations and the country's limited expe- access to emergency care, but must wait one year for surgery rience with this form of regulation have made implemen- or two years for medical care. Under pressure from the tation difficult. Indeed, the regulations place a large burden IAMCs, the government passed a decree that imposes entry on regulators, who must deal with institutions that are not restrictions on individuals over the age of 50 or 60. These accustomed to systematic government oversight. restrictions can be total (that is, rejection of applicant because Among sickness funds in Argentina, South Africa, and of age) or partial (constrained access to services such as Turkey, allbut SouthAfrica's Medical Schemes (sickness funds) medication or outpatient services). run chronic deficits that are financed with government sub- The strict regulation of nonprofits and the absence of sidies (TAI 1997; World Bank and IDB 1997). Although the regulation of for-profits have divided the market in much Medical Schemes are experiencing financial difficulties, the the same way it has in Chile. The for-profits (ISAPREs in lack of regulation inhibits the government from stepping in, Chile) have designed contracts that appeal to the young and and the weaker schemes are in danger of collapsing. the healthy, and the IAMCs (Ministry of Health in Chile) Argentina's Obras Sociales have recently run high deficits, are increasingly serving the elderly and the chronically ill. and some are insolvent. While the Argentine regulatory Thus the costs of the IAMCs keep rising, as they do in the agency, Administraci6n Nacional del Seguro de la Saluid, is public system in Chile. The IAMCs, however, are overreg- responsible for overseeing the financial practices of the ulated; it is difficult to become efficient when prices, costs, Obras Sociales, it has been largely ineffective. As a result and beneficiaries are beyond the insurer's control. the government has periodically been forced to step in to prop up ailing Obras. The system is currently being restruc- Nongovemmental regulation tured to address these and other deficienies. In countries where government has not provided a regula- Prohibitions on underwriting tory framework for private health insurance, private groups have started to step in to try to maintain the industry's Underwriting by insurance companies is economically effi- credibility and integrity. The industry's goals are to protect cient for the reasons discussed in the first section of this its market and survival-but unlike government, it does not paper: it is efficient to group homogeneous risks and to price seek to protect consumers. The standards set by the health insurance coverage accordingly. The social consequences of insurance industry are necessarily voluntary, and enforce- such practices are considered inequitable and unacceptable ment is problematic. The limited experience in middle- in many countries, however. As a result health insurance income countries is consistent with this principle. regulations often include measures to prevent or discour- Financial integrity and fairness in access are neither mon- age such practices. itored nor required under any specific Brazilian legislation In Colombia insurance policies are either group or indi- or legal decree. The concerned trade group for prepaid vidual, but individuals cannot legally be denied a basic pol- group practice, ABRAMGE (Associacdo Brasileira de icy-although certain conditions cannot be claimed within Medicina de Grupo), officially requested government regu- the first six months or year of the insurance contract. lation to protect the integrity of the industry, but the gov- Uruguay's nonprofit IAMCs are tightly regulated in all ernment refused. The group has created its own regulatory financial and access issues; for-profits have relative freedom body, CONAMGE (National Council for Self-Regulation in the scope and pricing of health insurance products. The of Company-Based Prepaid Group Practice), which repre- IAMCs have defined enrollment procedures and premiums; sents about 90 percent of all enrollees in such plans. The copayments and physician fees are set by the Ministry of association is attempting to provide guidelines that will Economy and Finance. reduce fiscal irregularities and false advertising. A spin-off The IAMCs underwrite potential enrollees and must association, the National Council for Self-Regulation of grant enrollment rights if there are no preexisting condi- Advertising, is considering setting up a regulatory arrange- tions. People with preexisting conditions have immediate ment for advertising among its members. However, both INNOVATIONS IN HEALTHI CARE FINANCING 98 are voluntary programs, and punitive actions against trans- dated. Insurance contracts are then transferred to other gressors have not yet been spelled out, limiting the clout of insurers. Chile's regulatory body cannot intervene when such efforts (World Bank 1994). insurers falter, but upon exit it takes over the firm, pays off Korea, lacking regulation or careful monitoring, has seen creditors (including patients awaiting reimbursements), and a rise in inefficient risk pooling, excessive technology acqui- liquidates assets (Oyarzo 1997). In Uruguay IAMCs were sition, and fraudulent billings. Some self-regulation has been originally required to have a minimum of 20,000 enrollees; introduced by the Federation of Korean Medical Insurance those below that level were merged or exited from the Societies, which represents a large portion of insurers. The market, with the insured transferred to other IAMCs. federation has established a National Insurance Appeals Committee to handle enrollee problems, developed a process Consumer protection for identifying designated treatment facilities, and overseen a program that provides training, routine inspections, and Consumer protection regulation is just beginning in non- audits to ensure proper operation of member insurers OECD countries. India's 1986 Consumer Protection Act (FKMIS 1986). Although these mechanisms do not neces- was passed with great fanfare and was an important start. sarily work well (Yang 1996), they provide an infrastruc- However, professional groups claimed that the law under- ture on which to build a potentiallyviable regulatory system. mined the physician-patient relationship, and weak enforce- Turkey's private health insurance industry views regula- ment has resulted in few changes. Public interest groups are tion as essential to maintaining its integrity. Given the lack now pressing for judicial intervention and government of government action, health insurance companies are set- enforcement of the law. Despite progress in filing complaints, ting standards for their members' organization and payment there are serious backlogs in bringing these cases to adju- of providers. They are not, however, setting financial per- dication (Bhat 1996). formance standards, and enforcing their policies is diffi- A similar situation exists in Brazil, where consumer pro- cult (Fuenzalida-Puelma 1996). However, these initiatives tection laws exist but adjudication under these statues is suggest the importance of regulation in protecting both con- difficult to assure. Options for individuals who have encoun- sumers and the health insurance industry in Turkey, as in tered problems tend to be protracted, and often result in all countries. unsatisfactory decisions from a less than transparent process (World Bank 1994). Orderly entry and exit In Chile the Superintendencia does not regulate bene- fits, is not responsible for quality lapses in provision, and is Controlling entry into the insurance market promotes finan- prohibited from giving information to consumers to influ- cial viability and ethical business conduct. The careful mon- ence their choice among competing private health funds itoring that occurs in the United States is difficult to replicate (Oyarzo 1997). in developing countries. Still, simple controls such as lim- Uruguay has attempted to introduce competition into its iting the types of investors and banning previous abusers social insurance system.The nonprofit IAMCs are closely reg- can help establish a process of orderly entry. Exit regula- ulated, but consumers can choose among them (although, as tions ensure that firms leaving the business do not leave discussed, there are restrictions among the elderly). Enrollees their clients without coverage or with large debts (because can transfer among the IAMCs subject to certain enrollment they have not been reimbursed for services consumed). requirements and a maximum hiatus of two months between Strict entry regulations are rare in developing countries; exit departing from and enrolling in another IAMC. policies are more common. In Colombia entrants into the market must be licensed Enforcement experience by the government and have minimum capital reserves. Troubled insurers are counseled, and firms that exit are Many countries lack an adequate regulatory structure, and taken over by the Superintendencia and their assets liqui- those that have regulations in place often cannot enforce PRIVATE INSURANCE: PRINCIPLES AND PRACTICE 99 them. Indeed, enforcement at all levels of government has ner rather than regulate it. Rigid price controls on private encountered difficulties in the countries for which there are insurance, generic space requirements, and infrastructure data. and technology specifications can generally be classified as Despite the logic and strength of Colombia's new regu- nuisance regulations, and ultimately provide little in the way lations, enforcement is having mixed results. Part of the dif- of true regulation.33 The objective of health insurance reg- ficulty, and an issue that has emerged elsewhere, is the ulation is to move away from controls to a set of incentives politicization of the regulatory function, which ultimately and disincentives that ensure the proper conduct of busi- undermines the effectiveness and credibility of regulation. ness. Most countries, however, have focused on bureau- Still, the gains from establishing an appropriate structure cratic requirements and have ignored the more difficult are a significant achievement (Londofio 1997). but essential functions for regulating private payers. Sickness fund systems in Argentina, South Africa, and The concept of price controls is an important one. During Turkey share some of these difficulties in regulation. First, periods of high inflation in the 1980s Brazil froze private because these systems are employment-based, there is no health insurance prices, including those for managed care competition among funds. Second, government regulations plans. This led to overcompensation in price adjustments often control the decisionmaking of the sickness fund admin- when controls were lifted, in order to hedge against a return istrators, leading to problems. Third, few of these arrange- to price controls. A similar pattern occurred with price ments ensure that abuses associated with health insurance controls in Uruguay. Both initiatives' sole objective was to (moral hazard) and enforcement of the social contract stop price increases-the reasonableness of health insur- (adverse selection) are mitigated through legal and regula- ance prices was never considered. Such policies distort the tory measures. Finally, none of the funds is accountable to market for health insurance, and in the end have little impact enrollees, firms, government, or reguiatory bodies. on overall prices. Argentina's regulatory agency, ANSSAL, has had trou- In the United States some states approve insurance rates. ble with enforcement partly because of its complex rela- These typically involve general targets for premium increases tionship with the sickness fund owners who lead trade unions. based on loss ratios (that is, costs/revenues). This approach Indeed, recent reforms suggest that ANS SAL will need to allows profit levels to be contained while allowing health strengthen its role to include overseeing enrollment, pro- insurance companies maximum flexibility in management viding technical and financial assistance for mergers, and and sufficient profit to ensure competition in the market. closing Obras Sociales. Despite this increased role, ANSSAL Some regulations are too specific and have offered per- continues to be challenged by enforcement (World Bank verse incentives. In 1989 Uruguay's government decreed and IDB 1997). the necessary infrastructure, equipment, and inpatient capac- In Brazil there are no institutions to protect against finan- ity that the private sickness funds (IAMCs) were to possess cial insolvency or sloppy exits from the market, a threat based on an assessment of resource availability and the that could undermine the country's entire private insurance perceived effectiveness of those resources. Financial feasi- market. Some forms of regulation have been tried in vari- bility was defined according to national and regional health ous Brazilian states, but weak and politicized enforcement capacity as well as financial constraints. These were adopted has undermined the effectiveness of these experiments in another decree that set IAMC investment levels and (World Bank 1994). increased monthly premiums accordingly. The result has been overcapacity, rising maintenance costs, and higher uti- Inappropriate regulations lization, part of which government finances and part of which comes from private insurance premiums. Regulation is difficult for most countries, especially in the Most health sector regulations in developing countries health sector. While there is an important role for govern- are weak, and regulatory structures (where they exist) are ment in many areas, there is also a risk of inappropriate often unenforced or inappropriate.34 Weak regulations regulations that control the sector in an unreasonable man- discourage private insurance because there is a dearth of INNOVATIONS IN HEALTII CARE FINANCING 100 competent providers from whom to purchase health care professionals) has a direct bearing on health insurance reg- services. Moreover, when payers and providers merge (as ulation and the regulatory environment in general. they do under managed care), regulations on the organiza- These lapses-either in terms of impeding growth of tion and delivery of health services have an increasing impact the private sector or allowing it free rein with no controls on health insurance regulation, its structure, and enforce- or policy signals from government-lead to an uncertain ment. Hence, although a separate issue, regulation of business environment and an inhospitable climate for pri- providers (hospitals, clinics, physicians, and other health vate investment in health insurance. Conclusions and Lessons Learned Private insurance exists in almost every country, and its poten- Thus insurers have developed a number of underwrit- tial is significant. In countries that already have a large private ing and pricing practices to avoid accepting sick people when insurance sector, it is likely to remain, since reversing gov- they apply for coverage. From a purely economic perspec- enmnentpolicytoward amajorconsumerindustrycanbepolit- tive, these techniques can improve economic efficiency, ically difficult. Whether countries choose to encourage, although they may be socially unacceptable. These practices discourage, or simply accept private insurance, it is important expedite the formation of homogeneous risk pools in which that governments understand the health insurance industry. participants' expected costs are similar. When these risk Private health insurance has both benefits and pitfalls pools form and are priced efficiently, they reduce involun- relative to purely public financing systems. Private insur- tary cross-subsidies among participants-a result that is eco- ance can improve the availability and quality of health. It nomically efficient. But they can also reduce the ability of can help health care systems rebuild infrastructure and amor- sick people to find insurance, increase administrative costs, tize needed investment when government financing for and reduce the percentage of health care expenditures health care is inadequate. It can offer consumers a choice that actually finances health care. By attracting healthier of providers, and thereby generate incentives that reward risks, insurers can quickly erode the broad risk pooling high-quality providers and penalize poor quality (at least as that social insurance attempts to create, straining the social perceived by consumers). Similarly, private insurers are contract's ability to ensure the universal, equal coverage becoming increasingly adept at distinguishing efficient from that many countries have embraced. inefficient providers, and can develop methods that sys- Moreover, the ability of private insurance systems to tematically reward efficiency. (By contrast, public financ- improve access and quality, and to help recapitalize health ing systems often reward efficiency and inefficiency alike.) care capacity, can cause health care costs to accelerate as Finally, although private insurance can siphon off lower-cost private insurance grows. Greater access produces greater people and leave the public sector with higher average costs, use of health care services; the acquisition and use of costly it can reduce the total fnancial burden on government. technologies further increase costs. The challenge to regu- The pitfalls of private insurance are, in many cases, the lators-and to private insurers that hope to expand private flip side of their advantages. Private insurance systems are coverage-is to keep private insurance affordable while naturally dynamic. When private insurance plans compete encouraging ongoing improvements in the quality of health with one another for enrollees (or when one or more pri- care. This is not a simple task. vate insurance companies compete with a public insurance Other potential problems of private insurance are unre- program), the composition of any one insurer's risk pool lated to its advantages: financial instability, unethical behav- can change as consumers move from plan to plan. Plans ior, and unreliable consumer orientation. Governments can that experience adverse selection, even when they are well- resolve these problems, but only if they establish clear, prob- run and efficient, can fail. lem-focused regulations and enforce them systematically. PRIVATE INSURANCE: PRINCIPLES AND PRACTICE 101 Regulation of private insurance can take many forms. * Regulation can be tooperfect. Governments that have had Depending on its nature, scope, and enforcement, regula- very little regulation may find it impossible to leap to a tion can have different effects on consumers and providers, comprehensive system of insurance regulation. For exam- and therefore on costs. In Germany all aspects of the pri- ple, Colombia's new comprehensive set of insurance reg- vate, noncompeting sickness funds are so closely regulated ulation is an impressive move by the government to ensure as to be quasi-public institutions; company behavior and oversight of insurers and health care providers. But the performance are tightly controlled. At the other extreme, regulation imposes an abrupt shift in culture. As a result most non-OECD countries have almost no oversight or reg- the breadth of the changes and the adoption of sophis- ulation. The absence of regulation can have effects on the ticated methods have been difficult to establish and to industry that are as perverse as those caused by excessive enforce. A better approach may be to start with a core and irrational regulation. set of regulations that rectify egregious problems; once Regulation is essential for any country that has a private these regulations become effective, regulators can turn insurance sector. Some of the most important lessons for systematically to other areas of concern. Tradeoffs need regulators about health insurance include: to be made between regulatory needs and the ability to * Know your industry and the major players. Information effectively enforce regulations. about the practices of companies and their managers, * Regulations must be effective and offer proper incentives. and about the performance and financial integrity of In many countries nonfinancial regulations are nuisance health insurers, is essential to preserve the industry and regulations. They create significant bureaucratic red tape to protect consumers. Regulators who are unfamiliarwith and have little impact on quality or other desirable out- the specifics of their country's insurance industry will comes. For the insurance industry such regulations merely be unable to anticipate and prevent abuses. raise costs. To be effective, regulations must have a clear * Allhealth insurance systems must balance efficiency with cul- objective, and the laws and decrees put in place must tural notions offairness. Countries that have adopted a serve that end as directly as possible. The experience of sweeping social contract, and seek to ensure absolute a number of countries suggests the need to revise and equality to all, will have important problems with the eco- craft legislation to strengthen appropriate incentives and nomic efficiency of the insurance system they create. to discourage pernicious company behavior. Conversely, countries that allow insurers to isolate sick Regulation without enforcement is like a tigerwithoutteeth. people into high-cost insurance pools, and offerno means An effective regulatory system relies on an effective to subsidize people with high health care costs, will have enforcement strategy. Regulations that succumb to polit- a system that may be economically efficient (in that it ical imperatives or that allow exceptions undermine the does not force involuntary cross-subsidies) but that most process and purpose of regulation. Sound regulatory people will consider unfair or inadequately protective. All requirements will be ineffective if they cannot be enforced insurance systems, public and private, must strike a bal- or are not enforced systematically ance between economic efficiency and equity. Where they Finally, even countries that want to retain a broad pub- lie on the spectrum will be determined by each country's lic insurance program to finance health care can benefit social and political culture, and by its economic resources. from understanding and judiciously using principles of pri- * Health insurance companies, whetherfor-profitornonprofit, vate insurance. If it is encouraged and directed toward must make a profit. Government can regulate insurance explicit system goals, competition is a powerful tool for companies' profit levels (usually by regulating price improving efficiency and quality. For example, the Czech increases), but preventing profit will undermine the com- Republic has initiated competition within its public insur- pany and the industry. Without profit, insurers have no ance program, allowing regional government-sanctioned means of financing investment in efficient management insurance plans to form and compete with the national plan. systerns and effective systems of health care cost As might be expected, some of the new regional plans are management. doing well: they are attracting growing enrollment. They are INNOVATIONS IN HEALTH CARE FINANCING 102 also causing the national plan some discomfort, since it and culture, but in most countries it is well-established. believes that it is being left with sicker enrollees. This expe- The challenge to governments is to regulate private insur- rience bears observation, but in the short run it will moti- ance effectively: to assure the financial stability of insur- vate an ongoing search for ways to improve administrative ers and the integrity of insurance contracts, and to strike efficiency and customer satisfaction in all plans. a balance between the social acceptability of private insur- Private health insurance offers benefits and costs to ance and insurers' competitive drive toward economic society. Its importance and applicability vary by country efficiency. PRIVATE INSURANCE: PRINCIPLES AND PRACTICE 103 Annex table I Roles, characteristics, and extent of private health insurance coverage and main alternatives, OECD countries Share of population with Country Role of private insurance Characteristics of private insurance private insurance Main alternative source of financing Australia Supplemental to universal Pays only for hospital care in private 45 percent Public universal program (Medicare) financed from the public coverage facilities, including inpatient physician general income tax. charges that exceed the public fee rate for inpatient care. Belgium Supplemental coverage for Pays for ambulatory, nonsurgical care. n.a. Compulsory social insurance covers risks for the entire the self-employed populabon and minor risks for all but the self-employed. Canada Supplemental to universal Only covers services not covered n.a. Universal public insurance plan financed from nabonal public coverage by the provincial public plans. and provincial general revenues. Physirians are in private practice and predominantly paid on a fee-for-service basis. Balance billing to patients is prohibited by law. France Supplemental to universal Pays for private hospital and physician n.a. Comprehensive public insurance program. Private public coverage charges that exceed contractual fees provider fees are negobated by the public insurance paid by the public program. About 25 program. Supplementary private insurance to cover percent of physicians (mostly specialists) cost sharing is common. balance bill pabents for care. Covers required deductible. Germany Altemative to universal People who opt out of the statutory 9 percent Universal entitlement for the statutory health insurance public coverage system in favor of private insurance (1986) system comprising 1,200 self-financing sickness funds cannot ever rejoin the statutory system. organized by geographic area, finm, trade, or craft. Membership is compulsory for people within a specified income limit and for rebred people who participated while working. About three-quarters of participants in the statutory system are mandatory; one-quarter participate voluntarily. Ireland Supplemental to universal Voluntary private insurance, sold by a 30 percent Public program is funded from general taxation and public coverage, with two monopolistc statutory health insurer to (1991) provides comprehensive benefits. About one-third of levels of benefit coverage people who are partly eligible for public the population relies exdusively on the public insurance program coverage, covers physician fees program with no out-of-pocket payments for care; the and private hospital accommodabons rest of the population is partly eligible. Of these, more (including private beds in public hospitals). than half pay for some care out of pocket, but have no supplemental private insurance. INNOVATIONS IN HEALTH CARE FINANCING 104 Annex table I Roles, characteristics, and extent of private health insurance coverage and main alternatives, OECD countries (continued) Share of population vith Country Role of private insurance Characteristics of private insurance private insurance Main alternative source of financing Netherands Alternative to universal Mandatory regulated private insurance 30 percent Compulsory universal public coverage for chronic public coverage coverage for acute care for higher- (1991) care; public coverage for acute care is compulsory for income patients; insurance plans about 70 percent of the population. reimburse patients. Spain Supplemental to universal Typically covers supplementary n.a. Compulsory national system of public hospitals and public coverage voluntary payments for hospital and publicly salaried physicians, funded from general physician care. revenues and social insurance contributions. Some provinces have their own social insurance system. United Supplemental to universal Covers acute care services, supplements 10 percent Comprehensive, compulsory national system of public Kingdom public coverage or substitutes for payment by the (1990) hospitals and publicly salaried physicians, financed from National Health System. general revenues. Patients are assigned to primary and institutional care providers largely by place of residence. Efficiencies have emerged from recent reforms that introduced systems of provider accountability, not on competition among plans or providers. United Primary, voluntary Primarily a tax-exempt, voluntary 71 percent of Public assistance coverage is available to poor individuals States coverage for people benefit from self-insured or insured population in specific drcumstances (under age 13, over age 65, ineligible for the national employer plans. About 7 percent of under 65 or permanently disabled); eligibility and coverage varies social insurance program for people under 65 buy private individual (1995) by state. Neariy all persons over 65 qualify for cover- the elderiy (Medicare) or (nongroup) coverage. About one-third age from the social insurance program (Medicare). federal-state public assistance of elderly public insurance enrollees buy health program (Medicaid). supplemental private coverage. In any Public insurance enrollees year, i 6-19 percent of the population may buy supplemental under 65 is uninsured all year. private insurance. Source: Aftman and Jac6son 1991; Chollet 1996; Day and Klein 1991; Enthoven 1991; Hurst 1991; Reinhardt 1995. PRIVATE INSURANCE: PRINCIPLES AND PRACTICE 105 Annex table 2 Roles, characteristics, and extent of private health insurance coverage and main alternatives, non-OECD countries Share of population with Country Role of private insurance Characteristics of private insurance private insurance Main alternative source of financing Argentina Additional to social welfare White-collar workers typically redirect 19 percent Compulsory social welfare funds operated by funds run by trade unions 75 percent of mandatory health con- govemment-affiliated trade unions, as well as tributions to prvate, prepaid medical federal, provincial, and municipal health services plans that vary in coverage. The plans (22 percent), social insurance (36 percent), and normally contract with prvate hospitals. out-of-pocket payments (23 percent). Brazil Additional to publicly Most enrollment is in prepaid group 25 percent Compulsory public insurance program and federal, financed care from largely practices, medical cooperatives, or (includes state, and municipal services. Out-of-pocket private facilities employer plans that own or contract managed care): expenditures are significant. with health care facilities. indemnity plans are 4 percent of total Chile Altemative to the public Benefits are tied to amount of contri- 27 percent FONASA collects 7 percent compulsory health health system bution and age, sex, and number of insurance premium and allocates them to ISAPREs dependents. Large companies provide or the Ministry of Health. The ministry's system is complementary coverage or create subsidized by the govemment. their own private ISAPRE. Colombia Supplemental to public New entities called EPSs may be state I I percent Half the population is affiliated with the public coverage run (6), private (I 2), or nonprofit (7). insurance system, which is supported by a EPSs compete with the public system. compulsory wage tax that also finances the EPSs. Costa Pica Additional to public There are no private (health) insurance Low Good-quality public social security system. coverage companies. The wealthy use private facilities whose services are financed by new payers and providers. Cote Additional or supplemental Most insured are covered by their 7 percent Ministry of Health facilities. Social security system d'lvoire to public options employers. Insurance payments are for formal workers provides generous benefits and made to providers based on a "payment less generous mutuelle for govemment employees guarantee" issued to patients prior to and dependents. More than half of expenditures hospital admission. are private. Czech Complementary to public None Compulsory universal social insurance with Republic insurance contributions to I of 20 licensed insurance companies with generous, publicly mandated medical and dental benefits. Dominican Addiional to social security Largely employer-provided care in 14 percent Poor-quality social securty and Ministry of Public Republic and supplemental to other prepaid, PPO, and indemnity plans. Health services. Expenditures are 10 percent public services social security, 20 percent ministry, and 52 percent private (NGOs and for-profit). Ecuador Altemaive to public Employer-operated medical plans 12 percent Poor-quality social security system and Ministry of coverage; additional to finance use of private clinics and Health services. Expenditures are 17 percent social security doctors. social security, 14 percent ministry, 63 percent private, and 6 percent other. Egypt Altematve/supplemental Employer-run and -financed clinics. 3 percent Poor-quality compulsory Health Insurance to public coverage Typical private plans are noncontributory, Organization and Ministry of Health services. cover all employees, and reimburse Expenditures are 30 percent govemment, 54 costs up to a fixed level. percent households, and 9 percent social insurance. Guatemala Addiional to public Employer-provided, prepaid group 5 percent Poor-quality social security and Ministry of Health coverage health insurance for non-union services. employees. Comprehensive major medical plans. HMOs and PPOs have emerged as other options. INNOVATIONS IN HEALTH CARE FINANCING 106 Annex table 2 Roles, characteristics, and extent of private health insurance coverage and main alternatives, non-OECD countries (continued) Share of population with Country Role of private insurance Characteristics of private insurance private insurance r Main alternative source of financing Hungary Supplemental to public Many companies grant free medical Low Social security based on contributions. coverage consultations in-house or in private clinics. Recent modification of law will increase private provision. India Aftemative to public Expensive medical insurance provided 3.3 percent System funded and operated by federal and state coverage (operated at to the rich by employers. Govemment govemments. Expendiures are 75 percent state level) employees have their own system. out-of-pocket, 6 percent federal, and 16 percent state govemment. Indonesia Alternative to public Employer-provided plans wdh Minimal Public system OAMSOSTEK). Expenditures are coverage intemal limitations or coinsurance 70 percent out of pocket, 4 percent of which is for to limit costs. HMOs and PPOs private insurance. becoming increasingly popular. Jamaica Altemative to public Typically covers outpatient and some 15 percent Ministry of Health services with some user fees, coverage inpatient care. but free for the indigent. Jordan Altemative to public Mainly provided as a health benefit 12 percent CMI Insurance Program for govemment employees coverage of large firms. and the indigent, and Royal Medical Services for the military. Both are financed by payroll tax and general revenues; cover 58 percent of the population. About 53 percent of spending is private. Honduras Addiional to public Employer-provided plans through 1.5 percent Expenditures are 9 percent for social security coverage prepaid and indemnity plans. (formal sector workers), 29 percent for Ministry of Health, and 61 percent out of pocket. Kenya Altemative to public Employer-operated, noncontributory 1 1.4 percent State-run compulsory insurance system covers coverage plans of urban-based companies that people in formal employment and their offer coverage for dependents. Five dependents-about 25 percent of the population. companies are active. Direct employer contracting with providers (self-insuring) is more common. Korea, Compulsory health 417 autonomous health insurance 94 percent 6 percent of population covered by public Rep. of insurance funds; restricted access to high Medicaid insurance program. Govemment technology and ceilings on inpatient subsidizes private funds. care. All funds have deductibles and copayments for doctor visits, outpatient services (30-55 percent), and inpatient care (20 percent). Lebanon AJtemative to NGO and 90 private insurance companies have 8 percent Three public insurance schemes cover private public insurance entered the market recently anticipating employees (28 percent), civil servants (9 percent), economic recovery. Copayments are and the military (I I percent). Expenditures are 3 1 standard for most services. percent govemment, 61 percent private sources, and 8 percent donors. Malaysia Aftemative to public Large or foreign employers provide 3 percent Well-run public system. coverage private plans with different levels of coverage for categories of employees. Mexico Additional to public Large employers provide coverage to 1-2 percent Social security systems for formal sector workers coverage and compulsory non-union employees. Typical plans are and govemment employees, and Ministry of social insurance for formal noncontributory, but there is a trend Health for the poor. About 47 percent of sector workers toward providing higher benefits and expenditures are private. employee contributions. (Table continues next poge) PRIVATE INSURANCE: PRINCIPLES AND PRACTICE 107 Annex table 2 Roles, characteristics, and extent of private health insurance coverage and main alternatives, non-OECD countries (continued) Share of population with Country Role of private insurance Characteristics of private insurance private insurance Main alternative source of financing Nigeria Supplemental to public Recent introduction of medical 0.4 percent Free treatment at govemment dinics, though coverage insurance on a limited basis in tandem patients often pay for medication, medical aids, with a self-funded arrangement. and so on. Typically noncontributory. Pakistan Altemative to public Multinabonal and public sector Low State-run system through Ministry of Health and coverage employees receive almost blanket limited Employee Social Security Insurance. benefits from their employers. Limited otherwise. Panama Supplemental to public Companies commonly provide group Low Public social security system and Ministry of Health coverage medical insurance for white-collar and services. sometimes plant workers. Typically non- contributory. Peru Altemative to public Group medical insurance for white- 6 percent Expenditures are 36 percent public social securty coverage; additional to collar workers, characterized by system, 30 percent Ministry of Health, and 28 social security deductibles, coinsurance, cost sharing, percent out-of-pocket. and maximum benefit levels. Philippines Supplemental to public Two categories of group medical 1.6 percent Medicare program through compulsory social coverage insurance: basic hospitalizaton (for security system (covers 8.5 percent), and growing typical medical expenses) and major managed care options. About 54 percent of medical (pays for a percentage of covered spending is out of pocket. medical expenses in excess of a specified deductible amount). Usually paid by employers. Poland Limited High social security taxes (46 percent Minimal Free government-provided health care. Patients of gross salary) restrict health as an cover half of all costs through informal payments, employee benefit, but investment is and 29 percent indicate that they pay 'gratuities." pending. Singapore Altemative to public Insured or self-financed private Low High-quality public system of Central Provident coverage medical schemes are provided by Fund, Medisave, and Medishield allows consumers companies on a scheduled to choose extent of coverage. reimbursement basis for outpatient and hospitalization benefits. Major medical not common. South Altemative to employment- Traditional medical schemes (sickness 16 percent of Limited govemment system. Expenditures are 45 Africa based sickness funds; public funds) and insured plans, most of which population (69 percent public, 34 percent medical schemes, 3 coverage is not widespread require employee and employer percent of percent insurance, and 14 percent out-of-pocket. contributions (evenly split). AJI medical whites; 7 percent plans under severe cost pressures. of blacks) Sri Lanka Altemative to public Employer-provided medical insurance 2 percent Low-quality public hospitals. About 47 percent of coverage for executives; some contributory the population uses private health care services, insurance for others. Practice varies and 60 percent of total expenditures are out of among the five insurance companies, pocket. active mostly in Colombo. Thailand Addibonal to public Common for companies to provide 0.7 percent In 1992 expenditures were 24 percent public coverage medical insurance for all employees on services, 2 percent other public insurance, and 74 a noncontributory basis, in the form of percent out of pocket. scheduled indemnity programs. INNOVATIONS IN HEALTH CARE FINANCING 108 Annex table 2 Roles, characteristics, and extent of private health insurance coverage and main alternatives, non-OECD countries (continued) Share of population with Country Role of private insurance Characteristics of private insurance private insurance Main alternative source of financing Turkey Additional to compulsory Most insurance plans are individual 500,000 Three social security systems of variable quality. social insurance for 65 rather than group policies offered policyholders Ministry of Health covers the rest. Two-thirds of percent of eligible by 29 insurance companies. payment are private. population Multinational companies offer health care benefits to Turkish managers and some office workers and dependents. Uruguay Supplemental insurance Sickness funds cover 55 percent of the 5.9 percent Private sickness funds finance defined benefits population through own and contracted (48 percent of expenditures). Expenditures also facilities. 35 supplementary insurers offer covered by social securnty, Ministry of Public Health medical, surgical, dental, diagnostic, and (16 percent ), and out of pocket ( 14 percent). mobile/emergency plans. Venezuela Additional to public Employer-contracted medical insurance Low Public social security institute and Ministry of Health. coverage common in offices with 50 or more employees. Typically contributory and includes dependents. Zimbabwe Altemative to public Private mutual medical organizations 5 percent Poor-quality state-run system for low-income clinics with reasonable level of benefits. people. Employers and employees typically split costs. Membership common for middle and upper classes. Source: Bhat 1996; California Consortium 1996; Femandez 1997; Fuenzalida-Puelma 1996; La Forgia and Couttolenc 1993; La Forgia anrd Griffin 1993; Lewis 1988; Lewis and Medici 1995; William M. Mercer Lmited 1996; Nittayaramphong and Tangcharoensathien 1994; Rafeh in this volume; Shaw and Griffin 1995; TAI 1997; World Bank 1992, 1994, 1995, 1996, and forthcoming; WVorld Bank and IDB 1997; Yang 1996. PRIVATE INSURANCE: PRINCIPLES AND PRACTICE Notes adequately pool risk. Insurers that experience an underwriting spi- ral in a particular insurance product typically will cancel that 1. In competitive markets the price of insurance coverage is product if they are unable to stabilize experience by redesigning equal to the expected value of loss, plus administrative and mar- the product to retain low risks. keting expense, plus an additional fee (called a loading). In gen- 7. Over time this behavior will cause insurance pools to dis- eral, insurance loadings are larger as consumers are more risk solve and form in ways that reduce differences in the risk embod- averse (that is, as consumers are willing to pay more to avoid ied in any one pool. Called separating equilibrium (Rothschild and risk). As the probability of loss approaches 1 (that is, certainty), Stiglitz 1976), the propensity of consumers to separate themselves the cost of insurance may exceed consumers' cost of sustaining into relatively homogeneous risk pools is a well-recognized phe- the loss, even if consumers are risk averse. Consumers are more nomenon, especially in competitive insurance markets. likely to insure against high-probability losses and high-magnitude 8. Tiered rating helps stabilize insurance markets as follows. losses in competitive markets (where the price of insurance equals Because consumers gravitate toward risk pools that include mem- the marginal cost of producing insurance) than in monopolistic bers whose risk is similar to or less than their own, insurers that markets (where the price of insurance exceeds marginal cost). attempt to aggregate dissimilar risk (and set a single price regard- 2. For high-risk populations the price of insurance reflects the less of differences in individuals' risk) will see an exodus of low- high probability of high loss. However, the price may also include risk participants from the pool. By aggregate pricing only similar a deterrent surcharge, reflecting the absence of a competitive mar- risk and pricing insurance accurately at the outset, insurers hope ket for these risks. to avoid an underwriting spiral (see note 6). 3. In the professional literature consumers' greater informa- 9. Many insurance plans exclude coverage for any procedure tion about risk is called asymmetric information. Ligon and Thistle that is deemed experimental. In the United States insurance exdu- (1996) provide a technical presentation of possible contractual sions for experimental procedures are controversial; many patients remedies to asymmetric information problems in insurance mar- (and their physicians) feel that the transition from experimental kets, although not specifically with regard to health insurance. procedure to accepted practice is a matter of professional judge- 4. In the extreme case consumers would buy health insurance ment. No single agency, public or private, judges which procedures not to manage risk, but to gain a subsidy for perfectly anticipated are experimental and which are not. While the insurance contract medical expenditures. The magnitude of the subsidy to a con- may identify specific procedures that it deems experimental and sumer with asymmetric information is positively related to the will not cover, in most cases each insurer's medical director makes difference between the anticipated expected value of personal loss a decision about whether a specific procedure is deemed experi- and the "pure" premium-that is, the average expected value of mental at the time that the insured patient requires care. loss among the insured population. Consumer demand for insur- 10. In the broadest sense, the rapid adoption of new and costly ance in markets with asymmetric information generally is assumed medical technology that occurs when most consumers are insured to be greater as the opportunity for such a subsidy increases. (despite improvements in the quality of care) also may be viewed Such consumers would include people who have either an unusu- as moral hazard. Thus private insurance in many countries may not ally high probability of illness or even a slightly higher probability cover advanced forms of medical technology. For example, in Korea of a very costly illness. private insurance typically excludes expensive high-technology med- 5. An insurance pool (or risk pool) is a collection of individual ical services such as computerized axial tomography (CAT) scan- risks that corresponds to a specific price for an insurance prod- ning, magnetic resonance imaging (MRI), positron emission uct. Each insurer may forn many pools, each of which is defined tomography (PET) scanning, and some chemotherapy (Yang 1996). by the specific insurance contract and the group or class of insured 11. This source of moral hazard is generally considered indis- risks over which the price is calculated. A risk pool may include tinguishable from the phenomenon of supplier-induced demand: either unrelated individuals or groups. the propensity of health care providers to recommend more care 6. The logic of this is as follows: If an insurer sets the price of when consumers are insensitive to price. Of course, full insur- coverage too low relative to the plan's medical losses, it will be ance is the main reason for consumers' price insensitivity forced to raise the price at the first opportunity. But by raising the 12. Critics of managed care argue that managed care plans also price the insurer will lose low-risk enrollees and retain high-risk can discourage sick people from enrolling in any number of ways- enrollees, because the new price is high relative to low-risk con- for example, by establishing a relative small network of specialty sumers' expected value of loss. When that happens, the insurer providers, by excluding primary care providers who have estab- may enter an underwriting spiral (sometimes called a death spiral), lished caseloads of sick people, and by limiting coverage for high- a repeating sequence of setting higher insurance prices and los- cost services such as mental health care. ing low-risk enrollees. The ultimate consequence of an under- Indeed, reviewing the research literature on selection bias in writing spiral is the loss of the insurance arrangement's ability to U.S. managed care plans, Hellinger (1995) concludes that health INNOVATIONS IN HEALTH CARE FINANCING 110 plans that restrict the choice of enrollees to a specific set of providers an unlicensed insurer. However, when they do buy from an unli- (that is, health maintenance organizations of all types and exclu- censed insurer, they are not protected by regulatory standards. sive provider organizations) often do experience favorable selec- 19. Most U.S. states allow only certain classes of assets to be tion. This result holds for both the elderly and nonelderly population admitted to the balance sheet, and many restrict the share of assets enrolling in managed care plans. However, over time the selec- that insurers can hold in certain categories. For example, in most tion advantage experienced byHMOs tends to decline as enrollee states common stock cannot account for more than 10 percent of utilization regresses toward the mean (Welch 1985). alife insurer's assets. Some states alsolimitinvestmentsin medium- 13. Extensive research on the costs of managed care plans con- to lower-grade bonds (Skipper 1992). sistently concludes that the main reason for the lower cost of 20. In the United States the National Association of Insurance managed care relative to financial insurance plans is the lower Commissioners-a voluntary association that advises states about rate of hospitalization, controlling for patient diagnosis and other regulatory standards and practices-is developing risk-based relevant characteristics. capital measures and standards. Risk-based standards would vary 14. In addition, large employers usually employ professional minimum levels of insurer capital and surplus by the financial risk benefits managers who are accessible to the insurer and able to of the instruments in which these funds are invested and by the promptly resolve questions about enrollment and claims. U.S. insurer's operational characteristics. This initiative is expected to insurers consider the ability to communicate efficiently with large- encourage greater uniformity of capital and surplus requirements group customers to be an important aspect of the lower adminis- among states. trative cost of that business. 21. Some U.S. states do not require that reinsurers be licensed, 15. The reasons for this are the inverse of those that make large- since direct insurers are considered sophisticated buyers. However, group customers attractive. Specifically, marketing to many small uncollectible reinsurance has been a source of some direct insurer groups is more expensive than marketing to fewer large groups, insolvencies. Some states indirectly regulate reinsurance opera- and may not differ appreciably from marketing to individuals. tions by not allowing direct insurers reserve credit for reinsurance Moreover, small groups are much less likely than large groups to placed with unauthorized reinsurers. be able to afford a full-time professional plan administrator, mak- 22. Such actions generally require a court order. In some states ing the resolution of questions and disputes about coverage more the regulator may assume control of an insurer without a prior difficult and time-consuming for the insurer. Because of the greater court order if the regulator decides that immediate action is needed risk and cost of insuring small groups, insurers typically under- to preserve assets. However, such "quick take" actions are sub- write small groups very carefully to avoid adverse selection. ject to subsequent judicial review (Skipper 1992). In some cases 16. For at least two reasons, insurers may redline workers in where an insurance regulator has assumed control of an insurer, some occupations or industries even if those workers have sepa- the regulator has discovered illegal behavior by members of the rate insurance to pay for job-related injuries or illnesses. First, insurer's management or governing board, and has initiated civil such workers are more likely to pose difficult (and therefore costly) or criminal prosecution. problems for claim adjudication: every claim would have to be 23. A holding company is a parent corporation that typically examined as beingpotentiallyrelated to theiremployrment. Second, owns all or most of the stock of a number of constituent sub- workers who suffer health problems from past exposure or injury sidiaries, but occasionally may own less than a majority share based cannot easily bring a claim against past insurers of job-related ill- on control or investment motives (Munn, Garcia, and Woelfel ness. Instead, they will bring the claim against their current health 1991). insurer. Thus insurers mayredline to protect themselves frominad- 24. U.S. state insurance regulators commonly require that an equacies in other insurance sectors (specifically, insurance for job- affiliated insurer construct a corporate "firewall"-that is, a pro- related injuries) and from adverse selection by workers who are vision of incorporation that formally protects the insurer's finances aware of past exposures or injuries but may conceal them from an from insolvent affiliates. insurer. 25. Each U.S. state operates two guarantee funds (by con- 17. In the United States the best managed care plans are increas- vention, for life and health insurers and for property and casualty ingly active in developing and using medical practice protocols to insurers) to compensate policyholders for financial loss due to improve the efficiency and quality of care. However, the devel- insurer insolvency. As an artifact of their historical development, opment and implementation of practice protocols can generate health insurers may be cdassified in either group. Differences among significant administrative costs. In general, financial insurance states in the amounts guaranteed, persons eligible for protection, plans are much less inclined to invest in measuring and improv- extent of coverage, premium base for assessment, caps on assess- ing the quality of care, since they do not affiliate with any partic- ments, and rights to recoup guarantee payments can lead to very ular set of health care providers. uneven economic results and drawn-out litigation over large 18. In the United States and many other countries individu. amounts of money when a multistate insurer becomes insolvent als and businesses are not prohibited from buying insurance from (Skipper 1992). PRIVATE INSURANCE: PRINCIPLES AND PRACTICE 111 26. In the United States the National Association of Insurance risk is nonrandom, and reinsurers will be reluctant to do business Commissioners maintains two electronic databases to help states in those markets. Such a situation may arise if the introduction of in this effort. The Regulatory Information Retrieval System con- health insurance is likely to cause a sudden acceleration in providers' tains information on people and companies against whom regu- charges-eitherbecause more care is delivered orbecause providers latory actions have been taken. The Special Activities Database are able to shift costs (that is, to charge privately insured patients facilitates the confidential exchange of information among regu- more than publicly insured or uninsured patients for the same ser- lators, allowing them to inquire into the activities of companies vice). Because managed care companies typically pay negotiated and individuals of regulatory concern, including people who may rates to hospitals and specialty providers and may capitate primary be involved in fraudulent activities. health care services, they may be more attractive to reinsurers in 27. Every U.S. insurer is required to retain an outside firm of developing markets than financial insurance plans that have fewer certified public accountants that is accountable to the insurer's gov- options for controlling plan costs. erning board. Regulators sometimes have the right to approve the 32. During the past decade manyU.S. states have enacted laws choice, and insurers typically retain a large national accounting firm. requiring that health care providers prove their services are a lower- The accounting firm periodically audits the insurer (reviewing its cost alternative to more conventional medical treatment before fnancial statements) and presents the audited financial statements the legislature will consider proposals to require that their ser- to the insurer's board with an opinion on whether the insurer's vices be covered. Few states that have enacted such a require- financial statements conform to generally accepted accounting prin- ment have subsequently enacted any laws requiring coverage for ciples as well as statutory accounting principles. The accounting specific services or providers. firm usually also submits a management letter to the board-in 33. Variants of these controls are obviously valid in some set- effect, a "report card" on the insurer's fnancial management. tings-for example, space requirements for technology to pro- This system is not perfect, however. Outside accountants are tect health workers and patients and minimal standards for specific not always thorough, and insurers can hide financial distortions. providers (for example, requiring that providers have the neces- The accounting firm that conducts the audit may be friendly to saryequipment before government finances those services), among the insurer's management, especially if the insurer also retains others. them for consulting services that generate more revenue than the 34. Other elements of the health system-controlling hospi- audit. While the board should expect the accounting firm to be tal construction, licensing health-related businesses, medical prac- responsive to management in operational matters, but commit- tice, consumer protection-are much less common, not enforced, ted to the board in evaluating fiscal condition and management, and/or counterproductive. Among these the most common are striking that balance can be difficult (Akula 1997). licensing of health business (as in Bangladesh, Hungary, Jordan, 28. De Sa (1996) summarizes the current state of quality mea- Philippines, Thailand, Turkey, and Uganda). South Africa licenses surement and prospects for quality competition among managed new hospitals, but responsibility for enforcement is at both the care plans. federal and province levels, allowing enforcement to fall between 29. No state regulates underwriting or pricing in the large- the cracks. The same is true in India. Regulations on space allo- group market (typically defined as groups of 50 to 100 or more). cations are often enforced in Brazil and the Czech Republic, but In general, this market functions very differently than small- have little or no impact on quality, effectiveness, or efficiency of group and individual markets. Specifically, insurers "experience health service delivery in either the public or private sector (TAI rate" large-group clients; experience rating allows them to assume 1996; Bhat 1996). little or no risk and to negotiate a fee to cover administrative cost and profit. References 30. For example, smaller insurers may offer insurance agents incentives (such as salary or fee bonuses) to screen informaUy for Abel-Smith, Brian. 1995. 'Assessing the Experience of Health poor health status and to send applicants with health problems to Financing in the United Kngdom." In David W Dunlop and another insurer. Even if such practices are illegal, they are extremely Jo. M. Martins, eds., An International Assessment of Health difficult for regulators to detect and document. Care Financing: Lessons for Developing Countries. Washington, 31. Reinsurance is a set of products sold to insurers to help them D.C.: World Bank Economic Development Institute. finance unusually high risk. It may be written as aggregate rein- Akula John L. 1997. "Insolvency Risk in Health Carriers: sua , (pyn all clim afe th inue exed a thehl of kl,Jh .197 Islec ls elhCres surance (paying all dlains after the insurer exceeds a threshold Of Innovation, Competition and Public Protection. " Health Affairs total liability) or as specific reinsurance (paying all claims per enrollee 16(1): 9-33. above a threshold amount). Reinsurance companies pool risk across Altman, Stuart, andTerrijackson. 1991. "Health Care in Australa: insurers, but all of the basic principles of insurance apply in rein- Lessons from Down Under." Health Affairs 10(3): 129-46. surance markets. Specifically, if reinsurers believe that all or many American Academy of Actuaries. 1993. 'An Analysis of Mandated insurers in a market are likely to exceed the reinsurance threshold, Community Rating." Monograph 3. Washington, D.C. INNOVATIONS IN HEALTH CARE FINANCING 112 Behlil, Ali. 1995. "Private Health Sector Experience in Turkey" Department of Health and Human Services and U.S. Agency Working paper. Halk Yasam, Sigorta, Istanbul. for International Development, Washington, D.C. Bennett, Sara. 1992. "Promoting the Private Sector: A Review of Hellinger, Fred J. 1995. "Selection Bias in HMOs and PPOs: A Developing CountryTrends." Health Policy and Planning 7(2): Review of the Evidence." Inquiry 32(2): 135-42. 97-110. Hurst, Jeremy. 1991. "Reforming Health Care in Seven European Bennett, Sara, George Dakpallah, Paul Garner, Lucy Gilson, Nations." Health Affairs 10(3): 7-21. Sanguan Nittayaramphong, Beatriz Zurita, and Anthony Zwi. Jensen, Gail A. 1993. "Regulating the Content of Health Plans.' 1994. "Carrot and Stick: State Mechanisms toInfluence Private In Robert B. Helms, ed., American Healtb Policy: Critical Provider Behaviour." Health Policy and Planning 9(1): 1-13. Issuesfor Reform. Washington, D.C.: American Enterprise Bhat, Ramesh. 1996. "Regulation of the Private Health Sector in Institute. India."InternationalJournalofHealth PlanningandManagement Jensen, Gail A., and Michael A. Morrisey 1990. "Group Health 11: 253-74. Insurance: A Hedonic Price Approach." Review of Economics California Consortium. 1996. "Options for Health Reform in and Statistics 72(1): 38-44. Ecuador." Paper prepared for the Ministry of Health and the Jensen, Gail A., Kevin D. Cotter, and Michael A. Morrisey. 1995. National Health Council of Ecuador. San Francisco. "State Insurance Regulation and Employers' Decisions to Self- Chollet, Deborah J. 1995. "Redefining Private Insurance in a Insure." Journal of Risk and Insurance 62(2). Changing Market Structure." In Stuart H. Altman and Uwe La Forgia, Gerard M., and Bernard Couttolenc. 1993. "Guatemala E. Reinhardt, eds., Strategic Choicesfor a Changing Health Care Health Sector Reform Lending Program: A Review of the System. Chicago, Ill.: Health Administration Press. Epiodemiologic Context, Government Resource AUlocation 1996. "The Changing Rate of Employer-Based Coverage." and Efficiency, and the Private Sector." Report presented to Working paper. Alpha Center, Washington, D.C. the Inter-American Development Bank. Urban Institute, Chollet, DeborahJ., and Rebecca R. Paul. 1993. Community Rating: Washington, D.C. Issues and Experience. Washington, D.C.: Alpha Center. La Forgia, Gerard, and Charles Griffin. 1993. Health Insurance in Collins, David H., Jonathan D. Quick, Stephen N. Musau, and Practice: Fifteen Case Studies. Washington, D.C.: Urban Institute. DanielL. Kraushaar. 1996. "Health Financing Reform inKenya: Levit, Katharine R., Helen C. Lazenby, Lekha Sivarajan, Madie The Fall and Rise of Cost-Sharing, 1989-94." Stubbs W Stewart, and others. 1996. "Data View: National Health Monograph Series 1. Management Sciences for Health, Boston, Expenditures." Health Care Financing Reviewv 17(3): 205-42. Mass. Lewis, Maureen A. 1988. Financing Health Care inJamaica. Report Day, Patricia, and Rudolf Klein. 1991. "Britain's Health Care 3714-04. Washington, D.C.: Urban Institute. Experiment." Health Affairs 10(3): 39-59. Lewis, MaureenA., andAndre CesarMedici. 1995. "PrivatePayers DeSa, Jeanne. 1996. The Market for Accountability: Measuring of Health CareinBrazil: Characteristics, Costs, and Coverage." and Managing Plan Performance. Washington, D.C.: Alpha Health Policy and Planning 10(4): 362-75. Center. Ligon, James A., and Paul D. Thistle. 1996. "Information Dunlop, David W, andJo. M. Martins. 1995. "Lessons Learned." Asymmetries and Informational Incentives in Monopolistic In David W Dunlop and Jo. M. Martins, eds., An International Insurance Markets." The Journal of Risk and Insurance 63 (3): Assessment of Health Care Financing: Lessons for Developing 434-60. Countries. Washington, D.C.: World Bank Economic Lohr, Katherine N., Robert H. Brooks, and others. 1989. "Use of Development Institute. Medical Care in the Rand Health Insurance Experiment: Enthoven,Alain C. 1991. "Intermal Market Reforms of the British Diagnosis- and Service-Specific Analyses in a Randomized National Health Service." Health Affairs 10(3): 60-70. Controlled Trial." Medical Care 24(9): S1-87. Fernandez, Jesus M. .1997. "Revisi6nde Estudios de Financiocion Londofio,Juan Luis. 1996. "Managed Competition in the Tropics?" del Sistema de Salud Del Perui. Implicaciones Para el Proceso Paper presented to the International Health Economics de Reforma del Sector." Consultant report. Overseas Association Inaugural Conference, May, Vancouver, British Development Administration, London. Columbia. FKMIS (Federation of Korean MedicalInsurance Societies). 1986. . 1997. Personal communication. Inter-American "Outline of Medical Insurance in Korea." Seoul. Development Bank, Washington, D.C. Fuenzalida-Puelma, Hernan L. 1996. "Regulation of Health McIsaac, DonaldA., and David F Babbel. 1995. "TheWorld Bank Insurance: Conceptual Framework, Case Studies on Hungary Primer on Reinsurance Policy." PolicyResearchWorking Paper and Turkey, Notes on Croatia and Lithuania." World Bank, 1512. World Bank, Financial Sector Development Department, Human Development Department, Washington, D.C. Washington, D.C. Gross, Alfred W, and Carlos Zarabozo. 1994. "Reports to USAID Munn, Glenn G., F.L. Garcia, and Charles J. Woelfel. 1991. The on January 20-28, 1994 visits to the Czech Republic." U.S. St. James Encyclopedia of Banking and Finance. Chicago, Ill. PRIVATE INSURANCE: PRINCIPLES AND PRACTICE 113 Musgrove, Philip. 1996. Public and Private Roles in Health: Theory U.S. House of Representatives, Committee of Energy and and Financing Patterns. World Bank Discussion Paper 339. Commerce, Subcommittee of Oversight and Investigations. Washington, D.C. 1990. Failed Promises. Washington, D.C.: U.S. Government Nittayaramphong, Sanguan, and Viroj Tangcharoensathien. 1994. Printing Office. "Thailand: Private Health Care Out of Control?" Health Policy . 1994. Wishful Thinnking: A World View of Insurance and Planning 9(1): 31-40. Solvency Regulation. Washington, D.C.: U.S. Government Oyarzo, Cesar. 1997. Personal communication. Director of the Printing Office. faculty of economic sciences and administration, University Welch, W Pete. 1985. "Regression Toward the Mean in Medical of Chile. Care Costs: Implications for Biased Selection in Health Pauly, Mark V 1986. "Taxation, Health Insurance, and Market Maintenance Organizations." Medical Care 23: 1234-41. Failure in the Medical Economy."JournalofEconomicLiterature Wllliam M. Mercer Ltd. 1996. InternationalBenefit Guidelines 1996. 24: 629-75. 19th ed. Brussels. Price, Max. 1994. 'The Impact of Political Transformationin South World Bank. 1992. "Indonesia: Health Insurance Issues in the Africa on Public/Private Mix Policy Debates." Health Policy 1990s." Green Cover Report 9999-IND. Washington, D.C. and Planning 9(1): 50-62. . 1994. "The Organization, Delivery, and Financing of Reinhardt, Uwe E. 1995. "Germany's Health Care and Health Health Care in Brazil: Agenda for the 90s." Report 12655- Insurance." In David W Dunlop and Jo. M. Martins, eds., An BR. Latin America and the Caribbean Region, Human International Assessment of Health Care Financing: Lessons for Resources Division, Washington, D.C. Developing Countries. Washington, D.C.: World Bank Economic . 1995. "India: Policy and Finance Strategies for Development Institute. Strengthening Primary Health Care Services." South Asia Rothschild, Michael, and Joseph E. Stiglitz. 1976. "Equilibrium Region, Country Department II, Population and Human in Competitive Insurance Markets: An Essay on the Economics Resources Division, Washington, D.C. of Imperfect Information." Quarterly Journal of Economics 90: . 1996. "Hashemite Kingdom of Jordan: Health Sector 626-49. Study" Report 15418-JO. Middle East and North Africa Region, Schneider, Markus, Juerg Sommer, and Aynur Keceki. 1987. Country Department II, Human Resources Division, Gesundheitssysteme im Internationalen Vergleich. Augsburg: Washington, D.C. BASYS GmbH. . Forthcoming. "Honduras National Health Accounts." Shaw, R. Paul, and Charles C. Griffin. 1995. Financing Health Care Preliminary report. Washington, D.C. in Sub-SaharanAfica through UserFeesandInsurance. A Directions World Bank and IDB (Inter-American Development Bank). 1997. in Development book. Washington, D.C.: World Bank. "Argentina: Facing the Challenge of Health Insurance Reform." Skipper, Harold D. Jr. 1992. "Insurer Solvency Regulation in the Gray cover report produced jointly by the World Bank's Latin United States." Research Report 92-4. Georgia State University, America and the Caribbean Region, Human and Social Center for Risk Management and Insurance Research, Atlanta, Development Group, and the Inter-Arnerican Development GA. Bank's Regional Department I, Social Programs Division, TAI (Taylor Associates International). 1997. 'Country Profiles in Washington, D.C. Private Hospital Investment Opportunities: A Global Study Yang, Bong-Min. 1996. "The Role of Health Insurance in the Conducted on Behalf of the International Finance Corporation." Growth of the Private Health Sector in Korea." International Annapolis, Maryland. Journal of Health Planning and Management 11: 231-52. INNOVATIONS IN HEALTH CARE FINANCING 114 Private Health Insurance in Egypt Nadwa Rafeh T his paper reviews health care financing in Egypt, That same year, all private voluntary associations were particularly the role of private insurance. It begins brought under the regulation of the Ministry of Social Affairs, by providing an overview of Egypt's health care sys- and those providing health services were required to regis- tem, reviewing the political and socioeconomic environ- ter with the Ministry of Health. Under Nasser the govern- ment, current health indicators, the epidemiological ment owned, operated, and financed health care facilities. transition of disease over time, resource distribution, and In addition, two hospital networks nationalized by the Nasser equity in service delivery. regime were formed: the Health Insurance Organization The paper then describes the various sources of health and the Curative Care Organization. Until then Health care financing-public and private-that are in place, includ- Insurance Organization hospitals had been private hospi- ing government financing, public financing and social insur- tals run by foreigners and Curative Care Organization hos- ance schemes, and private financing. Attention then focuses pitals had been run by the Charitable Association of Modem on the current status of the private health insurance sector, Women (Kemprecos 1993). analyzing its strengths and weaknesses from the perspec- In 1973 Egypt entered a new phase of economic and tive of society, consumers, and insurers. political development under President Sadat, who initi- Finally, the paper concludes by reviewing the factors that ated policies aimed at increasing economic growth by encour- will influence the development of the private health insur- aging foreign investment and private sector development. ance market in Egypt and describing the public sector's role These policies, continued today under President Mubarak, in providing group coverage. have had a major impact on the provision and financing of health care services. In the past twenty years the number Egypt's Health Care System of private medical facilities has increased dramatically. Today a significant portion of health care is delivered through pri- Egypt's Ministry of Health was founded in 1936. The min- vate hospitals and clinics. istry quickly became responsible for awide variety of health Since 1986, under the guidance of the Intemational services and began developing new services. The first new Monetary Fund and the World Bank, Egypt has been imple- service was a health insurance program for schoolchildren menting significant economic reform. However, reforms aimed in Cairo (Kemprecos 1993). During this period significant at ending public sector dominance of the economy are pro- portions of health care services in Egypt were provided by ceeding slowly. There is little confidence in govemrent pri- private voluntary associations (gami'yat). vatization efforts, and legislation to support and expand In 1962 President Nasser's National Charter declared these efforts has not been well developed or implemented. medical care, education, employment, minimum wages, and The government continues to play a major role in pub- health insurance benefits to be basic rights for all citizens. lic health through a wide range of programs and special- Nadwa Rafeh is a health services management and policy consultant at the World Bank. 115 ized centers. The Ministry of Health's strategy emphasizes circulatory, respiratory, and infectious and parasitic diseases. prevention, primary health care, drug manufacturing, free By 1990 circulatory diseases accounted for 42 percent of care for the indigent, and environmental protection. all deaths among men and 44 percent among women, and Successful attempts have also been made to improve health the proportion of respiratory and infectious and parasitic indicators through maternal and child health programs, pop- deaths had decreased. ulation and family planning programs, vaccination programs, Although infectious diseases are no longer the leading laboratories and blood banks, and control of infectious cause of morbidity and mortality, there is still a great deal diseases. of work to be done in the prevention of infectious diseases such as hepatitis, trachoma, and schistosomiasis. Basic health and epidemiological profile Epidemiological data suggest that there are areas where modest investment in health care can significantly reduce Historically, infectious diseases were the main cause of sick- infectious disease rates. Population growth will continue ness and premature death in Egypt. Government-sponsored to exert pressure on all aspects of the economy for years to efforts to control infectious disease have been successful, come. In 1990 more than 39 percent of the population in however. The prevalence of schistosomiasis, a major health Egypt was under the age of 15 (World Bank 1992). With problem, dropped from 36 percent in 1981 to 10 percent the successful control of infectious disease, the population in 1991 (MOH-IDC 1993). The incidence of neonatal is living longer and chronic diseases are becoming the main tetanus also has dropped, from 20.7 percent in 1986 to 93 contributors to morbidity and mortality. percent in 1990, reflecting significant government efforts through the Child Survival Project. Hospital and provider profile In addition, since 1980 the government has implemented a series of family planning and child and maternal health During the 1980s the government began allowing the estab- care programs that have helped improve health indicators. lishment of private hospitals, leading to a significant increase The crude birth rate fell from 37 per 1,000 people in 1981 in the number of beds in the private sector. Between 1975 to 28 per 1,000 in 1993. Infant mortality fell from 70 deaths and 1990 the number of hospital beds in Egypt rose by 60 per 1,0001 ive births in 1981 to 38 per 1,000 in 1990 (MOH- percent, to more than 110,000 beds. During the same period IDC 1994). Vaccination coverage is more than 80 percent the number of private beds increased by 180 percent, to for every antigen, and the percentage of children fully vac- about 11,000 (Boutros 1992). Almost half the private hos- cinated has risen to 75 percent, from 58 percent in 1987. pital beds are in Cairo. Coverage for tetanus toxoid jumped from 12 percent in Hospital occupancy rates are generally low, with a national 1987 to 63 percent in 1992. occupancy rate of 49 percent. Although there are no accu- MOH-IDC (1994) indicates that 28 percent of hospi- rate data on occupancy rates in private hospitals, evidence talized men were admitted as a result of accidents, poison, suggests that occupancy rates range between 60 and 70 per- or violence. An additional 21 percent had diseases of the cent, and many of these hospitals are struggling to main- digestive tract, 14 percent had respiratory diseases, 7 per- tain profitability (Kemprecos 1993). Public hospitals have cent had circulatory diseases, and 6 percent had genitouri- occupancy rates as low as 40 percent. nary tract diseases. Women were mainly admitted for There are many private clinics throughout the country, childbearing, with obstetrical-related hospitalizations particularly in rural areas. Many of these clinics provide accounting for 35 percent of female admissions. Other limnited inpatient services for recovery after minor proce- causes of female hospital admission induded digestive dis- dures. Many clinics are attached to mosques, churches, eases, accidents, and respiratory and genitourinary diseases. and charitable organizations and provide a wide range of Mortality data also show that an epiderniological transi- outpatient services. tion has been taking place over the past decade. In 1982 There are 19.6 physicians, 2.5 dentists, 5.6 pharmacists, the distribution of deaths was fairly evenly distributed among and 19.6 nurses per 10,000 people in Egypt. About half INNOVATIONS IN HEALTH CARE FINANCING 116 the physicians are employed by the Ministry of Health. hold (out-of-pocket) expenditures on health care, which The government policy guaranteeing a job for each physi- account for 55 percent of spending. Another 33 percent is cian upon graduation has led to overstaffing of physicians financed by government ministries, 9 percent by public within the ministry. financing (mainly through the Health Insurance Organization), and 3 percent by private firms, insurance Public Sources of Financing companies and unions, and professional organizations. Out- for Health Services of-pocket expenditures mainly cover outpatient care. The average household spends LE 380.5 (about $113) In 1991 Egypt spent about 4.7 percent of GDP on health a year on outpatient care, compared with LE 35.4 (about care. Although the health care system is predominantly pub- $9) on inpatient care (table 2). Drugs account for 53 per- lic, several different government, nonprofit, and private cent of outpatient expenditures. Thus, of the average LE organizations provide and finance health care. There are 410 spent on health care each year, 92 percent is spent on four main financing mechanisms: outpatient care and 8 percent on inpatient care. Per capita * Governmentftnancing-direct payments made by the gov- expenditures in urban areas (LE 106) are almost twice as emnment for health care. The Ministry of Health is the high as those in rural areas (LE 59). main government agency funding health care. Other min- istries that own and operate health facilities include the Ministry of Health Ministries of Education, Defense, Interior, Transportation, and Social Affairs. The Ministry of Education plays an The government guarantees all citizens the right to free important role in financing medical education and uni- health care through a network of 225 hospitals and 2,000 versity hospitals, thus funding a significant portion of clinics operated by the Ministry of Health. The free health tertiary care. care policy serves as a safety net for a large segment of the * Publicfinancing-including social insurance (such as the population, mainly low-income groups. The government Health Insurance Organization) that provides care to also provides free medical and nursing education and, selected groups as well as the Teaching Hospital through its employment policy, guarantees jobs in Ministry Organization, Curative Care Organization, and other of Health facilities to all graduating physicians. public firms. These organizations have several sources of funding, including revenues, premiums paid by TABLE I enrollees and government contributions. The Healt Distribution of health care expenditures in Egypt enroJIees, and government contributions. The Health bysueofiacn,19 Insurance Organization is the largest source of public financing, providing care to public and private sector Millions of employees. Institution Egyptian pounds Share (%) * Private fnancing-including privately owned organiza- Government Ministry of Finance 182 4 tions, private insurance companies, unions, cooperatives Ministry of Health 782 19 and professional organizations, and nonprofit non- Ministry of Education 270 7 governmental organizations (NGOs). NGOs are one the Oter ministries 107 3 fastest-growing sectors in Egypt. Funding for these orga- Public 370 9 nizations is provided by national and intemational donors, Prvate mosques and churches, and individuals. NGOs are con- FIm 70 2 sidered more cost-effective than public providers and Private insurance/unions 30 1 provide higher-quality services for the charges. Total 100 3 * Household payments through direct payments. Household poyments 2,263 55 The distribution of expenditures on health care is shown Total 4,115 100 in table 1. Most health care financing is through direct house- Source: World Bank data. PRIVATE HEALTH INSURANCE IN EGYPT 117 In principle, ministry facilities provide comprehensive About 60 percent of the ministry's budget is allocated coverage, including emergency care. In practice, the care to ministry operations, 26 percent to the Health Insurance provided in ministry-run hospitals is limited in volume and Organization, and 8 percent to the Curative Care quality due to budget constraints. The ministry also subsi- Organization. Ministry support for Curative Care dizes costly tertiary care to indigent patients, including open- Organization operations is in the form of loans and inter- heart surgery, renal dialysis, and treatment of malignancy. est payments. Additional support is provided for operating The ministry is the main provider and financier of health and capital expenses. The ministry does not support Curative care. Almost two-thirds (62 percent) of the hospitals in Egypt Care Organization salaries. are run by the ministry. In urban areas, where most private, About 42 percent of the ministry's budget is allocated university, teaching, and Curative Care Organization hos- to curative care, of which about half is allocated to hospi- pitals are concentrated, ministry facilities account for 25 tals. Curative care is defined as treatment of acutely ill percent of hospitals. In rural areas such as Upper Egypt and patients, including pregnancy and childbirth, on an outpa- the Sinai, ministry facilities account for 83 percent of tient or inpatient basis. Most of the curative care budget is hospitals. allocated to urban areas such as Cairo and Alexandria. Through the various ministries, the government accounts Primary care receives about 37 percent of the budget, and for 33 percent of the country's annual health care expen- preventive care only 8 percent. Almost three-quarters (71 ditures (see table 1). The largest portion is spent by the percent) of the primary care budget is allocated to rural Ministry of Health (19 percent of the total). In 1993 the areas. Most of this money is spent on rural health center ministry's budget was about LE 1 billion (MOH-ILDC 1993). operations and construction or renovation of rural health The ministry's budget includes government allocations, rev- care units. enues generated by ministry facilities, and grants from donor The government's "free health care for all" policy has agencies. The budget has shown a steady nominal increase significant implications for the delivery of health care through for each of the past five years. These increases have not kept government-owned facilities. Budget limitations, a rapidly pace with inflation, however, so in constant terrns the min- growing population, the inability to charge fees for ser- istry's budget has been decreasing. The budget has also been vices, and the policy guaranteeing jobs for all graduated decreasing as a share of government spending; during this physicians are among the biggest constraints facing the period the ministry's budget was about 1.8 percent of the Ministry of Health. Moreover, the ministry cannot provide overall government budget. Nearly two-thirds-65 per- the comprehensive health care coverage available through cent-of the ministry's budget is used to pay salaries, 21 private insurers and private providers. The ministry also can- percent is used for operating costs, and 14 percent for cap- not compete with the private sector in terms of quality of ital costs. care and patient satisfaction. And because the ministry is poorly staffed and poorly funded, it is the provider of last TABLE 2 resort for people who cannot afford to purchase care from Annual household expenditures for inpatient and other sources. outpatient care in Egypt Inpatient Outpatient Health Insurance Organization Egyptian Share of Egyptian Share of pounds total (%) pounds total (%) The Health Insurance Organization (HIO) is the largest Doctor fees 20.0 32 76.2 20 insurance organization in Egypt providing health insur- Drugs 7.2 28 101.2 27 Other 8.2 40 203.1 53 ance to a defined beneficiary population. HIO is a manda- Average household tory social insurance program; participation is required by expenditure 35.4 100 380.5 100 all company employees. Article 32 of Egyptian law requires Annual number of visits 0.034 4.62 the participation of government workers and Article 79 Source: Adapted from Berman 1995. requires the participation of private workers in MHO's health INNOVATIONS IN HEALTH CARE FINANCING 118 insurance program. 