H N P D I S C U S S I O N P A P E R Role of Communities in Resource Mobilization and Risk Sharing A Synthesis Report Alexander S. Preker, Guy Carrin, David M. Dror, Melitta Jakab, William Hsiao and Dyna Arhin September 2001 ROLE OF COMMUNITIES IN RESOURCE MOBILIZATION AND RISK SHARING A Synthesis Report Alexander S. Preker, Guy Carrrin, David M. Dror, Melitta Jakab, William Hsiao and Dyna Arhin September 2001 Health, Nutrition, and Population Discussion Paper This series is produced by the Health, Nutrition, and Population Family (HNP) of the World Bank's Human Development Network (HNP Discussion Paper). The series provides a vehicle for publishing preliminary and unpolished results on HNP topics to encourage discussion and debate. The findings, interpretations, and conclusions expressed in this paper are entirely those of the authors and should not be attributed in any manner to the World Bank, to its affiliated organizations, or to members of the Bank's Board of Executive Directors or the countries they represent. Citation and the use of material presented in this series should take into account this provisional character. For free copies of papers in this series, please contact the individual authors whose name appears on the paper. Inquiries about the series and submissions should be directed to the Editor in Chief. Submissions should have been previously reviewed and cleared by the sponsoring department that will bear the cost of publication. No additional reviews will be undertaken after submission. The sponsoring department and authors bear full responsibility for the quality of the technical contents and presentation of material in the series. Since the material will be published as presented, authors should submit an electronic copy in a predefined format as well as three camera-ready hard copies (copied front to back exactly as the author would like the final publication to appear). Rough drafts that do not meet minimum presentational standards may be returned to authors for more work before acceptance. The Editor in Chief of the series is Alexander S. Preker (apreker@worldbank.org); For information regarding this and other World Bank publications, please contact the HNP Advisory Services (healthpop@worldbank.org) at: Tel (202) 473-2256; and Fax (202) 522-3234. ISBN 1-932126-10-4 © 2001 The International Bank for Reconstruction and Development /The World Bank 1818 H Street, NW Washington, DC 20433 All rights reserved. ii Health, Nutrition, and Population Discussion Paper Role of Communities in Resource Mobilization and Risk Sharing A Synthesis Report Alexander S. Preker,a Guy Carrin,b David Dror,c Melitta Jakab,d William Hsiao,e and Dyna Arhin Tenkorangf aChief Economist for Health Nutrition and Population, World Bank, Washington DC bSenior Economist, Evidence Based Policy and Information, World Health Organization, Geneva cHealth Specialist, Social Security Department, International Labour Organisation , Geneva dPhD Candidate, School of Public Health, Harvard University, Boston, Massachusetts eProfessor, School of Public Health, Harvard University, Boston, Massachusetts fResearcher, Center for International Development, Harvard University, Boston, Massachusetts Report Submitted to Working Group 3 of the Commission on Macroeconomics and Health, Jeffrey D. Sachs (Chairman), September 2001 Abstract: Most community finance schemes have evolved in the context of severe economic constraints, political instability, and lack of good governance. Usually government taxation capacity is weak, formal mechanisms of social protection for vulnerable populations absent, and government oversight of the informal health sector lacking. In this context of extreme public sector failure, community involvement in financing health care provides a critical though insufficient first step in the long march toward improved access to health care by the poor and social protection against the cost of illness. It should be regarded as a complement to--not as a substitute for--strong government involvement in health care financing and risk management related to the cost of illness. Based on an extensive survey of the literature, the main strengths of community financing schemes are the extent of outreach penetration achieved through community participation, their contribution to financial protection against illness, and increase in access to health care by low-income rural and informal sector workers. Their main weaknesses are the low volume of revenues that can be mobilized from poor communities, the frequent exclusion of the very poorest from participation in such schemes without some form of subsidy, the small size of the risk pool, the limited management capacity that exists in rural and low-income contexts, and their isolation from the more comprehensive benefits that are often available through more formal health financing mechanisms and provider networks. The authors conclude by proposing concrete public policy measures that governments can introduce to strengthen and improve the effectiveness of community involvement in health care financing. This includes: (a) increased and well-targeted subsidies to pay for the premiums of low-income populations; (b) use of insurance to protect against expenditure fluctuations and use of reinsurance to enlarge the effective size of small risk pools; (c) use of effective prevention and case management techniques to limit expenditure fluctuations; (d) technical support to strengthen the management capacity of local schemes; and (e) establishment and strengthening of links with the formal financing and provider networks. Keywords: Community financing; informal sector; financial protection; health care financing; social exclusion. Disclaimer: The findings, interpretations, and conclusions expressed in the paper are entirely those of the authors, and do not necessarily represent the views of the World Bank, its Executive Directors, or the countries they represent. Correspondence Details: Alexander S. Preker, The World Bank, 1818 H Street, Mail Stop G7-702, NW, Washington DC, 20433, USA; Tel: (1 202) 473-2327; Fax: (1 202) 522-3234; Email: apreker@worldbank.org; Web: www.worldbank.org/hsd iii iv Contents PREFACE.............................................................................................................................................. VII ACKNOWLEDGMENTS ......................................................................................................................IX I. OVERVIEW AND CONTEXT............................................................................................................ 1 A. THE EXCLUSION OF LOW-INCOME RURAL POPULATIONS AND INFORMAL WORKERS...................... 1 B. ORIGINS OF RICH-POOR DIFFERENCES IN FINANCIAL PROTECTION .................................................. 2 C. ROLE OF COMMUNITIES IN PROVIDING FINANCIAL PROTECTION...................................................... 4 II. CONCEPTUAL UNDERPINNINGS FOR COMMUNITY-BASED ACTION IN HEALTH CARE FINANCING ................................................................................................................................. 5 A. LINKS TO EXISTING MICROFINANCE ORGANIZATIONS...................................................................... 5 B. LINKS TO COMMUNITY-LEVEL SOCIAL CAPITAL............................................................................... 7 C. LINKS TO MAINSTREAM PUBLIC ECONOMICS............................................................................. 7 III. METHODOLOGY FOR ASSESSING IMPACT, STRENGTHS, AND WEAKNESSES.......... 8 A. METHODOLOGY FOR SURVEY OF LITERATURE ON COMMUNITY FINANCING ................................. 10 B. METHODOLOGY FOR REGIONAL REVIEWS OF SELECTED ASIA AND AFRICA EXPERIENCE............. 13 C. METHODOLOGY FOR MICRO-LEVEL HOUSEHOLD SURVEY ANALYSIS ........................................... 13 D. METHODOLOGY FOR MACRO-LEVEL CROSS-COUNTRY ANALYSIS ................................................ 16 III. DISCUSSION OF MAIN FINDINGS FROM BACKGROUND REVIEWS ............................. 17 A. DISCUSSION OF SURVEY OF EXISTING LITERATURE ON COMMUNITY HEALTH FINANCING ........... 17 Assessment of Impact. ....................................................................................................................... 18 Identification of Determinants .......................................................................................................... 20 B. DISCUSSION OF MAIN FINDINGS FROM ASIA REGIONAL REVIEW.................................................... 22 C. DISCUSSION OF MAIN FINDINGS FROM AFRICA REGIONAL REVIEW ............................................... 24 D. DISCUSSION OF MAIN FINDINGS FROM MICRO-LEVEL HOUSEHOLD SURVEY ANALYSIS............... 26 Determinants of Social Inclusion in Community Financing............................................................. 26 Determinants of Financial Protection in Community Financing ..................................................... 27 E. DISCUSSION OF MAIN FINDINGS FROM MACRO-LEVEL CROSS-COUNTRY ANALYSIS..................... 29 IV. CONCLUSIONS AND RECOMMENDATIONS.......................................................................... 31 REFERENCES........................................................................................................................................ 35 APPENDIX A.......................................................................................................................................... 41 v vi PREFACE In January 2000, Dr. Gro Harlem Bruntland, Director General of the World Health Organization (WHO), established a Commission on Macroeconomics and Health (CMH) to provide evidence about the importance of health to economic development and poverty alleviation. This HNP Discussion Paper is based on a Report on community financing submitted in September 2001 to Working Group 3 of the CMH. The mandate of Working Group 3 was to examine alternative approaches to domestic resources mobilization, risk protection against the cost of illness, and resource allocation. The working group was chaired by Professor Alan Tait (Former Deputy Director of Fiscal Affairs, International Monetary Fund, and currently Honorary Fellow at University of Kent at Canterbury and Honorary Fellow at Trinity College, Dublin) and Professor Kwesi Botchewey (Director of Africa Research and Programs at the Harvard Center for International Development). Professor Jeffery D. Sachs (Chairman of the Commission and Director of the Harvard Center for International Development) presented the findings of the CMH in a Report that was submitted to WHO on December 20, 2001--Macroeconomics and Health: Investing in Health for Economic Development. The Report of the CMH recommended a six pronged approach to domestic resource mobilization at low- income levels: "(a) increased mobilization of general tax revenues for health, on the order of 1 percent of GNP by 2007 and 2 percent of GNP by 2015; (b) increased donor support to finance the provision of public goods and to ensure access for the poor to essential health services; (c) conversion of current out- of-pocket expenditure into prepayment schemes, including community financing programs supported by public funding, where feasible; (d) a deepening of the HIPC (Highly Indebted Poor Countries) initiative, in country coverage and in the extent of debt relief (with support form the bilateral donor community); (e) effort to address existing inefficiencies in the way in which government resources are presently allocated and used in the health sector; and (f) reallocating public outlays more generally from unproductive expenditure and subsidies to social-sector programs focused on the poor." Most community financing schemes have evolved in the context of severe economic constraints, political instability, and lack of good governance. Usually government taxation capacity is weak, formal mechanisms of social protection for vulnerable populations absent, and government oversight of the informal health sector lacking. In this context of extreme public sector failure, community involvement in the financing of health care provides a critical, though insufficient, first step in the long march toward improved access to health care by the poor and social protection against the cost of illness. The CMH stressed that community financing schemes are no panacea for the problems that low-income countries face in resource mobilization. They should be regarded as a complement to--not as a substitute for--strong government involvement in health care financing and risk management related to the cost of illness. Based on an extensive survey of the literature, the main strengths of community financing schemes are the degree of outreach penetration achieved through community participation, their contribution to financial protection against illness, and increase in access to health care by low-income rural and informal sector workers. Their main weaknesses are the low volume of revenues that can be mobilized from poor communities, the frequent exclusion of the very poorest from participation in such schemes without some form of subsidy, the small size of the risk pool, the limited management capacity that exists in rural and low-income contexts, and their isolation from the more comprehensive benefits that are often available through more formal health financing mechanisms and provider networks. vii The work by the CMH proposed concrete public policy measures that governments can introduce to strengthen and improve the effectiveness of community involvement in health care financing. This includes: (a) increased and well targeted subsidies to pay for the premiums of low-income populations; (b) use of insurance to protect against expenditure fluctuations and use of reinsurance to enlarge the effective size of small risk pools; (c) use of effective prevention and case management techniques to limit expenditure fluctuations; (d) technical support to strengthen the management capacity of local schemes; and (e) establishment and strengthening of links with the formal financing and provider networks. The report presented in this HNP Discussion Paper has made a valuable contribution to our understanding of some of the strengths, weaknesses, and policy options for securing better access for the poor to health care and financial protection against the impoverishing effects of illness, especially for rural and informal sector workers in low-income countries. Alexander S. Preker Chief Economist Health, Nutrition, and Population World Bank viii ACKNOWLEDGMENTS Working Group 3 of the Commission on Macroeconomics and Health is particularly grateful to the following individuals for their direct contribution to the report: (a) preparation of the synthesis report by Alexander S. Preker, Dr. Guy Carrin, David M. Dror, and Melitta Jakab; (b) survey of the literature by Melitta Jakab and Chitra Krishnan; (c) analysis of macroeconomic-level data by Guy Carrin, Riadh Zeramdini, Philip A. Musgrove, Jean-Pierre Poullier, Nicole Valentine, and Ke Xu; (d) analysis of microeconomic-level data by Melitta Jakab, Alexander Preker, Chitra Krishnan, Allison Gamble Kelly, Pia Schneider, François Diop, and A.K. Nandakumar; Dr. Johannes Jutting, Dr. Anil Gumber, Kent Ranson; and Dr. Siripen Supakankunti; (e) review of selected Asia and Africa experiences by Bill Hsiao and Dyna Arhin-Tenkorang; (f) review of reinsurance of community schemes by David Dror. Valuable guidance on methodological issues was provided by Adam Wagstaff. We are also indebted to the following individuals for data access, guidance on research methodologies, reviews, and other indirect contributions to the report: Christian Jacquier, Christian Baeaza, Michael Cichon, Chris Murray, Kei Kawabatak, Christopher Lovelace, Helen Saxenian, Jack Langenbrunner, David Peters, George Schieber, Charlie Griffin, Agnes Soucat, Abdo S. Yazbeck, Mariam Claeson, Flavia Bustreo, Steve Cummings, and Shanta Devarajan. The authors of the report are also grateful for the access provided to parallel and ongoing research on community financing by the Bank, World Health Organization, and International Labour Organisation with important inputs from Harvard University, London School of Hygiene and Tropical Medicine, University of Lyon, Partnerships for Health Reform--Abt Associates Inc.