Z ~~25987 CH~~I' -O __S W ~ S2II MR O March 2003 ________________ WORLD BANK INSTITUTE rton Promoting knowledge and leaming fore better world T H E W O R L D B A N K Social Safety Net Primer Series Waivers and Exemptions for Health Services in Developing Countries Ricardo Bitran and Ursula Giedion Bitran & Associates March 2003 WOR[LD BANK INSTITUTE T E -ia. Promohng knowledge and learning for a better world rot tion T H E W O R L D B A N I 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. Social Safety Net Primer Series The World Bank Social Safety Nets Primer is intended to provide a practical resource for those engaged in the design and implementation of safety net programs around the world. Readers will find information on good practices for a variety of types of interventions, country contexts, themes and target groups, as well as current thinking of specialists and practitioners on the role of social safety nets in the broader development agenda. Primer papers are designed to reflect a high standard of quality as well as a degree of consensus among the World Bank safety nets team and general practitioners on good practice and policy. Primer topics are initially reviewed by a steering committee composed of both World Bank and outside specialists, and draft papers are subject to peer review for quality control. Yet the format of the series is flexible enough to reflect important developments in the field in a timely fashion. The primer series contributes to the teaching materials covered in the annual Social Safety Nets course offered in Washington DC as well as various other Bank-sponsored courses. The Social Safety Nets Primer and the annual course are jointly supported by the Social Protection unit of the Human Development Network and by the World Bank Institute. The World Bank Institute also offers customized regional courses through Distance Learning on a regular basis. For more information on the primer paper series and papers on other safety nets topics, please contact the Social Protection Advisory Service; telephone (202) 458-5267; fax (202) 614-0471; email: socialprotection@worldbank.org. Copies of related safety nets papers, including the Social Safety Nets Primer series, are available in electronic form at www.worldbank.org/safetynets. The website also contains translated versions of the papers as they become available. An ambitious translation plan is underway (especially for Spanish and French, some in Russian). For more information about WBI courses on social safety nets, please visit the website www.worldbank.org/wbi/socialsafetynets. Recent and Forthcoming Papers in the Safety Nets Primer as of August 20021 Theme Author Program Interventions Cash transfers Tabor Food related programs Rogers and Coates Price and tax subsidies Alderman Fee waivers in health Bitran and Giedion Fee waivers in housing Katsura and Romanik Public works Subbarao Micro credit and informal insurance Sharma and Morduch Cross-cutting Issues Overview Grosh, Blomquist and Ouerghi Institutions de Neubourg Targeting Coady, Grosh and Hoddinott Evaluation Blomquist Political Economy Graham Gender Ezemenari, Chaudhury and Owens Community Based Targeting Conning and Kevane Country Setting/Target Grozup Very Low Income Countries Smith and Subbarao Transition Economies Fox Non-contributory pensions Grosh and Schwarz I Papers may be added or deleted from the series from time to time Abstract In response to shortages in public budgets for government health services, many developing countries around the world have adopted formal or informal systems of user fees for health care. In most countries user fee proceeds seldom represent more than 15 percent of total costs in hospitals and health centers, but they tend to account for a significant share of the resources required to pay for non-personnel costs. The problem with user fees is that the lack of provisions to confer partial or full waivers to the poor often results in inequity in access to medical care. The dilemma, then, is how to make a much needed system of user fees compatible with the goal of preserving equitable access to services. Different countries have tried different approaches. Those which have carefully designed and implemented waiver systems (e.g., Thailand and Indonesia) have had much greater success in terms of benefits incidence than countries that have improvised such systems (Ghana, Kenya, Zimbabwe). Key to the success of a waiver system is its financing. Systems that compensate providers for the revenue forgone from granting exemptions (Thailand, Indonesia, and Cambodia) have been more successful than those who expect the provider to absorb the cost of exemptions (Kenya). Where waiver systems exist, performance will improve with the timeliness of the reimbursement. Other success factors include the widespread dissemination of information among potential beneficiaries about waiver availability and procedures; the awarding of financial support to poor patients for non-fee costs of care, such as food and transportation (as in Cambodia); and the existence of clear criteria for the granting of waivers, thereby reducing confusion and ambiguity among those responsible for managing the system and among potential recipients. Those facing the task of adopting a system of waivers face multiple design options. These include the following, among others: should exemptions be granted to whole groups or on the basis of individual targeting (the review finds that most systems are based on the latter)? Should waivers or exemptions be permanent or temporary? How frequently should eligibility be reassessed? Should waiver eligibility be determined ex-ante, in the household, or when individuals seek care in the facility? The review examines various approaches taken by countries, but assessing their relative practical merits is difficult, as the evidence is scattered and mixed. iv Acknowledgements This paper was made possible thanks to the contributions and support of many individuals. Margaret Grosh commissioned the work, made insightful remarks on an earlier draft, and provided needed links with country experts. Marty Makinen, John Blomquist, and David Gwatkin also offered critical input on an earlier version. William Newbrander, David Collins and Lucy Gilson kindly made their work on this subject available, Ensuring Equal Access to Health Services (Newbrander et al., 2000), a must-read for those interested in the issue of poor protection mechanisms under cost recovery in health. The authors are grateful to all those who provided first hand country information and who facilitated documents on user fees and poor protection mechanisms: Helen Dzikunu, Randy Ellis, Liliana Escobar, Wim Hardeman, James Knowles, Keese Kosterman, Tom McGuire, Wasunna Owino, Menno Pradhan, Gonzalo Sanhueza, and Robert Sparrow. Finally, they thank Alex Preker, Maria Luisa Escobar, Paul Shaw, Tim Johnston, Jerry La Forgia and all the other participants in the World Bank Seminar on this subject in June 2002. v Table of Contents Abstract ............. . . iv Table of Contents ............. vi List of Tables ............. . . viii List of Figures ............... ix Acronyms ............... x I User fees for health services: concepts . . . . I 1.1 Introduction .. 1.2 The emergence of user fees .. 1.3 The policy debate about user fees .. 2 1.4 Equity in health.. 7 1.5 The rationale for waivers and exemptions .. 7 2 Mitigating the effect of fees: Identifying and protecting the poor .... 11 2.1 Identification of beneficiaries of protection mechanisms . .11 (a) Individual identification ................................ . ....................................................... 11 (b) Identification based on group characteristics ............................... .................. ................ ..... 12 (c) Self-identih cation ................... . ...................................................................................... 12 (d) Self-selection by type of service ........................................................................................ 12 2.2 Protection mechanisms: Waivers, exemptions, and design features .. 16 (a) Waivers 16 (b) Exemptions ....... ........................................... .................. ... ...................................... 17 (c) Waivers and exemptions combined ..18 (d) Design features for waivers and exemptions ........1.8.................. ...................................... 18 Waivers: Design and implementation issues ............................... 20 2.3 Performance of protection mechanisms ................................ 24 (a) Errors and accuracy in beneficiary identification .24 (b) Accuracy-cost tradeoff in beneficiary identification .25 3 Empirical evidence on protection mechanisms ..28 3.1 Introduction .28 3.2 Kenya .29 (a) Context 29 (b) User-fee policy and its impact on the poor .29 (c) Protection mechanisms ................3................................. ................ 30 (d) Results 32 (e) Lessons learned ................. 32 3.3 Cambodia ............... 33 (a) Background ................ ............................... ............... 33 (b) User fee policy. ............... 33 (c) Protection mechanisms ............... 34 (d) Results 37 (e) Lessons leamed ................. 37 3.4 Ghana ............... 38 (a) Background . ................ ..................................... ................ 38 (b) User-Fee policy ................. 39 (c) Protection mechanisms. ................ 39 (d) Results 39 (e) Protection mechanisms ................................. 40 (fl Lessons learned ................................ 43 3.5 Zimbabwe ............................ 44 (a) Background ................................. 44 (b) User-fee policy. ................................ 44 (c) Protection mechanisms. ................................................... ............... 47 (d) Results 48 (e) Lessons learned ................................ 49 3.6 Indonesia: The Kartu Sehat Program ............................. 49 (a) Background ......... ................... ........................................................... ............4.................. ... 49 (b) User-fee policy ........... ...............................5............................................. 50 (c) Protection policy ............................................. 50 vi (d) Results 51 (e) Lessons learned ........................................ 53 3.7 Thailand's Low Income Card Scheme (LICS) ................................... 53 (a) Context 53 (b) User Fee Policy.......................... 54 (c) Protection mechanism ...................................... .................. 55 (d) Results 57 (e) Lessons learned ......................... .59 3.8 Chile's Nacional Health Fund ...................... 60 (a) Context 60 (b) User fee policy ..61 (c) Protection mechanisms ..62 (d) Results 62 (e) Lessons learned ..65 4 Lessons and best practice . . .66 4.1 Performance of waiver and exemption systems . .66 (a) Performance monitoring and evaluation ..66 (b) Targeting methods used ........... ........... .......... .... ............ 66 4.2 Financing of waivers and exemptions and design and implementation features.69 (a) Funding 69 (b) Health benefits covered ..69 (c) Existence and clarity of national policy of waivers and exemptions . . 69 (d) Taking into account the multiple dimensions of vulnerability of the poor . .70 (e) Assigning responsibility for determination of eligibility ..70 (f) Updating fees and income eligibility thresholds ..71 (g) Institutional aspects ..71 4.3 Design and implementation features influencing the supply of waivers and exemptions ..73 (a) Provider compensation ..73 (b) Timeliness of compensation .....................................................7.......3...............3....... ............. 73 (c) Harmonizing the incentives created by different payment mechanisms . ...... ......... 73 4.4 Design and implementation features influencing the demand for waivers and exemptions .......................................... 76 (a) Disseminating pro-poor protection policies and mechanisms ............................................. 76 4.5 Conclusion .......................................... 78 Bibliography .............. 79 vii List of Tables Table 1 Requirements to obtain health services waiver: example from Latin America ............. ... 21 Table 2 Waivers and exemptions: design and implementation issues ...... .................................. 22 Table 3 Monitoring and evaluation of waiver and exemption systems .......................................... 24 Table 4 Selected indicators, case study countries* .................................................................... 28 Table 5 Kenya: Cost-sharing revenue as a percentage of Ministry of Health's recurrent expenditure ....................................................................... 29 Table 6 Kenya: Categories of patients and illnesses exempted, 1989-98 .................................... 31 Table 7 Kenya: Eligibility criteria for waivers in selected MOH health care facilities .... ................ 31 Table 8 Ghana: Official and actual fees charged in government district hospitals, 1997 (cedis).. 40 Table 9 Ghana: Revenue by funding source (in million cedis) .. ................................................... 40 Table 10 Ghana: Percentage of patients paying full user fee by patient category and by region 41 Table 11 Ghana: Proportion of facilities charging for exempted services ................... ............... 41 Table 12 Ghana: Average duration of fund transfer at various levels, Jan-Dec. 1999 (months)... 42 Table 13 Zimbabwe' Expected increase in MOH resources, 1990 (Z$ millions) ................ ................... 46 Table 14 Zimbabwe: Eligibility criteria for waivers in real and nominal terms ............ .............. .. 47 Table 15 Zimbabwe: Comparison of poverty lines and eligibility criteria .................... ................ 47 Table 16 Indonesia: Relationship between Kartu Sehat eligibility criteria and income-based poverty criterion, 2000 ................................................. ............. ................................ .. 52 Table 17 Indonesia: Coverage of social safety net programs across expenditure quintiles ......... 52 Table 18 Thailand Characteristics of health insurance schemes, circa 2000 ......................... ..... 54 Table 19 Thailand: Coverage of LICS and VHCS .........................5.......4................................... 54 Table 20 Thailand: National poverty line and cut off points for LICS ........................................ ... 55 Table 21 Thailand: Number of low-income cardholders by region, 1981-1998 (millions) ............. 57 Table 22 Thailand: Resources for LICS, 1988-97 ..................................................................... 58 Table 23 Thailand: Budget and expenditure of the LIC, 1987-1997 ........................................ .... 58 Table 24. Thailand: LICS per capita budget allocation by region (baht), 1992-1999 ....... ...... ... 59 Table 25. Chile: Classification of affiliates of the National Health Fund (FONASA) ..... ................ 62 Table 26: Chile' Self-declared and imputed FONASA group, 2000 .......... . ................. .......... .... 63 Table 27 Chile: Per capita benefits and contributions in FONASA (Ch$ of December 1995) . ..... 64 Table 28 Waivers and exemptions in case study countries. Performance indicators ................... 67 Table 29 Waivers and exemptions in case study countries: Funding, design and implementation features ................................................................. 72 Table 30 Waivers and exemptions in case study countries: Summary of Supply-side design and implementation features ................................................................. 75 Table 31 Waivers and exemptions in case study countries. Summary of Demand-side design and implementation features and of monitoring and evaluation efforts .......................... ............. 77 viii List of Figures Figure 1 Flat fee and sliding fee policies: Effect on the demand by the poor and the non-poor ..... 8 Figure 2 Steps in the adoption of pro-poor protection mechanisms ............ ............................... 11 Figure 3 Universal vs. targeted provision of health care ................ ...... .................................. 14 Figure 4 Share of benefits accruing to poorest 40 percent, by sector ... . ....1.........................5.... 15 Figure 5 Share of benefits accruing to the poorest 40 percent, by targeting mechanism ............. 15 Figure 6 Identification methods .................................................................. 16 Figure 7 Waiver .................................................................. 17 Figure 8 Exemption .................................................................. 17 Figure 9 Waiver and exemption combined ............................18 ........... ............................... 18 Figure 10 Identification errors. ................................... 25 Figure 11 Cambodia: Operation of equity funds .................................. 36 Figure 12 Cambodia. Per capita allocation of government health budget, 1999 ..... ..................... 68 ix Acronyms Acronym Country Meaning Ausaid Australia's Agency for International Development CASEN Chile Socioeconomic Characterization Survey CIMU Indonesia Central Independent Monitoring Unit CSMBS Thailand Civil Servants Medical Benefits Scheme DIY Indonesia Province of Daerah ltimewah Yogyakarta EF Cambodia Equity fund FONASA Chile National Health Fund GTZ Deutsche Gesellschaft fOr Technische Tusammenarbeit GOT Government of Thailand IADB Inter-American Development Bank IDT Indonesia "Left behind" villages (poorest villages) ISAPRE Chile Private Health Insurance Firm LEB Life expectancy at birth LIC Thailand Low Income Card MOF Thailand Ministry of Finance MOH Ministry of Health MOPH Thailand Ministry of Public Health MSA Suriname Ministry of Social Affairs MSH Management Sciences for Health NESDB Thailand National Economic Social Development Board NIDA Thailand National Institute of Development NTB Indonesia The province of Nusa Tenggara Barat OD Cambodia Operational District ODA Official donor assistance PAF China Poverty Alleviation Fund (PPHUP) Cambodia Phnom Penh Urban Health Project SDF Zimbabwe Social Development Fund SISBEN Colombia Beneficiary Identification System SSS Thailand Social Security Scheme USAID United States Agency for International Development VHCS Thailand Voluntary Health Card Scheme WHO World Health Organization x 1 User fees for health services: concepts 1.1 Introduction Following a dramatic rise in living standards over the last decades, extreme poverty -defined as living on less than $1 per day- declined only slowly in developing countries during the 1990s, and the number of poor people remained roughly constant as the population increased. Poverty has been rapidly rising in Europe and Central Asia, and remains on the increase in Sub-Saharan Africa. In Asia, where most of the world's poor live, the proportion of people living in poverty had shown sharp declines over the past two decades, but the recent crisis has slowed progress (World Bank, 2001 a). International agencies are putting poverty on the forefront of their agendas. At the United Nations' International Millennium Summit a declaration was issued containing main development goals and targets. Foremost is the Summit's commitment to reduce to half the proportion of people whose income is less than $1 a day. The world community is increasingly concerned about ways to combat poverty and to construct social safety nets that help the poor to get out of poverty while preventing the near poor from falling into it. Ill health is related to poverty and it is no coincidence that 4 out of the 18 development targets in the Millennium declaration are directly related to health. In its poverty reduction strategy sourcebook, the World Bank states: "poverty is both a consequence and a cause of ill health".' Poor people often lack the financial resources to pay for some health services and ill health can undermine a household's ability to cope financially. In many developing countries formal or informal user fees for health services are widespread and tend to account for a significant share of the financing for non-personnel costs. Thus, health systems in developing countries have come to rely on user fees. Yet where there is a lack of provisions to confer partial or full waivers to the poor user fees may lead to inequity in access to medical care. The dilemma, then, is how to make a much needed system of user fees compatible with the goal of preserving equitable access to services. Hence the interest in analyzing different mechanisms to mitigate the impact of user fees on access to health services by the poor. This paper is devoted to an analysis of this issue. Chapter I examines the consequences of user fees on the poor. Chapter 2 discusses at the conceptual level ways of protecting the poor when user fees are in place. Chapter 3 presents detailed case studies from seven developing countries with poor protection mechanisms in health. To conclude, Chapter 4 offers lessons learned and best practices. The remainder of this chapter is organized as follows. Section 1.2 briefly provides a historical perspective on the emergence of user fees for health care in developing countries. Section 1.3 reviews the arguments found in the theoretical and empirical literature in favor and against user fees in health care. These two sections introduce the role of waivers and exemptions in the context of health financing. Section 1.3 defines equity in health. Section discusses the economic rationale for the commonly used protection mechanisms of waivers and exemptions. 1.2 The emergence of user fees Although there is widespread recognition of the role that good health plays in human development, not all governments have had the ability or willingness to increase spending in the health sector, or to improve spending efficiency. Over the past two decades, as government budgets for the social sectors have failed See World Bank 2001 Poverty Reduction and the Health Sector, The Nutrition and Population Network's Chapter in the World Bank's Poverty Reduction Strategy Sourcebook 1. User Fees: Concepts to keep up % ith population growth and demand, many poor countries have resorted to the widespread implementation, of formal or informal user fees for health care in government health systems.2 Today user fiees constitute an important source of financing for health care in most regions of the developing woild. In some countries, such as India, Nigeria, Pakistan, Cambodia, and Vietnam, out-of- pocket spending accounts for more than half of total (public and private) health expenditure. User fees play a predominant role in the financing of HIV/AIDS health services as well. The AIDS epidemic has had profound implications for health systems around the world, including their financing and user fees have also emerged as a main source of funds for HlV/AIDS care. In Rwanda, they represent over 90 percent of all resources devoted to the treatment of HIV/AIDS (Schneider 2001). In the mid 1980s, the World Bank, along with USAID and UNICEF, was a leading international development agency promoting the adoption of user fees for health care in the public health sector of developing countries (World Bank 1987). Whereas many nations, such as Cambodia, had had user fees for government health services well before the 1980s, the influence of these development agencies was partly responsilble for the proliferation of user fees. The World Bank recognized that fees could limit access to health services by the poor, and therefore most of its policy papers prescribed that fees should be accompanied by appropriate systems of waivers (De Ferranti 1985; Griffin 1992). But despite their theoretical appeal, the viability of waivers and exemptions is an empirical question. Some authors were skeptical throughout that such mechanisms could be adopted, and criticized the World Bank for overlooking the practical difficulties of implementation (Gilson 1988). Since the early 1990s the World Bank has de-ernphasized user fees in the context of health financing and instead begun to promote risk sharing (Worlcl Bank 1997, Dror and Preker 2002, Preker forthcoming). Waivers and exemptions -the subject of this paper- are mechanisms intended to boost equity in access and in financing of health services when user fees are in place. The pervasive and in places growing presence of user fees in the health sector of developing countries, and the prospect that waivers and exemptions may improve equity in health, are the justification for this research. 1.3 The policy debate about user fees There have been numerous theoretical and empirical studies about the consequences of user fees for health care in developing countries.3 Below is discussion on the subject which draws on this literature; it attempts to sumnmarize the major arguments for and against user fees. Proponents of user fees for health care argue that fee revenue can make public spending more efficient: by improving the availability of complementary inputs such as medicines, user fees help put to better use the otherwise underutilized government-financed health workers and infrastructure. The expansion in output volume that user fees bring about also improves the availability of services to the population. In addition, by financing medicines and other supplies, fees make it possible for government health providers to imnprove the quality of care (Birdsall 1983, Ainsworth 1984, Attah 1988, Nickson 1990). Supporters of fees also note that even where government-provided health services are nominally free of charge, in practice not only quality is low but there are hidden payments and additional user costs (such as long waiting and private purchases of medicines) that result in low or unmet demand. As a consequence, low fees for low-quality public services constitute a less desirable policy than some fees for better quality care. Thus, user payments below private sector levels may suffice to make health services acceptable and available to the poor. In many cases, the adoption of fees in public facilities helps to reduce poverty, as 2 For a discussion of informal fees in the government health sector see Lewis 2001 3 Wood (1997) offers an excellent annotated bibliography on the subject. 1. User Fees Concepts 2 modest public fees substitute much higher payments made by the poor in the private sector (Akin 1986). Evidence from the Dominican Republic and El Salvador indicates that nominally free, low quality government health services face low demand (Bitran 1987, 1989). At the same time, even the poorest pay substantial fees for better or more accessible private care, showing thereby that even the poor often prefer to make some user fee payments rather than consume low quality, free services (see Boxes I and 2). Some also believe that fees convey a signal of higher service value that boosts demand above that achieved when the same services are offered free of charge. Fees are also said to promote a sense of ownership of the services received, thus empowering consumers to demand greater quality and higher provider accountability (Birdsall 1986). Other authors also allude to the gains in consumption efficiency that user fees for health care may bring about. User fees can deter spurious demand and, by reflecting the true relative costs of production, they may promote more appropriate (i.e., cost-effective) demand patterns along the referral system (de Ferranti 1985, Barnum and Kutzin 1993). Finally, some authors argue that user fees, when accompanied by a well-functioning system of waivers or exemptions, can help set up a pricing system with which to improve the targeting of public subsidies to the poor (Gertler and Hammer 1997). Detractors of user fees for health services state that fees do more harm than good. They argue that in many cases where fees are imposed, the extra revenue drawn represents only a small and irrelevant share of total revenue (WHO Study Group 1993); yet at the same time the fees have a substantially detrimental effect on the demand by the poorest (Gilson and Russell 1994). Further, they argue that often fee revenue displaces public subsidies, and therefore does not contribute to expand the revenue base of public providers. Those opposing fees also question the assertion that fees inhibit spurious demand in the health sector (Abel Smith 1993). They claim that the main problem in the health sector of poor countries is, in fact, under-utilization of medical services, instead of excess use. Rather than deterring spurious demand, fees would inhibit appropriate demand, thus keeping use of preventive and curative services below a social optimum, particularly among the poorest members of society. While there is evidence that specific fee structures may improve health care demand patterns -for example higher consultation fees in public hospitals for those bypassing lower-level facilities- existing demand pattems are often said to be a rational response by consumers to poor functioning services at the primary level. Consequently, some argue, a policy that would be more efficient and equitable than bypass fees would be the reallocation of funds away from public hospitals toward the under-funded primary level. This would improve the ability of providers at that level to attract demand and meet needs at a relatively lower cost (Dercon and Ruttens 1998). Accepting the argument that people are drawn by higher rather than lower (or no) prices for the same service would be questioning the fact that, in general, consumption of goods drops as their price rises. Finally, fees may confer users the right to demand better health services, although if services were improved through higher public funding and offered free of charge, demand would be higher than with the fees. 1. User Fees: Concepts 3 Box 1. Provider characteristics and consumption of health care services in Santo Domingo (Dominican Republic, 1987) and San Salvador (El Salvador, 1989) Even where user fees in public facilities are nomiinally very low or zero, poor quality of care, informal Figure 0 Provider choice, by income quintile fees, and accessibility problems Santo Domingo San Salvador often limit the use of public services 10__ by the poor. At the same time, iW significant use of subsidized public | services by the non-poor reveals a l leakage of public subsidies, or the ij 40%j jIj ] greater ability of 1 he non-poor to 20% pay informal fees in governmentc% * * * * , * * * * * facilities. 