1HO covers eligible employees, widows By 1993, 561 companies had received waivers to opt out of deceased beneficiaries, and pensioners. It does not cover of HIO. The characteristics of these companies and their spouses, children, or other family members of employees. reasons.for not participating in HIO have not been ana- In 1993, however, the program was extended to cover about lyzed. Anecdotal evidence suggests that companies request- 10 million schoolchildren. Today the program covers more ing waivers do so because of widespread dissatisfaction with than 15 million beneficiaries, almost a fourth of Egypt's the quality of health care provided under HIO. HIO is population. also criticized for making it too easy for employees to take HIO operates twenty-five hospitals containing about sick leave. 4,500 beds as well as 116 outpatient clinics. Overall bed Most companies opting out of HIO become self-insured, occupancy in 11O hospitals is about 69 percent, and the providing health care coverage through contracts with the average length of stay is 5.9 days. FIO staffs small aid sta- Curative Care Organization, private hospitals, and health tions with one doctor and one nurse at work sites with care providers. Other companies purchase group health care more than 3,000 employees. coverage through private insurance companies. Most employ- Revenuefor HO is provided by employee and employer ers that opt out of HIO are in the private sector and are contributions and government subsidies. HIO is predom- financially sound. Similarly, HIO members who can afford inantly an employment-based insurance program, with higher payments often choose to pay out of pocket for ser- employers and employees paying a portion of salary as pre- vices or to buy alternative private health insurance. mniums. Under Article 32 government employees pay 0.5 Individuals choosing not to use HIO services are more likely percent of their base salary and their agency or ministry to be eligible for health insurance provided through pro- employer pays 1.5 percent to -H1O. In addition, governmnent fessional organizations or cooperative private voluntary employees must make small copayments. The copayments organizations. are quite small and do not discourage inappropriate or excess RIO provides a comprehensive package of health care use of RIO services. benefits. Currently, however, there is a substantial gap Private sector workers covered under Article 79 must between the premiums paid to HIO and the costs incurred pay 1.0 percent of their base salary to RIO and their employer by the program. In 1993 1IO experienced a net operating must contribute 3.0 percent. Because of the higher pre- loss of LE 14.9 million after accounting for all premiums mium, no copaymnents are required for private sector work- paid and all government subsidies. There are several rea- ers. Pensioners and widows are required to pay 1.0 percent sons EIO has been unable to operate profitably. of their basic pension as a health insurance premium. First, health care coverage through HIO is provided to Pensioners and widows are the fastest-growing segment of all eligible beneficiaries regardless of preexisting conditions IO beneficiaries, increasing by 15.5 percent between 1991 or other high-risk characteristics. Denying coverage based and 1992. Other beneficiary groups increased by 4-6 per- on health status and requiring higher premiums for high- cent during the same period. Pensioners and widows are risk groups is contrary to the government's goals for the also the most frequent users of health care services, posing HIO social insurance program: HIO's inability to apply an ever increasing financial burden on the 11O program. standard underwriting practices makes it extremely diffi- Since 1995 the Student Medical Insurance Programrhas cult for the program to meet its financial objectives. been another source of funding for HIO. Small annual Second, RIO faces significant problems regarding inap- subscriptions are paid by students at every level of educa- propriate and excess use of services. For example, HIO pro- tion as part of their tuition. vides a generous drug benefit over which it exercises little In 1984 the government permitted employers to request or no control. As a result many beneficiaries use their HIO waivers from RIO participation if their employer provides benefits to obtain medications at minimal or no cost. similar health insurance coverage to all employees. However, The government is likely to expand health care cover- this law requires employers to continue to pay HIO a pre- age through HIO. The recent expansion of HIO coverage mium equal to 1.0 percent of each employee's basic salary. to students is an example of the government's policy objec- PRIvATE HEALTH INSURANCE IN EGYPT 119 tives. Yet the premiums and subsidies collected by HIO There are three main forms of private insurance cover- are insufficient to pay for the care provided to the increas- age: policies purchased through private health insurance ing beneficiary population. It is unclear whether HIO has companies, group insurance policies purchased through the management capabilities to deal with an increasing ben- unions, professional organizations, and cooperatives, and eficiary population and implement the reforms needed to self-insurance policies where care is provided under con- allow it to play an expanded role in health insurance cov- tract with hospitals and physicians and funded from inter- erage in Egypt. nal resources. About 100,000 people are covered under policies purchased from private insurance companies and Curative Care Organization 160,000 people are covered under union or professional organization policies. These policies may be used to sup- The Curative Care Organization (CCO) is a parastatal over- plement other social health insurance programs or to pro- seen by-the Ministry of Health. The CCO runs twelve hos- vide comprehensive coverage for people who can afford it. pitals containing a total of 4,846 beds. The hospitals are located in six governerates, with the largest in Cairo and Household out-of-pocket expenditures Alexandria. As noted, CCO hospitals were private until they were nationalized in 1964 under the Nasser regime. As noted, household out-of-pocket expenditures are the CCO provides services to four groups of users: employ- largest source of health care spending. In 1991 household ees through contracts with employers, individuals on a fee- out-of-pocket payments accounted for 55 percent of health for-service basis, low-income groups, and accident victims care expenditures, or LE 2.3 billion. In addition, employ- free of charge. In 1993 the CCO signed a contract with HIO ees must contribute to HIO premiums or to private health to provide health care services to students attending voca- insurance plans (for companies that opt out of the HIO tional schools. The CCO is financed through contracts with program). Employees spend an estimated LE 90 million employers opting out of the HIO insurance program, con- on HIO premiums and LE 10 million on other health tracts with FRO, out-of-pocket hospitalization fees, and firom insurance. Ministry of Health grants that cover the free treatment of low-income patients. Almost half of inpatient admissions to Health maintenance organizations the CCO are covered by contracts with companies, and more than a third of patients pay out-of-pocket fees. Several attempts have been made to establish health main- Sources of revenue for CCO hospitals include fees from tenance organizations (HMOs) in Egypt. The Middle East laboratory and inpatient services, premiums and fees from Medicare Plan, established in 1989, was the first nonprofit companies with contracts, a percentage of drug sales, gov- organization operating according to principles of managed ermnent subsidies for free beds, and donations and grants. care. This plan now covers 10,000 subscribers through contracts with thirty-six companies. The plan offers a full Private Sources of Financing for Health range of services, including comprehensive outpatient and Services inpatient services as well as emergency care. The policy does not, however, cover dental care, care associated with preg- Private financing of health care is limited. As noted, private nancy, home visits, regular checkups, or eyeglasses. Marketing funds accounts for about 3 percent of national health care of the Medicare plan has focused on group enrollment, expenditures. The private sector includes privately owned which tends to minimize problems of adverse selection. organizations, private health insurance companies, and Members or their employers pay the annual Medicare pre- NGOs. Privately owned organizations serve as financiers miums on a per capita basis. and providers of health care, while private health insurance The Medicare plan has adopted a number of practices companies finance health care through funds collected from to control costs and remain profitable. The company under- individuals and employers on behalf of their employees. writes coverage and maintains the right to evaluate indi- INNOVATIONS IN HEALTH CARE FINANCING 120 viduals above the age of 45 and to exclude high-risk indi- ers. Beneficiaries are given five vouchers for outpatient care viduals and those with preexisting conditions. To minimize each year. The plan maintains a list of 20,000 providers, 160 overutilization, the company uses affiliated physicians to hospitals, and 300 laboratories accredited for payment by act as gatekeepers and to continuously evaluate medical the health plan. need and appropriateness. Premiums are adjusted each year Members pay a fixed annual premium to participate in based on historical utilization,. the plan. Fees are based on a sliding scale that increases Medicare premiums are lower than those of other pri- with length of membership. This system makes participa- vate insurance companies. This HMO is considered a suc- tion in the program more affordable to young physicians cess, and the number of groups or companies contracting and other union members at the beginning of their careers. with it increases each year. However, most of Medicare's In addition to premiums, plan participants must make copay- contracts are with international companies and with ments each time medical care is received. The copayment embassies. Thus it is unclear whether Medicare's experi- is determined by the type of facility and service used. ence can be expanded to the national and public organiza- Specialty care requires higher copayments. tions that form the bulk of the market. The plan has a good reputation among its beneficiary population. Most beneficiaries use their participation in the Professional organizations, unions, and cooperatives plan to replace services available through -HO. The Medical Union plan provides members with the freedom to choose Health insurance plans provided through professional orga- their providers from the accredited list and receive high- nizations and unions and cooperatives are common in Egypt. quality care at a price they can afford. 'here are about twenty-one professional organizations (med- ical practitioners, engineers, lawyers), most of which make Privote health insurance a health insurance plan available to their members. Health coverage in many of the union plans can be extended to Egypt has a small private health insurance sector account- cover family members and retirees. ing for about 1 percent of annual health care expenditures. Medical Union is the largest and most successful health Insurance companies established prior to the revolution insurance plan. The plan is open to medical association were nationalized in 1961. All others were developed by physicians, dentists, pharmacists, and veterinarians. Like major Egyptian banks. Although insurance companies pro- other professional-based health care plans, the medical asso- vide a wide range of insurance coverage, health plans form ciation provides eligibility to ali its members and their fam- a small part of the insurance market. Thus health insur- ifies. The plan also covers retirees and widows of deceased ance plans are not preferred by insurers. Only three public members. Medical Union has about 160,000 members and organizations and one private company offer health covers a wide range of services, induding hospitalization, insurance. outpatient visits, diagnostic procedures, emergency care, Private health insurance mainly covers people who are and dental care. ineligible for public insurance (such as family members of The plan uses a number of measures to control costs employees and the unemployed) and employers and groups and utilization of services, including requiring copayments, who elect not to use public insurance because of its lower setting moderate reimbursement limits for hospitalization quality. Private health insurance serves mostly private, well- and key procedures, providing information to beneficia- financed companies and high-income individuals who can ries about the cost of services and procedures offered by afford to pay the high cost of premiums. Private health insur- different providers, and requiring preauthorization for ance is also used by the international community residing hospitalization and major outpatient procedures. in Egypt. Preauthorization allows beneficiaries to verify their remain- As in many developing countries, private health insur- ing credit with respect to the reimbursement limit. Utilization ance is growing. Since the 1980s Egypt has reluctantly imple- of outpatient services is controlled through the use of vouch- mented structural health sector reforms accompanied by PRIVATE HEALTH INSURANCE IN EGYPT 121 shrinking central spending on health care. Today Egypt is Security Forces, and employees of specific ministries. passing through a transitional phase, focusing on reform of Other employees received mandatory health care cov- the public health care system to decentralize and provide erage through the HIO social insurance program. greater autonomy to Ministry of Health facilities. The min- From a political perspective, public financing of health istry is exploring different financing mechanisms-such as care services has been important because it has gener- cost recovery for curative care-while maintaining preven- ated a sense of social solidarity and equity and guaran- tive care, primary health care, social health insurance, and teed universal access to care. But these social objectives care for the elderly and the indigent within the domain of cannot be achieved without paying a price. Funding for the public sector. the Ministry of Health has been limited to less than 2 Healh system reforms are also aimed at increasing free- percent of the government budget, and these funds are dom in the private health insurance sector. However, it is insufficient to meet the government's commitments to unclear whether social, legal, and economic factors will pro- health care. The provision of health care services is not vide the necessary incentives for expansion of the private equitable, with far worse access for the poor and the insurance sector. The following sections analyze the fac- unemployed. Low-wage workers must accept poor-qual- tors affecting the development of private health insurance ity care and pay many expenses out of pocket. and the impediments this industry faces. The market is ana- * Economic concerns. Significant portions of the Ministry lyzed from the perspective of the government, consumers, of Health's budget go to fixed costs such as salaries. and insurers. The government policy guaranteeing employment to all medical school graduates results in increasing salary costs Government perspective. Current government policies regardless of the number of beneficiaries of public pro- do not support the growth of private health insurance, nor grams. Thus the government must maintain high levels do they encourage beneficiaries to leave the social insur- of participation to cover these fixed costs. ance program. Indeed, for a number of reasons government The government is also forced to continue requiring com- policies are intended to maintain the largest possible pool panies opting out of the public health insurance program of beneficiaries in public programs: to pay a 1 percent salary tax to support HIO. However, the * Social concerns. In any health care system, basic values companies that opt out of HIO are likely to have employ- underlie policy decisions. To understand the health insur- ees in good health. Thus an increasing number of low-income ance sector, its funding mechanisms, scope of benefits, and less-healthy individuals (such as widows and pension- and extent of coverage, the social objectives of health ers) are participating in HIO and increasing costs. care must be defined. Egypt's constitution guarantees It is difficult to achieve an equitable public health care all citizens access to free health care services. Over the delivery system with a stable share of spending while rely- years the government has pursued this objective using ing predominantly on private insurance. The combination two parallel strategies. The first included direct govern- of finance, regulation, and mandates needed to achieve such ment financing and service provision. The second an outcome may be theoretically feasible but are politically extended coverage using mandatory social insurance. and operationally impossible. The Nasser regime strengthened the right to free care. The government made efforts not only to finance health Consumer perspective. A large portion of health care in care but also to provide free services through a network Egypt is financed out of pocket, which explains why there of health clinics and hospitals owned and operated by is a potential market.for private insurance. However, three the Ministry of Health. The government financed med- consumer-related factors limit the growth of Egypt's pri- ical education and guaranteed all graduating medical vate health insurance market. First, the affordability of pri- doctors and nurses employment in public facilities. The vate health insurance. Most Egyptians cannot afford private government also guaranteed medical coverage to spe- insurance premiums, and most public sector companies lack cific population groups, including the Army, Internal sufficient capital to seek group policies through private insur- INNOVATIONS IN HEALTH CARE FINANCING 122 ance. Thus private insurance is likely to serve only private, (such as HIO) to seek more efficient private insurance plans. high-income employers and wealthy individuals. For most As noted, companies that opt out of HIO are still required of the population private health insurance premiums with- to pay a 1 percent premium to HIO. out public support is still beyond reach. From an economic perspective, health insurance offers Second, the availability of private insurance. For most limited profitability to insurers because premiums are too Egyptians private health insurance is not provided as an low relative to the costs of providing health care. Evidence option for health care coverage. Insurance coverage is not suggests a minimal, if not negative, return on investment universal, and those who are not provided with coverage for most private insurance plans. Insurance premiums are through their employer or through another affiliation are low because companies try to make them affordable or less likely to seek private health insurance. The unemployed, because companies cannot accurately assess risks. Moreover, the self-employed, and workers in small companies are likely health insurance plans create many time-consuming prob- to seek coverage through Ministry of Health facilities or lems and contribute little or nothing to companies' profit through private providers who are paid out of pocket. margins (Kemprecose 1993). Finally, the acceptability of private health insurance. Health insurance management and administration poses Cultural and social factors play an important role in deter- yet another constraint on the insurance industry. The indus- mining the acceptability of private health insurance. Although try has poor information technology and weak underwrit- the public recognizes the protection against risk that insur- ing skills and lacks expertise in processing claims and ance can provide, there is a lack of confidence in private designing policies to meet the needs of Egyptian clients. insurance plans to provide coverage when needed. For pri- vate insurance to become a major source of coverage, the Conclusion design and implementation of insurance plans must be simplified. Health care financing in Egypt and the role of private health insurance must be viewed in perspective to the overall econ- Insurers' perspective. Although the insurance market is omy. For years Egypt's centralized health care system has well developed, private health insurance still lags behind. faced significant problems. The government has been unable For insurers, the current marketplace provides few incen- to provide high-quality and efficient public services to its tives to invest in private health insurance plans. Legal, man- citizens, while private initiatives have been constrained by agerial, and economic constraints hinder the ability of private legal roadblocks, lack of a skilled workforce, poor technol- health insurance to be profitable and to grow. ogy and infrastructure development, and a complex and There are legal constraints to starting a new insurance burdensome regulatory environment. company, domestic or foreign. Government regulations In the past few years Egypt has embarked on substan- require capital reserves to enter the health insurance mar- tial economic and political reform that is likely to affect the ket that are beyond the capacity of many investors. These health care sector in general and the health insurance mar- regulations serve as barriers to market entry and reduce ket in particular. These reforms have encouraged privati- competition. zation, increased international trade, and loosened legal Because private health insurance is not legally mandated constraints on foreign investment. The government recently and there is no obligatory mechanism to insure the self- signaled its commitment to world trade by signing the employed, private insurance is less important and many GeneralAgreement on Tariff and Trade (GATT). It also has Egyptians are uninsured. Furthermore, the law guarantees loosened laws on-private insurance. A law allowing the devel- employees the right to opt out of insurance plans if they opment of new insurance companies has been proposed obtain similar insurance through other providers. This option and is being studied. The government also has lowered cap- diminishes the size of the risk pool and reduces the eco- ital reserve requirements for insurance companies to LE nomic advantage to the insurer. Government laws also dis- 60 million to comply with GATT regulations (Kemprecos courage businesses from opting out of social insurance plans 1993). PRIVATE HEALTH INSURANCE IN EGYPT 123 Economic reform is likely to increase the demand for people. However, the ministry is underfunded and cannot private health insurance as incomes increase, employment hope to achieve this goal. Available services in ministry facil- rises, and the number of national and international com- ities are likely to remain scarce, and the quality of care in panies operating in the country increases. These compa- these facilities will remain poor. Two efforts are needed to nies are expected to be financially strong and better informed improve coverage in ministry facilities: about health insurance plans, benefit packages, and the * The population served by the ministry should be better value received for their money. The larger market will stim- defined and the basic benefits package provided to this ulate competition between health insurance companies and population should be specified. Continued and increased providers and reduce the cost of care. As noted earlier, many government funding should be given to ministry facili- private organizations opt out of HIO and contract with the ties that are able to meet government targets for specific CCO, private hospitals, and private health insurance plans. measures of health status, utilization, and quality of care. This trend is likely to continue, especially if the quality of * Decentralization of ministry facilities is essential to HIO services continues to be poor. improve the efficiency of public services. Through cost Still, private health insurance coverage will continue to recovery efforts, public hospitals today are testing dif- be limited to small segments of the population, mainly ferent financing mechanisms, including contracting with employees of financially strong companies and high-income employers and HIO for group coverage. Although pub- individuals, and will likely remain beyond the reach of lic hospitals lack the expertise to design benefit plans, most Egyptians. Most will continue to rely on public orga- pool risks, and calculate premiums, these attempts are nizations and public financing to meet health care needs. likely to raise awareness, understanding, and demand However, for the government to meet its social objectives for health insurance. in health care, significant reform is needed in public sector Finally, the development of knowledgeable consumers finance, management, and quality of care. As things stand, is imperative for the development of a private health insur- public health care coverage is likely to take two approaches. ance market. Many people do not understand their health First, HIO will likely play an expanding role in health risks, the importance of risk sharing, and prepayment for insurance coverage. The government is considering increas- unforeseen illnesses. Cultural acceptability of the basic ing the number of HIO beneficiaries to cover more seg- values of health insurance also remains weak. Unless soci- ments of the population (such as children under the age of etyperceives the need for insurance coverage, private health five). HIO will also remain the primary health insurance insurance will continue to serve a limited segment of the option forworkers and schoolchildren. However, itis unclear market in Egypt. whether HIO has the capability to manage an increase in the number of beneficiaries. Significant reform is needed References to improve the efficiency of HIO services. HIO should con- sider different service delivery models and strengthen its Badran, Ahmed. 1992. "Expenditure on Health in Egypt." Paper ability to design and regulate benefit plans. HIO has a staff presented to the National Democratic Party. Cairo. -_____ 1993. "The Economics of Health Services in Egypt." model managed care structure that provides a fertile set- Translated from Arabic. Cairo, ting for adopting contemporary managed care approaches. Berman, Peter. 1995. "Egypt: Strategies for Health Sector Change." Through reform, HIO may gain the expertise needed to Harvard University, The Financing of Health Services in Egypt design and implement comprehensive group coverage under Data for Decision Making Project, Cambridge, Mass. a unified social policy. Boutros, Samir G. 1992. "The Health Delivery System in Egypt: Second, the goverrnment wil continue to provide health The Crucial Role of the Private Sector." Paper prepared for Scond,a the grovarnent wicinue tor prOovierhea the Fourth National Conference for Junior Physicians, April coverage to groups that are not eligible for HIO coverage. 13-15, Cairo. Ministry of Health medical facilities must continue to meet CAPMAS (Central Agency for Public Mobilization and Statistics). the social objectives of providing free access to health care 1992. Statistical Yearbook 1992. Cairo. services and serving as a safety net, especially for low-income . 1995. "Egypt Household Health Care Use and INNOVATIONS IN HEALTII CARE FINANCING 124 Expenditure Survey." Cairo. MOH-IDC (Ministry of Health Information and Documentation Donabedian, Avedis. 1976. Benefits in Medical Programs. Center). Various years. "Basic Health Statistical Data." Cairo. Cambridge, Mass.: Harvard University Press. Musgrove, Philip. 1996. Public and Private Roles in Health: Theory Jeffers, James. 1982. "Health Policy Review." U.S. Agency for and Financing Patterns. World Bank Discussion Paper 339. International Development, Health Sector Assessment Phase Washington, D.C. I, Cairo. World Bank. 1992. World Tables. Baltimore, Md.: Johns Hopkins Kemprecos, Louise. 1993. "Health Care Financing in Egypt." University Press. Ministry of Health and U.S. Agency for International . 1993. World Development Report 1993: Investing in Health. Development, Cairo. New York: Oxford University Press. PRIVATE HEALTII INSURANCE IN EGYPT 125 Strategies for Pricing Publicly Provided Health Services Paul J. Gertler and Jeffrey S. Hammer M ost governments spend a lot on health. How access to medical care. While informative, this literature these public expenditures are financed is a cru- does not fully prescribe optimal policy. Optimal policy needs cial element of successful health policies, because to be based on the benefit that the policy would have for it determines the budget available for public activities and society above and beyond what would have happened in has implications for how expenditures are allocated. Public the absence of public intervention. The benefit of a pro- expenditures are financedbyboth public and private sources, posed policy is the extent to which it ameliorates individ- with public subsidies from the general government budget ual and social losses from private market failures. Priorities supplemented by private revenue from user fees. This com- should be based not only on the effectiveness of the policy, bination affects how public subsidies are allocated and deter- but also on the importance governments place on the types mines who gets them. Subsidy allocation decisions also of losses and the individuals who incur the losses. determine the extent to which the poor are cross-subsidized. We examine the implications of current policies and pos- The structure of fees creates financial incentives that affect sibilities for policy reform in the context of competing gov- utilization patterns and health outcomes, and affects how ernment priorities. Governments can intervene to correct well individuals are insured against the risk of large eco- private market failures that cause health outcomes to be nomic losses associated with unexpected illness. lower than they otherwise could be, to cross-subsidize the This paper examines the way governments finance and poor's access to medical care, and to correct insurance and allocate public expenditures on health. Much of the policy medical care market failures. Since governments have bud- debate has focused on the extent to which governments get constraints, they cannot fully subsidize all programs and are able to mobilize private resources to supplement pub- activities. This paper argues that: lic subsidies in financing these expenditures. Proponents of * Public spending on health can improve health outcomes, private resource mobilization argue that individuals are will- promote non-health aspects of well-being (for example, by ing to pay for medical care and that additional financing will insuring that individuals are not at risk for the large unex- allow governments to expand and improve crucial programs pected economic losses associated with random adverse (World Bank 1987; Jimenez 1996). Opponents argue that health events), and redistribute purchasing power to the the poor are unable to pay for medical care and will be worse poor. Optimal subsidy and fee policy will differ depend- off if governments expand private resource mobilization ing on how much weight government places on these (Cornia, Jolly, and Stewart 1987; Gilson 1988). competing objectives. Subsidies need to be reallocated Most of the literature contributing to this debate has toward the poor and toward public health programs. focused on the technical issues of how much money can be However, increasing public subsidies can finance only a mobilized and what impact it has on health outcomes and fraction of the resources needed to expand the health sector. Paul J. Gertler is professor of economics, finance, and public policy at the University of Califomia at Berkeley, where he holds joint appointments in the Haas School of Business and School of Public Health. Jeffrey S. Hammer is principal economist in the Policy Research Department at the World Bank. 127 *Prices for curative services (user fees) have two distinct because it can be centralized and needs to be done only roles. They can raise revenue, freeing public resources periodically, outside the pressure of having to treat an that can be reallocated to public health activities and to illness. Despite their promise, however, prepayment plans limited cofinancing of quality improvements in curative often introduce inefficient medical care cost inflation care. But perhaps more important, they can increase effi- that developing countries may be unable to afford. ciency in the use of public facilities and in the health care The next section discusses the role fees play in the bud- system as a whole. However, these gains must be weighed getary process, considering how fees can stretch the gov- against evidence that higher fees can compromise the emient's budget for various programs and how they can objectives mentioned above. The literature on user fees be used to allocate public subsidies. From this discussion has tended to focus on raising revenue (and its conse- comes a set of conditions that determine how fees affect quences for the poor), but their more important effect budgetary flexibility-most of which concern how price and is likely to be in the allocation of resources. In general, quality affect utilization. The third section reviews the empir- user fees at the point of service can play an important ical literature on utilization. The fourth section uses the role in cofinancing health care, but not as the primary information presented in the previous sections to recom- means of finance. mend optimal fee policy, reflecting government objectives. * Revenue from user fees is sometimes used to finance The final section presents conclusions. improvements in the quality of and access to curative medical care. Individuals are willing to pay at least some Role of Fees in the Budgetary Process of the cost of improving quality and access, especially for dirugs. However, the rich are willing (and able) to pay a Public expenditures on health are financed by revenues lot more than the poor. Thus if governments use the aver- from private sources and allocations from the general gov- age "willingness to pay" amount to finance quality ermnent budget (general tax revenues and donor assistance). improvements, the rich will use more services and the In many countries, especially developing countries, the pub- poor will use less. lic sector collects private revenues through fees charged at * Optimal policies also depend on the behavior of con- the point of service. sumers, private providers, and civil servants. Consumers Much of the literature justifies increasing user fees in and private providers determine the market environment terms of mobilizing resources (or achieving cost recovery) in which policies operate-defining limits to or in some and in terms of creating incentives for more efficient use cases additional opportunities for what can be achieved. of public medical services (World Bank 1987). But an equally Civil servants determine the ability of governments to important role for fees is in determining the allocation of implement policies. Policies that are optimal in one con- subsides from the general government budget across ser- text should not be generalized to all. Countries differ vices (hospitalization, primary care, vector control) and types significantly in the relative weights they place on policy of individuals (the poor, the elderly, children). The alloca- goals, and in the constraints they face in their resources tion of subsidies is one of the main policy instruments gov- and in the reactions of markets. Serious policy forma- ernments have to correct health care market failures and tion requires considerably more analysis relative to ide- improve welfare. ology than has characterized debates on the topic. This section describes the role of user fees in determin- * Social insurance plans, which enable governments to ing the government's budget constraint. The structure of mobilize private resources for health by collecting pre- fees determines not only the amount of resources available payments and charging for the health services provided and the amount spent on each program, but also the extent to beneficiaries, hold promise, particularly for middle- to which a particular program's expenditures are publicly and high-income countries. These schemes mobilize pri- subsidized. It is important to remember that this discus- vate resources with no loss in the insurance value of the sion is limited to how fees affect the budget. It says noth- public health care system. Price discrimination is easier ing about which programs should be funded, how much INNOVATIONS IN HEALTII CARE FINANCING 128 should be spent on them, and how much of the expendi- grams is determined by the residual amount left over from ture should be financed by public subsidies. That discus- spending on prevention and treatment services. sion, covered later in the paper, requires information on the The government's budget constraint, which sets expen- benefits of such allocations and the objectives of govern- ditures equal to revenues, can be expressed as: ment intervention. A + Xc1Uj + -cjoUjo + CDC = XfœUi + -fioUio + G The budget constraint i i i i where A is administrative costs, cij is cost of inpatient ser- Most of the resource allocation decisions that public health vice i, Uig is utilization of inpatient service i, cio is cost of care systems must make are related to one another through outpatient service i, Uio is utilization of outpatient service the government's budget constraint. The two main types i, CDC is expenditures on programs that are not utilization of decisions are: What services should be offered, and of driven (vector control, research, sanitation, water treat- what quality? And what should the user fee or copayment ment),fij is user fee charged for inpatient service i,fio is be for each service? user fee charged for outpatient service i, and G is subsidies These decisions are relevant to all levels of government from general tax revenues. where officials have to make finance and resource alloca- Many policymakers promote user fees as a way of mobi- tion decisions. In many countries such decisions are made lizing private resources for public expenditures. It is impor- at high levels of government-either the central or provin- tant to note, however, that charging user fees is not the same cial level. Other countries are devolving resource alloca- as forcing individuals to pay out of pocket. Here we define tion and financing decisions to local levels. The analysis user fees as the price received by a facility or program- below applies to local officials and public facility-level man- not necessarily what individuals pay at the point of service. agers as well as central and provincial officials. The extent Indeed, individuals could contribute to prepayment or insur- to which it applies to the local level depends on the degree ance plans to finance their payment of fees at the time of of autonomy in the system. treatment. Moreover, fees are not necessarily paid to Thbe level of services and the fee structure cannot be set providers in the form of fee-for-service. Prepayment and independent of one another, but rather must be set to sat- insurance plans could just as easily pay providers by capi- isfy the budget constraint-that is, total expenditures must tation. But since insurance affects people's utilization deci- be less than or equal to total revenues. Revenues come from sions and the form of payment affects provider behavior, public subsidies and general tax revenue and from user the source and form of payment need to be taken into charges for services provided. This budget is spent on admin- account when deciding on the structure of fees. istrative costs, inpatient services, outpatient services, and Although the above characterization of the public bud- public health disease prevention and control activities. get constraint is described in the context of centralized Government spending on health covers a wide range of decisionmaking, it can easily be generalized to a less cen- services, from public health activities (such as the preven- tralized structure. The simplest and most efficient bud- tion and treatment of communicable diseases) to curative getary model is one in which the entity that collects fees services that benefit only the individual. Spending on some keeps them and is free to use them as it sees fit. In this programs (hospitalization, primary care, prenatal care) case fee revenues expand available resources and local man- depends on the number of people who demand care. agers, if competent and publicly motivated, can use the Although there may be short-term rationing of these ser- resources to improve welfare. In this case the central gov- vices, in most cases the public sector is obliged to provide erinent must decide how to allocate subsides from the these services to all who request them. The costs of other central budget among national programs and to lower lev- programs, such as mass information campaigns and vector els of government (provinces, districts, states, and so on). control programs, do not depend on the number of users. These lower levels of government then combine these In many cases the amount of funds available for these pro- allocations with subsidies from the local budget and decide STRATEGIES FOR PRICING PUBLICLY PROVIDED HEALTH SERVICES 129 how to allocate the combined resources among facilities to retain and spend the fees that they collect, and dedicate and local programs. Then each facility and program com- few resources to administering and monitoring collection. bines these allocations with fee revenue and makes expen- In addition, these data usually come from national infor- diture allocation decisions among programs and services. mation systems that suffer from serious underreporting. Thus each decisionmaker in the process receives an allo- These data do not imply that well-designed and -admin- cation from a higher level of government-G in the above istered programs cannot mobilize resources-just that many equation-and combines it with local resources to finance of the countries studied lacked the political will to do so. expenditures. By contrast, China has been extremely successful in mobi- lizing resources through fees (World Bank 1996). Even in Resource mobilization 1978, before recent reforms were initiated, subsidies from general tax revenues financed just 28 percent of public health The classic approach to resource mobilization is to raise expenditures. By 1993 public subsidies accounted for even prices (user fees) to generate private revenues that can be less-14 percent of public health expenditures. The rest used to finance a service or finance improvements in the was financed through fees charged to both insured and unin- quality of that service. The practice of charging user fees sured patients. Indeed, cost recovery ratios are also high in for medical services at public facilities has been adopted local initiatives, where the revenue is typically retained and throughout much of the world (Griffin 1987; Nolan and where it is easier to implement and evaluate resource mobi- Turbat 1995; Jimenez 1996). However, fee structures and lization efforts. For example, in 1993 the revenues from fees control over revenues vary greatly across settings. charged to insured and uninsured patients accounted for Much of the evaluation of resource mobilization focuses 91 percent of hospital expenditures and 84 percent of health on cost recovery-that is, the percentage of costs covered center expenditures in China (World Bank 1996). In Senegal by fees. However, it is not clear how to evaluate resource private revenues amounted to 127 percent of recurrent mobilization efforts when cost recovery is limited. In par- expenditures in health centers (UNICEF/BIMU 1995). ticular, the value of private resources needs to be mea- Similar levels have been reported in LatinAmerica and else- sured in terms of freeing up scarce public subsidies that where (Richardson and others 1992; Olave and others 1992; can be reallocated to high-priority programs (such as con- Barnum and Kutzin 1993; Lewis 1993). In addition, McPake, tagious disease prevention and vector control) and in terms Hanson, and Mills (1993) found that a number of African of providing facility managers with enough additional countries used drug fee revenues to obtain tangible improve- resources (such as drugs) to make up the difference between ments in health services. effective and ineffective treatment. The value of private Still, the evidence that some locales are mobilizing sub- resources depends on the extent to which the funds pro- stantial resources does not make up for the administrative vide needed budgetary flexibility at the margin, where small costs of collecting the fees, including the time (opportunity) amounts of money go a long way. costs of administering fee exemption policies. There is lit- Creese and Kutzin (1995) examined national cost recov- tle if any credible data on this important issue. Most assess- ery ratios from fifteen (mostlyAfrican) countries and found ments take place in the context of schemes that have been that eleven have fee revenues that finance less than 5 per- funded through external assistance, which biases downward cent of public expenditures. While such levels are sub- estimates of real-world administrative costs (Creese and stantially less than the 10 to 20 percent potential cited in Kutzin 1995). World Bank (1987), it is hard to evaluate these data out- The extent to which raising fees mobilizes private side the countries' institutional and policy environments. resources depends on the extent to which individuals are A number of questions come to mind. For example, how willing to pay the higher price for services. Patients are not do these cost recovery ratios compare with government willing to pay any amount for curative care. As fees rise, targets? And how much potential revenue is not being col- utilization will fall. The question is, by how much? The lected, and why? Many countries do not allow local units less sensitive demand is to price increases-that is, the INNOVATIONS IN HEALTIH CARE FINANCING 130 more price inelasticl-the more revenue is mobilized There are a number of ways the government might want through price increases. This is because a price increase to reallocate its public subsidies. Consider an increase in has two effects on revenues. It increases revenues by rais- public spending on public health activities such as vector ing the revenue per patient visit, but it lowers revenues by control or sanitation-that is, CDC spending. To increase reducing the number of visits. If the reduction in visits is CDC expenditures, the government must reduce subsidies great enough, price increases actually reduce revenues. to other programs; otherwise it would spend more than its Similarly, the less sensitive is demand, the less prices will available resources and violate the budget constraint. To change service use. do so, it raises the fee for those services-thereby lowering The story is somewhat more complicated with respect the subsidy for beneficiaries of the program and inducing to increasing user fees to finance quality improvements. In some to stop using the service. this case there are two effects on utilization-the negative The amount of subsidies that can be reallocated depends effects of the price increase and the positive effect of the on the amount freed up by the price increase, which depends quality increase. Both need to be measured to assess the on how sensitive utilization is to price. The more price amount of resources that can be mobilized. The less price elastic is demand, the greater is the drop in utilization for elastic and the more quality elastic is the demand, the greater a given price increase. Thus the greater is the amount of are the resources that can be mobilized from a fee increase subsidies that can be reallocated through reductions in both used to finance quality improvements. unit subsidies and volume provided. In essence, the more price elastic is demand, the more easily the government Allocating public subsidies can reallocate subsidies-that is, the greater is its budgetary flexibility. However, the more price elastic is demand, the In addition to mobilizing private resources, fees determine fewer the amount of private resources that can be mobilized. the allocation of public subsidies. This point is extremely Another reallocation priority may be to shift subsidies important because it is through the allocation of public sub- from a lower-priority patient care program to a higher-pri- sidies that government is able to pursue its objectives and ority program. To increase public subsidies to a care pro- correct market failures. Increases in fees free up subsidies gram, the government lowers the fee charged, thereby that can be reallocated to other programs. The more sub- increasing the subsidy rate. The amount of public subsi- sidies a given fee increase frees up, the greater is the gov- dies going to that program increases for two reasons. First, ernment's budgetary flexibility in allocating subsidies. users of the program receive a higher subsidy. Second, the Interestingly, the condition that increases budgetary flexi- lower fee attracts new users who otherwise would not have bility is exactly the opposite of the condition that mobilizes received the subsidy. This discussion implies that reallo- more private resources-namely, the more price elastic is cating public subsides across care programs is a careful demand, the greater is the amount of subsidies that are freed balance of raising and lowering user fees. up. To see this, we rewrite the budget constraint as follows: Revenue retention A + X (c1I - fj1 )Uij + A (c1i - f1I )Ui0 + CDC = G, An important assumption in this discussion is that any rev- enues raised from private sources are kept in the health sec- where (cij -fii) is the public subsidy rate per unit of a ser- tor. If the fee revenue must be returned to the general vice. Then the amount of public subsidies spent on a pub- treasury, then the fee increase effectively does not increase lic program is the subsidy rate times the amount of services resources for health care; the same outcome holds when provided. In this formulation of the budget constraint, local health units are forced to return revenues to central administrative costs plus the sum of subsidies to each of ministries of health. It is as if the government lowered pub- the services and public health program costs cannot exceed lic subsidies by one dollar for every dollar raised privately. total subsidies allocated from the general govemment budget. This approach implies that no resources are mobilized and STRATEGIES FOR PRICING PUBLICLY PROVIDED HEALTH SERVICES 131 provides no incentive for health care providers to collect ities and programs have little or no incentive to collect them. fees, resulting in substantially less revenue being collected All that remains is the threat of audit and punishment, which than could be raised-an important administrative issue in is costly and rarely credible. But if local facilities and pro- implementing a successful user fee strategy. grams are able to keep and use at least some revenues, In most cases the claim on fee revenues collected at the they have an incentive to collect them. local level is likely to come from higher levels of govern- Despite this argument, there has been little systematic ment. In the worst case fee revenues leave the health sec- evaluation of the hypothesis that fee retention increases tor and are returned to central or local treasuries, and health the efficiency of collection. A few case studies, such as Chisvo sector resources do not expand. This is the case in African and Munro (1994) in Zimbabwe, claim that the lack of local countries such as Eritrea, Ethiopia, Namibia, and Zimbabwe retention of fees explains why only a small percentage of (Creese and Kutzin 1995). public health expenditures are financed through fee rev- Amore subtle outcome that is harder to document occurs enues. In a study of Africa, Nolan and Turbat (1995) com- when fee revenues merely displace public subsidies by one pare percentages of public expenditures financed through dollar for every dollar of revenue raised. Indeed, ministries fee revenues and find no correlation with fee retention of health in Cambodia, Chile, China, Iran, Jordan, Nepal, authority. But the comparisons are not clear, and many other and Thailand have cut budgets as fee revenues have increased factors (including different fee policies) could confound the (Leighton 1996). In contrast, health care providers in the observed cross-national relationship. Central African Republic and Kenya refused to restructure There is, however, evidence on the expenditure side that fees until they had received explicit assurances from their local control is important. Parker and Kippenberg (1991) ministries of finance that public subsidies would not be examined fourteen countries'experiences underthe Bamako reduced. Initiative and found that stronger local management of In both the direct and the subtle cases, fee revenues do resources increased the availability of essential medical care not augment the abilities of policymakers and facility man- inputs and improved service utilization. Mwabu, Mwanzia, agers to expand and improve programs. Rather, user fees and Liambila (1995) found that while most revenues from simply become a way to cut public subsidies to the health a 1989 fee increase were locally retained, Kenya's central sector. treasury retained authority over spending. As a result 40 Even when fee revenues remain in the health sector, percent of Kenyan facilities reported that they did not spend many governments do not allow local facilities and programs the revenues. to control the funds. In many countries this is because local facility and program managers have no authority on What Effect Do Prices Have on the Budget how to spend resources. For example, staffing and infra- and on Individual Welfare? structure decisions are usually handled at higher levels. At best local managers can alter the mix of drugs and sup- This section examines the extent to which public health care plies; few are allowed to shift resources between budgetary systems are able to mobilize private resources and maintain items. The reasons for this lack of local control include budgetary flexibility to reallocate public subsidies, and the lack of budgetary skills and concern about potential graft. consequences of these policies on health outcomes and access However, as the current decentralization trend gathers steam, to medical care. The first two issues are essentially a ques- such rationales may diminish. tion of the extent to which raising a fee lowers utilization. Indeed, one reason to speed decentralization is that it The first subsection reviews empirical evidence on how price increases the efficiency with which fees are collected. The increases affect the utilization of the institution that raises fact that a government mandates that fees will be charged the price-that is, own price elasticities of demand. does not mean that facilities and programs will collect the Whether inelastic demand is good or bad depends on fees and remit them to the government. Indeed, if rev- government objectives. When subsidies are reallocated from enues are simply passed on to higher authorities, local facil- one program to another, the more a given fee increase reduces INNOVATIONS IN HEALTIH CARE FINANCING 132 utilization, the greater is the amount of subsidies that can Another important result of a number of these studies be reallocated. When fees are increased to finance a pro- is that price sensitivity differs by economic and demographic gram, the more price inelastic is demand, the greater is the group. The poor appear to be more price sensitive than the amount of resources that can be mobilized. When fees are rich, and children's utilization seems to be price sensitive increased to finance improvements in quality, the less the than adults'. Thus increasing fees may reduce the utiliza- fee increase reduces utilization and the more the quality tion of the poor and children by more than the utilization improvement raises utilization, the greater is the amount of the rich and adults. of resources mobilized. In this case the question is essen- Policymakers should view these resultswith caution, how- tially how much people are willing to pay for quality improve- ever, since the studies suffer from several methodological ments, which is examined in the second subsection. problems. First, the countries studied typically had public The discussions in the next two subsections should be institutions that charged very low fees, with little geographical interpreted only in terms of implications for the budget variation in those fees. In some studies the investigators and in terms of welfare. A decrease in utilization of public used travel costs to measure price elasticities, since time facilities does not necessarily mean that health outcomes costs ration the market when fees are low. These studies or access to all medical care was reduced. For example, if used the estimated models of demand to conduct policy a price increase induces individuals to substitute private simulations that forecast how increases in fees were likely care for public care, access is not reduced. In addition, indi- to affect utilization and revenues. However, the forecasts viduals who decide not to seek care because of a price were based on price changes that were far outside the increase may have only minor illnesses that do not require observed range of the price data. Thus they are highly unre- treatment or that can be treated at home. Own price effects liable. In particular, while an individual's utilization deci- tell us nothing about access and health outcomes, only about sions may be insensitive to prices when prices are low, they resource mobilization and budgetary flexibility in allocat- might be very sensitive to prices when prices are high. ing public subsidies. The effects of price increases on access Several studies have evaluated actual price increases by and health outcomes are analyzed in the third subsection. comparing utilization before and after a fee increase at pub- None of the discussion in this section is devoted to what lic facilities. In the Ashanti-Akim region of Ghana, governments should do. This discussion provides informa- Waddington and Enyimayew (1990) found that after an tion about the budgetary constraints on government actions increase in user fees, long-term utilization fell at clinics serv- and implications of actions on outcomes. To take the dis- ing the poor but did not fall at clinics serving the nonpoor. cussion a step further, we must understand the objectives Mwabu, Mwanzia, and Liambila (1995) found a 52 per- of government policy-the subject of the next section of cent drop in outpatient visits to government health centers the paper. after Kenya introduced fees in 1989. After the fees were suspended in 1990, visits increased by 41 percent-almost Price elasticity of demand for medical care to the original level. Yoder (1989) reports a 32 percent drop in visits to government health care facilities after Many studies have used cross-sectional household surveys Swaziland increased fees. Kahenya and Lake (1994) found to estimate the price elasticity of demand for outpatient ser- that attendance at eleven clinics in Zambia fell by an aver- vices (table 1). Although a few early studies using ques- age of 64 percent after fees were raised, with larger reduc- tionable data found completely inelastic demand-that is, tions occurring at facilities in poorer areas. Finally, Bennett prices had no effect on utilization-most studies have found (1989) found that the utilization of government facilities that higher prices are associated with lower utilization, but dropped by 40 to 51 percent after fees were increased in that overall own price elasticities are low and well below Lesotho, and that the drops were greatest among children. unity. This finding suggests that increases in fees will mobi- As noted, however, these studies should be viewed with lize substantial private revenues, but that flexibility in real- extreme caution. Although the results are consistent with locating subsidies is limited. econometric findings that demand is sensitive to price, STRATEGIES FOR PRICING PUBLICLY PROVIDED HEALTH SERVICES 133 TABLE I Econometric estimates of own price elasticities of the demand for medical care in developing countries Own price elasticity Country Data period/type Service type Overall Low income High income Source Burkina Faso 1985 Public provider Sauerbom, Nougtara, and Latimer All ages -0.79 -1.44 -0.12 1994 Age 0- 1 -3.64 Age 1-14 -1.73 Age 15+ -0.27 C6te d'lvoire 1985 Health clinic -0.61 -0.38 Gertler and van der Gaag (1990) Hospital outpatient -0.47 -0.29 C6te d'lvoire 1985-87 Health clinic -0.37 Dow (1996) Hospital outpaient -0.15 Ghana 1987 Hospital inpatient -1.82 Lavy and Quigley (1993) Hospital outpatient -0.25 Dispensary -0.34 Pharmacy -0.20 Health clinic -0.22 Kenya 1980-81 Govemment provider -0.10 Mwabu, Ainsworth, and Nyamete Mission provider -1.57 (1993) Private provider -1.94 Indonesia 1991-93 Gerder and Molyneaux (1997) Children Health center -1.07 Health subcenter -0.35 Adults Health center -1.04 Health subcenter -0.47 Elderly Health center -0.47 Health subcenter -0.11 Mali 1982 -0.98 Birdsall and others ( 1 983) Pakistan 1986 Alderman and Gertler (1997) Female Traditional healer -0.43 -0.24 Children Public clinic -0.43 -0.23 Pharmacist -0.44 -0.25 Private doctor -0.17 -0.09 Male Traditional healer -0.60 -0.26 Children Public dinic -0.61 -0.27 Pharmacist -0.63 -0.27 Private doctor -0.25 -0.10 Peru 1985 Private doctor -0.44 -0.12 Gertler and van der Gaag (1990) Hospital outpatient -0.67 -0.33 Health clinic -0.76 -0.30 Philippines 1981 Public providers -2.26 -1.28 Ching (1995) Private providers -3.93 -2.23 Philippines 1981 Prenatal care -0.01 Akin and Griffin (1986) Philippines 1983-84 Urban matemity -0.24 Schwartz and others (1988) Rural matemity -0.05 INNOVATIONS IN HEALTH CARE FINANCING 134 they report only the fall in utilization, and say little about The estimated Indonesian price elasticities of demand the degree of price elasticity. Large drops in utilization could are shown in table 1. Although a price increase signifi- be associated with big price increases, and demand could cantly lowers utilization, the effect on children is greater still be relatively price insensitive. Moreover, because these than on adults, and the effect on adults is greater than on studies lacked control groups for which fees did not change, the elderly. Demand for health center care is more price there is no way of knowing how much of the fall in utiliza- elastic than for health subcenters. This finding is not sur- tion was due to the price increase and how much was due prising, since the health subcenters serve rural populations to factors such as changes in quality in the disease that have fewer alternative (public or private) providers than environment. in urban areas, where health centers tend to be located. In Most of the results discussed so far analyze data that fact, the price elasticity of the demand for health center care reflect the rules that governments use to set prices and locate is close to unity, whereas it is well below 1 for subcenters- facilities (Rosenzweig and Wolpin 1986; Pitt, Rosenzweig, suggesting that little revenue will be mobilized by raising and Gibbons 1993; Gertler and Molyneaux 1993; health center fees, but that a lot will be mobilized by rais- Frankenberg 1995). Since government policy is trying to ing subcenter fees. achieve some objective, the variation in fees is unlikely to Sirilarly, Cretin and others (1992 and 1996) report results be random. Rather, it reflects government policy In many from a rural health insurance study in two rural Chinese cases governments set fees and locate facilities based on counties in which copayments, another form of user fee, population characteristics such as economic status and health were experimentally varied to estimate price elasticities of problems. If the multivariate analysis does not explicitly demand. During 1989 and 1990 twenty-six villages in two account for government policies for setting fees and deter- rural counties of Sichuan province, China, participated in mining the locations of facilities, the estimates of the effects an experimental longitudinal study to provide an analytic of the fee on utilization will be confounded with the effects basis for developing sound health care financing mecha- of utilization on government policy2 In addition, most of nisms. The experiment assigned each village two health these studies have only rudimentary controls for quality of insurance plans, one for 1989 and one for 1990. Eight dif- care. Although they usually distinguish between levels (hos- ferent plans were assigned, with outpatient and inpatient pital, health center, and so on) and sector (public or pri- coinsurance rates ranging from 30 to 75 percent. Three of vate) of care, they do not control for qualityvariations (such the plans emphasized the coverage of outpatient care, three as drug availability and provider training) within provider emphasized the coverage of inpatient care, and two offered types. balanced coverage of outpatient and inpatient care. Although Three studies that are not subject to these criticisms ana- participation in the insurance plans was voluntary, each lyze the effect experimentally designed fee increases have household had to enroll as a unit, and participation rates on individual utilization in experimental and control areas. were more than 95 percent. The first, Gertler and Molyneaux (1997), estimated price As expected, higher coinsurance rates for outpatient ser- elasticities of demand for outpatient services in Indonesia vices (used by about two-thirds of the population each year) using longitudinal panel data in which public sector user were associated with significantly lower probability of use fees were varied experimentally in two of Indonesia's twenty- and significantly lower expenditures. For inpatient services seven provinces. The study design was integrated with local (used by just 3 percent of the population each year) higher political decisionmaking authority, which was already plan- coinsurance also led to less utilization and less expenditures, ning to increase user fees. Rather than raising fees every- although the decrease was not as strong. These results sug- where, fee changes were staggered to generate price variation gest that demand for outpatient services is more price elas- based on an explicit experimental design. User fees were tic than demand for inpatient services. Since more serious increased in some districts (treatment areas) but not in illnesses are treated by inpatient hospitalization, these results others (controls) and in both government health centers are consistent with the hypothesis that demand for med- and health subcenters. ical care is less price sensitive for more serious illnesses. This STRATEGIES FOR PRICING PUBLICLY PROVIDED HEALTH SERVICES 135 finding is somewhat reassuring because it suggests that 1995), Nigeria (Akin, Guilky, and Denton 1995), Kenya reductions in utilization resulting from price increases are (Mwabu, Ainsworth, and Nyamete 1993), and the likely to be lower for less serious illnesses. Philippines (Hotchkiss 1993). Manning and others (1987) and Newhouse (1995) report The estimated quality effects are quite large. For exam- the results of a large health insurance experiment in the ple, if thepercentage of Ghanaian public facilities with drugs United States conducted in the late 1970s. More than 20,000 increased from 66 percent to 100, utilization of public individuals in six sites were randomly assigned to one of facilities would increase by nearly 44 percent (Lavy and fourteen health insurance plans that had different copay- Germain 1995). Simultaneous improvements in drugs, infra- ment structures. This experiment differed from the structure, and services would increase the use of public facil- Indonesian and Chinese experiments in that its design was ities by 127 percent. Much of the increase in utilization, based on controlled random assignment of a large number however, comes from substituting public for private care. of individuals. Although the Asian experiments were con- The same quality improvements that increase public uti- trolled, individuals were not randomly assigned. Rather, the lization by 127 percent would only reduce self-care by 14 intervention was at the community level. Randomization percent. Thus the net effect on utilization is quite small. at the individual level provides a better and more robust However, users of public services would have higher-qual- design. ity care than before. Since quality improvements mainly The U.S. experiment was one of the first scientifically affect the choice among providers rather than whether to valid studies that convincingly documented that individu- obtain treatment, the net effect on health outcomes is deter- als' medical care utilization decisions were influenced by mined solely by the increase in quality to current public prices. The results indicated that prices had a bigger influ- sector patients, not by increased access. ence on decisions to initiate treatment that on the amount These studies use estimated parameters from demand obtained once treatment began. Overall, the experiment models to calculate the willingness to pay for quality and found price elasticities of about -0.2, and the price elas- access improvements. Methodologically, this is equivalent to ticity increased with the coinsurance rate. Moreover, demand asking how much a fee would have to increase to offset the for acute care and inpatient services was less sensitive to increase in utilization from the improvement in quality or price than chronic care and outpatient care. This is consis- access.3 Gertler and van der Gaag (1990) found that the geo- tent with the Chinese results, and with the hypothesis that graphic distribution of individuals in Peru and C6te d'Ivoire demand for the treatment of more severe illnesses is less makes them willing to pay (on average) about 20 percent of sensitive to price. the cost of operating a health facility if it cuts travel time to public health centers from two hours to zero. The poor, Willingness to pay for better quality and access however, were wiling to pay substantially less than the rich. In Ghana individuals were willing to pay, on average, half There is evidence that people are willing to pay at least a the cost of improved access, with the poor again wiling to share of the cost of improving access and quality, especially pay substantially less (Lavy and Germain 1994). Thus if the for drugs. (See Alderman and Lavy 1996 for a review of government asked consumers to paythe "average willingness the literature.) For example, studies of cross-sectional house- to pay" for improved access, the nonpoor's utilization would hold data have found that individuals are willing to pay at increase while the poor's utilization would decrease. least some of the cost of improving access to medical care, Although these studies demonstrate a significant statis- as measured by the distance that they have to travel to reach tical correlation between quality and utilization, they raise the closest public facility (Gertler and van der Gaag 1990; several questions about the direction of causality. First, the Lavy and Germain 1995). In addition, four studies analyz- studies used cross-sectional household data to investigate ing cross-sectional data have found that a number of struc- the effects of price, travel time, and quality on utilization tural quality indicators, especially drug availability, and then used the estimated models to simulate the effects significantly affected demand in Ghana (Lavy and Germain of price and quality changes on utilization. Thus they suf- INNOVATIONS IN HEALTII CARE FINANCING 136 fer from the same problem as the cross-sectional demand Effect of fees on access and health outcomes studies discussed earlier: the results confuse the effects of prices and quality on utilization with the effects of utiliza- The preceding subsection focused on the budgetary impli- tion on government geographical pricing and quality policy. cations of different user fee policies in terms of measuring Manylongitudinal studies do not suffer from this method- the extent to which raising fees increases govemment's abil- ological problem. For example, Litvack and Bodart (1993) ity to mobilize resources and allows flexibility in allocating carried out a field experiment in the Adamaoua province of subsidies. Although knowing the own price elasticity of Cameroon that investigated people's willingness to pay for demand for public facilities is essential for forecasting drugs. In one treatment area facilities charged user fees to expected revenues, it is insufficient for evaluating the effect finance a revolving drug fund, which increased drug avail- higher user fees have on individual welfare. As a step toward ability. As a result utilization increased in the treatment area this goal, this subsection reviews the empirical evidence on relative to utilization in a control area. Thus it appears that the effect of prices and quality on health outcomes and consumers in the treatment area were willing to pay for drugs; access to medical care. hence their utilization increased. Using a similar methodol- The first question regards the effect of price increases ogy,Yazbeck and Leighton (1995) investigated the effect of on overall access-that is, did the individuals who chose to introducing fees to finance better-quality prenatal care in stop seeking treatment at public facilities switch to self- Niger. They found that prenatal care enrollments in the treat- treatment or private sector treatment? To measure the effect ment area increased relative to the control area, and that price increases have on access, we examine the effect they the increase was greatest among the poor. However, in an have on the utilization of all providers, public and private. analysis without a control group, Haddad and Fournier (1995) In their research in Indonesia, Gertler and Molyneaux found that user fees led to a drop in utilization in Zaire- (1997) examined the effect public health center fees have despite the fact that the supply of drugs and the physical on total visits, including visits to all public and private condition of the facility were increased at the same time. providers (table 2). They found that price elasticities of total Measuring quality is a problem in all these studies. Most demand were less than the elasticities of health center of the studies used structural measures of quality, such as demand, implying that higher fees caused some individu- availability of drugs, personnel, physical infrastructure, als to switch to other providers rather than treat themselves. and equipment. But it is not what people have that mat- Similarly, Bennett (1989) reports that after fees increased ters, it is what they do with it. Indeed, several studies have in Lesotho, about half the reduction in public sector uti- shown that process measures of quality are better predic- lization was a reallocation to private facilities. tors of health outcomes than structural measures (Peabody, Returning to Indonesia, in urban areas (where there are Gertler, and Liebowitz 1995). Moreover, the most impor- more private alternatives) the total visit elasticity is about tant measure, availability of drugs, confounds supply and half the health center elasticity This finding implies that demand effects. Facilities may have drug shortages because about half of the reduction in utilization switched to other they provide high-quality services, and have high utiliza- providers and about half to self-treatment. In rural areas tion that depletes the drug stocks. the total elasticity is about two-thirds the health center TABLE 2 Own and total price elasticities in Indonesia (percentage change with a I percent increase in public health center fees) Children Adults Seniors Urban Rural Urban Rural Urban Rural Visits to health centers -1.07 -0.63 -1.04 -0.01 -0.45 -0.47 Visits to all providers -0.48 -0.49 -0.70 0.0 -0.22 -0.39 Source: Gertler and Molyneaux 1997. STRATEGIES FOR PRICING PUBLICLY PROVIDED HEALTHI SERVICES 137 elasticity, implying that a much larger percentage of patients Reductions in total utilization that are caused by fee switched to self-treatment than in urban areas. These results increases can have negative health effects. In the Indonesia suggest that public sector fee increases reduce access more user fee experiment, Gertler and Molyneaux (1997) found in rural areas, where there are fewer private alternatives. that the observed reductions in utilization were not only Simulating the effect of fee increases in public facilities on for minor illnesses, but also for medical problems that mea- total utilization also requires information on how competing surably affect health status indicators. The fee increase private providers respond to the increased prices at govern- caused an increase in the duration of illness for all age groups ment facilities. When government providers raise their prices, and in illness symptoms associated with infectious dis- some patients may shift to the private sector, which may lead eases. These results indicate that an important channel private providers to respond to the increased demand by through which prices hurt health is by delaying treatment raising their prices. The extent of the price response depends to the point of reducing the efficacy of medical interven- on the extent of the increase in demand and the slope of the tion. In an extreme case this may cause people to delay seek- private provider supply curve. The larger is the private sector ing help for emergency care. In addition, the fee increases price response, the fewer is the number of people who will impaired older (50 years and above) Indonesians' ability to switch to the private sector, implying that more individuals function physically. The increase in fees had a large enough will choose self-treatment or remain in the public sector. negative effect on health that it reduced labor force par- Private sector price responses are likely to be very impor- ticipation among women (Dow and others 1997). tant. When public user fees were increased experimentally In addition, empirical evidence suggests that increases in Indonesia, Gertler and Molyneaux (1997) found that in access and quality improve health outcomes. In a cross- while the fee increases caused some individuals to substi- sectional analysis of household data, Benefo and Schultz tute self-treatment for care at public facilities, others turned (1994) found that child mortality was lower among famni- to the private sector instead. The resulting increase in demand lies that lived doser to government health facilities in C6te caused private doctors and private nurses and paramedics d'Ivoire and in Ghana. They also found that a doubling of to increase their fees in response to the increased demand drug prices was associated with a 50 percent increase in (table 3). In general, private sector responses were greater child mortality. Thomas, Lavy, and Strauss (forthcoming) in semiurban and rural areas, where there is more direct found in an analysis of cross-sectional data from Ghana that competition between public and private providers. Similarly, improving drug supplies significantly improves the nutri- private nurses and paramedics, who are closer substitutes tional status of children. In an analysis of cross-sectional for public primary care facilities, had larger relative price data fromJamaica, Peabody, Gertler, and Liebowitz (1996) responses than private doctors. The price elasticity estimates found that the birth weight of babies was 500 grams higher shown in table 2 reflect both the increases in public sector in communities that offered good prenatal care services fees and the consequent increases in private sector fees. using process measures of quality. TABLE 3 Private providers' price responses to 100 percent public sector fee increases in Indonesia (percentage change) Urban area Semiurban health Rural health health centers centers and subcenters subcenters Private doctors Health center fees 4.4 18.4 Health subcenter fees - 3.5 20.1 Private nurses and paramedics Health center fees 23.8 9.5 Health subcenter fees - 16.7 57.9 Source: Gertler and Molyneaux 1997. INNOVATIONS IN HEAL-ri CARE FINANCING 138 As with the cross-sectional studies of demand, however, to supplement the public subsidies allocated from general it is hard to say how much the associations between health tax revenues to finance public programs. Here we consider outcomes and quality of care reflect the impact of quality the task of setting user charges and allocating the total bud- and access on health outcomes or the effect of outcomes get (public subsidies plus user fee revenues) while maxi- on government policy toward the geographical allocation of mizing government objectives. This section considers optimal facilities and quality. Frankenberg (1995), in one of the policywhen health outcomes are the main objective of pub- few explicit attempts to determine the direction of causal- lic policy and discusses how the policy should be adjusted ity, used longitudinal data from Indonesia to show that infant when equity and insurance considerations are added. mortality was lower in families located near public health centers. What do prices do? These results suggest that there are real returns to pub- lic programs and public subsidies in terms of health out- One of the main messages of this paper is that the level of comes. Raising fees-thereby lowering subsidies-can have fee charged determines the degree to which a particular pro- negative health consequences. Thus if governments choose gram (or group) is subsidized. Much of the discussion about to raise fees, unless the freed subsidies are reallocated to user fees is couched in terms of whether government should more effective programs, health outcomes may deteriorate. raise fees closer to the cost of providing the service. To Evidence suggests that investing the subsidies in better help guide optimal fee policy, we turn this question around access and quality can improve health outcomes. and askwhen government should subsidize services in order to lower prices below the cost of providing the service. How Should Governments Set Fees? For most commodities, there is a certain "rightness" about the level of use (demand) when consumers face a price While the above analysis suggests that governments may reflecting the true resource cost of producing it. People be able to mobilize private resources to cofinance public ask themselves whether it is worth buying the commodity programs, it does not say how governments should choose given all the other things they can do with their money. If the optimal combination of user fees and allocation of the answer is yes, then they get more value out of the com- public subsidies across programs. Optimal policy needs to modity than it costs society to produce it. If not, then they be based on what is best for furthering social objectives, decide to spend their money on something else that they subject to the limits imposed by medical, behavioral, and think is more worthwhile for the price. Individuals do not economic constraints. Three groups of objectives are often purchase good or services whose prices exceed the value cited: improving health status, improving equity in access they place on it. Similarly, if someone is in a position to to medical care, and irnproving individuals' insurance against provide a good or service, they will do so only if the price the risk of large financial losses due to ill health (Hammer exceeds the cost of their making it. So prices tell both pro- and Berman 1996). In many cases the policy prescriptions ducers and consumers how much something really costs to that best achieve each of these objectives are in conflict. produce and how much people really value it. Thus, because resources are limited, governments must There are several ways prices can help guide resources make tradeoffs in financing programs and base those deci- efficiently in the health