(USA), National Council For Economic Research (India), Center for Development Research (ZEF) (Germany), and Chulalongkorn University Faculty of Economics (Thailand). The authors are grateful to WHO for having provided an opportunity to contribute to the work of the Commission on Macroeconomics and Health and to the World Bank for having published the Report as an HNP Discussion Paper. ix x I. OVERVIEW AND CONTEXT This century has witnessed greater gains in health outcomes than any other time in history. These gains are partly the result of improvements in income with accompanying improvements in health-enhancing social policies (housing, clean water, sanitation systems, and nutrition) and greater gender equality in education. They result also from new knowledge about the causes, prevention, and treatment of disease and from the introduction of policies, financing, and health services that make such interventions more equitably accessible. Improving ways to finance health care and protect populations against the cost of illness has been central to this success story (see Preker, Langenbrunner, and Jakab 2002; and Preker et al. 2002a and 2002b). The share of the world's population protected against the catastrophic cost of illness rose significantly during the twentieth century, with global spending on health increasing from 3 percent to 8 percent of global GDP (US$2.8 trillion) or 4 percent of the GDP of developing countries (US$250 billion). At the current global growth rate for GDP of 3.5 percent, spending on health-enhancing activities will increase annually by about US$98 billion a year worldwide, or US$8 billion a year in low- and middle-income countries. A. THE EXCLUSION OF LOW-INCOME RURAL POPULATIONS AND INFORMAL WORKERS Today, the population in most industrial countries (except Mexico, Turkey, and the United States) enjoy universal access to a comprehensive range of health services that are financed through a combination of general tax revenues, social insurance, private insurance, and charges (Preker 1998). A number of low-income countries (such as Sri Lanka, Malaysia, Zambia, and Costa Rica) have tried to follow a similar path, but the quest for financial protection against the cost of illness in middle- and low- income countries has been a bumpy ride. Many of the world's 1.3 billion poor still do not have access to effective and affordable drugs, surgeries, and other interventions because of weaknesses in the financing and delivery of health care (ILO 2000a; WHO 2000; World Bank 1993; 1997). See Figure 1. Less Pooling of Revenues in Low-Income Countries Share of world's 1.3 billion living on less than US$1 day indicated by size of blue bubbles Although 84 percent of the world's poor shoulder 93 percent of the global burden of disease, only 11 percent of the US$2.8 trillion spent on health care reaches the low- and middle-income countries. Vaccination strategies of modern health care systems have reached millions of poor. However, when ill, low-income households in rural areas continue to use home remedies, traditional healers, and local 1 providers who are often outside the formal health system. The share of the population covered by risk- sharing arrangements is lower at low-income levels (see Figure 1). As a result, the rich and urban middle classes often have better access to the modern health care advances of the twenty-first century. B. ORIGINS OF RICH-POOR DIFFERENCES IN FINANCIAL PROTECTION The flow of funds through the health care system, and the public/private mix, is complex (see Figure 2-- modified from Schieber and Maeda 1997). It can be differentiated into three discrete functions: (a) collection of revenues (source of funds); (b) pooling of funds and spreading of risks across larger population groups; and (c) purchase of services from public and private providers of health services (allocation or use of funds) (see also WHO 2000). A combination of general taxation, social insurance, private health insurance, and limited out-of-pocket user charges has become the preferred health financing instruments for middle- and higher income countries, where income is readily identifiable and taxes or premiums can be collected at the source. Flow of Funds through the System Revenue Pooling Resource Allocation Service Collection or Purchasing Provision Taxes Government Agency ic Public Charges Social Insurance or Public Publ Sickness Funds Providers Mandates Grants Private Insurance Organizations Private Providers Loans etavi Employers Private Insurance Pr Communities Individuals And Households Out-of-Pocket Different issues arise in the case of the public and private engagement in health care financing and service delivery. The need for collective arrangements and strong government action in health care financing is often confused with public production of services. The poor and other excluded populations often seek care from private providers because public services in rural and low-income urban areas are often scarce or plagued by understaffing, supply shortages, and low-quality care. Poor households and community financing schemes therefore often turn to private providers for the care they need. Such engagement by private providers can still be pro-poor if there are mechanisms to exempt the poor or subsidize user fees (Preker, Harding, and Girishankar 2001) and if purchasing arrangement include coverage for the poor (Preker et al. 2001). Several factors make the policy options for financing health care at low-income levels different from those at higher income levels. Low-income countries often have large rural and informal sector populations, limiting the taxation capacity of their governments. When a country's taxation capacity is as low as 10 percent of GDP or lower, it would take 30 percent of government revenues to meet a target of 3 percent of GDP health expenditure through formal collective health care financing channels. In most countries, public expenditure on health care is much lower than this, often not surpassing 10 percent of public expenditure, hence less than 1 percent of GDP of public resources available for the health sector (see Figure 3--modified from World Bank 1997). 2 Low-Income Countries Have Weak Capacity to Raise Revenues PDG 100 Governments in many % countries often raise less sa 80 than 20% of GDP in public revenues; and seun ve 60 The tax structure in many Retne 40 low-income countries is mnr often regressive. veoGlatoT 20 0 100 1,000 10,000 100,000 Per capita GDP (Log scale) A related set of problems is faced during the pooling of financial resources at low-income levels. Pooling requires some transfer of resources from rich to poor, healthy to sick, and gainfully employed to inactive. Tax evasion by the rich and middle class in the informal sector is widespread in low-income countries, allowing higher income groups to avoid contributing their share to the overall revenue pool. Without such pooling of revenues and sharing of risks, low-income populations are exposed to serious financial hardship at times of illness (Diop et al. 1995). Figure 4 (Wagstaff et. al. 2001) indicates households whose income drops below the poverty line (horizontal bar indicating poverty line) due to out-of-pocket expenditure on health care (vertical drop bars on the income distribution curve). Any pooling that does occur tends to be fragmented along income levels, preventing effective cross-subsidies between higher and lower income groups. In many poor countries, local community financing schemes have emerged partially as an informal sector response to these shortcomings in revenue pooling at low-income levels. Out-of-Pocket (OOPs) Expenditure and Poverty without Risk Sharing 10 9 LPfo Vietnam 8 lepit 7 mulsa 6 5 ureitdnepxe 4 3 HH 2 1 0 1 500 999 1498 1997 2496 2995 3494 3993 4492 4991 5490 5989 Pov line = 1789870 dongs/day Pre OOP HH income Post OOP HH income ` Faced with overwhelming demand and very limited resources, many low-income countries use nonspecific broad expenditure caps that push rationing and resource allocation decisions to lower levels of the provider system. This often leads to serious drug shortages, equipment breakdowns, capital stock depreciation, and lowering of standards of hygiene. Politically and ethically difficult rationing decisions about the targeting of public expenditure to the poor are also difficult in such an environment. As a result of such difficulties, the rich often benefit more from public subsidies and public expenditure than the poor (Figure 5--Peters 2001; see also Gwatkins 2001). 3 Pro-Rich Bias of Public Subsidies in Many Low-Income Countries 0.50 0.40 xednInoitartne 0.30 Pro-Rich Distribution 0.20 0.10 ncoC 0.00 -0.10 Pro-Poor Distribution -0.20 -0.30 ae in rah ra ur ia a a sa eil ac y In sc iab Pe ar in ny liza alar m Ch Ri iasy ua Gu Bi nahtsa hseda aid ana taraj anay Gh aisen do ga mant Ch Ke Br ailogn acirfA Ke Raj Prr akatan Vie lguB hsedarP udaNli Gu S Gu ats laaM anitneg ugrU ta Kar In daaM lagneBts a Mo nduroH looC Ar We Tam Co Ut Andhr It has been less difficult for national policymakers to design effective health financing schemes for individuals and households in formal employment whose income is readily identifiable and who can be taxed at the source. Unfortunately, the formal sector in most low-income countries is small, compared with populations in rural areas and informal employment. In low-income countries, large segments of the population in informal employment remain without effective collective arrangements to pay for health care or to protect them from the cost of illness (Guhan 1994; Midgley 1996; Van Ginneken 1999; and World Bank 1995). C. ROLE OF COMMUNITIES IN PROVIDING FINANCIAL PROTECTION Community initiatives have recently begun to bridge the large gap in social protection between people covered by formal schemes and those with no protection at all against the cost of illness who are exposed to the impoverishing effects of user charges. (Arhin-Tenkorang 1994, 1995, 2000; Atim 1998, 1999; Bennett et al. 1998; Jakab and Krishnan 2001; Musau 1999; and Ziemek and Jutting 2000). In the literature, the term "community financing" has evolved into a generic expression used to cover a large variety of health financing arrangements (Abel-Smith 1988; Dror 1999; Foster 1982; Hsiao 2001; McPake 1993; Muller 1983; Navarro 1984; Rifkin 1988; Stinton 1982;). On one hand different authors use the term "community financing" in different ways. On the other hand, similar--more specific-- terms are often used to describe similar financing arrangements. Microinsurance, community health funds, mutual health organizations, rural health insurance, revolving drug funds, community involvement in user fee management have all been referred to as community-based financing. Yet, each of these risk-sharing arrangements has different objectives, policy, management, organizational, and institutional characteristics, and different strengths and weaknesses. The Oxford dictionary defines community as (a) "joint or common, ownership, tenure or liability"; (b) "common character"; (c) "social fellowship"; (d) "life in association with others"; (e) "common character"; (f) "common or equal rights or rank"; and (g) "people organized into common political, municipal or social unity". Community-based health care financing reflects most of these concepts. One common feature of the definitions is the predominant role of collective action in raising, pooling, allocating/purchasing, and/or supervising the management of health financing arrangements, even when there is interface with government programs and services in terms of subsidies, supplemental insurance coverage, or access to public provider networks. Some community financing schemes cover common geographic entities, while 4 others are based on professional affiliations, religion, or some other kind of joint activity. A second common feature of community financing schemes relates to the beneficiaries of these schemes, which tend to be populations with no other financial protection or access to collective financing arrangement to cover the cost of health care. A third common feature is the voluntary nature of these schemes, and tradition of self-help and social mobilization that are embraced by the poor in many low-income countries. II. CONCEPTUAL UNDERPINNINGS FOR COMMUNITY-BASED ACTION IN HEALTH CARE FINANCING If both markets and governments fail to provide financial protection mechanisms for the poor, what is it about community-based initiatives that makes them turn to such arrangements? The growth of community-based health financing arrangements rests on developments in three related areas (Table 1-- see Dror, Preker and Jakab 2002): · microfinance (microsavings, microcredits, microinsurance, financial intermediation); · social capital (community, network, institutional, and societal links); and · mainstream theories (welfare economics, public finance, health economics, and public health). A. LINKS TO EXISTING MICROFINANCE ORGANIZATIONS The role of microfinance in poverty alleviation for low-income groups has become a prominent theme in recent years (ADB 2000; Brown and Churchill 2000; Otero and Rhyne 1994; and Zeller and Sharma 2000). Poor and rich households are equally exposed to a range of events beyond their immediate control that put them at financial risk. Such events range from predictable life-cycle events such as marriage, childbirth, education, and death to less predictable events such as droughts, fire, floods, and catastrophic illness. The difference between poor and non-poor households is the availability of mechanisms to cope with the financial consequences of unpredictable events. Non-poor households take advantage of a wide range of risk-protection mechanisms that are available even in the lowest income countries. This includes savings, access to credit, insurance, and other financial intermediation mechanisms. Until recently, few risk-protection mechanisms were accessible to the poor. It was assumed that the poor--living on less than a dollar a day--were neither willing nor able to save or contribute to insurance against the risks they faced. In sum, the poor were thought to be "unbankable" and "uninsurable" (Zeller and Sharma 2000). This led to the growth of informal risk-protection mechanisms through families, friends, and community networks. However, the past decade has witnessed a steady expansion of successful initiatives to provide the poor with savings, credit, and insurance services. Growing experience with these mechanisms suggests that the poor can be creditworthy, can save, and can buy insurance. In particular, four microfinance instruments have been developed to improve the productive needs of low- income households. They are: (a) microcredits that help improve the immediate human, physical, and social capital of the poor (e.g., small short-term loans to help pay for training, a piece of farm equipment, and access to social networks); (b) savings to be used to build up the medium-term capital of the poor such as education, the down payment on a piece of land, and dowry for marriage of a daughter into a good family; (c) insurance to stave off the unpredictable expenses such as theft, loss, and illness); and (d) financial intermediation (payment systems to facilitate trade and investments). 