0% 1 2 3 4 5 1 2 3 4 5 In Santo Domingo and San hcomqu h.- q.rWe Salvador, for example, formal user M PnvatelE]_Mnistry_of_ealth___Soial_Secu fees in Ministry of Health and Social I Pnvate C Mnisiry of Health * Soaat Secunly Security public facilities were insignificant by the late 1 980s, and remain so today. The price of a curative medical consultation was 20 to 40 times higher in the private sector than in public facilities. Yet, as is seen Figure 0, in both cities and within all Table 0 Payments to private health care providers income groups, the majority of individuals obtained care from private providers, despite the relatively much higher private prices. High use of public providers by non-poor individuals (say, from quintiles Santo Domingo San Salvador 3 and up), revealls a significant leakage, or a poor Income quinble (1987 pesos) (1989 pesos) targeting of public subsidies. Qumntle 1 (poorest) 41 20 As is shown in Table 0, the out-of-pocket Quintile 2 31 48 payments for private care did not differ greatly by Quintile 3 44 46 income for the bottom four quintiles in Santo Domingo and for the top four quintiles in San Quintle 4 57 36 Salvador. As a proportion of household monthly Quintile 5 (richest) 82 45 income, however, these payments varied significantly Source Bltran, 1987 and 1989 across quintiles, and were heavily regressive. 1. User Fees. Concepts 4 Box 2. User fees and equity in health Under the right circumstances, user fees may improve Figure 0 Adoption of user fees and quality the welfare of the poor. If fees are used to improve quality improvement: Effect on demand and if as a result of the quality improvement demand for basic health care by the poor increases, then their welfare F unambiguously increases. Newbrander et al. (2000) present this issue graphically as in Figure 0. If user-fee Demand before proceeds are kept locally by the provider and are used to user fee and quality improve quality -for example, to purchase more and better improvement medicines- then demand for the improved service by the poor may expand. This shift in demand can be high B Demand afer user enough such that even with the higher price, quantity improvement demanded (Qi) is greater than it was when the price was zero and quality was low (Qo). \ --.... Outward shift in This result requires that cost recovery revenue allow a demand from quality quality improvement sizable enough to promote a P, D improvement E significant shift in demand. Relative to a scenario of no fee and low quality, consumer surplus increases with a higher A \C \ fee and better quality if the area of the triangle DEF is 0QO C Quantity greater than that of the triangle ABC. The gain in consumer Increase in quanbty demanded surplus from the user fee policy equals the difference in the when pnce and quality are higher area of the two triangles, or DEF-ABC. Gilson (1997), in her review of the experience with user fees in the health sector in Africa, notes: Evidence suggests that if fees are associated with quality improvements, as in community financing schemes of the (Bamako Initiative] type, this offsets their negative impact on utilization, and the introductton of fees plus quality improvements may even generate utilization increases among the poorest.4 Consumer surplus does not always increase when user fees are adopted, however, if fee revenue does not permit, or is not used to finance a sizable improvement in quality. In the same review Gilson concludes: Countries have not realized many of the theoretical benefits of user fees because of implementation difficulties. Thus presenting case studies of good practice is difficult. Rather, the most discernible lessons pertain to implementation problems and requisites for surmounting them Thus, these conclusions suggest that user fees should not always be discarded as undesirable, although in practice achieving the potential benefits of user fees is not a trivial problem. There is by now a growing body of literature suggesting that user fees may adversely affect the poor. In health, several studies have shown that demand for health services is more price-elastic, and that the price elasticity of demand -or people's sensitivity to price changes- is higher the lower a person's income. In Cote d'lvoire, for example, the price elasticity of demand varied between -0.3 for the highest income quintile and -1.8 for the lowest quintile (Gertler and Van der Gaag, 1988). Gertler, Locay and Sanderson (1987) used empirical results from their analysis of health care demand in Peru to suggest that the adoption of a flat consultation user fee in Ministry of Health facilities would promote inequity in access. Demand by the poor would drop significantly in response to a price increase and a simultaneous improvement in quality, while demand by the non-poor would actually increase. Consequently, the welfare of the poor would fall with the flat fee policy and that of the non-poor would go up. These authors recommended the adoption of a system of sliding fees on the basis of ability to pay. There are several other examples of important reductions in access after the introduction of user fees in the health sector. To sum up, proponents of user fees have argued that these * Generate additional revenue with which to improve health care quality 4 Gilson offers these thoughts on the basis of results from an experience in Cameroon, documented by Litvack and Bodart (1993). 1. User Fees: Concepts 5 * Increase demand for services owing to the improvement in quality * May reduce out-of-pocket and other costs, even for the poor, by substituting public services sold at relatively modest fees for higher-priced and less accessible private services * Promote more efficient consumption pattems, by reducing spurious demand and encouraging the use of cost-effective health services * Encourage patients to exert their right to obtain good quality services and make health workers more accountable to patients * When combined with a system of waivers and exemptions, serve as an instrument to target public subsidies to the poor and to reduce the leakage of subsidies to the non-poor. Detractors of user fees argue that these: * Are rarely used to achieve significant improvements in quality of care, either because their revenue generating potential is marginal or because fee revenue is not used to financing quality gains. u Do not curtail spurious demand because in poor countries there is a lack not an excess demand * Fail to promote cost-effective demand pattems because the govemment health system fails to make cost-effective services available to users * Hurt access by the poor, and thus harm equity, because appropriate waivers and exemption systems are seldom implemented. The opposing views about the desirability of user fees reflect both a difference in ideology as well as diversity in empirical circumstances. For user fees to produce welfare gains for the poor, certain conditions must be met, and the evidence shows that in practice that is not always the case. Still, the adoption of user fees tends to be a policy of choice in most developing countries. This preference seems to respond prirnarily to practical considerations. Despite their potential adverse effect on equity, user fees are relatively easy to implement and therefore they tend to be preferred over other, harder to adopt policies. Optional policies, but ones that generally are viewed as less viable than user fees, include an increase in government health budgets, additional taxation earmarked for health, the reallocation of government health funds from richer to poorer regions, risk sharing arrangements, the reallocation of public funds from urban hospitals to rural primary level facilities, and the targeting of public health subsidies toward the poor. 1. User Fees. Concepts 6 1.4 Equity in health A commonly accepted principle in the Figure 0 Equity in the delivery and financing of basic curative context of health services is that equity holds health care when consumption is in accordance with need and financing is in accordance with ability to pay (Figure 0).5 Need for health care varies from one individual to another, and to some Consumpton Payments extent is random and largely unpredictable. Over large population groups, however, { health care needs are negatively correlated - ___ with income, i.e., poorer individuals tend to Er E have a lower health status and thus need more , 0 health care. If health care consumption were in accordance with medical need, one would expect to find that, on a per capita basis, the poor would consume greater amounts of curative health care services than the non- poor. A necessary condition for Poorest Income consumption to match need in health is that access -physical, financial, and cultural- to appropriate services be should favor those with greatest need. In particular, there should be no financial barriers for those wishing to obtain basic health care. For the poor, this means that direct payment for basic services, in the form of medicines or consultation fees should be lower than for the non-poor. Equity in financing holds when those with equal ability to pay make equal payments for basic health care (horizontal equity in financing) and those with greater ability to pay make higher payments (vertical equity in financing). Payments include direct and indirect taxes, payments to social security and to insurance (mainly for health care), prepayments for health care, and all out-of-pocket payments for services. 1.5 The rationale for waivers and exemptions For normal goods, such as quality curative health care, demand by the non-poor is greater, on a per capita basis, than demand by the poor. This means that at any given price, such as a flat fee (Pflatjfee) in the left panel of Figure 1, the non-poor will demand higher quantities (Qnonpoor) than the poor (Qpor). The higher demand by the non-poor responds to their higher income. It also reflects their greater education and the higher associated awareness about the benefits of timely, quality health care. Thus, with a flat fee the poor will be at a relative disadvantage. Sliding fees, or price discrimination, can help solve this problem. To simplify, it is assumed that per capita health care need is equal for the poor and the non- poor. If a sufficiently lower fee can be charged to the poor, such as Ppoor in the right panel of Figure 1, then their per capita demand will equal that of the non-poor. Protecting the poor from user fees entails exactly this kind of mechanism: the structuring of a fee system that promotes equal per capita consumption by the poor and the non-poor.6 In theory, then, the problem of mitigating the negative consequences of user fees on the poor is solved quite simply. It suffices to know the slope and position of the demand curves, something that can be estimated empirically through household survey data. In practice, however, it is necessary to know who is poor and who is not, and to This is the principle proposed by van Doorslaer, Wagstaff, and Rutten (eds ) (1993). More generally, with greater per capita health care need by the poor, equitable pricing is one that leads to higher per capita consumption by the poor. 1. User Fees: Concepts 7 apply differential prices appropriately. Experience shows, alas, that this is an extremely difficult practical problem, wilh rnajor administrative and logistical requirements. Further, it has been shown that fees are not the only costs faced by potential users of health services. For example, in Cambodia transportation and food costs associated with the use of hospital services can greatly exceed the fees charged by the provider (Hardeman 2001). Thus, while fee exemptions may improve equity in access, to be effective in some cases the, must be accompanied by supplementary subsidies to defray additional and significant non-fee costs associated with consumption. Figure 1 Flat fee and sliding fee policies: Effect on the demand by the poor and the non-poor Flat-fee polic.y the poor demand less per capita than the Sliding-fee policy: the poor and the non-poor have equal per non-poor capita demand Pnce Pnce Demand by lOfpOOr Demandby the non-poor the non-poor Demand by p o__ Demand by the poor P0O the poor Q,o Q-poo, Quantty Qpor=Qn-Fpo Quantity Difference In per capita equal per cap,ta demand demand between poor and for poor and non-poor, wth non-poor with flat-fee policy sliding fees A main reason behind the practical difficulty of Figure 1 Propensity to misrepresent one's telling poor from non-poor resides in the incentives economic status as a function of the size of facing the non -poor to misreport their identity. If a the benefit and the magnitude of the sanction subsidized service wanted by the poor and the non- if cauaht Ivina poor is made available to all, it will be demanded by both groups. But fiscal constraints in poor countries expected imply that good quality, fully subsidized health care sanction cannot always be made available to all. There is ° Middie need to ration the subsidized benefit, and the E sanction appropriate policy decision is to keep the non-poor out of the subsidized scheme. Yet if the subsidized _/ High c expected services are wanted by all, chances are that the non- ssanctin poor will seek to misrepresent their status, to be santi/ classified as poor and thus obtain the subsidy. This e is illustrated in Figure 1. The upward slope of the "propensit) to lie" curve shows that the higher the Size of the benefit (eg amount of subsidized, quaiity healh care) payoff (i.e., the desirability of the subsidized service), the greater the propensity of the non-poor to lie. If it were somehow possible for the provider of the subsidized social service to tell poor from non-poor, for example through a home visit by an incorruptible social worker, and to sanction those lying, then the liars would be facing a known sanction with a known probability. This is not different, conceptually, to the problem of tax evasion. Figure I shows that the higher the expected value of the sanction, the lower the propensity of the non-poor to misrepresent their status. Unfortunately there is no such thing as a perfect system of sanctions, and often the optimal policy is the adoption of a combination of incentives -both positive and negative- that lead 1. User Fees: Concepts 8 most individuals to behave in accordance with what the managers of social programs expect. Constructing and implementing such a set of incentives requires skill, resources, and information, and constitutes the great challenge of implementing effective and efficient targeted social programs for enhancing the well- being of the poor. The experience of Colombia helps to illustrate the difficulties of avoiding the leakage of subsidies to the non-poor (Box 3). 