sector. First, people often bypass sions on the relative value it places on each objective. lower-level clinics to go to hospitals even when the clinic Regardless of the objectives a government is pursuing could have handled their problem. They do so because through its involvement in the health sector, most countries they will get better care at the hospital, and if both are free have limited public resources to invest in health. In allocat- or have the same price, they have no incentive not to use ing their limited budgets government officials and program the hospital. If prices reflected service costs, hospital prices managers must use the resources wisely and get as close as would be higher than clinic prices, and only those who val- possible to their goals within a fixed budget. In previous ued the hospital service more than its unit cost would bypass sections we considered the possibility of charging user fees the clinic. STRATEGIES FOR PRICING PUBLICLY PROVIDED HEALTH SERVICES 139 A second example is when prices limit the use of ser- One of the main ways in which ministries of health can vices by people who do not think their health problem is improve health status is to encourage or discourage uti- serious enough to be worth the trouble and cost of seeking lization by the way they set the price of health care ser- help. Seeking help takes up real resources in terms of the vices. Ministries may want to stratify their price subsidies time of trained professionals as well as increased conges- to encourage utilization of specific services (immunizations, tion and waiting time for others. If prices are less than the prenatal care) and by specific groups (the poor). But not cost of providing treatment, then treating minor problems all increases in utilization are from new utilization. Some whose value is less than the cost is inefficient. The main may be substituting for private sector services (or other, less point is that prices make people's choices efficient-putting public services) that have been substituted for by the sub- resources where they are most valuable to them. sidized public services. The degree to which the increased What is wrong with this picture? For any of several mar- utilization improves health depends on the efficacy of the ket failure reasons (see below), the amount demanded or additional health care consumed. To determine the amount the value placed on goods may not be "right." The value to of additional care consumed, we have to subtract any reduc- society for seeking care or providing a service may be higher tion in private sector services that the individual would have than that privately judged by the individual. If that is the purchased had there been no subsidy. case, society benefits if more people use the service than However, ministries of health do not have unlimited pub- would based on individuals' private benefits. Thus public lic resources that they can spend on their various activities. subsidies to lower the price are justified. How much of a Ministries have fixed budgets that they can relax only by subsidy depends on the degree to which the social value charging fees for their services. Although price increases exceeds the private value. The benefits that can be obtained may generate substantial revenue, they also deter people from subsidies also depend on how responsive to prices con- from seeking care who might have sought the treatment sumers are. The more price elastic is demand, the greater when it was priced lower. is the social benefit from a given subsidy. To translate this discussion into a set of policy rules, we As with government intervention in any sector, the allo- need to establish the link between policy levers and policy cation of government subsidies needs to be justified in terms objectives. Thus, since a ministry of health's main objec- of the benefit the investment has for society above and tive is in terms of health and its policy levers are in terms beyond what would have happened without public inter- of prices that determine the level of private resources and vention. The way to assess the benefit of a proposed pub- the allocation of public subsidies, we need to establish lic intervention is to identify the failures of private markets links between health and prices. By altering prices, gov- and quantify the loss from these failures. Priorities should ernments affect the utilization of medical care and the be based on the degree to which the subsidy ameliorates amount of money spent on public health activities. Utilization these losses and the importance government places on those of medical care and public health programs influences health losses and the individuals who incur them. Important mar- outcomes. ket failures in the health sector that justify public subsidy With improved health as the objective and the links include public goods, inequity in access to medical care, and between policy and objectives established, we can identify insurance market failure due to asymmetric information. four pricing principles that need to be balanced for the government to get the best health outcomes for its subsi- Improving health status dies (see Hammer forthcoming for formal derivation and details of the pricing rules). Most countries' ministries of health try to improve health * Subsidiesshould be higherforservicesfor which public care by funding public health activities and delivering health care is better than private-that is, for services that yield the services in public health centers and hospitals. This is why best health outcomes compared with people's alterna- policymakers become concerned when fee increases lead tives. If the alternative to public care is a traditional healer to big reductions in utilization. of dubious quality, fees should be raised with great cau- INNOVATIONS IN HEALTH CARE FINANCING 140 tion. If the alternative is a reasonable private sector (in subsidize other activities or groups and provide more ser- Indonesia the private sector consists of public providers vices. Services or groups for which prices discourage large working during the afternoon), raising fees may make numbers of (price elastic) individuals from getting treat- more sense. If health is the objective, it is better to encour- ment should have lower prices. Conversely, when demand age people to use the most productive services at subsi- is more price inelastic, higher prices affect health status dized prices. less and mobilize more revenue that can be used to cross- * Subsidies should be higherforservicesfor which total (pub- subsidize other beneficial activities. The basic idea in set- lic and private) demand is most elastic with respect to fees ting prices is to push public subsidies as far as they can go inpublicfacilities. Governments cannot mandate the use in achieving health gains. Thus price subsidies should be of health care. They can only provide incentives for use. assessed in terms of their effect on health outcomes and Subsidies encourage use of a service by lowering the their impact on the budget, rather than relative to the price. The more price elastic is demand, the larger is the resource costs of service delivery. increase in utilization from a given price subsidy. However, The first and fourth principle point out that interaction demand may be more price elastic for less efficacious between the public and private sectors is crucial in setting services. Thus subsidies should be higher for services prices. If the private sector offers comparable quality ser- that produce the best health outcomes. These services vices and individuals are willing to pay the private sector are most successful in producing the most health because price, government subsidies will not improve health. All they of the combination of efficacy and of the volume of will do is cause individuals to substitute public for private patients generated by the introduction of the subsidy. care. In this case the ministry of health should not provide e Subsidies should be higher for people whose demand is the care, or it should at least price the services so that few more price elastic. For similar reasons as in the previous subsidies are absorbed. This is clearly the case for luxury point, subsidies produce better health outcomes for rooms in hospitals because the rich are the only group that groups for whom the subsidy is likely to encourage use. uses such rooms, and these services are usually available in This finding implies that subsidies should be higher for the private sector. poor individuals whose demand is more price elastic. When the public sector lowers its prices because of An interesting implication of this pricing principle is subsidies and draws patients away from the private sector, that it is optimal to lower prices for the poor even if the it is in essence competing with the private sector. Subsidies government is concerned with neither equity nor wel- to public providers lower the profitability of private fare but with health status per se. providers. Public subsidies affect the prices that the pri- • Subsidies should be higher for services and in areas where vate sector can charge and raise speculation on whether it there are few private sector alternatives (competition). is profitable for private providers to locate in the same Subsidies will produce substantially weaker health out- area as the public provider. The fact that there are no pri- comes if they only cause individuals to substitute out of vate providers in an area does not necessarily indicate that the private sector into the public sector. The best health private providers would not serve the area if there were no outcomes are achieved when subsidies encourage new public services available-it merely indicates that the pri- utilization, so that illnesses that would not otherwise have vate sector does not find the area profitable. As the pub- been treated are treated. Thus certain preventive ser- lic sector raises its prices, however, the competitive vices and health care services in rural areas should be constraints on the private sector are eased. As a result the more heavily subsidized because there are fewer private private sector may raise its prices, and new private providers sector alternatives. may enter the market. These supply responses will affect The first three principles argue that setting prices for ser- the demand for public and private services and, therefore, vices or for particular groups must balance two competing affect health outcomes and resource mobilization. Thus goals: limiting the adverse health effect from a reduction these supply responses should be factored into the setting in utilization and mobilizing resources that can be used to of public sector prices. STRATEGIES FOR PRICING PUBLICLY PROVIDED HEALTH SERVICES 141 One clear message is that the government should sub- Some countries fully subsidize the prevention and treat- sidize services that the private sector is unlikely to provide. ment of communicable diseases. For example, Creese and Public goods are the most obvious candidates for public Kutzin (1995) report that Ethiopia, Ghana, Jamaica, Mali, subsidies. A pure public good is one for which a private mar- Niger, Papua New Guinea, and Zimbabwe do not charge ket cannot exist because beneficiaries cannot be made to for the treatment of tuberculosis. Moreover, all these coun- pay for benefits (nonexcludable) and one person's benefits tries expect Papua New Guinea do not charge for the treat- are not reduced when others benefit as well (nonrivalrous). ment of sexually transmitted diseases. There is evidence Health sector examples include some forms of vector con- that subsidizing the use of public goods programs leads to trol (for example, draining swamps), some forms of sani- improvements in utilization. In China and Zambia child tation (especially in urban areas), and provision of health immunization rates fell dramatically after user fees were information and education for activities such as washing introduced (Booth and others 1995; Sheng-Lan and oth- hands, which have no product associated with them that ers 1994). advertising would promote. Research, epidemiological sur- Taiwan (China) provides an example of a dramatic improve- veillance, and food and drug safety are other examples. A ment in health indicators through public investments at low health service has a positive externality if its use generates income levels. In the 1950s Taiwan (China) was extremely benefits to society above and beyond the benefit to the pri- poor, with a per capita income of less than $150 in today's vate individual. The most common externality in the health terms. Associated with this low living standard were wide- sector comes from prevention and treatment of infectious spread incidences of infectious and parasitic diseases. In 1952 diseases. In the Gambia, for example, pesticide-treated bed- the main causes of death were gastritis, duodentitis, enteri- nets reduced the incidence of malaria even among people tis, clotitis, pneumonia, and tuberculosis. About 1.2 million who had not used them, suggesting that the societal bene- people (in a population of 7.8 million) were infected with fit from bed-nets was greater than the private benefits malaria. In 1962, 383 cases of cholera were reported. About (Tropical Disease Research Program 1995). 90 of the population was infected with hepatitis B by age 40, Left to their own devices, individuals will prevent and and 15 to 20 percent were hepatitis B carriers. The infant treat infectious diseases less than is socially optimal. Many mortality rate was 45 per 1,000 live births; the maternal individuals are unwilling to pay the full cost of immuniza- mortality rate was 197 per 100,000 live births. tion because they know that they will be protected if enough The govermment responded to these problems with exten- other people are immunized.4 Even when immunization sive improvements in water supply and sanitation, disease offers important medical benefits, the cost may impede indi- control programs, and immunization campaigns. Free vac- viduals from seeking treatment soon enough to prevent cinations against the main infectious diseases were made the spread to other individuals or from completing the full available to infants and preschool children. To expand immu- course of treatment. When drug therapies are not com- nization, health education, and treatment, the government pleted, it may lead to a resurgence of the disease, to an also set up primary care facilities throughout the country. increase in transmission, and to resistance to known drug These efforts, combined with better living conditions, therapies. For example, tuberculosis is a virulent, commu- were able to control infectious diseases by the mid- 1960s. nicable disease, and although the drug therapy is available No cases of smallpox or rabies have been reported since and effective, it is expensive. Individuals feel better after 1959. In 1965 Taiwan (China) was declared free of malaria partial treatment and tend to want to stop treatment long by the World Health Organization. By 1970 Taiwan (China) before the course of drugs is completed. They remain a pub- had health indicators similar to those in most industrial lic hazard because they can still transmit the disease. To get countries today. Life expectancy increased from 55 years in individuals to obtain proper prevention and treatment, the 1951 to 69 years in 1970. Neonatal mortality fell by more government needs to use public subsidies to lower the price than half between 1955 and 1970. Infant mortality dropped of these services to encourage utilization. In some cases by about two-thirds over the same period, and maternal the government must fully subsidize the activities. mortality had similar improvements. INNOVATIONS IN HEALTII CARE FINANCING 142 The most striking feature of Taiwan's (China) achieve- or that the health centers provide few services of real ments is that they were realized despite very low income value-that is, they have no drugs and few qualified med- levels. In 1970 real per capita income was $389 (in 1993 ical personnel. In the second case the quality-adjusted dollars), which would place 1970 Taiwan (China) among price differential is too low and, in any event, there is no the poorest countries in today's world. A second clear point health benefit of sending people to health centers. This is that governments should not expend resources where a situation would require improvements in quality to jus- well-functioning market exists. If the private sector provides tify keeping health centers open. an acceptable and affordable alternative to a public ser- A fourth message is that governments should not use the vice, there is little justification for public sector subsidies same consultation fees for each diagnosis and demographic for that service. This situation is most likely in the market group. Fees should be lower-that is, subsidies should be for outpatient services and drugs for noncommunicable dis- higher-for the prevention and treatment of illnesses that eases. In this case the benefits of treatment accrue mostly have large public health externalities and for which demand to the individual; thus there should be a private market for is most elastic. Across-the-board fee increases in Kenya led these services. Moreover, this situation justifies shifting more to a 40 percent reduction in the treatment of sexually trans- subsidies to rural areas where there are fewer private alter- mitted diseases; similar results were found in Zambia (WHO natives. 1994). Similarly, across-the-board fee increases led to reduc- A possible role for government in the market for indi- tions in child immunizations in China (Sheng-Lan and oth- vidual (non-public good) health care services such as cura- ers 1994) and Swaziland (Yoder 1989). Moreover, there is tive care is in cases where private providers have sufficient strong econometric evidence that children's demand for market power to set prices above marginal (incremental medical care is more price sensitive than adults'. Combined unit) costs, as in the case of a monopoly.5 When private with the importance of prevention and treatment early in prices are higher than marginal costs, utilization is lower life, this finding suggests that children's health care ser- than would be warranted by the cost of providing the ser- vices should receive higher subsidies. A similar price elas- vice and there is (deadweight) loss in economic efficiency. ticity argument can be made for directing more subsidies In this case the government could either regulate private to the poor. However, charging the poor a lower fee is admin- prices or directly provide services (priced at cost) through istratively difficult (see below). the public sector. A third message is that prices should be used to direct Adjusting policies to increase equity individuals to the most efficient treatment location. Illnesses and prevention activities that can be treated at Recognizing that poor individuals may not be able to afford health clinics should not be treated at hospitals. Because health care, most countries subsidize their access to care. demand for the treatment of more serious illnesses is In countries where health care is delivered through public less price elastic, increasing the price of hospital care rel- delivery systems, subsidies are used to keep user charges ative to health center care will induce people with less low so that even the poorest families can afford medical serious illnesses to not bypass health centers in favor of care. Support for this use of public subsidies is often based hospitals. Such cascading systems of charges exist in a on the idea that nobody, regardless of income, should be number of countries, including Kenya, Indonesia, Namibia, denied access to basic minimal health care. Although these Zambia, and Zimbabwe (Barnum and Kutzin 1993). Criel commitments are not boundless, they are pervasive through- and Van Balen (1993) found that these price structures out the world. Such arrangements have important implica- succeeded in moving people out of hospitals and into tions in that redistribution policies are inseparable from health centers in Zaire. In Zambia and Zimbabwe, how- health care policies. Unless private health care and insur- ever, hospitals are still overcrowded and health centers ance markets are able to guarantee universal access, gov- still underutilized. This implies that either the price dif- ernments will intervene and subsidize certain services and ferential between health centers and hospitals is too low groups to varying extents. STRATEGIES FOR PRICING PUBLICLY PROVIDED HEALTH SERVICES 143 The health sector is not an effective vehicle for general then move to price discrimination strategies that try to poverty alleviation, however. Studies of the demand for exempt the poor from paying fees. health care show it to be an income-elastic good-that is, the rich spend a larger portion of their income on health Across-the-board subsidies. Many governments try to pro- care than do the poor (Gertler and van der Gaag 1990; mote equity by subsidizing the public health care system. Baker and van der Gaag 1993).6 Thus health care subsi- Because low-income countries have trouble implementing dies accrue more to the rich more than to the poor. Other means testing (that is, identifying the poor individually by goods that are more income inelastic (such as food) would examining their financial resources), they keep fees low for be better vehicles for general poverty alleviation. everyone. This approach amounts to across-the-board sub- Much of the concern about user fees derives from the sidies from the average taxpayer to the average user of health fear that increasing fees may reduce utilization by the poor- facilities. If the average user is poorer than the average tax- a reasonable concern given the strong empirical evidence payer, there is a net redistribution of income. that poor people's demand for health care is more price Many countries, however, allocate most public subsi- elastic than rich people's. Moreover, this finding extends dies to the services used least by the poor-hospital ser- to the case where fees are used to generate revenues that vices. These services are expensive and are rationed byprice, finance improvements in the quality of and access to cura- travel time, and social status rather than by clinical need. tive care. The extent to which this policy improves welfare As a result public subsidies tend to benefit the rich more depends on how willing individuals are to pay for the qual- than the poor. Governments can better target subsidies to ity and access improvements. If individuals are willing to the poor by more heavily subsidizing services that are used pay the full cost of the improvement, the improvements can by the poor. be fully financed through increased user fees without reduc- Indonesia is typical of countries that try to subsidize the tions in utilization. But if the rich are willing to pay but the poor's access to medical care through low-fee public health poor are not, this policy could lead to a reallocation of pub- care systems. The wealthiest 20 percent captures about 29 lic subsidies from the poor to the rich. Thus equity propo- percent of government health care subsidies; the poorest nents are concerned that increased user fees would become 20 percent captures just 12 percent. This is partly because a financial barrier to the poor and reduce their access to the wealthy use hospital services in much greater rates than care (Cornia,Jolly, and Stewart 1987; Gilson 1988). In this the poor. One reason is that hospitals tend to be located in case there is a tradeoff between using subsidies to pursue urban areas close to the wealthy and far from the rural equitable access to medical care and overall improvements poor. Moreover, hospital services are subsidized at much in health. higher levels than are health centers and health subcenters.7 The current situation needs to be remedied because most The situation is similar in Vietnam-the allocation of curative public expenditures are used to care for the non- public subsidies increases with income (World Bank 1995b). poor. Moreover, the poor actually pay higher prices than the These results are driven by the fact that the rich capture a nonpoor when transport and other time costs are taken into much larger share of both hospital inpatient and outpa- account. This is because the current geographic distribu- tient subsidies. This is because they use more hospital ser- tion of public facilities requires the poor to travel much vices, and hospital services receive the highest unit subsidies. further than the nonpoor. These higher prices are in part Although the poor use commune health centers at much responsible for the poor having low utilization rates and greater rates than the nonpoor, this has little impact on the obtaining a small share of public subsidies. benefit incidence distribution because public subsidies to This subsection considers pricing policy in the context commune health centers account for a small portion of pub- of the government expanding its objectives beyond health lic expenditures. to include equity concerns in its objectives for pricing pol- The subsidies that leak to the nonpoor are a major cost icy. We begin with the common approach of using across- of subsidizing the poor with across-the-board subsidies. The the-board subsidies, which are used by many countries, and greater is the income elasticity of demand, the higher is INNOVATIONS IN HEALTII CARE FINANCING 144 this cost of targeting. Jamaica, like Indonesia and Vietnam, more precise. Administrative methods vary from inex- heavily subsidizes hospital care. In order to target one dol- pensive procedures such as geographic price discrimi- lar to the poor, the government must give the nonpoor about nation and targeting by age and gender to costly $3.25 in subsidies (Gertler and Sturm 1997). Similarly, van procedures such as a sliding fee systemwith socialworker der Gaag (1995) shows that while espousing equity as a goal, verification. The additional benefits of better targeting countries such as China, C6te d'Ivoire, Peru, and Tanzania methods need to be compared with the additional admin- also provide higher subsidies to services used by the wealthy. istrative costs of implementing them. Solon and others (1991) shows that high-income individu- Individual price discrimination based on means testing is als in the Philippines receive much more in public health the ideal method for minimizing the revenue loss from care benefits than they pay in taxes, protecting the poor. However, administrative costs and past experiences make means testing ineffective in most coun- Pice discrimination. The extent to which the government tries. For example, in Indonesia and Vietnam the poor can is able to price discriminate and only raise fees that the have user fees waived through an affidavit of indigence. Few nonpoor pay mitigates the severity of this health-equity people, however, seem to take advantage of this mechanism tradeoff.8 To improve equity, the government must develop (World Bank 1995a and 1995b). It is not clear why these policies that lower the price paid by the poor relative to the systems are failing. There are several possibilities: people price paid by the nonpoor by even more than is indicated may not know about the benefit, prices may be so low that by the optimal pricing policies developed in the previous the benefit is not worth the opportunity cost of obtaining section. There are a number of ways to do this. it, local officials may be charging a fee to issue the affi- The government's ability to implement a pricing policy davit, facilities may charge a fee to accept the affidavit, and that maximizes health care outcomes and redistributes sub- there may be a social stigma associated with using the sidies toward the poor depends on its ability to identify affidavit. the poor (in order to price discriminate and target pro- One of the biggest obstacles to means testing lies in mea- grams). Here we consider four common types of target- suring economic well-being in an economy where most peo- ing: individual means testing, geographic targeting, ple pay no income tax and a sizable portion of economic self-selection, and indicator targeting. The targeting effec- resources are home produced. Without accurate, fast, and tiveness criteria: administratively simple methods of identifying the poor, * Type 1 error. failing to exempt someone who should be an individual exemption mechanism may exempt too many exempted. The greater is the type 1 error, the fewer people and sacrifice substantial revenues. More important, poor are protected by the price discrimination method. identifying the poor when they seek treatment is costly and An extreme example of type 1 error would be if facili- difficult. It is certainly beyond the capacity of health care ties charged everyone the full cost of delivering a service. providers to do so, and is impractical outside the context In this case type 1 error would be 100 percent. of a general governmentwide means testing program. Facility- * Type 2 error: exempting someone who should not be based individual exemption programs in the health sector exempted. The greater is the type 2 error, the greater is are too costly, very difficult to administer, and typically are the leakage of potential revenues from the nonpoor and not good at identifying the poor in all developing regions the lower are the subsidies that reach the poor. An extreme (Booth and others 1995; Chaulagai 1995; Ensor and San example of type 2 error would be if everyone were given 1995; McPake and others 1993; Mwabu, Mwanzia, and free care. In this case all potential revenue would be lost Liambila 1995; Nolan and Tubat 1995; Richardson and oth- and type 2 error would be 100 percent. ers 1992; Stinson 1982; and Vogel 1988). * Administrative costs: the costs of identifying the poor Geographic targeting is an alternative way of implement- and implementing price discrimnination can swamp all ing a pricing policy that protects the poor. This approach the gains from price discrimination. There are dimin- attempts to tailor the fee structure to the socioeconomic ishing returns to making price discrimination methods composition of the population served by each health care STRATEGIES FOR PRICING PUBLICLY PROVIDED HEALTH SERVICES 145 facility. If the poor live in more rural than in urban areas Using geographic price discrimination in rural areas where and facilities are located in more urban than in rural areas, the poor are concentrated is a promising way to protect the with uniform fees the poor face higher access costs than poor without sacrificing considerable revenue. Long travel the nonpoor. Thus geographic targeting attempts to locate times prevent people living in wealthier areas from switch- facilities closer to where the poor live. Otherwise the fees ing to the lower-fee facilities in poorer areas once fees in at facilities that serve the poor must be lower in order to the more affluent areas have been increased. However, geo- compensate for the higher time costs so that the price of graphic price discrimination has limited potential in urban access is the same for the poor and the nonpoor. areas where the poor live alongside the nonpoor and most With this in mind, facilities that serve primarily poor facilities are easily accessed by both groups. households would charge zero or near-zero fees, and facil- An alternative approach to protecting the urban poor is ities that serve primarily nonpoor households would charge through differential pricing by level of service and self-selec- higher fees. The fees charged by a facility would rise with tion. The idea is to have low subsidies for services valued the average economic status of the households in its ser- and used mostly by the nonpoor, and high subsidies for vice region. Indeed, facilities in wealthy areas could charge services used mostly by the poor. These are the services for fees equal to or in excess of unit costs. A facility-level fee which demand is income inelastic. schedule increasing with the economic status of the house- One approach is to shift subsidies toward the preven- holds in the facility's service region would imply that gov- tion and treatment of infectious diseases. Since the poor emnment subsidies are pro-poor in that they are largest in tend to suffer proportionally more from infectious diseases, the poorest areas. subsidizing their treatment and prevention not only helps In principle geographic price discrimination is straight- meet public health objectives but also improves the distrib- forward; in practice it is quite complex. Populations within ution of public subsidies across income groups. For exam- a region are not homogeneous. Every region has some house- ple, table 4 shows the distribution of mortality from different holds whose income is below the government's poverty line. causes across different income groups of adult women in In regions where a large portion of the population is poor, China. While poorer women have higher mortality rates from the government can keep fees low enough to protect most all causes, the poor die from infectious diseases propor- of the poor without experiencing high levels of type 2 error. tionately more than do other income groups. Poor women But in regions with a small portion of poor residents, the are 3.5 times as likely as rich women to die from infectious government must choose between forgoing substantial diseases, but are only 1.3 times as likely to die from non- revenues from those able to pay in order to protect a small communicable diseases. Basic principles of targeting (Besley number of poor, or failing to protect the poor in order to and Kanbur 1993) suggest that, if the costs of treatment are reduce revenue loss from the nonpoor. In this case it would the same, reallocating subsidies from noncommunicable dis- be cost-effective to screen the poor at health care facilities eases to the prevention and treatment of communicable or to use an individual discrimination method. diseases would better target public subsidies to the poor. As a general rule, the government should subsidize ser- TABLE 4 vices for which demand is income inelastic-that is, ser- Female adult mortality rates by cause of death vices that are used more by the poor and for which demand and income group in China does not increase much with income. For example, in (percentage dlying between ages 1 5 and 60) Vietnam demand for commune health centers is highly Infectious Noncommunicable income inelastic, while demand for hospital care is income Income quartile diseases diseases Injuries elastic (Gertler and Litvack 1996). Thus keeping subsidies Richest 0.4 6.7 1.2 high for health center care' and low for hospital outpatient 2 0.46 7.6 2.4 services will better target subsides to the poor. Poorest 1.4 8.9 2.7 Vietnam's results are likely to be true for most coun- Source: Murray, Yang, and Qiao 1992. tries-that is, demand for health center care is the most INNOVATIONS IN HEALTHI CARE FINANCING 146 income inelastic, especially in rural areas. This finding sug- Seemingly healthyindividuals can be struck by cancer, injured gests that increasing subsidies for rural health centers best in accidents, or experience bouts of severe diarrhea. This targets subsidies to the poor. It also suggests a general pric- uncertainty is compounded the longer one looks into the ing structure in which fees are lower (and subsidies higher) future and the less one knows about one's current health. if the patient enters the system at the lowest level, and are While most families are able to finance routine care out of progressively higher the further up the system the patient pocket, few are able to finance rare but expensive incidents. enters. Thus if an individual first goes to a commune health In fact, all countries' health care expenditures are extremely center and requires a higher level of care at a hospital, the skewed in that a small portion of the population accounts registration fees should be waived (or at least lowered) at for a large portion of total expenditures. Thus, while most the hospital. This pricing structure provides an affordable families have only small expenditures in a given year, a small portal of entry into the health care system (through the com- number have very large expenditures. mune health centers) and allows people who are willing to Risk-averse individuals prefer to have predictable health pay to go directly to higher levels of care. Since the non- care expenditures. Predictability relieves the worry of how poor are willing to pay to bypass the lower levels, they will to finance costly unexpected illnesses and allows families be charged higher prices and receive lower subsidies. to better plan other consumption. Thus individuals will seek Policies can take advantage of self-selection if a wider to insure themselves against the financial loss associated range of instruments is considered. If adequate medical with uncertain illness. In the absence of formal health insur- treatment is maintained, government facilities may delib- ance people have to informally finance the losses out of erately offer fewer amenities so that only the poor will choose accumulated savings, transfers from relatives and friends, to use them. This approach, however, requires accepting credit markets, or help from charities. different levels of service (at least from the consumers' sat- However, informal insurance seems inadequate. Using isfaction standpoint) in order to concentrate more resources household panel data from Indonesia, Gertler and Gruber on the poor. (1996) show that these informal sources of insurance are Finally, targeting identifiable groups through indicator insufficient for Indonesians to fully finance the costs of targeting is one way to charge those most able to pay at least severe illnesses. When illnesses are severe enough to affect the full cost of care. One group that may be able to afford labor supply and income, their economic costs are partly care is the insured population. Insurance status is a good financed by a reduction in the family's consumption of non- indicator of ability to pay because the insured are wealth- medical goods and services. As a result there is demand for ier than the general population. For example, in Indonesia insurance for both the medical and income costs associ- civil servants are covered by insurance, and almost all civil ated with illness. servants are in the top half of the income distribution (World Despite the demand, most people are unable to buy insur- Bank 1995a). However, the insurance company pays less ance from private sources .because of market failure from than the full cost of care, so the public system still subsi- adverse selection (Rothschild and Stiglitz 1976). Adverse dizes the wealthy insured population. This situation exists selection arises because insurers are unable to observe het- in most countries with mixed public-private systems. By erogeneity in a population's health status. People are born charging the full cost of care to insured patients, public with different genetic makeups that make them more or systems could reduce subsidies to the nonpoor and reallo- less predisposed to illness, and have different life experi- cate them to services used by the poor. ences in terms of exposure to environmental contagion and accidents. For both reasons there is substantial varia- Adjusting resource allocation to insure against financial risk tion in the propensity to become ill. Because insurers do not observe each individual's propensity to become ill, The inherent uncertainty in health status is the classic rea- they cannot write individual contracts. Rather, they are son most industrial countries intervene in health markets forced to offer the best community-rated insurance plans. (Arrow 1963). No one knows what tomorrow will bring. The terms of these contracts can be quite unfavorable to STRATEGIES FOR PRICING PUBLICLY PROVIDED HEALTH SERVICES 147 healthy individuals. Good risks (healthy people) tend to higher user fees tax families "when they are down," impos- subsidize the bad (unhealthy people), and the value of insur- ing higher costs at precisely the point where the marginal ance to the good risks drops significantly. Good risks have utility (value of the next unit) of consumption is highest. an incentive to drop out of the market, leaving the bad Second, subsidies may help mitigate the loss of income from risks to insure among themselves-and substantially dri- illness by financing medical care that improves health and ving up the cost of insurance, making it a financially bad productivity. In essence, public subsidies relax credit con- deal for both insurers and beneficiaries. In many cases it is straints on the purchase of medical care that may help peo- such a bad deal that the insurance market fails to exist. ple get back to work faster. The opposite problem is risk rating (or "cream skim- Public subsidies for medical care can correct failure in ming"), which occurs when unhealthy people are observ- the insurance market, because private markets are unlikely able. Competing on their ability to select good risks leads to supply adequate insurance because of adverse selection. insurers to avoid insuring individuals with preexisting con- Insurance principles suggest that the subsidies should go ditions (such as cancer or AIDS) who are "certain" bad to the services that provide care for the rare, high-cost ill- risks-that is, they will have predictably high medical care nesses that wreak the most havoc on household budgets. expenditures. Insurers do not want to provide these indi- Given that the rich disproportionately use hospital ser- viduals with coverage at the community-rated (average) pre- vices under current systems, there is a distinct tradeoff mium. Instead they either explicitly deny coverage or between equity and efficiency in subsidized hospital care. effectively deny coverage by charging a premium approxi- This tradeoff can be mitigated by enforcing strict referral mately equal to the cost of care. In many high-risk cases rules requiring high charges for people entering hospitals the actuarially fair cost of insurance (expected expenditures directly, with generous exemptions for people who are prop- plus a loading factor to cover administrative costs) may be erly referred. prohibitively expensive and these individuals are effectively Insurance carries with it another market failure called uninsured.9 moral hazard: insured people may use more services than Insurance market failure due to adverse selection occurs they would otherwise because their price at the point of when insurance is voluntary rather than compulsory. Adverse service is lower than without insurance. In this case patients selection and cream-skimming do not occur when everyone tend to consume medical care beyond the point where the is in the insurance pool. Most countries correct for insur- additional benefit is greater than or equal to the additional ance market failure through a universal public system with cost. In this sense too many resources are being allocated subsidized low prices or through compulsory social insur- to treatment. The greater is the price elasticity of demand, ance in which the poor's enrollment is subsidized. the larger is the welfare loss from moral hazard. In public systems heavily subsidized public hospitals pro- These market failures have very different implications vide insurance against large financial loss associated with for pricing policy. When adverse selection prevents the emer- catastrophic illness. However, public systems provide lower gence of insurance markets, public policy should promote levels of insurance if they provide lower-quality services than a pricing structure that protects against catastrophic loss. could be bought in the private sector with social insurance Such a structure typically includes significant copayments funds. by the insured for small expenditures with stop-loss provi- The debate over whether to increase user fees in hospi- sions (caps on out-of-pocket costs) for large expenditures. tals has ignored the crucial role public subsidies play as insur- This takes care of the welfare loss from assuming too much ance. Subsidies can reduce risk in two ways. First, they can risk. But when moral hazard is a serious problem, people make uncertain health care costs more predictable by spread- should face the true costs on the margin to limit overuse of ing them across healthy and sicktimes. Taxes that are incurred services. The risk problem needs to be handled by rela- in all states of health help finance medical care that is pur- tively large, inframarginal payments (Zeckhauser 1970). chased when sick. Thus raising user fees in a world of imper- Thus adverse selection suggests that an optimal payment fect consumption insurance has an important welfare cost: policy would require a large copayment for small expendi- INNOVATIONS IN HEALTII CARE FINANCING 148 tures and none for large expenditures, while moral hazard because families must pay higher premiums at the expense argues for low (or no) copayment for most expected costs of other consumption or savings, but that loss is predictable with a large exposure (full cost burden) beyond that point. and can be spread over the year and across individuals. Since such different policy conclusions follow from differ- The problem of equity in access to health care does not ent institutional structures, a great deal of knowledge about disappear with the introduction of social insurance. However, the way markets work is essential for good policy formulation. using government subsidies to increase equity in access is In most countries the allocation of public subsidies is much easier with social insurance-the government simply consistent with efforts to ameliorate losses from private subsidizes the poor's enrollment in the insurance plans. insurance market failure, since the bulk of public subsidies For this approach to be budget neutral, subsidies provided is spent on hospitals. However, these subsidies are insuffi- directly to facilities have to be reduced to finance the poor's cient to adequately insure families against the risk of finan- enrollment. Facilities recoup the lost revenues by provid- cial loss from unexpected ill health. In fact, despite large ing care for insured patients and being reimbursed. In this subsidies to public hospitals, people are still incurring large way public subsidies are better targeted to the poor, and the out-of-pocket expenditures. Moreover, Gertler and Gruber facilities that get the subsidies are the ones that care for the (1996) show that in Indonesia, a country with a heavily poor. Administering such a program is easier than price subsidized public health care system, families finance the discrimination by facilities at the time care is needed because economic costs of illness by reducing consumption. it is centralized, only needs to be done periodically outside the pressure of having to treat an illness, and can be done Is social insurance the answer? by a trained staff that does not have other responsibilities. Although social insurance can correct some of the prob- Although it is possible to mobilize resources through user lems created by resource mobilization policy, it creates a fees in ways that improve welfare, there are potentially large host of other problems that, if not addressed as part of insur- costs to such a policy. Specifically, the scope for mobilizing ance design, could outweigh its benefits. The most obvious private financing for public expenditures is limited by two problem, mentioned earlier, is that social insurance cannot costs: reduced insurance coverage against the risk of finan- be voluntary. Voluntary insurance markets fail because of cial loss from unexpected illness and reduced utilization adverse selection. For social insurance to be financial viable, and possible consequent adverse health outcomes (espe- enrollment must be compulsory. This is not to say that such cially for the poor). However, these costs are much lower plans must enroll entire populations, but rather segments in health systems financed through social insurance. such as formal sector workers. In fact, most countries already With social insurance, individuals are still insured against have compulsory social insurance for civil servants, many the risk of financial loss from illness when governments raise have expanded compulsory coverage to wage sector employ- user fees. Under social insurance, individuals prepay their ees, and a few have achieved universal coverage. In addi- medical care expenditures (that is, premiums) into a fund tion to adverse selection, social insurance raises a number that is used to pay for their medical care if and when they of important design and administration issues that are beyond become ill or injured. Thus people can avoid paying unex- the scope of this paper. pected fees when they are ill and so are insured against the risk of financial loss from illness. With insurance, individ- Conclusion uals' medical care expenditures (premiums) are predictable and can be planned. Governments mobilize private resources User fees have strong potential for improving the effi- by raising the fees charged by insurance plans for health ser- ciency of health care systems in developing countries. Still, vices provided to the plan's beneficiaries. Because fee they raise many questions and concerns: increases raise premiums and not the out-of-pocket charges * Optimal fee structures can differ dramatically from one at the time of treatment, raising fees causes no loss in the institutional setting to another. Countries differ signifi- insurance value. Raising fees still causes a loss in welfare cantly in terms of the size and performance of the pri- STRATEGIES FOR PRICING PUBLICLY PROVIDED HEALTH SERVICES 149 vate sector, the nature of insurance, credit, and health must come from targeted interventions, as imperfect as care markets, and the administrative capacity of the pub- current targeting methods may be. Health care is highly lic sector to run or regulate health services. Thus much income elastic, and any uniform subsidy system is likely more analysis based on accurate data is needed to define to be regressive. better policies. • In many countries public health budgets are determined Notes reactively: demands for curative care must be honored and funds for population-based public goods are deter- 1. If a price elasticity is small-between 0 and 1-then demand mined residually. In such cases fees for curative care can is said to be inelastic because the percentage reduction in demand is less than the percentage increase in price. When demand is conserve resources for use on these public goods, imnrov- inelastic, price increases raise revenues because the positive price ing efficiency and equity and increasing the effect pub- effect is larger than the negative demand effect. If a price elas- lic spending has on health status. ticity is large-less than -1-then demand is said to be elastic, * Most countries have a large private sector providing pri- and price increases reduce revenues because the negative demand mary healthcare. Public funds should complement rather effect outweighs the positive price effect. Finally, if demand is uni- than crowd out privaeacivityLargsubtary elastic-equal to -1-the percentage decrease in demand is thansiv cerowdou private actiityleto Large subsiesr ineat exactly equal to the percentage increase in price and there is no pensive services may do little to Improve overall health change in revenues. status if they are strong substitutes for private care. 2. For example, if fees are set low in areas where people have * In the many countries in which insurance markets are serious illnesses, the observed correlation between prices and uti- not going to be fixed in the near future, hospitals should lization reflects the fact that sicker individuals use more health remain an important item for public expenditures. This care and the effect of price on utilization. Alternatively, if facili- helps protect people from catastrophicdloss in the absence ties are located near urban areas where individuals are wealthier, helps porotec people frotems. caTastrophiclossinhe abm nce f the correlation between travel costs and utilization reflects the relationship between income and utilization and the effect of travel services, with a small number of people requiring expen- costs on utilization. In both cases the price elasticity estimates are sive care, implies that hospitals will be a large part of the biased, since they are confounded with other omitted factors related budget. to government policy choices. X Many countries will find that the optimal allocation of 3. Strictly speaking, the authors compute the compensating public health subsidies will involve large subsidies for differential, which is the amount of income the individual would pubclinchealth subsicgoodi,fees will prin argheah s ies forfl- be willing to give up for the quality improvement so that there is nonclinical public goods, fees at primary health care facil- lno change in welfare. ities to conserve public resources for these goods, and 4. Transmission of an infectious disease is affected mainly by large per unit subsidies to hospital services with a proper the number of people who are immunized. Thus individual immu- referral system. Such a system entails patients referred nization confers a benefit to people who are not immunized. on the basis of clinical need and high fees for those 5. Another reason for government intervention in the market using hospitals as the first point of contact. This is atlmost for individual health care services arises when the private provider using hospitals asrthe firsubstpint ofscontact.This is aplms b has more information about the patient's illness than the patient the opposite price and subsidy scheme as iS implied by' and the provider is not a perfect agent for the patient. By perfect the current emphasis on primary health care. agent we mean that the provider cares about things other than * In countries with social insurance or adequately regu- the patient's health. In this case the provider could induce the lated private insurance markets, optimal policies would patient to buy more services than he or she might otherwise buy. involve few subsidies to hospital services. These would 6. The best candidates for redistribution through subsidies be coveredl by actuarialy fair prepayments. are goods and services that have low or negative income elastici- ties of demand-that is, goods and services whose use does not increase with income. These are the things that poor people tend ing quality improvements from user fees. to consume relatively more of than other things. * Uniform price increases reduce facility use more for the 7. The subsidies were 206,000 rupiah for a hospital inpatient poor than for others. To the extent that this outcome is visit, 8,100 rupiah for a hospital outpatient visit, 3,400 rupiah for socially undesirable, adequate protection for the poor a health center visit, and 2,200 rupiah for a health subcenter visit INNOVATIONS IN HEALTH CARE FINANCING 150 (World Bank 1993). Subsidies through health centers and sub- van Doorslaer, Adam Wagstaff, and Frans Rutten, eds., Equity centers are much more equitably distributed than hospital subsi- in the Finance and Delivery of Health Care: An International dies because utilization rates of these facilities are more evenly Perspective. New York: Oxford University Press. distributed across income groups. Barnum, Howard, and Joseph Kutzin. 1993. Public Hospitals in 8. As discussed in the previous section, even if the govern- Developing Countries: Resource Use, Cost, Financing. Baltimore, ment is not concerned with equity, there are still good reasons Md.: Johns Hopkins University Press. (related to the desire to improve overall health indicators) to Benefo, Kofi Darkwa, and T Paul Schultz. 1994. "Determinants price discriminate in favor of the poor. of Fertility and Child Mortality in C6te d'Ivoire and Ghana." 9. This problem is exacerbated by the fact that insurance con- Living Standards Measurement Study Working Paper 103. tracts are written for limited terms (for example, one year). Over World Bank, Washington, D.C. time, as more high-risk individuals contract serious illness, the Bennett, Sara. 1989. "The Impact of the Increase in User Fees." number of high-risk individuals able to obtain insurance declines. Lesotho Epidemiological Bulletin 4. With an aging population increasing the number of individuals Bennett, Sara, and Manengu Musambo. 1990. "Report on with long-term chronic illnesses, and improvements in genetic test- Community Financing and District Management Strengthening ing and long-range diagnosis, the number of individuals denied in Zambia." Bamako Initiative Technical Report 5. United coverage can only grow. Nations Children's Fund, Bamako Initiative Management Unit, New York. References Besley, Timothy, and Stephen Coate. 1991. "Public Provision of Private Goods and the Redistribution of Income." American Abel-Smith, Brian, and Pankaj Rawal. 1992. "Can the Poor Afford Economic Review 81 (September): 989-84. 'Free' Health Services? A Case Study of Tanzania." Health Besley, Timothy, and Miguel Gouveia. 1994. "Alternative Systems Policy and Planning 7(4): 329-41. of Health Care Provision."Economic Policy (October): 201-58. Abu-Zied, H., and W Dann. 1985. "Health Utilization in Ismaila, Besley, Timothy, and Raui Kanbur. 1993. "Principles of Targeting." Egypt." Social Science and Medicine 21: 451-61. In Michael Lipton and Jacques van der Gaag, eds., Including Akin,John S., and Charles C. Griffin. 1985. The Demandfor Primary the Poor. Baltimore, Md.: Johns Hopkins University Press. Health Services in the Third World. Totowa, N.J.: Rowan and Birdsall, Nancy, and P Chuan. 1986. "Client Choice of Health Allenhead. Treatment in Rural Mali." Population, Nutrition, and Health - 1986. "Demand for Primary Health Care in the Bicol Department working paper. World Bank, Washington, D.C. Regionof thePhilippines." EconomicDevelopmentandCultural Birdsall, Nancy, F. Orivel, M. Ainsworth, and P Chuhau. 1983. Change 34(July): 755-82. "Three Studies on Cost Recovery in Social Sector Projects." Akin,John S., D. Guilkey, and H. Douton. 1995. "Quality of Services Country Policy Department Discussion Paper 1983-8. World and Demand for Health Care in Nigeria: A Multinominal Probit Bank, Washington, D.C. Estimation." Social Science and Medicine 40(11): 1527-37. Blakney, R., J. Litvack, and J. Quick. 1989. "Financing Primary Alderman, Harold. 1996. "Household Responses to Public Health Health Care: Experiences in Pharmaceutical Cost Recovery." Services: Cost and Quality Tradeoffs." The WorldBank Research Management Sciences for Health, Arlington, VA. Observer 11(1): 3-22. Booth, D., and others. 1995. "Coping with Cost Recovery." Alderman, Harold, and Paul Gertler. 1989. "The Substitutability Stockholm University, Department of Anthropology, of Public and Private Health Care for the Treatment Children's Development Studies Unit, Stockholm. in Pakistan." Living Standards Measurement Study Working Carrin, Guy. 1988. "Community Financing of Health Care." World Paper 57. World Bank, Washington, D.C. Health Forum 9(4): 601-06. - 1997. "Family Resources and Gender Differences in Chaulagai, ChetNath. 1995. "Community Financing for Essential Human Capital Investments: The Demand for Children's Drugs in Nepal." World Health Forum 16(1): 92-94. Medical Care in Pakistan." In HaroldAlderman and Lawrence Ching, Panfila. 1995. "User Fees, Demand for Children's Health Haddad, eds., How Families Make Decisions. Baltimore, Md.: Care and Access across Income Groups in the Philippines." Johns Hopkins University Press. Social Science and Medicine 41(1): 37-46. Alderman, Harold, and Victor Lavy. 1996. "Household Responses Chisvo, M., and L. Munro. 1994. A Review of Social Dimensions to Public Health Services: Cost to Quality Tradeoffs." The of Adjustment in Zimbabwe, 1990-94." United Nations World Bank Research Observer 11 (1): 3-22. Children's Fund, Harare. Arrow, Kenneth. 1963. "Uncertainty and the Welfare Economics Collins, C., and A. Green. 1994. "Decentralization of Primary of Medical Care." American Economic Review 53: 537-38. Health Care: Some Negative Implications in Developing Baker, J., and L. van der Gaag. 1993. "Equity in Health Care Countries." International Journal of Health Services 24(3): Financing: Evidence from Five Developing Countries." In Eddy 459-75. STRATEGIES FOR PRICING PUBLICLY PROVIDED HEALTH SERVICES 151 Cornia, A., R. Jolly, and F. Stewart, eds. 1992. Adjustment with a Conference on Financing Human Resource Development, Human Face. Oxford: Oxford University Press. May, Asian Development Bank, Manila. Creese, Andrew. 1991. "User Charges for Health Care: A Review Gertler, Paul, andJennieLitvack. Forthcoming. "Access to Medical of Recent Experience." Health Policy and Planning 6(4): 603-19. Care during Transition: The Role of the Private Sector in Viet Creese, Andrew, and Joseph Kutzin. 1995. "Lessons From Cost- Nam." Forthcoming in David Dollar, Jennie Litvack, and Paul Recovery in Health." Discussion Paper 2. World Health Glewwe, eds., Poverty and the Transition to the Market in Viet Organization, Division of Strengthening Health Services, Nam. Oxford: Oxford University Press. Geneva. Gertler, Paul, and John Molyneaux. 1993. "How Economic Criel, Bart, and Harrie Van Balen. 1993. "Paying for the Kasongo Development and Family Planning Programs Combined to Hospital in Zaire: A Conceptual Framework." Health Policy Reduce Indonesian Fertility." Labor and Population Program and Planning 8(1): 61-71. Working Paper 93-08. RAND, Santa Monica, CA. de Bethune, X., S. Alfani, and J. Lahaye. 1989. "Influence of an - . 1997. "Experimental Evidence on the Effect of Raising Abrupt Price Increase on Health Service Utilization: Evidence User Fees for Publicly Delivered Health Care Services: from Zaire." Health Policy and Planning 4(1): 76-81. Utilization, Health Outcomes, and Private Provider Response." de Ferranti, David. 1985. "Paying for Health Services in Developing RAND, Santa Monica, CA. Countries: An Overview." Staff Working Paper 721. World Gertler, Paul, and Orville Solon. 1996. "Who Benefits from Social Bank, Washington, D.C. Health Insurance in Low Income Countries?" RAND, Santa Diskett, P, and P. Nickson. 1991. "Financing Primary Health Care: Monica, CA. An NGO Perspective." Development in Practice: An Oxfam Gertler, Paul, and Roland Sturm. 1997. "Private Health Insurance Journal 1(1): 43-51. and Public Expenditures." Journal of Econometrics 77(1): Dor, Avi, Paul Gertler, and Jacques van der Gaag. 1987. "Non- 237-57. Price Rationing and the Demand for Medical Care in Rural Gertler, Paul, and Jacques van der Gaag. 1990. The Willingness to C6te d'Ivoire." Journal of Health Economics 6(December): Pay for Medical Care in Developing Countries. Baltirnore, MD: 291-304. Johns Hopkins University Press. Dow, Wllliam H. 1996. "Unconditional Demand for Health Care Gertler, Paul, Luis Locay, and Warren Sanderson. 1987. "Are in Cote d'Ivoire: Does Selection on Health Status Matter?" User Fees Regressive?"Journal of Econometrics 36(S): 67-88. Living Standards Measurement Study Working Paper 127. Gertler, Paul, B.Y Yang, S.W Lee, andJ. Peabody. 1995. "Financing World Bank, Washington, D.C. Health Care in as Aging Society: A Crisis on Korea's Horizon." Dow, William, and Paul Gertler. 1997. "Insurance and the Demand RAND, Santa Monica, CA. for Public and Private Medical Care in Indonesia." RAND, Gilson, Lucy. 1988. "Government Health Care Charges: Is Equity Santa Monica, CA. Being Abandoned?" EPC Publication 15. London School of Dow, William, Paul Gertler, R. Schoeni, J. Strauss, and D. Thomas. Hygiene and Tropical Medicine, Evaluation and Planning Center 1997. "Experimental Results on the Impact of Medical Care for Health Care, London. Prices on Health and Labor Supply" RAND, SantaMonica, CA. Gilson, Lucy, and Anne Mills. 1995. "Health Sector Reform in Ellis, R. 1987. "The Revenue Generating Potential of User Fees Sub-Saharan Africa: Lessons of the Last Ten Years." In Peter in Kenyan Government Health Facilities." Social Science and Berman, ed., Health Sector Reform in Developing Countries: Medicine 25(9): 995-1002. Making Health Development Sustainable. Cambridge: Harvard Ensor, Tim, and P San. 1995. "Health Charges and Exemptions University Press. in North Viet Nam." United Nations Children's Fund, Bamako Griffn, Charles. 1987. "User Charges for Health Care in Principle Initiative Research Program, New York. and Practice." Economic Development Institute Seminar Paper Fiedler,John. 1993. "Increasing Reliance on UserFees asa Response 37. World Bank, Washington, D.C. to Public Health Financing Crisis: A Case Study of El Salvador." . 1992. Health Care in Asia: A Comparative Study of Cost Social Science and Medicine 36(6): 735-47. and Financing. A Regional and Sectoral Studies book. Frankenberg, Elizabeth. 1995. "The Effects of Access to Medical Washington, D.C.: World Bank. Care on Infant Mortalityin Indonesia." Health Transition Review Haddad, Slirn, and Pierre Fournier. 1995. "Quality, Cost and 5(2): 143-62. Utilization of Health Services in Developing Countries: A Fryatt, R., R. Rai, S. Crowley, andY Gurung. 1994. "Community Longitudinal Studyin Zaire." SocialScienceandMedicine 40(6): Financing of Drug Supplies in Rural Nepal: Evaluating a Fee 743-53. Per Item Drug Scheme." Health Policy and Planning 9(2): Hammer, Jeffrey. 1997. "Economic Analysis for Health Projects." 193-203. The World Bank Research Observer 12(1): 47-72. Gertler, Paul. 1996. "On the Road to Social Insurance: Lessons . Forthcoming. "Prices and Protocols in Public Health from and for Asian Economies." Paper presented at the Care." The World Bank Economic Review. INNOVATIONS IN HEALTI-I CARE FINANCING 152 Hammer, Jeffrey, and Peter Berman. 1996. "Ends and Means in Lewis, Maureen. 1993. "User Fees in Public Hospitals: Comparison Public Health Policy in Developing Countries." In Peter of Three Country Case Studies." Economic Development and Berman, ed., Health Sector Reform in Developing Countries: Cultural Change 41(3): 513-32. Making Health Development Sustainable. Cambridge, Mass.: Light, Donald. 1992. "Equity and Efficiency in Health Care." Harvard University Press. Social Science and Medicine 35(4): 465-69. Hotchkiss, D. 1993. "The Role of Quality in the Demand for Litvack,Jennie, and Claude Bodart. 1993. "UserFeesPlus Quality Health Care in Cebu, the Philippines." PhD thesis. University Equals Improved Access to Health Care: Results From a Field of North Carolina, Department of Economics, Chapel Hill. Experiment in Cameroon." Social Science and Medicine 37(3): Hongoro, Charles, and S.K. Chandiwana. 1994. "The Effects of 369-83. Enforcement of User Fees on the Health Care Delivery System Makinen, M., and S. Block. 1986. "Pricing for Cost Recovery in in Zimbabwe." Ministry of Health and Child Welfare, Harare. Primary Health Care in Guinea." REACH, Arlington, VA. Huber, J. 1993. "Ensuring Access to Health Care with the Manning, Willard,J. Newhouse, N. Duan, B. Keeler,A. Leibowitz, Introduction of User Fees: AKenyan Example." Social Science and M. Marquis. 1987. "Health Insurance and the Demand and Medicine 36(4): 485-94. for Medical Care." American Economic Review 77(3): 251-77. Jarrett, Stephen, and Samuel Ofosu-Amaah. 1992. "Strengthening Mbugua,J. Karanja,GeraldBloom, andMalcolmSegall. 1995. 'impact Health Services for MCH in Africa: The First Four Years of of User Charges on Vulnerable Groups: The Case of Kibwezi in the Bamako Initiative." Health Policy and Planning 7(2): 164-76. Rural Kenya." Social Science and Medicine 41(6): 829-35. Jimenez, Emmanuel. 1987. Pricing Policy in the Social Sectors: McPake, Barbara. 1993. "User Charges for Health Services in Cost Recoveryfor Education and Health in Developing Countries. Developing Countries: A Review of the Economic Literature." Baltimore, MD.: Johns Hopkins University Press for the World Social Science and Medicine 36(11): 1397-1405. Bank. McPake, Barbara, Kara Hanson, and Anne Mills. 1993. 1 1990. "Social Sector Pricing Policy Revisited A Survey of "Community Financing of Health Care in Africa: An Evaluation Recent Controversies." In Stanley Fischer and Dennis de Tray, of the Bamako Initiative." Social Science and Medicine 36(1): eds., Proceedings of the World Bank Annual Conference on 1383-95. Development Economics 1989. Washington, D.C.: World Bank. Murray, Christopher J. L., Gonghuan Yang, and Xinjian Qiao. 1. 1996. "Human and Physical Infrastructure." In Jere 1992. 'Adult Mortality: Levels, Patterns and Causes." In Richard Behrman andT.N. Srinavasan, eds.,HandbookofDevelopment G. A. Feachem, Toud Kjellstrom, Christopher J. L. Murray, Economics. Amsterdam: North Holland Press. Mead Over, and Margaret Phillips, eds., The Health of Adults Kahenya, G., and S. Lake. 1994. "User Fees and their Impact on in the Developing World. New York, Oxford University Press. Utilization of KeyHealth Services." United Nations Children's Mwabu, Germano. 1989. "Non-monetaryFactors inthe Household Fund, Lusaka, Zambia. Choice of Medical Facilities." Economic Development and lCani, N. 1989. "Charging for Drugs inAfrica: UNICEF's Bamako Cultural Change 37(2): 383-92. Initiative." Health Policy and Planning 4(2): 110-20. . 1990. "FinancingHealth ServicesinAfrica: AnAssessment Korte, R., H. Richter;F. Merkle, and H. Gorgen. 1992. "Financing of Alternative Approaches." Population, Health, and Nutrition Health Services in Sub-Saharan Africa: Options for Decision Working Paper 457. World Bank, Washington, D.C. Makers duringAdjustment." Social Science and Medicine 34(1): Mwabu, Germano, andJosephWang'ombe. 1995. "User.Charges 1-9. in Kenya." World Bank, International Health Policy Program, Lake, S. 1994. "User Charges in the Health Sector: Some Washington, D.C. Observations on the Zambian Experience" United Nations Mwabu, Germano, Martha Ainsworth, and Andrew Nyamete. Children's Fund, Lusaka, Zambia. 1993. "Quality of Medical Care and Choice of Medical Lavy, Victor, and Jean-Marc Germain. 1995. "Trade-offs in Cost, Treatment in Kenya: An Empirical Analysis." Journal Of Human Quality and Accessibility in Utilization of Health Facilities: Insights Resources 28(4): 838-62. fromn Ghana." In R. Paul Shaw and Martha Ainsworth, eds., Mwabu, Germano, James Mwanzia, and Wilson Liambila. 1995. Financing Health Services through User Fees and Insurance: Case User Charges in Government Health Facilities in Kenya: Effect StudiesFromSub-SaharanAfrica. Washington, D.C.: World Bank. on Attendance and Revenue." Health Policy and Planning 10(2): Lavy, Victor, and John Quigley. 1993. "The Willingness to Pay for 164-70. the Quality and Intensity of Medical Care." Living Standards Newhouse, Joseph P., and the Rand Corporation Insurance Measurement Study Working Paper 94. World Bank, Experiment Group. 1995. Freefor All? Lessonsfrom the Rand Washington, D.C. Health Insurance Experiment. Cambridge, Mass.: Harvard Lavy, Victor,J. Strauss, D. Thomas, and P de Vreyer. 1996. "Quality University Press. of Health Care, Survival and Health Outcomes in Ghana." Nolan, Brian, and Vincent Turbat. 1995. "Cost Recovery in Public Journal of Health Economics 15(3): 333-57. Health Services in Sub-SaharanAfrica." EDITechnical Series. STRATEGIES FOR PRICING PUBLICLY PROVIDED HEALTIH SERVICES 153 World Bank, Economic Development Institute, Washington, Solon, Orville, and others. 1991. "Health Sector Financing in the D.C. Philippines. " Research Triangle Institute and University of the Olave, M., D. Gimenez, R. Weerhoff and others. 1992. "Study of Philippines School of Economics. Manila. the Recurrent Unit Costs of Primary Health Care Services in Stanton, Bonita, and John Clemens. 1989. "User Fees for Health a Sample of Bolivian Ministry of Health Facilities, 1990-91." Care in Developing Countries: A Case Study of Bangladesh." U.S. Agency for International Development, Bolivia. Social Science and Medicine 29(1): 1199-1205. Parker, David, and Rudolf Kippenberg. 1991. "Community Cost- Stinson, Wayne. 1982. Community Financing of Primary Health Sharing and Participation: A Review of the Issues." Bamako Care. Washington, D.C.: American Public Health Association. Initiative Technical Report Series 9. United Nations Children's Thomas D., Victor Lavy, and J. Strauss. Forthcoming. "Public Fund, New York. Policy and Anthropometric Outcomes." Journal of Public Peabody, John, Paul Gertler, and A. Liebowitz. 1995. "The Effect Economics. of Quality of Prenatal Care on Birthweight." RAND, Santa Tropical Disease Research Program. 1995. TDR News (February). Monica, CA. van der Gaag,Jacques. 1995. Privateand Public Initiatives: Working Pitt, Mark, M. Rosenzweig, and D. Gibbons. 1993. "The TogetberforHealthandEducation. ADirectionsinDevelopment Determinants and Consequences of the Placement of book. Washington, D.C.: World Bank. Government Programs in Indonesia." The WorldBank Economic Vogel, Ronald J. 1988. Cost Recovery in the Health Care Sector: Review 7(3): 319-48. Selected Country Studies in West Africa. World Bank Technical Richardson, P, J. Galloway, and M. Olave. 1992. "Quality, Costs Paper 82. Washington, D.C. and Cost-Recovery: A Comparative Study of the Unidad .1991. "Cost Recovery in the Health Sector in Sub-Saharan Sanitaria of the Ministry of Health and Nutrition Sustainability." Africa." InternationalJournal ofHealth Planning and Management URC, Washington, D.C. 6:167-91. Rosenzweig, Mark, and Kenneth Wolpin. 1986. "Evaluating the Waddington, Cj., and K.A. Enyimayew. 1990. 'A Price to Pay: Effects of OptimallyDistributed Public Programs: Child Health The Impact of User Charges in the Volta region of Ghana." andFamnilyPlanningInterventions." AmericanEconomicReviewv International Journal of Heath Planning and Management 5(4): 76: 470-82. 287-312. Rothschild, Michael, and Joseph Stiglitz. 1976. "Equilibrium in Weaver, Marcia, Holly Wong, Amadou Sese Sako, Robert Simon, Competitive Insurance Markets: An Essay on the Economics and Felix Lee. 1994. "Prospects for Reform of Hospital Fees of Imperfect Competition." Quarterly Journal of Economics in Sub-Saharan Africa: A Case Study of Niamey National 80(November): 629-49. Hospital in Niger." Social Science and Medicine 38(4): 565-74. Sauerborn, Rainer, Adrien Nougtara, and Eric Latimer. 1994. World Bank. 1987. FinancingHealthServicesin Developing Countries: "Elasticity of Demand for Health Care in Burkina Faso: AnAgendaforReform. AWorldBankPolicyStudy.Washington, Differences across Age and Income Groups. " Health Policy and D.C. Planning 9(2): 185-92. . 1993. World DevelopmentReport 1993: Investing in Health. Schwartz,J. Brad,John S. Alin, and BarryM. Popkin. 1988. "Price New York: Oxford University Press. and Income Elasticities of the Demand for Modern Health . 1995a. "Indonesia: Public Expenditure, Prices, and the Care: The Case of Infant Delivery in the Philippines." The Poor." Washington, D.C. World Bank Economic Review 2(1): 49-76. .1995b. "VietNamPovertyAssessment."Washington, D.C. Shaw, R. Paul, and Charles C. Griffin. 1995. Financing Health Care . 1996. "China: Issues and Options in Health Financing." in Sub-Saharan Africa through User Fees and Insurance. A Washington, D.C. Directions in Development book. Washington, D.C.: World Yoder, Richard. 1989. "Are People Willing and Able to Pay for Bank. Health Services?" Social Science and Medicine 29(1): 35-42. Sheng-Lan, T, and others. 1994. "Financing Health Services in Zeckhauser, Richard. 1970. "Medical Insurance: A Case Study of China: Adapting to Economic Reform." Research Report 26. the Tradeoff between Risk Spreading and Appropriate Institute of Development Studies, Sussex. Incentives." Journal ofEconomic Theory 77(1): 10-24. INNOVATIONS IN HEALTH CARE FINANCING 154 Cost Recovery Strategies in Sub-Saharan Africa Joseph Wang'ombe C ost recovery is an essential component of health * Increase efficiency in the provision of health care by mak- care reform-one that requires a major overhaul ing providers cost conscious and encouraging cost-effec- of health care financing systems. For the past ten tive provision of care. years most countries in Sub-Saharan Africa have been actively reforming their health sectors. Before that, these countries The General Model had deteriorating health indicators and weak health infra- structure, partly because of persistent shortfalls in health Many studies have evaluated the effects of cost recovery and care financing. other measures (such as decentralization) on health care rev- Most Sub-Saharan countries share a common agenda enue, quality, equity, and utilization (Creese and Kutzin 1995; forhealth reform that includes increasing economic growth, Shaw and Griffin 1995; Nolan and Turbat 1995; Berman expanding basic schooling (especially for girls), reallocat- 1995; McPake 1993). Yet most studies are hesitant to assess ing government spending on health from tertiary care and the effects of cost recovery on health outcomes for two rea- specialty training to public health measures and essential sons. First, healh outcomes (especially in the long run) are care packages, encouraging diversity and competition in the the result of inputs from many other sectors-including pol- provision of care and the development of cost contain- itics, agriculture, the environment, and national security. ment approaches to insurance, increasing the efficiency of Second, it is difficult to isolate the effects of health out- government health services, and fostering the participa- comes within the health sector. A country that makes good tion of communities and households in promoting health- policy choices and adopts appropriate strategies in the health ier behavior and in managing local health care services (World sector may not generate better health outcomes if policies Bank 1993). are weak in other sectors. Mwabu (1996) demonstrates how In most countries cost recovery efforts are intended to: countries that pursued structural adjustment policies dur- * Raise revenue for health care by imposing user charges ing 1980-93 recorded gains in the health status of their pop- for public health services that used to be provided free ulations. The convergence of cross-sectoral compensatory of charge. effects on health is given as the explanation. Adjustment in * Improve the coverage and quality of care by increasing agriculture, education, and other sectors, combined with resources for the health sector. adjustments in the health sector, increase life expectancy * Enhance equity in the provision of health care by tar- and reduce infant mortality. This paper reviews Sub-Saharan geting spending toward services for the poor and other African countries' cost recovery, cost sharing, and user fee vulnerable groups. reforms in the health sector in the context of general struc- * Improve service utilization patterns and control frivo- tural adjustment efforts, bearing in mind the complex inter- lous demand. actions among different sectors as they affect health. Joseph Wang'ombe is associate professor and chairman of the Department of Community Health at the University of Nairobi. 