5 Table 1. Conceptual Underpinnings of Community Financing Schemes Key Conceptual Underpinnings Microfinance 1. Microcredit Risk taking (take advantage of opportunity, avoid over-cautious behavior) Current liquidity management (smooth out consumption, increase choice) Short-term shocks (drought, famine) 2. Microsavings Predictable life-cycle events (education, marriage dowry, childbirth, death) Capital formation (purchase of equipment, down payment on land, growth) Future liquidity management (smooth consumption, increase choice) 3. Microinsurance Long-term income support (life and disability insurance, pensions) Short-term income support (sick pay, unemployment insurance--not well developed) Unpredictable health expenditure (health insurance) Replacement of loss (fire and theft insurance) 4. Financial intermediation Payment and money-transfer services (facilitate trade and investments) Social capital 1. Community links Between extended families, local organizations, clubs, associations, civic groups 2. Network links Between similar communities (horizontal) and different communities (vertical) 3. Institutional links To communities' political, legal, and cultural environment 4. Societal links Between governments and citizens through public/private partnerships and community participation Mainstream 1. Welfare of society theories Income and growth 2. Public finance Taxation and social insurance 3. Social policy Social services and safety nets 4. Health policy Public health priorities and health systems Life, casualty, and crop insurance is often used to secure loans for low-income populations. Microfinance instruments help the poor avoid having to invest in less cost-efficient means of saving, credit, and insurance such as jewelry, livestock, and staple food, or to resort to inefficient barter systems of payment (payment in-kind). And these instruments contribute to the early transformation of barter transactions into more formal economic exchange and formalization of property rights. The extension of such techniques to the health sector is now being observed in many microfinance and development organizations in low-income countries, especially in the case of microinsurance (Brown and Churchill 2000; Dror and Jacquier 2000; and ILO 2000b, 2001). Extending microinsurance techniques to health care presents a unique set of challenges under exploration. While life and crop insurance deals mainly with the financial cost of income loss, health insurance presents an additional set of issues related to financing tangible services for which the cost is neither fully predictable nor constant. This includes the range and severity of different illnesses, the range and scope of services provided, and the behavior of both patients and providers (the latter influenced particularly by the payment mechanism due to moral hazard, adverse selection, and fraud, especially in the form of supplier-induced demand). 6 B. LINKS TO COMMUNITY-LEVEL SOCIAL CAPITAL Why have microfinance organizations been able to reach low-income individuals and households while more formal national systems continue to fail to do so? Clues to answer this question come from the social capital literature of the 1990s, which can be summed up as "it is not what you know, but whom you know" that counts (Platteau 1994, Woolcock 1998, Woolcock 2000). When hard times strike, it is often family and friends that constitute the ultimate safety net for low-income groups. Evidence suggests that social capital has four dimensions with both potential positive and negative impacts on development. The four dimensions include: · community links such as extended families, local organizations, clubs, associations, and civic groups--people in small communities helping each other (Dordick 1997) · network links between similar communities (horizontal) and between different communities (vertical) such as ethnic groups, religious groups, class structures, gender (Granovetter 1973) · institutional links such as communities' political, legal, and cultural environment (North 1990) · societal links between governments and their citizens through complementarity and embeddedness such as public/private partnerships and the legal framework that protects the rights of association (e.g., chambers of commerce and business groups) and community participation in public organizations (e.g., community members on city councils and hospital boards) (Evans 1992, 1995, 1996). Low-income households are likely to have more trust in microhealth insurance programs that are linked to the community credit, savings, and insurance organizations to which they already belong and feel they have some control over. And, they often regard national systems as impersonal and distant and think they will never benefit from them. This view is reinforced when the national programs ration care to focus on "global" public health priorities that--although they may have large externalities and benefits to society as a whole--often do not respond to the immediate, day-to-day health care needs of the poor. But such social capital has both benefits and costs. The downside of social capital occurs when communities and networks become isolated or parochial or work at cross-purposes to societal collective interests (e.g., ghettos, gangs, cartels). Inter-community ties or bridges are needed to overcome the tendency of communities and networks to pursue narrow, sectarian interests that may run counter to broader societal goals. (Narayan 1999) Community financing schemes are vulnerable to a number of these shortcomings associated with social capital: · Community-financing schemes that share risk only among the poor will deprive its members of much needed cross-subsides from higher income groups. · Community-financing schemes that remain isolated and small deprive their members of the benefits of spreading risks across a broader population. · Community-financing schemes that are disconnected from the broader referral system and health networks deprive their members of the more comprehensive range of care available through the formal health care system. C. LINKS TO MAINSTREAM PUBLIC ECONOMICS Community-financing schemes--in addition to their links to microfinance and social capital--benefit from interconnectivity to the overall welfare of the society in which they exist, the system of public financing (no matter how weak it may be), and the broader social policy underpinning the prevailing national health system. Schemes that build such connections at an early stage are better able to evolve in 7 terms of expanding the number of members covered, level of resources mobilized, size of the risk pool, and range of benefits they can cover as the local community they serve grows and evolves. Their members have more to gain through such connectivity than they would through isolation. Principal-agent problems also explain why community-based initiatives are expected to be more successful than purely market-based institutions at providing financial protection products. These problems can be overcome in two ways, by: (a) designing incentives that align the interest of the agent (insurer) with that of the principal (member), and (b) designing monitoring systems that allow the principal (member) to effectively observe the actions of the agent (insurer). The proximity of community schemes (agents) to their members (principals) allows effective monitoring, which is much more difficult at the national level. Proponents of linkage between community involvement and public finance argue their case on both philosophical and technical grounds. In most societies, care for the sick and disabled is considered an expression of humanitarian and philosophical aspirations. But one does not have to resort to moral principles or arguments about the welfare state to justify collective intervention in health. The past century is rich in examples of the failure of the private sector and market forces alone to secure efficiency and equity in the health sector. There is ample justification for such an engagement on both theoretical and practical grounds. In the case of efficiency, there is ample evidence of the significant market failure that exists in the health sector (information asymmetries; public goods; positive and negative externalities; distorting or monopolistic market power of many providers and producers; absence of functioning markets in some areas; and frequent occurrence of high transaction costs (Arrow 1963; Atkinson and Stiglitz 1980; Bator 1958; Evans 1984; and Musgrave and Musgrave 1984). In the case of equity, there is equally good evidence that individuals and families often fail to protect themselves adequately against the risks of illness and disability on a voluntary basis (Barer et al. 1998; and van Doorslaer et al. 1993) III. METHODOLOGY FOR ASSESSING IMPACT, STRENGTHS, AND WEAKNESSES To assess the impact, strengths and weaknesses of community-based involvement in health care financing we will use a modified version of the Bank's Poverty Reduction Strategy Paper (PRSP) framework (Claeson et al. 2001). See Figure 6. According to this framework, community financing can be seen as having three independent objectives: (a) it mobilizes financial resources to promote better health and to diagnose, prevent, and treat known illness; and (b) it protects individuals and households against direct financial cost of illness when channeled through risk-sharing mechanisms; and (c) it gives the poor a voice in their own destiny and makes them active participants in breaking out of the social exclusion in which they are often trapped. We will not deal with the indirect impact of illness on loss of income due to interruption of employment although this is clearly another important dimension of financial protection against the cost of illness. This framework is consistent with the three goals of health systems that were emphasized by the World Health Report 2000 (WHO 2000): financial fairness (an indicator that measures inequality of the financial contribution for health across households ); disability-adjusted life expectancy (DALE, an indicator that combines life-expectancy and disability measures); and responsiveness (a consumer- satisfaction indicator that combines ethical and consumer quality dimensions). This framework is also consistent with the International Development Goals (IDGs) relating to achievement of better health and protection against impoverishment by the year 2015. 8 The determinants of financial protection, improved health, and social inclusion are complex. The PRSP framework emphasizes the following causal links: (a) close tracking of key outcome measures relating to improved financial protection, health, and social inclusion: (b) demand and utilization patterns; (c) supply in the health system and related sectors; and (d) policy actions by governments, civil society, the private sector, and donors. Determinants of Financial Protection, Health, and Social Inclusions Demand/Utilization Supply Policies Outcomes Individuals, Financing and Other Inputs Governments Households and Services Communities Communities Other Sectors Donors Financial Willingness and Ability to Pay Health Services Financing Government Protection Public health Revenues Stewardship Ambulatory Care Pooling Coordination In-Patient Care Purchasing Information Regulation Input Markets M&E Household Behavior Enforcement Human Resources HNP Risk Factors Knowledge Communities Needs Pharmaceuticals Expectations Equipment Self-Regulation Demand Consumables Information Health Capital M&E Outcomes Enforcement Household Assets Other Sectors Donors Human (education, health status) Social Physical (house and livestock) GDP, growth, prices, inflation Loans Inclusion Financial (loans, savings, insurance) Tax systems, Insurance, Micofinance Grants Social (communities, networks) Water, food, roads Advise · Outcome indicators. Much work is still needed to develop a meaningful set of indicators for improving health, protection against impoverishment, and combating social exclusion. For this report, we have used both the financial fairness, DALE, and responsiveness indicators recommended by the World Health Organization (WHO) and several intermediate indicators (see next section for details). · Demand and utilization in influencing financial protection. There is a complex interplay between household assets (human, physical, financial, and social), household behavior (risk factors, needs and expectation for services), ability and willingness to pay, and the availability of insurance or subsidies (Soucat et al. 1997). This part of the analysis emphasizes the importance of household and community behavior in improving health and in reducing the financial risks. · Supply in health system and related sectors. There is a hierarchy of interest from non-health sector factors in improving financial protection--such as GDP, prices, inflation, availability of insurance markets, effective tax systems, credit, and savings programs--to more traditional parts of the health system (a) preventive and curative health services; (b) health financing; (c) input markets; and (d) access to effective and quality health services (preventive, ambulatory, and in- patient). In respect to the latter, organizational and institutional factors contribute to the incentive environment of health financing and service delivery systems in addition to the more commonly examined determinants such as management, input, throughput, and output factors (Harding and Preker 2001). · Policy actions by governments, civil society and the private sector. Finally, through their stewardship function, governments have a variety of policy instruments that can be used to strengthen the health system, the financing of services, and the regulatory environment within which the system functions (Saltman and Ferroussier-Davis 2000). This includes regulation, contracting, subsidies, direct public production, and ensuring that information is available. In 9 countries with weak government capacity, civil society and donors can be encouraged to play a similar role. Four levels of analysis were used to assess the impact, strengths, and weaknesses of community involvement in financial protection against the cost of illness and improved health. This includes: (a) survey of the literature on the impact, strengths, and weakness of different types of community involvement in health financing; (b) macro-level cross-country analysis of the impact of different health care financing mechanisms on national health systems' performance indicators--health, financial fairness, and responsiveness; (c) micro-level household data analysis of the specific impact of community financing schemes on overall welfare of the poor--financial protection and access to health services for the poor; (d) regional reviews of the Asia and Africa experience of community involvement in health care financing, including different public policy options such as subsidies, reinsurance, linkages to formal public financing systems, and management-capacity building. A. METHODOLOGY FOR SURVEY OF LITERATURE ON COMMUNITY FINANCING Despite the recent growth in research on community-based health care financing, there is a paucity of systematic evidence regarding the performance of these schemes in terms of their impact on broad outcome goals such as health, protection against impoverishment, and combating social exclusion. In particular, little is known about their: (a) effectiveness in mobilizing resources and improving access to effective and quality health care; (b) role in sharing risks across population groups; and (c) impact on addressing the problems associated with social exclusion. And, despite progress made through the World Health Report 2000, experts are still debating which indicators best capture progress toward achieving these goals. The review looked at any past studies whose main focus had been to examine community involvement in health care financing. Based on this broad criterion, 43 past studies were include in the review. The selected papers included articles published in peer-reviewed journals, reports published in formal publication series of international organizations (e.g., WHO, International Labour Organisation, United Nations Children's Fund), internal unpublished documents of international organizations and academic institutions, and conference proceedings. Table 2 presents the breakdown of the reviewed studies, by publication type. 10 Table 2. Summary Statistic of the Literature Reviewed, by Publication Type Peer reviewed Published report Internal document Conference journal article of International proceeding Organizations or Academic Institutions Number of studies 20 15 4 4 Of these 43 studies, five were conceptual papers, seven were large scale comparative papers (analyzing five or more community-based health financing schemes) and the remaining 31 were case studies. The regional breakdown of the case studies was 15 in Africa, 11 in Asia, and 4 in Latin America. Language barriers and time constraints created a certain selection bias--Spanish literature was not included in our search while French literature was (Jakab and Krishnan 2001 ). Assessment of Performance Since past research of community financing schemes varies considerably both in the issues examined and methodologies used, a standard set of questions were asked relating to both the review of impact assessments and review of determinants (key strengths and weaknesses of various types of schemes). The following three questions were asked relating to the impact of community involvement on health, financial protection, and social inclusion. Question 1: What and how robust was the evidence on the amount of resources that could be mobilized through community involvement to pay for health care and the sustainability of this source of financing? Question 2: What and how robust was the evidence on the effectiveness of community involvement in protecting individuals against the impoverishing effects of illness? Question 3: What and how robust was the evidence on the role that community involvement played in combating social exclusion by allowing low-income groups to have a more direct role in the financing of their health care needs and protecting them against the financial burden of illness? A number of studies offered conclusions on resource mobilization, financial protection, and social exclusion based on the experience of authors or review of other studies but did not provide actual evidence in support of their conclusions. The review excluded studies of performance assessments from the analysis. The review also excluded studies that did not use controls from the performance evaluation. This approach yielded 11 studies for the performance assessment of the review. Assessment of Institutional Determinants of Performance The direct and indirect determinants of improved health, financial protection against the cost of illness, and social inclusion are complex. As described earlier by the PRSP framework, policy actions by governments, civil society, and the private sector are mediated though supply and demand factors related to both the health sector and other sectors that affect the outcome measures that are being examined. This would include indicators of the service delivery system (product markets), input generation (factor markets), the stewardship or government oversight function (policymaking, coordination, regulation, monitoring, evaluation) and market pressures. The current body of literature on community financing is not comprehensive so the report looked only at factors directly related to health care financing. 