1. User Fees: Concepts 9 Box 3. Colombia's experience trying to tell the poor from the non-poor In the mid 1990s Colombia began implementing a The effort appears largely successful in reaching the major health reform known as Law 100. A main poor in those municipalities that have received reform aim was achieving an improvement in the adequate resources. Yet it has been reported that targeting of public health subsidies for the poor. To some families that would otherwise not qualify for the achieve this, municipalities around the country were subsidy, rent for a few hours a poor family's house given resources and were assigned the task of prior to being subject to the test, to qualify as, thus carrying out surveys of all households in their defeating the targeting tool. Also, some individuals commune to identify those that were poor. Poverty subject to assessment by social workers in public was determined on the basis of various household hospitals are said to purchase false utility bills on the characteristics, such as quality and size of the basis of which hospital staff make a determination construction, number of household members, about who pays what amount for health services. education levels, and so on Falsifiers of bills station themselves in kiosks outside of the hospitals and sell customized bills that make the buyer look poor. 1. User Fees Concepts 10 2 Mitigating the effect of fees: Identifying and protecting the poor Preserving equitable access to health services under a system of user fees can be accomplished in three steps (Figure 2). First, the poor population to receive preferential treatment with respect to user fees must be identified. Second, a protection mechanism must be selected and implemented. Third, to ensure that the protection policy is working adequately, an evaluation of its performance is required. Issues pertaining to these three steps are addressed in this chapter. Figure 2 Steps in the adoption of pro-poor protection mechanisms Ste) I Ste1 2 Step 3 Identif benefiiariesSelect and Evaluate Identfy beneficnaries implement performance of ofpection protection protection mechanism mechanism mechanism 2.1 Identification of beneficiaries of protection mechanisms There are several methods available to identify those in need of protection and thus to target public subsidies towards those requiring protection. These methods constitute an alternative to the traditional approach of health ministries consisting of universal provision (see Box 5). Which targeting method works best in practice depends on its administrative feasibility and costs, political viability, impact on demand, and other factors (see Box 6).' Different authors offer different typologies of targeting methods.8 The following four categories, depicted in Figure 6, are retained in this paper (for a further description and comparison of these targeting methods see Appendix A): * Individual identification * Identification based on group characteristics * Self-identification * Self-selection by type of service (a) Individual identification With this method, who is entitled or eligible to subsidized health services is determined through an assessment of individual characteristics such as income, health status, behavior, nutritional status, or other criteria. Where ability to pay is the basis for selection, the assessment, which is typically carried out by a 7 See Sen (1995) Grosh (1995) recognized three main methods (1) individual assessment mechanisms, (2) group or geographic targeting, and (3) self-targeting, Glewwe and van der Gaag (1988) and Willis (1993) distinguish between two types of targeting. (1) characteristic targeting (equivalent to Grosh's group targeting), and (2) direct targeting (based on direct individual or household assessment). The World Bank (1993) identifies (1) individual targeting, (2) group targeting, (3) self-targeting, and (4) targeting by type of service. 2. Mitigating the effect of fees 11 social worker or a specially trained health worker, may be based on income. But where target beneficiaries work mostly in the informal economy assessing income can in practice be very difficult. In those circumstances, other methods of assessing ability to pay are used, including means tests and proxy means tests (see Willis 1993 and Appendix A). (b) Identification based on group characteristics Assessing individual characteristics is not always easy or efficient. For example, suppose that a carefully conducted household survey carried out on a sample of households reveals that the vast majority, say 85 percent, of the inhabitants in a particular geographic region are poor and should be target beneficiaries of a subsidized program. Given the high percentage of target beneficiaries, it may be more efficient to confer protection to all area residents than trying to assess individual characteristics to single out, at a high administrative cost, the non-target population (the non-poor 15 percent). The cost of identifying the poor and the non-poor could exceed the amount of subsidies that would leak out to the non-poor if no targeting effort were in place. In such circumstances, protection is provided at the group level, where a group may be a geographic area known to be predominantly very poor. The leakage to the non-poor is tolerated on efficiency grounds. (c) Self-ideintification In self-identification, no effort is made by the agency granting the protection to identify its recipients because individuals self-select. That is, the health services are provided in such a way that it is mostly those individuals eligible for protection who will come forth and demand the services, whereas non-target persons will mostly demand services elsewhere. For example, this can be achieved by operating a health center in a predominantly poor neighborhood. Most users will be the very poor who inhabit the area whereas fewer non-poor will decide to seek care there, because of stigma, safety concerns, long waiting lines, the lack of amenities, or other reasons. A system designed for self-identification may sometimes impose high private costs to the beneficiaries, such as safety problems and the opportunity cost of the wait. (d) Self-selection by type of service The idea here is to offer subsidized services that, for epidemiological, cultural or other reasons, are demanded disproportionately by the poor, or target group, given the special health circumstances and income constraints they face. This is the main idea imbedded in the supply of a basic package of health services in developing countries. The treatment of dehydration from diarrhea, nutritional supplements, and the caring for sexually transmitted diseased and tuberculosis are all examples of services that, if made universally available, would especially benefit the poor. 2. Mitigating the effect of fees 12 Box S. Universal versus targeted provision Despite its theoretical appeal and favorable results around the developing Figure 2 Indonesia: The allocation of government world in other social sectors targeting health care subsidies. 1987 still remains a relatively infrequent method of directing government health 200 - - - - - - subsidies. In fact, the most widely Subsidies practiced method in the developing received from world for reaching the poor with free or E 150 govemment subsidized government social services a health centers K 8 42.3 is universal provision, also known as a -_ 5. 100 _ general price subsidies. Universal 7 435 provision refers to the traditional X 38 provider arrangements of national or : 50 388 regional health ministries around the a. developing world, which offer free or subsidized care to all patients, often 0 1 2 3 4 5 6 7 8 9 lo regardless of ability to pay (see Figure 3 Income decde Subsidies below). However, van de Walle (1995a) received from states that govemment hospitals uuniversal provision is too costly and fails to have much impact on poverty subsidized ambulatory services. Yet the non-poor obtained a much and that targeting can promote cost- larger share of hospital care than they should (i.e., if all citizens, effectiveness. So, in the context of irrespective of income, were entitled to an equal amount of subsidy). pressures to reduce public expenditures, This is attributable to the fact that government hospitals were mostly the view has become widespread that located in urban areas were most of the non-poor tend to live, and targeting allows governments to reduce also to the greater ability of the non-poor to secure public subsidies. poverty more effectively and at a lower Recent data from Morocco signal the occurrence of a similar cost", phenomenon there. In Morocco the poor can obtain indigence certificates through their local governments, thereby becoming There is ample evidence from the entitled to free care in government health facilities. The process for developing world that universal provision the awarding of such certificates, however, is subject to abuse, and of health services leads to a situation thus subsidies appear to be poorly targeted. It is estimated that the where only a share of the public subsidy richest quintile captures 40 percent of MOH spending while the reaches the poor while a sizable portion poorest 40 percent receives less than 20 percent of MOH resources of it leaks out to the non-poor. Such (World Bank 2001 b). When there is a leakage there is a case for was the case in Indonesia in 1997, as is improving the efficacy of targeting, as this will help to reach the poor shown in Figure 2. The poor and the with the subsidy and thus exempt them from the fees non-poor consumed similar amounts of 2. Mitigating the effect of fees 13 Figure 3 Universal vs. targeted provision of health care SUBSIDIES UNIVERSAL PROVISION Universal provision offers subsidized services to the all citizens, irrespective _ _ _ r__________________________________________ of their socioeconomic status or needs lTi i i Il I Thus, some of the poor obtained 1 t s r t 1 t fl 1subsidized services (A) while some A A A C C B nnon-poor also do so (B) Owmng to the A B B D limited nature of subsidies, and of the 64f % 6"fi ( % tefi l) (P), ffi XPIk to fi m (6 t I services they finance, some people are _ / ~~~~~~~~~~~~ ~~~~~ /wl ~~~~~~~~~~left wwthout subsidized services _ _ _ _ _ _ _ I 1/ 1/ 1/ 1 Among them are some of the poor, POORAi A NON-POORA / ""_' A (C), and some of the non-poor (D) POOR NON-POOR ~~~~~~~.. . . . . . . . .... ..... . . . . ...... ..... .... .... .....a.a.......*X X ADMINISTRATIVE one rei 0fi one ffi (0fi\ t-fIN n$o " EXPENSES TARGETED PROVISION Targeted provision adopts mechan- isms to channel resources toward the _ _ __ X * poor It is successful if it increases the A A A A A A C B D D D G amount of subsidies reaching the poor A A~~~~ ~ A (A) while reducing the flow of subsidies 4 to the non-poor (B) But sometimes target-ing procedures have an _i _T _ _administrative cost (G) -for example | the salaries of social workers who Identify the poor through individual interviews This administrative cost reduces the amount of subsidies _~~j A- CORRECT INCLUSION OF THE POOR available to pay for social services, SUBSIDIES B= WRONG INCLUSION OF THE NON.POOR thus leaving some of the poor without C= WRONG EXCLUSION OF 'HE POOR access to the subsidized services D.CORRECT EXCLUSIONOF THE NON POOR Source Bitran 2000 2. Mitigating the effect of fees 14 Box 6. Targeting effort and incidence: Empirical evidence from Latin America To assess the tradeoff between administrative terms of incidence of benefits, as is shown in Figure costs of targeting and targeting outcomes of 5. Individual assessment mechanisms achieved a govemment-subsidized social programs, Grosh median of benefits going to the poorest 2 quintiles of (1995) carried out an empirical review of 30 such 73 percent. The equivalent figure for geographic programs in Latin America, including ones in health targeting mechanisms was 72 percent, while for self- and education. Most were government programs with targeting it was 71 percent. In addition, there were national coverage; several of the newest programs only minor differences in the performance of these were specifically motivated by increases in poverty in different mechanisms among the various countries in the 1980s or by the need to mitigate the social costs the sample. of macroeconomic adjustment programs. Seventeen Figure 5 Share of benefits accruing to the of the programs used individual assessment mechanisms, seven used group or geographic poorest 40 percent, by targeting mechanism mechanisms, and six were self-targeted (see Figure 6). Grosh's two research questions were (1) which .o targeting mechanisms provide the best targeting outcomes? And (2) what are their administrative costs? , i She found that targeted programs had better S. incidence, i.e., a greater proportion of the subsidy reached the poor than with the untargeted general food price subsidies (Figure 4). The best performing 2. general food price subsidy program (that is, the least regressive of them), delivered 37 percent of the 0 subsidy to the poorest 40 percent of all households. INDIVIDUAL GEOGRAPHIC SELF- The worst performer of the targeted programs (i.e., ASSESSMENT. ASSESSMENT, ASSESSMENT the least progressive of them) channeled 59 percent N - IDO of the benefits to the poorest 40 percent of the I___ HIGH __ MID___ LOW__ population. On average, 33 percent of the benefits of general food price subsidies went to the poorest two Although targeted programs had a much better quintiles, while for the targeted programs the figure incidence that untargeted ones, Grosh had to was 72 percent. examine the relative costs of both types of program to be able to infer their relative merits. Figure 4 Share of benefits accruing to The median cost of targeting as a share of total poorest 40 percent, by sector program costs was 6 percent for programs relying on self-targeting, 7 percent for those using geographic assessment, and 9 percent for programs that relied on individual assessment. Not all program administrative n . , costs are associated with the actual effort of targeting, 2 . . . . however. Still, targeting costs (beneficiary identification, or screening) were relatively modest, I I , , accounting for a share of total program costs that I l varied between 0.4 percent and 8 percent. Finally, Grosh examined whether programs that succeeded in channeling a greater proportion of benefits to the GEOERALFODO TARGETED PNNMARY PMMARY poorer income groups cost more administratively. 