155 Most recent surveys evaluating Sub-Saharan Africa's per- Country progress with cost recovery can be classified in formance in cost recovery rely on data from 1992 and, in a four categories: a national system of user fees, some national few cases, from 1993 (Shaw and Griffin 1995; Gilson and system of fees but with minimal enforcement, some facili- Mills 1995; Creese and Kutzin 1995; Bennett and Ngalande- ties and communities collecting fees, and no apparent form Banda 1994; McPake 1993; Nolan andTurbat 1993). During of user fees (table 1). The countries in the first category are 1993-95, however, many countries in the region imple- implementing clear policies on cost recovery and are run- mented important health sector reforms. These efforts ning a unitary health care delivery system; that is, the cen- may reflect the release of the World Bank's World ter controls or oversees regional efforts. Countries in the Development Report 1993: Investing in Health, and the cam- second category also have a unitary system, but cost recov- paign that ensued thereafter. Thus attempts should be made ery policies are less evolved, and institutions have not been to compile more recent survey data. reoriented to implement cost recovery policies. Countries in the third category do not have a national system, but indi- Recent Reform Efforts vidual communities and health facilities are implementing cost recovery schemes. (In these countries the Bamako ManyAfrican countries have adopted health sector reforms, Initiative is implemented separate from the general policy including: of cost recovery.) Countries in the fourth category are still * Charging user fees at public facilities to generate rev- formulating policies and establishing structures and insti- enue from out-of-pocket payments, social insurance, pri- tutions (at this point only Botswana occupies this category). vate insurance, and community financing (for example, Cost recovery for public health care services was ini- the Bamako Initiative). tially viewed as a panacea for inadequate revenue genera- * Increasing funds for the health sector and raising the tion in the health sector. However, it has become apparent level of spending for public health care for the poor and that although they can increase revenue in absolute terms, other vulnerable groups. direct fees have little proportionate impact on the public * Adopting innovative financing arrangements in the orga- health system's financial requirements (Waddington and nization and management of health care delivery sys- Enyimayew 1989; Shaw and Griffin 1995). Most countries tems. The most common strategy has been to decentralize in Sub-Saharan Africa recover 3-5 percent of ministry of health care delivery systems and privatize public insti- health recurrent expenditures through user fees (table 2). tutions. The literature espousing user fees, however, claims that they * Providing incentives to expand nongovernmental deliv- should recover as much as 20 percent of recurrent costs ery of health care (for example, tax incentives for pri- (World Bank 1993). vate health care providers and subsidies to religious Such levels-and some much higher-have been groups, programs, and institutions). achieved, but only in small-scale projects and community- based schemes similar to the Bamako Initiative. In Ghana, User Fees As a Source of Additional for example, user fees under some programs have recov- Revenue ered 52 percent of costs, in Guinea-Bissau 32 percent, in Mali 55 percent, in Senegal 50 percent, and in Uganda 19 A recent World Bank review of user fee schemes in Sub- percent. SaharanAfricafound thatin 1993 onlyfive countries-Angola, These high levels of cost recovery are partly explained Botswana, Malawi, Sao Tome and Principe, and Tanzania- by the special circumstances of these initiatives. Such pro- did not have any apparent form of user fees or cost recovery jects have access to essential (often foreign) management in the public health care system (Shaw 1995). By 1994 these skills and can finance necessaryinfrastructure. Inthe Bamako countries (except Botswana) had started to implement some Initiative projects, external funds from the United Nations form of cost recovery, and by now have acquired some expe- Children's Fund (UNICEF) are in some places used to rience with cost recovery (Bennett and Ngalande-Banda 1994). buy the initial stocks of drugs that establish the base for INNOVATIONS IN HEALTH CARE FINANCING 156 TABLE I Cost recovery in public health facilities in Sub-Saharan Africa, 1996 Category I Category 2 Category 3 Cost recovery in place and Some national system of user No national system of user Category 4 dominated by national system fee but minimal or poor fees but some facilities and No apparent form of user fees of user fees enforcement communities collect fees or cost recovery in place Anglophone,usophone countnies Angola Equatorial Guinea Botswana The Gambia Ethiopia Ghana Guinea-Bissau Kenya Malawi Lesotho Nigeria Malawi Sao Tom6 and Principe Mozambique Sierra Leone Namibia Sudan Swaziland Tanzania Francophone countries Benin Burkina Faso Central African Republic Burundi Mauritania Congo Cameroon Togo Madagascar C6te d'lvoire Rwanda Niger Guinea Zaire Mali Senegal Source: Derived from Shaw 1995. the revolving drug fund. Moreover, the projects are cush- demand that such fees generate can improve equity by ioned against inadequacies in the general health care system. increasing the availability of services and encouraging appro- The potential for revenue generation from direct user priate utilization (World Bank 1987; de Ferranti 1985; fees has not been realized. Shortfalls may be explained by Mwabu and Mwangi 1986). The increased revenue should inefficiencies, implementation costs, lack of skills, an inabil- be reinvested and targeted toward services that improve the ity to pay for full service cost, and incomplete institutional health of the poor. To ensure that vulnerable groups are adjustment and realignment (McPake 1993). Yet even if not excluded from public services, exemption schemes fully realized, the potential of user fees would not result in should be developed. Moreover, these groups should be full cost recovery. Hence work continues on alternative entitled to public subsidies. sources of finance, including social insurance, private insur- Although recent studies on the effect of user fees on uti- ance, and community-based schemes (Shaw and Griffin lization rates have found negative consequences for equity, 1995). this conclusion is somewhat controversial given earlier stud- ies by Heller (1982) and Akin and others (1985), which User Fees and Equity found that demand was inelastic with respect to price and income. Yet later work, using different econometric for- Discussions of user fees and equity in Sub-Saharan Africa mulations from Heller and Akin and others, found that generally focus on the utilization, accessibility, and avail- users are sensitive to price changes (Gertler and van der ability of services, considerations for indigent and low- Gaag 1990; Waddington and Enyimayew 1990; Mwabu income groups, initiatives to deal with health problems and Wang'ombe 1995; Bennett and Ngalande-Banda 1994). specific to the region (for example, childhood diseases such Even in countries where price elasticities are low, large as diarrhea, measles, and whooping cough), and support for drops in utilization have been demonstrated-in Kenya public health care approaches in general. Proponents of user by up to 38 percent (Mwabu and Wang'ombe 1995). Huge fees claim that the increased revenue and price effects on drops in utilization in countries where price elasticities are COST RECOVERY STRATEGIES IN SUB-SAHARAN AFRICA 157 TABLE 2 Means testing is one method that is used to identify the Ministry of health recurrent expenditures recov- poor (see Levine and others 1992 for examples of means ered through user fees, various countries testing in Ethiopia, Ghana, and Senegal). Most countries, (percentage of total) however, lack sufficient skills to use means testing to deter- Country Share Year mine the extent of poverty and implement appropriate Botswana i1.3-2.8 1983 exemption policies. As a result fee exemption programs are often exploited by the nonpoor (civil servants, medical C8ted'lvoire 3.1-7.0 1986 Ghana 7.9 1986 workers, the military, and so on). In Kenya, for example, 7.8 1992 large amounts of potential revenue were lost wheen exemp- Guinea-Bissau 0.5 1988 tions were granted to civil servants and health workers Kenya 2.1 1993 (Collins and others 1995). This practice has since been replaced with a medical allowance to civil servants with Lesotho 5.8 1986/87 9.0 1991/92 which it is hoped they will pay for medical care. Civil ser- Mali 1.2-7.0 1986 vants in Ghana, Mali, Niger, St. Lucia, and Yemen also Mozambique 8.0 1985 receive non-income-related exemptions (Bennett and Lessthan 1.0 1992 Ngalande-Banda 1994). Senegal 4.4-7.0 1986 The increased revenue from user fees was expected to Swaziland 2.2 1985 help deliver services of public health importance (immu- 4.6 1988/89 nizations, communicable disease control) and to support Memo items public health care in general. As noted above, however, China (exclude insurance reimbursements) 24.0 1980 36.0 1988 revenue potential has not been realized. No African coun- try has managed to use funds from cost sharing to finance public utility health care services; public health care is still Salvador 4.0 1990 financed by the budget. Retaining fees in the institutions Yemen 3.3 1983 and districts where they are collected might help address Source: Creese and Kutzin 1995. this problem, so long as districts are able to decide on the low can be explained by the fact that fees were introduced services mix for funding. Retaining fees where they are col- where they previously did not exist or were very low. In lected is also a necessary condition of decentralization, which some cases declines in utilization reversed slightly after is becomning popular. This approach is now operational in users adjusted to the initial price shocks (Collins and oth- Uganda (Tmdyebwa 1997), Kenya (Collins and others 1995), ers 1995). and Mozambique, Nigeria, and Zambia (Bennett and In most cases two groups cut back on their use of pub- Ngalande-Banda 1994). Other countries-Burundi, Congo, lic services when user fees are introduced: people who are Central African Republic, Ghana, Guinea, Malawi, Mali, completely excluded from modern health care because they Niger, Togo, Zaire-split revenues between health facilities depend entirely on government services, and people who and the ministry of health (Bennett and Ngalande-Banda are picked up by other providers. Equity considerations 1994). require that the first group be protected, usually with fee In these countries only health centers retain all their fee exemptions. All Sub-Saharan Africa countries that are pur- revenue; hospitals remit some of their income to the cen- suing cost recovery have some type of exemption program. tral treasury. In Ethiopia and Namibia all the money reverts These programs attempt to identify people who cannot to the treasury. Retention of fees in districts and institutions afford the services they need and to make sure that people could also increase equity in another area-namely, by who can afford them are charged for the services they use. improving the quality of care. Revenues are expected to be In many countries the task of identifying the deserving poor reinvested not only to increase (or maintain) the volume of is left to communities or managers of health care institutions. services but also to improve their quality. Moreover, users INNOVATIONS IN HEALTH CARE FINANCING 158 will demand quality improvements if they have to continue lic health care services (such as family planning). Nationally, paying. The slow recovery of service utilization rates in user charges are being implemented on a gradated scale. Ghana and Kenya after user fees were imposed can be partly Gate fees are higher at provincial hospitals and lower at attributed to the fact that facilities could not retain fees to health centers. There are no user charges at village dispen- invest in quality irnprovements (Waddington and Enyimayew saries (Collins and others 1995). Cbte d'Ivoire, Ghana, Mali, 1990; Collins and others 1995). Namibia, Zambia, and Zimbabwe also use gradated user Policies have since been changed in Kenya, however. charges (Barnum and Kutzin 1993; Bennett and Ngalande- Facilities now retain 100 percent of fees and are supposed Banda 1994). Namibia is encouraging proper use of a refer- to use 75 percent for quality maintenance and 25 percent ral system by exempting referral clients from charges at for public health care. Commnunity projects and the Bamako higher-level institutions (Creese and Kutzin 1995). The rev- Initiative tend to be successful at increasing quality, partic- enue collected at various levels is used to improve quality ularly in francophone countries such as Cameroon and and to encourage client conformity with the referral system. Guinea (Litvack and Bodart 1993; Nolan and Turbat 1993). Mwabu and others (1995) show the importance of avail- The revolving fund feature of these projects allows funds ability of drugs on demand of services. Bamako Initiative to be used to maintain drug supplies. These successes help projects are credited with maintaining quality by making legitimize decentralization policies. drugs available. If the selection of drugs covers local dis- eases, clients only need go to higher-level facilities for refer- User Fees and Efficiency of Service ral services. Drugs maybe cheaperat the communityfacility Delivery than at the referral facility. But even if they are not, having drugs available at the community level allows clients to It is often claimed that user fees increase efficiency in the escape other consumption costs (such as transport and travel delivery of health care, for several reasons. First, user fees and waiting time). cause service providers and users to behave more efficiently. As noted above, user fees can also lower utilization rates. Second, providers have an incentive to allocate revenues Presumably, part of that reduction is due to a reduction in to produce appropriate services at the appropriate levels, frivolous demand. Although lower utilization rates may and to choose appropriate production techniques. Finally, increase efficiency, there is no evidence showing what por- user fees send price signals to which clients respond by using tion of the reduction is due to frivolous demand. only the services that they need. As a result clients will respect the referral system and seek only primary care at low-level User Fee Implementation institutions (health centers and dispensaries) and only ter- tiary care at high-level institutions (district and provincial The context in which cost recovery is implemented-that hospitals and other referral institutions; World Bank 1987; is, the institutional, administrative, and management frame- de Ferranti 1985). Until cost sharing was adopted, man- work-is as important for the success of the policy as the agers of public services were not concerned with allocative ability of users to pay for services or the quality of those and technical efficiency. It was common to see tertiary services. Since health reform depends on successful reforms institutions providing primary care. The pyramid referral in other, related sectors, it must be implemented as part of system was failing. a broader process of structural adjustment. But success in Evidence on cost sharing in Sub-Saharan Africa reveals the health sector also depends on internal structures and a dynamic picture of policy initiatives affecting people's capabilities. Success in cost recovery requires a public sec- choices of levels and types of services and gradation of tor that is capable of admrinistering and managing the process. fees. In Kenya, for instance, the national referral hospital Most examples of administrative and management prob- (Kenyatta National Hospital) has in the past four years lems come from Ghana and Kenya, where districts and stopped providing adult outpatient care except emergency regions hold funds for long periods in non-interest-earn- casualty services. The hospital also no longer provides pub- ing accounts and spend too little in the face of shortages COST RECOVERY STRATEGIES IN SUB-SAIIARAN AFRICA 159 and financial needs (Waddington and Enyimayew 1990; less than the 20 percent anticipated. This potential is not Collins and others 1995). These problems can be traced to realized because of organizational and management prob- a lack of knowledge and skills in financial management or lems. Moreover, implementation of cost recovery efforts to bottlenecks in making decisions on expenditures. must ensure that quality, equity, and efficiency are Ministries of health in anglophone Africa traditionally trans- maintained. ferred spending authority from headquarters to regions and Different types of potentially sustainable and effective districts and imposed strict timeframes and amounts. policies and programs for achieving quality, equity, and Although this system worked well when ministries provided efficiency are being tried in Sub-Saharan Africa. These free services and districts spent money granted by head- include fee exemption schemes that target vulnerable groups, quarters, it is no longer effective. fee retention at the point of collection to reinvest in qual- Most attempts to respond to these problems have focused ity, and reorganization and decentralization of health care on two policies. The first frees the provision of health ser- delivery systems to enhance efficiency vices from the encumbrances of the civil service by creat- These efforts face an apparent catch-22, however. Because ing autonomous or semiautonomous boards. Zambia has systems are not raising enough revenue from user fees, there proposed creating health service boards, and Ghana is con- is not enough money to reinvest in quality improvement sidering developing Ghana Health Services, which would and system adjustments. But these problems are not insur- be similar to Ghana Education Services. Management of mountable, as evidence by the Bamako Initiative and other large hospitals and medical centers-including Muhimbili community-based projects that have reported success at Hospital in Tanzania, Mulago Hospital in Uganda, and the local levels. Properly managed projects have been able to Royal Victoria Hospital in the Gambia-also may be put use initial external funding to lay a foundation for a sus- under independent boards (Cassels 1995). tainable revolving fund. This model should be applied to The second response decentralizes responsibility for national-level programs where policies are clear. An exter- the management or provision of health services to local gov- nal injection of funds targeted at quality improvements ernments or health care agencies. However, this approach should enable health care delivery systems to attract higher creates many difficulties, mainly due to lack of resources fees from clients. At the same time, however, governments and managerial capacity at the local level. For example, in must demonstrate the political will to pursue these policies Nigeria and Tanzania health budgets are raided to finance as well as restructure public health services. Structural adjust- local administration. In Zambia the reform program envis- ment in other sectors is essential as well. aged that district and hospital boards would set their own fees and employ staff. This approach, however, could increase References the inequity that decentralization is meant to address. The center needs to retain some control over the distribution Akin, J. S., C. C. Griffin, and D. K Guilkey. 1985. The Demand of resources, but without imposing too many rules and reg- for Primary Health Services in the Third World. Totowa, NJ.: ulations on the civil service (Cassels 1995) Rowan and Alenheld. Arhin, Dyna. 1992. "Health Financing in Zambia: An Approach to Cost-Sharing." Mission report. U.S. Agency forInternational Conclusion Development, Washington, D.C. Barnum, Howard, and Joseph Kutzin. 1993. Public Hospitals The debate on user fees for health care services has moved in Developing Countries: Resource Use, Cost, Financing. from advocacy toward issues involving the programs' effi- Baltimore, Md.: Johns Hopkins University Press for the ciency and effectiveness. So far the experience with user World Bank. Bennett, Sara, and Ellis Ngalande-Banda. 1994. "Public and Private Roles in Health: A Review and Analysis of Experience in Sub- pointing. Except in cases of vertical programs and com- SaharanAfrica." Strengthening Health Services Paper 6. World munity-based projects, user fees have failed to provide more Health Organization, Division of Strengthening Health Services, than 5 percent of recurrent ministry of health costs-much Geneva. INNOVATIONS IN HEALTH CARE FINANCING 160 Berman, Peter, ed. 1995. Health Sector Reform in Developing Five Case Studies." London School of Hygiene and Tropical Countries: Making Health Development Sustainable. Cambridge, Medicine, Department of Public Health and Policy, Health Mass.: Harvard University Press. Policy Unit, London. Cassels, Andrew. 1995. "Health Sector Reform: Key Issues in Less Mwabu, Germano. 1996. "Health Effects of Market-Based Developed Countries." Discussion Paper 1. World Health Reforms in Developing Countries.' World Institute for Organization, Geneva. Development Research Working Paper 120. Helsinki. Collins, D.H.,J. D. Quick, S. N. Musau, and D. L. Kraushaar. 1995. Mwabu, Germano, and W Mwangi. 1986. "Health Care Financing "Health Financing Reformn in Kenya: The Fall and Rise of Cost- in Kenya: A Simulation of the Welfare Effects of User Fees." Sharing, 1989-1994." Management Sciences For Health. Nairobi. Social Science and Medicine 22(7): 763-67. Creese, Andrew, and Joseph Kutzin. 1995. "Lessons From Cost Mwabu, Germano, andJosephWang'ombe. 1995. "User Charges Recovery in Health." Discussion Paper 2. World Health in Kenya: Health Service Pricing Reforms in Kenya, 1989-1993. Organization, Geneva. A Report on Work in Progress with Support from the de Ferranti, David. 1985. "Paying For Health Services in Developing International Health Policy Program. " IHPP, Washington, D.C. Countries: An Overview." Staff Working Paper 721. World Mwabu, Germano, Martha Ainsworth, and Andrew Nyamete. Bank, Wshington, D.C. 1995. "The Effects of Prices, Service Quality, and Availability Gertler, Paul, and Jacques van der Gaag. 1990. The Willingness to on the Demand For Medical Care. Insights From Kenya." In Pay for Medical Care. Baltimore, Md.: Johns Hopkins Press. R. Paul Shaw and Charles C. Griffin, eds., Financing Health Gilson, Lucy, and Anne Mills. 1995. "Health Sector Reforms in Care in Sub-Sabaran Africa through User Fees and Insurance. Sub-Saharan Africa: Lessons of the Last 10 Years." In Peter Washington, D.C.: World Bank. Berman, ed., Health Sector Reform in Developing Countries: Nolan, B., and V Turbat. 1995. "Cost Recovery in Public Health Making Health Development Sustainable. Cambridge, Mass.: Services in Sub-Saharan Africa." World Bank, Economic Harvard University Press. Development Institute, Washington, D.C. Griffin, Charles. 1989. "User Charges for Health Care in Principle Quick,Jonathan. 1995. "Impact of Health CenterFees and Update and Practice." EDI Seminar Paper 37. World Bank, Economic on Hospital Utilization." Management Sciences For Health, Development Institute, Washington D.C. Boston. HeJler,P S. 1982. 'A Model of theDemand forMedical and Health Shaw, R. Paul. 1995. "User Fees in Sub-Saharan Africa: Aims, Services in Peninsular Malaysia." Social Science and Medicine Findings, Policy Implications." In R. Paul Shaw and Charles 16(3): 267-84. C. Griffin, eds., Financing Health Care in Sub-Saharan Africa Kenya Government. 1994. "Kenya's Health Policy Framework." through User Fees and Insurance. Washington, D.C.: World Bank. Nairobi. Shaw, R. Paul, and Charles C. Griffin. 1995. Financing Health Care -.1996. "Kenya's Health Policy Framework: Implementation in Sub-Saharan Africa through User Fees and Insurance. A and Action Plans." Nairobi. Directions in Development book. Washington, D.C.: World Korte, R., and others. 1992. "Financing Health Services in Sub- Bank. Saharan Africa: Options for Decision Makers during Tindyebwa, Dennis. 1996. Personal commnunication. Ministry of Adjustment." Social Science and Medicine 34(1): 1-9. Health, Uganda. Lecher, M., R. Morar, B. Mohlala, F. Sithole, E. Semenya, and D. Waddington, C., and K. Enyimayew. 1989. "A Price to Pay: The Mphatsoe. 1995. "Modelling the Effects of National Health Impact of User Charges in Ashanti-Akim District, Ghana." Insurance on a Health District." University of Cape Town, InternationalJournal of Health Planning and Management 4(1): Department of Community Health, South Africa. 17-47. Levine, R., and others. 1992. "Means Testing in Cost Recovery: . 1990. 'APrice to Pay, Part 2: The Impact of UserCharges A Review of Experience." Technical Note 23. U.S. Agency for in the Volta Region of Ghana." International Journal of Health International Development, Health Financing and Sustainability Planning and Management 5(4): 287-312. Project, Washington, D.C. Waters, H. 1995. "Literature Review: Equity in Health Sector in Litvack, Jennie, and Claude Bodart. 1993. "User Fees plus Quality Developing Countries, with Lessons Learned for Sub-Saharan Equals Improved Access to Health Care: Results of a Field Africa." Basics Trip Report. U.S. Agency for International Experiment in Cameroon." Social Science and Medicine 37(3): Development, Washington, D.C. 369-83. World Bank. 1987. Financing Health Services in Developing Countries: McPake, Barbara. 1993. "User Charges For Health Services in An Agenda For Reform. Washington D.C. Developing Countries: A Review of the Economic Literature." . 1993. World Development Report 1993: Investing in Health. Social Science and Medicine 36(11): 1397-1405. New York: Oxford University Press. McPake, Barbara, Kara Hanson, and Anne Mills. 1993. . 1994. Better Health in Africa: Experience and Lessons "Implementing the Bamako Initiative in Africa: A Review and Learned. A Development in Practice book. Washington D.C. COST RECOVERY STRATEGIES IN SUB-SAHARAN AFRICA 161 Rural Risk-Sharing Strategies Andrew Creese and Sara Bennett R isk-sharing strategies in health care relate not to ill physicians cluster in the provincial capitals ... the qual- health, but to its financial consequences. The costs ity of rural specialists is lower ... the types of health facil- of health care, like those of other goods and ser- ities provided almost rule out adequate emphasis on vices, are commonly assigned to the individuals using the villages ... referrals of the rural sick to urban hospitals services, and these costs may be reassigned or shared using do not happen on any substantial scale. (pp. 265-66) a variety of risk-sharing institutions and mechanisms. Risk sharing can be defined as the reduction or "elimination of Of potential significance to risk-sharing potential in rural the uncertain risk of loss for the individual or household by areas, Lipton also established that welfare disparities were combining a larger number of similarly exposed individu- typically smaller in rural areas than in urban areas. als or households who are included in a common fund that Despite recent rapid urban growth, rural areas still dom- makes good the loss caused to any one member" (ILO inate developing countries. In 1988, 74 percent of Asia's 1996a, p. 3-1). These concepts apply equally to rural and population (including China and India) and 73 percent of urban populations. Where rural and urban areas may dif- Sub-SaharanAfrica'swas rural; in the least developed coun- fer substantially is in the implementation conditions facili- tries as a group this figure was 80 percent, with 69 percent tating or impeding risk-sharing efforts. Industrialization, of the rural population below the poverty line (Jazairy, urbanization, high and rising per capita incomes, and pop- Alamgir, and Panuccio 1993). It will be well into the third ulation density-typically urban characteristics-facilitate decade of the next century before more than half of Africa's the growth of insurance (Ensor 1997). population is urban (UN 1993). Writing twenty years ago, Lipton (1976) showed that Rural residents work primarily in agricultural activities- earnings and leisure-that is, welfare-are substantially much of it seasonal, self, or family employment. Cash income lower in rural than in urban areas. In advancing his theory is seasonal and subject to large annual fluctuations. In poor of "urban bias," Lipton also argued that these gaps are clear- countries many rural residents face cash liquidity constraints est in the field of medicine: for much of the year. In both urban and especially rural areas a large portion of the economically active population is The townsman has nine times as good a prospect of engaged in informal employment. The size of the formal medical attention as the villager in India, eleven times sector, and its rate of expansion or contraction, has been in Ghana, thirty-three times in Ethiopia. The poorer, argued to be an important background factor in the suc- the larger in area, and the less densely populated a coun- cess or demise of national health insurance schemes (WHO try is, the greater in general is this disparity... Most rural 1995; Preker and Feachem 1995). It remains to be seen Andrew Creese is health economist in the Division of Analysis, Research, andAssessment at the Wrld Health Organization. Sara Bennett is applied research director at the Partnerships for Health Reform Project in Washington, D.C., and lecturer in health economics and financing at the London School of Hygiene and Tropical Medicine. The authors are grateful to David Dunlop, Wouter van Ginneken, Joseph Kutzin, and Roeland Monasch for comments, suggestions, and practical assistance in preparing this paper. 163 whether developing countries can develop mechanisms to and all health care systems are characterized by some risk- spread small-scale rural risk-sharing schemes to the major- sharing arrangements. Thus risk-sharing arrangements in ity of the population. Although the rural sector in develop- rural areas could include publicly financed health care, com- ing countries has special features that affect risk sharing munity-based health insurance, and even private health for health care costs, these issues are not fundamentally dif- insurance. ferent from those faced by informal workers in urban areas. Focusing on individual schemes, or even broad types of Urban informal employment is of increasing importance insurance, impedes an overall assessment of the role of insur- in developing countries, and in recent years has grown faster ance and the performance of all actors. In large part the than formal employment. In Latin America the informal debate remains straight-jacketed because insurance is treated sector now accounts for most urban employment (table 1). wholly or predominantly as a source of health care finance. Similar trends are reported for South Asia and Sub-Saharan Saltman (1995), for example, distinguishes finance, allo- Africa. Thus, although this paper's focus is on risk-sharing cation mechanisms, and production components in com- arrangements for the rural informal sector, it also analyzes paring health care systems. But health insurance is not, some urban and urban-rural schemes. strictly speaking, a source of finance, but rather a type of Informal employment creates substantial difficulties in allocative mechanism. the development of health insurance: it is difficult to iden- Experience with rural or informal risk sharing is of major tify beneficiaries, to assess their incomes, and to collect con- potential relevance to countries at different levels of devel- tributions. Mandating coverage, which offers substantial opment. With a growing recognition of central governments' advantages in terms of the size of the risk pool and control limited ability to finance and manage health care, new forms of adverse selection, is also much harder for informal than of finance and, perhaps more important, new forms of orga- formal workers. nization are being introduced. In low-income countries with The scope of the concept of risk sharing also warrants poor growth prospects and large informal sectors, a major consideration. Discussions about rural risk sharing-or goal of health reform is to find new ways to organize acces- about health insurance more generally-typically analyze sible care of good quality, using a maximum of nongovern- particular schemes (as this paper does). Inmost discussions ment resources in a transparent manner. of the health insurance prospects of developing and tran- The work presented here is preliminary. It reviews recent sition econornies, the analysis quickly moves from concep- evidence on the organization and performance of health tual definitions of risk sharingto analysis or recommendations insurance schemes for the informal sector. Despite our argu- on design and performance characteristics of particular ment that risk sharing should be viewed as a function rather schemes. Seldom is insurance considered in broad terms than as a category of schemes, it has proved necessary at as afunction of health care systems, and as an overall pol- this point to confine our analysis. At a later date we intend icy objective. But the pooling of risks is a policy objective, to develop the analysis to consider a wider variety of risk- sharing mechanisms. TABLE I Thirty-six risk-sharing schemes for the informal sector Urban informal nonagricultural employment were reviewed. These schemes are in no way representa- in Latin America, 1990 and 1994 with rural and urban informal health (percent)tieoalexrens insurance. The schemes included here were selected based Country 1990 1994 on the availability of adequate documentation. Thus our Argentina 47.5 52.5 sample is likely biased toward successful schemes. Failed Boliva 56.9 61.3 Brazil 52.0 56.4 schemes, while equally instructive, are less likely to have Chile 49.9 51.0 been documented. Colombia 59.1 61.6 The schemes examined spread risks to different degrees Ecuador 51.6 54.2 Peru 51.8 56.0 and in different ways. The level of risk sharing depends on Source: ILO 1996b. the risks that are included in an insurance benefits pack- INNOVATIONS IN HEALTII CARE FINANCING 164 age and on how large and diverse the risk pool is. The high- Thus the relatively small set of documented recent expe- est level of risk pooling occurs in tax-funded, mandatory riences in this area should be seen as a subset of this older, national health service systems that are capable of provid- more diverse set of risk-sharing arrangements. Given that ing accessible services to the entire population. At the oppo- they have barely been analyzed in terms of their potential site extreme is out-of-pocket payment by individuals for for expanding into health insurance funds, the lessons that their care. An individual's risk of incurring the financial costs can be drawn from both sets of experiences are limited. of health care can also be distributed over time. The sim- But history shows that people have organized and man- plest example occurs when prices for health care are set on aged cash-based risk-sharing mechanisms for big expendi- a "fee per episode" basis, which allows the patient an unlim- ture ("catastrophic") events, often with complex contribution ited number of visits until a defined illness episode is over. and benefit arrangement schedules. Membership is usually More ambitious intertemporal risk sharing is offered through individual (rather than household), voluntary, and tightly nontransferable health care benefits for a defined period, limited. Kinship and trust is usually important. The impor- rather like a season ticket for health care. The most sys- tance of social homogeneity and group confidence signal tematic intertemporal risk-redistributing mechanisms are the limitations of such schemes as a basis for national schemes offered in schemes such as Singapore's Medisave, an indi- unless they are supported as a set of initiatives that other vidual earmarked medical care savings account that is avail- actors (government and nongoverrment) join in a coordi- able over a lifetime. This program allows people to build nated manner. Of the schemes reviewed below, the mutual up credit for health care when they are well, to cushion or funds of Yaounde, Cameroon, and Yoffe, Senegal, are clos- cover the increasing costs of care in old age. est to such traditional arrangements. The German, Japanese, and Korean insurance systems Long Traditions of Risk Sharing originated in small schemes of employed people in the same craft, town, or industry. Coverage grew to the whole pop- The risk-sharing schemes analyzed below share certain char- ulation as these countries moved toward full- or high-employ- acteristics. They involve (with one exception) voluntary mem- ment industrial economies. Employment in agriculture fell bership, prepayment of contributions into an identifiable sharply and its productivity and earnings rose, enabling a fund, some (often loose) notion of entitlement to benefits, larger portion of agricultural workers to organize themselves and a defined set of service providers. To varying degrees into insurance schemes or to buy into industrial insurance they have attracted international interest and support. But funds. Government, at least in Korea, subsidized rural par- they represent only a fraction of risk-sharing experience in ticipation. In recent years a reverse trend has been taking protecting rural populations against the costs of unexpected place in Central and Eastern Europe, where falling employ- bereavement, disability, and illness. Numerous traditional ment and shrinking tax revenue have stymied attempts to structures also spread financial risks among groups or are establish employment-based national health insurance in linked to nonspecific savings schemes. Some schemes allow countries such as Bulgaria, Kazakstan, the Kyrgyz Republic, both nonspecific savings and indemnify members against the Slovak Republic, and Russia. the costs of unpredictable events, such as marriage or death. The oldest documented accounts ofvoluntaryprepayment Typology of (Mainly Rural) Risk-Sharing associations for health care date from the fifth century B.C. Schemes in Greece, where trade and craft groups organized mutual help schemes based on regular pooled savings to protect mem- There are many ways to typologize the various schemes involv- bers in case of death, ilness, or incapacity (WSM 1996). Today ing ruralrisk sharing. Theycanbe defined accordingto whether voluntaryprepaymentschemesexistinmanypartsoftheworld, they are voluntary or mandatory, the kind of benefits pro- sometimes with health care as a specific claim on resources. vided, the degree of interpersonal risk sharing, or the cir- (See Lukholo 1996 and Alain, Tchente, and Guillerme- cumstances that led to their creation, such as government Dieumegard 1991 for other descriptions of such schemes.) collapse, extemal assistance, or local community initiative. RURAL RISK-SHARING STRATEGIES 165 The niain dimension for categorization used here is fund Education Resources and Training (ORT, an international ownership and management-that is, the location of con- NGO), West Bengal's Saheed Shibsankar Sabar Samity trol and decisionmaking regarding the use of resources (SSSS, alocalNGO), andZaire's Bwamandaare examples. and collection of contributions in the scheme (table 2). Fund ownership and management were used to catego- The schemes examined were predominantly insurance rather rize the schemes for several reasons. First, fund ownership than personal prepayment schemes, all but one were vol- often reveals a scheme's initial motivation and objectives, untary, most were rural (though five had an urban focus), which can differ substantially. For example, some facility- and most received external financial or technical support. based schemes are driven primarily by the need to raise Health facility schemes are generally initiated by hospi- revenue. Others may combine revenue-raising goals with tal staff and cover catastrophic hospital care costs. Such efforts to increase service utilization levels. Community- schemes have geographically defined beneficiary groups based schemes, on the other hand, tend to have popula- based on the hospital catchment area, often including both tions' needs as their principal motivation, although concern rural and urban communities. Examples include India's is often focused at one level or on one benefit, such as drugs. Kasturba hospital scheme and Zaire's Masisi hospital scheme. Other schemes have a demonstration motivation, such as Community schemes usually focus on primary care, espe- China's Sichuan Rural Health Insurance experiment. cially drugs, but also may include referral services and often Second, typologizing schemes according to fund own- have a broad community development orientation. Examples ership often reveals a scheme's design details. Design is an include Guinea-Bissau's Abota, Indonesia's Dana Sehat, important factor in a scheme's performance. Several risk- Taiwan's (China) Farmer's health insurance, and Vietnam's sharing schemes have serious design flaws from a sustain- Quang Nam Da Nang (QNDN). ability perspective (see below). Third, ownership may Cooperative schemes are often linked to local labor mar- determine a population's overall trust and confidence in the kets and based on individuals' place of work. Contributions scheme and in the services provided. Finally, ownership may to the health fund may come from the sale of cooperatively be an important factor in determining the way a scheme can produced goods. Examples include China's former coop- be complemented and supported by government, and can erative medical system, India's Mathew Milk Cooperative, affect approaches to health care financing and organization. and Senegal's Mutuelle de Yoffe. Cooperative schemes are A second key dimension for classifying schemes relates to the successor to the craft-related associations that devel- the distinction between schemes that focus on providing oped in ancient Greece, medieval Europe, and seventeenth- coverage for high-cost, low-frequency events and those that century Chile. focus on low-cost, high-frequency events (designated type 1 NGO schemes vary, reflecting the origin and purpose of and type 2 in table 3). There is no easy way to categorize type the sponsoring NGO. Indeed, NGOs may be the motivat- 1 and type 2 schemes, but they represent ends of a spectrum ing force behind facility, community, or cooperative schemes. of different risk-sharing schemes. Schemes such as Bwamanda Ghana's Nkoranza, the Philippines's Organization for (Zaire), Chogoria (Kenya), and Nkoranza (Ghana) are clearly TABLE 2 Rural risk-sharing schemes by owner and region Southeast and Latin America and Owner Africa South Asia East Asia the Caribbean Total Health facility 3 3 0 i 7 Community 4 1 5 1 11 Cooperative I 1 4 1 7 Govemment I 0 4 0 5 NGO 2 3 1 0 6 Total I1 8 1 4 3 36 Note: In some countries there are a large number of similar but slightly different schemes, at operating under the same umbrella name (such as Dana Sehat in Indonesia). Because of the similarities between these schemes, they are cournted only once. INNOVATIONS IN HEALTHI CARE FINANCING 166 type 1 schemes. Type 2 schemes include Dana Sehat (Indonesia) clearly depends on the context, demand for type 1 and and United Mission (Nepal). Some schemes incorporate ele- type 2 schemes appears to be rooted in different contex- ments of both types. For example, several schemes coverboth tual factors. Many type 1 schemes have arisen in circum- high-cost, low-frequency events and low-cost, high-frequency stances where user fees for hospital services were high. In ones, often without setting premiums on an actuarial basis. this context people were interested in buying insurance The distinction between type 1 and type 2 schemes is primarily for its risk-sharing benefits. In contrast, demand important because different conditions are required for for type 2 schemes does not stem from a desire for risk shar- these types to succeed. In particular, type 1 and type 2 ing. Instead these schemes have focused on improving qual- schemes differ in terms of: ity of care, particularly by expanding services to previously * The degree of social cohesiveness and trust required for unserved communities. their effective operation Much of the debate about risk sharing has also focused * The level of demand for insurance on the administrative feasibility of implementing such * Administrative complexity. schemes. Again, there is a critical difference between type A cohesive community is much less of an issue for type 1 and type 2 schemes. Type 1 schemes that aim to cover cer- I schemes than type 2 schemes. Moreover, beneficiaries in tain variable costs require actuarial estimates of premiums type 1 schemes tend to be distributed over a wide area and and defined payment mechanisms (rather than simply sup- to be relatively heterogeneous. Thus it is unlikely that strong plementing budgets) and are more difficult to manage than feelings of solidarity can be established under these schemes. type 2 schemes. Management structures for type 1 schemes Some type 1 schemes appear (mistakenly) to emphasize reflect this difference. They tend to be more complex, and community solidarity. For example, the Nkoranza evalua- the management problems encountered are more substan- tion report states that "the concept of risk sharing in the tial. Thus capable management may be crucial for type 1 community must be well explained for the people to under- schemes, but is probably not required for type 2 schemes. stand that if you insure and do not benefit directly, your A formal model for type 2 schemes has been proposed 'neighbour will benefit from your contribution"' (Somkang by Hsiao and Sen (1995), under the title of cooperative and others 1994, p. 12). health care. Under this model strong social bonds and mutual Such an appeal to social solidarity may be effective at trust are viewed as the defining elements of the commu- the village level but seems unlikely to be successful at the nity that undertakes the financing and provision of health district level. On the whole type 1 schemes are less con- care. Based on rural China and India, services are expected cerned with altruism and community development. This to include preventive care, immunization, family planning, feature and the nature of the benefits they offer mean that maternal and child care, health education, and basic ambu- they are more likely to be more susceptible to adverse selec- latory curative care. Communities are expected to contain tion and moral hazard than type 2 schemes. about 1,000 people and to be capable of paying for a health The level of effective demand for insurance in develop- worker and a stock of generic essential drugs. Amalgamations ing countries has been hotly debated. Although demand of these communities would establish a primary health center for about 15,000 people staffed by a doctor, mid- TABLE 3 wife, clinical nurse, pharmacist, and assistant. A mix of Two types of risk-sharing schemes annual premiums and copayments is envisaged; member- Type I schemes Type 2 schemes ship would idealiy be compulsory, and referrals would be High-cost, low-frequency events Low-cost, high-frequency events covered through packaged fees or capitation. Tend to be hospital owned Tend to be community owned Although nearly all the schemes examined are owned by Tend to cover whole district Tend to be based at the village level a single party, autonomy in fund management varied. Some Use actuarial basis or variable costs Premium set mainly according to for calculating premium ability to pay program, such as the Thai health card, are centrally driven Committed to meeting certain Committed only to raising extra schemes and operate within tight government guidelines. designated costs revenue for services In Indonesia and the Philippines guidelines exist on the use RURAL RISK-SHARING STRATEGIES 167 of fundsbut are not very tight. Some schemes, such as Zaire's management, NGO activity (such as religious missions) or hospital-based schemes, had almost complete autonomy. other forms of technical assistance, and monetization in the There was little discussion, in the documentation of the local economy (such as cash or subsistence crops). The exist- schemes reviewed, on the accountability of fund manage- ing health care system also may lead to risk-sharing mech- ment. Where schemes were community owned, fund man- anisms, as well as limit or define the kind of insurance agers were sometimes held accountable to the local response. For example, the quality and availability of health community through community meetings or simple account- services, particularly government services, might define ing procedures (such as showing receipts to the commu- the need for additional or better service providers. Insurance nity; Chabot, Boal, and Da Silva 1991; Mogedal 1984). maybe seen as a mechanism forimproving or extendingthe Ghana's Nkoranza hospital scheme had an insurance advi- provision of health services. In addition, the cost of exist- sory board that included members of the hospital manage- ing services may enable or. inhibit the development of risk ment team and twenty-five community members (Somkang sharing. Bennett and Ngalande-Banda (1994) observe, in and others 1994). On the whole, however, systems for the context of Sub-Saharan Africa, that user fees for gov- accountability to beneficiary communities appeared weak. errnent services may be a prerequisite for wider popular demand for or recognition of the potential role of insurance. Overview of Schemes The availability, price, and quality of private provision may help determine which services are covered in the ben- This section reviews the thirty-six schemes, focusing on: efits package. In Korea and Thailand the scale of private • The context in which the scheme was developed provision made it clear that rural insurance, like urban insur- * Membership and coverage ance, would have to cover services provided by private prac- * Financing mechanisms titioners, as well as by government clinics and hospitals. * Administration and fund management Thus the configuration, quality, and price of existing health * Provider payment mechanisms services help shape health insurance schemes. The national * Arrangements for health care provision. or local context is important in understanding both the pur- pose and performance of risk-sharing schemes, and in iden- The context tifying the barriers or opportunities to their replicability nationwide or in other countries The risk-sharing schemes compared below are drawn from In nearly all the schemes examined, people were accus- contexts that sometimes differ widely Zaire's Bwamanda tomed to paying fees prior to the introduction of risk shar- scheme and Guinea-Bissau's Abota scheme were developed ing, or did not have any real access to health care (as in in response to the near collapse of govemment-funded health Guinea-Bissau and Lalitpur district, Nepal). Moreover, care. Korea's class II rural schemes, on the other hand, were fees were often so high that a large portion of the popula- initiated at a time of accelerating economic growth and urban- tion could not afford them. This was particularly the case ization, as a way to protect rural residents under an increas- for facility-based schemes; facilities often faced declining inglynational insurance system. China's Cooperative Medical use because of high fees and low revenues. Such arrange- System was developed as communal agricultural production ments led to considerations of insurance. In Zaire, for exam- was emphasized as the main mechanism for economic devel- ple, health zones were meant to be self-supporting; thus, opment. When macroeconomic policy shifted to the "social- other than funds received from external donors, zones had ist market" and central and provincial government subsidies to recover full operating costs. In other African countries were cut and redirected, the system collapsed (Hsiao 1995). (notably Ghana and Kenya) schemes were initiated by mis- The local context may be shaped by the overall economic sion hospitals. Although detailed information on the finances situation and the direction of policy. Other factors that of these hospitals is not available, in general mission hospi- may encourage risk-sharing mechanisms include decen- tals in Africa have been forced to rely on fee income as gov- tralization policy, traditions of community initiative and emment subsidies and external support has dried up (Gilson INNOVATIONS IN HEALTH CARE FINANCING 168 and others 1994). InJapan, Korea, and Taiwan (China) health ment variables (whether occupation, place of work, or how care is dominated by private providers charging high fees. produce is sold). Only one of the schemes reviewed, India's The existence of other types of insurance schemes in a Kasturba Hospital scheme, was targeted at the poor. country did not appear to affect the uptake of rural risk Most schemes use the household as the unit of mem- sharing, but many people were able to understand the notion bership. Many schemes that initially allowed individual enroll- of risk sharing because they were used to traditional mutual ment often faced problems of adverse selection and switched self-help mechanisms (Wong share in Thailand, Gotong to household enrollment. In Nkoranza, Ghana, premiums Royong in Indonesia, Abota in Guinea-Bissau). were set individually but the entire household had to join. Some of the schemes examined were run by NGOs that The failure of insurance scheme workers to sign up all mem- appeared to be quite autonomous from government. In bers of a household, however, contributed to the failure of other instances schemes were initiated in the face of a break- the scheme (Somkang and others 1994). In Taiwan (China) down in government finance for health care (for example, individual enrollment was allowed under both the Farmers in Zaire, Guinea-Bissau, and recently in China). Although Health insurance scheme and labor insurance, since only some schemes (such as that in Guinea-Bissau) operated in workers (not their dependents) could be enrolled. However, small, close-knit communities, others covered large districts the Farmers scheme appears to have been compulsory and with diffuse communities. In Guinea-Bissau the cohesive- the labor insurance scheme probably attracted few self- ness and small size of local villages were seen as factors employed workers. In Vietnam at least two-thirds of the supporting the success of the schemes (Chabot, Boal, and household was required to join (Ensor 1995). Da Silva 1991). However, the importance of this factor Other measures that prevent adverse selection include depends on the ownership of the scheme. Where funds are requiring that a minimum number or portion of households community owned, trust within the community is critical. in a village or administrative area join a scheme. In the Most schemes covering larger areas were facility or gov- Kasturba program at least 75 percent of poor households erminent owned, so accountability within management sys- in a village must join. In Thailand at least 30 percent of tems is more important than trust in the community. households in a village must join in order for the village to Schemes were located both in areas where most bene- participate in the health card scheme. ficiaries were subsistence farmers (as in Nepal and Guinea- If enrollment in a scheme is allowed over a long period Bissau) and in areas where a large number of farmers were and there is no waiting period, then people tend to enroll organized in cooperatives. Cooperative and mutual schemes when they need care. At the VH-S hospital in Madras, India, develop where labor is more organized. where enrollment was allowed throughout the year with- Economic growth often supports the growth of risk shar- out a waiting period, less than a quarter of subscribers had ing. In East and South-East Asia rapid expansion of health renewed their membership. The remaining three-quarters insurance in the nonformal sector coincided with rapid eco- probably joined when they got sick, eroding the insurance nomic growth. Even in Bwamanda, Zaire, one of the fac- effect of the scheme (Dave and Berman 1990). Several tors contributing to the success of the scheme was a buoyant schemes that planned to have a limited enrollment period local economy, attributable to a donor-funded rural devel- later extended it because enrollment rates were low. Users opment project that established the health insurance scheme who enroll only when they get sick create big problems in (Moens 1990). hospital-based schemes, where the need for service is most unpredictable and bears large financial consequences. For Membership and coverage community schemes covering mainly primary care it is eas- ier for households to predict utilization and decide whether Membership. Potential beneficiaries of schemes are it is worth joining. defined both by geographic location (particularly catchment areas for hospital-based schemes and village, ward, or dis- Coverage. Except for the schemes in China, coverage of trict of residence for community schemes) and by employ- the target population tends to be low (table 4). Zaire's RURAL RISK-SIIARING STRATEGIES 169 TABLE 4 There were exceptions to this general pattern. Two hos- Coverage of schemes pital-based schemes in India (Kasturba and VHS) and one Number of Coverage of scheme in Bangladesh set premiums on a sliding scale accord- Scheme Country people covered target population ing to income. More sophisticated schemes in Japan and Facility-owned Taiwan (China) generally set premiums as a percentage of Chogoria Kenya 9,000 17% earnings. In Korea premiums were based on a complex CIMIGen Mexico 600 n.a. Katurba India 4,390 n.a. assessment of income and assets. Masisi Zaire 3,500 n.a. Some villages in Guinea-Bissau, the scheme in Lalitpur, Raigarh India 75,000 n.a. Nepal, and the Kasturba hospital scheme in India allow pay- VHS India 3,800 n.a. ment in kind. Interestingly, very few poor agricultural com- Community-owned Abota Guinea-Bissau 200,000 n.a. munities in Nepal chose to pay in kind (Donaldson 1982). Assaba Guatemala 65,000 40 families The Dana Sehat scheme in Indonesia and the ORT Boboye Niger 250,000 i00% (compulsory) scheme in the Phiippines were the only insurance schemes Dena Sehat Indonesia 12,000,000 13 villages Healh card Thailand 2,700,000 6.5% that allowed payment of premiums more than once a year. Lalitpur Nepal 6,000 5-10% Still, monthly payments create difficulties for workers with Rural Cooperative China s46,000 31-100% Medical System project seasonal income. Ile ORT scheme adopted a flexible pay- Sichuan rural health China 40,443 >90% ment schedule because it was felt that many households insurance would not be able to afford the annual premium in one lump Govemment-owned sum. A number of families, however, dropped out of the Muyinga Burundi n.a. 23 households QNDN vietnam 10,000 n.a. scheme because they failed to keep up payments (Ron and Cooperative-owned Kupferman 1996). Cooperative Medical China About -100% Few schemes had built-in exemption policies. In Boboye, Scheme 700,000,000 Mallur Milk India 5,000 n.a. Niger, the indigentcould get specialwaivers (Diop,Yazbeck, Mutuelle Famille Cameroon 455 22% and Bitran 1995). In Guinea-Bissau's Abota scheme vil- Babouantou de lagers could choose to give the indigent access to drugs Yaounde Yoff Senegal 150 families n.a. despite the fact that they had not paid (Chabot, Boal, and NGO-owned Da Silva 1991). In the ORT scheme project staff tried to Bwamanda Zaire 80,000 66% seek supplementary funds to subsidize the premiums of G.Kendra Bangladesh 20,000 n.a. poor families (Ron and Kupferman 1996). InLalitpur, Nepal, Goalpara India 1,000 n.a. Nkoranza Ghana 37,000 32% poor households could get a free health card if they had a ORT Philippines 1,300 n.a. letter from a community leader (Donaldson 1982). ssss India 6,800 n.a. Elsewhere people who could not afford premiums would n.a. is not available. simply pay user fees when using the service or not seek care. Bwamanda hospital scheme stands out as having strikingly East and South-East Asian economies provide cover- high coverage relative to other schemes. Many commu- age for the poorby issuing special low-income cards ordevel- nity-owned schemes (even those that were successful, such oping schemes specifically for the poor, rather than as Thailand's health card scheme) failed to cover more integrating them into the main health insurance system. In than 10 percent of the target population. Korea, however, the poor are theoretically integrated into the system through subsidized premiums. Financing-premiums, copayments, and cost recovery All the schemes relied on funds other than those received from premiums (table 5). Again, Zaire's Bwamanda scheme Premiums were generally flat-rate premiums, paid on an stands out, with a cost recovery ratio of about 80 percent. annual basis. In-kind payments generally were not accepted. Cost recovery ratios are much lower in other schemes. In Few schemes took proceeds from cooperative sales. Thailand about 35 percent of recurrent costs are covered. INNOVATIONS IN HEALTH CARE FINANCING 170 However, this needs to be placed in the context of high gov- of recurrent costs, and the same level of revenue would oth- ernment cost recovery (on the order of 50-60 percent of erwise be collected through user fees. recurrent costs at district hospitals). Thus less cost recov- In some cases, however, management structures were ery is achieved under the health card scheme than under very simple. In Guinea-Bissau one of the main reasons for the user fee system. implementing a prepayment (as opposed to a user fee) sys- Guinea-Bissau's Abota scheme is generally perceived to tem was its administrative simplicity A village leader would be successful. For many years community health insurance simplyvisit each household once a year and ask for payment was the sole source of finance for drugs in many commu- of a fixed amount. After an initial learning period, villagers nities. It should be emphasized, however, that the drugs managed the scheme well, alhough increasing economic stocked at the community level were very basic (just twelve pressures eventually led to misuse of funds that threatened essential drugs), and that drugs were sold to health posts the credibility of the scheme. at donor-subsidized prices. About 90 percent of the cost of In Nepal's United Mission scheme village committees essential drugs was covered by donors (Eklund and Stavem made some very good decisions, particularly relating to 1996). exemptions, butwere poor at accounting for funds and com- As noted, the VHS scheme in Madras, India, charged municating their plans to the rest of the village. premiums on a sliding scale, which led to very different rates of cost recovery. Predictably, most of the people (74 per- Fund management. Few details were available about man- cent) that joined the scheme were uninsured and seeking agement structure or management capacity for government- care at the health center level. Health centers record all run, mutual, and cooperative schemes. However, more insured persons on separate registers and at the end of formal administrative structures are likely to be in place for each month present claims to the fund holders, who pay mutual and cooperative funds, which may facilitate them. This procedure covers no more than about 3 percent management. TABLE S Cost recovery under the schemes Scheme Costs covered Cost recovery ratio Abota, Guinea-Bissau Full cost recovery on drugs n.a. Boboye, Niger Not specified 149% of drug costs Bwamanda, Zaire Aiming to cover operating costs (that is, excluding capital 80% of operating costs (from patient revenue, including and staff benefits in kind) 20% copayments) CAM, Burundi Not specified (funds go to local govemment) 34% of drug costs for card holders Farmers health insurance, Costs fully covered by premiums, government Copayment is 10-50% of fee (govemment pays 70% of Taiwan (China) contributions, and copayments premium, beneficiary pays 30%) Health card scheme, Thailand Not specified Estimates differ; about 35% of recurrent costs Kasturba, India Drug costs, village health worker remuneration, n.a. and travel costs of hospital team Health insurance for informal Costs fully covered by premiums, govemment Copayment is 40-60% of fee (govemment pays 50% of sector, Korea contributions, and copayments premium, beneficiary pays 50%) Lalitpur, Nepal Aimed to cover any health post expenditure on drugs Less than 10% of health post recurrent costs over Rs 8,500 a year (the standard govemment drugs budget) ORT scheme, Philippines Self-supporting except ORT finance of one doctor n.a. and two nurses VHS, India Unclear Range from 1I.1% of recurrent costs for low-income to 59.9% for high-income subscribers n.a. is not available. RURAL RISK-SHARING STRATEGIES 171 Where premiums are collected at one point in time and trictwide meetings, campaigns, and information sheets. must meet financial commitments for an entire year, it is Under the ORT scheme in the Philippines a registration essential (particularly in high-inflation environments) that campaign was held. In Chogoria (Kenya) a major market- the funds are invested. In its first year of operation Ghana's ing campaign was planned to launch the new scheme Nkoranza scheme ran into difficultybecause ithad no invest- (McFarlane 1996). ment policy and high inflation rates rapidly eroded the value of the fund. In later years, however, the scheme bought trea- Provider payment mechonisms sury bonds. Several innovative ways to combat inflation have been found. In Zaire's Masisi scheme funds were held by All the hospital-based schemes paid the hospital on a case- the district pharmacy, which immediately converted rev- based or fee-for-service basis. For most of the primary care enues into drugs. In Zaire's Bwamanda scheme funds were schemes all funds collected were allocated to the nearest capitalized by the NGO. This problem does not occur in provider on a lump-sum basis. The Thai scheme is an inter- more stable economic environments, particularly where esting exception. The scheme provides access to all levels investment opportunities are safe and accessible (Japan, of care if patients are referred. Initially, therefore, there Korea, and Taiwan, China). Under the Thai scheme fund was a fixed formula for allocating funds between different holders did not have to pay providers until the end of the levels of the system. For example, in Chiang Mai 15 per- year and could invest the funds during the year. Investments cent of funds were ultimately retained by the village com- often took the form of interest-bearing loans to commu- mittee, 20 percent by the health center, 33 percent by the nity members (Myers 1989; Supachutikul 1996). community hospital, and 32 percent by the provincial hos- Few of the studies reviewed provided much informa- pital (Supachutikul and Sirinirund 1993). These allocations tion on management information systems. In general varied somewhat between provinces. Since there were strict schemes focused first on developing adequate financial man- referral procedures in order to be eligible for fund cover- agement systems in order to account for finances and ensure age, these fixed proportions could be estimated to some that only insured people could access benefits. Protecting degree. However, under this payment system higher service the scheme against fraudulent claims often proved diffi- levels, particularly district hospitals, often felt that they cult because in many contexts it was difficult check the iden- received an unfair share of funds. In more recent years the tity of the person seeking care (McFarlane 1996; Somkang referral procedures have become more lax and there is and others 1994). now discussion about linking allocation to actual utiliza- After the basic systems were in place, they might expand tion of facilities. In order to institute such a payment mech- to cover utilization, which is useful in setting premiums. Only anism, however, stronger information systems are required. after these mechanisms were operational did attention turn to more complex issues, such as monitoring quality of care. Provision of health care Still, only the most developed of the schemes examined, such as the one in Korea, had information on these aspects. Benefts package. On the whole, benefits packages were Some community-owned schemes (the Thai health card poorly defined. Some schemes had exclusions (for exam- scheme, the prepayment scheme in Boboye, Niger) had infor- ple, sexually transmitted diseases in Vietnam, dental ser- mation on quality from special evaluations, but not from rou- vices in the Philippines's ORT scheme), but otherwise tine data. In Taiwan (China), under the Farmers Health schemes tended to cover all the services available at the par- Insurance scheme, routine patient data now include fees ticipating facilities. The main problem with this approach charged by category, diagnosis, surgery, and length of stay. was high enrollment rates among people with preexisting Marketing and information and education strategies var- conditions, particularly chronic illnesses. Kenya's Chogoria ied immensely, depending partly on a scheme's catchment hospital scheme initially had a very broad benefit package area. In Nkoranza substantial efforts were devoted to mar- covering all such conditions. But when the scheme was keting and information and education, with a series of dis- recently reviewed the benefits package was defined much INNOVATIONS IN HEALTIH CARE FINANCING 172 more tightly in order to exclude both the elderly and those with preexisting conditions (box 1). Very few of the schemes Box I Exclusions and limits under Kenya's used revenue to provide non-personal services. Chogora Hospital Health Insuance Scheme * Diseases that were diagnosed prior to joining the scheme Integration ofhealth care services. Hospital-based schemes or within the waiting period are not covered. tended to focus exclusively on the hospital level and have * Coverage is discontinued when a person turns 65. limited connections with primary care. There were, how- * Treatment of patients with AIDS is provided up to a maxi- ever, some exceptions. The Chogoria scheme covered pri- mum of Sh 36.00 a year. every providers andused Themas goatekeepers tovered hospial * Treatment of psychiatric illness is limited to Sh 68.00 per mary providers and used them as gatekeepers to the hospital policy per year. level. In Zaire's Bwamanda scheme primary clinics acted * Expenses that are associated with normal or abnormal preg- as gatekeepers to the hospital scheme, but fees for services nancy are not covered (although the operation fee for a first at this level were not covered by the scheme (Moens 1990). Caesarian section is covered). Community-owned schemes tended to have clearer and * Reading glasses, eye and ear tests, and hearing aids are not stronger referral structures. In Taiwan (China) and Korea covered. * Self-inflicted injuries are not covered. providers are predominantly private and referral systems * Birth defects and cosmetic surgery are not covered. are extremely weak (if they exist at all). The lack of a gate- * Dental procedures are not covered. keeper has contributed to rapid cost escalation in these * Medical examinations are not covered. economies. * Procedures carried out for nonmedical reasons are not cov- ered (circumcision, for example). Quality of care. Several community-owned and NGO Source: McFarlane 1996. schemes used revenues to expand access to health ser- vices, but few made efforts to improve other aspects of qual- Efficiency ity of care (the exceptions being Niger's Boboye, Vietnam's QNDN, and Mexico's prepayment scheme). None of the Administrative efficiency. In some cases (such as Guinea- facility-owned schemes had explicit links between the intro- Bissau's Abota) the prepayment route was taken because duction of the scheme and attempts to improve quality of it was seen as being administratively more efficient than care. user fees. In other instances (such as Ghana's Nkoranza) Under Vietnam's QNDN scheme special efforts were there appear to be high administrative costs (particularly made to improve the quality of care because there were the use of time of scarce skilled personnel), yet substantial (officially) no fees at hospitals. Thus special efforts had to problems remain in the administration of the scheme. be made to attract people to the scheme. In both Mexico and Vietnam quality improvements focused on the hotel Allocative and technical efficiency. Fee-for-service reim- aspects of care. bursement has a number of well-known shortcomings. It Few schemes adopted special pharmaceutical policies, provides little incentive for efficiency on behalf of the hos- alhough the ORT scheme in the Philippines managed to pital. It does not guard against problems of cost escalation. negotiate favorable prices for essential drugs purchased And it is administratively complex. Fee-for-service payment from local suppliers. gives providers incentives to overservice and overprescribe. In most of the cases examined overprovision was unlikely Assessment of Performance to be an issue because the provider was a government or mission facility, and staff were paid on a salary basis. However, Health status improvement it was a concern in Masisi, Zaire, where part of hospital revenue was used as incentive payments for doctors No studies have been made evaluating the impact of these (Noterman and others 1996). At Chogoria hospital the med- schemes on health outcomes. ical officer in charge noted that "it has taken some time to RURAL RISK-SHARING STRATEGIES 173 educate our prescribers to treat patients on the scheme in the insurance schemes examined set premiums on a com- a similar manner to other patients, keeping in view the cost munity-rated basis and thus entailed a subsidy from the of treatment" (McFarlane 1996, p. 7). healthy to the sick. Few schemes, however, adopted sliding All the schemes examined were fairly weak purchasing scales, and the use of flat-rate premiums implies regressiv- agents; few defined cost-effective packages of care, few ity in financing. A sliding scale system in Bwamanda allowed implemented strong referral and utilization control systems lower copayments for more distant residents but was aban- to optimize efficient use of different levels of the health care doned because of the extra administrative cost and because system, and few implemented a management information it did not appear to affect utilization. It did, however, improve system that monitored cost-effectiveness or appropriate- enrollment and thus would have effectively reduced adverse ness of care delivered. selection. Not only were the schemes weak purchasers, they also More important is the question of whether insurance is sometimes introduced inefficiencies into the system. For more or less regressive than other financing alternatives. example, many hospital-based schemes largely ignored pri- Although flat-rate premiums are likely to be less regressive mary care. Although the documentation does not explore than user fees, they may be more regressive than general the impact of this, it would seem likely to result in under- tax revenue financing. Empirical work in a number of indus- utilization of health centers (leading to facilities operating trial countries has established that social insurance is more at low capacity and rising unit costs) while patients are treated regressive. less efficiently at the hospital level. Again, Zaire's Bwamanda Affordability is probably the key issue in terms of equity. scheme differs, because access under insurance to the hos- Few schemes made special allowances for people who could pital requires a referral from a health center. not afford to pay the premiums. In most cases people who The hospital-based schemes in Nkoranza, Ghana and could not afford premiums were required to pay user fees Masisi and Bwamanda, Zaire, experienced rapid cost esca- instead. Thus the effectiveness of insurance in protecting lation, at least in their early years. In Korea and Taiwan the poor raises the question of the effectiveness of exemp- (China), where fee-for-service payments also exist, such tion mechanisms. problems have persisted despite efforts to contain costs Several schemes that examined the issue of affordabil- through copayments. The lack of a gatekeeper has also con- ity acknowledged that it could be a problem, although the tributed to rapid cost escalation in these economies. evidence was not always clear cut. For moderate to large Presumably, if funds are held at the community level, low-income households in Nkoranza, Ghana, the estimated then it is in the interest of the fund holder to make sure cost of premiums amounted to 5-10 percent of the annual that unnecessary utilization of expensive secondary-level household budget, which may well constitute a financial services does not occur. barrier to membership (Somkang and others 1994). In Muyinga, Burundi, 27 percent of respondents to a house- Financial efficiency. Overtime most of the schemes were hold survey stated that financial inability to purchase a able to devise ways to invest the revenue raised from the card was one of the main reasons they did not participate insurance scheme. However, scheme managers had not in the scheme (Arhin 1994). In Mexico about 20 percent always thought out this aspect adequately prior to imple- of enrollees in the perinatal prepayment scheme dropped mentation, and heavy financial losses in the first year could out, and this was mainly attributed to financial inability to adversely affect the financial efficiency of any scheme for keep up payments (Ensor 1995). a long period. In utilization. Only the pilot project in Boboye, Niger, Equity analyzed how utilization patterns varied by income group. In Boboye it was found that utilization rates among the poor In financing. Risk sharing has been promoted as a means had gone up since implementation of the scheme. Moreover, of encouraging more equitable financing of health care. All when payments by the poor who used government facili- INNOVATIONS IN HEALTH CARE FINANCING 174 ties were compared before and after the scheme, total pay- to be a "market response" to the establishment of insurance ments had gone down (Diop, Yazbeck, and Bitran 1995). in remote areas. In Korea, however, where private providers The scheme in Boboye was, however, very different from were the main mode of service delivery, the establishment those implemented elsewhere. And although the Boboye of insurance schemes was found to be insufficient to encour- experience suggests that it is possible to design an insur- age providers to relocate to rural areas. The government ance scheme that has positive equity effects in terms of had to initiate separate programs to finance remote health who benefits from the service, it by no means demonstrates care centers, subsidize insurance societies in rural areas, and that this is always the case. The compulsory (tax-based) use tax incentives to encourage the development of clinics system in Boboye performed better than a user fee exper- and hospitals in these areas (Peabody, Lee, and Bickel 1995). iment in a comparison district. Finally, exclusions and limits on the benefits of schemes Many of the case studies discussed utilization and enroll- have implications for equity in the utilization of services. ment patterns by geographical location of the household. In schemes that cover catastrophic costs of care, setting cer- There is fairly substantial evidence that utilization increases tain exclusions may be essential to guard against adverse far more among insured households located near a health selection. However, these exdusions are likely to affect more care facility and that these households are also more likely vulnerable groups, such as the elderly and people with AIDS to join such a scheme (Criel 1992; Donaldson 1982; (see box 1). Noterman and others 1996). The private costs of seeking care form a considerable barrier to accessing care. Under Equity between schemes. In the few countries where there most schemes people pay the same premium wherever they is substantial coverage by rural risk-sharing schemes (for live; thus those located far from the facility (who might example, in China prior to the breakdown of the Cooperative be part of poorer, more remote rural communities) end Medical System and in Thailand) equity between schemes up cross-subsidizing those who live close. This situation has become a major issue. In China the type of care to which might also lead to a form of adverse selection in which people had access varied substantially according to the wealth remote households drop out of the scheme because the of the community; poor communities could often only afford premiums are so high that it is not worthwhile for them to cover primary care services and did not cover inpatient to join. This can lead to a vicious circle in which average services at county hospitals (World Bank 1996). utilization rates rise even higher and more people drop Korea now has more than 600 insurance funds. There is out. substantial government subsidy and regulation of the funds The only documented attempt to implement a sliding but it is not clear that this intervention promotes equity scale based on geographical proximity to the facility took between schemes. The government regulates reimburse- place in Bwamanda, where for one year the level of copay- ment rates and provides a 50 percent subsidy to premiums ment was based on a sliding scale depending on distance charged. However, premiums are set by the individual insur- from the hospital (a 20 percent copayment was levied for ance society, which implies that schemes covering more the nearest group, down to 5 percent for the most remote affluent groups can buy bigger and better benefit pack- group; Criel 1992). Under this system enrollment went up ages, and that the government's subsidy of this benefit pack- among the remote group but their utilization of the hos- age will be higher in absolute terms (Peabody, Lee, and pital did not. Furthermore, the implementation team Bickel 1995; Yu and Anderson 1992). thought that enrollment among this group might have gone Few of the countries examined provided evidence on up in any case because the scheme became better known. equity issues between formal and informal schemes. In The sliding scale was dropped because it appeared to have Japan, Korea, and Taiwan (China) the schemes that cover no impact on utilization and was administratively more the informal sector are now fully integrated with the national complex. health insurance scheme. In many other countries formal Most of the schemes in Sub-SaharanAfrica allowed access schemes are nonexistent or very limited. Thailand is an only to public health care facilities. Thus there is unlikely exception, and substantial questions are being posed about RURAL RISK-SHARING STRATEGIES 175 the different benefits and government subsidies provided Schemes that allowed individual membership often faced to participants in formal and informal schemes (Khoman problems of adverse selection. In Nkoranza, Ghana, pre- in this volume). In 1994 it was estimated that total expen- miums were set on an individual basis but the entire house- diture per capita under the Civil Servants Medical Benefit hold had to join. However, the failure of insurance scheme Scheme was 9.4 times higher than that under the Health workers to sign up all members of a household contributed Card Scheme, and expenditure under the Social Security to the failure of the scheme (Somkang and others 1994). Scheme was 3.7 times that under the Health Card Scheme. Further prevention measures against adverse selection Differences are even more marked if the level of govern- include requiring that a minimum number or portion of ment subsidy to the schemes is considered. Government households in a village or administrative area join a scheme. subsidy per capita to the Civil Servants Medical Benefit If enrollment in a scheme is allowed over a long period Scheme and the Social Security Scheme was 27.0 times and and there is no waiting period, then people tend to enroll 4.4 times that to the Health Card Scheme (Supachutikul when they need care. This is a big problem in hospital-based 1996). schemes, where the need for service is most unpredictable and bears higher financial consequences. For community Consumer satisfaction schemes covering mainly primary care it is easier for house- holds to predict utilization and decide whether it is worth- Most of the schemes paid little attention to consumer sat- while to join. isfaction, or even to what consumers wanted, during the The main problem with benefit packages that do not design phase of the scheme. None of the studies reported exclude preexisting conditions was high enrollment rates surveys of consumer satisfaction, and few had carried out among those with preexisting conditions, particularly chronic marketing surveys prior to implementation. As judged by illnesses. demand for the schemes, consumer satisfaction was often Two of the schemes, the ORT scheme in the Philippines low. and the RCMS scheme in China, claim to fully recover costs, but they did not set premiums on an actuarial basis. Sustainability and replicability Presumably in these cases input levels or copayments were adjusted to reflect the revenue received. The advantages The evidence from these experience suggests that there of this approach over a traditional actuarial approach are are several threats to the scope for raising revenue through not clear. rural risk-sharing schemes: With one exception, the schemes examined were vol- * The small scale of the majority of the schemes exam- untary schemes. Many of the problems (particularly adverse ined selection) associated with the schemes stemmed from their * Adverse selection, leading to progressively smaller risk voluntary nature. It has been argued that in developing coun- pools and higher costs try contexts mandatory schemes for informal sector work- * Heavy administrative structures and costs in some ers are unlikely to be feasible because there is insufficient schemes. knowledge about the number and location of rural house- These constraints often led to low levels of cost recov- holds. Identification, income assessment, and contribu- ery. Furthermore, schemes often received substantial input, tion collection can rapidly become an expensive process in particularly technical input, from donors and expatriates, rural areas. However, authorities in Boboye, Niger, man- which suggests that they might otherwise be unsustain- aged to implement a mandatory insurance scheme through able. Most of the schemes examined were relatively short an earmarked tax. More investigation of the prospects for lived or, if still operating, were recently initiated. The main implementing mandatory schemes in developing countries exceptions are some of the Indian NGO schemes and those is needed. Clearly, mandatory forms of risk sharing will be schemes in East Asia that have become part of government easier to implement in areas where local government taxa- national health insurance schemes. tion systems are extensive and well developed. INNOVATIONS IN HEALTH CARE FINANCING 176 The inclusion criteria for schemes in this review effec- * The minimum action that should be taken is making tively excluded many of the better-known Latin American the household or even the village the unit of member- schemes. For example, the Igualas scheme in the Dominican ship and implementing this policy strictly. Republic was excluded because the insurance carrier (or * Making the scheme compulsory (although this does not health maintenance organization) is essentially a private seem to be feasible in many contexts). for-profit business (La Forgia 1990). Ecuador and Mexico * Stipulating that a certain portion of households in the both have schemes for informal sector workers that are village must join the scheme before the village is allowed subsidized entirely by premiums paid by formal sector to enter the scheme (as is done under Thailand's health workers (DeRoeck and others 1996). It would be inter- care project and India's Kasturba hospital scheme). esting to explore whether special conditions in Latin * Preventing people with preexisting conditions from reg- America made the typical risk-sharing scheme discussed istering, or limiting the benefits available for such con- here inappropriate. ditions (although the advantages of such measures need to be weighed against the equity implications). Lessons from Rural Risk-Sharing Experience If enrollment throughout the year is allowed, establish a wait- ingperiod before services can be accessed. Households should Context not be able to join the scheme when they get ill and decide to seek care. Although most schemes recognized this in their The first lesson is that context matters. Some schemes initial design, in many cases few people were enrolled at have come into existence in response to economic and polit- the end of the registration period and the period was ical crisis (Bwamanda, Zaire). The success of others has extended in the hope that more people would join. been facilitated by economic factors (Korea) or terminated Administrative capacity is required to manage required wait- by them (China). The external environment has an influ- ing periods. ence on whether and what type of risk-sharing initiative might be propitious, and on the replicability within a coun- Support the referral system and define a primary point of try or transferability between countries of such experience. contact. Many schemes, particularly hospital-owned schemes, A consideration of context provides a wider opportunity to have paid little attention to the effect the scheme has on review how the overall risk-sharing function in health is other levels of the health care system. Such a segmented being implemented, including the use of tax-based health approach adversely affects not only the providers excluded expenditure by government and other public and private from the scheme but also the providers in the scheme. schemes. Referral systems exist to ensure that patients are treated in the most appropriate and cost-effective manner and to Design protect the financial viability of hospitals. Insurance schemes cannot afford to ignore these issues. Design is critical. It is tempting to conclude that the expe- rience with rural risk sharing has been a litany of disasters. Develop an investment strategy forfunds prior to receiving Many of the schemes examined here had fundamental design them. In high-inflation environments delays of even a few flaws. Although many of these flaws appear obvious, they months can quickly deplete insurance funds. An investment clearly were not to the people designing and implementing strategy is essential to guard against erosion of funds. the schemes. Thus a number of points need to be stated clearly and disseminated widely. Becoming an active purchaser Protect against adverse selection. Ways to ensure against The schemes examined have largely failed to: adverse selection include: * Negotiate special prices with providers RURAL RISK-SHARING STRATEGIES 177 * Define benefit packages to ensure delivery of only cost- Risk sharing and the rural poor effective services * Monitor the quality and appropriateness of care Most risk-sharing schemes appear not to be targeted at the * Use payment mechanisms to encourage efficient, qual- rural poor, but at the rural middle classes: ity service * They seldom allow payment in kind * Develop strong essential drugs policies. * They have flat-rate premiums Recent health sector reform literature from both indus- * They have no exemption policy trial and developing countries emphasizes the importance * They may require substantial copayments. of informed purchasers in the health care sector (Saltman Geographical and income inequities in financing could 1995). But most rural insurance funds remain pure finan- be reduced through the use of a sliding scale. In principle cial intermediaries-that is, collectors of contributions and it should be easier to implement a sliding scale for payment payers to a single provider. Considerable scope exists for of an annual premium than for multiple user fee-type the development of a more active role, including using the charges. However, the user fee literature has highlighted payment system to change incentives, to include new the problems involved in successfully targeting exemptions providers (such as primary care or NGO providers), and to based on income (Parker and Knippenberg 199 1;Willis and reward quality improvement. Risk-sharing schemes should Leighton 1995). Sliding scales based on geographical loca- not be seen simply as a source of finance, but rather as tion would be much less susceptible to targeting errors, ways to organize health services financing and delivery. and appear to be worth trying. The potential that risk-sharing mechanisms have for improv- ing system performance is often untapped. Questioning focility-owned schemes Listening to consumers About 30 percent of the schemes identified were owned and initiated by health facilities, mainly hospitals. Hospital A substantial amount of the literature on rural insurance management often has a strong incentive to implement focuses on whether demand exists among informal sector insurance; if financial support from government is limited workers and rural people for health insurance. In cases where and the population cannot afford to pay cost-recovering health facilities are charging substantial amounts for care it fees, then health insurance schemes may appear to solve would seem likely that demand for health insurance does many problems. However, hospital managers have few incen- exist. Yet few schemes have rooted their design in a survey tives to design and implement health insurance schemes of consumer demand or have evaluated whether schemes that protect the interests of beneficiaries in the most cost- match people's expectations. effective way, or that cover the entire population. Marketing efforts are likely to be weak in this context. The review of facility-owned schemes suggests that, with Too many project documents seem to assume that market- the exception of Bwamanda, hospital-owned schemes: ing techniques explaining the principles of health insur- * Have little incentive to improve the quality of care ance will convince people to join the schemes. But consumers * Tend to overlook primary care really need to be consulted during the design phase. More * Tend to seek overly favorable remuneration (particularly important, they need to have confidence that the scheme through retrospective fee-for-service payment) is managed in their interest, and that it will ensure they have * Have few incentives to improve efficiency. access to quality services when they need them. Confidence Even if a scheme is designed principally to cover the in a risk-sharing mechanism is an abstract notion: people's costs of hospital care, it may be preferable that responsi- assessment of the competence, quality, and professionalism bility for managing the scheme and ownership of the insur- of a health care provider is usually accurate. If the right ance fund rest outside of the hospital, or that the fund type of care is not available, insurance initiatives cannot itself own or be responsible for managing all district health expect to succeed. services. INNOVATIONS IN HEALTH CARE FINANCING 178 The role of govemment Government general revenue financing can be used to solve some of the problems associated with rural insur- A number of important roles emerge for government in ance. For example, it can be used to purchase health ser- the development of rural risk-sharing schemes. vices for the poor orto offset regional inequities. Government capacity to do this depends on whether it is subsidizing 15 Financing. Policymakers should recognize that the rev- percent or 65 percent of the population. enue-raising potential of rural risk-sharing schemes, par- ticularly in very poor countries, is likely to be limited. Thus Policy framework and operational guidelines. By their nature they should not set ambitious cost recovery targets under most insurance schemes are independent local efforts. such schemes. Some of the literature appears overly ambi- However, in countries'that have had the most success in tious in terms of the potential revenue gains from health increasing rural insurance coverage (China, Indonesia, Korea, insurance. For example, Shaw and Griffin (1995, p. 55) Thailand) the schemes have taken place within clearly defined claim that "health insurance is virtually the only practical policy frameworks and have often benefited from specific instrument governments can use to get out of the expen- operational guidelines (box 2). sive business of providing across-the-board subsidies for The effectiveness ofnational guidelines in aiding the devel- hospital care." And De Ferranti (1985, p. 41-2) says that opment of insurance depends on how sensible the guidelines "there has been recognition of the high cost recovery poten- are. In the Philippines pilot projects have been used to help tial of such schemes, since relativelymodest coverage charges, develop national policy and legislation on rural insurance. when spread across an entire participant population, can raise substantial revenue." Training. Many countries with dearly defined policies The evidence from most of the schemes examined, indud- and operational guidelines have also developed training ing those in affluent economies such as Korea and Taiwan packages to help community members manage insurance (China), suggests that cost recoverylevels under rural insur- schemes. Such training packages help ensure that lessons ance are likely to be limited in most developing countries. learned by one community are passed on to others. Training Like user fees, insurance should be seen as a way to top up packages should cover, among other things, the design rec- government budgetary funding and to introduce or ommendations described above. strengthen management of the health system. In countries where rural schemes are widespread and have been inte- Ensuring accountability of fund holders. With increased grated with national health insurance programs, there are decentralization, particularlythe establishment of autonomous substantial government contributions to the scheme, and health care facilities, there are critical questions about to there are often also substantial copayments. whomhealth insurance fund managers are responsible. Where Government subsidy can be made directly to the provider funds are owned by government, cooperatives, or commu- (for example, in most community-owned schemes the gov- nities these issues may be less pressing, but for funds owned emnment continues to fund the bulk of service providers' recur- by NGOs and facilities the lines of accountability to benefi- rent costs, and revenue from the insurance scheme provides ciaries maybe extremely weak. Government must ensure that a top up) or be directed to the insurance fund itself (as in fund managers are accountable. In few of the cases exam- Japan, Korea, and Taiwan, China, and is proposed in Thailand). ined was this responsibility carried out effectively. The preferred govemment strategy depends principally on the ownership of provider units. In instances where the Conclusion private sector is dominant, the government subsidizes the fund or poor households buy into the fund. But by con- Debate about the potential for rural insurance has provoked tributing directly to the fund, rather than to the provider, quite extreme positions. The diversity of schemes found in governments can help develop effective purchasing power the literature means that many of these positions can be and strengthen fund management. supported. While some schemes have operated with com- RuRAL RISK-SHARING STRATEGIES 179 service delivery In the Bwamanda (Zaire) and CAM (Burundi) Box 2 Guidelines for Indonesia's Dana Sehat schemes, prepayment, subsidy, and fee-for-service elements community insurance scheme coexist. Neither user fees nor voluntary prepayment strate- * The scheme is run by the community for the health benefit gies deal adequately with the needs of the poorest people. of its members. Local institutions (family welfare agencies, Although some rural risk-sharing schemes have been suc- village cooperatives, religious organizations) can applyto man- cessful, several common failings are evident: age the Dana Sehat. * Schemes in low-income countries have generally achieved * Premium payments are supported by local economic activ- limited population coverage ities such as cooperatives of crops, handicrafts, moneylend- ing an soon * With few exceptions, cost recovery rates under the ing, and so on. * Vision, mission, objectives, and program identification should schemes are low be based on deliberation and agreement among commu- * The schemes examined have a limited ability to protect nity members. the poorest members of society. * The scheme is controlled primarily by its members. The gov- These criticisms are damning since they strike at the most emment provides tools and guidelines on how its should oper- central reasons for promoting insurance-that is, the sup- ate; community members should monitor these procedures. * Three different levels of Dana Sehatdevelopment have been posed ability of such schemes to raise significant amounts identified, rising from simple community management of of revenue in an equitable manner. small-scale schemes to large-scale, complex, professionally Many of the schemes examined suffered from poor managed schemes. design. It could be argued that with better design, some of Source: Suwandono, Brahim, and Malik 1995. the core problems identified above could be resolved. I Certainly, widely disseminating the lessons of experience may alleviate some of the problems that have been described. plex administrative structures, others have had very simple But many of the schemes reviewed had extensive external ones. While some schemes have had substantial problems technical assistance from well-informed experts. It would with adverse selection, others have avoided such problems be impossible to widely replicate health insurance schemes almost completely. While some schemes have devised incen- for the informal sector and to provide such intensive tech- tivesto promote efficient use of the health care system, oth- nical support. The few success stories, such as Zaire's ers have probably increased inefficiencies in the system. To Bwamanda scheme and Nepal's Boboye scheme, demon- engage in further discussion about the potential for rural strate that it is possible to design and operate a successful insurance we need terms that distinguish more carefully health insurance scheme for the informal sector. They do among different types of schemes and their objectives. not, however, address the question of whether it is feasible The framework presented here is a first step along this path. to do so on a widespread basis. The evidence suggests that Well-designed insurance schemes may have even greater it is probably not feasible. potential for improving health system performance-partic- The schemes reviewed in this paper were mainly volun- ularlyquality and efficiency-than forraising substantial addi- tary schemes run by hospitals, communities, or NGOs. tional finance. This is particularly likely in poor communities, Other approaches warrant further exploration. In particu- where there simply is not a lot of extra money available. lar, many of the problems associated with the schemes Delicate organizational changes-including tighter refer- stemmed from their voluntary nature. More information is ral control, contracting arrangements between purchasers needed on both the feasibility and desirability of compul- and providers, accreditation and service quality improve- sory schemes. Cooperative and mutual insurance organi- ments, and performance-related pay-all might be intro- zations formed the foundations of social health insurance duced under the guise of a shift toward a health insurance systems in Japan and in many Western European coun- system. Insurance should be seen as a supporting strategy, tries. This review unearthed little information on such orga- not as an exclusive financing alternative that might dose off nizations in developing countries. Finally few schemes appear other options and divert attention from the need to improve to have used exemption mechanisms or premiums that were INNOVATIONS IN HEALTH CARE FINANCING 180 graduated according to income or household location. If Dave,P. 1991. "Communityand Self-FinancinginVoluntaryHealth insurance is to succeed in protecting the poor, then much ProgrammesinIndia."Health PolicyandPlanning 6(1):20-31. greater use of these mechanisms will be required. Dave, P, and PeterBerman. 