11 Table 3 provides a list of the core technical design, management, organizational, and institutional characteristics related to health care financing in general. Based on this framework, the study reviewed 43 assessments of community financing schemes for their impact, strengths, and weaknesses.[Suggestion: for the reader who will not turn to the background papers or referenced paper, perhaps a line here or later about the difference between `organization' and "institution" would be helpful. (adapted from Preker et al. 2001).] Table 3. Core Characteristics of the Community-Based Financing Schemes Key Policy Questions Technical design · Revenue-collection mechanisms characteristics Level of prepayment compared with direct out-of-pocket spending Extent to which contributions are compulsory compared with voluntary Degree of progressivity of contributions Subsidies for the poor and buffer against external shocks · Arrangements for pooling revenues and sharing risks Size Number Redistribution from rich to poor, healthy to sick, and gainfully employed to inactive · Purchasing and resource allocation Demand (for whom to buy)? Supply (what to buy, in which form, and what to exclude?) Prices and incentive regime (at what price and how to pay?) Management · Staff characteristics Leadership Capacity (management skills) · Culture Management style (top down or consensual?) Structure (flat or hierarchical?) · Access to information Financial, resources, health information, behavior Organizational Organizational forms (extent of economies of scale and scope, and contractual characteristics relationships?) Incentive regime (extent of decision rights, market exposure, financial responsibility, accountability, and coverage of social functions?) Linkages (extent of horizontal and vertical integration or fragmentation?) Institutional Stewardship (who controls strategic and operational decisions, regulations?) characteristics Governance (what are the ownership arrangements?) Insurance markets (rules on revenue collection, pooling, and transfer of funds?) Factor and product markets (from whom to buy, at what price, and how much?) Outcome Indicators Health Protection Against Impoverishment Social Inclusion 12 B. METHODOLOGY FOR REGIONAL REVIEWS OF SELECTED ASIA AND AFRICA EXPERIENCE The main objective of the reviews of selected Asia and Africa experiences was to provide additional insights about several key issues from the perspective of the two regions of the world that carry the heaviest burden of mortality and morbidity, that have the weakest risk-sharing arrangements to protect their populations against the impoverishing effects of illness, and that have the greatest number of poor living in absolute poverty and social exclusion (Arhin-Tenkorang 2001 and Hsiao 2001). In addition to contributing to an understanding about the current roles of community involvement in health care financing, the regional reviews also focused on future policy options. Key questions asked include the following: · Using the same framework described under the survey of the literature, what are the main characteristics of existing community involvement in financing health care in the Africa and Asia regions in terms of impact, strengths, and weaknesses of existing schemes (describe successful and unsuccessful features)? · To what extent do community financing schemes serve the objective of securing adequate, equitable, and sustainable financing for the low-income and rural populations served (impact on the poor)? · What are the main challenges and obstacles to improving community arrangements to provide adequate, equitable, and sustainable financing? · Are there other viable alternatives to community financing in the country settings where they exist today? · In the context of these study findings, what role could the international donor community play to improve financing for rural and other low-income population groups? C. METHODOLOGY FOR MICRO-LEVEL HOUSEHOLD SURVEY ANALYSIS The aim of the micro-level household survey analysis was to shed light on two questions (Jakab et al. 2001): (i) what characteristics affect the decision of households to join community-based prepayment schemes; and (ii) do community health financing schemes provide financial protection for their members against the cost of illness. Eleven household budget surveys, four Living Standard Measurement Surveys (LSMS) and nine Demographic and Health Surveys (DHS) were screened for community financing data. Most of these surveys did not allow an identification of households with access to community-based health financing. Of the 11 smaller scale nonstandardized surveys that matched the requirements for the core list of variables, 5 were available for further analysis and were included in this report. Table 4 summarizes the key characteristics of these surveys. The remaining 6 were either not accessible for further analysis (4), data collection incomplete (1), or authors were not available to collaborate (1). 13 Table 4. Characteristics of 5 Survey Instruments Name of scheme Year of Sample size Organization data (Households) associated with the collection survey Rwanda 54 prepayment 2000 2,518 Partnerships for Health schemes in 3 districts Reform (PHR) in of Kabutare, Byumba collaboration with and Kabgayi National Population Office Senegal 3 Mutual Health 2000 346 Institute of Health and Insurance Schemes Development, Dakar in (Thiés Region) collaboration with ILO India (1) Self-Employed 1998­99 1,200 National Council of Women's Association Applied Economic (SEWA) Research (NCAER) India (2) Self-Employed 1,200 London School of Women's Association Hygiene and Tropical (SEWA) Medicine Thailand Voluntary Health Card 1994­95 1,005 National Statistics Office Scheme (HCP) The five household surveys identified and accessible for analysis for the purposes of this report represent nonstandardized relatively small-scale data-collection efforts with a sample size of 346 to 1,200 households. None of the surveys were nationally representative; they were a random sample of the local population. With the exception of Thailand, four surveys are very recent. Determinants of Inclusion To assess the determinants of social inclusion in community financing schemes, we assume that the choice of whether to enroll is influenced by two main determinants: (i) individual and household characteristics and (ii) community characteristics. Individual and household characteristics influence the cost and the benefit calculation of the rational individual decision maker. This choice is moderated, however, through certain social characteristics of the community. The individual rational choice model of weighting costs and benefits of joining a prepayment scheme is altered by the social values and ethics of the local culture. For example, two individuals with similar individual and household characteristics (e.g., income, household size, assets, education level, health status) may decide differently about joining or not joining a prepayment scheme depending, for example, on encouragement from community leaders, availability of information, ease of maneuvering unknown processes. To estimate the weight of these determinants, a binary logit model was applied to four of the data sets and a binary probit was applied to the Senegal data set. The model can be formally written as follows. Prob (membership>0) = X11 + X22+ (1) The independent variable takes on a value of 1 if the individual belongs to a community financing scheme and 0 if s/he does not. X1 represents a set of independent variables that are characteristics of the individual and the household such as income, gender, age, marker on chronic illness or disability. X2 represents a set of independent variables that approximate the social values in the communities: religion, marker on various communities where appropriate. Other variables specific to the surveys as well as 14 interaction terms were included where appropriate. 1 and 2 are vectors of coefficient estimates and is the error term. The two variables of primary interest are income (measure of social inclusion) and a marker for community factors (dummy variable). Control variables also included gender, age, disability or chronic illness, religion, and distance to the health center under the scheme. Some of these variables are important to control for the different probability of health care use (e.g., age, health status, distance from provider) These variables also allow us to test the presence and importance of adverse selection that all voluntary prepayment schemes are subject to. Others variables included control for the different individual and household attitudes toward investment in health at a time when illness is not necessarily present (e.g., gender, religion). Literature has shown that the distance to the hospitals and local health centers and existence of outreach programs influence the decision to purchase membership to the scheme. Determinants of Financial Protection To empirically assess the impact of scheme membership on financial protection, a two-part model was used.1 The first part of the model analyses the determinants of using health care services. The second part of the model analyses the determinants of health care expenditures for those who reported any health care use. There are several reasons for taking this approach. First, using health expenditure alone as a predictor of financial protection does not allow capture of the lack of financial protection for people who choose not to seek health care because they cannot afford it. As the first part of the model assesses the determinants of utilization, this approach allows us to see whether membership in community financing reduces barriers to accessing health care services. Second, the distribution of health expenditures is typically not a normal distribution. Many non-spenders do not use health care in the recall period. The distribution also has a long tail due to the small number of very high spenders. To address the first cause of non-normality, the study restricted the analysis of health expenditures to those who report any health care use. As the first part of model assesses determinants of use, we will still be able to look into whether scheme membership removes barriers to care. To address the second part of non-normality, a log-linear model specification is used. Part one of the model is a binary logit model for the Rwanda, Thailand, India data sets and a probit model in the Senegal model. The model estimates the probability of an individual's visiting a health care provider. Formally, part one of the model can be written as follows: Prob (visit>0) = X + (2) Part two is a log-linear model that estimates the incurred level of out-of-pocket expenditures, conditioned on positive use of health care services. Formally, part two of the model can be written as follows: Log (out-of-pocket expenditure | visit>0) = X + µ (3) where X represents a set of individual and household characteristics that are hypothesized to affect individual patterns of utilization and expenditures. and are vectors of coefficient estimates of the respective models. and µ are error terms. 1 This model is similar to the two-part demand model developed as part of the Rand Health Insurance experiment to estimate demand for health care services (Duan et al. 1982; and Manning et al. 1987). For a recent application of the model that analyses the access impact of school health insurance in Egypt, see Yip and Berman 2001. 15 The two variables of primary interest are scheme membership status and income. Other control variables were also included in the estimation model to control for the differences in need for health care (e.g., age, gender); differences in preferences toward seeking health care (e.g., gender, religion); and differences in the cost (direct and indirect) of seeking health care (e.g., distance). D. METHODOLOGY FOR MACRO-LEVEL CROSS-COUNTRY ANALYSIS For the dependent variables of the macro-level country analysis, the study used the standard indicators proposed by WHO for health systems performance (WHO 2000). These are the disability-adjusted life expectancy (DALE), the index of level of responsiveness (IR), the index of fairness of financial contribution (IFFC), the index of distribution of responsiveness (IRD) and the index of equality of child survival (IECS). Only the observed data for these indicators were included in the analysis. For the independent variables of the macro-level analysis, countries were divided into three groups based on the extent of their risk-sharing arrangements. We assign countries to the advanced risk-sharing category when they have either a social health insurance scheme or a health financing scheme based on general taxation, and when these two schemes are associated with the principle of universal coverage. Second, there are countries with no explicit reference to overall coverage of the population. They usually have mixed health financing systems, with some part of the population partially covered via general taxation and specific population groups covered by health insurance schemes. These countries will be associated with medium risk-sharing. Finally, there are countries with general taxation systems but that incompletely cover the population; these will associated with low-risk sharing. This classification allows us now to define the two main organizational dummy variables: DARS = 1 when a country belongs to the set of advanced risk-sharing systems and 0 otherwise; DMRS = 1 when a country belongs to the set of medium risk-sharing systems and 0 otherwise. The methodology for this analysis is described by Carrin et al. 2001. The objective of the analysis is to examine the degree to which risk sharing has a beneficial impact on the five indicators of health systems performance. The analysis used the following specification for the impact of risk sharing on the level of health: Ln (80--DALE) = a1 + b1 Ln HEC + c1 Ln EDU + d1 DARS (1) HEC refers to the health expenditure per capita (in U.S. dollars). EDU refers to the educational attainment in society and is measured by the primary enrolment. The dependent variable is the logarithm of the difference between the observed DALE and a maximum. Several alternative models were also tested. The hypothesis is that advanced risk sharing (among indirect determinants such as education) is associated with a better definition of the benefit package of health services to which citizens are entitled, which translates into increased overall level of health. The analysis used two alternative functional forms to assess the impact of risk sharing on responsiveness; Ln [IR/(1- IR] = a21 + b21 HEC + c21 EDU + d21 DARS (2a) and Ln (1--IR) = a22 + b22 Ln HEC + c22 Ln EDU + d22 DARS (2b) 16 The hypothesis to be tested is that advanced risk-sharing systems are associated with a larger degree of