3UBSIDIEJ, PROGRAMS HEALTH CARE, EDUCATION N-7 NU1I N-11 N-11 She concluded that was no clear connection between HIGH _ MID LOW targeting results and administrative costs. Grosh also found that different targeting mechanisms produced rather similar outcomes in 2. Mitigating the effect of fees 15 Figure 6 Identification methods Methods Individual Group Eligibiliy for protection is established on an A collection of individuals is selected for protection individual basis based on location, ethnicity, gender, age, etc. Meet Do not meet Group - \ Get Protection - Do not - selection No..- get protection By tvDe of service Self The poor tend to demand the subsidized service For epidemiological reasons, the service provided is more frequently than others demanded mostly by those requiring protection No,- t: IDangerus area of L..J (~~~~~~~~~~~~9 t~~~~~. - i~~~~~~owri., long vri,ng Serirces Services Look L .0k for without or subject to belier, pa better, paid With low fees care' ae user fees 3isewh. elsewtere Mixed A combination of two or more of the four methods 2.2 Protection mechanisms: Waivers, exemptions, and design features (a) Waivers A waiver is a tight conferred to an individual that entitles him or her to obtain health services in certain health facilities at no direct charge or at a reduced price (see Figure 7). The subjects of waivers are individuals The existence of waivers in a health system implies that the system will discriminate between waiver holders and the rest of the population. By reducing the out-of-pocket cost of care to beneficiaries, waivers seek to improve both equity in access and equity in financing of health services. 2. Mitigating the effect of fees 16 Figure 7 Waiver Main target group: the poor and the near poor Example 1 - _ Non Poor Waiver on the Service I basis of poverty Service 2 Service n t t t People in these People Waiver: A right awarded to one or groups are in these more groups of individuals that waived the groups entitles the recipients to obtain requirement to pay pay user fees or regular health services at no fees or at are allowed to fees reduced fees pay reduced fees Other Groups Example 2 Servicei1 Waiver on the basis of ethnicity Service Service n Main target group: the disadvantaged ethnic groups (b) Exemptions Whereas waivers are associated to certain individuals, exemptions are associated to certain services. An exempt service is one that is to be provided at no charge (or at a reduced price) to patients. In its broadest form, a waiver entitles its holder to receive all services at no direct charge; in its broadest form, an exemption implies that the exempt service will be provided to all individuals at no charge. Exemptions are adopted mainly for efficiency reasons and thus seek to correct some market failures. Their purpose is to promote the consumption of specific health services, including those whose benefits are under-valued by the population, those that have externalities, or those that are pure public goods. Vaccinations are a typical example of an exemption whose purpose is to correct information failures (uneducated people tend to fear vaccines or under-value their benefits) and to promote consumption for a medical service that has an externality. Figure 8 Exemption Exemption: The Main target group: the poor and the near poor condition assigned to a Near Non service whereby Poor Poor Poor All people, no user fee is Service i trrespective of charged for it to their ability to consumers, or I pay, can whereby the same _ obtain this reduced fee is _ service at no charged to all ior at a consumers, subsidized irrespective of Sepcn sbiiee their ability to pay 2. Mitigating the effect of fees 17 (c) Waivers and exemptions combined Waivers and exemptions may be combined, thus setting a system where certain individuals are entitled to obtain certain health services for free or with a subsidy (Figure 9). Figure 9 Waiver and exemption combined exemDtion Main target group: the poor and the near poor combined: Waiver Non Poor Some holders Service I individuals can obtain this service are granted at no direcl= the right to charge or obtain some ata services at no subsidized 1 1 11n direct charge prce n or at a reduced price People in these groups are waived the requirement to pay user fees or are allowed to pay reduced fees (d) Design features for waivers and exemptions Designing and implementing a system of exemptions is considerably simpler than doing so with a system of waivers. A system of exemptions requires one initial basic decision, namely determining which services will be offered for free or at reduced prices. Such a determination may respond to specific economic and medical criteria, such as those mentioned in the preceding paragraph. Once that decision has been made, managing the system of exemptions tends to be administratively simpler than managing a system of waivers. This relative simplicity stems from the fact that a system of exemptions imposes no need for identification of beneficiaries -a major hurdle in a system of waivers. Adopting exemptions thus can be as easy as imparting the order among the concerned providers that the services to be exempted are to be offered at no charge to all customers. Most developing countries have systems of exemptions, and these function quite well and impose only minor administrative demands. For example, in some countries some primary and preventive services, such as child immunizations, are offered for free in all public facilities and to all patients country-wide. Aside from deciding which services are to be waived, designing a system of exemptions entails also determining who will deliver the exempt services -it can be public, private for profit, and private non- profit providers- and how the costs of delivery will be financed by the payer of services. A common situation is that the providers of exempt services are all public and their funding comes from their fixed monthly budget, in which case the payment system remains easy to manage. Universal provision of health services free of charge to all in government facilities that are financed through historic budgets is a generalized system of exemptions. In some situations, the funding of exempt services may come from payments made by the financing agency to the provider of exempt services on the basis of a previously agreed on payrnent mechanism, in which case management may become more complex. Capitation requires that a list of registered beneficiaries be handed by the provider to the payer; fee-for-service requires that periodic invoices be submitted by the provider to the payer; and so on. Brazil's Sistema Unico de Sauide (SUS) offers a large set of exempt services to Brazil's population through a nationwide network of public and private providers, to which it has to pay for exempt services on the basis of diverse payment systems. 2. Mitigatinig the effect of fees is To sum up, designing and adopting a system of exemptions is relative simple because, unlike a system of waivers, it does not impose the need to identify individual beneficiaries. A system of exemptions shares with a system of waivers the administrative requirement that a payment system be in place to defray the costs of the services that are offered at reduced prices or for free. Designing and implementing a system of waivers is, in contrast, considerably more complex because it imposes the adoption of different rules for different individuals. The need to classify individuals as beneficiaries and non-beneficiaries of waivers, along with the need to identify them at the point of service imposes major administrative demands on the health system. Here, many design options arise and these impose a series of consequences for the health system, in terms of incidence of benefits and administrative costs. Table 2 outlines in the form of questions the design options that the policymaker wishing to adopt a system of waivers faces. The last two columns of the table show that some questions apply both to waivers and exemptions, whereas others apply only to one or the other system. The design options of Table 2 have been classified in four distinct categories, as follows: * Resource availability: How much money will be available to finance waivers and exemptions? Often, policymakers adopts waiver and exemption systems without making the necessary provisions to ensure that there is a match between the total production cost of the services to be waived or exempted and the funds available to finance the policy. Achieving such a match requires that estimates be available about the total beneficiary population, their demand for the waived or exempt services, and the cost of those services. * General design and implementation: Adopting a system of waivers or exemptions involves a series of design issues. One central issue is the revision of the existing system of user fees with the aim of furthering the objectives of the waiver or exemption system. This means that user fees must be in place for those with an ability to pay (patients not entitled to waivers must pay the ongoing fee) and that only exempt services must be provided free of charge (non-exempt services are charged to patients). Revising the user fee policy is important whether the facility is allowed to retain the revenue locally or whether user fee proceeds must be returned to a regional or central authority; in either case user fee revenue will eventually ease the fiscal impact of a waiver or exemption policy. There are other key design and implementation issues and these are outlined in Table 2. They involve determining which services are to be exempted or which individuals are to be waived, the duration of the exemption or waiver, choosing an identification (or targeting) instrument, and deciding who will administer the system, among other things. * Supply-side design and implementation features: As already noted, the particular design of the waiver or exemption system conveys specific incentives to the providers of services. A good design is one which confers incentives leading providers to behave in ways that further the objectives of the waiver or exemption system. That is, a good design leads to the provision of the appropriate amount of waivers to the target beneficiaries, the denial of waivers to non-beneficiaries, and the provision of the right kinds of exemptions. Clear and realistic definitions of beneficiaries, and appropriate dissemination of information about procedures, are factors that contribute to the success of a waiver system. In contrast, a faulty design may defeat the waiver or exemption policy. For example, an ambiguous or complex definition of a beneficiary in a waiver system can lead to failure. Examples of successful and unsuccessful experiences are provided in Chapter 3. * Demand-side side design and implementation features: Waivers and exemption systems confer incentives to consumers, or should be accompanied by incentives which must be aligned with the aims of the systems. For instance, where exemptions exist for some services, such as institutional deliveries, appropriate information must be disseminated among pregnant women, often with the help of promoters and community leaders, to foster an understanding about the health benefits of institutional deliveries and, thus, to promote demand for the exempt service. Simply exempting institutional deliveries may in some circumstances be an insufficient incentive to 2. Mitigating the effect of fees 19 boost demand for this service. As regards waivers, a system which provides free curative health services to holders of an indigence card may fail to draw target beneficiaries if access costs other than the user fee (e.g., travel costs and the opportunity cost of time) are high. In such a case, the waiver system may also have to envision funds and a mechanism for defraying those other consumption costs of target patients. As already noted, relative to exemptions waivers have the added difficulty that they are awarded on the basis on persorial characteristics. Therefore they require an administration for the waiver process and the adoption of decisions regarding the awarding of waivers. These are discussed next. Waivers: Design and implementation issues Who grants waivers? Whose responsibility is it to identify those to be protected: An epidemiologist with the MOH, a specialist with the National Statistical Institute, a social worker with the agency delivering the social services, or a doctor working in the hospital? In El Salvador, MOH officials determine on the basis of a broad set of poverty and health status indicators which health districts are entitled to a free package of health services. In Zaire's health zones, it is the nurse running the health center who makes the determination about who is fully or partially exempted from fees. Thus, waivers may be granted by a special