1990. The CostsandFinancingofHealth Care: Experiences in the Voluntary Sector. Case Study 1: The Voluntary Health Services, Madras. New Delhi: Ford Foundation. References de Ferranti, David. 1985. "Paying for Health Services in Developing Countries: A Call for Realism." World Health Forum 6:99-105. Alain, Henry, Guy-Honore Tchente, and Philippe Guillerme- DeRoeck, D., and others. 1996. Rural Health ServicesatSeguridad Dieumegard. 1991. Tontines et Banques du Cameroun: Les Social Campesino Facilities: Analyses of Facility and Household Principes de la Societe des Amis. Paris: Karthala. Surveys: Ecuador Bethesda, Md.: Abt Associates Inc. Arhin, Dyna. 1994. "Health Card Insurance Scheme in Burundi: Diop, F., A. Yazbeck, and R. Bitran. 1995. "The Impact of A Social Asset or A Non-Viable Venture?" Social Science and Alternative Cost Recovery Schemes on Access and Equity in Medicine 39(6): 861-70. Niger." Health Policy and Planning 10(3): 223-40. . 1995. "Health Insurance in RuralAfrica." TheLancet345: Donaldson, D.S. 1982. 'An Analysis of Health Insurance Schemes 44-45. in the Lalipur District, Nepal." Thesis. University of Wshington, Arrow, Kenneth. 1963. "Uncertainty and the Welfare Economics Seattle. of Medical Care." American Economic Review 53(5): 941-73. Eklund, P, and K. Stavem. 1990. "Prepaid Financing of Primary Atim, C. 1997. "Social Movements and Healthlnsurance:A Critical Health Care in Guinea-Bissau: An Assessment of Eighteen Evaluation of Voluntary, Non-Profit Insurance Schemes with Village Health Posts." Policy Research Working Paper 488. Case Studies from Ghana and Cameroon." Paper presented World Bank, Washington D.C. at a University of Antwerp conference on the economics of Ensor,T. 1995. "IntroducingHealthInsuranceinVietnam."Health health insurance in low- and middle-income countries, January Policy and Planning 10(2): 154-63. 17-18, Antwerp. . 1997. "Macro Issues in the Development of Health Bennett, Sara, andE. Ngalande-Banda. 1994. "Public and Private Insurance: World Experience and Lessons forTransitionalAsia." Roles in Health: A Review and Analysis of Experience in Sub- Paper presented at a University of Antwerp conference on the Saharan Africa." Strengthening Health Services Paper 6. World economics of health insurance in low- and middle-income coun- Health Organization, Division of Strengthening Health Services, tries, January 17-18, Antwerp. Geneva. Gilson, L., and others. 1994. "The Potential of Health Sector Bloom, G., and others. 1995. Financing Health Services in Poor Non-Governmental Organizations: Policy Options." Health Rural Areas: Adapting to Economic and Institutional Reform in Policy and Planning 9(1): 14-24. China. Brighton: Institute for Development Studies. Hsiao, W C. L. 1995. "The Chinese Health Care System: Lessons Bogg, L., and others. 1996. "The Cost of Coverage: Rural Health for Other Nations. " Social Science and Medicine 41 (8): 1047-55. Insurance in China." Health Policy and Planning 11(3): Hsiao, W C. L., and P D. Sen. 1995. "Cooperative Financing for 238-52. Health Care in Rural India." Paper presented at an International Carrin, G., and others. 1996. The Reforn of the Rural Cooperative Workshop on Health Insurance in India, sponsored by the Medical System in the People's Republic of China: Initial Design Indian Institute of Management, September, Bangalore. and Interim Experience. Geneva: World Health Organisation. Ikegami, N., and T. Hasegawa. 1995. "The Japanese Health Care ChabotJ., M. Boal, and A. Da Silva. 1991. "National Community System: A Step-Wise Approach to Universal Coverage." In Health Insurance at the Village Level: The Case from Guinea- David W Dunlop and Jo M. Martins, eds., An International Bissau." Health Policy and Planning 6(1): 46-54. Assessment of Health Care Financing: Lessons for Developing Creese, Andrew, and Joseph Kutzin. 1995. "Lessons from Cost Countries. Washington, D.C.: World Bank Economic Recovery in Health." Discussion Paper 2. World Health Development Institute. Organization, Geneva. ILO (International Labour Office). 1996a. "India: Social Protection Criel, B. 1992. "Community Financing Schemes: Give Them Tume. for the Unorganized Sector." South Asia Advisory Team. New Discussion of Two Prepayment Schemes Conducted at the Delhi, India. District Level in Zaire." University of Antwerp, Belgium. . 1996b. World Employment Report 1996/1997. Geneva. Criel, B., and W van Lerberghe. 1997. "The Bwamanda Hospital Jajoo, U.N. 1992. "Risk-Sharing in Rural Health Care." World Insurance Scheme: Effective for Whom? A Study of its Impact Health Forum 13: 171-75. onHospitalUtilisationPatterns."PaperpresentedataUniversity Jazairy, I., M. Alamgir, and T Panuccio. 1993. The State of World of Antwerp conference on the economics of health insurance Rural Poverty. New York: New York University Press for the in low- and middle-income countries,January 17-18,Antwerp. International Fund for Agricultural Development. RuRAL RISK-SHARING STRATEGIES 181 Kutzin, Joseph. 1995. "Experience with Organizational and Ron, A., and A. Kupferman. 1996. Community Health Insurance Financing Reform of the Health Sector." Strengthening Health Scheme in the Philippines: Extension of a Community-Based Services Paper 8. World Health Organization, Geneva. Integrated Project. Geneva: World Health Organization. La Forgia, G. M. 1990. Health Services for Low-Income Families: Saltman, R. 1995. "ApplyingPlanned Market Logic to Developing Extending Coverage Through Prepayment Plans in the Dominican Countries' Health Systems: An Initial Exploration." Discussion Republic. Bethesda, Md.: Health Financing and Sustainability Paper 4. World Health Organization, Geneva. Project. Shaw, R. Paul, and Charles C. Griffin. 1995. Financing Health Care Litpon, Michael. 1976. Why PoorPeople Stay Poor London: Temple in Sub-Saharan Africa through User Fees and Insurance. A Directions Smith. in Development book. Washington, D.C.: World Bank. Lukholo A. 1996. Stockvels in South Africa. Johannesburg: Amagi Sheng-Lan, T., and others. 1994. Financing Health Services in China: Books. Adapting to Economic Reform. Brighton: Institute for McFarlane, G. A. 1996. "Chogoria Hospital Insurance Scheme." Development Studies. Chogoria Hospital, Kenya. Somkang, E., P. Akanzinge, T. Apau, and F. Moens. 1994. McPake, Barbara, Kara Hanson, and Anne Mills. 1993. "Nkoranza Health Insurance Evaluation Report." Child Health "Implementing the Bamako Initiative in Africa: A Review and and Development Centre, Uganda. Five Case Studies." London School of Hygiene and Tropical Stinson, W 1982. Community Financing of Primary Health Care. Medicine, Department of Public Health and Policy, Health Washington, D.C.: American Public Health Association. Policy Unit, London. Supachutikul, A. 1996. Situation Analysis of Health Insurance and Mills, Anne. 1983. "Economic Aspects of Health Insurance." In its Future Development. Bangkok: Health Systems Research K. Lee and Anne Mills, eds., The Economics of Health in Institute. Developing Countries. Oxford: Oxford University Press. Supachutikul, A., and P. Sirinirund. 1993. "A Report on Moens F. 1990. "Design, Implementation and Evaluation of a Thai-German Health Card Project." Community Financing Scheme for Hospital Care in Suwandono, A., Brahim, and R. Malik. 1995. "The Indonesian Developing Countries: A Pre-Paid Health Plan in the Experiences on Rural Health Financing." Ministry of Health, Bwamanda Health Zone, Zaire." Social Science and Medicine Indonesia. 30(12): 1319-27. UN (United Nations). 1993. World Urbanization Prospects: The Mogedal, S. M. 1984. "Local Health Insurance Supplementing 1992 Revision. New York. National Efforts-Nepal." Public Health Review 12: 286-93. UNICEF (United Nations Children's Fund). 1995. The Bamako Myers, C.N. 1989. "Thailand's CommunityFinance Experiments: Initiative: Rebuilding Health Systems. New York. Experience and Prospects." Harvard Institute for International Waddington, C., and K Enyimayew. 1990. "A Price to Pay, Part Development, Cambridge, Mass. 2: The Impact of User Fees in the Volta Region of Ghana." Noterman,J., and others. 1996. 'A Prepayment Scheme for Hospital International Journal of Health Planning and Management 5(4): Care in the Masisi District in Zaire: A Critical Evaluation." 287-312. Social Science and Medicine 40(7): 919-30. WHO (World Health Organization). 1995. International Parker, D., and R. Knippenberg. 1991. Community Cost-Sharing Consultation on Health Insurance Reform. Seoul, Republic of and Participation: A Review of the Issues. New York: United Korea. WHO/SHS/NHP/95.9. Nations Children's Fund. Wibulpolprasert, S. 1991. "Community Financing: Thailand's Peabody, J.W, S. Lee, and S.R. Bickel. 1995. "Health for All in Experience." Health Policy and Planning 6(4): 354-60. the Republic of Korea: One Country's Experience with Willis,C.Y, and C. Leighton. 1995. "Protecting the Poor: The Role Implementing Universal Health Care." Health Policy 3 1: 29-42. of Means Testing." Health Policy and Planning 10(3): 241-56. Preker, Alexander, and Richard Feachem. 1995. Market Mechanisms World Bank. 1987. Financing Health Services in Developing Countries: and the Health Sector in Central and Eastern Europe. World Bank AnAgendaforReform. AWorld Bank Policy Study. Washington, Technical Paper 293. Washington, D.C. D.C. Ramirez, T., and others. 1997. "The Market for Maternal and Child - 1991. Indonesia: Health Planning and Budgeting. A World Health Care Services in Mexico City and the Role of an NGO." Bank Country Study. Washington, D.C. In Sara Bennett and others, eds. Private Health Care Providers . 1996. China: Issues and Options in Health Financing. in Developing Countries: Serving the Public Interest? London: Washington, D.C. Zed Press. WSM (World Solidarity Movement). 1996. Mutulles de Sante' en Ron, A. 1997. "Community Health Insurance Schemes: Experience Afrique. Senegal. inGuatemalaandthePhilippines."PaperpresentedataUniversity Yu, S., and G. F. Anderson. 1992. "Achieving Universal Health of Antwerp conference on the economics of health insurance in Insurance in Korea: A Model for Other Developing Countries?" low- and middle-income countries, January 17-18, Antwerp. Health Policy 20: 289-99. INNOVATIONS IN HEALTH CARE FINANCING 182 Rural Health Care Financing in kThailand Sirilaksana Kboman H T ealth care financing has become an issue of global poverty has declined, the distribution of income has not interest and concern in recent years, as both improved (Krongkaew 1995). Moreover, rural and urban capitalist and socialist countries grapple with populations have very different health indicators. Rural pop- rising costs, dwindling resources, poor-quality care, ineffi- ulations are afflicted with nearly twice as many poverty- cient resource use, and unequal distribution of services, related diseases-such as infectious diseases-per capita Transition economies in Indochina, Eastern Europe, and as urban populations. The Noitheast, the country's poor- China, as well as capitalistic strongholds like the United est region, has the highest incidence of infectious diseases- States, face formidable challengesindeterminingwho should 53.2 cases per 1,000 people, compared with 25.5 per 1,000 pay for health care and how it should be managed. in Bangkok. Urban residents have a higher incidence of dis- In developing countries rural areas are of particular ease only for endocrine disturbances (such as diabetes) concern because regional disparities in income put rural and circulatory diseases (such as heart disease). populations at a disadvantage in terms of living standards Differences in illness patterns are also apparent in hos- and access to health care. Moreover, limited administra- pital admission statistics (Thailand Ministry of Public Health tive capacity in rural areas makes it harder to manage what- 1995). Among inpatients at government hospitals, diges- ever financing scheme is implemented, especially in areas tive disorders are the largest cause of admission in the where large portions of the population are engaged in sub- provinces; in Bangkok the largest cause is malignant tumors. sistence activities, cut off from the formal sector. Infant and maternal mortality rates also show marked This paper reviews Thailand's experience in developing, differences between the different regions and Bangkok. implementing, and attimes experimentingwithvarious health Infant mortality in the North, Northeast, and South is esti- care financing schemes, with a focus on projects affecting the mated to be about twice the rate in Bangkok (where it is rural population. The next section summarizes the disparity 18 deaths per 1,000 live births). Similarly, maternal mor- between rural and urban areas with respect to health outcomes tality in all regions, particularly the South, is two to four and available resources. The third section assesses the range times higher than in Bangkok (where it is 0.1 per 1,000 live of financing schemes that are in place. Schemes that include births). Since the country's urbanization rate is only 27 and affect rural populations are discussed in the fourth sec- percent, with 100 percent urbanization for Bangkok, the tion. Finally, condusions are presented in the final section. regional differences also reflect rural-urban differences. The Northeast also records the country's highest inci- Rural-Urban Differentials dence of first-degree malnutrition, although the situation has been improving over the past ten years. About 25 per- For the past two decades Thailand has been one of the cent of the Northeast's population aged five years and under world's fastest-growing economies. But although absolute suffers from first-degree malnutrition, compared with a Sirilaksana Khoman is dean of the faculty of economics at Thammasat University in Bangkok. 183 countrywide average of less than 19 percent (8 percent in Financing Schemes the Central region). Thais prefer physician services to any other form of treat- There are four main types of health care financing schemes ment; 54 percent of sick people seek treatment at health in Thailand: voluntary health insurance, mandatory schemes, outlets staffed by physicians. Again, however, a rural-urban social welfare schemes, and fringe benefit schemes such as breakdown reveals that a much higher percentage of the health coverage for government officials and state enter- urban population seeks care from physicians. In urban areas prise employees (table 2). The distinction between types of 81 percent of sick people have consultations with physi- schemes has not always been clear, however, because cians, compared with 47 percent in rural areas. Moreover, Thailand experiments with ways to provide its population rural populations are almost twice as likely to rely on self- with secure and accessible health care. treatment as urban populations (32 percent compared with Voluntary health insurance consists of private commercial 17 percent). The behavior of urban residents is almost iden- insurance, which covers about 0.9 million people almost exclu- tical, whether they live in Bangkok or other urban areas sively in the formal sector, and the Health Card Program, (Khoman 1992). which was implemented in rural areas in 1983 as a voluntary The Ministry of Public Health is the main provider of scheme. Over the years the Health Card Program has evolved health services, particularly primary care. It has a network and can now also be considered a kind of social welfare pro- of hospitals at the regional, provincial, and conmnunity lev- gram, since it receives an explicit contribution from the gov- els, and health centers at the subdistrict level. Most of the ernment equal to the contribution of the card purchaser. mninistry's services are in rural areas, as well as in urban Coverage is still fairly limited, however, and has fluctuated centers besides Bangkok. Almost half of the ministry's annual between 1.3 and 2.7 million people. These fluctuations occurred budget is spent in rural areas, and an increasing share is mainly because of lapses in policy direction, and the often ad used for primary health care. hoc way in which the program is implemented (see below). Still, rural-urban differentials in the supply of physicians, Mandatory schemes indude the Workmen's Compensation hospitals, and hospital beds remain striking. In Bangkok there Fund and the Social Security Scheme. The recently formed is one physician for every 958 people; in the Northeast the Ministry of Labor and Welfare manages both schemes, which rate is more than ten times higher (table 1). Moreover, pri- cover workers in firms with ten or more employees. These vate households, both rural and urban, remain the largest schemes require extensive record keeping on employment to source of finance for health services. The share of total health verify eligibility, and thus are confined to formal sector expenditures coming from households and private companies employees. increased from 63 percent in 1977 to 72 percent in 1986, to The Workmen's Compensation Fund covers job-related nearly 74 percent in 1992 (Khoman and Mongkolsmai 1993). injuries and (theoretically) work-induced illnesses. The lat- TABLE I Distribution of medical and public health resources, 1992 (population per unit) Region Physician Dentist Pharmacist Nurse Hospital beda North 6,316 41,176 24,910 964 Northeast 10,970 78,211 45,020 1,606 Central 5,804 29,181 25,854 815 South 6,079 31,574 21,143 806 Bangkok 958 4,599 2,142 363 260 Regional average excluding Bangkok 7,326 42,811 29,608 1,050 800 Whole Kingdomn 4,425 25,530 13,076 885 666 a. Beds in government hospitals for general services only. Source: Thailand Ministry of Public Health 1995. INNOVATIONS IN HEALTH CARE FINANCING 184 ter, however, are difficult to prove-particularly if adverse could be expanded to accommodate rural populations, it health effects are evident only after many years (as is the is discussed in greater detail in the next section. case with manganese poisoning, silicosis, lead poisoning, The third type of scheme is social welfare programs, con- and others). Even job-related injuries, particularly if they sisting of several specifically targeted schemes. First, the are perceived as being minor, are underreported if work- Low Income Support Program provides free medical care ers are not assertive or well informed, and if the employer to poor rural families in government health facilities. The is concerned with loss of work time or the hassles of the cutoff point for coverage is 2,000 baht per person or 2,800 required paperwork (Kultap 1983). All employers with baht per family per month. In 1992 this program covered ten or more employees are required to contribute to the 11.7 million people, or about 20 percent of Thailand's fund. population. The Social Security Scheme, implemented under the Free medical services are also given to the elderly. In 1992 Social Security Act of 1990, provides health insurance as about 3.5 million were covered, and this number is likely part of an overall package of benefits designed to provide to increase considerably in the near future because of an security to populations not covered by other benefit pro- aging population. In addition, the Ministry of Education grams. However, implementation is confined to the most provides free medical care to primary schoolchildren in manageable group-employees in formal sector establish- schools under its jurisdiction. This program benefits chil- ments. In its first three years the scheme covered employ- dren in both urban and rural areas, but provides propor- ees in firms with twenty or more employees. Coverage has tionately higher benefits for rural populations, who are less since been extended to firms with ten or more employees, likely to have other coverage. More than 5 million chil- increasing the population covered to 4.5 million people, or dren, or about 9 percent of the population, are covered 7.6 percent of the population in 1995. Since this scheme under this scheme. TABLE 2 Coverage of health care financing schemes, 1992 Share of Population population Subsidy Scheme Target population covered (percent) Source of finance per capita Voluntary health insurance Private insurance Mainly urban 0.9 million 1.6 Insurer Health card Mainly rural 1.3 million 2.3 Card holder and Ministry of Public Health 63 baht Mandatory schemes Workmen's Compensation Fund Formal sector employees 1.8 million 3.2 Employers and Ministry of Labor and Welfare - Social security Formal sector employees 2.5 million (1992) 4.4 Employers, employees, and Ministry of 541 baht 4.5 million ( 1995) 7.6 Labor and Welare Social welfare schemes, Low-income support Low-income, mainly rural 1 1.7 million 20.7 Ministry of Intenor 214 baht Support for the elderly Population over 60 3.5 million 6.2 Ministry of Public Health 72 baht Schoolchildren Primary school children 5.1 million 9.0 Ministry of Educaton Fringe benefit schemes Govemment reimbursement Govemment officials and 5.6 million 9.9 Govemment (various agencies) 916 baht employees and their families State enterprise benefits State enterprise employees 0.8 million 1.4 815 baht and their families Insured populabon 33.2 million 58.7 Uninsured population 23.3 million 41.3 a. Other wefare recipients include veterans, monks, and those deemed needy. Source: Thailand Ministry of Public Heaith 1992; Thailand Ministry of Labor and Wefare data; calculated from Hsiao 1994 and Mongkolsmai 1993. RURAL HEALTH CARE FINANCING IN THAILAND 185 Other welfare recipients include veterans and monks. Health Schemes Affecting Rural People who are considered needy (as determined by social Populations workers in government hospitals) are also routinely given free care. In addition, the Ministry of Interior offers free- Low-income support care cards to low-income urban families. Finally, health coverage is offered as a fringe benefit in Government policy on providing free medical care to low- large private companies, government agencies, and state income groups was initiated in 1975. The goal of this pol- enterprises. About 5.6 million government officials and their icy was to reduce the prevailing inequity in access to health families are covered, and nearly 1.0 million state enterprise services. Known as the Low Income Support Program, it employees and dependents. Combined, these two groups offers free medical care at government hospitals to low- make up about 11 percent of the population. income groups and has become the main health scheme Government officials tend to be in middle- to high-income for rural populations. groups. In fact, they receive the largest subsidy from gov- Coverage initially was limited to people with monthly emnment expenditures on health (see table 2). Such cover- incomes below 1,000 baht. Upon implementation, the tar- age has little effect on rural populations unless they are get population was set at 7-8 million people. Until 1980, sufficiently educated to qualify for positions in govern- however, no identification cards were issued, and free care ment agencies or state enterprises. was given at the discretion of health facilities' staff. Since Altogether, about 59 percent of Thailand's population then low-income cards have been issued to eligible citi- is protected by some kind of health care coverage-and zens, now defined as families with monthly incomes below 41 percent, or 23 million people, is not covered by any 2,800 baht and individuals with monthly incomes below scheme. This group includes subsistence farmers, the self- 2,000 baht. The cards entitle holders to free medical care employed, rural workers, and urban dwellers engaged in at all government health facilities operated by the Ministry informal sector activity such as street vending and small- of Public Health, the Bangkok Metropolitan Administration, scale commercial undertakings. Those living on the fringes the Red Cross Society, Pattaya City, and municipalities. of society (slum dwellers, homeless urban migrants) typi- Cards are valid for three years. cally also have to fend for themselves. A survey of Bangkok The government provides block grants to health facili- transients who dwell in makeshift accommodations under ties based on the expected distribution of the eligible pop- bridges and at railway stations found that none was cov- ulation and past records of service to patients. Specifically, ered by a health scheme (Khoman 1995). Yet they say they the budgetary allocation is based on the number of low- are fit and well. income people living in less-developed villages, as defined The Ministry of Public Health is trying to expand insur- by the National Economic and Social Development Board. ance coverage to slum dwellers, the self-employed, and In addition, the number of users of health care facilities is highly mobile groups such as construction workers, service taken into account, and the number of veterans and their workers, and prostitutes. Some studies, however, indicate families. that construction workers and prostitutes have little inter- The free medical care budget for low-income groups est in purchasing any form of health care coverage increased from 521 million baht in 1982 to 1,911 million (Mongkolsmai and others 1994). Prostitutes already visit baht in 1993. As a share of the health budget, the low-income clinics that treat sexually transmitted diseases, a service pro- budget has ranged from 7.7 percent in 1980 to 7.9 percent vided by most government provincial hospital as the need in 1993, down from 12.5 percent in 1976. In real terms the arises. Construction workers also showed little interest, budget per patient dropped from 155 baht in 1976 to 45 partly due to the complicated nature of health coverage and baht in 1980, stabilizing at around 50 baht over the past the low perceived need. Current research on slum dwellers ten years. The budget allocated per card holder has been andtransients should provide further insight into this group's rising, however, increasing from 68 baht in 1984 to 163 baht willingness to pay for health coverage (Rojvanit 1995). in 1992. INNOVATIONS IN HEALTH CARE FINANCING 186 Still, the budgetary allocation is invariably insufficient The cutoff level used for card eligibility is almost four to cover the cost of providing services. Satsanguan and times the poverty line, defined as the minimum income Leopairote (1992) report a unit cost of 85 baht for an required for subsistence. Yet in 1988/89, when about 29 outpatient visit and 1,200 baht per case or 360 baht per percent of the rural population was below the poverty line, day for inpatient care at district hospitals in 1991. As a the Low Income Support Program covered only 7.65 mil- result hospitals routinely cross-subsidize low-income cov- lion people, or just under 20 percent of the rural popula- erage using other sources of revenue, such as reimburse- tion. The card covered just 28 percent of the low-income ments for government officials and hospital fees. The extent group as defined by the income cutoff level, and 49 per- of the subsidy varies by hospital. For example, Nan cent of the poor as defined by the poverty line. In 1990 Provincial Hospital in the North reported that between coverage improved as a result of expanded efforts to reach October 1992 and March 1993 the costs of providing targeted groups and increased screening of card recipients. care for low-income patients totaled 1.4 million baht a Nevertheless, coverage remains low, with up to 20 percent month, while the annual budgetary allocation was 10.5 of those below subsistence still left out (Mongkolsmai 1993). million baht, or 0.87 million baht a month. Thus the cross- Even with limited coverage of the poor, however, Thailand subsidization was about 6 million baht a year (Mongkolsmai still has a safety net that protects the needy, since many 1993). In the province of Samutsakorn the budget allo- low-income people routinely receive free medical care even cation was 2.3 million baht while actual expenditures were without a card. In 1987, 13.7 million people received free 5.8 million baht. medical care, but of these only 7.6 million were low-income The scheme does not require cost sharing on the part of card holders. The proportion is believed to be roughly the the eligible population and has been subject to much crit- same today, so that people seeking health care can access icism since its inception, particularly with respect to the health facilities without income being a constraint. distribution of the card. A 1980 study by the Ministry of Moreover, the Low Income Support Program has prob- Public Health's Rural Health Division found that 12 per- ably improved access to health services among the poor. cent of the supposedly low-income beneficiaries using provin- The Northeast has the highest share of low-income card cial and district hospitals and 9 percent visiting health centers coverage, covering about 30 percent of the population (table had monthly incomes above 2,000 baht. Another study found 3). And coverage of the poor (those living below the poverty that about 20 percent of card holderswere not poor (Thailand line) increased in all regions between 1987 and 1990, with Ministry of Public Health and Faculty of Health Services, the largest increase in the Northeast. Mahidol University 1988). Mongkolsmai and Khoman The problem that remains is the difficulty of properly (1993) also found that some nonpoor families, possiblywith identifying eligibility. Given that much of the rural popula- connections to officials, possess the card. tion is engaged in agricultural activity, with a substantial pro- TABLE 3 Coverage under the Low Income Support Program, 1987 and 1990 1987 1990 Population Share of Share of Population Share of Share of covered population poor' covered population poor' Region (millions) (percent) (percent) (millions) (percent) (percent) North 1.8 18.8 73.7 2.4 23.4 100.8 Northeast 3.5 20.4 42.3 5.3 30.4 81.1 Central 1.3 11.3 72.6 1.5 12.1 75.7 South 1.0 12.4 45.6 1.5 18.2 84.7 Bangkok 0.0 0.5 14.3 0.0 0.3 8.7 Whole Kingdom 7.6 14.5 49.2 10.7 19.2 81.0 a. Determined using poverty line for 1988/89. Source: Thailand Ministry of Public Health 1995. RuRAL HEALTIH CARE FINANCING IN THAILAND 187 portion of income received in kind, assessing and imput- baht. Each cardwasvalid forayear. The familycard allowed ing income is difficult. Defining a household is also prob- eight illness episodes a year for four family members; the lematic in cases where family members work in cities and individual card allowed four illness episodes. To prevent remit earnings. against moral hazard behavior, ceilings were imposed. Coverage was limited to 1,000 baht for accidents, with a Health Card Program 10 percent discount for amounts over 1,000 baht. Chronic diseases, cancer, and "self-inflicted" diseases were excluded. The Health Card Program was started in 1983 to promote Card holders who had not used the card in one year were maternal and child health. It was an innovative program entitled to renew it without charge for an additional year, because it involved selling health cards to villagers who not exceeding two renewals. had been accustomed to receiving free care. Card buyers In mid-1984 the ceiling was raised to 2,000 baht per visit. prepaid a fixed annual premium in return for free services. A strict referral system was also enforced whereby initial Proceeds from card sales went into a health card fund that contact had to be with the village drug fund, and subse- was managed by a village committee. Thus the program quently the health center in the subdistrict, the community familiarized rural populations with concepts of preventive hospital at the district level, and finally the general or regional behavior, insurance, risk pooling, fund management, and hospital at the provincial level. This approach was intended community self-help, though the initial emphasis was on to increase efficiency in the use of different levels of health improving health. services, since over-utilization of high-level hospital care T'he program was initiated in eighteen villages in seven was occurring because villagers routinely bypassed the health provinces. At its inception the program's primary objective centers in favor of physician services. was to improve health among rural populations, with an From the outset, the health card fund was designed to emphasis on primary care-including health education, envi- be a village-level fund in order to foster grassroots partici- ronmental health (sanitation and water supply), maternal pation and management skills. Of this fund, 15 percent and child health, family planning, nutrition, immunization, was allocated to the health center, 30 percent to the com- prevention and control of diseases, treatment of common munity hospital, and 30 percent to the regional or provin- ailments, and provision of essential drugs. In addition, the cial hospital, reflecting the pattern of utilization in the referral program incorporated a referral system that required card system and the severity of illness and cost to the provider. holders to visit primary care centers instead of tertiary care Of the remaining 25 percent, 10 percent went to person- institutions, so that care-seeking behavior would be more nel in the provider institutions and 15 percent was used as rational and efficient in terms of utilization of health operating expenses of the fund. resources. The program was also intended to involve local The Health Card Program later gave more attention to villagers in self-help as well as in managing the health card medical treatment, and the price of the family card was fund. Self-help and communal participation were also fos- increased to 300 baht. tered by recruiting village health volunteers and village health During 1985-90 some health card funds were initiated communicators who worked with health personnel at the at the district and subdistrict levels: a 300 baht family card, village health center. a 100 baht maternal and child card, and a 200 baht indi- The features of the Health Card Program were constantly vidual card. In addition, the number of episodes allowed adjusted, and policymakers sometimes vacillated over time. was cut from eight to six. At issue were the number of free episodes allowed, the Confusion arose about the different types of cards, the coverage ceiling per visit, the number of family members terms and conditions of use, the losses incurred due to the included, the types of diseases covered, the price of the card, inability of hospitals to recover costs from the health card fund and renewability. In 1983 there were three types of cards: contribution, and the problems with the strict referral sys- a family card priced at 200 baht, a maternal and child card tem, which tended to disregard geographical proximity priced at 100 baht, and an individual card priced at 100 Tantiserani (1988) found that the Health Card Program was INNOVATIONS IN HEALTH CARE FINANCING 188 active in about 33 percent of the villages and 70 percent of Social security the subdistricts in 72 provinces, with about 2.7 million peo- ple holding about 550,000 family cards. A 1992 survey by the The Social Security Act of 1990 provides health insurance Ministry of Public Health found that coverage had extended as part of an overall package of benefits covering illness to 36 percent of the villages, but the population covered was unrelated to work, maternal benefits, disability unrelated reduced to 30 percent of the subdistricts in 68 provinces, to work, death, child benefits, old age, and unemployment. with 1.3 milion persons holding 260,000 family cards. Companies with ten or more employees and employees This reduction in coverage has been attributed to the themselves are each required to contribute 1.5 percent of ministry's lack of policy direction during this period the employee's wages to the Social Security Fund, with an (Kiranandana 1990). The program was implemented in an equal 1.5 percent provided by the government. Since the ad hoc manner and thrived only in provinces that actively contribution is based on income and not the expected risk encouraged villager participation. Moreover, the program or incidence of illness, risks are pooled and benefits are was never lucrative for large hospitals, and had to be sub- skewed in favor of high-risk individuals. sidized by other sources of hospital revenue. Of this contribution, 2;45 percentage points is used to In 1990 the health card scheme was modified and provide medical care for the insured for illness and mater- renamed the New Health Card Approach, which empha- nity and 2.05 percentage points are used for disability and sized the concept of risk sharing. Price restructuring (rang- death benefits. Expanding benefits to include child and ing from 200 baht for the individual card, and from 300-500 the elderly would require an increase in the contribution, baht for the family card, covering up to three generations) and so has been postponed. Unemployment benefits will occurred in five provinces (Manopimoke 1995). In addi- be implemented at a much later stage for fear of moral tion, the referral system was relaxed and coverage was hazard behavior-that is, providing unemployment bene- expanded, with unlimited numbers of free visits and no cov- fits could induce unemployment. erage ceiling per visit. About 20 percent of the population As noted, social security is mandated for the most man- in the selected areas participated in the project. ageable group, namely employees in formal sector estab- In 1991 the different types of cards were discontinued, lishments with ten or more employees. The scheme covers with only family cards (priced at 500 baht) offered. Since about 4.5 million workers, or 7.6 percent of the population. 1994 the Ministry of Public Health has provided an equal The Social Security Act also provides for expansion of contribution of 500 baht per card. In addition, no limits the scheme on a voluntary basis to include the self-employed, are imposed on the number of episodes or the cost cover- such as farmers, own-account workers, and other uninsured age per visit. Moreover, more flexibility was built into the groups. However, this provision was probably included for referral system, and each province could impose whatever political expediency to protect against charges that the conditions it deemed appropriate. Some provinces, such program only covers the well off. Even if the self-employed as Rachaburi, allow card purchasers to pay in installments were allowed to participate, the problem remains of how (Mongkolsmai and others 1994). Administrative changes to deal with their contribution. Thus the program has been were also implemented, with the health card fund man- proposed as a practical form of coverage for the self- aged by a committee at the district level in coordination employed, but Thailand is still experimentingwith this idea. with village-level bodies. This approach was intended to Amore immediate concern is the current system of pro- expand the enrollment base beyond the village to the dis- viding medical care. Insured workers are required to regis- trict level. In addition, 80 percent of the card price now goes ter at a specified hospital, called the "main contractor," to providers of medical care, while the remaining 20 per- where they receive free medical care (except for certain cent is retained for marketing and sales incentives. types of treatment, such as cosmetic surgery and optome- The health card is likely to be used to expand coverage try). These restrictions are similar to those under the Health for people who currently lack insurance. Studies are being Card Program. The main contractor receives an annual pre- undertaken to ascertain its feasibility in urban areas as well. payment or capitation fee from the Social Security Fund, RURAL HEALTH CARE FINANCING IN THAILAND 189 initially equal to 700 baht per insured person registered, was 0.32 visits per insured person per year, and in 1992 the regardless of actual utilization. Capitation was chosen over rate dropped to 0.22 visits. In calculating the capitation a fee-for-service payment scheme (to providers) for its admin- fee, it was estimated that there would be 3.0 visits per insured istrative simplicity and to prevent the cost escalation that person per year. Inpatient utilization was also lower than invariably occurs with fee-for-service payments. The expected; in 1991 and 1992 the utilization rate was 0.016 Workmen's Compensation Fund, for example, pays hospi- and 0.012 visits per insured person per year. The estimated tals on a fee-for-service basis, and has incurred substantial utilization rate was 0.05 visits. cost escalation, as well as the administrative burden of deal- Apart from the inconvenience of workers receiving care ing with massive claims documents. Capitation, however, at a hospital they were not able to select, and the limited may not be attractive enough to induce medical providers number of participating hospitals to choose from, several to enroll in the scheme. The relative merits of alternative other problems emerged. Many workers were ignorant of payment methods are sunmmarized in table 4. their rights, some were not aware of the contributions they The main contractor is able to contract with subcon- made to the Social Security Fund because of automatic tractors, which provide lower levels of care, as well as with deductions from their wages, and many did not under- supracontractors, which provide higher levels of care. Both stand the procedures that had to be followed to obtain med- public and private hospitals participate in the scheme. One ical care (Patichon 1995). Further, providers of medical care of the main problems with. the scheme is confusion with were not prepared to manage the scheme in terms of health respect to insured persons receiving care in hospitals in care delivery. As a result efforts have been made to improve which they are not registered. Accident victims, for exam- the delivery system and the quality of care so that insured ple, made headlines when they were allegedly denied treat- persons have better access to the care to which they are ment by noncontracting hospitals. entitled. At first employers chose the main contractor, which may Specifically, the Social Security Office issued a policy explain low utilization in the first year of the scheme's imple- directive in 1992 to grant insured persons the right to mentation. In 1991 the utilization rate for outpatient care choose their own hospital. This policy was implemented TABLE 4 Strengths and weaknesses of alternative methods of paying health providers Payment method Strengths Weaknesses Fee for service Provider's reward closely linked to level of effort Tends to cause cost inflation and output Requires processing and verification of massive numbers Allows for easy analysis of provider's practice of claims documents Creates incentive for excessive and unnecessary treatment Per case (for example, using Provider's reward fairly well tied to output Technical difficulty of forcing all cases into standard list diagnostic-related groups) Gives provider incentive to minimize resource can lead to mismatch between output and reward Providers may misrepresent diagnosis in order to receive use per individual treated higher payment Capitation (per patient Administratively simple; no need to break down Gives provider incentive to select patient based on risk under continuous care) physician's work into procedures or cases and reject high-cost patients Facilitates prospective budgeting May create incentives for provider to underservice Gives provider incentive to minimize cost of treatment accepted patients Allows for consumer clout if patient can select own provider Difficult to analyze provider's practice Salary (straight payment Administratively simplest Loss of patient influence over provider behavior unless per period of work) patient choice links provider salary to patient sabisfaction Facilitates prospective budgeting Can easily create incentives for provider to underservice patient and to reduce productivity Source: Wodd Bank 1993. INNOVATIONS IN HEALTH CARE FINANCING 190 in one province in 1992 and expanded to nineteen and There is also evidence of abuse. That is, the medical forty-four provinces in the following two years. The pol- providers that some networks have recruited are so far icy is being implemented in sixty provinces at present. As apart-in some cases, providers are in provinces separated a result the portion of insured persons choosing their by several hundred kilometers-that insured persons have own hospital increased from 2 percent in 1992 to 37 per- few opportunities to use them. Providers can claim the addi- cent in 1995. tional payment that is calculated on the basis of services The Social Security Office is also encouraging the for- provided, but these services are provided to different peo- mation of provider networks to increase efficiency in health ple, and the insured person does not benefit from the appar- care delivery, improve accessibility of services (particularly ently wide range of providers available. These problems if network members are geographically dispersed), and pool are being studied and further adjustments made to the risks. Moreover, additional payment is given to the main system. contractors in proportion to the inpatient and outpatient care that they provide. Thus medical providers have an Conclusion incentive to increase their market share and engage in mar- keting for contracts with workers. As a result the number The basic issue with respect to financing health care involves of main contractors increased from 137 in 1991 to 1,879 determining how much of a country's resources should be in 1995. Among private providers, the number of network devoted to health services, how much should be spent by members increased from 69 in 1992 to 620 in 1995. In the the govemment, and how much reliance should be placed public sector the increase has been more moderate: from on private sources of finance. The main questions are: 671 in 1991 to 1,257 providers in 1995. The switch from Who should pay for the cost of providing services (recipi- public to private hospitals has also been increasing. In 1991 ents through user charges, government through subsidies, only 16 percent of insured persons chose private hospitals, or other funding sources such as private business, collec- but by 1994 that share was 59 percent. tive bodies, and charitable organizations)? How should Two pattems of network formation are used. The first is health care be organized? And what role should financial a direct contract network, in which the main contractor con- intermediaries play? tracts directly with network members and assumes respon- In Thailand various schemes are used to address these sibility for managing the capitation fee received from the issues. Some overlap, covering the same population, while Social Security Office and acting as a "secondary medical other population groups are left unprotected. There are also care provider." The second is an indirect contract network, differences between the benefits that can be obtained under in which the main contractor transfers the capitation fee to each scheme. Some schemes, such as the fringe benefit a network office responsible for managing the funds received scheme that covers government officials and the Workmen's from several main contractors. The network office is also Compensation Fund, pay for health care on a fee-for-ser- responsible for marketing and recruiting network members. vice basis. Others, like the social security scheme, use a cap- In the second case the main contractor is responsible for itation method of payment. Some schemes, such as the social medical services only. security scheme and the Workmen's Compensation Fund, In practice, however, there is considerable confusion in are compulsory. The Health Card Program, on the other the system, with great variation between networks with hand, is voluntary, and raise questions of adverse selection. respect to network coverage of the three levels of care Some schemes require copayments from beneficiaries, while (subcontractor, main contractor, supracontractor), ability others restrict the types of services covered. to manage funds and the payment mechanism to encour- Some schemes, such as the Low Income Support Program age cost containment and foster financial feasibility for the and coverage for veterans and the elderly, require no direct network, quality and standard of medical services pro- contribution from beneficiaries. Others, such as the health vided by the networks, and coordination between networks card and social security, require explicit contributions. and network management at the national level. Government subsidies exist for most schemes, some in terms RURAL HEALTH CARE FINANCING IN THAILAND 191 of explicit budgetary allocation (such as for veterans, the conducive to risk pooling, which requires a large number elderly, low-income groups, social security, and the health of members. Thus the second precondition is government card). Implicit government subsidies also pervade the pub- initiative and partial subsidy. This is the course that the lic health system through the pricing of services at govern- Health Card Program has taken, and even though it has ment hospitals, giving different benefits to different types been beset with problems, it is nevertheless a viable scheme of users (Khoman 1995). to start off with in a rural community Although some schemes were designed specifically for the rural population, the distinction between the formal and References informal sectors is even clearer. In fact, many groups in the informal sector are excluded from existing schemes. Hsiao, William C. 1994. "Health Care Financing in Thailand: There is little coordination between schemes, and Thailand Challenges for the Future." Paper presented at a faculty sem- wiT have to figure outhoration esetween cscrepanies, ind filanc inar at the College of Public Health, Chulalongkom University, will have to figure out how to ease the discrepancies in financ- Bangkok. ing. Different schemes contain varying elements of subsidy, Khoman, Sirilaksana. 1992. 'Household Choice of Health-Care and could cause the allocation of resources toworsen between Provider in Thailand." Research Report 92-02. International urban and rural areas. All these aspects must be considered Health Policy Program, Washington, D.C. as Thailand continues to make adjustments to increase effi- . 1995. "The Pricing of Govemment Health Services in ciencyand equity in health care financing. Thailand: Equity Versus Efficiency?" Paper prepared for the ciency lessons m bead rawn fromg. Regional Conference on Health Sector Reform in Asia, held Some lessons can be drawn from Thailand's experi- at the Asian Development Bank, May 22-25, Manila. ence, however, particularly in terms of the types of financ- Khoman, Sirilaksana, and Dow Mongkolsmai. 1993. "Public-Sector ing schemes that are viable in rural or informal sector settings Health Financing in Thailand: A Synthesis of Findings." and the preconditions that are required for such schemes Research Report 93-04. International Health Policy Program, to succeed. Among other things, given the problems of Washington, D.C. assesinidby semi-subsistence Kiranandana, 'Thienchay 1990. Evaluation ofthe Health Card Project. assessing incomes in areas dormnated by seml-subslsten(In Thai.) Ministry of Public Health, Bangkok. or informal sector employment and the need to avoid cre- Krongkaew, Medhi. 1995. "Thai Society and the Distribution of ating excessive fiscal burden, a workable system could be Opportunity and Income." Paper presented at the annual sym- based on some form of community financing, such as the posium organized by the faculty of economics, March 30-3 1, Health Card Program. But certain preconditions must be Thammasat University, Bangkok. in place for such a scheme to get off the ground. Kultap, Praneet. 1983. "A Study of Industrial Injury: A Case Study The most important of these preconditions is sufficient of the Textile Industry" M.A. thesis. Thammasat University, faculty of economics, Bangkok. s^,ocial capital. The concept of social capital stems from the Manopimoke, Supachit. 1995. "Voluntary Health Insurance in recognition that social actions and the development of social Thailand: Development and Achievement." Paper presented organizations-such as a community health scheme- at the Regional Conference on Health Sector Reform in Asia, depend on expected mutual benefits that are affected by held at the Asian Development Bank, May 22-25, Manila. social norms and obligations within a community Social Mongkoismai, Dow. 1993. "The Social Welfare for Health Care." capital is accumulated through the coordination and coop- Paper presented at the National Workshop on Health Financing erapitaionsof aumuter wthinrsocial stheucturessuh and fp- in Thailand, November 12-13, Phetburi. eration of members within social structures such as fami- Mongkolsmai, Dow, and Sirilaksana Khoman. 1993. "Study of lies, associations, clubs, and communities. The closer a Beneficiaries of Public Hospitals in Thailand." Research Report community is, the stronger are the norms that require com- 93-02. International Health Policy Program. [city?] pliance; and the more stable a community is, the stronger Mongkolsmai, Dow, Plearnpit Satsanguan, Anong Rojvanit, is the likelihood of social capital accumulation. Moreover, Sirilaksana Khoman, Praphatsorn Leopairote, and Thanakorn social structures that are well-endowed with social capital Ungswad. 1994. "Features ofVoluntaryHealth Insurance: Case are moretamenable to coarective well-en .owet the smaal num Study of the Health Card Project for Selected Population are more amenable to collective action. Yet the small num- Groups in Rachaburi Province." Research report submitted to ber of members required for community trust and cohe- the Thailand Ministry of Public Health, Office of Health siveness-the building blocks of social capital-is less In.urance, Bangkok. INNOVATIONS IN HEALTEI CARE FINANCING 192 Patichon, Preeya. 1995."Choice of Hospital and Utilization of Thailand Ministry of Public Health. 1980. Rural Health Division Medical Services Among Insured Persons Under the Social Survey Report. (In Thai.) Bangkok. Security Scheme in Nonthaburi Province." (In Thai.) Master . 1992. Evaluative Reporton Social Security. (In Thai.) Office of Economics thesis. Thammasat University, faculty of eco- of the Permnanent Secretary, Bangkok. nornics, Bangkok. - . 1995. Public Health Statistics 1993. Bangkok. Population Studies Institute. 1988. "Thailand Demographic and Thailand Ministry ofPublic Health and Faculty of Health Sciences, Health Survey' Chulalongkom University, Bangkok. Mahidol University. 1988.A Study ofthe Coverage ofFree Medical Rojvanit,Anong. 1995. "The Epidemiologyof Inequity: Preliminary Care Servicesfor the Low-Income People in Thailand. (In Thai.) Findings." Paper presented at the Intemational Health Policy Bangkok. Program Annual Meeting, October 2-7, Beijing. Thailand National Statistical Office. 1990. Survey of Population Satsanguan,Pleampit, and Praphatsom Leopairote. 1992. "A Study Change 1987-88. Bangkok. of Unit Costs in Public Hospitals in Thailand." Research Report Thailand Office of the Prime Minister. 1995. Government Budget 92-02. Intemational Health Policy Program, Washington, D.C. 1995. Bangkok. Tantiserani, Pichai. 1988. "Health Card in the Rural Areas." (In World Bank. 1993. World Development Report 1993: Investing in Thai.) Paper presented at the Workshop on Health Insurance Health. New York: Oxford University Press. System for Thailand, August 18-19, Pattaya. RURAL HEALTH CARE FINANCING IN THAILAND 193 Market-Based Reform of U.S. Health Care Financing and Delivery: Managed Care and Managed Competition Alain C. Enthoven T he U.S. health care financing and delivery system between insurer and provider, is quickly disappearing. This is undergoing a profound and rapid transformation paper describes the traditional system as a point of depar- from a model characterized by fee-for-service pay- ture and as a point of contrast to the new system. It then ments, indemnity insurance coverage of completely free describes the emerging paradigm of managed care and man- choice among providers, and a fragmented and nonac- aged competition. countable delivery system, to a model made up of com- peting, integrated, accountable comprehensive care systems The Traditional System and Its generically known as managed care. This transformation is Consequences not being driven by public policy, which for the most part has been protective of the status quo. Rather, it is being The traditional U.S. health care system was the product of driven by consumer and employer choice and market forces interactions among different groups pursuing their own in the private sector. interests without any overarching public policy to guide Indeed, the large government health care programs- them or general agreement on the goals of the system. As Medicare (the federal government's program for the elderly a result the system ultimately became an example of pro- and the disabled), Medicaid (the federal and state govern- found and multifaceted market failure. ments' programs for some of the poor), and the Veterans Administration Health System-have barely started the Natural market failure transformation and will be the last to complete it. (In July 1994 just 7 percent of Medicare beneficiaries and 8 percent The market failure began with what might be called nat- of Medicaid beneficiaries were in health maintenance orga- ural or inherent market failure. As Arrow (1963) noted, nizations; InterStudy 1995; Health Care Financing Review the incidence of illness and the efficacy of treatment are 1996) The health maintenance organization (HMO) rep- very uncertain, creating a natural desire among risk-averse resents the more advanced type of managed care. (HMOs people to insure against large and unpredictable medical are defined below.) In 1980 about 9 million Americans were expenses. But with insurance comes moral hazard-that enrolled in HMOs. At the end of 1996 about 65 milion is, insured people become unconscious of the costs of the were, and the number is growing by about 12 percent a year. care that they receive. Over time the prices and standards Preferred provider insurance, a minimal form of managed of U.S. health care adapted to a market of cost-unconscious care, almost did not exist in 1980. Today, it covers about 90 patients. Information on medical conditions and their treat- miflion Americans. The traditional model of indemnity insur- ment is costly, and doctors and their patients have very dif- ance, characterized by a lack of any contractual relationship ferent amounts of it. Patients depend on their doctors for Alain C. Enthoven is the Marriner S. Eccles Professor of Public and Private Management in the Graduate School of Business of Stanford University. 195 advice and for treatments, and doctors, having a financial lack information on the quality, price, and treatment pat- interest in treatment decisions, are imperfect agents. terns of other doctors, they are unlikely to be able to Complicating matters, different people have very dif- negotiate with their doctor as an equal. ferent medical needs. In a voluntarymarket for health insur- * Solo practice-meaning that multispecialty group prac- ance, the healthy do not want to be pooled with the sick. tice was resisted because such a group could break the And pooling arrangements can be exploited by insurers seamless web of mutual coercion (through referrals) that selectively offering products that are more attractive to the the profession used to maintain the guild model. healthier members of the pool, effectively isolating the sick Moreover, group practice introduced an important ele- from affordable coverage. Finally, in the U.S. mixed but ment of quality management through peer review. Doctors partly free market for health insurance, there is a safety net who deviated from the solo practice model were denied in the form of county hospitals, laws requiring hospitals to hospital staff privileges, medical society membership, evaluate and stabilize uninsured patients who appear at their and referrals. From the 1940s through the 1960s, what doors, and a great deal of free care given by doctors and we now call HMOs were few in number and compara- hospitals. Moreover, health care coverage has become costly tively small, and experienced intense opposition from relative to the incomes of many low-income people. In a organized medicine (Weller 1984). voluntary system of health insurance these problems encour- The enforcement of these principles prevented the devel- age "free riders." As a result about 40 million Americans opment of an ordinary economic market in which alterna- now have no health insurance (Employee Benefit Research tive approaches to cost-effective care could develop (Stephan Institute 1996). 1978). "Guild free choice": Doctor-created morket failure Other humon-mode morket failures These market failures were compounded by a system of These market failures were reinforced by the behavior of health care finance and organization created and enforced employers and of organized labor. Most Americans who by the medical profession, a system characterized by Weller are not retired, disabled, or poor get health insurance through (1984) as "guild free choice." The principles of the med- their employers (Employee Benefit Research Institute 1996). ical guild, in the United States and other countries, had sev- From the 1940s through the 1970s (and even beyond), most eral defining features: employers offered their employees no choice of health insur- F Free choice ofprovider-meaning that every health insur- ance plan. It was an uphill battle for HMOs to persuade ance plan let every patient freely choose among doctors employers to offer them. Employers saw no advantage to and hospitals for covered services without discrimina- offering a choice among plans and preferred to stick with tion. This approach destroys the bargaining power of the fee-for-service model. Employers that did offer choices insurers: they cannot tell doctors that their enrollees typically paid in full the cost of the fee-for-service plan, will not be covered for the doctors' services if the doc- depriving employees who chose HMOs the economic tors do not agree to price and quality controls. rewards of choosing a more economical health plan. * Free choice of treatment-meaning no practice guide- Employers avoided HMOs because their premiums were lines or quality management. often higher, since in most cases HMO coverage was more * Fee-for-service payments-meaning that no outside entity comprehensive than fee-for-service coverage (including doc- can control doctors' incomes. Doctors could always earn tor office visits, preventive services, and the like). The financ- more by doing more, regardless of whether more was ing system created open-ended demand and a perpetual necessary or beneficial to the patient. shortage of doctors, so HMOs did not have much bargaining * Direct negotiation offees between doctor and patient- power when it came to doctor incomes. Organized labor meaning that if patients are in pain or worried about their saw comprehensive health insurance coverage as a bar- health, depend on the good will of their doctor, and gaining prize and demanded that employers pay the full cost INNOVATIONS IN HEALTH CARE FINANCING 196 of fee-for-service coverage. Trade union leaders acted as if hospitals for care that might appropriately occur in vari- they believed that health insurance premiums came at the ous settings, but the different providers were paid out of expense of employers rather than wages. Employers and separate accounts, and there was no management control labor failed to see health services as a purchasing problem or responsibility to see that the pieces were put together in which they had a common interest in getting value for effectively or economically. The system was nonaccount- their money. Moreover, most employers were too small to able. Providers could waste resources or make decisions spread risks and serve as effective purchasers of health that were bad for the quality of care without negative care coverage. Market failure is particularly severe in the consequences. There was no contractual link between market for small employment groups and individuals. providers and third-party payers. (The patient was the first Insurance carriers compounded the market failure, using party, the provider the second, and the payer the third.) marketing strategies and "product design' to try to select Except for a few HMOs, no one was responsible for link- the best risks or to avoid the worst risks. They differenti- ing resources to the needs of the population served. The ated coverage contracts in an attempt to create inelastic government created health systems agencies that were sup- demand by segmenting markets and making it very costly posed to do comprehensive health planning at the com- for consumers to compare value for money and to switch munity level, but no one knew how to do health planning; insurance plans. the agencies could not overcome the cost-increasing incen- Finally, government was a major contributor to market tives described above, and they were ineffective (Enthoven failure. First, federal and state governments excluded (with- 1980). Hospitals were largely independent nonprofit orga- out limit) employer contributions to employee health insur- nizations that competed for doctors by offering more ance from the taxable incomes of employees. Thus employers advanced technology and convenience. (Such competi- and employees had a strong incentive to agree on more, tion was referred to as the "medical arms race.") The rather than less, costly insurance plans (Enthoven 1984 and links between doctors and hospitals were weak. Many doc- 1985). Medicare and Medicaid were frozen in the "guild tors practiced in two or three hospitals and played them free choice" model from which they are only now begin- off against each other. ning to emerge. Until 1983 Medicare paid hospitals for inpa- Information on treatments and costs was decentralized tient services on a retroactive basis. Some U.S. states and to the local care site. Doctors had records of what went on the federal government passed laws requiring that all insur- in their offices, hospitals had records of what went on in ance contracts cover the services of providers who sought the hospital, pharmacies had records of prescriptions, and protection from the legislatures (for example, insurers in so on. But outside of HMOs, nobody had access to the com- Rhode Island must cover pastoral counseling), even though plete picture. Outcomes were rarely systematically followed competent buyers and sellers would have preferred less outside the context of a few studies. In the absence of costly coverage that did not include those services. And comprehensive longitudinal records that stored informa- finally, the federal government generously subsidized the tion on all the treatments a patient had received, there was expansion of medical schools and hospitals and the train- no way to link outcomes to treatments. ing of specialists far beyond what was needed, and acade- One consequence of this system was cost-unconscious mic health centers became important protectorates for some (or inelastic) demand. Providers had incentives to resolve members of Congress. all doubts about care by doing more. A phenomenon devel- oped known as "physician-induced demand"-that is, where Consequences of market failure there were more surgeons per capita there was more surgery per capita, unmediated by a reduction in price. Because of The result of these market failures was that, although health their superior information, physicians could shift the demand care is the joint product of many components (doctors, curve for their services (Fuchs and Kramer 1972, Fuchs hospitals, laboratories, pharmacies), the U.S. health care 1978). Providers had little responsibility for quality or cost. system was fragmented. Patients depended on doctors and There were, and remain, wide variations in medical prac- MARKET-BASED REFORM OF U.S. HEALTH CARE FINANCING AND DELIVERY 197 tice. Wennberg and Gittelsohn (1982) found tenfold vari- Manoged care: Organizing the supply side ations in the incidence of medical practices with no mea- surable difference in need or health outcomes. This finding The main origins of managed care in the United States can suggests that many people are being overtreated or under- be traced to several sources. First, there was the prepaid treated. group practice movement, whose foundations were laid in Another consequence was too many hospitals and beds, the years after World War II with the creation of the Kaiser too many specialists, and too many specialized facilities. For Permanente Medical Care Program and Group Health example, California has 120 hospitals that perform open- Cooperative of Puget Sound (Somers 1971). Kaiser heart surgery, half of them with annual volumes of fewer Permanente was the direct descendant of the medical care than 200 cases. There is also a great deal of inappropriate programs organized in the 1930s and 1940s to care for work- care. Winslow (1988) found that about one-third of the ers in Henry J. Kaiser's industrial enterprises. These non- carotid endarterectomies performed in the United States profit organizations combined multispecialty group practice, were inappropriate-that is, the patients would have been per capita prepayment, voluntary enrollment, and physician better off without them. There was little effective quality responsibility for the management of care. These organi- management. And by 1994 U.S. spending on health care zations are described in greater detail later in this paper. accounted for 13.7 percent of GDP, far more than any other Second, in some communities in which prepaid group country (Levit and others 1996). This level of spending seri- practiceswere successful and growing, doctors in fee-for-ser- ously strains public finances, and puts health care coverage vice solo practice were feeling competitive pressure. They out of reach for many families of moderate means. formed individual practice associations through which they could offer patients the financial equivalent of the prepaid Managed Care and Managed Competition: group practices while continuing to practice in their own Correcting Market Failure and Getting the offices and also see patients with traditional insurance. Incentives Right (These associations became humorously known as "defen- sive alliances against Kaiser.") Managed care and managed competition are strategies used The term health maintenance organization was coined in by the purchasers of health care services to: 1970 by Dr. Paul Ellwood as part of a national strategy to * Create health services delivery organizations capable of solve America's problems of uncontrolled health expendi- acquiring appropriate health care resources, obtaining ture growth, fragmentation, and lack of accountability by value for money, deploying the resources to care for fostering competition among nongovernmental compre- enrolled populations, designing and executing care hensive care organizations (Ellwood, Anderson, and McClure processes that produce good outcomes and value for 1971). In 1973 Congress passed the HMO Act, which money, and measuring and monitoring performance (out- defined HMOs as being of either the group practice or indi- comes, satisfaction, and cost) and continuously improv- vidual practice variety; provided grants and loans to help ing it (that is, managing care). start nonprofit HMOs; required that employers with twenty- * Develop a framework of incentives for such organiza- five or more employees that were offering traditional insur- tions to improve quality and lower costs. ance offer to their employees the choice of one group practice * Use market forces to transform the health care delivery and one individual practice HMO as alternatives to tradi- system from its fragmented, nonaccountable mode to tional health insurance (if such HMOs served the areas efficient, integrated, comprehensive care organizations where their employees lived and asked to be offered); and constantly striving to improve (Enthoven 1988 and overruled state laws that inhibited HMO growth. This act 1993a). had an important effect in opening the market to compe- In brief, managed care organizations are the players on tition. this field, and managed competition refers to the rules of Seeking to bring soaring health care costs under con- the game in which they play. trol, some employers wanted to be able to offer their employ- INNOVATIONS IN HEALTH CARE FINANCING 198 ees health insurance based on selective provider contract- concurrent review). A more advanced form of utilization ing-that is, insurance that resembled the traditional model, management is based on the recognition that medical uncer- except that employees would be offered preferential terms tainty is often great and practice variations are wide. Teams of coverage if they used contracting providers. Such con- of physicians study particular medical conditions, review tracting enabled employers and insurers to negotiate prices the medical literature, analyze their own data, and recom- and utilization controls with providers. But until 1982, in mend practice guidelines based on consensus within the compliance with the principles of "guild free choice," this team. These guidelines typically reflect the least costly way kind of insurance was illegal under the insurance codes of of achieving the best possible outcomes. most states. In 1982, in a major legislative battleinCalifornia, The third principle is negotiated payment. The basic idea employers, insurers, and labor unions joined forces to defeat is to trade patient volume for better prices. Compared the California Medical Association and secure the enact- with the usual fees in the fee-for-service system, managed ment of new legislation