stewardship. The latter, in turn, is likely to positively influence the mechanisms and incentives that entail a greater responsiveness. The analysis used three measures for distributional impact. This included: an index of equality of child survival (IECS), an index of fairness of financial contribution (IFFC), and an index of distribution of responsiveness (IRD). Several models were tested. A model was developed that examined the impact of the dummy variable (DARS) on the distributional variables for health, fairness, and responsiveness. We have adopted the same functional forms as in equations 2a and 2b: Ln [Ij/(1- Ij)] = a31 + b31 DARS (3a) and Ln (1--Ij) = a32 + b32 DARS (3b) where Ij (j=1,...,3) refers to the three above-mentioned indices, respectively. The effect of DARS on the indicator of fair financing is expected to be positive when using the logit form of the equation. The hypothesis to be tested is that in countries with advanced risk sharing, more so than in other systems, people pay financial contributions according to their capacity to pay. This would be associated with a higher IFFC. Second, systems with universal coverage generally pay more attention to the objective of equal treatment for equal need. It is therefore assumed that such systems also respond to people's expectations as to the nonmedical aspects of health care in a more equal way. Hence, the effect of DARS on the distribution of responsiveness is anticipated to be positive as well. Third, it is assumed that universal coverage systems are more likely to provide people with a similar benefit package than in other systems, irrespective of their socioeconomic background, with a resulting positive impact on the distributional aspects of child health. III. DISCUSSION OF MAIN FINDINGS FROM BACKGROUND REVIEWS A. DISCUSSION OF SURVEY OF EXISTING LITERATURE ON COMMUNITY HEALTH FINANCING Based on a review of 43 papers discussing community-based health care financing, the first and foremost conclusion is that there is a paucity of systematic empirical work regarding the performance of these financing mechanisms or the determinants of good outcomes in achieving good health (Jakab et al. 2001). Although several authors have tired to create a typology for community-based schemes (Atim 1998; Bennet al. et 1997; Criel 1999; and Hsiao 2001), the possibilities for variations is almost limitless, given the great diversity in objectives, design, context and implementation arrangements. Nevertheless, the review revealed four commonly encountered and well-identifiable types of schemes. In the first type of scheme, resource mobilization relies mainly on out-of-pocket payments at the point of contact with providers but the community is actively involved in designing these fees and managing the collection, pooling, and allocation of the funds mobilized in this way (community cost-sharing). In the second type, the community collects payments in advance of treatment (prepayment) and then manages these resources in paying for providers (community prepayment or mutual health organization). In the third type, 17 providers serving a particular community collect the prepayments themselves (community provider-based health insurance). In the fourth type, the community acts as "agent" to reach rural and excluded populations on behalf of the formal government or social health insurance system (government or social insurance) via contracts or agreements. Table 5 summarizes these four types of community-based financing schemes based on their core design features, management, organizational and institutional characteristics. Table 5. Types of Community-Based Financing Four Community-Based Finance Modalities Type of Government Type 4 Type 3 Type 2 Type 1 Direct Scheme schemes: user social- Linked Community Community Community fees insurance community provider- -based -managed and tax- health fund, based prepaymen user fees (spot based revolving health t schemes market) fund, or insurance prepayment Assessment of Impact. Following the framework presented in Table 3, the survey of the literature looked at three indicators of performance of community-based financing schemes (Jakab et al. 2001): (a) their effectiveness in mobilizing resources and improving access to effective and quality health care; (b) their role in sharing risks across population groups; and (c) their impact on addressing the problems associated with social exclusion (Table 5). This is followed by a discussion on the key conclusions from the performance review of the literature. Table 6. Number of Studies that Examined Each Core Health Financing Subfunctions Financing Function Revenue Pooling of Resource Allocation or Collection revenues Purchasing Type 1 5 2 3 Type 2 6 4 9 Type 3 2 2 3 Type 4 3 3 2 Multiple 10 2 3 Resource Mobilization. There is good evidence from the literature that community financing arrangements make a positive contribution to the financing of health care at low-income levels, thereby improving access to drugs, primary care, and even more advanced hospital care (Dave 1991). Such community involvement allowed rural and low-income populations to mobilize more resources to pay for health care than would have been available without this involvement (Diop 1995; McPake 1993; and Soucat 1997). But there are great variations in the volume of resource that can be mobilized this way, constrained largely by the low income of the contributing population (Atim 1998; Bennett 1998; Hsiao 2001; Jutting 2000--see Box 1). This is particularly true when most members of the community schemes are already below the poverty line. None of the studies reviewed reported the share of aggregate national resources that were mobilized through community financing arrangements. There is an urgent need to strengthen the evidence base of community financing arrangements through more rigorous registration, monitoring, and evaluation of the resource mobilization capacity of these schemes. 18 60% BOX 1. REVENUE MOBILIZATION 50% prepayment Based on data from Bennett et al. (1998), 40% this graph shows the cost recovery from from 30% prepayment of six Modality II schemes. 20% The range is from 12 percent to 51 costs 10% percent of recurrent expenditure. This shows that, for these schemes, the 0% resources collected contribute recurrent of (India) ia) Bissau) (Nepal) significantly to the full recurrent costs % GK (B'desh)SSSS (Boliv (B'desh) but do not fully cover them, thereby Caja inea pur necessitating other sources of funding, Abota (Gu Grameen Lalit such as out-of-pocket spending, government subsidies, and donor grants ( l ) Financial Protection. Where household survey data have been analyzed, a consistent observation was that community-based health financing has been effective in reaching more low-income populations who would otherwise have no financial protection against the cost of illness (Litvack and Bodart 1992). Improved financial protection is achieved through reduced out-of-pocket spending of the membership while increasing their utilization of health care services. (Atim 1998; Criel 1999; Desmet 1999, Gumber and Kulkarni 2000; Jutting 2000; and Supakakunti 1997). At the same time, some of the research suggested that the poorest and socially excluded groups are often not included in community-based health financing initiatives (Arhin-Tenkorang 1994; Criel 1999; and Jutting 2000). Those studies that compared the level of financial protection of scheme members with that of nonmembers found that belonging to some from of prepayment scheme reduced the financial burden of seeking health care (Arhin-Tenkorang 1994; Diop 1995; DeRoeck 1996; and Gumber and Kulkarni 2000). Two studies indicated that community financing does not eliminate the need for broader coverage in the case of catastrophic health care expenditures (Pradhan 2000, and Xing-Yuan 2000). Table 7. Studies that Looked at Ways to Prevent Impoverishment Studies that Confirmed Key Utilization of Members Level of Out-of-pocket Hypothesis Being Tested Relative to Non-Members expenditure of members relative to non members Increase Decrease Increase Decrease Type 1 Community User Fees 3 1 0 1 Type 2 Community Prepayment 4 2 0 6 Type 3 Provider Prepayment 3 0 0 0 Type 4 Linked to Formal System 3 0 0 2 Combating Social Exclusion. Community-based health financing schemes appear to extend coverage to many rural and low-income populations who would otherwise be excluded from collective arrangements to pay for health care and protect them against the cost of illness. However, the poorest are often excluded even from community financing arrangements, and higher income groups often do not belong, thereby segmenting the revenue pool by income group. 19 Table 8. Studies that Looked at Ways to Combat Social Exclusion Studies that Confirmed Key Sche Poorest Inability Rich do Distance Hypothesis Being Tested me not to pay, not gradient to reaches covered main partici- scheme the reason pate provider poor for not being covered Type 1 Community User Fees 3 1 1 0 0 Type 2 Community Prepayment 5 1 2 1 1 Type 3 Provider Prepayment 2 2 1 1 1 Type 4 Linked to Formal System 3 1 2 1 1 Identification of Determinants The survey of the literature also looked at factors that would contribute to strengths and weaknesses of the schemes (Jakab et al. 2001) in the following four areas: (a) technical design characteristics; (b) management characteristics; (c) organizational characteristics; and (d) institutional characteristics. The key advantage and disadvantages of community-based schemes lie in their ability to fill the policy, management, organizational, and institutional void left by extreme government failure to secure more organized financing arrangements for the poor. In this context, a number of strengths (Box 2) and weaknesses (Box 3) of community financing schemes have been identified by various authors. Box 2 Strengths of Community Financing Schemes The following highlight some of the key advantage of community-based schemes: Technical Design Characteristics · Revenue Collection Mechanisms - Shift away from point-of-service payment to increasing prepayment and risk sharing. - Flat rate premium, which facilitates revenue collection, reduces the scope for manipulation, and contributes to low transaction costs - Contribution payment that accommodates the income-generating patterns of households employed in agriculture and the informal sector (irregular, often non-cash) - Modest degree of household-level affiliation - Pro-poor orientation even at low-income levels through exemptions of premiums and subsidies, despite flat rate contribution rate - Some buffering against external shocks though accumulation of reserves and links to formal financing schemes · Arrangements for Pooling Revenues and Sharing Risks - Some transfers from rich to poor, healthy to sick, and gainfully employed to inactive through some pooling of revenues and sharing of risk within community groups · Purchasing and Resource Allocation - Most community schemes make a collective decision about who is covered through scheme, based on affiliation and direct family kinship (for whom to buy). - Many community schemes define the benefit package to be covered in advance (what to buy, in what form, and what to exclude). - Some community schemes engage in collective negotiations about price and payment mechanisms. Management · Most community schemes are established and managed by community leaders. Community involvement in management allows social controls over the behavior of members and providers that mitigates moral hazard, adverse selection, and induced demand. · Many schemes seek external assistance in strengthening management capacity. 20 · The management culture tends to be consensual (high degree of democratic participation). · Most schemes have good access to local utilization and behavior patterns. Organizational Structure · Most community schemes are distributed organizational configurations that reach deep into the rural and informal sectors. · Incentive regimes include: (a) extensive decision rights; (b) strong internal accountability arrangements to membership or parent community organization; (c) ability to accumulate limited reserves if successful but unsuccessful schemes often ask governments for bailouts; (d) mainly factor-market exposure since few overlapping schemes compete with each other in the product market: and (e) some limited coverage of indigent populations though community or government subsidies. · Vertical integration may lead to increased efficiency and quality services. Schemes that have a durable partnership arrangement or contractual arrangement with providers are able to negotiate preferential rates for their members. This in turn increases the attractiveness of the scheme to the population and contributes to sustainable membership levels. · Better organized schemes use horizontal referral networks and vertical links to formal sector. Institutional Environment · Stewardship function is almost always controlled by local community, not central government or national health insurance system, which is apt to making the schemes responsive to local contexts. · Ownership and governance arrangements (management boards or committees) are almost always directly linked to parent community schemes; free-standing health insurance schemes are rare. · There is little competition in the product market. · Competition is limited in factor markets and through consumer choice. 21 Box 3. Weakness of Community Financing Schemes The following weaknesses of community financing schemes have been identified by several authors (Carrin, Desmet, and Basaza 2001; and Bennett et al. 1998). Technical Design Characteristics · Revenue Collection Mechanisms - Without subsidies, resource mobilization is limited when everyone in the pool is poor. - Many of the poorest do not join since they cannot afford premiums. - Pro-poor orientation is undermined by the regressive flat rate contributions and by a lack of subsidies or premium exemption, which create a financial barrier for the poor. - Community-based voluntary prepayment schemes are also prone to adverse selection. - Few schemes have reinsurance or other mechanisms to buffer against large external shocks. · Revenue-Pooling and Risk-Sharing Arrangements - The scope for transfers within very small pools is limited (often fewer than 1,000 members per scheme). · Purchasing and Resource Allocation - Without subsidies, the poorest are often left out (for whom to buy). - The benefit package is often very restricted (what to buy, in what form, and what to exclude). - Providers can often exert monopoly power during price and payment negotiations. Management · Community leaders are as vulnerable to adverse incentives and corruption as national bureaucrats. · Even with external assistance, absorptive capacity in management training is limited. · Extensive community consultation is time consuming and can lead to conflicting advice. · Most schemes do not use modern information management systems. Organizational Structure · Even very distributed organizational configurations may have difficulty reaching deep into the rural and informal sectors. · There are often conflicting incentives, especially among extensive decision rights, soft budget constraints at time of deficits (bail outs by governments and external sources of funding such as nongovernmental organizations), limited competitive pressures in the product markets, and lack of financing to cover the poorest population groups. · The less organized schemes are often cut off from formal sector networks. Institutional Environment · Government stewardship and oversight function is often very weak, leading to a poor regulatory environment and lack of remedies in the case of fraud and abuse. · Ownership and governance arrangements are often driven by non-health and financial protection objectives. · Choice in strategic purchasing is limited by small number of providers in rural areas. · True consumer choice is often limited by lack of a full insurance and product market, leading to (a) adverse selection (signing on only the better-off, working age, and healthy); (b) moral hazard (members making unnecessary claims because they have insurance coverage); (c) free-rider effect (households waiting until they think they will be sick before joining); and (d) information asymmetry (e.g., concealing preexisting conditions). B. DISCUSSION OF MAIN FINDINGS FROM ASIA REGIONAL REVIEW The review of selected Asia experiences emphasized the heterogeneity of community financing schemes and the fact