administrative body that is separate from a health facility, in which case waiver beneficiaries must carry some kind of identification that distinguishes them from non-beneficiaries when demanding health services. Or waivers may be provided health facility staff and in that case identification may be unnecessary. When an entity other than the health provider grants waivers a conflict of interest presents itself if the cost of the waived services is not faced entirely by the entity issuing the waiver and if the waiving agency is rewarded on the basis of the volume of waivers granted. In Cambodia, for example, some equity funds granting waivers do not have to reimburse health providers for the full cost of waived services and therefore they may over-provide waivers. Suriname's Ministry of Social Affairs used to over-provide waivers because it did not face the full cost of waived services, and its staff derived personal benefits from issuing waivers. When that situation changed, and MSA begun to face the full cost of health care, the agency became much more stringent and reduced the number of waivers granted. Are waivers supply- or demand-driven? Does the identification system actively search for prospective beneficiaries, or does it passively wait for them to show up and request protection? In Suriname, the Ministry of Social Affairs (MSA) issues MSA cards that exempt cardholders from the payment of user fees in govermnent hospitals. To obtain the card individuals must apply as the MSA does not actively search for potential beneficiaries. In Colombia, in contrast, the Beneficiary Identification System (SISBEN) run by municipalities actively interviews all households in their jurisdiction, and selects the ones to enter the Subsidized Regime. All beneficiaries of the Subsidized Regime are automatically entitled to a gcvernment subsidy that covers their health insurance premium. When are waivers granted? Waivers may be provided at different moments in time. For example, they may be distributed to beneficiaries in their homes or in the premises of an administrative agency when individuals are in no imminent need for health services. Beneficiaries thus carry the waiver document with them in case they will need to produce it at a health facility in the future, should the need for health services arise. Chile's National Health Fund (FONASA) provides waiver cards to the indigent and these present the cards to public providers when demanding health services, thus receiving free care. Alternatively, waivers may be provided by health facility staff or by a special agency at the time when health services are demanded. Such is the case in Cambodia's Takeo Hospital, where specially trained health staff subject waiver applicants to the means test on site, asking them to fill out a form answering various questions about their employment, house characteristics, number and age of household members, possession of vehicles and other durable assets, and ownership of domestic animals. Waiver granting staff thus compute a score on the basis of the answers given and provide a waiver if the score is below a certain threshold. 2. Mitigating the effect of fees 20 When a waiver is provided -ex ante or at the time the services are needed- may affect health care seeking behavior if the waiver application costs, also known as "participation costs", are high or if prospective applicants are uncertain about their chances of getting a waiver. In situations where waivers are granted ex-ante, waiver beneficiaries know with certainty that if they demand health services in certain locations they will be entitled to free care. Thus, their demand for care will not be diminished by the prospect of an out-of-pocket payment. In contrast, in situations where a waiver determination is made each time a health service is demanded, some prospective patients may get discouraged from seeking care if waiver participation costs are high or if their perceived likelihood of getting a waiver is uncertain. Participation costs refer to the economic, time, and other costs that an individual has to incur to apply for a waiver. For example, Table I presents some of the requirements set forth by a government agency in a Latin American country to grant waiver cards for health care or to provide free care to those not entitled to a waiver card. This is an example of a waiver process with high participation costs. If the waiver granting entity is known by the target population to always awards waivers to applicants, as was the case of Suriname's Ministry of Social Affairs,9 then when the waiver is provided -ex ante or at the time the services are needed- makes no difference in terms of people's health care seeking behavior. Those needing care will demand services equally, whether they already have a waiver card in their possession of whether they do not have a card but are certain that they will obtain free care. Table 1 Requirements to obtain health services waiver: example from Latin America Individuals entitled to a waiver health Dependent Worker Dependent Worker card: requirements to obtain the card Indefinite Contract Temporary Contract Retired Application form Yes Yes Yes Photocopy of work contract Yes Yes Contribution certificate Yes, last payment Yes, payments for at least No, but must present 6 months pension payment coupon Personal national identification card Yes Yes Yes Family dependent's national identification card Yes Yes Yes Certificate of dependence Yes, from birth registry Yes, from birth registry Yes, from pension Fund manager Duration of waiver 2 years 12 months Indefinite Individuals not entitled to a waiver health card: requirements to obtain free health services Independent Worker Unemployed Worker Indigent Certificate of indigence No No Yes Contribution certificate Yes, payments for at least 6 months Personal national identification card Yes Yes Yes Family dependent's national identification card Yes Yes Yes Certificate of dependence Yes, pension fund manager Pension fund affiliation certificate Yes Other Unemployment subsidy Health condition certificate card from the health service or municipality Eligibility requirements. What information is required for identification: data from a household survey, individual tax forms, proof of income, and assessment of assets or enrollment in other social protection programs? If administrative requirements for identification are major (as in the case of Kenya described in Chapter 3) then participation costs might be too high, thus leading to a modest level of identification (under-coverage). If, on the other hand, the requirements are too few, then leakage may be high. Eligibility requirements may be as simple as responding to and passing an on-site interview to a longer and more demanding process such as the one described in Table 1. In an effort to increase precision, those designing and managing eligibility processes tend to adopt complex, demanding, and time Giedion, Bitran, and Mufloz 2002 2. Mitigating the effect of fees 21 consuming lesis and interviews. These raise the participation cost of waivers and thus discourage some prospective applicants. Those who choose to comply may be lower-income persons whose opportunity cost is lower than the value of the benefits that the waiver carries, but such cumbersome procedures may also discourage application from some low income individuals wary of questions and bureaucracy. Finding the right balance between the stringency of a means test for waivers and the coverage of the waiver prograrn (percentage of the poor who actually benefit from waivers) is a major challenge for policymakers. Duration of Naiver Waivers can be temporary or permanent. The most short-lived waiver is that provided for a specific service within a health facility. For example, a patient who is hospitalized may have to request a waiver for the payment of a laboratory exam, and then again, during the same hospitalization, he or she may have to apply repeatedly for waivers on other medical procedures, such as the surgical fees or the medicines received. Such an atomized waiver procedure, illustrated in an example in Chapter 3, which tends to cause confusion and frustration among patients, is common in those health facilities where there are separate medical departments, each with its own budget and waiver policy. A somewhat broader waiver but one that is still short term in nature is that which is provided for the treatment of a specific illness episode, covering all inputs and services given to the patient during that episode, but expiring once the treatment is completed. Prospective patients must reapply for a waiver in the future to treat new illness episodes. Long-tenm waivers are those awarded to individuals with a validity of one year or more, including those that are indefinite. In some countries the handicapped or the elderly are entitled to permanent waivers in certain health facilities. Extent of waiver. Waivers may be full or partial. A full waiver entitles its beneficiary to obtain for free the services it covers; a partial waiver may require of the applicant a partial payment, that is, one below the full price. Chile's National Health Fund confers full waivers to indigent beneficiaries, but has a sliding fee scale for non-indigent beneficiaries, thus granting a price subsidy that decreases with income. Thailand's fonner low income card system (LICS) fully exempted the poor from payment in health facilities. Table 2 Waivers and exemptions: design and implementation issues Waiver Exemption Resource availability What amount of resources is available to finance waivers/exemptions? X X General design and implementation Is user fee policy effectively used to further objectives of waivers/exemptions policy? X X Which services are waived or exempted? X X Who is entitled to a waiver? X At which level of health system is waiver/exemption provided? X X On what basis is waiver provided (individual assessment, group, self-selection)? X How well informed are implementers about waiver/exemption policy? X X If waiver provided on individual basis, what kind individual assessment method is used (means test, proxy means test)? X Who provides waivers? X Is waiver provided on an individual or a family basis? X Is a systenm of identification cards used? X Are identification criteria revised regularly? X How long is waiver good for? X Supply-side design and implementation features Are providears compensated for revenue forgone? X X What paynient mechanism is used to compensate providers? X X 2. Mitigating the effect of fees 22 Table 2 Waivers and exemptions: design and implementation issues What is the significance of compensation relative to customary fees and to actual cost? X X How timely is compensation? X X Does compensation cover administrative costs? X X Are compensation amounts revised regularly to keep up with costs? X X Is user fee revenue kept at facility? X X Is staff income tied to user fee revenue? X X Are records kept on user fee revenue and on waivers and exemptions provided? X X Demand-side design and implementation features How well disseminated is waiver/exemption policy? X X Do agents responsible for providing waivers actively screen for potential beneficiaries or do they passively wait for them? X When is waiver provided ex-ante, when service is demanded, or ex-post? X How important is social stigma? X X Are beneficiaries discriminated (i e, are they mistreated or treated differently)? X X How accessible are services for non-beneficiaries? X X How well informed are potential beneficiaries about policy? X X What is the magnitude of individual access costs to the system? X X Does program finance all or part of patients' access costs? X X What penalties are imposed on those cheating the system? X Reach. A waiver may be provided to some members of the household or to all. Typically the latter is more common, but there are several examples of waiver systems which are provided at the individual level -for example a waiver covering a single mother and her child, both of whom live with the mother's parents, but where these do not qualify for a waiver. Who provides program benefits? Do protection mechanisms restrict the set of providers where program benefits are delivered or can beneficiaries freely select their providers? In Colombia the beneficiaries of the Subsidized Regime get a demand subsidy from the municipal govemment that allows them to enroll with a public or with a private health insurer of their choice. But they can obtain health services only from those providers that have a contract agreement with their insurer. Chile's National Health Fund entitles its higher income beneficiaries to a partial waiver for private health services. Beneficiaries can freely select the provider from a broad list established by FONASA. In Honduras, beneficiaries of the Matemal and Child Food Coupon program can receive health benefits financed by the program only from public facilities, but they can redeem the food coupon in any private grocery store. Financing of waived benefits. Up to here all issues in this section have dealt with the process of identifying beneficiaries and managing the waiver process. A distinct but equally critical component of a waiver system is its financing. A successful waiver program requires that the value of the services waived equal the funding available for the program. Such a balance is seldom contemplated in the design of a waiver policy and thus rarely holds. As already noted, achieving that balance requires knowledge about the number of beneficiaries of the waiver program, their demand for the waived services, and the production cost of the services. These information requirements increase with the number and complexity of the services included in the waiver program. Lack of clarity about funding needs for a waiver program often means that financing is insufficient and that the rationing of services emerges as the de facto tool which imposes a balance. For example, Guatemala's Coverage Extension Program implemented in the late 1990s sought to deliver about 15 different kinds of preventive and primary health services free of charge to rural poor populations through provision contracts with NGOs. Poor prior knowledge about actual delivery costs meant that the program was underfinanced and as a consequence only few of the services contained in the basic package of 15 interventions were actually delivered to the 2. Mitigating the effect of fees 23 target population. Carrying out a thorough analysis of financing needs for a waiver program is a key to its success. Provider compensation. Does the provider of services get financial compensation from the program for the delivery of subsidized services to beneficiaries of protection programs? Or does it have to collect revenue from non-program customers to cover program costs? Or does it have to ration delivery and limit free care to live with its limited existing budget? Many waiver programs fail because they do not fully compensate providers for the revenue they forego by delivering free services to program beneficiaries (see examples in Chapter 3). Compensation obviously requires knowledge about production costs and the existence oi a payment system through which the waiver program channels funds to the provider. Which payment system is chosen has implications for the performance of the waiver program. For example, if fee-for-service is selected, and if the fee reflects full marginal cost, then the provider will have an incentive to provide free care to beneficiaries. In an attempt to boost revenue, it may also attempt to give free care to non-beneficiaries, if the reimbursement is higher than the regular price, and to bill the financing agency for it. To avoid cost escalation, the financing agency may have to impose a budget ceiling for the payment of waived services. Capitation, in contrast, consists of the periodic payment to the provider of a fixed amount of money for each registered beneficiary of the waiver program. Since this payment is not tied to the volume of services actually delivered, the incentive of the provider may be to under-provide free care -just the opposite effect of fee-for-service payment. The selection of an appropriate payment system has been the subject of a recent debate in the context of Cambodia's equity funds (Knowles 2001; Bitran 2002). Protection with specific benefits versus unconditioned cash transfers Do the benefits of protection programs come in the form of free or subsidized health services, or do they come in the form of income supplements -cash- which beneficiaries can use freely, on health services or on other consumption? 