allowing insurers to contract selec- care organizations typically obtain discounts of 20 to 40 per- tively and pass the savings on to the insureds. Most other cent. These negotiated payments often include some states followed. This move authorized preferred provider bundling of services-for example, all-inclusive payments insurance, the other form of managed care. perinpatient day (for different types of patients) or per inpa- tient hospital case. Essential principles of managed care. Managed care has The fourth principle is quality management. For exam- four essential principles. The first is selective provider con- ple, a managed care organization is likely to survey patient tracting. Insurers can select providers for quality and econ- satisfaction. They may reward providers who score well with omy. Quality is important because employers care about the bonuses, and may not renew contracts with providers who health and satisfaction of their employees (or, if they do not, score poorly. Sophisticated managed care organizations will trade unions are likely to), insurers care about their repu- attempt to measure outcomes of care or performance of tations, and mistakes cost money More often than not, qual- processes of care and report them to consumers and pur- ity and economy go hand in hand. Providers are also chosen chasers. for their willingness to cooperate with a managed care orga- nization's quality and utilization management programs, Preferred provider insurance. Minimal managed care. and its reporting requirements. Preferred provider insurance is the form of managed care The second principle is utilization management. This most like the traditional model. (Preferred provider organi- varies from the crude to the sophisticated. For example, zation is sometimes used to parallel the better-established some managed care organizations have retained actuarial health maintenance organization.) In most cases such insur- consulting firms to develop guidelines on how long various ance is not provided by medical care organizations; rather inpatient cases should remain in the hospital, and these it is provided by insurance companies that contract with guidelines are translated into limits on what the insurance large numbers of providers that are not otherwise related will pay. Many managed care organizations employ "primary (Boland 1985). care gatekeepers"-primary care physicians who control The typical preferred provider insurer contracts with a referrals to specialists. Many managed care organizations large number of doctors, hospitals, laboratories, home health dealing with doctors from the fee-for-service sector (who agencies, and the like. It creates incentives for insured are thought to be overutilizers) require preauthorization patients to choose contracting providers. For example, the before a nonemergency patient can be hospitalized. A man- insurance contract might pay in full for the services of con- aged care insurance contract may include a deductible for tracting providers, but pay only 80 percent of what it would hospitalizations that is waived if the patient obtains autho- have paid contracting providers for the services of non- rization. Some managed care organizations employ utiliza- contracting providers; the patient must pay the rest. The tion management nurses to check on the hospital inpatient's insurer negotiates discounted fees, and the provider agrees condition and plan prompt discharges (a process known as to accept those fees as payment in full from contracting MARKET-BASED REFORM OF U.S. HEALTH CARE FINANCING AND DELIVERY 199 patients. Finally, the insurer adopts utilization management The HMO contracts with employers and individual sub- tools such as preauthorization for hospital admissions, scribers on the basis of per capita prepayment-that is, all length-of-stay guidelines, review of provider credentials, the medical care the patient needs for a periodic per capita and so on. payment fixed in advance and independent of the person's Some preferred provider insurers cover comprehensive actual use of services. Thus the HMO bears the full risk health care services. Others specialize, focusing on a sub- for the cost of medical care. The amount and type of risk set of comprehensive services such as mental health, phar- sharing with providers vary widely. But in most cases the macy, cardiology, or radiology. These specialized insurers HMO shares risk with providers, explicitly or implicitly. An serve as subcontractors to insurers that cover comprehen- explicit risk-sharing arrangement might be a contract with sive services. They can offer greater detailed knowledge of a medical group to provide all necessary professional ser- their specialty. And they may contract with several insurers vices for a fixed per capita payment. An implicit risk-shar- that cover comprehensive care and subcontract the com- ing arrangement might pay individual doctors on a ponents. discounted fee-for-service basis, but the HMO keeps track Preferred provider insurance often serves as an impor- of the per patient costs of each doctor, adjusted for age, sex, tant part of the transition from the traditional unman- and possibly diagnosis. Doctors whose costs consistently aged fee-for-service system to the HMO that uses per exceed norms might find themselves receiving extra coun- capita prepayment instead of fee-for-service. A group of seling on practice patterns, or not have their contracts doctors may begin with a discounted fee-for-service con- renewed. tract, acquire experience on which they can base a per There are now many HMO models, as the competitive capita payment, and eventually convert to per capita pre- marketplace has motivated a great deal of innovation in payment (see below). the search for better ways to organize and finance medical care. Historically, the first main category of HMO included Health maintenance organizations. A more fundamental prepaid group practice and "staff model" HMOs. Under change from the traditional system is represented by health the prepaid group practice model HMOs are based on a maintenance organizations (HMOs). The term bealth main- medical group that contracts with the HMO. The medical tenance organization was originally used to describe prepaid group accepts the risks of costs of care and usually rewards group practice, the main example of which was Kaiser the partners if the group is successful in managing costs. Permanente. It subsequently was applied to individual prac- Under the staff model the doctors are salaried employees. tice associations. Now the term is used to describe a remark- Though the two models are usually grouped together, there able variety of organizations. Some are based on is a difference. The doctors in the group model are more multispecialty group practices, some on doctors in individ- likely to see themselves as part owners of the enterprise ual practice, and some on both. Some are merely insurance and feel more responsible for its success. These are HMOs carriers that comply with the regulatory definition of HMO. "from the ground up." They attempt to organize compre- In general, an HMO is a health insurance carrier that hensive care systems. Their doctors care exclusively for covers a comprehensive list of health care services: physi- patients enrolled in their affiliated health insurance plan. cian and hospital care, laboratory testing, diagnostic imag- And they generally care for patients in HMO-owned or - ing, and usually prescription drugs. The coverage provides leased facilities. The prepaid group practice model has been for nominal copayments at the point of service-for exam- much more efficient and effective than the fee-for-service ple, Stanford University employees pay $10 per doctor office model, and in some states (California, Massachusetts, visit, but there is no deductible and no limit on the amount Minnesota, Oregon) its role was to force fee-for-service the HMO will pay for necessary acute care. Copayments providers to join network and individual practice associa- are not supposed to be so large as to constitute a barrier to tion models to compete. care. The HMO is supposed to control costs by managing The second category of HMO includes the network and care, not by imposing deterrent fees. individual practice association models, which contract with INNOVATIONS IN HEALTHI CARE FINANCING 200 former fee-for-service providers and enable them to compete ment controls. HMOs are empirically tuning their meth- with the group and staff models. In these models indepen- ods to find what works in their marketplaces. dent medical group practices and individual doctors contract A third category of HMO that is only just beginning to with insurance carrier HMOs to see the patients enrolled emerge is called thephysician-hospitalorganization, in which with those carriers, while also continuing to see patients one or a group of hospitals team up with their medical staffs enrolled in traditional insurance, Medicare, Medicaid, or to offer subscribers comprehensive health services on a per other or no coverage. The doctors continue to practice in capita prepayment basis. Physicians and hospitals are moti- their own offices. The medical groups are paid on a per capita vated to do so by what appear to them to be the large profit basis for professional services under contracts that include margins of for-profit HMOs, and by their perceptions of incentives for efficient hospital use. A typical contract might the way that they are treated by them. Such organizations provide a fixed monthly amount per enrolled person for reflect a desire by providers to control their own destinies. professional services plus a risk-sharing arrangement for hos- Thus, in principle, managed care could be introduced into pital costs. Individual doctors are paid negotiated fees for a country without a health insurance industry. services, given incentives for economical behavior, and are Of course, physician-hospital organizations will find that usually monitored for the economy of their practice pat- they need to develop functions and capabilities that are usu- terns. A typical contract might pay primary care doctors 80 ally associated with insurance companies, including the abil- percent of their fees soon after delivery of services, with the ity to enroll members, collect premiums, set prices on their other 20 percent withheld to ensure that enough money is services, make actuarial estimates of the costs of caring for left in the pool. At the end of the year the doctors are paid different groups, make arrangements for covered services in proportion to their billings if there is money left over. In that are beyond their capabilities (for example, to contract addition, the pool of primary care doctors may share in the with national or regional centers for advanced care), provide savings from efficient specialist referrals and hospital use. coverage for enrolled members when they are outside the The trend in these models is toward more integrated sys- organization's direct service area, and reinsure exceptional tems of care (see below). The original individual practice medical costs. And they will need financial reserves to cover associations were sponsored by county medical societies unplanned losses. Thus the physician-hospital organization and included every fee-for-service doctor who wanted to ends up creating (or partnering with) an insurance company. participate. Today's individual practice associations select Managed care, especially the HMO, requires a change doctors and drive hard bargains with them. For example, in patterns of access to doctors, from complete free choice U.S. Healthcare, founded in Pennsylvania, is based on to choice from among the managed care plan's contracting doctors in individual practice. It selects a panel of partici- providers. People who are used to the traditional system pating primary care doctors, evaluates them continuously often do not understand this change and the reasons for it. (through surveys of patient satisfaction, reviews of sample However, experience in California and other states with records and referral patterns, office inspections, and so on), high penetrations of managed care shows that people even- and pays them an age- and sex-adjusted capitation fee for tually get used to the new pattern. Surveys by large employer primary care services. Good performers win cash bonuses; coalitions in California report that employees are highly sat- poor performers are dropped from the program. U.S. isfied with some HMOs. Healthcare contracts with selected specialists on a discounted When they are seriously ill, some patients want to be fee-for-service basis, and gives primary care doctors as a able to go to a famous regional or national referral center group a financial incentive to control specialist referrals. and take their insurance with them. And they often do not HMOs translate the broad incentive of capitated pay- understand why they cannot. The comparatively free mar- ment into payment to doctors in a great variety of ways. ket for managed care in the United States has generated an Some pay salaries. Others pay salaries with bonuses for pro- innovation to address such concerns: a new kind of insur- ductivity, patient satisfaction, and overall economic success. ance plan called the "point-of-service HMO" that func- Others pay various forms of fee-for-service with manage- tions as an HMO for people who want to stay with the medical MARKET-BASED REFORM OF U.S. HEALTH CARE FINANCING AND DELIVERY 201 group that they have chosen within their HMO, but also ical care organization. The premiums that are paid reflect includes a preferred provider insurance plan, which the mem- the overall efficiency of the provider organization as well ber can access by paying a deductible (typically the first sev- as the health risks of the enrolled population. Per capita eral hundred dollars of expense) and a portion of each medical prepayment implies a reversal of the economic incen- bill. Such patients can also opt for a traditional insurance tives in the fee-for-service system. Doctors prosper by plan with somewhat stronger financial disincentives. These keeping patients healthy and by diagnosing and treating add-ons allow the covered person the full range of choice their medical problems promptly and effectively. Tertiary of provider, but grant more favorable financial terms for care (open-heart surgery, organ transplants, services that using the HMO's primary care group. Such arrangements are usually performed in regional referral centers, and are proving to be very popular. The point-of-service HMO so on), considered a major profit center in the traditional was introduced in the mid-1980s. In March 1987 eleven system, has become a "cost center." Under the tradi- HMOs reported point-of-service enrollment of nearly tional system filled hospital beds were an indicator of 400,000. ByJuly 1995, 318 HMOs served more than 5 mil- success; under the new paradigm empty beds are good. lion enrollees in point-of-service plans. Most people in these Per capita prepayment holds providers accountable arrangements end up staying with their primary care group for costs, and for the costs of poor quality. If an opera- for more than 90 percent of the services that they use. tion is done poorly and leads to complications and the In 1978 the HMO industry was made up almost entirely need for more treatment, providers pay the extra costs- of local nonprofit HMOs and Kaiser Permanente (then a not the insurers or patients. In that sense per capita pre- large national organization serving 3.5 million enrollees in payment can create incentives for high-quality treatment. six states). By 1995 thirty-five national HMOs served 42.7 Patients with unsolved or poorly managed problems con- million of the industry's 53.8 million enrollees. Some data tinue to impose costs on the health care system. Per capita on the HMO industry's growth are shown in tables 1 and 2. prepayment facilitates the alignment of doctors' incen- tives with patients' interests in high-quality, economical Integrated delivery systems: The seven integrations. The term care. It pays for and rewards cost-effective preventive ser- integration, as applied to health care financing and delivery, vices, such as increased outreach of prenatal care to reduce refers to the (at least) seven ways in which the system's com- thecosts of neonatologyandbettermanagementof chronic ponents are being linked more closely to realize important diseases to minimize acute episodes. Prepayment also gains in quality and economy. Integration need not mean provides a framework for cost-benefit analysis, helping common ownership. Indeed, the trend is toward integra- tion by contractual relationships. TABLE 2 1. Between financial responsibility and delivery, through per U.S. HMO industry composition: model types and capita prepayments by the purchaser to the chosen med- tax status, 1978, 1985, and 1995 (millions of members) TABLE I U.S. HMO growth: local and national firms, 1978, 1978 1985 1995 1985, and 1995 Model type Staff a 3.0 0.8 1978 1985 1995 Individual practce associations 0.6 6.4 22.1 Network n.a. 5.0 3.3 Number of HMOs Group 6.7 6.6 9.1 Local HMOs 183 298 192 Mixed n.a. n.a. 18.1 Branches of national HMOs 12 187 385 Total 7.3 21.0 53.4 Total 195 485 577 Tax status Number of members (millions) Nonprofit n.a. 13.6 22.1 Local HMOs 3.7 10.1 11.0 For-profit n.a. 7.4 31.2 Branches of national HMOs 3.6 10.9 42.7 Total n.a. 21.0 53.3 Total 7.3 21.0 53.8 Total_7_3_21_0_53-8 a. Included in group model. Source: InterStudy data. Source: InterStudy data. INNOVATIONS IN HEALTFI CARE FINANCING 202 to determine the most effective ways to spend limited cialists may serve as consultants to generalists who actu- resources. And it rewards cost-reducing innovation, such ally deliver the care. And it means efficient use of para- as the many innovations that have dramatically reduced medicals, for example, nurse practitioners and social the length of hospital stays for total hip replacement oper- workers to work in teams with primary care physicians. ations (Keston and Enthoven forthcoming). 5. Between doctors and hospitals, giving doctors a serious 2. Between providers and populations. This integration facil- interest in reducing hospital costs. Under the fee-for-ser- itates and encourages population-based medicine, which vice system doctors had incentives to behave in ways that adds an epidemiological perspective to encounter-based increased hospital costs. In a well-integrated system doc- medicine. Providers examine their encounters with tors develop practice patterns that facilitate efficient hos- patients to understand the underlying causes of the pital operations. They work with hospitals to reduce patients' complaints and to determine if there are effec- unnecessary record keeping. And they support "value for tive methods of prevention. One of the enduring leg- money" investments. ends of Kaiser Permanente is that when its founding 6. Among hospitals. Groups of hospitals in a region combine doctor, Sidney Garfield, was treating construction work- to share administrative support functions, induding man- ers who had nail-puncture wounds in their feet, he went agement personnel, to consolidate volume-sensitive clin- to the construction sitewith a hammer and pounded down ical services such as open-heart surgery, neonatology, and nails. Some HMOs now give children bicycle helmets and laboratories. videotapes explaining why they should always be used. 7. With patient information. In the traditional model infor- Thus HMOs can allocate resources to maximize the well- mation on diagnoses, treatments, and costs was scattered ness of their enrolled populations through preventive and throughout the system; integrated systems develop com- patient education services. Moreover, the defined pop- prehensive longitudinal records for each patient so that ulation base enables HMOs to match the numbers and every provider who contacts a patient can have a com- types of doctors in their groups, as well as other resources, plete picture of the patient's medical history. This to the needs of the enrolled population. approach helps to avoid duplicate tests and unfavorable 3. Among the full spectrum of health care services, including drug interactions. Moreover, the information can serve inpatient care, outpatient care, doctors' offices, and home as a basis for research on the relationship between diag- nursing, as well as with drugs and other services. Thus noses, treatments, and outcomes that is impossible to the resources that HMOs have been devoting to improved perform without such information. preventive services and outpatient care are more than being paid for by reduced inpatient costs. As noted, Controversy overmanaged care. Managed care has become HMOs seek to deliver care in the least costly appropri- extremely controversial in the United States even while (or ate setting. They are motivated to organize seamless com- perhaps because) it is growing with remarkable speed. prehensive care so that patients are not left to their own Concerns are expressed almost daily in the press. Perhaps devices when they leave the hospital. In the best-man- the greatest concern, often expressed by doctors, is that aged HMOs, committees of doctors and pharmacists the incentives of per capita prepayment will motivate doc- choose drugs that produce the best outcomes and min- tors to do less than they otherwise might do to improve imize total costs of care, not merely the least costly drugs. patients' health ("underservice" or "skimping"). This poten- 4. Among doctors and with other health professionals. This tial shortcoming is often seen as the mirror image of the means the right numbers and types of professionals and incentives for excess treatment under the fee-for-service the right mnix of specialists to ensure that patients have system. good access and to ensure that specialists are proficient For all the sound and fury, there is remarkably little evi- in caring for the patients they were trained to see. It dence of reductions in services at the expense of patients' means rational referral patterns and an efficient health, at least in established HMOs that serve the employed specialist-generalist division of labor. For example, spe- middle class. (Managed care has been more problematic MARKET-BASED REFORM OF U.S. HEALTH CARE FINANCING AND DELIVERY 203 when states have contracted with or created organizations appointments and unhappy surprises. For example, in an that specialize in caring fof the poor.) For example, a recent effort to reduce the inappropriate use of hospital emergency survey of the literature found that "the HMO and [fee-for- departments, HMOs have sometimes refused to pay non- service] plans provided enrollees with roughly comparable contracting providers for services that they did not consider quality of care, according to process or outcomes measures. to be related to genuine emergencies. What is needed (and Fourteen of 17 observations from 16 studies showed either are now emerging) are industry standards that define clearly better or equivalent (same or a mixture of better and worse) what is and is not covered in ways that people can under- quality-of-care results for HMO enrollees compared with stand and accept, and procedures for the prompt resolu- [fee-for-service] enrollees for a wide range of conditions, tion of disputes. diseases, or interventions" (Miller and Luft 1994, p. 1,516). The arrival of large-scale managed care in the United The framework in which managed care organizations com- States coincides with the recognition that the resources pete (managed competition, discussed below) needs to devoted to health care must be limited. With limited include measurement and oversight of quality, as well as dis- resources, some kind of management is inevitable. The incentives to skimp on or underprovide services, while choice is whether management will be done by govern- encouraging cost-reducing innovation-a delicate balance. ment as the single payer, with all its rigidity, insensitivity to Related to the underservice issue is a question of trust local conditions, and susceptibility to "pork barrel" politics in doctors. Some managed care arrangements (for exam- and corruption, or whether it will be done by competing ple, a typical prepaid group practice) leave salaried doctors private entities from which people can choose and that can in a neutral financial position between doing more and doing be held accountable by subscribers and government. less. But some arrangements give doctors strong financial Managed care isa complexbundle of innovative solutions incentives to provide less care, raising questions of whether to the problems that characterized the traditional fee-for-ser- people can trust their doctors. Compounding this prob- vice system. Innovation means trial and error; in hindsight, lem, a few HMOs indude wording in their contracts with some efforts turn out to be poor ideas that generate much-pub- doctors that are intended to limit what doctors can say to licized complaints. Doing managed care well is a complex their patients-so-called gag rules. There is nothing intrin- business. Not everyone understands it, especially when it is sic to HMOs that requires gag rules, and the HMO indus- growing so rapidly. "Good" managed care seeks to substi- try association opposes them. Apparently, this practice is tute better, more satisfying, less costly care processes; "bad" fast disappearing in the face of public protest and, in some managed care seeks to limit costs byiimposing across-the-board cases, legislation. numerical limits that apply poorly to individual cases. Over A related controversy inappropriately targets managed time, in an appropriately structured market serving informed care for something that is part of every health insurance people, the good will drive out the bad. But in the short run contract and public insurance program-that is, exclu- the poor performers may be confused with the good and try sions from coverage. The federal government's Medicare the public's patiencewith the entire endeavor. Countries con- program and private health insurers, whether managed or templating managed care should recognize that a great deal not, have typically excluded coverage of unproven or exper- oflearningwill be needed on the part of patients and providers. imental therapies, and individual tragic cases have inspired Of course, few developing countries will have to deal with intense controversy. The U.S. media has often confused this the expectations of a population that has had the freedom situation with managed care. and lack of personal responsibility for health care costs that Traditional insurance contracts were remarkably open in most U.S. citizens have had. their coverage. Insurers were usually not really at risk for Finally, followers of the U.S. debate over managed care the costs of care-they passed them back to employers or should remember that the growth of managed care in the to the government. So they did almost nothing to control United States is very much to the economic disadvantage costs. Under pressure to cut costs, insurers are tightening of doctors, nurses, hospital workers, and other providers their definitions of covered services, often leading to dis- of care, who now find themselves in a more normal eco- INNOVATIONS IN HEALTII CARE FINANCING 204 nomic market than the one they previously enjoyed. This Managed competition: Organizing the demand side is bound to influence their judgments about managed care. Managed competition is a market-making function per- For-profit and nonprofit arrangements. Until about 1980 formed by large group purchasers of health insurance cov- the HMO industry was dominated by nonprofit organiza- erage such as large employers, coalitions of large employers tions. Since then for-profit carrier HMOs have grown much (such as the California-based Pacific Business Group on faster than nonprofits (see table 2). In 1995 for-profit car- Health), coalitions of small and medium-size employers riers covered 58.5 percent of HMO members. The situa- (such as the Health Insurance Plan of California, a pooled tion is complex, however. purchasing cooperative created by the state for employers For example, Stanford University employees are covered with three to fifty employees), the federal government for by four HMOs. One, Kaiser Permanente, is made up of a Medicare beneficiaries as well as for its own employees, and nonprofit insurance plan, a nonprofit hospital corporation, state governments for state-sponsored Medicaid benefi- and the Permanente Medical Groups, which are organized ciaries as well as their own employees. These entities are as for-profit professional corporations. But the corporations' sometimes referred to as sponsors. In 1986 I introduced shares are held only by the doctors, and are not publicly these concepts with the following paragraphs: traded. Another HMO, Blue Shield, is a nonprofit carrier that markets the services of the nonprofit Palo Alto Medical Many proponents and critics of the competition idea Foundation and Stanford Faculty Practice Plan. (In another share the misconception that competition means a mar- community, nonprofit Blue Shield might contract with the ket made up of health care financing and delivery plans for-profit Columbia Hospital Corporation of America for on the supply side and individual consumers on the hospital services and with a medical group organized as a demand side, without a carefully drawn set of rules professional corporation.) The two other HMOs are designed to mitigate the effects of the market failures for-profit carriers that market the services of these two non- endemic to health care financing and delivery, and with- profit provider groups. Their shares are publicly traded. out mediation by some form of collective action on Inmyview Stanford employees arewell-served bythis mix. the demand side. Such a market does not work. It The nonprofits offer important benefits, such as more research cannot produce efficiency and equity. and community service. In 1980 there was large unmet demand Managed competition must involve intelligent, active for HMOs and cost-contained medical care. With a few excep- agents on the demand side, contractingwith health care tions, such as Kaiser Permanente, the nonprofit HMO sec- plans and continuously structuring and adjusting the tor did not expand fast enough to supply the demand. The market to overcome attempts to avoid price competi- for-profits, with their superior access to capital and stronger tion. I call these agents sponsors. .. . A sponsor is an incentives to grow, expanded to meet the demand, bringing agency that assures the members of a defined popula- the lively competition and lower costs we now enjoy. tion group of the opportunity to buy health care cov- The setup that works will depend on a country's busi- erage. The sponsor contracts with health plans ness culture, regulatory institutions, and so on. Developing concerning benefits covered, prices, enrollment proce- countries that are interested in creating a rapidly growing dures, and other conditions of participation.... Sponsors HMO industry will likely need to include for-profit com- may be public or private. (Enthoven 1986, p.106) panies in the mix because of their access to capital and stronger incentives to grow and innovate. Concems about Sponsors perform several functions to manage compe- these companies' performance in meeting social objectives tition, outlined below. can be addressed by regulation (community rating, rules governing access to doctors, and so on), contractual provi- Selecting the competitors. Sponsors must first decide sions with purchasers, and measurement and oversight by which managed care organizations and carriers should be employers, employer coalitions, and other payers. allowed to compete to serve the sponsored population. MARKET-BASED REFORM OF U.S. HEALTH CARE FINANCING AND DELIVERY 205 Criteria for selection should include quality of care, finan- Next, subscribers need to be fully responsible for pre- cial solvency and stability, competence of management, mium differences. That is, if subscribers choose plans with and willingness to operate in support of the goals of the higher premiums, they should be required to pay the full program. U.S. employers have had an important advan- difference so that they have an incentive to seek value for tage in this role. They can contract at will and use their money. This principle is now observed by about one-quar- judgment to decide subtle issues of quality and employee ter of American employers. In the United States a compli- satisfaction. cating factor in creating subscriber responsibility for In general, no health plan has an entitlement to be offered, premiums is the fact that employer-based health insurance and an employer can decide not to renew a contract with- contributions are tax-free to the employee, without limit. out having to prove poor performance in court. However, As a result, at the margin, the government is subsidizing employers have exhibited deficiencies in selecting com- people who choose more costly health plans. This short- petitors, one of which is to prefer one or two carriers to coming could be corrected by a limit on the tax-free amount, more choices because doing so creates less work for the set at the premium of the low-priced plan. employee benefits office. Government purchasers are usu- Next, sponsors should standardize coverage contracts. ally constrained by laws and rules that preclude the exer- Ideally, this would mean one standard contract offered by cise of judgment. (However, the statute establishing the all the health plans competing in a sponsored group. Federal Employees' Health Benefits Program freed this pro- However, HMOs rely on provider incentives and care man- gram from normal procurement laws and regulations and agement to control costs, while preferred provider insur- allows its managers to exercise discretionaryjudgment, which ance relies more on patient cost sharing (coinsurance, they have done with some distinction.) Public purchasers copayments, and deductibles) to deter excess use. So in a usually have to specify the criteria and then contract with group in which HMOs and preferred provider organiza- any organization or carrier that meets them- even if they tions compete, it may be necessary to offer two coverage are poorly qualified in the judgment of the officials man- contracts; one for each type. And if some of the HMOs aging the program. Moreover, public officials usually must offer a point-of-service feature, there may be a need for try to enforce the letter of the contract in a situation in which three. Standardization simplifies the choice, shifting atten- the 'product" is difficult to define. The buyer-seller rela- tion from details of coverage to overall quality and price. tionship gives the sponsor a tool for enforcing good-faith Standardization also lowers the information costs of switch- compliance with the spirit of the contract even when its ing plans, and is an important way of combating market seg- terms are imprecise. mentation and making different health plans better substitutes. Creating price-elastic demand. To create price-elastic Next, sponsors must provide information on the qual- demand, sponsors begin by organizing coordinated annual ity of care and service. (Sponsor measurement of quality is open-enrollment periods, which give all participants an discussed below.) However, the information requirement opportunity to consider alternative health plans, review should not be overstated. HMO competition has worked information about their prices and performance, and change quite well for decades in some large U.S. employment groups plans if desired. (Experience in the United States has shown with no formal quantitative measures of quality or perfor- that annual enrollments work well; in principle, a longer mance. People got the information they needed by asking period could also work.) To a point, a longer list of com- their friends. The need for performance information applies petitors creates more substitutes, and closer substitutes equally to a competitive market model and a govern- help make demand more elastic. Beyond that point, more ment-managed monopoly if the monopoly is to do a good choices, cspecially if they merely replicate the choices already job of managing. being offered, may create a bewildering array of options Finally, for demand elasticity and other reasons, there that raises switching costs and discourages people from should be no waiting periods or exclusion of coverage for making changes. preexisting conditions. When people switch plans, they INNOVATIONS IN HEALTH CARE FINANCING 206 must be fully covered as of the day the coverage becomes Insurance Plan of California, a purchasing pool for small effective. employers, has installed a risk-adjustment system based on age, sex, geographic area, and medical diagnostic informa- Managing risk selection. The incentives of health care orga- tion obtained from hospital records (Health Insurance Plan nizations need to be directed toward giving better care at of California 1995). However, this program is far too small lower cost, not toward selecting risks. Without careful design to have a significant effect on the incentives of the partic- to offset risk selection, avoiding the coverage of poor health ipating health plans and their behavior. For risk adjust- risks can be the shortest route to profitability, and it is impor- ment to really work, a majority of the market would have tant that managed competition design against this. to be using it. Still, the Health Insurance Plan of California's The first component of this design is to have the enroll- experience shows that the practical issues of doing risk ment process run by an independent clearinghouse rather adjustment are manageable. than requiring would-be subscribers to deal directly with competing health plans, because in the latter case there Establishing equitable rules for enrollment and pricing. would be opportunities for the plans to be hospitable to Certain rules of enrollment and pricing need to be applied apparent good risks and inhospitable to poor risks. An essen- within each sponsored group (for example, employees of tial rule of enrollment is that any member of the sponsored the U.S. government). These rules are usually embodied in group can join or re-enroll in any participating plan that he the contracts between health plans and sponsors. First, every or she chooses. The health plans cannot pick and choose covered person is guaranteed the right to enroll in the plan enrollees. Any attempt to do so should be viewed negatively of his or her choice, and to renew coverage at each annual and punished by the sponsor. enrollment. The second component is to standardize the coverage Second, the enrolled population is divided into rating contract, because nonstandard contracts can be used to categories, and the health plans quote uniform prices for select risks. everybody in a category. For example, at Stanford University The third component is risk-adjusted premiums. That we have separate premiums for single adult employees, for is, the health care costs of the populations enrolled in the single parents with one or more children, for couples with- different plans need to be estimated, based on demographic out children, and for families with children. All the single and diagnostic factors, and compensatory payments made adults enrolled in a particular health plan pay the same from surcharges on the premiums of the plans enrolling premium, regardless of their health status. The federal the good risks to the plans enrolling the bad risks so that government simply uses "individuals" and "families" for its risk selection is removed from the competitive market. For employees. The Health Insurance Plan of California also example, health plans that enroll a disproportionate share uses age categories, because in their market environment of diabetics should be compensated so that they are not they could not succeed if they tried to make young groups punished in the marketplace, giving them no incentive to subsidize older groups. cut back their endocrinology department and make them- selves less attractive to diabetics. A great deal of sophisti- Measuring and reporting quality. Both to improve quality cated research has been done on this topic, but so far there and to make the market work better, major sponsors are has been little practical application (Weiner and others 1996; investing in efforts to measure performance and quality and Ellis and others 1996). to report it to consumers. First, they have developed sur- In fact, the competitive market in California has been veys of consumer experience and satisfaction. These sur- working quite effectively in recent years without risk adjust- veys range from pointed questions about quality to general ment, and there is little evidence of the negative effects that questions about satisfaction with doctors. (One of my favorite economic theory would predict. But real price competition examples is, "If you were hospitalized in the past year, did is in its early days in California, and these problems could you acquire any illness or injury in the hospital?") The become more serious as competition heats up. The Health California Public Employees Retirement System recently MARKET-BASED REFORM OF U.S. HEALTH CARE FINANCING AND DELIVERY 207 started analyzing the replies of people who had been hos- ers and about 1 million employees, retirees, and pitalized or were high users of care to see if they were as dependents). satisfied as the healthy consumers. The methods for such The situation is much more challenging for people who surveys need not be very complex. do not belong to large groups: small employers, the Second, the National Committee for Quality Assurance, self-employed, and the unemployed. To help small employ- a private nonprofit organization sponsored by employers ers, the states of California and Florida have created pooled and HMOs, has developed a data set that all HMOs sup- purchasing arrangements that small employers can join. The ply. Data include measures of preventive services, such as Health Insurance Plan of California is open to employers the percentage of two-year-olds who have had their required with three to fifty employees that agree to contribute at least immunizations, the percentage of women who have had half of the premium of the least costly managed care plan periodic mammograms, and the percentage of diabetics who on offer. About twenty-five managed care plans participate, have had an annual retina examination. In California these giving most employees a choice of six to eight plans serv- data are audited. The California Public Employees ing their area. The plan has worked well, although it has Retirement System presents these data to consumers in a not grown rapidly. Its main problem is a lack of a powerful "report card" format. financial incentive to hold the pool together and prevent Some states have systems that measure risk-adjusted the lowest-risk groups from splitting off and seeking lower mortality rates (that is, mortality rates from coronarybypass premiums on their own. graft surgery, childbirth, and the like, adjusted for the patient In a system of completely voluntary insurance, adverse mix). This will gradually become a widespread practice in selection makes such a scheme unfeasible for unsponsored the United States. individuals such as the self-employed and unemployed. If Finally, major purchasers (government and private participation is voluntary and is not motivated by access employers) have created the Foundation for Accountability to the employer contribution (or the tax subsidy that sup- to explore and endorse more advanced quality measures ports it in the United States), people who do not antici- related to specific medical conditions and outcomes. pate medical costs will choose not to enroll, while people who expect medical costs will. Costs and premiums will Poolingpurchasingand sponsoringtheunsponsored. Managed become prohibitive in a phenomenon known as a death competition is well suited to large employment groups that spiral. can mobilize the resources to manage, contract with health Such a scheme could work for small employers, the plans on more or less equal terms, enforce demands for self-employed, and the unemployed if there were a univer- information, and achieve economies of scale in adminis- sal voucher that could be used only to buy health insurance tration. In large employment groups the employer contri- through a group plan like the Health Insurance Plan of bution to employee health insurance is the glue that holds California. In the United States, however, we are nowhere the group together as a purchaser, preventing people with near that stage. the lowest health risks from splitting off and buying cheaper insurance on their own. Health insurers can only gain access Simple models can work. The preceding discussion may to the employer contribution by contracting with the give the impression that managed competition is necessar- employer. ily complex and must be supported by advanced data sys- The most successful managers of competition are large tems. This is not the case. Simple models of managed coalitions of employers such as the Pacific Business Group competition have done quite well. For example, a basic on Health (a California-based coalition of large private survey of patient experience and satisfaction can be very employers), a national multimarket consortium organized informative. And premium risk adjustment might be based by American Express, and the Health Benefits Program on easily measurable demographic factors plus the presence of the California Public Employees' Retirement System (a of a few costly chronic conditions. In fact, a general advan- pooled purchasing arrangement for 1,000 public employ- tage of decentralized private markets applies here: such INNOVATIONS IN HEALTH CARE FINANCING 208 markets can function with less information than is needed help patients make good decisions about the care they are by a centralized system. seeking. They are investing in call centers to shorten tele- phone waiting times and expedite appointments. The emerg- Consequences of managed competition. Large-scale man- ing standard of access to doctors is same-day appointments aged competition is a recent development in the United for patients who think they need them (with some doctor, States. It has been introduced more extensively in California if not one's own doctor). than elsewhere, led mainly by a few large employers and purchasing coalitions. Yet even though most employers in Downsides of managed competition. Effective managed California have not applied all or even most of the princi- competition requires a good understanding of how health ples outlined here, competition has become quite active. insurance and managed care markets work. One of the main In real terms the 1997 premiums for competitive HMOs problems has come from incomplete implementation of are about 15 percent less than in 1994. HMO premiums managed competition concepts-for example, if employ- have for the most part stopped rising across the United ment groups do not make their employees sensitive to pre- States, although recent newspaper reports have forecast mium costs, they can be disappointed if they offer a choice renewed increases (Freudenheim 1997; Wmslow 1997). among HMOs and do not see their costs come down Medical practice is changing rapidly, shortening hospi- (Enthoven 1993b). tal stays and finding ways to avoid them altogether. The sur- Over the longer run the incentive (created by a lack of plus of hospital beds has been exposed, and efforts are under good risk adjustment) to avoid or underserve people who way in California to find ways to overcome the political have costly chronic conditions is a matter of serious con- obstacles to closing hospitals. A few have been closed, and cern. When provider organizations enter this competitive it is likely that many more will be in the next few years. framework, unless risk adjustment is done well, they have (Too many hospitals can be bad for the quality of care because powerful incentives to make themselves unattractive to a minimum volume of patients is needed to ensure the patients with costly chronic conditions. This can happen in proficiency of health professionals.) The surplus of specialist subtle ways: generous pay and budgets for pediatrics depart- physicians is also being exposed, and specialists are leaving ments, which attract healthy young families, and tight bud- California. Service levels are improving. There is much inno- gets for endocrinology (diabetes) and infectious diseases vation to improve quality and service and to lower costs (AIDS). This is unfortunate because the patients with costly (Enthoven and Singer 1996). Nationally, HMO member- chronic conditions are the ones that have the most to gain ship has been growing by about 12 percent a year since 1990. from well-coordinated care. A wise public policy would The best managed care organizations are doing a great encourage good risk adjustment. deal to take advantage of the opportunities created by man- A related concern stems from the fact that Americans aged care to improve medical care. They are creating infor- move about and change jobs a great deal. Thus an HMO mation systems to study practice variations, measure considering a more costly but effective way of treating a dia- outcomes, identify and adopt best practices, create clinical betic today in order to forestall serious complications ten practice guidelines, and monitor progress. They do (or col- years from now has reason to believe that the benefits will laborate with organizations that do) ongoing technology occur when the patient is a member of another HMO. (This assessment. They inform patients about healthy behaviors, may mean that the quality measurement program should chronic disease management, and the risks and benefits of monitor such processes of care in detail.) alternative therapies. Some apply continuous quality improvement that employs interdisciplinary perspectives Managed competition without managed care. I have con- (Berwick, Godfrey, and Roessner 1990). sistently linked managed care and managed competition Facing competitive pressures in Califomia, HMOs are because I see managed competition in the United States as working hard to improve customer service. For example, a framework of incentives intended to drive insurers and they offer convenient access to "advice nurses" who can providers away from the traditional model and toward more MARKET-BASED REFORM OF U.S. HEALTH CARE FINANCING AND DELIVERY 209 effectively organized and managed systems of care. Yet some * Measures to make demand price-elastic, including peri- analysts have asked whether governments and other spon- odic choice, ease of switching, subscriber responsibility sors could use the concepts of managed competition to for premium differences, and information about com- improve the functioning of traditional insurance markets petitors. even in the absence of managed care. (By traditional insur- * Management of risk selection, including an independent ers I mean insurers that do not have selective contracts clearinghouse for enrollment, rules ensuring that con- with providers and therefore lack effective tools to modify sumers can choose their plan, standard contracts, and their behavior.) I believe that the answer is yes, but in that eventual and progressive implementation of risk-adjusted event insurers would be competing on their efficiency and premiums. profit margins, which account for about 10 percent of health * Equitable rules regarding pricing, access, and enrollment. insurance premiums, and on customer service. * Information on health plan performance for consumers. Ibelieve that competition on "retentions" (that is, admin- * Pooled purchasing arrangements (where needed). istrative expenses and profits as a percentage of premi- ums) contributed to undermanagement and inflation in Relevance for Developing Countries the traditional health care market. Certainly, it is appropri- ate to subject such services to competition, and some man- The relevance of this experience to developing countries is aged competition concepts are needed to make that happen. neither simple nor obvious. I certainly do not want to sug- For example, insurance contracts can be exceedingly com- gest that other countries should transplant U.S.-style man- plex to understand and compare. Insurers often use this aged care and managed competition to their health systems complexity to raise switching costs, to deter people from without a great deal of careful thought about which ideas making the effort to compare the alternatives and to con- would or would not work in their societies. sider switching. One danger is that the best way for health I am uncertain how these concepts would apply in devel- insurers to compete under those circumstances, without the oping countries. I fear the consequences of a naive belief ability to manage costs, is by selectingrisks. Apoorly designed that terms like competition, marketforces, and HMOs are incentive framework might drive insurers to great lengths magic incantations whose invocation will make things bet- to select risks. Another danger is that purchasers might ter without the long, hard struggle that is required to improve assume that the lowest administrative expenses are best, health care systems; such naive beliefs have plagued poli- leading to undermanagement of care. The absence of the cymaking in the United States. possibility of managed care greatly reduces the advantages In 1994 William Hsiao wrote a sobering editorial that of the private sector in health insurance. If only traditional began: fee-for-service is allowed, the government as a single payer has several advantages-economies of scale, simplicity in Propelled by a declining faith in government, many administration, efficiencies of pooling risks widely, monop- developing nations have searched for a "magic pill" to sony power-that could make that model more effective. cure the ills of their underfunded and inefficient pub- lic sector-dominated health systems. Allured by the Summary of managed competition. For managed compe- success of free market mechanisms in promoting eco- tition to work, several elements must be in place: nomic growth, conservative politicians and economists, * The existence of managed care organizations in suffi- starting in the early 1980s, pushed many developing cient numbers to make markets competitive. Awide vari- countries to turn to the free market to finance and ety of entities might participate in this competition. provide their health services.... The magic of marke- * Sponsors acting as active, intelligent agents that can tization often seduces governments into action with- create markets, contract with managed care organiza- out a critical understanding of the conditions required tions, set rules, monitor compliance and progress, and for efficient markets and with no reference to empiri- offer subscribers periodic choice. cal evidence. (p. 351) INNOVATIONS IN HEALTII CARE FINANCING 210 Hsiao cites four countries in which elements of what do so. (But it can bring lower costs, making it easier to pay might be called a market strategy were attempted-with for.) In the absence of carefully designed rules and strong poor results. He concludes that "to contain cost inflation institutions, it will not prevent health plans from seeking to and improve allocative efficiency, the government has to serve only the most profitable customers. regulate the use of expensive technology" (p. 356). Each country's health care system reflects its history, cul- One problem with Hsiao's conclusion is that in the United ture, and political, social, and economic systems. One coun- States our experience has been that federal and state gov- try cannot simply adopt another's health care system and ernments are incapable of regulating to improve allocative expect success. The managed care and managed competi- efficiency. Indeed, many of our worst inflationary excesses tion model is a good fit for the U.S. culture, which values can be traced to public policy. Perhaps other countries pluralism, diversity, multiple competing approaches, and have developed a superior form of government that is capa- individual choice and responsibility, is suspicious of con- ble of regulating to improve allocative efficiency. But I doubt centrated power and dislikes government, and draws on it. constitutional arrangements that favor free markets. The Indeed, Hsiao goes on to observe that: considerable cultural diversity within the U.S. health care system requires a pluralistic approach. public sector provision of health services often suffers There are, however, important features of the U.S. sys- from bureaucratic inefficiency, long waiting time, and tem that other countries should try to avoid. These include unresponsive public sector workers protected by their a lack of broadly based social insurance, an overemphasis unions. Patients may not get value for money from a on complex technologies that yield small gains in health sta- rigid government-run system with inefficient and deper- tus at the expense of primary care and prevention (we spend sonalized government clinics and hospitals. Often a cen- billions on neonatal intensive care but seem unable to ensure trally planned health delivery system also lacks.... the every pregnant woman good prenatal care), and an inabil- motivation to innovate. (p. 356) ity to organize collective action to make the health insur- ance market work well for small employers and people I would add pervasive corruption to Hsiao's list of public who want to buy their own coverage. However, managed sector problems. care and managed competition could be built into a system So, implementation of managed competition will not of universal coverage with equitable financing (Enthoven be easy, and good results will not come quickly-a serious 1978; Enthoven and Kronick 1989). Still, I believe that the problem in democracies with politicians whose horizon U.S. experience supports several insights that are of gen- extends only to the next election. These markets are com- eral relevance for developing countries. plex and interrelated. Policymakers must think through the entire system carefully, and be prepared to make mid- Change should be incremental course corrections as problems arise and progress unfolds. Implementing managed competition will not occur with- By incremental change, I mean continuous evolutionary out disagreement and struggle. In the United States man- change in which each step builds on the success of the pre- aged competition has been and remains intensely vious steps. Incremental does not necessarily mean slow. controversial, mainly because it is undermining the eco- For example, HMO membership in the United States is nomic interests of providers. The changes have come at a growing by 12 percent a year. This is a high rate of change, time when health spending is growing rapidly. Employers but it is a continuous process. and government were forced to respond. Incrementalism is one of the first laws of democracy. Managed competition is not a magic pill. It will not raise We make changes in small steps so that people can under- money from workers in the informal economy to finance stand and adjust to them. We avoid public actions that cre- their health care. It will not insure the uninsured unless ate large windfall gains and losses. Democratic government someone is willing to pay for it and can raise the money to cannot be seen as inflicting direct harm to people (Schultze MARKET-BASED REFORM OF U.S. HEALTH CARE FINANCING AND DELIVERY 211 1977). Institutions take time to develop. Under managed which has been so successful in California, is making some care and managed competition, people require a great deal fundamental changes to adapt to market conditions on the of time to learn how to deliver and how to receive man- East coast. Developing countries should adopt health care aged care. As noted above, managed care requires a great frameworks that allow for the development of multiple com- deal of learning, and managed competition requires changes peting approaches, so that they can try them all and see what in expectations. People in developing countries sometimes works best. speak of health care reform as though it were one single sweeping change after which no further change is needed. There are many routes to managed care "Big bang" reform that is sudden, wrenching change is unlikely to work well. It is better to think in terms of Managed care organizations can have a wide variety of ori- "mid-course corrections" in a process of continuous change. gins. In addition to the U.S. examples that I have presented, consider these: Pilot projects and multiple competing approaches are * Primary care or multispecialty clinics that are branches useful of the ministry of health might be transformed into free-standing nonprofit entities that accept capitation Theories about health system organization and manage- payment and compete with similar entities. ment are not strong enough to predict what will work in a * As has been occurring in the United Kingdom, groups completely new context. What looks and sounds good in of primary care physicians might be formed that accept theory may turn out to be bad in practice. In the 1992-94 a capitation payment for a long list of services (backed U.S. health care reform debate most academics and health up by reinsurance for high-cost cases). policy experts were skeptical that managed competition * Traditional insurers might be allowed to contract selec- could moderate the growth of health expenditures-even tively with providers and to offer patients incentives to after it had started to work. Social science research is always use contracting providers. backward looking. When prepaid group practice started in * Traditional insurers might study the performance and the United States, it was denounced by organized medicine prices of providers, select good performers, and offer and greeted with suspicion by most Americans. Today, Kaiser preferential insurance coverage to patients who use them. Permanente receives some of the highest consumer satis- This would be a form of preferred provider insurance faction ratings among California HMOs, and the rest of the with implicit contracts. (Until the legislative changes of health care system has had to change fundamentally to the 1980s the laws governing traditional insurance in respond to the competitive threat it poses. Through its prac- the United States prohibited insurers from discriminat- tical, demonstrated experience it has proved itself in a com- ing among providers.) petitive marketplace of health care and ideas. * Public hospitals might be transformed into free-stand- Individual practice associations started with built-in fatal ing nonprofit entities that accept capitation contracts, defects-namely, they were dominated by county medical in a manner reminiscent of the U.K. National Health societies that used them in an attempt to preserve the sta- Service Trust Hospitals. tus quo. They were not selective in their physician mem- * Large employers might create clinics for their employ- bership. Only out of failure came the fundamental changes ees and dependents, then open them up to the public that enabled them to compete effectively. They changed to on a prepaid capitation basis, in a manner reminiscent become selective in numbers and types of doctors and of the industrial origins of Kaiser Permanente. changed theirmanagement styles from"hands off" to "hands In addition to national managed care firms, for-profits, on." nonprofits, and mutual insurers, HMOs in the United States Finally, one size does not fit all. Different models work have been sponsored by businesses, community coopera- for different people in different circumstances in different tives, county governments, hospitals, multi-hospital corpo- parts of the United States. For example, Kaiser Permanente, rations, physician-hospital joint ventures, medical societies, INNOVATIONS IN HEALTH CARE FINANCING 212 physician group practices, unions, universities, and others per capita prepayment basis. HMOs make major changes (InterStudy 1995). to increase economic efficiency when they are subjected to competition. Ultimately, individuals and institutions will do A developmental approach can foster innovation and what is rewarded by society (especially what is rewarded by increase popular support more resources) and move away from behavior that is not rewarded. Health care reform does not have to happen all at once. There is no perfect incentive scheme. Managed compe- Indeed, it should not. Managed care and managed com- tition is an attempt to create a balance of incentives in petition in the United States should be seen as historical favor of providers producing value for money Incentives processes. They started with some doctors, employers, can be used to correct serious problems, as happened in consumers, and trade unions wanting to try a different the United States when growth in health care spending approach, prepaid group practice. To survive, prepaid group was excessive. New incentives may create new problems practice had to sell employment groups on the idea of the that require adjustments in the incentive scheme. individual employee having a choice of a plan. Americans believe in choice. As increasing numbers of people began Integration of delivery systems and care processes is choosing prepaid group insurance, fee-for-service doctors important began to suffer and so organized competing alternatives, individual practice associations. As explained above, the process of care is the orchestration Then the government made some strategic interventions. of many components for the purpose of improving the health First, the federal government offered its employees a wide of a patient. Effectiveness can be enhanced and costs reduced range of choice, including prepaid group plans and indi- considerably if the components are integrated well-that vidual practice associations. Then, in 1973, a lawwas passed is, if theywork together toward a common goal. Developing authorizing grants and loans to nonprofit HMOs, requir- countries should develop integrated systems in their over- ing employers to offer them as a choice and overriding pro- all strategies to see if they can realize these advantages. visions of state laws that inhibited HMO growth. Only in Managed competition is not a simple, static model. It is 1991 did my employer, Stanford University, adopt standard a journey guided by microeconomic principles, empirical benefits and make employees sensitive to premium prices. research, and careful observation of what works. Two years later, having seen that Stanford survived the change, the University of California system followed suit. References And so the story unfolds. The key is to create a climate in which desirable innovations can occur. Arrow, K J. 1963. "Uncertainty and the Welfare Economics of Medical Care." American Economic Review 53: 941-73. Incentives matter Berwick, D. M., A.B. Godfrey, andJ. Roessner. 1990. Curing Health Care. San Francisco, Calif.: Jossey-Bass. Boland, Peter. 1985. "The New Healthcare Market: A Guide to Perhaps because the subject has been dominated by health PPOs for Purchasers, Payers and Providers." Dowjones Irwin, professionals, much if not most of the world's thinking about Homewood, 11l. health system organization is uncontaminated by economic Ellis, R. P, G. C. Pope, L. I. lezzoni, J. Z. Ayanian, D. W Bates, insight. Doctors and nurses even consider it insultingto sug- H. Burstin, and A.S. Ash. 1996. "Diagnosis-Based Risk gest that economic incentives have anything to do with their Adjustments for Medicare Capitation Payments." Health Care behavior. But the fact is that economic incentives do make Financing Review 17(3): 101-28. behavior. But thatstateooEllwood, P M., N. Anderson, and W McClure. 1971. "Health a difference to the behavior of individuals and organiza- Maintenance Strategy." Medical Care 291 (May): 250-56. tions. The U.S. experience clearly shows that. Doctors for- Employee Benefit Research Institute. 1996. "Sources of Health merly practicing in the fee-for-service model change their Insurance and Characteristics of the Uninsured." Issue Brief behavior markedly when their services are purchased on a 70. Washington, D.C. MARKET-BASED REFORM OF U.S. HEALTH CARE FINANCING AND DELIVERY 213 Enthoven,A. C. 1978. "Consumer Choice Health Plan." TheNew Measures: Results of Simulation No.2, Revised." Sacramento, England Journal of Medicine 298 (12 and 13): 650-58 and Calif. 709-20. Hsiao, W. C. 1994. "Marketization-The Illusory Magic Pill." . 1980. Health Plan. Reading, Mass.: Addison-Wesley. Heath Economics 3: 351-57. . 1984. "A New Proposal to Reform the Tax Treatment of InterStudy. 1995. The InterStudy Competitive Edge, Part II: Industry Health Insurance." Health Affairs 3(1): 21-39. Report. Excelsior, Minn. .1985. "Health Tax Policy Mismatch." Health Affairs 4(4): Keston, VJ., and A. C. Enthoven. Forthcoming. "Re-Engineering 5-14. Total Hip Replacement: A Case History of Innovations to - 1986. "Managed Competition in Health Care and the Improve Quality While Reducing Costs." Health Care Unfinished Agenda." Health Care Financing Review (annual Management Review. supplement): 105-19. Levit, K. R., H. C. Lazenby, L. Sivarajan, and others. 1996. - 1988. "Theory and Practice of Managed Competition." "National Health Expenditures, 1994." Health Care Financing In Health Care Finance. Amsterdam: North-Holland. DeVries Review 17(3): 205-42. Lectures Vol. 9. Lipsett, S. M. 1991. Continental Divide. New York- Rutledge. 1993a. "History and Principles of Managed Competition." Miller, R. H., and H. S. Luft. 1994. "Managed Care Plan Health Affairs 12(supplement): 24-48. Performance since 1980." Journal of the American Medical -. 1993b. "WhyManaged Care Has Failed to Contain Health Association 271(19): 1512-19. Costs." Health Affairs 12(3): 27-43. - Schultze, C. L. 1977. The Public Use ofPrivate Interest. Washington, Enthoven, A. C., and R. Kronick. 1989. "A Consumer-Choice D.C.: Brookings Institution. Health Plan for the 1990s: Universal Health Insurance in a Somers, A. R. 1971. "The Kaiser-Permanente Medical Care System Designed to Promote Quality and Economy." New Program." Commonwealth Fund, New York. England Journal of Medicine 320(2): 29-37 and 94-101 Stephan,J-C. 1978. EconomieetPouvoirMedical. Paris: Econornica. Enthoven, A. C., and S. J. Singer. 1996. "Managed Competition Weiner, J. P, A. Dobson, S. L. Maxwell, K. Coleman, B. Starfield, and California's Health Care Economy." Health Affairs 15(1): and G.F. Anderson. 1996. "Risk-Adjusted Capitation Rates 39-57. Using Ambulatory and Inpatient Diagnoses." Health Care Freudenheim, M. 1997. "Health Care Costs Edging Up and a Financing Review 17(3): 77-100. Bigger Surge is Feared." New York Times, 21 January, p. 1. Weller, C.D. 1984. "Free ChoiceAsaRestraint of'TradeinAmerican Fuchs, V R. 1978."The Supply of Surgeons and the Demand for Health Care Delivery and Insurance." Iowa Law Review 69(5): Operations." Journal of Human Resources 12(supplement): 1351-91. 35-55. Wennberg, J. 1996. The Dartmouth Atlas of Health Care. Chicago, Fuchs, V R., and M. J. Kramer. 1972. "Determinants of Ill.: American Hospital Publishing, Inc. Expenditures for Physicians' Services in the United States, Wennberg, J., and A. Gittlelsohn. 1982. "Variations in Medical 1948-68." NBER Occasional Paper 117. National Bureau of Care among Small Areas." Scientific American 26(4): 120-34. Economic Research, Cambridge, Mass. Winslow, C. M., D. H. Solomon, M. R. Chassin, R. H. Brook, and Health Care Financing Review. 1996. "Medicare and Medicaid others. 1988. "The Appropriateness of Carotid Endarectomy" Statistical Supplement." New England Journal of Medicine 318(12): 721-28. Health Insurance Plan of California. 1995. "Methods for Winslow,1R. 1997. "HealthCareCostsMayBeHeadingUpAgain." Calculating and Applying RiskAssessment and RiskAdjustment Wall Street Journal, 21 January, p. B 1. INNOVATIONS IN HEALTH CARE FINANCING 214 Managed Care and Managed Competition in Latin America and the Caribbean Andre Cezar Medici, Juan Luis Londono, Oswaldo Coelho, and Helen Saxenian D issatisfaction with the performance of health care Managed care programs integrate financial responsibil- systems has led to an intense search for more effec- ity for and delivery of health care services. Instead of using tive means of using public and private resources solely fee-for-service payments, health care purchasers use to improve people's well-being. Policy innovations in this area techniques such as per capita prepayments to providers, depend to a large extent on the history and starting points which put providers at risk for the cost of services provided. of different health systems-particularly with respect to the Well-developed managed care programs also use utilization role of the state in the provision of health care services. In review and quality control management to contain costs and Europe, where the state traditionally has played an impor- ensure quality care. tant role, countries have started to question the extent of Advocates of managed care argue that it offers consid- the state's involvement in health care financing, and to look erable cost savings over traditional insurance and improves for more effective and equitable ways to separate the financ- the quality of health care. The realization of these poten- ing of health care from the provision of services. In countries tial benefits, however, depends on the context in which man- like the United States, where the private sector plays an active aged care develops-particularly the regulatory framework role in the funding and provision of health services, efforts and its enforcement. Managed care programs can grow with have been made to restructure private markets. Enthoven (in or without a regulatory framework based on managed com- this volume), for example, proposes various models of man- petition. Managed competition refers to a regulatory frame- aged competition for different types of systems. work (rules of the game) that uses competition and choice Over the past two decades managed health care programs to enhance the functioning of health care markets. 1 Managed in the United States have grown rapidly and received con- competition often refers to government regulation, but it siderable attention. Although many other countries also have can also refer to the regulatory framework that a large group managed care programs, they have received less attention. purchaser of health insurance puts into place. California's Such programs are becoming increasingly important in Latin Pacific Business Group on Health is an example of the lat- America and the Caribbean. Although these programs dif- ter (Enthoven in this volume). The performance of man- fer, they share several characteristics. They typically either aged care-in terms of its contribution to efficiency, coverage, provide services directly or use selective contracting to pro- and equity-depends on the nature of the market and on vide services to covered individuals. In addition, covered indi- the regulatory framework and its enforcement. Assuming viduals are more constrained in their choice of service providers a variety of forms and approaches, the concepts of man- than under traditional indemnity health insurance plans. aged care and managed competition are highly relevant for Andre Cezar Medici is health economist in Social Programs Division 1 and Juan-Luis Londonio is lead economist in the Office of the Chief Economist at the Inter-American Development Bank. Oswaldo Coelho is health care consultant and Helen Saxenian is principal economist in the Human Development Department at the Wrld Bank. The authors are grateful to Deborah Chollet, Jillian Cohen, and George Schieber for helpful comments and to Madelyn Ross for editorial assistance. This paper is dedicated to the memory of Jose-Luis Bobadilla. 