that their performance is highly dependent on the nature of their technical design, management, organizational and institutional characteristics. For the purpose of this review, Hsiao (2001) classified community involvement in health care financing into five types: (a) direct subsidy to 22 individuals (Thai Health Card and Tanzania Community Health Fund): (b) cooperative health care (Jiangsu Province and Tibet); (c) community-based third-party insurance (Rand Experiment in Sichuan Province and Dana Sehat); (d) provider sponsored insurance (Dkaha Community Hospital, Gonoshasthya and Bwamanda ; and (e) producer/consumer cooperative Grameen). Based on this typology, the Asia review ranked the community financing schemes examined according to their potential impact on several intermediate outcome indicators (coverage, equity in financing, efficiency and cost containment in service delivery, access, quality, and degree of risk sharing). The results are summarized in Table 9. Table 9. Potential Value Added by Types of Community Financing Schemes Type of Who Popula- Equity Increase Improve Improv Greater Community controls use tion to be in efficiency access e risk Financing of fund covered finan- and quality pooling Scheme & raise cing reduce funds cost Prepay User Government Low Low None Low Low Low Fees Individual. Modest Low None Modest Low Low households Cooperative Local High Low High High High Modest Healthcare community unless w/ and special gov't purpose NGO subsidy Community- Community Cover Low Low High for Low High based third- higher those party income insured Insurance families Provider Hospitals Cover Low Low High for Low High sponsored higher insured Insurance income families Provider or Cooperatives Cover Low High High High Medium consumer member cooperative Source: Hsiao (2001). Based on this framework, the review made the following observations: · Rural households and urban poor households are willing to prepay a portion of their health services. The resources that can be raised in this manner depend on both economic and social factors. · Since the membership of many community financing schemes consists of poor households, their ability to raise significant resources to pay for health care is limited by the community's overall income, exposure to out-of-pocket payment when not enrolled, availability and size of subsidies, and satisfaction with the services provided. The poor and near poor are more motivated to prepay if their contributions are supplemented by government or donor subsidies. For the poorest households, this subsidy has to be a large share of the total payment. · The social factors that influence membership rates include a sense of kinship, mutual community concern, and trust and confidence in the management of the scheme. · A major value added of well-performing community financing schemes is expanded access to quality services, improved efficiency of management and service delivery, and cost containment. · Government and nongovernmental organizations (NGOs) often catalyzed the startup of the community financing schemes in question and contributed to its management and sustainability 23 · Finally, members appear to prefer coverage for both primary care and more expensive hospital care. Since many schemes do not mobilize sufficient resources to pay for both, many communities opt for primary care coverage, which they will use regularly for their basic health care needs, rather than insurance coverage for rarer and more expensive events that may only happen once or twice in a lifetime and whose concept is often poorly understood. This creates a tension or trade-off between individual needs/demands for basic care and household/community needs for financial protection (see Figure 7). Hospitalization and Impoverishment BIHAR UTTAR PRADESH PUNJAB RAJASTHAN GUJARAT MADHYAPRADESH WEST BENGAL NORTH EAST ALL INDIA MAHARASHTRA ORISSA HARYANA ANDHRAPRADESH KARNATAKA TAMIL NADU KERALA 0 5 10 15 20 25 30 35 40 Percent Falling Into Poverty Source: Peters et al. (2001). C. DISCUSSION OF MAIN FINDINGS FROM AFRICA REGIONAL REVIEW The review of selected Africa experiences (Arhin-Tenkorang 2001) emphasized that a common feature of many of the reforms introduced during the past two decades in this region have consisted of copayments to influence utilization patterns and direct out-of-pocket user charges to mobilize much needed additional resources (Vogel 1990). Most of the population currently does not benefit from formal insurance coverage, and government expenditures often do not meet the basic health needs of the poor, let alone the whole population (Abel-Smith and Rawal 1994). These user charges add significantly to the financial hardship of poor households, often fully exposed to the financial risks associated with illness. This has been especially true during recent years, due to the rising incidence and prevalence of HIV/AIDS, tuberculosis, and other communicable diseases. A central premise of the Africa review is that individuals in the informal sector of poor countries cannot access appropriate health care--particularly curative care--at the time of need partially due to lack of adequate insurance coverage (Arhin-Tenkorang 2001). Although preventive measure may have long-term payoffs in improving the overall welfare and productivity of the population, the income shock associated with seeking access to curative and palliative care has become such a great financial burden for the poor that some form of insurance coverage has to be considered an essential part of any serious poverty alleviation strategy. The first section of the paper conceptualizes how scheme performance in terms of risk protection and resource mobilization is influenced by the interaction between several design features and institutional factors. In the absence of risk protection, several African studies demonstrated that poor households often deferred visits to formal health facilities until their illness became quite severe or used ineffective self- 24 medication that was sometimes injurious to their health, leading to both more severe health and financial consequences than had they sought care earlier. Key design features included the methodology, nature, and quality of the data used to determine contribution levels, benefit package, and subsidy level. The argument is that appropriate specifications require: (a) data on the target population's willingness-to-pay (WTP) and ability-to-pay, often not collected or available; (b) data on projected costs of the benefits to be consumed; and (c) operational modalities that facilitate interaction between individuals in an informal environment and in a range of formal organizations. The review concludes that in an informal environment, decisions cannot rely on written such information since the needed data are usually not available in this form. To be effective and affordable, activities undertaking by community financing schemes must be based on simple and directly observable behavior patterns that have a low transaction cost. Key institutional features included the degree of congruence between the scheme's operating rules and the normal behavior patterns of the participating population. It also included the degree of past experience of participating health care providers with third-party payments and contractual arrangements. The review found that these institutional factors had a significant influence on the nature and extent of community participation in any given scheme, as well as the quality of its management and monitoring of performance. The review did not examine other institutional factors such as government regulations and laws governing insurance and health care provision. The second part of the paper proposes the design features of several potential "high population schemes" for the informal sector in Africa and assesses their performance with respect to risk protection and resource mobilization. Potential "high population" schemes examined included the Abota Village Insurance Scheme (Guinea-Bissau), Bwamanda Hospital Insurance Scheme (former Zaïre), Carte d'Assurance Maladie (CAM) program (Burundi), Dangme West Health Insurance Scheme (Ghana), Nkoranza Community Financing Health Insurance Scheme (Ghana), and Community Health Fund (CHF) (Tanzania). These schemes had large target populations and provided a comprehensive range of benefits and geographically accessible care to its members. Key factors influencing enrolment appeared to include: (a) a matching of the premium to the willingness and ability to pay; (b) availability of government subsidies for the poor who cannot afford the basic premium; (c) ready access to basic care for common health problems and emergency care--both geographic proximity and availability of range of basic services appeared to significantly impact enrolment. The final part of the paper presents a set of policy measures that national and international health policy makers may consider implementing to increase the level of risk protection provided for informal sector populations. The financial risk protection and resource mobilization that can be achieved by any given scheme appears to be influenced by the compatibility between the way it is designed and operated with the behavior of the individuals and households from the informal sector that enrol in the schemes. The enrolment rate of a given population with such schemes appears to reflect the target population's willingness to pay (WTP), in turn, closely related to ability to pay (ATP). In most cases, some central government support in the form of fiscal transfers and/or budget allocations is necessary, given the low volume of resources available at low-income levels in poor communities. Schemes that are operated as solidarity-based partnerships with service providers appear to create additional incentives to increase efficiency and accountability. The authors conclude that national government policies, a legal framework, and financial support for these organizations are likely to be a good investment of scarce government resources. The authors emphasize that, in the absence of established best practices in the design of community financing schemes, donor funding, procedures, and regulations supporting community financing through communities, local governments and local NGOs still need further pilot testing to identify the elements that would be needed to expand or to go to scale. 25 D. DISCUSSION OF MAIN FINDINGS FROM MICRO-LEVEL HOUSEHOLD SURVEY ANALYSIS Determinants of Social Inclusion in Community Financing The results are varied from the micro-level household survey analysis in terms of the determinants of social inclusion through community financing. Table 10 presents the determinants that were found statistically significant in the five household surveys (Gumber 2001; Jutting 2001; Ranson 2001; Schneider 2001; and Supakankunti 2001). The key finding from this part of the study include the following (Jakab et al. 2001): Table 10. Statistically Significant Determinants of Inclusion in Community Financing Rwanda Senegal India (1) India (2) Thailand Model Logit Probit Logit Logit Logit Dependent variable Dependent variable Proportion of Proportion of Proportion of Proportion of Proportion of population population population population population enrolled in 1 enrolled in 1 enrolled in enrolled in purchased new of 66 schemes of 4 schemes SEWA- SEWA- health card, insurance insurance continued, dropped out, never purchased Independent variables: individual & household characteristics Income/assets No Yes No No Yes Age No No Yes Yes No Education Yes No No No Yes Gender No No -- -- No Health status No -- Yes Yes Yes Household size Yes No Yes No -- Marital status Yes No No Religion -- Yes -- No -- Distance of household from Yes -- -- -- -- scheme provider Independent variables: community characteristics Community marker for unobservable Yes Yes -- -- -- ch. Solidarity N/A Yes N/A N/A N/A Yes: variable is significant at least at the 10 percent level. No: Variable is not significant. (--) : not included in the particular model · Income and other socioeconomic determinants. In Senegal and Thailand, household income was a significant determinant of being member of a prepayment scheme while in Rwanda and India income was not significant. · Other individual and household characteristics. Health status was included in the analysis of the Rwanda, Thailand, and both India surveys. In all three surveys, the analysis confirmed the presence of adverse selection that characterizes voluntary prepayment schemes. Patients with recent illness episodes or with chronic illnesses are more likely to purchase a prepayment plan. Distance of the household from the provider of the scheme was included in the Rwanda analysis. 26 Households less than 30 minutes from the health facility of the scheme were four times more likely to belong to the prepayment scheme than households living farther away. · Community characteristics. Dummy variables for community characteristics were significant predictors of the probability of enrolling in the prepayment scheme (Senegal and Rwanda). Determinants of Financial Protection in Community Financing The results are varied in terms of the determinants of financial protection through community financing. Table 11 presents the determinants found statistically significant in four of the household surveys. The household survey conducted in Thailand did not permit analysis of the determinants of out-of-pocket payments and was therefore excluded. The key findings from this part of the study include: · Insurance effect. In three of the five household surveys, membership in a community financing scheme was a significant determinant of the probability of using health care and in reducing out- of-pocket payments. This confirms our original hypothesis that even small-scale prepayment and risk pooling reduce financial barriers to health care (Rwanda, Senegal, and India). · Socioeconomic determinants. The analysis indicated that even with insurance low-income remains a significant constraint to health care utilization and ability to pay out-of-pocket payments (Rwanda, Senegal, and India). · Other determinants. Distance from scheme provider was a significant determinant of the likelihood of using health care (Rwanda, and Senegal). 27 Table 11. Summary Findings: Statistically Significant (at Least at 10 Percent) Determinants of Utilization and Out-of-Pocket Expenditure Patterns Rwanda Senegal India (1) India (2) Utilisation OOPs Utilisation OOPs Utilisation OOPs Utilisation OOPs Model Logit Log-linear Logit Log-linear Logit Log-linear Logit Log-linear conditional on conditional on conditional on conditional on (use>0) (use>0) (use>0) (use>0) Dependent variable Dependent variable Proportion Total illness Proportion of Out-of-pocket Proportion of Total annual Proportion of Total annual of sample w/ related out-of- sample w/ at spending of sample reporting direct and sample w/ at out-of-pocket at least one pocket least one hospitalization any health care indirect cost of least one payment for visit to payment per hospitalization use health care use hospitalization use of hospital professional episode of care health care illness for the provider full episode Independent variables: Insurance effect Scheme membership Yes Yes Yes Yes Yes No No Yes Independent variables: Individual & household characteristics Income/assets Only for Only for Yes poorest Only for richest Yes No Yes Only for richest richest quartile terzile quintile quintile Age Yes No No Yes No -- Only for oldest Only for group oldest group Education No No No No No -- No Yes Gender No Yes Yes No No -- -- -- Health status/severity of Only for very illness Yes No -- Yes severe Yes Yes Yes Household size No No -- -- No Only small hh size No Yes Marital status -- -- -- -- No -- No No Religion -- -- -- -- -- -- Yes Yes Distance of household from scheme provider Yes No Yes No -- -- -- -- Note: Other control variables were included in some of the studies but, as they are not discussed in the paper, we did not include them in this table. 