2.3 Performance of protection mechanisms Assessing the performance of a protection mechanism should be an integral part of a poor-protection policy. Questions to be asked include the following (see Table 3): Are the intended beneficiaries actually protected; has there been an increase in their use of health care services; are some non-poor inappropriately benefiting from the protection? What is the actual administrative cost of the protection mechanism'? How efficient is the protection mechanism, in terms of the ratio between the protection granted and the administrative costs of this protection? What are the outcomes of the program, in terms of better health status? Below is a definition of concepts related to performance and a discussion of performance measurement issues. Table 3 Monitoring and evaluation of waiver and exemption systems Monitoringl and evaluation issues Are perforrnance evaluations conducted regularly or sporadically? X X Is provider compensation system working effectively and is compensation amount in accordanos with the actual cost of service? X X Are prospective beneficiaries actually waived? X Do waived beneficiaries make greater use of services than poor, non-waived individuals? X Do the beneficianes of waivers get financial protection? X Is there an increase in utilization of exempt services? X (a) Errors and accuracy in beneficiary identification A successful beneficiary identification program is one that identifies those in need of protection with precision, neitlher missing any of those in the target group nor allowing any benefits to go to those outside 2. Mitigating the effect of fees 24 of it. Identification is seldom perfect. Those running a beneficiary identification program can make the two following possible types of errors (it is assumed that those needing protection are the poor, as will almost always be the case for social programs such as basic health care): To classify as non-poor someone who is truly poor. This means denying protection to someone who deserves or requires it. This kind of inaccuracy is called a Type I error (also known as a false negative) or "under-coverage". Type I errors can arise from an overly stringent screening mechanism, from the reluctance of potential beneficiaries to be tagged as poor, or from lack of knowledge among prospective beneficiaries about the program. Reducing Type I errors promotes equity by improving access to social services by the poor. To classify as poor someone who is not poor. This implies that protection is awarded to non-target individuals. This type of mistake is generally referred to as a Type II error (also known as a false positive), and more commonly called "leakage". Figure 10 summarizes the four possible outcomes of a beneficiary identification effort. The accuracy of the method can be calculated in terms of Type I and Type II errors. Both types of error can be expressed either as the number of misclassified people or as amounts of money over or under spent. The extent of Type I error or under-coverage is measured by dividing the number of poor wrongly excluded from the benefits by the total number of poor. It can also be measured by dividing the value of benefits wrongly denied by the amount of money required to provide benefits to all of the poor. Grosh (1992a) states "The complement of under-coverage is coverage, or the percent of those who ought to be served who are served. This is sometimes called the participation rate". Type 11 error or leakage is measured by dividing the number of non-poor beneficiaries by the total number of beneficiaries, or the value of benefits wrongly awarded to the non-poor (upper right quadrant) by the total value of benefits (upper left and right quadrants). Figure 10 Identification errors Appropnate- Number of Amount of ness of action people stubsidy given or denied ACTUAL STATUS POOR NON-POOR ALL PEOPLE Correctly given subsidies I y glien subsUes ------ No error Typ11 error 0- EL~~~~~~~~~~~~~N Lu n P. *~~-- . N, P.+N. X _ sp, --------P, + s-Ni s (P,+N,) L Incyrec derded subsidies Correctly denied subsidies Z o U a B Type I wor No error O P2 Nz P24N2 S-P2 s N2 _ L P,<,P2,N,+N2 a. P,+Pz ,N 0 0. 5~~~~~~~~~~~~~~~~~~~~ (P,4N,) saP. sN, Amount of total subsidy required for full coverage of the poor S = s (P.+P2) Actual coverage of the poor, or participation rate P./(P.+P2) Extent of Type I error, or under-coverage PI/(PI+P2) = s P2Js (P.+P2)) = PZJ(Pi+P2) Extent of Type 11 error N,/(P,+N,) = P,t(P.+N,) = s N.4s (P,+N.)] = N,/(P,+N,) (b) Accuracy-cost tradeoff in beneficiary identification 2. Mitigating the effect of fees 25 In principle. one would like the identification effort to be as precise as possible, ideally reducing leakages to zero while reaching all target beneficiaries. In practice, however, this is seldom possible. The more accurate a beneficiary identification effort is, the higher its coverage (smallest Type I Figure 10 Tradeoff between accuracy and cost of error) and the lower its leakage (minimal targeting Type II error). Achieving a high degree of Q accuracy however requires costly t o information gathering and compliance E targetmgI efforts to identify intended beneficiaries f j Program benes and exclude the rest. s I tet pou ion This tradeoff between the identification _ T I effort and the incidence of the program is illustrated in Figure I0.'° Consider a c , subsidized health program, worth $T E Program benefits million, dehvered without any targeting A going to t ta t effort. Universal, free-of-direct-charge I provision ofigovernment-financed services is an example. In the absence of a MlmwII M MSt3m targeting effort, part of the subsidy, or the Targeting effort amount A in thte figure, will reach the intended population group (defined, for example, as the poorest 40 percent of the population); and, because of a lack of targeting, part of the subsidy, equal to Be, will leak out to the non-poor, such that A+B=T. If an effort were made to identify target beneficiaries -for example those below a certain income or socioeconomic status threshold- the amount of subsidy reaching the poor would increase from its original value of A, while the subsidies reaching the non-poor would decrease from the initial value of B. But at the same time, a new cost, the administrative cost of the beneficiary identification program, begins to eat up part of the total subsidy amount available. For a while, the greater the identification effort, the larger the volume of subsidies that reaches the poor, the smaller the share of the subsidy that filters to the non- poor, and the higher the share of the total subsidy that is devoted to the identification effort. If the effort were maximum (right-hand side of the figure), that is if somehow program personnel were able to identify with precision all target and non-target beneficiaries, then there would no longer be any leakage of program resources: subsidies would be divided into benefits going to target beneficiaries and costs of the targeting effort. But with a decreasing marginal return of identification, engaging in the highest possible identification effort clearly vvould not be optimal, because the share of program benefits going to the poor would not be maximal. A smaller beneficiary identification effort, or optimum, denoted by M in the figure, would be preferred: a maximum of benefits equal to the distance MN would go the to poor, a total of resources equal to NL would leak to the non-poor, and an amount equal to LQ would be eaten up by the identification effort." Both accuracy and costs determine the efficiency of beneficiary identification. Thus, the most accurate strategy is not necessarily the most efficient because it might entail high costs. The "output" to be produced as efficiently as possible is defined by the policy objective (for example, providing the biggest transfer to the poor). In this case, the most efficient targeting strategy is the one that transfers the largest 10 This scheme has been suggested by Willis (1993) It has bieen suggested that to gain political support for a targeted program sometimes it is necessary to allow some leakage. 2. Mitigating the effect of fees 26 portion of a fixed budget to the poor. This requires minimizing the sum of targeting costs and costs in the form of leakage to the non-poor. 2. Mitigating the effect of fees 27 -~~~~~~~~~ 3 Empirical evidence on protection mechanisms 3.1 Introdujction Chapter 1 was a conceptual discussion of user fees in health and their likely effect on the poor. Chapter 2 examined theoretical ways of preserving equitable access to health care when user fees are in place. This third chapter offers information from seven countries on their actual experience with protection mechanisms. Criteria for choosing a case include novelty, success or failure of protection mechanisms, diversity in socioeconomic circumstances, and availability of data. Gathering sufficient information to write a case was a demanding task, as sources are dispersed and often informal. Most cases were written on the basis of multiple sources, written and oral. Table 4 presents a list of selected indicators for case study countries, with the countries organized in ascending order of purchase-power-adjusted per capita GDP (1997). The cases that follow are presented in the same order, each with the following common structure: (a) context; (b) user fee policies; (c) protection mechanisms; (d) results (e) lessons leamed for case study countries and for others wishing to implement protection mechanisms. Table 4 Selected indicators, case study countries* Kenya Cambodia Ghana Zimbabwe Indonesia Thailand Chile Demographic indicators Total population 2001 (millions) 298 13 1 19 9 11 4 206 1 624 154 Life expectancy at birth 1999 (years) 47 7 53 7 57 9 40 4 65.7 6&6 75 5 Urban population 1999 (percent) 32 2 15 6 37 9 34 6 39.8 21 3 85 4 Economic indicators GDP growth raite 1997 (percent) 2.1 1 0 4 2 2 8 4.7 -1.5 7 4 Per capita GDi 1997 (1995 PPP $) 1,020 1,330 1,760 2,680 3,130 6,440 8,440 Poverty indicators Year 1992 1999 1992 1990 1999 1998 1994 Population uncler poverty line, Total (percent) 42 0 35 9 31 4 25 5 23 4 12 8 23 2 Rural 469 252 343 310 195 172 218 Urban 29 3 40.0 26 7 10.0 26 1 1 5 30 6 Health expenditure Per capda health expenditure 1997 (current PPP $) 81 96 54 186 55 347 533 Private health expenditure as percentage of total 1997 35 9 90 6 53 38 2 47 4 65.4 48 6 Structure of health expenditure 1997 (percent of GDP) Private 54 64 1 6 3.7 1 1 39 3 1 Public 2.4 0 7 1.4 2 9 0 6 1 8 3 Total 78 71 30 66 17 57 61 Organized in ascending order of PPP-adjusted per capita GDP in 1997 Data from World Bank World Development Indicators, 2000/2001 and WHO Report 2000, unless otherwise noted Data on poverty for Chile was obtained from USAID, 1998 3. Empirical evidence on protection mechanisms 28 3.2 Kenya12 (a) Context Kenya is the poorest country of the selected case studies. Forty-two percent of its population lives below the national poverty line (World Development Indicators, 2001). In 1997, total health expenditure was 7.8 percent of GDP. Kenya relies heavily on private expenditure which represents 70 percent of total health expenditure. A severe economic crisis in the early 1990s led to a steady decrease of government allocations to health, from $29 per capita in 1990 to $17 in 1993. By 1998, total public per capita expenditure in health had returned to its pre-crisis level of $30. (b) User-fee policy and its impact on the poor Beginning with independence in 1963, Kenya committed itself to free public education and health care. Owing to the economic crises of the late 1980s, however, in 1989 the country introduced cost-sharing policies to raise revenues in outpatient wards in hospitals and in health centers. Over time, user fee revenues have increased substantially from Ksh. 28 million in FY 1990/91, to Ksh. 325 million in FY 1998-99 (Table 5). In FY 98/99 they accounted for 3.4 percent of non-wage recurrent health expenditures (Owino, 2000).'3 Table 5 Kenya: Cost-sharing revenue as a percentage of Ministry of Health's recurrent expenditure Cost sharing Cost sharing MOH recurrent MOH recurrent revenue revenue expenditure expenditure Fiscal year (Ksh. millions) ($ millions) (Ksh. millions) ($ millions) Percentage 1990/91 28 1 1 3,040 117 092 1991/92 34 1.1 3,500 116 0.97 1992/93 61 1.4 4,580 102 1 33 1993/94 130 2.3 5,900 103 2 20 1994/95 238 4 4 7,420 138 3 21 1995/96 200 3.7 8,280 153 2 42 1996/97 205 3 6 8,760 