215 LatinAmerica and the Caribbean. This paperreviews expe- 1995, for example, Haiti's per capita income was $250- riences with managed care and the regulatory frameworks less than one-tenth Costa Rica's $2,610. In 1990 Haiti's per in which it operates in selected countries in the region. capita health spending, at $62 a year, was 13 percent of The paper draws two basic conclusions. First, there is Costa Rica's $460. In 1995 life expectancy for women in ample scope for the development of managed care in both Haiti was 57 years, compared with 79 years in Costa Rica. public and private health care systems in Latin America and In 1994 the region spent an estimated 6.3 percent of the Caribbean. Traditional social insurance institutions GDP on health, about half of which came from public and private prepaid medicine have much to learn from man- sources. Total health spending ranges from 2.7 percent of aged care modes of financing and delivery. Second, several GDP in Guatemala to more than 8.0 percent in Argentina, varieties of managed competition are relevant for the region's Belize, Costa Rica, and Uruguay (World Bank forthcoming). health care systems. Under an appropriate regulatory frame- Although the region's health markets and health insur- work greater competition among integrated delivery sys- ance institutions differ significantly, there are four basic tems, together with greater consumer choice, can improve systems: health care providers financed directly by private the efficiency and quality of services. The specific methods out-of-pocket payments, private health insurance markets used to manage competition will depend as much on coun- financed by prepaid contributions, social insurance mar- tries' size and level of development as on the methods kets financed by mandated employer and employee con- used to mobilize financing. Given the unequal distribution tributions, and publicly delivered health care services of income and wealth in the region, the state needs to play financed by general taxes and operated by ministries and an important role in mobilizing resources to ensure equity decentralized regional health offices (table 1). Managed and efficiency. The weakness of markets also requires that care appears in both private insurance and social insurance the state play an active role in enhancing market develop- systems. ment, to increase the chances of improving the coverage, These four systems serve different but overlapping groups. efficiency, equity, and quality of services. The poor tend to use public hospitals and clinics, supple- mented by out-of-pocket expenditures on private practi- The Region's Health Care Systems tioners and drugs. Formal sector workers are typically covered by social insurance systems, although many supplement this Latin America and the Caribbean is a heterogeneous region: coverage with out-of-pocket expenditures and, in some income levels vary enormously, as do health status, health cases, private insurance. The rich tend to buy private insur- care spending, and health care financing and delivery. In ance, although they also may be covered by social insur- TABLE I The four main health service delivery systems in Latin America and the Caribbean Source of Cost Provider payment Consumer Criteria for System Regulation financing controls mechanism options Ownership coverage Private out-of-pocket Minimal Out of pocket None Fee for service Many Private Ability to pay spending Private insurancea Some financial Employers and Range of Range of Many Private Ability to pay regulation households cost controls payment mechanisms Social insurancea Little or no extemal Mandated Range of Range of Limited Collective Mandated regulation payroll taxes cost controls payment mechanisms contributions Public services financed No extemal General tax Public Installed capacity Restricted Public Universal, but by general tax revenue regulation revenue budgets especially as a last resort for the poor a. Can include managed care. Source: Adapted from IDB 1996. INNOVATIONS IN HEALTH CARE FINANCING 216 ance and may rely on the public sector for high-technology In countries where the public sector has played a dom- services. This broad classification leaves out important vari- inant role in the funding and provision of health care, the ations, however. In Brazil, for example, the government government will have to reorient its role, becoming a con- contracts to the private sector many services that are pub- sumer advocate and working to ensure greater autonomy licly financed. for health care providers, greater freedom of choice for con- Universal access, quality, and efficiency are major issues sumers, and more efficient modes of resource allocation. in the region's health care systems. The poor have the worst In countries with better-developed markets for private access. And despite rapidly rising health care costs in many finance and provision, as well as more experience with countries, some analysts believe that the quality of public integrated delivery systems, it will be possible to experiment services has declined. Many countries' health care systems with more competitive systems using the financial resources are fragmented. Populations are served by overlapping and provided under social insurance schemes. And finally, in uncoordinated systems. Few providers-pubic or private- countries where population groups are highly segmented have the incentives to adopt an integrated approach to the within the health care system, and that typically exclude health problems of the populations they serve. Governments the poor from health care, competition may best be pro- generally have failed to coordinate and regulate the diverse moted within a publicly financed and regulated system segments of the health system. that expands coverage and improves the quality of service. Analysts in the region have traditionally emphasized the public sector's role in financing and providing health care. Emerging Managed Care Models Only recently have they started to examine the importance of private participation in health services. In the early 1990s The development of integrated health care delivery systems total spending on health was believed to total 4.0 percent is not new in Latin America and the Caribbean. In Argentina of regional GDP, including 1.6 percent that was not part of and Uruguay the development of organizations of health public spending (World Bank 1993). Researchers have since service users (mutuales) dates from the end of the nineteenth estimated that public spending on health services was much century. Since World War II many countries have experi- higher-more than 3.1 percent of GDP-and that private enced growth in social insurance systems that have inte- spending was more than twice as high-3.5-4.0 percent of grated finance and delivery of health services for formal GDP Moreover, recent research has illustrated the tremen- sector workers. These institutions were constituted as pub- dous progress made in the organization of health services lic monopolies in most countries, though not in Argentina in the Americas. Almost 100 million Latin Americans are and Uruguay, were pluralism prevailed. What is new in the now served by integrated health care delivery systems, and region is the recent rapid growth in private insurance. more than 60 million are registered with prepaid, non- A number of managed care models have been developed governmental, integrated delivery organizations-almost as in the region, including Chile's Provisional Health Institutions, many as are served by health maintenance organizations Uruguay's Collective Institutions of Medical Assistance, (HMOs) in the United States. Colombia's Health Promotion Entities, the Dominican Thus public financing and provision of health care is Republic's Igualas Medicas, Brazil's Medical Group just one part of the health sector, albeit an important one. Organizations, and Argentina's ObrasSociales. Characteristics Although the organization and structure of the private health of these models are summarized in table 2. More than 75 mil- sector are weak, the concepts of managed competition are lion people are enrolled in these organizations. highly relevant to the development of the region's health These managed care programs are similar to preferred care systems. The specific methods used to develop man- provider organizations in the United States in that, in almost aged care in Latin America and the Caribbean will depend all cases, covered individuals must use providers within the on countries' size and level of development as well as on program's network. Providers in the network are often the modalities of health care systems. In this regard, there paid on a capitated basis, which puts them at risk for the are three types of countries in the region. cost of services provided-and can create incentives to MANAGED CARE AND MANAGED COMPETITION IN LATIN AMERICA AND THE CARIBBEAN 217 TABLE 2 Characteristics of managed care organizations in selected Latin American and Caribbean countries Integrated Selective Country/ delivery provider Utilization Negotiated Quality Number organization systems? contracting? management? payments? management? enrolled ChileASAPREs Yes Some Yes Yes Heterogeneous 27% of population (3.8 million) UruguayAAMCs Yes, in most cases Some Some Fixed prices set by Heterogeneous 65% of population (2 million) govemment (in some cases) Colombia/EPSs Yes Some Some Yes, risk-adjusted Heterogeneous 50-60% of population (I18-22 capitation set by million) govemment Dominican Republic/ Yes, in most cases Some Some Yes No 7% of population (0.5 million) Igualas Medicas Brazil/MGOs Yes Yes Some Yesa Weak 25% of population (40 million) Argentina/ Some Yes Some Yes, but not uniform Weak 29% of population (10 million) Obras Sociales a. Price tables are used, with general agreements defined each year. Source: IDB and World Bank staff estimates. underprovide services unless countervailing forces check aged competition.2 At the same time, regulatory mecha- this tendency. Some of the managed care programs listed nisms must be adapted to a country's public sector man- in table 2 provide services directly, and in this way resem- agement capacity and ability to enforce regulations. Another ble health maintenance organizations (HMOs) in the United challenge for governments is improving access to and qual- States. Utilization review and quality control management ity of services for the poor in a competitive framework. Some are the two standard elements of managed care that are govemments, for example, are exploring demand-side sub- least developed in Latin America and the Caribbean. sidies so that the poor can choose from competing health plans (including, in some cases, public providers). How Can Countries Achieve Managed Enthoven (in this volume) argues that govemments and Competition? large purchasers of health insurance ("sponsors") must deter- mine, based on factors such as quality of care and financial Health care systems in Latin America and the Caribbean solvency and stability, which health insurers or managed reflect each country's history, culure, and political, social, care organizations should be allowed to compete to provide and economic development. Since many countries have health care services for the sponsored population. In Latin highly pluralistic systems, one of the main challenges that America and the Caribbean the sponsor might be the min- governments face is creating regulatory mechanisms that istry of health, social security institute, large private employ- encourage competition to induce equity and efficiency and ers, or other institutions. The sponsor's main roles include: improve the quality of services. * Defining the basic health care service package provided Unregulated health care markets generate many mar- by insurers or managed care programs. A standardized ket failures (Arrow 1963; Musgrove 1996). Unregulated benefits package simplifies consumer choice and focuses insurance markets, for example, result in adverse selection attention on quality and cost. by individuals and risk selection by insurers-both major * Providing information on quality and cost to consumers obstacles for efficient and equitable risk pooling. The goal so that they can make informed choices. of regulation should be to establish rules of the game that * Establishing rules of the game so that consumers can mitigate the market failures endemic to competitive health choose the insurer or managed care program that best care financing and delivery-an objective known as man- suits their needs. These rules typically require coordi- INNOVATIONS IN HEALTH CARE FINANCING 218 nated annual open enrollment and no waiting periods an average of 7.4 percent a year (in real terms). The aver- or exclusion of coverage for preexisting conditions. age annual domestic savings rate is 25 percent. And unem- Managing risk selection so that health care organizations ployment, which averaged 28 percent during 1983-89, can focus on providing health care efficientlyrather than dropped to 5.6 percent in the first half of the 1990s. on trying to enroll populations with the lowest risk of As a regional leader in terms of economic adjustment, health expenditures. Standardized benefit packages limit Chile's economic and social reforms, including those involv- health care organizations' opportunities to select risks. ing social security and health care, are considered models In addition, the sponsor can use risk-adjusted premi- for the rest of the continent. Between 1960 and 1995, for ums-that is, make compensatory payments to plans that example, the under-five mortality rate dropped from 155 enroll high-risk populations and cut payments to plans per 1,000 to 15 per 1,000 (World Bank forthcoming). that enroll low-risk populations. Moreover, the main causes of death are now similar to those Sponsors also have a role to play in increasing equity. in industrial nations. They can, for example, channel government subsidies so In 1981 Chile launched two parallel reforms: creating a that the poor can participate in the system and enroll in the system of private integrated delivery systems-ISAPREs- same plans as the rest of the population. They also can set and partly decentralizing public health care to twenty-six contribution rates in social insurance systems so that con- regional health centers. Public health care now covers 73 tributions from high-income workers and their employers percent of the population; the remaining 27 percent is cov- subsidize contributions from low-income workers and their ered by the ISAPREs. ll formal sector employees must employers. contribute 7 percent of their salaries to the national health No country has yet achieved this idealized version of system. (There is no employer contribution.) Workers whose managed competition. In Latin America and the Caribbean salaries reach a certain level can purchase an ISAPRE (pri- two different models of managed competition have emerged. vate) health insurance plan. If workers choose to buy a more In some countries (Colombia, Uruguay) the government, expensive health plan than their 7 percent contribution per- as sponsor, is playing a bigger role in managing competi- mits, they must pay the difference. tion, intervening in finance and regulation. In other coun- Although the government established a basic package tries (Brazil, the Dominican Republic) large private of services to be provided by the ISAPREs, the plans are employers are playing the role of sponsor, without any pub- increasingly selling more comprehensive packages to work- lic finance or regulation. These efforts are analyzed in the ers. The unemployed and workers whose salary contribu- case studies that follow. tion is not high enough to buy an ISAPRE plan are covered by the National Health Fund, which is generally consid- Case Studies of Health System Reforms ered to provide lower-quality care. This is not surprising and Managed Competition given that in 1995 per capita spending by the ISAPREs was $646, compared with $121 by the National Health This section analyzes reforms that are introducing elements Fund. Copayments are significant in the system, and with of managed competition in Latin American and Caribbean the growth in the system's costs they have become an impor- health systems-in Chile, Uruguay, Colombia, the tant source of financing for the services provided by both Dominican Republic, Brazil, and Argentina. the ISAPREs and the National Health Fund. In recent years the government has implemented other Chile's system has increased competition, but equity is a public sector reforms, such as provider payment reforms problem (payment by diagnosis). Budgets for local health posts are now capitated and are linked to the posts' capacity to recruit The economic and social development model followed by and maintain clientele. Public hospitals have been given Chile since the early 1980s is one of the most advanced in more autonomy, and are increasingly selling services to the Latin America. During 1991-95 Chile's economy grew by ISAPREs. MANAGED CARE AND MANAGED COMPETITION IN LATIN AMERICA AND THE CARIBBEAN 219 Although these reforms have introduced some compe- forthcoming). The health system is pluralistic, with many tition, in that the ISAPREs compete to provide health care private and public service providers. Although in theory all services to one segment of the population, the minimum Uruguayans are covered by a health plan, data indicate conditions for managed competition have not been achieved. that about 6 percent of the population is not enrolled in Most ISAPREs enroll healthy, young, high-income work- any health plan. In 1994 health spending was almost $300 ers. Moreover, the ISAPREs have powerful incentives to per capita-about 8.5 percent of GDP (World Bank forth- make themselves unattractive to patients with costly chronic coming). conditions. When individuals' risks for health expenses During the 1970s and 1980s Uruguay's health system increase-whether because of aging or health problems- underwent a series of reforms. Formal sector employees can the ISAPREs raise their premiums or exclude conditions now choose among health plans, which are provided by from coverage. The population is divided into rating cate- private Collective Institutions of Medical Assistance gories byindividual risk. As a result many high-cost patients (IAMCs). The system's frequent deficits are covered by gen- return to the public system. eral tax revenue. A regulatory office was recently established to regulate The Bank of Social Provision finances, on a capitation the behavior of the ISAPREs and to maintain a minimum basis, each worker's basic health plan. The IAMCs now number of providers in the system. But this office does not cover about 65 percent of the population. Worker contri- have the legal authority to ensure that the ISAPREs oper- butions are set by the government, which also has defined ate under equitable rules regarding pricing, access, and the basic package of services to be provided by the lAMCs. enrollment. Moreover, risk selection is not managed. Given fixed contributions, IAMCs control their costs by In summary, Chile's health system still has many short- adjusting the volume and quality of services and by exclud- comings. Equity is a major problem given that the public ing high-risk individuals. The degree of consumer choice system is poorly funded relative to the ISAPREs and receives among the IAMCs depends heavily on location. In Uruguay's no subsidies from the ISAPREs. In essence the country capital, Montevideo, many IAMCs compete in the market. maintains a two-tier system with differing services and qual- In rural areas, however, there are usually only two IAMCs ity of care. Although quality of care and consumer choice per district, and in several districts there are none. have improved for many high-income workers, the regu- Employee and employer contributions to the IAMCs latory framework needs to be reformed so that competi- only cover services provided to employees. Thus employ- tion is better managed to minimize market failures and to ees tend to select the IAMC that offers the best terms for improve equity, quality, and cost control. A managed com- family coverage, or they enroll their children and spouses petition regulatory framework is needed so that the in partial insurance institutions that are not required to offer ISAPREs have incentives to improve service quality and the basic package. reduce costs, not to maximize profits by attracting low- The public sector covers people who are not served by risk populations. Public sector networks could be orga- the IAMCs, including poor households whose head does nized to compete with the ISAPREs for consumers under not eam enough to purchase coverage by an IAMC or a par- comparable health care packages. Direct subsidies could tial insurance institution. As with Chile's ISAPREs, the pub- be given to low-income groups to enhance the demand side lic sector also covers people-usually the elderly-who have of the system. been forced to leave the IAMCs because of increases in their health risk. Uruguay provides extensive coverage, but needs to Both the public and private sectors of Uruguay's health strengthen its regulatory framework system use copayments to recover costs. Thus the IAMCs are financed by a combination of prepayments (in the form Uruguay, with 3.2 million inhabitants, has relatively good of monthlypremiums) and copayments for services received. socioeconomic indicators and income equality. In 1995 the The public system has many levels of copayment that vary under-five mortality rate was 21 per 1,000 (World Bank with family income. INNOVATIONS IN HEALTH CARE FINANCING 220 The Ministry of Economy and Finance sets prices for risk-adjusted payment formulas, that compete to cover low- copayments and contributions, and has avoided significant income groups. Risk-adjusted capitation formulas could price increases over the past ten years. As a result the oper- be used to avoid the cream-skimming behavior of the IAMCs. ational deficit of the lAMCs has increased substantially, and The efforts would, of course, require some adjustment of some have closed. Contributions do not vary by age, and Uruguay's rigid regulatory framework. the IAMCs do not receive any risk adjustments based on their enrolled populations. As a result many young people Colombia hos mode impressive progress in structuring voluntarily leave the system (because they can purchase competition in recent years cheaper plans elsewhere), while the elderly try to retain cov- erage. This imbalance has been a major factor in the finan- Colombia is in the middle tier of Latin America's develop- cial crisis facing the IAMCs. ing countries, with a population of 37 million people, per Young people have two alternatives to the IAMCs. They capita income of $1,800, and the region's most stable econ- can enroll in a partial health insurance plan offered by a omy. Having been spared the external shocks experienced private institution (which generally costs less than the by most other Latin American nations during the debt cri- IAMCs), or they can seek services from the public sector. sis, and with no deep recessions or inflationary spirals, it Since many choose the latter, the number of people served gradually managed to narrow pronounced income inequal- by the public sector has been growing. Public spending has ity during the 1980s. Still, nearly 30 percent of its people increased and service quality has deteriorated. Moreover, survived on less than $2.a day, and the country's social sec- an increasing number of people are covered simultane- tor institutions were undeveloped. ously by two or three modalities of assistance. Until the early 1990s health care systems were highly seg- Public resources are also strained by policies regarding mented, characterized by heavy out-of-pocket spending. the financing of high-technology services. Many high-tech- The neediest were left without coverage. An ambitious pack- nology services are covered by the publicly financed National age of institutional and financial reforms launched in 1993 Resource Fund and are supplied by the Institutes of Highly sought to implement a health insurance system with a strong Specialized Medicine. The cost of some of these services element of competition but guided by principles of soli- (renal transplants, for example) is almost five times the price darity. In the process Colombia proved itself a pioneer in charged in neighboring countries (such as Argentina and its pursuit of managed competition. In the first three years Brazil). In 1995 the National Resource Fund spent half of of the reform program, despite daunting technical and polit- its budget on high-technology services for about 8,000 ical problems, coverage has been vastly expanded, equity people. has been increased, and the quality of health services has As in Chile, Uruguay's pluralistic healh system has ele- been much improved. ments of competition, but the regulatory framework is not In years past Colombia had three parallel health care sys- sufficiently developed to enhance competition and choice. tems. A social insurance system run by the Social Security There are many market failures and problems in the sys- Institute provided health services to formal sector work- tem that put pressures on health care costs. Problems arise ers, and complementary agencies supplied health services from the way contributions are controlled, the lack of mea- to their families and to government employees. After forty- sures to control adverse selection by individuals and risk five years, this public health care system served 21 percent selection by insurers, the fact that dependents are not cov- of the population-mainly middle-class citizens, who were ered under a common framework, and policies on the reim- dissatisfied with service quality. A traditional public health bursement of high-technology procedures. system overseen by the Ministry of Health combined pub- The main suggestions for developing a more consistent lic health interventions and open-access hospitals. Repeated managed competition framework are related to the trans- efforts since the 1970s to decentralize personal medical care formation of public services and budgets. Public providers and emphasize primary health care for the poor had borne could be organized as autonomous networks, financed by little fruit, creating a situation in which the poor had scant MANAGED CARE AND MANAGED COMPETITION IN LATIN AMERICA AND THE CARIBBEAN 221 access at the same time that hospital utilization rates were tribution among population groups with differing economic very low (less than 50 percent). means and health needs. The fund assigns to the organiza- Operating alongside the two public systems was a pri- tion chosen by each family a capitation payment that is vate health care system, itself far from uniform. Privately run risk-adjusted for sex, age, and geographic location, with hospitals accounted for 25 percent of hospital discharges reinsurance for catastrophic illness. The average capitation and surgeries. Some 75 percent of outpatients were attended payment is $120 per person per year. In addition to the basic to by private doctors or practitioners. One million wealthy contribution, there is a copayment system to encourage Colombians had access to high-quality and expensive pre- rational use of services. Public funds do not finance the paid medical systems and private insurance companies, whose Solidarity Fund; they are gradually being turned to directly coverage essentially duplicated what was offered bythe social fund health insurance for the poorest 30 percent of insurance system. This segmented system cost a great deal Colombians. and generated sizable disparities between regions and At the core of health care reform is a new system of between segments of society. Health care expenditures were multiple organizations in charge of enrollment (by delega- enormous, equivalent in 1992 to about 6 percent of GDP, tion of the Solidarity Fund), insurance, and organization and more than half came out of families' pockets. The coun- of service delivery. Health Promotion Organizations (EPSs) try's infant mortality rate of 25 per 1,000 live births was an combine insurer functions with the type of service articu- average of the 15-20 per 1,000 rate in large cities and the lation performed by U.S. HMOs. Without privatizing the more than 200 per 1,000 in remote rural areas. At least 20 Social Security Institute, the reforms divested it of its monop- percent of the population-the poorest rural dwellers, res- oly in this area. They allow unrestricted competition among idents of marginal urban districts-had no access to health EPSs-public or private, nonprofit or for-profit, integrated care, a situation they attributed largely to their inability to delivery or by contract-subject to certain minimum finan- pay. The poorest 20 percent of Colombians paid about 18 cial conditions. Families can freely choose their EPS. Users percent of their income for health care. are guaranteed a universal package of high-quality health Between 1990 and 1993 Colombia laid the institutional services, including medicines. One aim of this arrange- foundations for a complete overhaul of this fragmented, ment is to stimulate competition for service delivery and segmented, and inequitable health system. A 1991 consti- spur the development of supplementary plans for inter- tutional reform created the framework for decentralizing ventions not covered under the universal package. social services and developing a social security system The new social security system seeks to integrate the pub- grounded in the principles of universality, solidarity, effi- lic health care delivery system. Government budget out- ciency, and private sector involvement. In 1993 Congress lays (bolstered by a "solidarity payment" of 1 percent of approved Law 60 and Law 100, the comerstone of a new contributions) are being strictly targeted to public health system made up of public finance, mandatory affiliation, interventions or to the poorest 30 percent of Colombians. decentralization and competition in service delivery, con- Public hospitals have become autonomous corporations, sumer choice, and broad participation. akin to state industrial enterprises. Regional agencies now At the heart of this new system was mandatory affilia- purchase services on behalf of disadvantaged groups. tion, by those able to pay, with a social security system guided Growing volumes of direct transfers from the national gov- by a uniform resource mobilization and allocation scheme emient to municipalities are intended to fund insurance and plurality of health service organizations. Thus finance for the neediest. To that end, a system of community coop- was separated from service delivery, and the purchaser- erative healh organizations has been actively promoted in provider split of services was developed. the poorest areas to purchase basic health care services or The system's funding scheme is grounded in social insur- to allow people to join an EPS. ance principles. All participants contribute 11 percent of In tandem with this separation of the system into three their earnings to enroll their families. These resources are streams of funding, service purchase, and service production, placed in the Solidarity Fund, a compensation fund for redis- the Ministry of Health, freed of direct responsibility for pro- INNOVATIONS IN HEALTIH CARE FINANCING 222 viding services, is now responsible for general oversight. The Growing violence, resulting in 30,000 homicides a year ministry is aided by a national board representing business, (more than in the United States, which has ten times labor, and health service users. This tripartite body has broad the population), is adding to the burden of disease. The powers to develop rules and regulations under the law. In associated skyrocketing demand for medical care is gen- addition, the now-autonomous Health Superintendency erating additional costs of more than 1 percent of GDP. has been strengthened to better perform its financial control * The process of identifying basic parameters for the new functions and supervise the workings of the system. system has been one of trial and error. The capitation Thus Colombia's new health system combines the two rate, for instance, must balance out system revenues while preeminent principles of Latin America's version of man- defraying the cost of universal service. Difficulties in aged competition: coordination of service delivery in inte- measuring costs and forecasting revenues of new mem- grated plans and consumer empowerment. At the core of bers have prompted more changes in these areas than the system are the EPSs, which assume the risk of ensur- would have been desirable. ing a universal package of services. The demand side is * The absence of strong political resolve to curb evasion bolstered by users' ability to freely choose their EPS, by has kept system revenues down. consumer organization through health alliances and coop- * The transformation from a system that allocated public erative health organizations, and by government media- funds to one based on demand fnancing has taken longer tion in health care funding (with risk adjustments for groups than expected, driving up costs unnecessarily. that cannot pay, and direct government subsidies for the * Not all the adjustmetits to the universal health package poor) and regulation. have been based on sound cost-effectiveness criteria. In the three years since Colombia adopted the new sys- * In the absence of fully developed information systems, tem, enormous progress has been made in its development. the allocation of public resources and conditions for con- More than fifty decrees have been issued to regulate the sumer choice are still not fully transparent. system and fine-tune its various elements. The Ministry of * Political problems stil arise. One result of the govern- Health has made local agencies responsible for service deliv- ment's political crisis was a succession of three health ery in 80 percent of the country, and 85 percent of hospitals ministers in less than three years, leaving the system with- are now run autonomously. With the new funding system in out strong direction at critical junctures. And the shift in place, nearly 20 milion people-55 percent of the popula- the balance of power between users, providers, organiz- tion-now has health insurance coverage. Some 14 milion ers, and fund holders has created friction, sometimes people have chosen to join one of thirty EPSs, just over half open and sometimes not, that has hampered decision- of them falling under the Social Security Institute. The new making and tested providers' support for the system. EPSs have an impressive record of innovation in service orga- In the final analysis, Colombia's innovative reforms to nization, information systems, contracting, payments, and introduce competition into its health care system have yielded quality assurance. Six miDlion of Colombia's poorest have results far more quickly than in other countries. The chal- received vouchers giving them access to the system: four mil- lenge in the years ahead will be to consolidate the gains lion through EPSs and two milion through cooperative health and step up the pace of the transition, to give the country organizations. According to a survey conducted in early 1997, a more efficient and equitable system overall. 83 percent of EPS members were satisfied or very satisfied with the services they were receiving, and 73 percent of the Costa Rica has achieved some elements of managed care population felt that the health system had improved since the enactment of Law 100. The Costa Rican government created the Costa Rica Social Colombia's mix of public funding and managed com- Security Fund (CCSS) in 1941 to provide sickness and mater- petition has substantially expanded coverage and made nity coverage and pension funds for urban salaried work- the system more equitable. Still, the transition has had its ers. In 1961 this mandate was expanded to cover the entire share of technical, institutional, and political problems: population within ten years, but political disputes delayed MANAGED CARE AND MANAGED COMPETITION IN LATIN AMERICA AND THE CARIBBEAN 223 implementation ten years. In 1971 the CCSS was required Under the medical cooperative program, the CCSS by law to proceed with the universalization of coverage. signed a contract with two health cooperatives, making them This goal was achieved in 1973 with the enactment of a responsible for two clinics in the SanJose metropolitan area. law that transferred all Ministry of Health hospitals to the Under one of these agreements, the CCSS constructed and CCSS and entrusted the CCSS with providing free health delivered a $1.5 million clinic to COPPESALUD, a physi- care to the poor. That same year, the National Health Plan cian's cooperative, which became responsible for outpatient defned the scope of work for the Ministry of Health and care in the Pavas district. The CCSS pays about $30 a year the CCSS. The ministry became responsible for providing for each insured resident in the area. COPPESALUD health preventive health services and the CCSS for curative ser- services are delivered by a basic health care team made up vices. The social security system's extensive coverage has of a physician, an assistant, and a community health worker left little room for the private sector. (Both the ministry for each 3,000-3,500 persons. Staff turnover is low because and the CCSS serve patients who are uninsured and do salaries are high. not deny services to anyone, regardless of income.) A Although Costa Rica's health system does not have 1982-85 survey, for example, found that the private sector enough insurers to create competitive market conditions, managed just 1.9 percent of hospital beds. and the private sector is small, the three private health pro- Costa Rica's economy, like many others in Latin America grams do contain elements of managed care. In the enter- and the Caribbean, suffered considerably during the economic prise physician program, doctors have to compete for the crisis of the early 1980s. In 1981 inflationwas 81 percent and salaried positions within companies. Like the general and the currencywas devalued by320 percent, and during 1982-83 family agents of the primary care network, enterprise doc- unemployment reached 9.4 percent. The country's capacity tors serve as gatekeepers to specialized services and as agents to pay its external debt was undermined as debt payments of cost control. Under the mixed medicine program, the started to consume more than half of export earnings. The insured must select a doctor from the list of physicians pro- government's capacity to collect taxes was reduced, and pub- vided by the insurance organization in which he or she is lic expenditures were cut. As a result the National Health Plan enrolled. This approach can save money for the system if was revised to redefine the legal roles of the Ministry of Health the insured signs up with the most economical providers. and the CCSS, to promote administrative decentralization, Finally, the medical cooperative program represents a type and to create altemative models of health delivery. of capitation model in which a group of physicians provide During the 1980s the government experimented with health care services in return for a fixed per capita pay- transferring part of the health delivery system to the pri- ment provided by the government. vate sector. In particular, the CCSS developed three pro- Given that Costa Rica's small population impedes the grams-enterprise physicians, mixed medicine, and medical development of competitive health plans, future innova- cooperatives-to incorporate the private sector in service tions could come from increasing choice for beneficiaries delivery. These measures sought to reduce costs and improve of the Family Health Program, using more managed care service efficiency and quality. techniques in primary care centers and public hospitals, and Under the enterprise physician program, private enter- increasing transparency and efficiency in the system of prises (630 to date) hire a doctor and provide him or her resource allocation. with an office and a nurse. Under the mixed medicine pro- gram, the insured chooses a doctor from among members The Dominican Republic is moving toward greater privote of the program and pays for the service. Both programs are provision supported by the CCSS. During 1986-87 these two pro- grams reached a combined 7 percent of the population cov- With 7.5 million inhabitants and a per capita income of eredbythe CCSS while consuming only4 percent of x-rays, about $1,000, the Dominican Republic spends about 5.3 3 percent of laboratory tests, and 6 percent of approved percent of its GDP on health (World Bank forthcoming). paid sick leave. Public spending accounts for 38 percent of the total. Despite INNOVATIONS IN HEALTH CARE FINANCING 224 the lack of in-depth studies on service coverage, it is esti- parameters. The second scheme is usually more expensive mated (based on a study conducted in the capital) that 56 and has a more sophisticated system of control. percent of the population is covered by the private sector, Finally, the self-administered insurers are nonprofit orga- 30 percent by the Sub-Secretariat of Public Health and nizations established by institutions, companies, and unions Social Assistance (SESPAS), 10 percent by the Dominican to manage health insurance plans for their specific popu- Institute of Social Security, and 4 percent by the Institute lations. Ten such insurers are operating, the largest of which of Social Security of the Army and Police Force. is the Medical Insurance of Teachers, with about 120,000 SESPAS plays a contradictory role since it regulates and enrollees. inspects all public and private health services while also pro- The Igualas Medicas are the fastest growing of these viding publicly financed health services through its 57 hos- three types of insurers. With the backing of USAID, these pitals, 79 health posts, and 587 rural clinics and medical enterprises have received technical assistance with admin- dispensaries. The Dominican Institute of Social Security istration, cost control, quality assurance, and actuarial capac- manages a maternity hospital, 20 polyclinics, and 13 urban ity. Although they are now trying to expand their services and 128 rural medical offices. Its financing comes from to low-income groups, they are doing so without the finan- voluntary contributions to social security accounts and trans- cial backing of the state, which continues to provide its fers from the federal government. It charges a fee for some own health services. services to partially recover costs. Despite the institutional forms of prepayment developed The private sector is composed of both nonprofit insti- in the Dominican Republic, most insurers lack basic tech- tutions (such as NGOs) and for-profit organizations. Over nical and administrative expertise in terms of information the past twenty years the organization of private services systems, methodologies for establishing risk, cost controls, has changed as coverage strategies and health plans simi- and cost-efficient methods of allocating resources (Santana lar to those of U.S. HMOs have been adopted. Although 1996). They also lack adequate mechanisms for respond- it controls just one-quarter of the country's hospital beds, ing to consumers' complaints and comments. the private sector accounts for more than half of outpa- Despite these deficiencies, the health services provided tient visits. by the private sector are relatively comprehensive and of The economic and institutional crisis experienced in higher quality than those provided by the public sector. the 1980s created problems for the Dominican Republic's Private insurers admit families and all categories of work- public health system. Since 1985 the government, with assis- ers, and their beneficiaries have access to better clinics and tance from the U.S. Agency for International Development services. Their units also offer greater choice of doctors (USAID), has been developing altemative financing schemes and more personalized attention. and extending private coverage to low-income families One frequent criticism of the private system is that its and workers in the informal sector (La Forgia 1990). package of health coverage excludes many services and high- There are at least three kinds of private insurance: Igualas cost patients. But this criticism should not overshadow the Medicas (prepaid HMO-type health plans), health insur- fact that the Igualas Medicas are a viable option for improv- ers, and self-administered insurers. The twenty-one Igualas ing health services and eventually reforming the Dominican Medicas administer the organization, financing, and pro- Republic's social security systems. vision of health services. Most have their own clinics. Further development of managed competition in the The health insurers are specialized departments of insur- Dominican Republic requires the development of a regu- ance companies organized to cover health risks within an latory framework that increases the accountability of the integrated insurance package. In general, they contract with Igualas Medicas while fostering competition with SESPAS medical clinics for service provision. The insurers provide and the Institute for Social Security. Consumer empower- two types of services: those based on a system of preferred ment can be increased by developing an essential package providers (where the user has limited choices) and a sys- of services and allocating public resources to subsidize the tem offering freedom of choice within established price poor's access to the competitive integrated delivery systems. MANAGED CARE AND MANAGED COMPETITION IN LATIN AMERICA AND THE CARIBBEAN 225 Brazil's three systems should be better integrated Despite a pluralistic structure of service providers, the Brazilian health model lacks well-defined rules separating Brazil's health care system is composed of a public system, the three systems. As a result there are conflicts among the a private system of supplementary medicine, and an out- system's participants, due both to the lack of adequate leg- of-pocket system (table 3). The resources gathered through islation and to the constitutional principles guaranteeing these three systems totaled $23.2 billion in 1994, with expen- the universal right to free and comprehensive health ser- ditures equivalent to 4.5 percent of GDP vices. Low-income citizens account for 78 percent of the The public system-known as the Unified Health public system's users, yet must share the system with the System-was established by the 1988 Constitution and is 22 percent of people who are also covered by private insur- financed by federal and local governments. In 1994 fed- ance. Otherwise, the poor have to pay for medical services eral health expenditures reached $10.4 billion, while those out of pocket. of states and municipalities reached $4.1 billion. Most of Since the public sector is not organized on a competi- the resources local governments spend on health are trans- tive basis, services are essentially rationed-as evidenced ferred to them by the federal government. In exchange, by long hospital lines and lack of materials and medication. local governments are responsible for the direct provision And because there are no barriers blocking use of the pub- of services or for contracting out services to private estab- lic system, people covered by private health insurance tend lishments. to use the most sophisticated services and exams provided The private system consists of four types of organiza- by the public sector. Thus the government is indirectly tions: medical group organizations (MGOs) that operate subsidizing private health insurance. like U.S. HMOs, health insurance institutions that func- A few innovative experiments are under way to reverse tion under similar criteria but do not provide services directly, this situation at the local level, where states and munici- medical cooperatives that use prepayment systems (though palities are assuming increasing autonomy in the adminis- their operations resemble those of preferred provider orga- tration of health services. One such experiment is the Health nizations), and medical services provided by companies, Plan of Action introduced in 1995 in Sao Paulo, Brazil's which may be within their facilities or contracted from any largest city (15 million inhabitants). of the other three organizations. The private system relies The goal of the Health Plan of Action is to organize Sao on prepayments, covers about 35 million Brazilians, and has Paulo's municipal services (including hospitals and basic annual revenues averaging $6.6 billion. health services) into cooperatives (managed by doctors The out-of-pocket system-meaning household payments and staff) that compete for clientele. Many such coopera- for services provided through medical networks and private tives have been organized. Doctors and staff have been hospitals-has annual revenues of $2.1 biDlion. The number transferred from the human resources division of the city of people covered underthis system is residual, since resources government to administrative and management positions from the public and private systems are the main source of in hospitals and other health institutions. Under this scheme financing for hospitals, health services, and doctors. citizens can choose the cooperative that best meets their TABLE 3 Health care services in Brazil Public system Group (Unified Health System) Private supplementary medicine Direct out-of-pocket payments Informal sector workers/ low income Primary care and None Complement Unified Health System hospitalization Formal sector workers/ middle income High technology and Primary care, high technology and Not used much sophisticated procedures sophisticated procedures High income High technology and Primary care, high technology and Used heavily sophisticated procedures sophisticated procedures INNOVATIONS IN HEALTH CARE FINANCING 226 needs, and the municipal government makes an annual pay- cent for each dependent) and 3-6 percent paid by employ- ment of about $225 per capita to each cooperative. ers. Taxable income is capped at $3,750 a month. The Experience with the Health Plan of Action is limited to national Obra Social system covers 10 million beneficiaries relatively uncomplicated services in Sao Paulo. Public hos- in about 300 Obras. In addition to the national system, pitals are responsible for more complex services. The Health twenty-four provincial Obras cover 5 million public employ- Plan of Action has four main weaknesses: ees and their dependents (World Bank 1996). * Lack of incentives to provide preventive health care to About 4 million retired, disabled, and pensioned per- plan members. sons and their families are covered by Integral Medical * Limits to competition, since the cooperatives are orga- Attention for Retirees, which is funded by an additional pay- nized into geographically separate networks. roll tax of 5 percent that is shared by employers and employ- * A blurred relationship between the Health Plan of Action ees. Altogether, this social insurance network covers nearly and the Unified Health System, which creates cross-sub- 20 million Argentines, or about 61 percent of the population. sidies between the two systems and prevents clear analy- There is also a large market for private health insur- sis of the Health Plan of Action. ance. About 200 private plans cover more than 2 million * The per capita payments that the city makes to the Health people. Another 1 million belong to nonprofit mutual insur- Plan of Action could be used to foster competition ance funds. Coverage is often duplicated, however. Many between public cooperatives and private health struc- workers who contribute to the national Obras Sociales also tures. buy voluntary health insurance because of dissatisfaction Despite these problems, 80 percent of users are satis- with the services provided by the Obras. In addition, some fied with the Health Plan of Action. The challenge is to employers provide private health insurance in addition to give continuity to the program and to better define its rela- the mandated Obra Social coverage. People that are not tionship with the other health systems. covered by either the Obras Sociales or by private insur- Brazil can learn a lot from the varieties of managed care ance-mainly the poor-rely on the publicly financed pub- and managed competition in the region. The government lic hospital system and on out-of-pocket purchase of health could develop a system of competitive purchasers of health services. care and simplify the system of allocating public resources. The health delivery system is mixed. About half of hos- The public system of providers could be transformed into pital beds are in private institutions. Although some of the integrated delivery systems that compete with private net- Obras Sociales provide health services directly through their works. A common system of regulating public and private own facilities, most purchase services from a network of pri- purchasers could then be developed, with attention to a vate providers. Because access to providers is restricted to comparable package of essential services and a system of those within the network, many Obras Sociales resemble quality assurance. Within this framework there is ample preferred provider organizations. Over the past few years space for the application of managed care techniques, espe- the Obras Sociales have moved away from fee-for-service daily in ambulatory services. Brazil, however, has not reached provider payments toward capitated payments. consensus about future reforms. Still, the system is facing financial difficulties. The national and provincial Obras Sociales (induding Integral Medical Argentina's system has weaknesses-but the government Attention for Retirees) have incurred huge deficits in recent is trying to address them years. Moreover, consumer dissatisfaction with the Obras Sociales is widespread. Except for white-collar workers, In Argentina most formal sector workers and their depen- until very recently workers have not been able to choose dents are required to participate in an Obra Social (statu- their Obra Social-it is dictated by their employment. The tory sickness fund) linked to their place of employment. Obras Sociales also have huge differences in both per capita The national Obras Sociales are funded by a compulsory funding and performance. Some workers receive almost payroll tax of 3 percent paid by employees (plus 1.5 per- no health care services from their Obras. Others belong to MANAGED CARE AND MANAGED COMPETITION IN LATIN AMERICA AND THE CARIBBEAN 227 Obras that provide comprehensive services. Many Obras Conclusion Sociales are probably too small to be economically efficient. Many have weak managers and are overstaffed. Managed care models of health service delivery now make The provinces, which are responsible for almost all pub- up a significant share of the health insurance market in lic hospitals, also have experienced sizable fiscal deficits. several Latin American and Caribbean countries. Although Equity and efficiency are both big problems. Yet the gov- these models share many features with managed care pro- ernment's approach to reform in one part of the system- grams in the United States-the country with the largest the national Obras Sociales-suggests that it recognizes and share of the population (57 percent) enrolled in managed is responding to at least some of these challenges. care in the world-they have distinct features that reflect The government is now establishing a regulatory frame- their political and socioeconomic contexts. Their growth is work that supports a competitive environment for the Obras likely to continue over the next decade given both the growth Sociales. The envisioned framework contains many elements in demand for private insurance and the reforms that sev- of managed competition. It would guarantee standard health eral governments are supporting to promote competition benefit packages, support consumer choice, improve equity, in health insurance markets. and minimize risk selection by workers and Obras Sociales. The concepts of managed care and managed competi- Many components of the reforms have already been imple- tion have found fertile soil in Latin America and the mented. The government has signed decrees allowingwork- Caribbean. Colombia and Uruguay have gone furthest in ers to choose their Obra Social and to take their contributions introducing varieties of managed competition, with about with them. A preliminary standard benefits package has been half of their populations covered. In Argentina, Chile, and defined. The formulas for redistributing Obra Social con- southern Brazil new institutional modalities have been tributions from high-income to low-income workers were extended to cover about a quarter of the population. In overhauled to make the system more accountable, trans- Costa Rica and some parts of the English-speaking Caribbean parent, and equitable. Efforts are being made to develop the introduction of new organizational models is still in the risk-adjusted payments for the Obras Sociales based on the experimental phase. risk profiles of those who enroll. Obras are restructuring to A country's ability to introduce managed competition cut staff and modernize their operations. Finally, steps are depends on its characteristics and institutional features. being taken to strengthen regulatory oversight of the Obras. Three basic determinants of managed competition's suc- Designing and implementing these reforms is difficult, cess include a country's size, its level of development, and and depends on both political and technical factors. Design the way its health system is organized. The smaller and and implementation is made more difficult by the weak poorer a country is, and the less developed its institutional information base that supports the Obras Sociales. The capacity, the more trouble it will have managing competi- Obras Sociales enrollment database, for example, is incom- tion. The achievements of Uruguay (a small country) and plete and inaccurate. The number of dependents is not dear. Colombia (a middle-income country with a highly hetero- Data to support risk-adjusted payments (age, sex, health geneous population) do not follow these general guidelines, risks) are limited. And much work will be needed to develop however. measures of Obra Social quality to support consumer choice. As noted, the thirty-three countries of Latin America The consolidation of managed competition in Argentina and the Caribbean show considerable diversity. At one end will require improving the system of payments (including risk of the scale are eleven middle-income countries with fewer adjustment) to integrated delivery systems and increasing the than 1 million inhabitants (mainly English-speaking scope of choice for consumers. Obras Sociales will need to Caribbean countries) and high levels of coverage financed become more accountable to their consumers and face more and operated by the public sector. These are followed by competition with private providers. Over the long run the seven countries with 1-5 million inhabitants and high pop- poor could be brought into the system by redirecting public ulation densities (for example, Costa Rica, El Salvador, resources toward plans that provide them with coverage. and Jamaica). In these countries discussions of internal INNOVATIONS IN HEALTH CARE FINANCING 228 TABLE 4 Trends in Latin American health care markets Higher-income countries Lower-income countries Small homogeneous countries Managed care in public systems Competition for primary health care with public funds for the poor Large heterogeneous countries Managed competition with a mix of public and Structured pluralism with a stronger element of public finance private funds and institutions markets and opportunities for managed care are proba- active configuration of managed competition could help bly more relevant for primary care, and must be com- overcome the segmentation of existing systems, expand cov- bined with incentive systems that improve the public erage, and improve the quality of services (table 4). administration of services. At the other end of the scale Latin America and the Caribbean lag far behind the rest are the large Southern cone countries (including south- of the world in terms of education and health indicators. em Brazil), which have achieved considerable economic Catching up over the next decade will require more than and institutional development. Given these countries' more increased financial and human resources. Above all, it will developed markets, the integration and plurality of their require redoubling the pace and scope of institutional inno- service providers, and their health care experiments over vations that the region has experimented with during the the past decade, full competition between integrated deliv- 1990s in order to use additional resources more efficiently ery systems, with public financing, has the greatest poten- and equitably. The challenges are particularly great in coun- tial. Rather than encouraging development of separate tries with low levels of public sector management capacity. private and public systems, innovations should promote The methods described here under the aegis of managed convergence among existing systems in order to mini- competition hold great promise in this respect. A long road mize duplication of coverage, control costs, and improve must still be traveled to define the new models for the health the quality of services. sector. Managed competition systems will play a major role In other large but poor countries (Bolivia, Guatemala, in this process. Haiti, Honduras) the empowerment of consumers would be more effective if the state redirected its efforts from Notes providing the middle class with public hospital services toward funding and purchasing primary health care services 1. Londofio and Frenk (1997) suggest a related concept of for the poor. To that end, community organizations could structured competition to characterize a system in which competi- fory the por.eT acthat roleind strucommurinimarety org ions cd btion is structured with an explicit and common set of rules that play a more active role in structuring markets from the begin- aelglyvldfraldlvr ytm n pnos hscn are legally valid for all delivery systerms and sponsors. This con- ning, drawing on some of the ideas of sponsors presented cept differs from managed competition, in which different spon- by Enthoven (in this volume) or the community financing sors can manage, with more discretion, different rules (see Enthoven schemes identified by Hsiao (1992). The experiences of in this volume). Colombia's Empresas Solidarias and Peru's community devel- 2. Of course, managed competition is not the only type of reg- opment cooperatives are particularly relevant in this regard. ulation possible. Regulation is often used to restrict competition. Other countries have more options for developing health care systems with greater competition, freedom of choice, References and solidarity. These indude the larger, higher-income coun- tries with good institutional development (Mexico, AmericanAssociationofHealthPlans. 1995.HMOandPPO Trends Venezuela) in which health reforms aimed at managing com- Report. Washington, D.C. petition would be highly effective. There are also smaller Arrow, Kenneth. 1963. "Uncertainty and the Welfare Economics petition would be highly effective. There are also smaller of Medical Care." American Economic Review 53 (5). countries with more diversified social and institutional struc- Birdsall, Nancy. 1993. "Pragmatism, Robin Wood and Other tures (the Dominican Republic, Ecuador, Peru), where an Themes: Good Government and Social Well-Being in MANAGED CARE AND MANAGED COMPETITION IN LATIN AMERICA AND THE CARIBBEAN 229 Developing Countries." In L.C. Chen,A. KNeinrman, and N.C. laAi.gentina, Chiley Uruguay. CuadernoTecnico36.Washington, Ware, eds., Health and Social Change in an International D.C. Perspective. Carnbridge, Mass.: Harvard School of Public Health. La Forgia, G. M. 1990. Health Services for Low-Income Families: Bobadilla,Jose Luis. 1996. "SearchingforEssentialHealth Services Extending Coverage Through Prepayment Plans in the Dominican in Low and Middle Income Countries: A Review of Recent Republic. Washington, D.C.: The Urban Institute. Studies on Health Priorities." World Bank, Human - .1995. "UnderstandingHealthSectorReform:AFinancial Development Department, Washington, D.C. and Service Flow Model and the Colombian Case." Inter- Bobadilla, Jose Luis, Peter Cowley, Philip Musgrove, and Helen American Development Bank, Social Programs Division 2, Saxenian. 1994. "Design, Content and Financingof an Essential Washington, D.C. National Package of Health Services." Bulletin of the World Larrafiaga, O., R. Miranda, and R.D.P Molina. 1996. "Organizaci6n Health Organization 72: 653-62. Industrial de los Servicios Sociales de Salud en Chile." Paper Bobadilla, Jose Luis, Julio Frenk, Peter Cowley, Beatriz Zurita, commissioned byInter-American Development Bank, Santiago. Juan Querol, EmestiVilarreal, and Rafael Lozano, 1994. "El Lewis, Maureen A. 1996. "Measuring Public Hospital Costs: Paquete Universal de Servicios Esenciales de Salud." In Julio Empirical Evidence from the Dominican Republic." Social Frenk, ed., Economia y Salud: Documentos para elAnalisisy la Science and Medicine 43 (2): 221-34. Convergencia. Mexico, D.F.: Fundaci6n Mexicana para la Salud. Lewis, Maureen A., andAndre Cezar Medici. 1995. "Private Payers Enthoven, Alain. 1993. "The History and Principles of Managed of Health Carein Brazil: Characteristics, Costs, and Coverage." Competition." Health Affairs 12 (supplement): 24-48 Health Policy and Planning 10 (4): 362-75. - 1996. "Managed Competition: An Agenda for Action." . 1996. "The Health Reform in Brazil: Phasing Changes." Health Affairs (spring). Inter-American Development Bank, Social Programs Division Enthoven, Alain, and SaraJ. Singer. 1996a. "Managed Competition 1, Washington, D.C. and California's Health Care Economy." HealthAffairs 15 (1): Londotio, Juan Luis. 1996a. "Estructurando Pluralismo en los 39-57. Servicios de Salud: La Experiencia Colombiana." Revista de - 1996b. "Market Based Reform: What to Regulate and Andlisis Econ6mico. 11 (2). Santiago de Chile. Whom." Health Affairs (spring). . 1996b. "Is There a Health Gap in Latin America? A Fiedler, J. L. 1996. "The Privatization of Health Care in Three Graphical Exposition." World Bank, Latin American and the Latin American Social Security Systems." Health Policy and Caribbean Region Technical Department, Washington, D.C. Planning 11 (4): 406-17. . 1996c. "Managed Competition in the Tropics?" Paper Frenk, Julio. 1995. "Comprehensive Policy Analysis for Health presented at the International Health Economics Association System Reform." Health Policy 32: 257-77. Inaugural Conference, May, Vancouver. Govindaraj, Rarnesh, Christopher Murray, and Gnanaraj Chellaraj. Londofho,Juan Luis, andJulio Frenk. 1997. "Structured Pluralism: 1995. Health Expenditures in Latin America. World Bank A New Model for Health Reform in Latin America." Health Technical Paper 274. Washington, D.C. Policy (JUly). Hsiao, W C. 1992. "Cooperative Health Care: A Comprehensive Londofio, Juan Luis, and 0. Icoechea. 1996. "Latin American Strategy to Finance Health Care for Developing Nations." In Health Care Systems: A Simple Description of its Financing Proceedings from the World Health Organization Leadership and Organization." World Bank, Latin American and the Conference. Geneva: World Health Organization. Caribbean Region Technical Department, Washington, D.C. IDB (Inter-American Development Bank). 1996. Economic and Luft, H. S. 1996. "Modifying Managed Competition to Address Social Progress in LatinAmerica: 1996Report. Washington, D.C. Cost and Quality. " Health Affairs 15 (1): 23-38. Janovsky, Katja, ed. 1996. Health Policy and Systems Development: Mazzafiero, VE. 1996. "Health Conditions in Argentina: Facing An Agenda for Research. Geneva: World Health Organization. the Twenty-First Century." Journal of Public Health Medicine Katz, J., and J. Laspiur. 1993. "Reflexiones Comparativas sobre 18 (2): 234-37. el Modelo de Organizaci6ny Comportamientodel Sector Salud Medici, A. C. 1994. "Economia e Financiamento do Secor Saude en la Argentina, Chile y Uruguay" In OPS, ed., Estructura y no Brazil." AdSaide S6rieTematica, Faculdadde Sa6dePblica, Comportamiento del Sector Salud en laArgentina, Chiley Uruguay. Universidade de Sao Paulo, Sao Paulo. Cuaderno Tecnico 36. Washington, D.C. . 1996. "Opcoes para Reformar o Financiamento dos Katz,J.,J. Laspiur, and H. Arce. 1993. "UnModelo de Organizaci6n Servicos de Saude." In OPS, ed., Reunion EspecialsobreReforma y Comportamiento Sectorial." In OPS, ed., Estructura y del Sector Salud. Washington, D.C. ComportamientodelSectorSalud enlaArgentina, Chiley Uruguay. Medici, A. C, and C. Czapski. 1995. "Evolucao e Perspectivas Cuademo Tecnico 36. Washington, D.C. dos Gastos Publicos com Saulde no Brasil." World Bank, Latin Labadie, G., and D. Sanchez. 1993. "El Sector Salud en Uruguay" America and the Caribbean Department 1, Human Resources In OPS, ed., Estructura y Comportamiento del Sector Salud en Division, Washington, D.C. INNOVATIONS IN HEALTH CARE FINANCING 230 Mills, A. 1993. "Economic Aspects of Health Insurance." In K. Economico 36 (summer). Lee and A. Mills, eds., The Economics of Health in Developing Rossetti, A. G. 1996. "The Political Economy of Policy Change: Countries. Oxford: Oxford University Press. Concepts for the Analysis of Health System Reform." World Miranda Radic, E., and Y Vergara. 1993. "Evoluci6n y Estructura Bank, Latin American and the Caribbean Region Technical del Sector Salud en Chile." In OPS, ed., Estructura y Department, Washington, D.C. Comportamiento del Sector Salud en la Argentina, Chiley Uruguay. Robinson,J. C., and L. P Casalino. 1996. 'Vertical Integration and Cuademo Tecnico 36. Washington, D.C. Organizational Networks in Health Care."HealthAffzain 15(1): Munioz, A., 1993. "Evoluci6n de laAtenci6n de Salud enArgentina." 7-22. In OPS, ed., Estructura y Comportamiento del Sector Salud en Santana, A. D. 1996. "Los Servicios de Salud Pre-pagos en la Argentina, Cbile y Uruguay. Cuademo Tecnico 3 6. Washington, Republica Dominicana." Santo Domingo, Tercera Vers. D.C. Schieber, George, Jean-Pierre Poullier, and Leslie M. Greenwald. Murray, Christopher. 1994. "Towards an Analytical Approach to 1994. "Health Systems Performance in OECD Countries: Health Sector Reform: Investing in Health Systems." Bulletin 1980-1992." Health Affairs 13 (4): 100-12. Of the World Health Organization 72(3): 663-74. World Bank. 1993. World Development Report 1993: Investing in Musgrove, Philip. 1995. "Cost Effectiveness and Health Sector Health. New York: Oxford University Press. Reform." Human Resources Development and Operational . 1996. "Argentina: Facingthe Challenge of Health Insurance Policy Working Papers 48. World Bank, Washington, D.C. Reform." LatinAmerica and the Caribbean Region, Vashington, .1996.Public and PrivateRolesinHealth: Theoryand Fnncing D.C. Patterns. World Bank Discussion Paper 339. Washington, D.C. . Forthcoming. "Health, Nutrition and Population Sector Panadeiros, M. 1996. "Organizaci6n del Seguro de Salud en la Strategy Paper, Statistical Annex." Human Development Argentina: Analisis y Propuesta de Reforma." Desarrollo Department, Washington, D.C. MANAGED CARE AND MANAGED COMPETITION IN LATIN AMERICA AND THE CARIBBEAN 231 Medical Savings Accounts for Developing Countries Len M. Nichols, Nicholas Prescott, and Kai Hong Phua Schieber and Maeda (in this volume) outline the fun- anism is some kind of insurance, for its use is contingent damental health care financing choices for any econ- on certified medical events. This mechanism typically has omy. Most systems rely on a mix of public and private either a large or no explicit size limit, and thereby provides funds, using variations along a continuum between two backup financing for medical needs that would exhaust extremes: reliance on government through public financ- the medical savings account fund. Backup financing can ing, public providers, and government price and resource be wholly or partly private or public. controls; and reliance on markets through private financ- Medical savings account proposals assume that both com- ing, private providers, market prices, and private invest- ponents-the fund and the backup insurance-are present. ment in health facilities. Like the other papers in this volume, Both components are needed because the distribution of this paper assesses the advantages and disadvantages of health care expenditures is highly skewed. In the United financing options along this continuum for countries at States, for example, 10 percent of the population accounts different stages of development. Under certain conditions for 70 percent of health care spending in a given year (Berk medical savings accounts could play an important role in and Monheit 1992). The health spending of this 10 per- increasing resources, efficiency, and equity in market-ori- cent is, on average, eight times the populationwide average, ented health systems. This paper clarifies the preconditions, and that of the top 1 percent is thirty times the national risks, and risk management strategies for increased reliance mean expenditure. Similar distributions are observed in all on medical savings accounts, insurance, and market forces technologically sophisticated health care delivery systems. generally. This skewed distribution of health spending means that some kind of pooling or insurance mechanism is needed to Features of Medical Savings Accounts guarantee access to high-cost but necessary health care. There are two ways to pool health risks: cross-sectional, Financial structure across individuals and families during a given year; and intertemporal, over many years for an individual or family. Medical savings accounts have two essential components: The main problem with comprehensive insurance in a cross- a fund with balances that are set aside for medical expenses sectional pool is moral hazard: sufficient resources can be and that may accumulate at the discretion of the individ- marshaled for health care needs, but low copayment require- ual; and a backup financing mechanism. The backup mech- ments can lead to excessive demand for health services Len M. Nichols is principal research associate at the Urban Institute in Washington, D.C. Nicholas Prescott is senior economist in the EastAsia and the Pacific Region at theWorld Bank. KaiHongPhua is seniorlecturerin health policyand management in the Department of Community, Occupational, and Family Medicine at the National University of Singapore, and adjunct fellow at the Institute of Policy Studies in Singapore. The authors are grateful to Deborah Chollet, Alex Preker, and George Schieber for helpful comments on earlier drafts, though the authors retain responsibility for any errors or ambiguities. The opinions expressed here are those of the authors and do not necessarily reflect those of the Urban Institute, World Bank, or National University of Singapore. 233 (Chollet and Lewis in this volume). Thus some type of ily income. This means that patients will have to pay 100 demand management is required. The main problem with percent of medical costs, either from the medical savings intertemporal risk pooling is cash flow: until an individ- account or other personal funds, until a high level of expense ual's annual contributions to the medical savings account has been incurred. Deductibles can vary with income, but have been accumulated over many years, resources are insuf- this adds considerable administrative complexity. ficient for unpredictable and catastrophic health care needs. Thus some type of cross-sectional pooling is required to Mechanics insure against these needs. Combined, medical savings accounts and high-deductible Medical savings account contributions are accumulated in backup insurance can deal effectively with both pooling a fund managed by a trustee-an employer, bank, insurer, problems. Medical savings accounts help make patients or the state. When workers need health care, they can use more aware of health care costs and so reduce the moral the balance of the medical savings account to paythe required hazard associated with cross-sectional pooling, and backup fees. If medical needs in a given year exhaust the balance insurance reduces the cash-flow problem of intertemporal of the medical savings account, then additional out-of-pocket pooling over an individual's lifetime. By itself a medical sav- funds must be expended until the backup insurance pol- ings account is a weak intertemporal risk-pooling device for icy's deductible amount is reached. At that point the backup individuals or families. Although it can accumulate and thus policy starts to pay for the health care needs of workers or be used for a variety of future purposes, it could never their families. Depending on the policy, the family's oblig- finance the costs of being in the top 1 percent or even 10 ation is either greatly reduced or eliminated. percent of people needing health care. By contrast, backup If a worker's family does not use much health care dur- insurance, by pooling health risks across individuals and ing the year, then the medical savings account fund could families in a given year, can easily finance the high health have a positive balance at year's end. This balance could care costs of the few who need them with relatively low per be withdrawn and spent on other goods, or it could be capita premiums or tax payments. retained and accumulated for future health care needs. This Contributions to the medical savings account fund can balance could be used as savings for retirement or long-term be either voluntary or compulsory (through a mandate or health care needs. This source of savings could be particu- tax), and they can be made by individuals or by employers larly important for countries that now have declining depen- (including governments) on individuals' behalf. To encour- dency ratios and relatively few elderly citizens, but expect age the spread of medical savings accounts, a tax subsidy to have much larger retired populations in the future. Interest maybe granted to income or expenditures devoted to these could be earned on these account balances and taxed or funds. Contribution and withdrawal limits may also be not, depending on policy objectives. imposed to prevent high-income individuals from using medical savings accounts to evade tax liabilities. Health Care Financing with Medical Backup insurance policies can be purchased in private Savings Accounts and Insurance markets if feasible. The state can adopt the principles of managed competition to facilitate the development of pri- Distribution of financial risks vate insurance markets (see Enthoven in this volume). Alternatively, the state can sell insurance, with subsidies This section illustrates different combinations of public and for the poor, or directly provide backup insurance by pro- private financing and describes three health financing mod- viding services or subsidizing providers. To encourage els with medical savings accounts and backup insurance patients to be parsimonious when demanding care, the policies. All the models are special cases of the general one policy's deductible and out-of-pocket limit should be set at depicted in figure 1 through T(x) and HH(x), which rep- high levels, higher than the annual medical savings account resent total and household or privately financed expendi- contribution limit-perhaps as much as 10 percent of fain- tures, respectively' INNOVATIONS IN HEALTH CARE FINANCING 234 FIGURE I amount to providers on behalf of all patients. Whether it Health expenditure distributions occurs through a public insurance program for certain groups or a publicly financed price subsidy for all, the difference between T(x) and HH(x) can be interpreted as a measure- ment of the public payments for health care aggregated over all individuals. T, 7 7 No household pays for health care with private sources H, beyond the catastrophic expenditure level X2. Care pro- vided beyond X2 is completely under the discretion of the state. The more comprehensive are private insurance and Pnivate \Public < medical savings accounts, especially if mandated, the greater \ H(x) T(x) is the share of health expenditures financed outside the gov- ernment budget, since expanding private financing instru- X, X2 $/HH ments will shift HH(x) up and to the right. Annual household expenditures Altemative medical savings account models Annual household expenditures are measured along the horizontal axis. The vertical axis measures the percentage Keeping this expenditure distribution in mind, we now turn of households with particular health expenditure levels. The to examples of alternative financing models. The first med- entire area under T(x) represents total national health spend- ical savings account system we discuss has a purely public ing. This general framework can accommodate any financ- backup insurance mechanism; the second has a limited ing structure, for the difference between total and household private backup insurance system; and the third has a purely (private) spending is public spending.2 Different financing private backup mechanism. In each case we assume that systems have these lines closer or farther apart, depending contributions on behalf of each worker are by employers as on the importance of public subsidies. well as by workers. Household expenditures are the sum of out-of-pocket payments, private insurance payments for health services, Medical savings accounts with public backup. In the first and payments for health services made out of medical sav- model, individual medical savings accounts with a public ings accounts. Any or all of these components could be sector backup, contributions fund only the medical savings zero in any particular health financing system. The differ- account. Workers draw from this fund as health care needs ence between T(x) and HH(x) can be explained in two ways, arise for them or their family, now or in the future. Under but each interpretation represents some form of public pay- this model the state essentially compels individuals and fam- ment for health care. Consider expenditure level X1. T, ilies to create intertemporal risk-pooling devices using med- households have total expenditures equal toXI. HI house- ical savings accounts, and provides the cross-sectional holds might pay all of X1 out of pocket, with private insur- risk-pooling backup mechanism to finance health care needs ance orwith medical savings accounts, inwhich case T1 -H1 if the medical savings account is exhausted. This backup households get their entire X1 from the government.These assistance can be through a public insurance program, households could be said to be insured by a government through direct provision of subsidized care by public program. They may be poor or privileged. providers, or through public subsidies to private providers. Alternatively, all households could be primarily respon- For people with positive balances in their medical sav- sible for their own health care, but a government health ings account at the end of any year, the state has an impor- service subsidy equal to (T1 -. H1) X1 is shared among all tant policy choice. (This choice is present in all the health T1 households. That is, the government controls health ser- system financing models discussed in this section.) If the vice costs below market levels or total cost and pays some primary goal of the compulsory medical savings account MEDICAL SAVINGS ACCOUNTS FOR DEVELOPING COUNTRIES 235 program was to ensure the accumulation of sufficient pri- try's health care expenses, since the distribution of health vate funds to finance future health care needs (and thereby expenditures is so skewed and highly correlated with the avoid intergenerational transfers as the population ages), need for hospitalization. then the state should not allow balances to be withdrawn in full for non-health care purposes. But if the main pur- Medical savings accounts with private backup. The third pose of the medical savings account scheme was to reduce financing model has a fully private backup insurance sys- current health care spending by making patients pay for tem for people with medical savings accounts. The insur- health care, then the state should allow withdrawals for non- ance and cross-sectional risk pooling in this system are health purposes. Although both motivations are likely to complete in the sense that they cover all the health care be present in many cases, analysts can infer which motiva- needs of the insured above those financed by out-of-pocket tion is stronger by noting which of these alternatives each payments and the limited intertemporal pooling created by state chooses. medical savings accounts. In this case the public sector's responsibility is reduced to guaranteeing access for the poor. Medical savings accounts with mixed backup. The second Again, it can do so through public insurance, direct service model requires larger contributions by employers and work- provision, or provider subsidies. ers because they must coverboth the medical savings account These different systems and backup mechanisms for and a backup insurance policy. This high-deductible backup medical savings accounts are illustrated in figure 2. The first insurance provides cross-sectional risk pooling for families bar represents a traditional system, in which there is no pri- but is not large enough to fully cover all contingencies once vate insurance or medical savings accounts. Instead there medical savings accounts are exhausted. This option can is only a mix of public financing combined with private be viewed as a medical savings account with a private insur- out-of-pocket payments. Introducing medical savings ance "corridor." The corridor lies between the medical sav- accounts can reduce public spending because private ings account and the public backup system for truly resources are mobilized. Medical savings accounts can also catastrophic events. Thus the cross-sectional risk pooling replace direct out-of-pocket spending. Private insurance lia- here is partly private (over the insured only) and partly bility is smaller in the mixed backup case than in the purely public (over the whole society), or mixed. Corridor insurance policies have explicit lifetime or annual FIGURE 2 benefit limits, and when they are reached the public sector Financing health care with medical savings finances extremely high-cost cases. This corridor of limited accounts and insurance liability insurance may be a good way to nurture a private Percentage of health expenditure insurance market, but the state could also sell or compel 100 the purchase of insurance products that would accomplish the actuarial pooling required. We consider this corridor 80 "private" regardless of where the insurance is purchased, since the basic goal is to pool risk among the insured group insurance of individuals and families and not the society as a whole, 60 and since the funding for this insurance comes from indi- vidual contributions made by the insured (even though they 40 may be compulsory). It is also private in the sense that individuals have discretion over how this additional health 20 1t care purchasing power is spent. Privately directed purchasing power may have many important implications for the health Status Public Mixed Private delivery system (see below). Corridor insurance could eas- quo backup backup backup ilybe structured to finance a considerable portion of a coun- Financing system INNOVATIONS IN HEALTII CARE FINANCING 236 private backup case because the "corridor" model preserves accounthigh-deductible insurance arrangement (Keeler and public liability for the catastrophically ill. The purely private others 1996; Nichols, Moon, and Wall 1996; American backup system has the smallest public sector role, but even Academy of Actuaries 1995; Ozanne 1996; O'Grady 1996). it has considerable public spending because the state is This section highlights analytical issues by discussing a set presumed to subsidize insurance premiums and health ser- of hypotheses about the effects of medical savings accounts vice copayments and to directly finance service provision that could be tested with proper data and research design. for the poor. The discussion of these hypotheses is organized around In all medical savings account systems that include backup three broad themes common to public finance in develop- insurance, workers and their families have strong incentives ing countries: resource mobilization, efficiency, and equity. to use health resources parsimoniously at least up to a point, while truly catastrophic health care expenses are covered Resource mobilization either by the public sector or by private insurance. In both cases more private funds are drawn into the health financ- In health systems that have heretofore been wholly or mostly ing system than in the traditional system, increasing the public, creating medical savings accounts and private insur- state's options for allocating health resources. Total national ance or private discretion over health care purchasing may health spending could be larger under all systems with med- be the most politically palatable way to make the middle ical savings accounts and much larger under the mixed and upper classes pay more for health services, since most backup or private backup options, since they mobilize con- consumers place considerable value on discretion over health siderable private resources to pay for health care. A larger services and providers. Thus one hypothesis is that intro- share of private resources would enable budgetary resources ducing medical savings accounts will increase total spend- to be better targeted to the health care needs of the poor. ing on health as private resources are mobilized. Introducing In each system basic ambulatory health services for a rea- medical savings accounts as well as a compulsory backup sonably healthy family with average income could be cov- insurance mechanism would increase health spending even ered from their medical savings account plus moderate more. Underlying these hypotheses is the assumption that out-of-pocket obligations. Sicker and poorer families may the state will not reduce the absolute level of its health spend- require additional subsidies. ing, but may redirect it, as discussed below. Another resource mobilization hypothesis is that private Medical Savings Accounts and Health demand for health services will encourage providers and oth- Policy Objectives ers to organize and invest in more efficient health care insti- tutions in order to serve the emerging block of purchasing Medical savings accounts are controversial in health policy power. In essence, private purchasing power may make it research circles. Singapore's experience has been used to profitable to invest in health facilities. Over time this could support a particular side in the contentious U.S. political considerably augment the supply of health facilities. debate (Hsiao 1995), and one of this paper's goals is to clar- ify what Singapore's experience means for other countries- Efficiency especially developing countries. With two notable exceptions (Bond, Heshizer, and There are three broad hypotheses about the efficiency effects Hrivnak 1996; CBO 1997), reviewed below, practically all of medical savings accounts and backup insurance. It has the literature on medical savings accounts outside Singapore been argued that medical savings accounts will lower costs, has been theoretical-either advocacy pieces (Tanner 1995; rationalize health professional and health facility supply, and Goodman and Musgrave 1988; Pauly and Goodman 1995) improve the quality of care. or hypothetical microsimulation exercises designed to show who would win or lose financially upon switching from a Costs. As noted, the extremely skewed nature of health comprehensive insurance policy to a medical savings expenditures makes risk pooling essential. This pooling of MEDICAL SAVINGS ACCOUNTS FOR DEVELOPING COUNTRIES 237 resources sets up a natural tradeoff between shielding indi- contend. First, the medical savings account may still be vidual patients from financial ruin and the moral hazard looked on as someone else's money to spend, especially if problem described in Schieber and Maeda (in this vol- some of the contribution comes from employers (American ume) and Chollet and Lewis (in this volume). Patients are Academy of Actuaries 1995). Furthermore, since medical inclined to use more health resources than is optimal if the savings accounts can be used to pay the first incurred marginal price they face is artificially lower than the true expense, some patients may actually use more health care marginal cost. With insurance and risk pooling, this gap is than they did when they faced coverage limits or had to typically financed by an external third party. Medical sav- make out-of-pocket copayments (Pauly 1994). ings accounts create incentives to consume fewer health Second, providers have the power, as in any market resources because patients are responsible for the full cost with asymmetric information, to influence consumer deci- of care and are spending their own money, at least until the sions, especially once the deductible has been reached and backup insurance takes over. the marginal cost to the patient drops dramatically. The bulk There are two other ways to manage the excess demand of national health spending could actually occur at levels that results from risk pooling and insurance financing. The above the backup insurance policy's deductible (Nichols, first is for the state to control the supply of health care Moon, and Wall 1996; Keeler and others 1996). Providers providers until supply constrains effective demand. This have incentives to encourage use when they are paid on a approach will keep spending down, but it may not provide fee-for-service basis. Furthermore, high spending is gener- a desirable allocation of resources, since triage decisions are ally associated with complex hospital inpatient stays. The essentially made by the state.3 nature of these illnesses and treatment options makes fam- The second mechanism for dealing with excess demand idies much less willing to sacrifice services to save money for health care, increasingly common in the United States, than they are in more discretionary ambulatory contexts is managed care (see Enthoven in this volume). Managed (American Academy of Actuaries 1995). care, in all its many forms, essentially enforces triage deci- Third, the combination of medical savings account and sions and non-price rationing by private physicians. backup insurance produces substitution and income effects Competition in the marketplace for health plan enrollees that counteract each other. The right to keep unused med- protects patients from providers' incentives to underpro- ical savings account balances and the requirement to pay a vide care. Markets and quality reporting standards need to high insurance deductible mean that patients initially have be well developed for competition to be effective, and it relative price incentives to reduce health service use. At has proven to be so in some markets. the same time, the backup insurance policy is like an income By themselves medical savings accounts attack the root effect that will increase consumption, especially if this cov- cause of excess demand: copayment obligations that are erage is new. Thus the net effect of medical savings accounts below marginal cost. But the key difference between this on the level and rate of growth of per capita costs is some- approach to the moral