28 E. DISCUSSION OF MAIN FINDINGS FROM MACRO-LEVEL CROSS-COUNTRY ANALYSIS A first observation was that most routine national statistical sources do not include data on the share of overall financing that is channeled through either community-based or private health insurance schemes (Carrin and Zeramdini 2001). The analysis therefore had to focus on the extent of collective risk sharing provided at low-income levels through different combinations of general tax revenues and social insurance. The equations have been estimated with the ordinary least squares method, using data for the explanatory variables HEC, EDU, and PHE percent that pertain to the year 1997. The Gini index pertains to specific years, depending upon the country, in 1986­99. In this synthesis paper, we present only the "best" regressions2 in summary Tables 1 and 2. Except for the functional form of the regression for DALE, we present only the results of the logit specification. The estimation results for the basic model presented in summary Table 1 are discussed next. First, concerning the level of health (DALE), the effects of DARS, HEC, and EDU are as expected and are statistically significant at the 1 percent significance level. Second, from the equation for the level of responsiveness (IR), we see that HEC and EDU do not have a statistically significant impact. One major reason is likely to be that the index of responsiveness contains elements of both respect for persons and client orientation and that both are influenced differently by HEC and EDU. For instance, HEC may be important in explaining client orientation, but it may not be when explaining respect for persons. Therefore, when analyzing the determinants of the overall index of responsiveness, the effect of HEC may disappear. Notice, however, that both the coefficients of DARS and DMRS have the expected sign and are statistically significant. Third, the explanatory power of the regression for the index of fair financing (IFFC) is minimal; DARS does not have a statistically significant impact on the IFFC. We submit that the major reason for this unsatisfactory result is the relatively small sample size. Moreover, the sample did not include sufficient data on countries with advanced and with low risk sharing. For instance, the (full sample) data on advanced risk sharing are those of Bulgaria, Jamaica, Kyrgyzstan, Romania, and Russia and do inadequately reflect the experience of high-income countries with either social health insurance or general taxation financing. Fourth, in the equation for the distribution of responsiveness (IRD), the coefficient of DARS is statistically significant. The impact of DSHI is statistically insignificant. Fifth, the results for the index of equality of child survival (IECS) show that both DARS and DMRS have statistically significant impacts. We next present the estimation results for the enlarged model with the Gini index as an explanatory variable in the equations for the distributional measures. The results are presented in Summary Table 2. In the fair financing equation (IFFC), which has very low explanatory power, the coefficient of the Gini index has the anticipated sign but is not statistically significant. The coefficient of DARS is also not statistically significant. 2"Best" according to the adjusted R-squared and/or the Akaike criterion, as well as the theoretical consistency of the model. In addition, we present only the results using restricted samples (these are samples where data points have been deleted from the "full" samples because of uncertainty in the risk-sharing classification) or restricted samples with additional deletion of influential data. Related to the distribution of responsiveness (IRD), the result shows significant impacts of both DARS and DMRS, as well as of the Gini index. All coefficients have the expected sign. One can conclude that these risk-sharing arrangements are efficient in counterbalancing the overall effect of income inequality. A threshold for the Gini indices can be computed, indicating the value above which risk sharing is no longer able to counteract the effect of overall income inequality. In the case of a country with an advanced risk-sharing scheme, the threshold value is 57.9. In the case of medium risk-sharing schemes, the threshold is 26.3. From these estimates, we can infer that advanced risk-sharing schemes are more effective in counteracting the effects of overall income inequality in society. For example, let us assume that a country has a Gini coefficient of 35. If this country has an advanced risk-sharing scheme, its effect will outweigh the impact of income equality: the combined effect will be +0.8588. However, if the country has a medium-risk sharing arrangement, the combined effect will be ­0.3252. In the regression result related to the inequality of child survival (IECS), the sign of the Gini coefficients is against our expectations. Surprisingly, the Gini coefficient is also statistically significant at the 10 percent level. The coefficient of DARS has the anticipated sign, however, and is statistically significant at the 1 percent level. Inclusion of the interaction variables with PHE percent in the equations did not result in a general improvement of the estimation results. For instance, in a number of cases, the coefficients of DARS have the correct sign but are statistically insignificant. In other instances, the coefficient of DARS has a negative sign. Further estimations were done with transformed interaction variables. In the case of the interaction between DARS and PHE percent, the variable constructed was DARS*(PHE percent­0.5). The coefficient associated with this variable reveals the impact of the difference between PHE percent and a threshold of 50 percent. The results for IR, IFFC, IRD, and IECS are not satisfactory: the coefficient of the new interaction variable has the wrong sign, is not statistically significant, or both. Only in the case of DALE did we obtain a satisfactory result: both the coefficients of DARS and the interaction variable have the expected sign and are statistically significant. This result is presented in Summary Table 2. In other words, for advanced risk-sharing systems with a PHE percent above 50, the level of PHE percent reinforces the "average" effect of DARS. For instance, in the case of Oman with a PHE percent of 63.31, the combined impact of DARS and DARS*(PHE percent ­ 0.50) becomes ­0.2694. For countries with a PHE percent below 50, (Chile, Republic of Korea, Brunei Darussalam, and United Arab Emirates), the initial effect of DARS is weakened. For instance, for Chile with a PHE percent of 40.10, the combined effect of DARS and DARS*(PHE percent ­ 0.50) on the dependent variable becomes ­0.1637. Key conclusions can be drawn from the various estimates. A first conclusion is that the extent of advanced risk sharing, as measured by the dummy variable DARS, is significant in the equations for four of the five goal measurements. No impact could be found in the case of the index of fair financing, but we submit this is due to the small sample size. In addition, in at least two of these measurements (level of responsiveness, distribution of health), the variable DMRS also has been shown to have a statistically significant impact. Second, when enlarging the set of explanatory variables in the models for the distributional measures with the Gini index, DARS remains statistically significant in the equations for IRD and IECS. In addition, DMRS has a statistically significant impact in the equations for IRD. An additional interpretation emerges from the results, namely that risk sharing corrects for, or may even outweigh, the negative effect of overall income inequality on the fair financing index and the index of distribution of responsiveness. Third, using interaction terms with PHE percent leads to plausible results for DALE only: the level of PHE percent reinforces the average positive effect of advanced risk sharing. 30 An analysis with preliminary updated data was also undertaken; since publication of the World Health Report 2000, WHO has developed updated estimates for the level (HEC) and share of public health expenditure in total health expenditure (PHE percent). When using updated data for HEC in the equations for DALE and IR, similar results to those presented here are obtained (in terms of explanatory power, sign, and statistical significance of coefficients). The use of the updated PHE percent does not significantly change the estimates for the equations with the interaction terms. Estimates of the index of fair financing (IFFC) were also obtained for an additional 30 countries. Reestimation of the equations, using an enlarged sample of 50, now leads to two interesting results: (i) the advanced risk-sharing dummy variable DARS exerts a statistically significant effect on the fair financing index; (ii) the Gini index has a statistically significant impact on IFFC but is counterbalanced by a health financing system characterized by advanced risk sharing. These preliminary results prove to be more in line with those obtained for the other distributional measures. IV. CONCLUSIONS AND RECOMMENDATIONS Most community financing schemes have evolved in the context of severe economic constraints, political instability, and lack of good governance. Usually government taxation capacity is weak, formal mechanisms of social protection for vulnerable populations absent, and government oversight of the informal health sector lacking. In this context of extreme public sector failure, community involvement in the financing of health care provides a critical albeit insufficient first step in the long march toward improved access to health care for the poor and social protection against the cost of illness. It should be regarded as a complement to--not as a substitute for--strong government involvement in health care financing and risk management related to the cost of illness. Based on an extensive survey of the literature, the main strengths of community financing schemes are the degree of outreach penetration achieved through community participation, their contribution to financial protection against illness, and increase in access to health care by low-income rural and informal sector workers. Their main weaknesses are the low volume of revenues that can be mobilized from poor communities, the frequent exclusion of the poorest from participation in such schemes without some form of subsidy, the small size of the risk pool, the limited management capacity that exists in rural and low- income contexts, and their isolation from the more comprehensive benefits that are often available through more formal health financing mechanisms and provider networks. The results of the macro-level cross-country analysis presented in this Report give empirical support to the hypothesis that risk sharing in health financing matters in terms of its impact on both the level and distribution of health, financial fairness, and responsiveness indicators. The results even suggested that risk sharing corrects for, and may outweigh, the negative effect of overall income inequality, suggesting that financial protection against the cost of illness may be a more effective poverty alleviation strategy in some settings than direct income support. The results of the micro-level household data analysis indicated that prepayment and risk sharing through community involvement in health care financing--no matter how small--increases access by poor populations to basic health services and protects them to a limited extent against the impoverishing effects of illness. Community involvement alone is not sufficient in preventing social exclusion since the very poorest often do not participate fully in these schemes. However, the study provided evidence that this constraint in reaching the poorest could be overcome through well-targeted design features and implementation arrangements. 31 The Asia regional review supported many of these conclusions. In particular, the review emphasized that, although income is a key constraint to participation by the very poorest, even they are often willing to participate if their contributions are supplemented by a government subsidy and if the benefits they receive provide access to quality services that address their most frequent health problems. In the context of extreme resource constraints, this creates a tension or trade-off between prepayment for basic services and the need for insurance coverage for rarer, more expensive, and life threatening events that may only happen once or twice in a lifetime. This highlights an area of market failure relating to voluntary community involvement in health care financing that needs to be addressed by appropriate government policies since it is precisely during hospitalization that many of the poor become even more impoverished. Figure 8. Stages of Financial Protection Renewed policy commitment BROAD INSURANCE COVERAGE Optimised subsidy of low- Advocacy and income by high income consumer protection INSURANCE POOL funding and CONSOLIDATION reinsurance Inter-pool subsidies and consolidation policies ESTABLISHED INSURANCE Framework for pool POOLS management and interactions Set-up funding and EVOLUTION OF reinsurance COMMUNITY- BASED Capacity building and Commitment to dissociation of POOLS technical support contribution from utilisation Evidence-based advocacy NATIONAL DONOR POLICY THRUSTS DOMINANCE OF OUT- POLICY THRUSTS POCKET PAYMENTS Source: Adapted from Arhin-Tenkorang (2001). More rigorous research is still needed on understanding the institutional strengths and weakness of community involvement in health care financing, and in monitoring and evaluating their impact on financial protection, increasing access to needed health care, and combating social exclusion of the poor. Yet the research for this report points to five key policies available to governments to improve the 32 effectiveness and sustainability of community financing: (a) increased and well-targeted subsidies to pay the premiums of low-income populations; (b) use of insurance to protect against expenditure fluctuations and use of reinsurance to enlarge the effective size of small risk pools; (c) use of effective prevention and case-management techniques to limit expenditure fluctuations; (d) technical support to strengthen the management capacity of local schemes; and (e) establishment and strengthening of links with the formal financing and provider networks. 33 34 REFERENCES Abel-Smith B. "The rise and Decline of the Early HMOs: Some International Experiences. Milbank Quarterly. 1988;66(4):694-719. Abel-Smith B, Rawal P. Employer's willingness to pay: the case for compulsory health insurance in Tanzania. Health Policy and Planning 1994; 9(4):409-18. ADB Finance for the Poor: Microfinance Development Strategy. Manilla: Asian Development Bank. 2000. Arhin-Tenkorang, DC. The Health Card Insurance Scheme in Burundi: A social asset or a non-viable venture? Social Science and Medicine, 1994; 39 (6):861-870. Arhin-Tenkorang, DC. Rural Health Insurance: A Viable Alternative to User Fees. London School of Hygiene and Tropical Medicine, Discussion Paper. 1995. Arhin-Tenkorang, DC. Mobilizing Resources for Health: The Case for User Fees Re-visited. Report Submitted to Working Group 3 of the Commission on Macroeconomics and Health, Jeffrey D. Sachs (Chairman), Geneva: WHO, 2000. Arhin-Tenkorang, DC. Health Insurance for the Informal Sector in Africa: Design Features, Risk Protection and Resource Mobilization. Report Submitted to Working Group 3 of the Commission on Macroeconomics and Health, Jeffrey D. Sachs (Chairman), Geneva: WHO, 2001. Arrow KW. Uncertainty and the welfare economics of medical care. American Economic Review 1963; 53(5):940-73. Atim C. Contribution of mutual health to financing, delivery, and access to health care. Synthesis of research in nine West and Central African countries. Bethesda, Maryland: Abt Associates; (Partnerships for Health Reform Project, Technical Report No. 18). 1998 Atim C.. "Social movements and health insurance: A critical evaluation of voluntary, non-profit insurance schemes with case studies from Ghana and Cameroon." Social Science and Medicine 1999; 48(7):881- 896. Atkinson AB, Stiglitz JE. Lectures on public economics. Maidenhead: McGraw-Hill. 1980. Barer LM, Getzen TE, Stoddart GL, editors. Health, health care and health economics: perspectives on distribution. Chichester, West Sussex, England: John Wiley and Sons. 1998. Bator F. The anatomy of market failure. Quarterly Journal of Economics 1958; 72(3):351-79 Bennett S, Creese A, Monasch R. Health insurance schemes for people outside formal sector employment. Geneva: World Health Organization; (ARA Paper No. 16). 1998. Brown W, Churchill C. Insurance provision in low-income communities. Part II. Initial lessons from microinsurance experiments for the poor. Microenterprise best practices. Bethesda: Development Alternatives Inc; 2000. 