hazard problem of health insurance what ambiguous. and the others is that consumers-not health plan profes- sionals or the state-get to choose. In essence, consumers Provider supply. To control costs, state-run health care ration their own health care instead of having elites ration systems have to limit the number of health professionals it for them. To the extent that health resources are free to and health facilities. In the discussion of resource mobi- follow the resulting dictates of consumer demand, medical lization above, we hypothesized that medical savings accounts savings accounts represent a tool for constructing a would lead to more private investment in health facilities. consumer-directed health care system. This is their advo- Here we argue that medical savings accounts and backup cates' ideal (Goodman and Musgrave 1988). insurance, by directing demand to private facilities, will raise Higher cost-sharing obligations will surely reduce health the prices of some professionals' services and thereby increase service utilization. But cost-containment results may not be the long-run supply of those professionals (assuming that as dramatic as some proponents of medical savings accounts the state relaxes artificial admission and certification lim- INNOVATIONS IN HEALTH CARE FINANCING 238 its). To the extent that this privately directed demand Another downside risk is that private demand may finance becomes dominant, the resulting supply of providers may a high-quality upper tier in a two-tiered health system, espe- be more reflective of consumer preferences than the provider cially where highly trained physicians and state-of-the-art mix produced by the state planning apparatus. hospitals are scarce. The existence of the upper tier could reduce willingness to support public financing of the lower Quality. Similarly, the privately directed purchasing power tier. Thus it could destabilize the entire system. Cross-sub- created by medical savings accounts and backup insurance sidies may be necessary, as well as vigilant attention to will likely increase demand for quality care outside state quality differentials between tiers. facilities. In some (mostly) public systems resource con- In addition, at least with the private backup insurance straints and rational triage require lower quality in some mechanism, there is some risk that out-of-pocket obliga- areas of care in order to provide adequate quality to the tions could be large relative to income. Ways to minimize larger number of patients in most areas. Medical savings this risk include limits on the out-of-pocket maximum accounts and backup insurance create the ability to pay for and policies that maintain competitive or regulated health higher quality in more limited areas. This may or may not service prices. Some analysts fear that the incentives of affect the average quality of care, depending on other pol- medical savings accounts will lead patients to neglect pre- icy decisions discussed below. ventive and basic care and thus frustrate the delivery of cost-effective care. There is some evidence that patients Equity in industrial countries are not well informed about com- plex health services, and that when faced with strong finan- Mobilizing private resources to finance medical savings cial incentives, they tend to reduce both necessary and accounts and insurance will cause middle-class citizens to unnecessary health services in equal measure (Newhouse pay more and low-income citizens to pay less for health care, and the Insurance Experiment Group 1993). At the same if the state targets spending on the low-income population. time, reports from U.S. companies that have implemented Thus medical savings accounts and private health insurance medical savings accounts in the past few years do not indi- make it possible to expand access to health care without cate any problems of this sort, although adequate data are increasing public resource requirements. There is no guar- not yet available for systematic and independent study antee, however, that the state will take this approach. (CBO 1997). Moreover, there are equity risks, at least during the tran- sition. Providers that suddenly have excess demand may raise Risk selection concems their prices as governments dismantle the price controls of regulatory public systems. In the long run, free entry into By far the most contentious issue in the U.S. debate over health professions and facility construction may guarantee medical savings accounts has been risk selection-the fear that competitive markets will prevail, but educating health that medical savings accounts would appeal mostly to the professionals and building health facilities take considerable healthy, and that payments for health care would then become time. The results could be enriched providers who leave the more correlated with health status than they are now (see controlled public sector for the more lucrative private sec- Nichols, Moon, and Wall 1997 and the references cited tor, abetter delivery system for the wealthy, and fewer health therein). This fear is relevant to all voluntary systems in resources available for everyone else. This risk may justify which people have the option of buying more comprehen- using state health personnel policy, at least during the tran- sive insurance. Although this maybe an important issue for sition to competitive health services markets, to guarantee developing countries that eschew compulsory participation, that enough providers continue to serve public sector patients. this paper focuses on a more common problem associated Such personnel need not remain state employees, but they with the skewed distribution of health expenditures: set- may be required to work one or two days a week at public ting the medical savings account contribution appropriately. facilities and receive lower remuneration for those days. Public and private backups are considered separately, MEDICAL SAVINGS ACCOUNTS FOR DEVELOPING COUNTRIES 239 To illustrate the problem, consider a system that is finance savings accounts to achieve spending and access targets. entirely by public funds. In this case, setting a medical sav- One way to do this is to dictate the parameters-contri- ings account level is equivalent to determining the amount bution levels, services covered-of the privately financed that will be given to each citizen in lieu of public coverage medical savings account options. This process is straight- for at least some health care needs. If the citizen uses no forward in compulsory savings or tax systems. If the state health services, then the state has "lost" money on that cit- instead allows voluntary medical savings account contribu- izen because it could have used that money to pay for some- tions, then it must respond to market-determined medical one else's health care. The theoryis that byproviding medical savings accounts. But medical savings accounts cannot savings accounts and their incentives for parsimony, peo- find a market equilibrium until the state reveals its pro- ple who do use health resources will use sufficiently fewer gram parameters. This creates a strategic situation in which of them to compensate for the public money that would be state and insurer interests may diverge. The general point, "wasted" on people who do not use them. Institutionalizing however, is that medical savings account and insurance para- mnedical savings accounts may help teach people that health meters will be coordinated one way or the other, for one care is not free for anyone, and this may be important in a set is contingent on the other. long-run development strategy. But as a short-run techni- cal point it should be noted that it is possible to set the Singapore's Experience publicly backed medical savings account too high and thereby "lose" money if more unspent health moneys go to the A detailed overview of Singapore's health care delivery healthy than the sick save (CBO 1996). and financing system is provided by Phua (in this volume). With private backup insurance, the medical savings This section highlights the salient policy choices and their account amount must be coordinated with the deductible effects. of the private insurance policy that accompanies it. The ana- lytical problem is similar to the previous case, except that Institutional features in this case the insurer, not the state, is the residual loser or beneficiary if the medical savings account amount is set Singapore is one of the East Asian "tiger" economies-a too high or too low relative to the deductible. If a medical small, high-income city-state with a population of just 3 mil- savings account contribution is too large, it will encourage lion but a per capita GNP of $22,500 (by comparison, per health care use that increases the insurer's costs. If it is too capita GNP in the United States is $25,880). Singapore's small, it will discourage use and lead to short-run insurer experience with medical savings accounts can be used to profits and lower premiums in a competitive market. The draw lessons for implementation of these accounts in other simplest way to solve this coordination problem is to let countries. In 1984 Singapore became the first economy to private insurers offer high-deductible premiums and med- implement medical savings accounts on a nationwide basis. ical savings account packages, for then both would be set And to this day the Medisave scheme remains the world's actuarially, at least in competitive equilibrium. only example of an applied medical savings account pro- With public backup insurance and a privately financed gram integrated with a country's health financing structure. medical savings account, the state faces the same contin- The medical savings account model that has evolved in gent liability problem as the insurer in the private backup Singapore over the past decade corresponds to the med- case. If the medical savings account is too small relative to ical savings account with mixed backup described earlier. the insurance program's deductible, it will discourage use In other words, individual Medisave accounts are embed- and could save the state money by not covering much health ded in a broader financing framework that backs up the care use. If it is too large, it could increase state expendi- medical savings accounts with a cross-sectional catastrophic tures by increasing demand for the high-end care that the risk-pooling scheme called Medishield (introduced in 1990) state finances. Thus the state needs to coordinate its insur- and a means-tested safety net for the poor called Medifund ance program parameters with privately financed medical (introduced in 1993). This three-tier package-Medisave, INNOVATIONS IN HEALTH CARE FINANCING 240 Medishield, and Medifund-is backed up by government Medifund Committee. This safety net is targeted at house- financing of supply-side subsidies to public providers aimed holds with two adults and three children earning less than at lowering the net prices charged to patients (table 1). S$ 1,400 a month (the lower one-third of the income The mechanics of the Singapore model are straightfor- distribution). ward. Contributions to Medisave are an integral part of Singapore's compulsory social security coverage, provided Effects on health policy objectives by the Central Provident Fund (CPF). Enrollment in the CPF is mandatory for all employees (and, since July 1992, Resource mobilization. The main policy objective guid- for the self-employed). The CPF is funded by a mandatory ing the introduction of medical savings accounts was to payroll tax equivalent to 40 percent of the wage bill, split mobilize nonbudgetary resources to help pay for the increas- evenly between employers and employees. Of this 40 per- ing health costs of Singapore's aging population. Medical cent contribution, between 6-8 percentage points are allo- savings accounts were expected to absorb pressures that cated to the member's Medisave account. These would otherwise drain the budget of the Ministry of Health, contributions are income tax-deductible and interest bear- freeing the ministry to focus on the priority government ing. The Medisave balance can accumulate up to S$ 19,000, functions of public health and protecting the poor. beyond which incremental savings are rolled over into the A complete assessment of Medisave's effectiveness in CPF member's ordinary account, from which it can be with- mobilizing resources for health must take into account both drawn after age 55. static and dynamic perspectives. A static view of Medisave's Medishield is also managed under the CPF umbrella- importance is given by its weight in the overall structure of all CPF members are automatically covered, and the annual health financing. In 1995 Singapore's ratio of total health premiums are deducted from their Medisave account unless care costs to GDP was estimated at just 2.7 percent. Of they choose to opt out. Medifund, however, is not managed this, private expenditures accounted for 60 percent and by the CPF. It is an autonomous endowment fund into which public budgetary spending, for 31 percent. periodic budget transfers are made, and from which invest- Public extrabudgetary expenditures financed 9.6 per- ment income is used to finance demand-side subsidies for cent of total health spending-mostly Medisave (8.5 per- the poor. cent), followed by Medishield (0.8 percent) and Medifund Withdrawals can be made from individual Medisave (0.3 percent). This static perspective shows that Medisave accounts to pay medical bills incurred by the account holder and immediate family members. However, withdrawals TABLE I Sources of health financing in Singapore, 1984-95 are subject to two important exclusions. First, because (percent) Medisave is designed to pay for hospitalization expenses, ambulatory care is generally not covered. Second, eligible 1984 1986 1988 1990 1992 1994 1995 hospitalization expenses are capped (S$ 300 a day for hos- Public sector 37.1 37.2 33.0 32.0 34.0 39.3 40.3 pital charges plus limits for each surgical operation). Thus Budgetary 35.3 27.8 21.3 21.2 24.4 29.7 31.0 the average hospital bill requires a significant copayment Capital 26 o 2.9 25 615 8.8 11I2 Current (net) 26.8 24.2 18.3 1 8.8 17.9 20.9 19.8 in addition to the portion paid by Medisave. Claims for Extrabudgetary 1.8 9.4 11.7 10.8 9.7 9.6 9.6 Medishield backup coverage for catastrophic expenses are Medisave 1.8 9.4 11.7 10.7 9.4 8.8 8.5 Medishieid n.a. n.a. n.a. 0.0 0.3 0.5 0.7 subject to a high annual deductible (S$ 1,000) as well as a Medishield Plus na. na. na. n.a. n.a. 0.0 0.1 20 percent copayment. In addition, Medishield coverage Medifund n.a. n.a. n.a. n.a. n.a. 0.3 0.3 exdudes preexisting conditions and is subject to claim lim- Private sector 62.9 62.8 67.0 68.0 66.0 60.7 59.7 its of S$ 20,000 per policy year and S$ 80,000 per lifetime. Out-of-pocket 62.9 62.8 67.0 66.5 63.9 58.7 57.7 Finally, Medishield coverage expires at age 75. As a last Prvateinsurance n.a. n.a. n.a. n.a. 1.6 1.8 2.0 . . ~~~~ ~ ~ ~~~~~Total 100.0 100.0 100.0 100.0 100.0 100.0 100.0 resort, patients unable to pay their bills at government hos- Tota0. 00 0 0 00 . 0 pitals can apply for a mean-tested grant from their Hospital Source: Woa d Bank staf estimates. MEDICAL SAVINGS ACCOUNTS FOR DEVELOPING COUNTRIES 241 has succeeded in playing an important though not at all not waste its resource pool on financing medical events that dominant role in mobilizing resources to finance health are generally affordable as part of routine consumption expen- expenditures. Indeed, if the role of Medisave is expressed ditures, and for which insurance does not produce any wel- relative to hospitalization expenditures-which is all it is fare gain. Since outpatient care could consume as much as intended to help finance-it appears much more significant two-thirds of health expenditures, this significant exclusion (almost 30 percent). frees a large amount of contributions to be rolled over into From a dynamic perspective, however, it seems likely individuals' lifecycle savings pool. This, in turn, allows that Medisave wiDl play an increasingly important role in financ- Medisave to achieve some degree of intertemporal risk pool- ing as Medisave balances continue to accumulate over indi- ing to cover intermediate financial risks, which could pro- vidual lifecycles. Today's flow of disbursements from Medisave duce a welfare gain to its beneficiaries (depending on their accounts understates the flow of contributions, and there- rate of time preference and the payout rate). In other words, fore the accumulating stock of Medisave balances available excluding outpatient coverage has an indirect dynamic to finance future claims. The actual Medisave payout ratio resource mobilization effect by lowering the payout rate, (expenditures as a share of contributions) has consistently which allows Medisave to perform an efficient intertempo- been around 20 percent. In fact, the payout ratio has been ral insurance function for hospitalization. Thus the outpa- faling because the nominal daily cap of S$ 300 has not been tient exclusion turns out to be a win-win policy on both counts. adjusted since inception. For example, in 1995 Medisave pay- Second, the insurance effect of Medisave coverage for ments totaled S$ 291 miDlion, compared with estimated con- intermediate (capped) hospitalization costs is accompanied tributions of S$ 1,809 million. That same year, the cumulative by efficiency incentives on both the demand and the supply savings generated by the low payout rate had already mobi- sides. On the demand side, the Medisave expenditure caps lized a total Medisave balance of S$ 12,700 million-equiv- mean that a significant portion of hospital bills are not cov- alent to four years' worth of Singapore's aggregate annual ered by Medisave. World Bank estimates suggest that 60 health spending. The dynamics of the underlying savings accu- percent of hospitalization costs in public hospitals are sub- mulation are illustrated by the rise in per capita balances (from sidized. The residual 40 percent that is charged to patients S$ 2,595 in 1985 to S$5,400in 1995) aswellasbytheincrease is split about evenly between Medisave and out-of-pocket in the share of employees who attain the required minimum payments. Thus patients feel the bite of individual respon- balance by age 55 (44 percent in 1995). sibility not only in the form of an average 20 percent coin- Medisave's exceptional resource mobilization potential surance fraction paid indirectly out of their Medisave account, results from two crucial features. First, its revenues are but of another 20 percent paid directly out of pocket. On supported by a sizable payroll tax base made possible by the the supply side, Singapore recently imposed revenue caps extensive formal employment in Singapore's rich, urban- on the government's restructured hospitals that limited allow- ized economy. Nearly universal coverage-the number of able increases in hospital revenue per patient per day to 5 individual Medisave accounts is equivalent to 80 percent percent a year during 1994-96. Singapore's remarkably low of the population, and dependent family members are also ratio of health care costs to GDP is consistent with these eligible-has been achieved by tapping into the well-estab- efficiency incentives for containing costs. lished social security operations of the Central Provident Fund. Second, Medisave's expenditures are sharply con- Equity. Medisave's effects on equity are not directly tained by its twin policy exclusions limiting coverage pri- observable in terms of different utilization rates or afford- marily to hospitalization, with defined expenditure caps. ability between income groups. There are, however, two sig- nificantly pro-poor elements built into the financing Efficiency. Medisave's design is consistent with principles framework in which Medisave is embedded. of economic efficiency in two important ways. First, by not First, government subsidies continue to play an important covering low-cost, high-probability outpatient care-unlike backup role in financing hospital care, and are targeted to comprehensive health insurance schemes-Medisave does the poor. As noted above, about 60 percent of public hospi- INNOVATIONS IN HEALTH CARE FINANCING 242 tal costs are subsidized. Subsidies are channeled to public kept in mind that the overriding goal of U.S. policy is to hospitals and reflected in public pricing policies. Explicit price contain costs, not to mobilize resources for the health sector. discrimination is based on the four types of accommoda- Ozanne (1996) focuses on how firms report their results tions in public hospitals, ranging in ascending order of com- differently, and how the overall satisfaction with medical fort from classes C, through B2 and B 1, to A. Subsidy ratios savings accounts differs. For example, some employers report are highly differentiated-averaging 84 percent of hospital that their costs have fallen substantially, while others report costs in class C, 71 percent in class B2, 36 percent in class modest savings. Some firms have discontinued medical sav- B 1, and 13 percent in class A. Public hospitals provide finan- ings accounts after using them for a while, though no rea- cial counseling to prospective inpatients to facilitate selec- sons are publicly stated. CBO (1997) also outlines the data tion of an affordable ward dass based on the expected size that need to be reported to properly evaluate medical sav- of the hospital bill and the individual's Medisave balance. ings accounts, concluding that "the reports generally do not These differential subsidy ratios help equalize the afford- provide enough information to fully assess how successful ability of dass-specific prices relative to the incomes and the plans were at reducing medical spending by and on Medisave balances of the patients who select them. For behalf of employees and limiting adverse selection" (p. 2). example, in 1995 the 75th percentile hospital bill charged Most of the reported cost data refer only to firm costs or to patients in class C was S$ 524, compared with S$ 854 in insurance claims filed, not total costs (including employees' class B2. These bills compare with median monthly house- out-of-pocket expenditures). Furthermore, there are little hold incomes of S$ 787 in the bottom quintile of house- data on the health status of workers before they chose their holds and S$ 1,657 in the second quintile. Government plan and the costs to employees of different plans. estimates suggest that at those income levels the corre- Bond, Heshizer, and Hrivnak (1996) asked two basic sponding bills were equivalent to 7.4 and 9.5 months of questions of firms that had switched from comprehensive Medisave contributions, respectively. insurance plans to medical savings accounts: Did employ- The second important pro-poor financing instrument is ers' costs and employees' maximum out-of-pocket liabili- Medifund. Although intended as a safety net of last resort ties go up or down? They found that employers had saved forthe poor, available data indicate that, during its first three an average of 12 percent and employees' maximum out-of- years of operation, Medifund responded favorably to 99 pocket liability had fallen by about a third. Although both percent of requests for financial assistance, and paid the results are partly due to higher than average baseline costs, entire medical bill in 87 percent of those cases. Patients the findings show that, on average, these firms are protecting receiving assistance accounted for 5 percent of hospital workers better than they were before medical savings admissions at the class C and B2 levels (3.3 percent in 1993, accounts. Whether this increased protection is commen- 4.2 percent in 1994, and 5.8 percent in 1995). surate with lower employer costs in long-run equilibrium is an empirical question that will be closely examined. Experience in Other Countries One reason the U.S. experience with medical savings accounts will be watched even more closely than usual is a Although China and Malaysia are exploring medical savings new law, The Health Insurance Portability and Accountability account options for their health systems, the United States Act, signed by President Bill Clinton on 8 August 1996. is the only other country where medical savings accounts Starting 1 January 1997, a limited number of employees of are part of the health financing structure and reports of small firms (fewer than fifty employees) and the self- their effects have been made. CBO (1997) surveys all the employed will be given tax preferences for setting up med- publidy available evidence from employers that have switched ical savings accounts along with their high-deductible from comprehensive insurance plans to medical savings insurance plans. The U.S. General Accounting Office is accounts and high-deductible insurance, and Bond, Heshizer, required to study this demonstration and report to Congress and Hrivnak (1996) surveyed twenty-seven firms offering by 1 January 1999. At that time the tax preference for med- medical savings accounts to their employees. It should be ical savings accounts may be expanded or eliminated. MEDICAL SAVINGS ACCOUNTS FOR DEVELOPING COUNTRIES 243 Lessons for Developing Countries effects that counterbalance the income effect of more cov- erage generally. The efficiency effect of these price incen- Singapore's experience and our theoretical discussion lead tives may be enhanced by educating the public about prices to five broad lessons for developing countries. and comparative costs-especially through financial coun- seling prior to utilization, as well as instant feedback on Lesson I Medical savings accounts cannot be used alone medical savings account balances upon billing and payment. It is not actuarially feasible to have complete self-insurance Lesson 4 Medicol sovings accounts pose equity risks for health care. Medical savings accounts can be important tools for mobilizing health resources, but only as part of a Having households pay more out of pocket at the point of comprehensive set of financing instruments. Cross-sectional service could raise already high barriers for low-income fam- backup insurance must also play a major role, and public ilies to fully participate in the health care delivery system. funding for catastrophic cases and for the poor will con- Thus public subsidies for medical savings account contribu- tinue to be necessary. Health personnel policies and health tions, price schedules that vary by income, and public subsi- service price controls are also likely to be useful policy dies for insurance premiums may be needed to ensure that tools to contain costs and maintain equity in the move toward the evolving health care system is not limited to the wealthy. a more market-oriented health system. Risk selection is a less serious problem in systems with mandatory participation (and high compliance) than it is in Lesson 2 Mobilizing resources will take a long time voluntary systems. Still, the medical savings account amount must be coordinated with the deductible of the backup The resource mobilization effects of medical savings accounts insurance policy that accompanies it. Uncoordinated lev- could help avoid the intergenerational transfer problems els are unsustainable, especially if the system mixes public that will be especially acute in developing countries with and private insurance mechanisms. Unstable insurance sys- rapidly aging populations and shrinking tax bases. But it tems always threaten the most vulnerable-that is, those may take quite a few years to accumulate sufficient med- with recognizably great health care needs. ical savings account balances to transfer significant finan- cial risk to households. Building public and provider Lesson 5 There are major institutional prerequisites for understanding of and support for systemwide change will implementing medical savings accounts take time and is essential for a system based on private ini- tiative and market incentives to work. First, per capita income levels must be high enough to finance individual contributions to medical savings accounts as well Lesson 3 Medical savings accounts can and should be as premiums to a backup catastrophic risk pool. Second, a designed to enhance efficiency high degree of labor force participation in formal sector employment is needed to provide a taxable transactions This is particularly important in systems that are mobiliz- base for resource mobilization. These conditions are unlikely ing more resources and rapidly expanding their health care to be met in many developing countries, and may rule out delivery systems. Increased resource mobilization and pur- any immediate possibility of nationwide coverage with med- chasing power will produce a significant income effect that ical savings accounts. However, such accounts could still could lead to substantial increases in health service utiliza- play a niche role for high-income urban employees, just as tion. Medical savings accounts can be structured to increase existing social insurance arrangements do for civil servants efficiency in these utilization decisions by insuring that sub- and industrial workers in many countries. Third, an effec- stantial out-of-pocket payments are required before backup tive system of payroll tax collection combined with effi- insurance policies take on the bulk of the financial burden. cient fund management and claims processing is needed to Medical savings accounts can create incentives or price implement the financial operations associated with medical INNOVATIONS IN HEALTH CARE FINANCING 244 savings accounts. This is likely to be much easier in coun- Bond, Michael T, Brian Heshizer, and Mary W Hrivnak. 1996. tries that already have a social security mechanism with "Reducing Employee Health Expenses with Medical Savings which medical savings accounts can be integrated. Fourth, Accounts." Cleveland State University, Ohio. CBO (Congressional Budget Office). 1996. "Medical Savings a well-developed computerized information system ii~lnk Accounts and Medicare." Washington, D.C. ingpersonal savings accountswith hospital providers, backed . 1997. "A Review of Reports of Employer Experiences by strict security and accounting controls, is essential. Finally, with Medical Savings Accounts. " Washington, D.C. in countries where medical savings accounts and their backup Goodman, John C., and Gerald L. Musgrave. 1988. "Freedom of insurance coverage are, for equity reasons, embedded in a Choice in Health Insurance. " National Center forPolicyAnalysis health financing structure that protects the poor through Policy Report 134. Dalas, Texas. Hsiao, Wlliam C. 1995. "Medical Savings Accounts: Lessons from self-targeted subsidies, the state must have the administra- Singapore." Health Affairs 14(2): 260-66. tive capacity to implement a policy of price discrimination. Keeler, Emmett B., Jesse D. Malkin, Dana P Goldman, and Joan This requires an ability to channel supply-side subsidies to L. Buchanan. 1996. "Can Medical Savings Accounts for the providers (public or private) while monitoring and regulat- Nonelderly Reduce Health Care Costs?"Journal ofthe American ing the resulting hospital price structure. Medical Association 275(21): 1666-71. Newhouse, Joseph P, and the Insurance Experiment Group. 1993. Notes FreeforAI1? Lessonsfrom the Rand Health Insurance Experiment. Cambridge, Mass.: Harvard University Press. Nichols, Len M., Marilyn Moon, and Susan Wall. 1996. "Tax 1 . lT(x) and HH(x) are drawn with the commonly observed, Preferred Medical Savings Accounts and Catastrophic Health approximately log norrnal shape for health expenditure density I N functions. ~~~~~~~~~~~~~Insurance Plans: A Numerical Analysis of Wlners and Losers. " functions. Working Paper 06571-002. Urban Institute, Washington, D.C. 2. We assume that employers finance their contributions to ____ 197 "Epria Wor on Meia aig cons employees' health insurance premiums through lower wages. Thus What We Know Now, What We Need to Know, and Howue households are the ultimate source for employer payments as well. Mht Wearn the Rest P e peed to the annual me 3. Triage is a process for sorting injured or sick people into ng of the Red Pal Sciente to Janual mew groups based on their need or likely benefit from immediate gorleans medical treatment. Triage is used on the battlefield, at disaster OGrady, Michael J. 1996. "Medical Savings Accounts asd the sites, in hospital emergency rooms, and anywhere else that lim- DynaMics of Advrs Selectn Congssional arch ited medical resources must be allocated. Today, medical resources Service Report for the US.S. Congress. Washington, D.C. are considered to be limnited virtually everywhere. SrieRpr o h ..Cnrs.Wsigo,DC Ozanne, Larry. 1996. "HowWil Medical Savings Accounts Affect Medical Spending?" Inquiry 33(3): 225-36. Pauly, Mark V 1994. Do Two Wrongs Make a Right? An Analysis References of Medical Savings Accounts. Washington, D.C.: American Enterprise Institute. AmericanAcademyofActuaries. 1995. "Medical SavingsAccounts: Pauly, Mark V, and John C. Goodman. 1995. "Tax Credits for CostImplications and Design Issues." Public Policy Monograph. Health Insurance and Medical SavingsAccounts." HealthAffairs Washington, D.C. 14(1): 126-39. Berk, Mark, and Alan Monheit. 1992. "The Concentration of Tanner, Michael. 1995. "Medical Savings Accounts: Answering the Health Expenditures: An Update."Health Affairs 11(4): 145-49. Critics." PolicyAnalysis 228. Cato Institute, Washington, D.C. MEDICAL SAVINGS ACCOUNTS FOR DEVELOPING COUNTRIES 245 Medical Savings Accounts and Health Care Financing in Singapore Kai Hong Phua ingapore maintains the world's longest-running exper- gency ambulance service to transport accident and trauma S iment with medical savings accounts.These accounts, cases and medical emergencies to the acute general described generally in Nichols, Prescott, and Phua hospitals. (in this volume), help defray public health care costs while Public health services are provided through three gov- ensuring that citizens receive adequate and affordable cov- ermnent ministries. The Ministry of Health is responsible erage. Moreover, the system encourages Singaporeans to for preventive, curative, and rehabilitative health services. stay healthy and minimize the use of unnecessary medical It formulates national health policies, coordinates private services. and public health care, and regulates health standards and legislation. The Ministry of Environment is responsible for Health Care System environmental public health services such as overseeing sew- erage and waste disposal systems, ensuring the safety of Singapore's health care delivery system is mixed. The pub- food that is prepared and sold in Singapore, and control- lic system is run by the government; the private system is ling infectious diseases, air and water pollution and toxic run by voluntary and private hospitals and practitioners. chemicals and poisons. The Ministry of Labor is responsi- The health care delivery system comprises primary health ble for the industrial and occupational health of workers. care provision at private medical clinics and government Singapore's health care philosophy emphasizes the build- outpatient polyclinics, and secondary and tertiary care at ing of a healhy population through preventive health care private and public hospitals. programs and the promotion of healthy living. Public health About 80 percent of primary health care is provided by education programs encourage Singaporeans to adopt a private practitioners; the government polyclinics provide healthy lifestyle and be responsible for their own health. the remaining 20 percent. For hospital care, which is more The public is made aware of the adverse consequences of costly, the situation is reversed: 80 percent of hospital care habits like smoking, alcohol consumption, unhealthy diets, is provided by the public sector and 20 percent is provided and sedentary lifestyles. The government polyclinics offer by the private sector. every child free immunization against tuberculosis, Patients are free to choose among providers in the dual poliomyelitis, diphtheria, whooping cough, tetanus, measles, health care delivery system and can walk in for a consul- mumps, and rubella. These immunizations are also provided tation at any private clinic or government polyclinic. For at private clinics for a small fee. Health screening pro- emergencies patients can go to the twenty-four-hour acci- grams have been introduced for the early detection of com- dent and emergency departments located in government mon ailments like cancer, heart disease, hypertension, and hospitals. Singapore's Civil Defense Force runs an emer- diabetes mellitus (Seow and Lee 1994). Kai Hong Phua is senior lecturer in health policy and management in the Department of Community, Occupational, and Family Medicine at the National University of Singapore, and adjunct fellow at the Institute of Policy Studies in Singapore. 247 The government ensures that good and affordable basic pharmacy services. About 770 private clinics are run by 1,060 medical services are available to all Singaporeans by pro- medical practitioners. The average outpatient consultation viding heavily subsidized medical services at public hospi- fee (including medication) is about S$ 10-15, well within tals and government clinics. All private hospitals, medical the means of every Singaporean. At the government poly- clinics, clinical laboratories, and nursing homes are required clinics children and the elderly (above 60 years) are given to maintain a good standard of medical services through up to a 50 percent concession in their payment. licensing by the Ministry of Health. To promote personal responsibility, Singaporeans are Hospitol services required to use the Medisave scheme to save for their hos- pitalization expenses, especially during old age. This is to There are about 10,500 hospital beds in twenty-four hos- avoid overreliance on state welfare or medical insurance. pitals in Singapore, or about 3.5 beds per 1,000 people. Under the scheme every employee puts 6-8 percent of their About 80 percent of these beds are in the twelve public hos- monthly income into a personal Medisave account. These pitals, which have between 200 and 2,500 beds. The twelve savings can be used to pay for hospitalization expenses private hospitals tend to be smaller, with 60 to 500 beds. incurred by the worker or their family As noted, this sys- The public hospitals set the standard of medical care and tem encourages individuals to stay healthy and minimnize benchmark for hospital charges. the use of unnecessary medical services. Of the twelve public hospitals, six are acute general The public system requires that patients make copay- hospitals. The others specialize in areas such as obstetrics ment for their medical services at the time of consumption and gynecology, psychiatry, and infectious disease. The pub- to discourage unnecessary use. For people who choose to lic general hospitals provide multidisciplinary inpatient be accommodated in the lower classes of wards in public and specialist outpatient services and twenty-four-hour acci- hospitals (there are four classes), hospitalization expenses dent and emergency service. In addition, there are spe- are subsidized up to 80 percent by the government. Smaller cialty institutes for cancer, heart, eye, and skin diseases. subsidies are given to people who prefer the comforts and Tertiary specialist care for cardiology, renal medicine, hema- personalized service of the higher-class wards. The indigent tology, neurology, oncology, radiotherapy, plastic and recon- can apply to hospital management for partial or full remis- structive surgery, pediatric surgery, neurosurgery, sion of their bills at public hospitals. No Singaporean is ever cardiothoracic surgery, and transplant surgery are central- denied access to the health care system or use of the acci- ized in two of the larger general hospitals, Singapore General dent and emergency services, and no one is turned away by Hospital and National University Hospital. The private hos- hospitals. Waiting time for an elective operation average, pitals have similar specialists and comparable facilities. between two and four weeks. There is no waiting for emer The government has also introduced low-cost community gency admissions. hospitals for intermediate health care for the convalescent sick and aged who do not require the more expensive care Primory health services of the acute general hospitals. In public hospitals, Singaporeans can choose from dif- Primary health services include primary medical care for ferent types of wards and accommodations upon admission. families, health screening and preventive health programs Patients pay more for a higher level of physical amenities, for schoolchildren, home nursing, day care, and rehabilita- although the provision of medical care is similar for all accom- tion for the elderly, and health education and promotion for modations. In most cases serious medical conditions are all. treated in the public hospitals by senior consultants or spe- The public sector comprises sixteen government poly- cialists, regardless of the type of ward chosen by the patient. clinics located throughout the country. Each clinic pro- The average length of stay in the general hospitals is about vides curative outpatient medical treatment, immunization, five days. Hospital beds are well utilized, with an average health screening and education, investigative facilities, and occupancy rate of about 80 percent. INNOVATIONS IN HEALTH CARE FINANCING 248 Since 1985 the government has restructured (or corpo, * Rising expectations and demand for better and more ratized) five of its acute hospitals and two specialty institutes sophisticated health services by an increasingly well- to be run as private companies wholly owned by the gov- informed and more affluent public. ermnent. This was done to give the corporatized hospitals * Rapid aging of the population, with the portion of peo- the managerial autonomy and flexibility to promptly respond ple 60 years and above estimated to increase from 9 to the needs of the patients. In the process commercial percent at present to 25 percent by 2030. accounting systems have been introduced, providing a more * Shortage of staff such as nurses and health therapists, accurate picture of operating costs as well as instilling greater as young people become more attracted to less demand- financial discipline and accountability. Corporatized hospi- ing jobs than those in the health care services. tals are different from other private hospitals in that they Singapore has sought to control supply factors in the receive an annual government subsidy to provide medical health sector through deliberate staffing and facilities plan- services. Moreover, they are expected to be managed like a ning. On the demand side, efforts have been made to mobi- nonprofit organization. The corporatized hospitals are sub- lize compulsory savings through the Central Provident Fund ject to broad policy guidance by the Ministry of Health. (CPF), thus rationing demand implicitly through consumer purchasing power. The government uses social planning to Dental health services avoid the problems faced by most countries in maintaining a balance between demand pressures and supply capacity. Under its dental health program the government provides The situation could potentially be more acute for a small free dental care for all schoolchildren. Both general and spe- nation like Singapore, where resources are limited and expec- cialized dental care are provided in five hospital dental tations are rising for more and better services. clinics, two community dental clinics, and the central Govemment Dental Clinic. Dental services are also avail- Health expenditure able in the private sector, where 320 dental clinics run by 500 dental surgeons provide specialized as well as basic den- In the early 1960s total health expenditures were more than tal treatment. 4.5 percent of GDP In the mid-1960s the level of public As part of the government program for preventive den- spending was similar to private spending. Since then gov- tal care, the national drinking water supply has been fluo- ernment health spending has dropped from 2.5 percent to ridated since 1956. Dental health education is provided to 1.0 percent of GDP, while private spending has ranged from all patients presenting for dental treatment and to special 1.5 to 2.5 percent. Thus total health expenditure did not groups such as schoolchildren. The government regularly increase as fast as GDP during 1960-80. organizes dental health campaigns to raise awareness of In the early 1980s health spending as a share of GDP dental conditions and their prevention. The dental health was just 2.5 percent (table 1). This low level was related to education program is credited with lowering caries among rapid growth in GDP2 There was also steady expansion in children to one of the lowest rates in the world. private health spending during this period, as reflected in increasing numbers of doctors in private practice as well as Major Concerns and Future Challenges private hospital admissions (table 2). Today, there are more private doctors than government doctors. Although Singapore has come a long way in improving its As redevelopment and rationalization of the government health indicators and standards of medical service, many hospitals progressed, government spending remained high concerns and challenges remain. These include: through subsidies to the Ministry of Health, although there * Increasing costs of providing health care as a result of were periodic increases in user changes to reflect rising costs. advances in medical knowledge and technology, leading Until 1985 and before substantial cost recovery effortswere to increased specialization and subspecialization and introduced,Ministry of Health subsidies were equal to about greater use of medical technology. half of private health consumption. MEDICAL SAVINGS ACCOUNTS AND HEALTH CARE COVERAGE IN SINGAPORE 249 TABLE I Health care expenditures in Singapore, 1975-95 (millions of Singapore dollars) Private Public Total Share of Share of Share of Year Amount GDP (%) Amount GDP (%) Amount GDP (%) GDP 1975 249 1.86 138 1.03 387 2.89 13,373 1980 396 1.58 223 0.89 635 2.53 25,091 1985 702 1.80 419 1.08 1.198 3.08 38,924 1990 1,526 2.25 457 0.67 2,032 2.99 67,879 1995 2,368 1.96 685 0.57 3,380 2.80 120,629 Source: Ministry of Health, Singapore. Total health spending now accounts for about 3 percent These considerations form the basis for the National of Singapore's GDP1 During the 1980s public spending Health Plan formulated by the ministry in 1983. The plan's was less than 1.0 percent of GDP, while private spending key proposal, the Medisave scheme, imposes compulsory increased from 1.5 percent to more than 2.0 percent. Thus savings and restructures the system of health care financ- there has been a discernible shift toward private spending, ing. The main objectives of the plan are to secure a healthy reflecting price increases and cost recovery efforts in the and productive population through active promotion of public sector as well as the consumption preferences of a healthy lifestyles, and to improve cost efficiency in the use affluent population for a perceived higher quality of ser- of health services. In addition to promoting individual respon- vice. The government currently subsidizes about one-quarter sibility for maintaining good health, the plan aims to build of total health care costs. up financial resources in order to create the means to pay for medical care during illness (Ministry of Health 1983). The National Health Plan Restructuring of government hospitals Until recently most medical costs were incurred by gov- ernment hospitals, as well as the growing private hospital Moves to restructure health care financing were made to sector, and financed by personal payments, limited insur- avoid the problems of a welfare state system financed by ance coverage, and employment benefits that include com- taxes (such as the United Kingdom's National Health pany plans for workers and their families (Fong and Phua Service) and to shift the burden of financing health care to 1985). The Ministry of Health's goal is to provide quality individuals, families, and employers (including the private health care that is not only available and accessible, but and voluntary sectors). The strategy used was to increase that is also affordable and must be paid for. cost sharing by users and to progressively move the provi- TABLE 2 Supply of health facilities and doctors in Singapore, 1960-95 Admissions Admissions Number of Government Private to govemment to private Govemment Private Year hospitals hospital beds hospital beds hospitals hospitals doctors doctors 1960 14 6,537 650 na na 282 358 1965 16 6,817 859 120,274 na 450 469 1970 16 6,891 869 135,952 na 496 867 1975 22 8,005 1,100 164,205 na 855 847 1980 26 8,078 1,507 234,502 31,326 914 1,121 1985 22 8,329 1,671 229,988 47,164 1,214 1,307 1990 21 7,922 1,837 259,541 77,562 1,831 1,593 1995 24 8,326 2,211 266,142 91,413 2,124 2,191 Source: Singapore Department of Statistics, Yearbook of Statistics, various years. INNOVATIONS IN HEALTH CARE FINANCING 250 sion of health care to the private sector. This approach was annual subsidy from the government to offset their oper- initially carried under the bannerof the privatization move- ating deficits. Since the restructuring program began in ment, but during its implementation the term restructur- 1985, the government has restructured five (of seven) acute ing was preferred in view of the sensitivities involved. hospitals and two specialty institutions. Although the extent of privatization was never explicitly defined in the health sector, one goal was to transfer the Medisave management and control of major public hospitals away from the government (Phua 1991). In 1981 the Ministry of Health held discussions with med- The hospital restructuring program first proceeded with ical professionals in the public and private sectors and in the formation of a government-owned subsidiary, incorpo- the National University of Singapore to gather ideas on how rated under the umbrella of Temasek Holdings, to manage to further develop Singapore's health care system. Additional the newly completed National University Hospital from research and data collection were then conducted, includ- June 1985 onward. This arrangement was later changed ing a review of health care systems elsewhere. Members of when the hospital was placed under the control of a new Parliament were also consulted. In 1982 the minister of government-owned structure, the Health Corporation of health proposed the Medisave scheme and its underlying Singapore. This corporation was created in April 1987 to philosophy, which was followed by extensive publicity. For acquire and manage all restructured government hospitals, the next year discussions were held with community lead- beginningwith the National Skin Centre byDecember 1988, ers from citizens' consultative committees, representative Singapore General Hospital byApril 1989, Kandang Kerbau employers' federations, trade unions, and health-related Hospital and Toa Payoh Hospital by April 1990, Singapore associations. National Eye Centre by October 1990, and Tan Tock Seng A paper on the National Health Plan was released in Hospital by April 1991. The National University Hospital, 1983 and was widely publicized. Feedback from the pub- Singapore's main teaching hospital, was eventually put under lic was actively solicited, and there was follow-up by the the governance of the National University of Singapore. media on issues related to health care financing and deliv- Under the restructuring program public hospitals and ery. Parliament then debated the National Health Plan specialty institutes have been incorporated as private com- before approving the Medisave scheme, which compels panies wholly owned by the government. The current aim Singaporeans to set aside their own savings to meet future of the restructuring program is to give greater managerial hospitalization expenses, and recommending that there be autonomy to government hospitals so that they can pro- periodic reviews of the scheme's implementation to ensure vide more efficient and higher-quality service, improve pro- that appropriate adjustments were made. ductivity, control costs, and have greater flexibility to rapidly The Medisave scheme was given wide media coverage respond to changing needs. to explain how it would work. Printed materials were widely Arestructured hospital is fully autonomous and can recruit distributed and public talks were held in community cen- its own staff, set its own terms of remuneration, and decide ters, hospitals, and companies. The scheme was implemented on the deplbyment of resources. Compared with a govern- in governmnent hospitals on 1 April 1984. ment hospital, a restructured hospital has significantly greater Before the scheme was expanded to indude approved autonomy over its operations. Managers of restructured private hospitals in 1986, it was introduced as a pilot pro- hospitals are accountable to the board of directors for the ject involving the new National University Hospital (run hospital's performance. by an autonomous quasi-government company) in 1985. Restructuring is no longer regarded as a privatization The scale of charges and Medisave withdrawals here were exercise. The hospitals are still 100 percent owned by the equivalent to those in govemment hospitals. Withdrawal government and continue to pursue the social objectives rates, however, were subjected to a daily maximum for of the Ministry of Health (WHO 1994). Restructured hos- hospital charges and a maximum rate for each surgical pro- pitals continue to subsidize their patients and receive an cedure. The same limits now apply to the private hospitals. MEDICAL SAVINGS ACCOUNTS AND HEALTH CARE COVERAGE IN SINGAPORE 251 The scheme has been modified several times based on welfare of its members (including the sick and the elderly) a the experience acquired. Initially, account holders were only collective but essentially personal responsibility. The aim is allowed to use their Medisave accounts to pay for the full to preserve and enhance the stability of an essential social cost of hospital stays in lower-priced wards, and for part of structure amid rapid environmental changes. Only if a fam- the costs in more expensive rooms. This ceiling was later ily is unable to share in the medical expenses of its sick mem- extended to cover alnost all hospital charges, subject to bers does the state use public taxes to subsidize health care. maximum daily limits. In addition, Medisave accounts that The idea is to promote self-reliance, although a safety net is are used to pay for higher-priced accommodations are not still available for those in need (Phua 1986). allowed to be overdrawn. Another feature of Medisave is that, unlike tax-based Implementation of the Medisave scheme, though for- financing, it does not place an unduly heavy burden on the mulated and coordinated by the Ministry of Health, required employed and the young, and does not subject public expen- the active participation of many groups, including medical diture to the vagaries of economic cycles. The current gen- and related professionals, academics, politicians, commu- eration of workers is obliged to save for the future, instead nity and grassroots leaders, employers and employees, and of relying on the uncertain taxes of the next generation for the media. It took more than two years to thoroughly debate support. This is in line with official policy to promote finan- the issues involved, to disseminate vital information, and cial independence among the elderly, whose medical needs to gather feedback from all levels. This approach points to are expected to increase (Phua 1987). the importance of bottom-up planning and community par- ticipation in the wide acceptance and successful adminis- Disadvantages. Although Medisave is designed as com- tration of any innovative public program. pulsory savings for predictable hospitalization needs, espe- cially among the elderly, it seems to have encouraged Advantages. In essence, Medisave serves as an additional imirediate spending among younger groups for expensive source of personal financing for medical expenditures incurred hospital services that are perceived to be of higher quality. by families. This shift in public cost sharing frees govern- There has been a dramatic shift in demand from the gov- ment tax revenue for more urgent ptiorities and contributes ernrneht hospitals to the restructured and private hospi- to better public health services. It is hoped that Medisave, tals, and a discernible upgrading from the lower- to acting as a personal health financing and payment scheme, higher-priced beds. Since Medisave does not cover ambu- will control effective demand through the price mechanism. latory care (except for certain procedures like hepatitis B Under Medisave most payments for health care are made vaccination and kidney dialysis), people may have an incen- at the point of consumption. This close link between pay- tive to shift more care and spending to the hospital sector. ment and use better reflects the real costs of health care and Moreover, there is considerable ignorance on the part helps prevent excess use. Medisave also caters to different of the public with regard to the limits of Medisave cover- consumer preferences for a range of accommodations in age (for example, that only S$ 300 a day is allowed for pri- public and private hospitals. Thus, within certain limits, it vate hospital bills), creating the illusion of more money being can be used to provide complete coverage in less comfort- available than in reality. Many people also view (wrongly) able wards and to subsidize more expensive charges. Medisave funds as frozen assets that, if unspent, are retained Thus the Medisave scheme represents a major departure by the government. Thus many patients opt for more lux- from the social security schemes of other countries in sev- urious services than they would otherwise consume. In prac- eral areas. First, Medisave is not a common pool of funds to tice, Medisave has acted more like a supplementary financing be used indiscrimninately by a government facing pressures scheme to enable the consumption of upgraded hospital from interest groups to respond to short-term problems; services, as well as a mechanism for recovering costs in the instead, it is a scheme that covers only dependent family mem- public hospital system. bers. This fits in with the concept that the family is the basic A critical assumption of the Medisave scheme is that social and economic unit of any society, with caring for the patients are the best judge of how their savings should be INNOVATIONS IN HEALTII CARE FINANCING 252 spent on health care. However, effective health care deci- After Medisave was implemented, however, it became sionmaking requires considerable knowledge about prices apparent that there was still a need for coverage of major ill- and the quality and probable outcome of medical treat- nesses requiring expensive and prolonged treatment. To cater ments, as well as a high standard of ethics on the part of to such needs, a low-cost national catastrophic illness insur- practitioners. Since many of these preconditions are not in ance scheme, known as the Medishield scheme, was intro- place, there are grounds to intervene for the public good duced on 1 July 1990. To avoid the problems associated with and to protect consumers' interests. prepaid insurance, there is a system of deductibles and coin- Most of the scheme's current disadvantages can be cor- surance. All members of the Central Provident Fund (CPF) rected by better educating the public and introducing checks who are citizens or permanent residents are automatically and balances, including cost control limits and incentives covered unless they opt out of the scheme. Participation of aimed at both the supply and demand sides. Revenue caps CPF members' immediate dependents is voluntary. Non- have been imposed on the public hospitals, which also CPF members may opt into the scheme by contributing toward have to maintain at least two-thirds of their hospital beds Medisave. CPF members and their dependents are insured for subsidized patients. Besides encouraging the prudent up to the age of 70. To encourage people to participate in use of Medisave, financial counseling of patients and fam- the scheme, premiums are low and affordable, and vary by ily members prior to admission and other such measures age group to minimize cross-subsidy. Annual premiums range (including medical audits) are being instituted to prevent from S$ 12 for people below 30 to S$ 96 for people between abuse and to maintain standards. To discourage unneces- 60 and 65, to S$ 132 for people in the 65-70 age group. sary hospitalization, Medisave has been extended to cover Premiums can be paid from Medisave funds. more ambulatory care (such as day surgery and certain expen- Reimbursement is on actual expenses incurred under the sive outpatient services). Although the Medisave account plan (up to a limit) less the initial deductible of S$ 500-1,000, covers the acute hospitalization needs of the typical family, which is borne by the insured. The copayment is 20 per- it is insufficient to cover major catastrophic illnesses. Such cent, and there are annual and lifetime claim limits. Medisave illnesses require greater risk sharing and financing that insur- can be used to pay the deductibles and copayments. ance could better provide for. Deductibles are intentionally kept high to avoid excessive demand for medical services. (Deductibles were pegged at Medishield the level where only 10 percent of hospitalizations would be eligible for Medishield claims.) Since the 1950s the issue of health insurance has cropped up in various discussions about its many shortcomings and Medifund lack of applicability for Singapore (Ministry of Culture 1982). In the deliberations that led to the formulation of the Despite the widespread coverage of Medisave and National Health Plan and the Medisave scheme, the health Medishield, a small number of Singaporeans lack adequate insurance option was dismissed because of the negative savings or family support to pay for health care. This group experiences of other countries' health insurance systems. includes older cohorts of low-income individuals without Among the main weaknesses were: families who have insufficient or no provident fund accounts. * The illusion of a free service at the point of consump- The idea of setting up a large endowment fund to help fill tion, encouraging overuse and escalating costs. this financial gap was broached in 1991, and parliamentary * Lack of incentives for the consumer to stay well and for approval was granted in January 1992. According to the the provider to economize since they are reimbursed Medical Endowment Act, the government will deposit grants for any utilization. in special accounts for public hospitals to defray part or all Thus the Medisave scheme was conceived to avoid the pit- of the bills incurred by eligible patients. falls of third-party reimbursement systems financed from Medifund was given start-up capital of S$ 200 million insurance premiums or taxes (Ng 1988). and is supplemented by S$ 100 million a year from bud- MEDICAL SAVINGS ACCOUNTS AND HEALTH CARE COVERAGE IN SINGAPORE 253 getary surplus. A Medifund Advisory Council was formred trolling costs. Suggestions were made on staffing needs, to advise on the use of income derived from the endow- medical specialization and training, and professional stan- ment. Each hospital has appointed a Medifund committee dards and quality of care. The recommendations focused to approve payments based on the guidelines established on health care financing, defining the government's role in by the council. financing health care and managing health care costs Since April 1993 needy patients have been able to apply (Ministry of Health 1992). for partial or full waivers of their medical fees through The govemment accepted the recommendations of the medical social workers. Prior to this, waivers and subsidies review committee, and a Ministerial Committee on Health had to be absorbed by individual hospitals through gov- Policies has finalized the course of action to be implemented. ernment subventions from general tax revenues. Thus Aparliamentary paper on health care, presented in October Medifund further relieves the dependency on taxation as a 1993, set out the government's approach to controlling means of health care financing. health care costs, in order to keep basic health care afford- able to all Singaporeans. Among the cost-containment mea- Recent health policies sures: * Defining a good basic medical package Demographic, epidemiological, social, and economic trends - Regulating the supply of doctors and hospitals indicate that the tax-based health care financing system - Regulating subvented hospitals through revenue caps will be strained even further in the years to come. The and subsidies main objectives of health care financing policies are to * Encouraging greater cost sharing in medical insurance strengthen the safety net to protect people against rising and employment benefits health costs as well as to control costs. Medisave forms the * Controlling prices in the private sector first layer of this safety net, based on compulsory savings - Coordinating medical research and development and a family support system. Medishield provides the next * Coordinating medical education and training (Singapore layer of the safety net, covering major chronic conditions 1993). requiring long-term and high-cost health care. Medifund is Although many of these policies are being implemented, it an additional layer, financed through an endowment fund is too early to evaluate their full impact. created from surplus revenue. This multilayered financing system should be able to withstand the increasing burden Conclusion and Recommendations of health care costs as well as provide a greater degree of social security in the years ahead. Through a multipronged Can Singapore continue to ensure that its health financing approach using a public-private mix of health services, a system will develop along a balanced path to achieve equity, varying level of subsidy and cost sharing, and Medisave, efficiency, and cost-effectiveness in delivering health care? Medishield, and Medifund, it is hoped that a more con- Given past experience with the traditional tax-funded sys- trolled increase in health expenditure will be achieved in tem, as well as with the Medisave and the Medishield line with the rate of inflation and economic growth (Kwa schemes, what other modifications are necessary? Current 1989). interest seems to focus not on whether alternative meth- A Review Committee on National Health Policies was ods of health care financing can play a greater role, but appointed in 1991 to review future policy directions for whether new forms of fnancing could be successfully imple- the country. The first part of its report, presented in October mented to contain costs. 1991, emphasized health promotion and disease preven- Any viable financing instrument must be able to balance tion as the basic philosophy guiding Singapore's health supply and demand on a sustainable basis, minimize moral care policies (Ministry of Health 199 1). Its main report, pre- hazard and abuse, avoid labor market distortions, and resolve sented in February 1992, recommended priority areas for distributional effects across social groups (such as intergen- action and measures for improving health care while con- erational income transfers). Against these criteria the checks INNOVATIONS IN HEALTH CARE FINANCING 254 and balances built into Singapore's health care financing When resources are pooled among family members or com- system seem to provide adequate security. Given continued bined with insurance mechanisms for catastrophic cover- robust economic growth, health care spending as a share of age, compulsory savings are a potentially powerful way to GDP may appear deceptively low at present. As the econ- finance health care and old age security on a sustainable omy matures, pressures can be expected from employers to basis (allowing for lifetime risk-adjusted contributions) with- lower health care consumption given rising wage costs. out distorting economic growth. Among the key issues that will continue to be addressed in the health sector is the need to meet the ever-rising expec- References tations of an aging, affluent population for the latest med- ical technology. Can the push toward achieving excellence Fong, N. P, and K. H. Phua. 1985. "Utilisation and Expenditure in health care and developing Singapore as a regional hi- on Medical Services in a Local Community." Singapore Medical Journal 26(2): 131-38. tech medical center be balanced with more socially accept- Kwa, S. B. 1989. 'Financing the Health Care System in Singapore." able means of cost sharing, while dampening the inflationary Singapore Medical Association Newsletter 20(8): 1-2. effects of rapid growth? In the future cost-containment poli- Ministry of Culture. 1982. "Health-A Crucial Concern. Excerpts cies aimed at both supply and demand will have to be imple- from Ministerial Speeches on Health 1980/82." Information mented judiciously throughout the health care system. Division, Singapore. Although elements of Singapore's health financing sys- Ministry of Health. 1983. "Blue Paper on the National Health Atemhmanouh belentioSnarely transferabletohealth finantries,sPlan." Singapore. tem may not be entirely transferable to other countries, the . 1991. "Healthy Famnily, Healthy Nation." Report of the system points to general lessons for health care financing Review Committee on National Health Policies. Singapore. strategies. The concept of compulsory savings has many - . 1992. "Towards Better Health Care." Main Report of advantages over traditional financing methods, but it requires the Review Committee on National Health Policies. Singapore. several prerequisites to work. Among them: Ng, A. L. 1988. "The Medisave Scheme." In Proceedings of the * A high level of labor force participation in formal employ- Seminar on Economic Issues in Health Care. Singapore: National University of Singapore. ment at sustainable growth rates. Phua, Kai Hong. 1986. "Singapore's Family Savings Scheme." • An effective system of payroll collection and contribu- World Health (May): 11-12. World Health Organization, tion, as well as efficient fund management and payment. Geneva. * Awell-developedcomputerizedinformationsystemlink- . 1987. "Saving for Health." World Health Forum 8(1): ingpesonl edialsavng acouts it hopialsan 38-41. ing personal medical savings accounts with hospitals and _____. 1991. "Privatization and Restructuringof Health Services health care providers, in addition to security features and in Singapore." Occasional Paper 5. Institute of Policy Studies, accounting controls. Singapore. These requirements may not be possible in large coun- Seow, Adeline, and H. P Lee. 1994. "From Colony to City State: tries where there are wide disparities in labor force partic- Changes in Health Needs in Singapore from 1950 to 1990." ipation rates and where there would be difficulties in Journal ofPublic Health Medicine 16(2): 149-58. Singapore. 1993. "Affordable Health Care." White Paper pre- enforcing mandatory contributions from employers on a setdoPalmn,Cd.1.Sgpr. sented to Parllament, Cmd. 16. Singapore. regular basis. But savings schemes are more politically accept- WHO (World Health Organization). 1994. 'Health System able, especially if they are personalized accounts with appre- Reforms." Report of a symposium sponsored by the WHO ciating interest and are portable despite employment changes. Western Pacific Regional Office, Wellington, New Zealand. MEDICAL SAVINGS ACCOUNTS AND HEALTH CARE COVERAGE IN SINGAPORE 255 I Distributors of COLOMBIA HAITI ITALY NEWZEALAND ROMANIA SWITZERLAND Inloenlace Ltda. Culure Diffusion Licosa Commissionaria Sansoni SPA EBSCO NZ Ltd. Compani De Librarii Bucuresti S A. Libraine Payot Service Insltutionnel W orld Bank Carrera6No.51-21 5, Rue Capois Via Duca DiCalabda, 1/1 Private Mail Bag 99914 Sir. Lipscani no. 26, seclor3 C6tes-de-Montberion30 Apartado Aereo 34270 C.P 257 Caselba Postale 552 New Market Bucharest 1002 Lausanne Publications Santalf de Bopta, D.C. Poa-au-Prince 50125 Firenze Auckland Tel: (40 1)6139645 Tel: (41 21)341-3229 Prices and credit terms zaryfrom Tel: (57 1)285-2798 Tel: (509 1)3 9260 Tel: (55) 645-415 Tel: (64 9) 524-8119 Fax: (401)312 4000 Fax: !41 21) 341-3235 country to country. Consult your Fax: (571) 285-2796 Fax: (55) 641-257 Fax: (64 9) 524-8067 local distributor before placing an HONG KONG, MACAO E-mail: licosa@Itbcc.ft RUSSIAN FEDERATION ADECO Van Diermen EdisonsTechniques order COTE DIVOIRE Asia 2000 Ltd. UA. httpf/www.Ibc.itlcosa NIGERIA Isdatelstvo Ch. de Lacuez 41 Center dEdilion et de Diffusion Africaines Sales & Circulation Department University Press Limied Sa, Lolpachniy Pereulok CH1807 Bbnay ARGENTINA (CEDA) Seabird House, usa 1101-02 JAMAICA Three Crowns Building Jericho Moscow 101831 Tel: (41 21)9432673 o0icina del Libro lntemacional 04 B.P 541 22-28 Wyndham Street, Central Ian Randle Publishenrs Ltd. Private Mail Bag 5095 Tel: ?7 095) 917 87 49 Fax: (41 21) 943 3605 Av. Cordoba 1877 Abidjan 04 Hong Kong 206 Ol Hope Road Ibadan Fax: (7 095) 917 92 59 1120 Buenos Aires Tel: (225)246510;24 6511 Tel: (852) 2530-1409 Kingston 6 Tel: (234 22)41-1356 TANZANIA Tel: (541) 815-8354 Fax: (225)250567 Fax: (852) 2526-1107 Tel:909-927-2085 Fax: (234 22) 41-2056 SINGAPORE, TAIWAN, Oxford Uniuersuty Press Fax: (54 1) 815-8156 E-mail: saleseasIa2000.mom.hk Fax: 809-977-0243 MYANMAR,BRUNEI Maktaba Street CYPRUS URL: hrapJwww.asia2000.com.hk E-mail irpl@colis.com NORWAY Asahgate Publishing Asia Pacdic Pte. Ltd. PO Box 5299 AUSTRALIA, FIJI PAPUA NEW GUINEA Center for Applied Research NIC Info A/S 41 Kaliang Pudding Road C04-03 Dar es Salaam SOLOMON ISLANDS, VANUATU, AND Cyprus College INDIA JAPAN Book Department Goden Wheel Building Tel: (255 51) 29209 WESTERN SAMOA 6, Diogenes Street, Engomi Allied Publishers Ltd. Easlem Book SeNice Postboks 6512 Enterstad Singapore 349316 Fax: (255 51)46822 D.A. Inlom1albn Services PO. Box 2006 751 Mount Read 3-13 Hongo 3-chome, Bunkyo-ku N-0606 Oslo Tel (65) 741-5166 0.48 Whiehonrse Rvie Nicosia Madras - 600 002 Tokyo 113 Tel: (47 22) 974500 Fax. (65) 742-9356 THAILAND Mcrhami3132 Tel: (357 2) 44-1730 Tel: (91 44) 852-3938 Tel: (81 3) 3818-0061 Fax: (47 22) 974545 E-mail: ashgate8asianonned.com Central Books Distribution Vidoria Fax: (357 2) 46-2051 Fax: (91 44) 852-0649 Fax: (81 3) 318-0864 306 Silom Road Tel (61(39210 7777 E-mail: orderssvo-ebos.co.jp PAKISTAN SLOVENIA Bangkok 10500 Fax: (68)392107799 CZECH REPUBLIC INDONESIA URL: hblpeWwwe.bekkoame.orjp/-svt-ebs Mirza Book Agency Gospodarski Vestnik Publishing Goup Tel (66 2)235-5400 E-mail: service lidadirectoomra National Inlormalion Center Pt. Indira LimRid 65, Shahrah-e-Quaid-e-Azam Dunajska cesta s Fax: (662)237-0321 URL: hop://www.dadirect.omrn.au prodejna, Konviktska 5 Jalan Borobudur 20 KENYA Lahore 54000 1000 Ljubljana CS- 11357 Prague 1 PO. Box 181 Africa Book Service (E.A.) Ltd. Tel: (92 42) 735 3601 Tel: (38601)133 0347.1321230 TRINIDAD& TOBAGO, AUSTRIA Tel: (422)2422-9433 Jakarla 10320 Quaran House, Mfangano Street Fax: (92 42) 756 5283 Fax (39661)1338030 AND THE CARRIBBEAN Gerold and Co. Fax: (42 2) 2422-1484 Tel: (62 21) 390-4290 PO. Boa 45245 E-mail: belicd@gvestnik.si Systematics Studies Unh Webthburggasse 26 URL: htpI1atwww.nisucl Fax: (62 21) 4214289 Nairobi Odord University Press 9 Waft Street A-1all Wien Tel: (254 2) 223 641 5 Bangalore Town SOUTH AFRICA, BOTSWANA Curepe Tel: (43 1) 512-47431-0 DENMARK IRAN Fax: (254 2) 330 272 Sharae Faisal For single titles: Trinidad. West Indies Fa:x (431) 51247-31-29 SamfundsLtteratur Ketab Sara Co. Publishers PO Box 13033 Odord Univershy Press Southem Atrica Tel. (809) 662-5654 URL: ht1pJvww .gqrold.co/at.onIine Rosenoems AlIe 11 Khaled Esiamboli Ave., KOREA, REPUBLIC OF Karachi-75350 PO. Box 1141 Fax: (609) 662-5654 DK-1970 Frederiksberg C 6thStreet Daejon Trading Co. Ud. Tel: (92 21) 446307 Cape Town 000 E-mail tobe@tnidad net BANGLADESH Tel: (45 31) 351942 Kusheh Delafrooz No. 8 PO. Box 34, Youida Fax. (92 21) 4547640 Tel. (27 21) 45-7266 Micro Industfies Development Fax: (45 31) 357822 PO. Box 15745-733 706 Seoun Bldg E-mail: oup@oup.khi.erum.com.pk Fax: (27 21) 45-7265 UGANDA A.ssistnce Sci;ety (MIDAS) Tehran 44-4 Yoouido-Dng, Yeongchengpo-Ku Gustro Ltd Houssistae S,Rociety 16 IDA EGYPT, ARAB REPUBLIC OF Tel: (98 21) 8717819; 8716104 Seoul Pak Book Corporation For subscription orders: PO Box 9997, Madhvani Building Hnuse 5, Road 16 Al Ahranm Distibution Agency Fax: (98 21) 8712479 Tel: (822)785-1631/4 Aziz Chambers 21 Intemational Subscription Service Plot 16/4 Jinja Rd. Danmon rea Al Galaa Street E-mail: ketab-sara@neda.net.ir Fax: (822)784-0315 Queens Road PO. Box 41095 Kampala TelDa8k2a 329427 Cairo Lahore Craighall Tel: (256 41) 254 763 Fax: (880 2) 811188 Tel:(202)5784083 KowkabPublishers MALAYSIA Tel: (92 42) 636 3222; 636 0885 Johannesburg 2024 Fax: (25641)251468 Fax: (20 2) 5784833 PO. Box 19575-511 UniversiftyofMalaya Cooperative Fax: (92 42) 636 2328 Tel: (27 11) 880-1440 BELGIUM Tehran Bookshop, Limated E-mail: pbc@brain.net.pk Fax: (2711) 880-6248 UNITED KINGDOM Jean De Lannoy The Middle East Observer Tel: (98 21) 258-3723 PO. Box 1127 E-mail: iss@is.co.za Microinfo Ltd. Av. du Roi 202 41, Sherr4 Street Fax: (98 21) 258-3723 Jatan Pantai Baru PERU PO. Box 3 1000 Brassels Cairo 59700 Kuala Lumpur Ednorial Desarrollo SA SPAIN Afon, Hampshire GU34 2PG Tel: (322) 538-5169 Tel: (202)393-9732 IRELAND Tel: (603) 756-5000 Apartado 3824 Mundi-Prensa Libros, S.A. England Fax: (32 2) 538-0841 Fax: (20 2) 393-9732 Govemment Supplies Agency Fax: (60 3) 7554424 Lima 1 Castello 37 Tel: (44 1420) (8486 Olig an tSolkIhair Tel: (51 14) 295380 28001 Madrd Fax. (44 1420) 89889 BRAZIL FINLAND 4-5 Harcourt Road MEXICO Fax: (51 14) 296628 Tel: (34 1) 431-3399 E-mail: wbank@ukminio.demon co.uk Publicac0es Tocnica Intemacionais Ltda aleeminen Kiriakauppa Dublin 2 INFOTEC Fax: (341) 575-3998 URL: http./www .microinso co uk RuaPeixotoGomida, 209 PO. Bax 128 Tel: (353 1)661-3111 Av. San Femando No. 37 PHILIPPINES E-mail: libreria@mundiprensa.es 01409 San Paulo, SR FIN-00101 Helsinki Fax: (3531) 475-2670 Col. Toiello Guerra Intemational Booksource Cenler Inc. URL: http:lwww .mundiprensa es/ VENEZUELA Teol:55 11) 259-6644 Tel: (358 0) 121 4418 14050 Mexico, D.F 1127-A Antipolo St. Tecni-Ciencia Libros, S.A. Fax: (5511) 258-6990 Fax: (358 0) 121443S ISRAEL Tel: (52 5) 624-2800 Barangay, Venezuela Mundi-Prensa Barcelona Centro Cuidad Comercial Tamanco E-rmail: postmaster(ptimol.ir E-mail: akatilaus8stockmann.fi Yozmot LRerature Ltd. Fax: (525)624-2822 Makati City Consel de Cent, 391 Nivel C2 URLE htipr//www .uol.br URL: hftp:#www.akateeminen.com/ PO. Box 56055 E-mail: infotec@4n.net.mx Tel: (63 2) 8966501:6505; 6507 08009 Barcelona Caracas URL:*hItp://wwm ual.br 3 YobnanHasandlar Street URL: ht1p://an.neo.mx Fax: (632)896 1741 Tel: (343)488-3492 Tel (582)9595547:5035;0016 CANADA FRANCE Tel Avivw61560 Fax (34 3) 487-7659 Fax: (58 2)F959 5636 Renouf Publishing Co. Ltd. Word Bank Publicatibos Tel: (9723)5285-397 NEPAL POLAND E-mail: barcelona@mundiprensa.es 5369 Canotek Road 66, avenue d1alna Fax: (972 3) 5285-397 Everest Media Inlematlonal Services (P)Ltd. Intemational Publishing Service ZAMBIA Ottawa, Ontario K1J 9,3 75116 Pars GPO Box 5443 Ul. Piekna 31137 University Bookshop, Unversty of Zambia Tel (613) 745-2665 Tel: (33 1) 40-69-30-56/57 R.O.Y Intemational Kathimandu 00-677 Warzawa SRI LANKA, THE MALDIVES Great East Road Campus Fax: (613) 745-7660 Fax: (33 1)40-49-3086 PO Box 13056 1el: (977 1)472 152 Tel:(482)628-089 Lake House Bookshop PO. Box 32379 E-mail: order.deptO renoufbooks.com Tel Aviv61130 Fax: (9771) 224 431 Fax: (48 2) 621-7255 100, Sir Chr tampalam Gardiner Mawatha Lusaka URL: hoprDe .dwwrenoudbooks.com GERMANY Tel: (972 3) 5461423 E-mail: books%ips@ikp.atm.com pl Colombo 2 Tel: (2601) 252 576 UNO-Verlag Fax: (9723)5461442 NETHERLANDS URL: htip:/Iwww.ipscg.wawpVips/export/ Tel: (94 1)32105 Fax. (260 1)253 952 CHINA Poppelsdoder Allee 00 E-mail: royil@netvision.net.il De LindeboorrilnOr.Publikates Fax (94 1)432104 China Financial & Economic 53115 Bonn PO. Box 202 PORTUGAL E-mail: LHL@si.ianka.net ZIMBABWE Publishing House Tel: (49 228) 212940 Palestinian Authorily/Middlo Cool 7400 AE Haaksbergen Livraita Portugal Longmnan Zimbabwe )Pvl.(Ltd. 8, Da Fo Si Dong Jie Fax: (49 228) 217492 Index Infornation Services Tel: (31 53) 5744044 Apartado 2081 SWEDEN Toude Road, Ardbennie nBaio nPO.B.19502 Jerusalem Faa: (31 53) 572-9296 Rua Do Camno 70-74 Wennergren-Williams AB PO. BoxoST125 Tel:(810) 63-257 GREECE Tel: (9722)6271219 E-mail: lindeboodlwoddonline.nl 1200 Lsbon PO Box 1305 Southenon Faa: (8610) 6401-7385 Papasonoiou S.A. Fax: (972 2) 6271634 URL: hup:/Awww .woddonline.nl-lindeboo Tel. (1) 347-4982 S-171 25 Solna Harare 35, Stoumara SIr. Fax: (1) 347-0264 Tel: (46 8) 705-97-50 Tel: (263 4) 6216617 100 82 Athens Fax: (46 0) 27-00-71 Fax. (263 4) 621670 Tel: (30 1) 364-1826 E-mail. mail@wwi.se %2ur9 Fax: (30 1) 364-8254 Recent World Bank Discussion Papers (continued) No. 331 Case Studies in War-to-Peace Transition: The Demobilization and Reintegrationi of Ex-Combatants in Ethiopia, Namiibia, and Uganda. Nat J. Colletta, Markus Kostner, Ingo Wiederhofer, with the assistance of Emilio Mondo, Tairni Sitari, and Tadesse A. Woldu No. 333 Participation in Practice: The Experience of the World Bank and Other Stakeholders. Edited by Jennifer Rietbergen- McCracken No. 334 Managing Price Risk in the Pakistan Wheat Market. Rashid Faruqee and Jonathan R. Coleman No. 335 Policy Optionsfor Reform. of Chinese State-Owned Enterprises. Edited by Harry G. Broadman No. 336 Targeted Credit Programs and Rural Poverty in Bangladesh. Shahidur Khandker and Osman H. Chowdhury No. 337 The Role of Family Planning and Targeted Credit Programs in Demographic Change in Bangladesh. Shahidur R. Khandker and M. Abdul Latif No. 338 Cost Sharing in the Social Sectors of Suzb-Saharan Africa: Impact on the Poor. Arvil Van Adams and Teresa Hartnett No. 339 Putblic and Private Roles in Health: Theory and Financing Patterns. Philip Musgrove No. 340 Developing the Nonfarm Sector in Bangladesh: Lessonsfrom Other Asian Couintries. Shahid Yusuf and Praveen Kumar No. 341 Beyond Privatization: The Second Wave of Telecommutnications Reforms in Mexico. Bjbrn Wellenius and Gregory Staple No. 342 Economic Integration and Trade Liberalization in Southern Africa: Is There a Rolefor South Africa? Merle Holden No. 343 Financing Private Infrastruicture in Developing Couintries. David Ferreira and Karman Khatami No. 344 Transport and the Village: Findings from African Village-Level Travel and Transport Sutrveys and Related Studies. Ian Barwell No. 345 On the Road to EU Accession: Financial Sector Development in Central Eutrope. Michael S. Borish, Wei Ding, and Michel Noel No. 346 Structural Aspects of Manufacturing in Sutb-Saharan Africa: Findingsfrom a Seven Country Enterprise Survey. Tyler Biggs and Pradeep Srivastava No. 347 Health Reform in Africa: Lessonsfrom Sierra Leone. Bruce Siegel, David Peters, and Sheku Kamara No. 348 Did External Barriers Cautse the Marginalization of Sub-Saharan Africa in World Trade? Azita Amjadi Ulrich Reincke, and Alexander J. Yeats No. 349 Sutrveillance of Agricultural Price and Trade Policy in Latin America during Major Policy Reforms. Alberto Valdes No. 350 Who Benefitsfrom Public Education Spending in Malawi: Resultsfrom the Recent Education Reform. Florencia Castro-Leal No. 351 From Universal Food Suibsidies to a Self-Targeted Program: A Case Study in Thinisian Reform. Laura Tuck and Kathy Lindert No. 352 China's Urban Transport Developmenit Strategy: Proceedings of a Symposiuzmn in Beijinig, Noveember 8-20, 1995. Edited by Stephen Stares and Liu Zhi No. 353 Telecommuinications Policiesfor Sub-Saharan Africa. Mohammad A. Mustafa, Bruce Laidlaw, and Mark Brand No. 354 Saving across the World: Puizzles and Policies. Klaus Schmidt-Hebbel and Luis Serven No. 355 Agricuilture and German Reuinification. Ulrich E. Koester and Karen M. Brooks No. 356 Evaluiating Health Projects: Lessonsfrom the Literature. Susan Stout, Alison Evans, Janet Nassim, and Laura Raney, with substantial contributions from Rudolpho Bulatao, Varun Gauri, and Timothy Johnston No. 357 Innovations and Risk Taking: The Engine of Reform in Local Government in Latin America and the Caribbean. Tim Campbell No. 358 China's Non-Bank Financial Instituitions:Trust and Investmenf Companies. Anjali Kumar, Nicholas Lardy, William Albrecht, Terry Chuppe, Susan Selwyn, Paula Perttunen, and Tao Zhang No. 359 The Demandfor Oil Produtcts in Developing Countries. Dermot Gately and Shane S. Streifel No. 360 Preventing Banking Sector Distress and Crises in Latin America: Proceedings of a Conference hield in Washington, D.C., April 15-16, 1996. Edited by Suman K. Bery and Valeriano F. Garcia No. 361 China: Power Sector Regutlation in a Socialist Market Economy. Edited by Shao Shiwei, Lu Zhengyong, Norreddine Berrah, Bernard Tenenbaum, and Zhao Jianping No. 362 The Regtulation of Non-Bank Financial Institittions: The United States, the Eutropean Union, and Other Coluntries. Edited by Anjali Kumar with contributions by Terry Chuppe and Paula Perttunen No. 363 Fostering Suistainable Development: The Sector Investment Program. Nwanze Okidegbe No. 364 Intensified Systems of Farming in the Tropics and Sutbtropics. J.A. Nicholas Wallis THE WORLD BANK 1 IS I I SIT-cct. .\\. \\.likliYtlEml I ).(. 2(1433 .l S \ ILIcph4)flc: 2012-477-1234 IFc'simic: 202-477-o3,)I ICl\: NCI (04145 \\(I(I.I)I, NIs \I( I 248423n \\MlRl.11)111\Nk \\\SIlI (IOI( \,1 \i.l. \\.1h: III t p: ENN X\ IT-,,,II) altIII-,l.I-8 3 I '-,l:1:11'al ml,l-, \ d(,, ll15:,i 1-/ ,' ISBN 0-8213-3964-8