35 Carrin G, Desmet M, Basaza R. Social health insurance development in low-income developing countries: new roles for government and non-profit health insurance organisations. In: Building social security: the challenge for privatization. Geneva: International Social Security Association. 2001. Carrin, G.; R. Zeramdini; P. Musgrove; J-P. Poullier; N. Valentine; and K. Xu. 2001. The Impact of the Degree of Risk-sharing in Health Financing on Health System Attainment. Report Submitted to Working Group 3 of the Commission on Macroeconomics and Health, Jeffrey D. Sachs (Chairman), Geneva: WHO, 2001. Claeson, M., Griffin, C.C., Johnston, T.A,, McLachlan, M., Wagstaff, A., Yazbeck, AS, 2001. Health, Nutrition, and Population Sourcebook for the Poverty Reduction Strategy Paper. Discussion Paper. Washington: World Bank. 2001 Criel B, Van der Stuyft P, Van Lerbergher W. The Bwamanda hospital insurance scheme: effective for whom? A study of its impact on hospitalisation and utilisation patterns. Social Science and Medicine 1999; 48(7):879-911. Dave P. Community and self-financing in voluntary health programmes in India. Health Policy and Planning 1991; 6(1):20-31. DeRoeck D, Knowles J, Wittenberg,T, Raney L, Cordova P. Rural health services at Seguridad Social Campesino facilities: analyses of facility and household surveys. Bethesda, Maryland: Abt Associates Inc., Health Financing and Sustainability Project (Partnerships for Health Reform Project, Technical Report No. 13). 1996. Desmet A, Chowdhury AQ, Islam K. Md. The potential for social mobilization in Bangladesh: the organization and functioning of two health insurance schemes. Social Science and Medicine 1999; 48: 925-938. Diop F, Yazbeck A, Bitran R. "The impact of alternative cost recovery schemes on access and equity in Niger." Health Policy and Planning 1995; 10(3):223-40. Dordick G. Something Left to Lose: Personal Relations and Survival among New York's Homeless. Philadelphia, Pa: Temple University Press. 1997. Dror D, Jacquier C. Microinsurance: extending health insurance to the excluded. International Social Security Review 1999; 52(1):71-97. Dror, D, AS Preker, and M Jakab. "Role of Communities in Combating Social Exclusion" In: Dror, D and AS Preker, (eds) Social Reinsurance: A New Approach To Sustainable Community Health Financing, Washington: World Bank/ILO, 2002. Duan N, Manning WG JR, Morris CM and Newhouse JP. A Comparison of Alternative Models for the Demand for Medical Care, report R-2754-HHS. The Rand Corporation: Santa Monica, California, 1982. Evans P. "The state as Problem and solution: Predation, Embedded Autonomy, and Structural Change." In Stephan Haggard and Robert Kaufman eds., The Politics of Economic Adjustment; International Constraints, Distributive Conflicts, and the State. Princeton: NJ: Princeton University Press. 1992. Evans P. Embedded Autonomy: States and Industrial Transformation. Princeton, NJ: Princeton University Press. 1995. 36 Evans P. "Government Action, Social Capital and Development: Reviewing the Evidence on Synergy." World Development 1996; 24(6): 1119-32. Evans RG. Strained mercy. Toronto: Butterworth. 1984. Foster G. Community development and primary health care: their conceptual similarities. Medical Anthropology 1982; 6:183-95. Granovetter M. "The Strength of Weak Ties." American Journal of Sociology 1973: 78: 1360-80. Guhan S. Social security options for developing countries. International Labour Review 1994; 33(1):35- 53. Gumber A, Kulkarni V. Health insurance for informal sector: Case study of Gujarat. Economic and Political Weekly, 2000: 3607-3613. Gumber A. 2001. Hedging the Health of the Poor: The Case for Community Financing in India. HNP Discussion Paper, Washington: World Bank. Gwatkin D. "Poverty and inequalities in health within developing countries: filling the information gap." In: Leon D, Walt G, eds. Poverty, inequality, and health: an international perspective. Oxford: Oxford University Press. 2001: 217-46. Harding, A and AS Preker, A Framework for Understanding Organizational Reforms in the Hospital Sector, HNP Discussion Paper, Washington: World Bank, 2001. Hsiao WC. Unmet health needs of two billion: is community financing a solution? Report Submitted to Working Group 3 of the Commission on Macroeconomics and Health, Jeffrey D. Sachs (Chairman), Geneva: WHO, 2001. ILO. World Labour Report: Income Security and Social Protection in a Changing World. Geneva: ILO. 2000a. ILO. Mutuelles de santé en Afrique: Charactéristiques et mise en place: Manuel de formateurs. Geneva: ILO, Programme Stratégies et Technique contre l'Exclusion sociale et la Pauvreté, Département de Sécurité Sociale (STEP). 2000b. ILO. Mutuelles de santé et associations de microentrepreneurs--Guide. Geneva: ILO. Programme Stratégies et Technique contre l'Exclusion Sociale et la Pauvreté, Département de Sécurité Sociale (STEP), Small Enterprise Development Programme (SEED). 2001. Jakab M, Krishnan C. Community involvement in health care: a survey of the literature on the impact, strengths and weaknesses. Report Submitted to Working Group 3 of the Commission on Macroeconomics and Health, Jeffrey D. Sachs (Chairman), Geneva: WHO, 2001. Jakab M, Preker, AS, Krishnan C, Schneider P, Diop F, Jutting J, Gumber A, Ranson K, Supakankunti S. Social inclusion and financial protection through community financing: initial results from five household surveys. Report Submitted to Working Group 3 of the Commission on Macroeconomics and Health, Jeffrey D. Sachs (Chairman), Geneva: WHO, 2001. Jütting J. Do mutual health insurance schemes improve the access to health care? Preliminary results from a household survey in rural Senegal. Paper presented at the International Conference on "Health 37 Systems Financing in Low-Income African and Asian Countries", Centre d`Etudes et de Recherches sur le Développement International (CERDI), Clermont-Ferrand, France Nov./Dec. 2000. Jutting J. 2001. The Impact of Health Insurance on the Access to Health Care and Financial Protection in Rural Areas of Senegal. HNP Discussion Paper, Washington: World Bank. Litvack JI, Bodart CI. User fees plus quality equals improved access to health care: results of a field experiment in Cameroon. Social Science and Medicine 1992; 37(3):369-83. Manning WG, Newhouse JP, Duan N, Keeler EB, Leibowitz A, Marquis MS. et al., "Health Insurance and the Demand for Medical Care: Evidence from a Randomized Experiment." American Economic Review 1987; 77(3): 251-277. McPake B, Hanson K, Mills A. Community financing of health care in Africa: an evaluation of the Bamako Initiative. Social Science and Medicine 1993; 3(11):1383-95. Midgley J, Tracey MB. Challenges to social security: an international exploration. Westport, CT: Auburn House. 1996. Musau S. Community-based health insurance: experiences and lessons learned from East and Southern Africa. Bethesda, Maryland: Abt Associates; (Partnerships for Health Reform Project, Technical Report No. 34). 1999. Musgrave RA, Musgrave PB. Public finance in theory and practice. New York: McGraw-Hill. 1984. Narayan D. Bonds and Bridges: Social Capital and Poverty. Policy Research Working Paper 2167. World Bank, Poverty Reduction and Economic Management Network, Washington, DC. 1999. Navarro VA. Critique of the ideological and political position of the Brandt Report and the Alma Ata declaration. International Journal of Health Services 1984; 2(14):159-72. North DC. Institutions, Institutional Change, and Economic Performance. New York: Cambridge University Press. 1990. Otero M, Rhyne E, eds. The new world of microenterprise finance: building healthy financial institutions for the poor. West Hartford, CT: Kumarian Press. 1994. Peters D, Yazbeck A, Ramana GNV, Sharma R. India--raising the sights: better health systems for India's poor. Washington, DC: World Bank. 2001. Pradhan M, Prescott N. Social risk management for medical care in Indonesia Paper presented at the International Conference on "Health Systems Financing in Low-Income African and Asian Countries", Centre d`Etudes et de Recherches sur le Développement International (CERDI), Clermont-Ferrand, France Nov./Dec. 2000,.Paris. 2000. Platteau JP "Behind the Market Stage Where Real Societies Exist" (Parts I and II). Journal of Development Studies 1994: 30: 533-77 and 753-817. 1994. Preker AS. "The Introduction of Universal Access to Health Care in the OECD: Lessons for Development Countries, in Nitayarumphong S, and Mills A. Achieving Universal Coverage of Health Care, Bangkok: Ministry of Public Health, 1998, p 103-124. 38 Preker AS, J Langenbrunner, and E Suzuki. September. The Global Expenditure Gap: Securing Financial Protection and Access to Health Care for the Poor. Report Submitted to Working Group 3 of the Commission on Macroeconomics and Health. Jeffrey D. Sachs (Chairman). Geneva: WHO. 2001 Preker AS, Harding H, and Girishankar N. "Private Participation in Supporting the Social Contract in Health: New Insights from Institutional Economics." In Ron A, and X Scheil-Adlung (eds), Recent Health Policy Innovations in Social Security, International Social Security Series, Volume 5, London: Transaction Publishers, 2001. p. 209-252. Preker, AS, J Langenbrunner, M Jakab, and C Baeza. Resource Allocation and Purchasing (RAP) Arrangements that Benefit the Poor and Excluded Groups, HNP Discussion Paper, Washington: World Bank, 2001. Preker, AS, J Langenbrunner, and M Jakab, "Development Challenges in Health Care Financing," In: Dror, D and AS Preker, (eds) Social Reinsurance: A New Approach To Sustainable Community Health Financing, Washington: World Bank/ILO, 2002. Preker AS, G Carrrin, D Dror, M Jakab, W Hiao, and D Arhin-Tenkorang. February 2002. "Effectiveness of Community Health Financing in Meeting the Cost of Illness." Bulletin of the World Health Organization. February 2002a. 8 (2) (143-150). . Preker, AS, G Carrrin, D Dror, M Jakab, W Hiao, and D Arhin-Tenkorang, (eds.) Chapter 4, "Alternative Ways of Financing Health in the Informal Sector," in Jeffrey D. Sachs, Alan Tait, and Kwesi Botchwey (Chairmen), Report of Working Group 3 of the Commission on Macroeconomics and Health, Geneva: WHO, 2002b. Ranson K. 2001. The impact of SEWA's Medical Insurance Fund on Hospital Utilization and Expenditure in India. HNP Discussion Paper, Washington: World Bank. Saltman RB, Ferroussier-Davis O. "The concept of stewardship in health policy". Bulletin of the World Health Organization. 2000;78(6):732-9. Schieber G, Maeda A. A curmudgeon's guide to health care financing in developing countries. In: Schieber G, editor. Innovations in health care financing. Washington DC: World Bank. (Proceedings of a World Bank Conference, 10-11 March 1997). 1997 Schneider P, Diop F. 2001. Synopsis of Results on the Impact of Community-Based Health Insurance (CBHI) on Financial Accessibility to Health Care in Rwanda. HNP Discussion Paper, Washington: World Bank. Soucat A, Gandaho T, Levy-Bruhl D, de Bethune X, Alihonou E, Ortiz C, Gbedonou P, Adovohekpe P, Camara O, Ndiaye JM, Dieng B, Knippenberg R. "Health seeking behavior and household expenditures in Benin and Guinea: the equity implications of the Bamako Initiative." International Journal of Health Planning and Management 1997; 12(S1):137-63. Supakankunti S. Future prospects of voluntary health insurance in Thailand. (Takemi Research Paper No. 13. Takemi Program in International Health. Harvard School of Public Health). Cambridge,MA: Harvard University. 1997. Supakankunti S. 2001. Determinants for Demand for Health Card in Thailand. HNP Discussion Paper, Washington: World Bank. 39 Van Doorslaer E, Wagstaff A, Rutten F, editors. Equity in the finance and delivery of health care: an international perspective. Oxford: Oxford Medical Publications. 1993. Van Ginneken W. Social security for the informal sector: new challenges for the developing countries. International Social Security Review 1999; 52(1):49-69. Vogel RJ. Health insurance in sub-Saharan Africa. Washington, DC: World Bank; (African Technical Department, Working Paper Series). 1990: 476. Wagstaff AN, Watanabe N, van Doorslaer E. Impoverishment, insurance, and health care payments. HNP Discussion Paper. Washington DC: World Bank. 2001 Woolcock M, ":Social Capital and Economic Development: Toward a Theoretical Synthesis and Policy Framework;" Theory and Society 1998; 27:151-208. Woolcock M, Narayan D. "Social Capital: Implications for Development Theory, Research, and Policy." World Bank Research Observer 2000 15(2): 225-49. World Bank. Sector strategy for HNP. Washington, DC: World Bank. 1997. World Bank. World development report 1993: investing in health. New York: Oxford University Press. 1993. World Bank. World development report. Workers in an integrating world. Washington DC: World Bank. 1995. World health report 2000. Health systems: measuring performance. Geneva: World Health Organization. 2000. Yip, W. and P. Berman. "Targeted Health Insurance in a Low-income Country and Its Impact on Access and Equity in Access: Egypt's School Health Insurance". Health Economics. 2001: 10(2): 207-20. Zeller M, Sharma M. Many borrow, more save, and all insure: implications for food and microfinance policy. Food Policy 2000; 25(2):143-67. Ziemek S, Jütting J. Mutual insurance schemes and social protection. (STEP Research Group on Civil Society and Social Economy). Geneva: International Labour Organization; 2000. 40 APPENDIX A Summary Table 1--Estimation Results1 for the Basic Models Explanatory variables 2 3 2 4 5 DALE IR IFFC IRD IECS Ln (80­ DALE) (Logit) (Logit) (Logit) (Logit) Constant 4.9423 ­0.4896 2.2874 1.6327 0.2798 (0.3328) (0.2160) (0.2786) (0.4507) (0.2038) (14.8493) (­2.2663) (8.2099) (3.6228) (1.3329) HEC ­0.1919 0.0000 (0.0197) (0.0003) (­9.7498) (0.1150) EDU ­0.2141 0.0032 (0.0834) (0.0026) (­2.5684) (1.2540) DARS ­0.2963 0.7244 ­0.1146 4.2257 0.6269 (0.0654) (0.2244) (0.6072) (0.8228) (0.3868) (­4.5321) (3.2275) (­0.1888) (5.1355) (17.1343) DSHI ­0.2521 ­1.4049 (0.1987) (0.9107) (­1.2688) (­1.5427) DMRS 0.2673 0.7217 0.6203 (0.1148) (0.5355) (0.2497) (2.3294) (1.3478) (2.4845) DMRS1 ­0.1079 (0.4607) DMRS2 (­0.2343) ­0.6458 (0.3995) (­1.6165) R­squared 0.7874 0.5678 0.0021 0.5749 0.8778 Adjusted R­squared 0.7821 0.4597 ­0.0566 0.5276 0.8671 S.E. of regression 0.2639 0.2134 1.0791 1.1924 0.7350 Ak. Info criterion 0.2049 ­0.0525 3.0894 3.3097 2.3149 Sample size 124 26 19 31 51 1 The first and second coefficients in the brackets refer to the standard error and t-statistic, respectively. 2 Restricted samples. 3 Bulgaria is excluded from the sample. 4 Chile and Poland are excluded from the full sample. 5 Uzbekistan is excluded from the restricted sample. 41 Summary Table 2 - Estimation Results1 for the Enlarged Models 2 2 2 IFFC IRD IECS DALE (Logit) ( Logit) Logit) Ln(80­ DALE)) Explanatory variables Constant 2.8260 3.0610 ­0.7471 4.9446 (1.3698) (0.7956) (0.9164) (0.3306) (2.0630) (3.8539) (­0.8153) (14.9580) Gini ­0.0119 ­0.0375 0.0355 (0.0296) (0.0180) (0.0206) (­0.4020) (­2.0853) (­0.8153) DARS ­0.2568 2.1713 5.3537 ­0.2088 (0.7162) (0.5222) (0.5531) (0.0843) (­0.3586) (4.1577) (9.6789) (­2.4774) DARS*[PHE percent­ 0.5] ­0.4556 (0.2798) (­1.6284) DMRS 0.9873 (0.4637) (2.1291) HEC ­0.1898 (0.0196) (14.9580) EDU ­0.2166 (0.0828) (­2.6155) R­squared 0.0121 0.5191 0.7053 0.7920 Adjusted R­squared ­0.1114 0.4590 0.6906 0.7850 S.E. of regression 1.1067 0.9320 1.1912 0.2621 Ak. Info criterion 3.1846 2.8286 3.2550 0.1990 Sample size 19 28 43 124 1 The first and second coefficients in the brackets refer to the standard error and t-statistic, respectively. 2 Restricted samples. 42 About this series... This series is produced by the Health, Nutrition, and Population Family (HNP) of the World Bank's Human Development Network. The papers in this series aim to provide a vehicle for publishing preliminary and unpolished results on HNP topics to encourage discussion and debate. 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. Citation and the use of material presented in this series should take into account this provisional character. For free copies of papers in this series please contact the individual authors whose name appears on the paper. Enquiries about the series and submissions should be made directly to the Editor in Chief Alexander S. Preker (apreker@worldbank.org) or HNP Advisory Service (healthpop@worldbank.org, tel 202 473-2256, fax 202 522-3234). For more information, see also www.worldbank.org/hnppublications. THE WORLD BANK 1818 H Street, NW Washington, DC USA 20433 Telephone: 202 477 1234 Facsimile: 202 477 6391 Internet: www.worldbank.org E-mail: feedback@worldbank.org