152 2.34 1997/98 186 3.1 9,100 152 2.04 1998/99 325 4.7 9,620 140 3 38 Source: Owino, 1998. Note. The figures exclude contributions by Kenyatta National Hospital Government health facilities have established official fees that are in principle uniform across the country and set nationally by the Ministry of Health (MOH). In practice, however, responding to declining government budget allocations and inflation, several District Health Management Boards have increased their fees without prior consultation with or approval from the MOH, resulting in wide divergence between official and actual fees (Error! Reference source not found., from Owino 1999). There is evidence that cost recovery may be adversely affecting access by the poor and that in some cases it may have had a negative impact on efficiency of health service provision. A study by Moses (1992) covering a nine-month period shows that the introduction of fees for patients attending Nairobi's Special Treatment Clinic for sexually transmitted diseases (STDs) resulted in a 40 percent drop in attendance by men and a 65 percent drop by women. Another study by Osuga and Nordberg (1993) examines the 12 Information for this case study is drawn mainly from Owino, 1998, 1999, and 2000, IPAR policy brief, 1999 and Newbrander 1995 13 . When Kenyatta Hospital is included this percentage rises to 14 5 percent for FY 94/95 3. Empirical evidence on protection mechanisms 29 "before after" effects of the introduction of fees at a rural ihospital, two health centers and Figure 10 Kenya: Official and actual fees in MOH two dispensaries in Kibwezi Division in facilities, 1994 (KSH) Machakos D)istrict in Kenya. Relative to previous years when services were free, T outpatient attendance dropped significantly Mortuary (ranging from 28 percent to 50 percent) in the Circumcision first six months at all fee charging facilities. Physiotherapy Actual Fee Admissions for inpatient care increased at the Pharmacy-drugs e hospital but length of stay fell, while in health X-ray Code F centers admissions, length of stay, and X-rayCodeA offiaal Fee deliveries declined. Attendance at those Lab Test/Specialized services that continued to be free remained Theatre/Minor Surgery _ _ stable or increased slightly at the hospital, but Normal delivery l _ at health centers trends were mixed, with Matemity ward declining use of under-fives care and ante- natal care and increased use of family Ksh. planning services. The drop in attendance was followed by a recovery, but levels remained 20 percent to 40 percent below pre-fee levels. Similarly, Mwabu and WVang'ombe (1995) show that the introduction of outpatient fees in Kenya's public hospitals brought about a substantial drop in demand. Finally according to a 1995 analysis of household interview data by Newbrander and colleagues (2000), 91 percent of poor households reported that they knew of someone who recently had not sought care because of inability to pay. Sixty percent of respondents reported that finances were a major factor behind the drop in demand in government facilities; 30 percent cited dissatisfaction with the service provider or service quality as a reason for seeking alternative treatments. (c) Protection mechanisms In an effort to mitigate the negative effects of user fees on access by the poor, MOH central level staff introduced a system of exemptions for certain categories of persons or patients afflicted with certain illnesses. Since 1992 exemption categories have been reduced (Table 6). For example, whereas before 1995 children under 5 years were waived from fees in all primary care facilities, after 1998 only about one-half of all facilities kept this waiver. 3. Empirical evidence on protection mechanisms 30 Facility staff (medical superintendents, matrons, Table 6 Kenya: Categories of patients and hospital secretaries,;or administrative officers) illnesses exempted, 1989-98 determines waivers locally and grant them to the poor on the basis of income and health status. 0-15 yrs 0-5 yrs 0-5 yrs1 Initially, local leaders and their assistants were in Prisoners Yes Yes Yes charge of screening and recommending patients Tuberculosis (TB) patients Yes Yes Yes for waivers. This process was deemed prone to Leprosy patients Yes Yes Yes political interference and led to delays in Patients from treatment (Owino, 1999). Also at the beginning charitable/destitute homes Yes Yes Yes of this policy, the MOH published rules for Family Planning Yes Yes - establishing eligibility on the basis of income, STDeAIDS Yes Yes - but the rules proved difficult to interpret and Unemployed (certified by their) implement. Measuring individual and family District Officer Yes Yes - cash income constituted a major hurdle leading Antenatal and postnatal clinics Yes Yes - health staff to use income proxies for waiver Civil servants Yes eligibility. A study of 9 facilities shows that in Unmarried children under age 22 (except for inpatient charges) Yes practice a wide variety of proxies are tused. Also, Source Owino, 1999 facilities seem to combine different patient Notes attributes to determine poverty (for example 'Yes' if category included in the exemption list sometime dunng mode of ranspor to hosptal andcorresponding period clothing, mode of transport to hospital and Children under 5 exempted in 50 percent of facilities visited, number of relatives accompanying the patient, though benefits were restricted to consultations. Plans reportedly Table 7).'4 Some criteria -such as occupation, underway to remove this group from exemption list (Owino, 1999) clothing, and hairstyles- seem common to all facilities, whereas others -number of dependents to family size, relationship with accompanying family members- are used only by a few entities. The staff issuing waivers is Table 7 Kenya: Eligibility criteria for waivers in selected MOH required to fill out a standard health care facilities MOH waiver form and Facility forward it to the medical Attributes relating to the patient 1 2 3 4 5 6 7 8 9 superintendent or his or her Occupation X X X X X X X X X delegate for approval. The Mode of dressing/hairstyles X X X X X X X X administrative process Mode of transport to hospital X X X involved in granting waivers Recommendation by local administration X X X X X Direct observation X X X X X X IS cumbersome, on average Drc bevto Number of dependents/family size X X lasting about 1-2 hours. Nature and type of relatives X X When the local Number and type of accompanying family administration is involved, members x x x x x x the approval process Length of stay after discharge* X X X X sometimes may take up to a Recommendation by social worker x x x Source: Owino, 1998 a full day. Often the process Notes: is hampered by the lack of 'X" means attribute used at facility for eligibility assessment (Owino, 1999) adequate stationery and * When patients did not pay the bill they were not allowed to leave the health facility qualified staff. Nurses, clinical and medical workers, and other professional staff are usually in charge of granting waivers, an activity that interferes and competes with their regular health care duties. The process of assessing and exempting patients is thereby often delayed or postponed. No explicit policy is in place to compensate facilities for revenue foregone due to waivers and exemptions, and thus more waivers and exemptions mean less revenue for the facilities. 14 3. Empirical evidence on protection mechanisms 31 (d) Results There is no systematic government monitoring or evaluation of the performance of protection mechanisms, and therefore there is a lack of data on coverage of the target group. A recent survey of 17 facilities reported by Owino (1999) shows that waivers rarely exceeded 2 persons per month -an insignificant figure given that 42 percent of Kenyans live below the poverty line. Beneficiaries of exemptions and waivers are mainly inpatients and outpatients with simple medical conditions and seldom include patients with costly treatments. This finding may reflect the reluctance of providers to forego significant r evenue though costly exemptions. It also indicates a situation where exemptions are absent where they are most needed: costly treatments have the most detrimental impact on the poor. The low level of exemptions contrasts with the high prevalence of poverty; it reveals the existence of serious problems of under-coverage, and thus points to major deficiencies with this protection mechanism. In a survey conducted by Owino (1998) 80 percent of inpatients and 86 percent of outpatients were not aware of waivers and exemptions. Even those who did know about them were unclear about eligibility criteria. Staff attitude towards waivers and exemptions was usually negative and they have been reluctant to publicize protection mechanisms. Fearing revenue losses, staff felt that information on waivers should not be easily accessible to patients and that patient relatives should assume the burden of fees. In fact, according to COwino, there seems to be a general reluctance by facilities to create awareness about the waiver system. In the same study, about 70 percent of the facility clerks interviewed were against publicizing the system. There have also been reports of leakage, especially in the form of exemptions for civil servants and health personnel (Ow ino, 1999). According to a study by Newbrander (1995), only one-third of all waivers and exemptions were accounted for by the poor, and a full two-thirds accrued to the non-poor. Leakage from waivers through individual targeting did not seem to be a problem. Based on a one-day exit interview, Newbrander reports that 100 percent of waivers granted through individual targeting were given to the poor. Still, theire seems to be other evidence of corrupt practices in the granting of waivers and exemptions C)wino (1999) reports that nurses in charge of wards were known to seek bribes in exchange for helping patients abscond without paying or for getting preferential treatment in the fee-determination process. It seems also to be common practice that health staff benefits from waivers and exemptions alleging that their salaries are too low. (e) Lessons learned User fees resulted in a sizable drop in demand in hospitals, and the poor reported access problems related to the imposition of user fees. Measures to preserve access by the poor through waivers and exemptions are characterized by under-coverage and leakage. Factors responsible for this poor performance include the following: * Lack of clear guidelines to determine eligibility. * Absence of staff incentives to grant and promote exemptions and waivers, including the lack of mechanisms for compensating health facilities for revenue foregone. * The protection system has not accounted for administrative costs involved in granting waivers. The approval and granting of waivers and exemptions is hampered by the lack of adequate stationery and qualified administrative staff -medical social workers. This lack of specialized staff has required that waiver functions be undertaken by nurses, clinical and medical officers, thus interfering with their medical duties and hampering the effectiveness of the waiver procedure. 3. Empirical evidence on protection mechanisms 32 * Lack of information and misinformation among potential beneficiaries about protection mechanisms. * Perceived low quality of services by all prospective patients, including the poor. * Ability of health workers and other non-target groups to capture subsidies through waivers and exemptions. 3.3 Cambodia (a) Background Cambodia is emerging from decades of civil war and unrest. It is estimated that between one-fourth and one-third of Cambodia's population died during the war, from violence and starvation. Since the early 1990s Cambodia has begun to rebuild its economy, achieving significant economic growth rates, and the government has pledged to improve the provision of social services. Still, Cambodia remains a very poor nation with a national prevalence of poverty of 36 percent and with a high concentration of poverty in rural areas. Meager public budgets for health have done little to reduce reliance of user fees in the public system and to alleviate the burden that fees represent for the poor. (b) User fee policy Cambodia's health sector relies heavily for its financing on private payments for health care. Knowles (2001) states that Overall health sector funding in Cambodia absorbed 12-13 percent of GDP in 1996-97, by far the highest share among Asian developing countries. Another striking feature of the financing of Cambodia's health sector is the large role played by out-of-pocket household expenditures, which accounted for 82-84 percent of total sector financing during the same period. In contrast, the government plays a relatively minor role in sector financing, accounting for only about 4-5 percent of the total. Official donor assistance (ODA) and direct funding by NGOs also contribute significantly to sector financing, accounting for about 8-12 and 2-3 percent respectively of total sector financing during this period. According to the 1996 Health Care Demand Study, on average health care accounted for 22 percent of all estimated household expenditures. This percentage increased with income, except for the poorest families that devoted 28 percent of their total spending to health care. In 1996 the monthly mean expenditure on health care per household was estimated at $13.90 (MOH / WHO / GTZ, 1998:20); adjusted by internal inflation this was equivalent to $27.10 per person per year in 1 999.15 The World Bank (2001) carried out an analysis of affordability of government-provided health services, by expressing health care spending in relation to household non-food expenditure per capita (a proxy for discretionary household income). They concluded that health care was extremely expensive for the poor. One outpatient visit to an ambulatory government facility cost $30.00, or the equivalent of one-third of a year's non-food spending for those in the poorest expenditure quintile; one hospital admission cost on average $87.00, or more than twice the average annual nonfood spending. In 1996 the MOH introduced the National Charter on Health Financing (NCHF). This initiative sought to formalize cost recovery in the form of user fees around the country. The NCHF also made possible the 15 . The Cambodia Socioeconomic Survey of 1997 found that health expenditure per household was US$18 60 per month, equivalent to a per capita out-of-pocket expenditure per year of US$42 3. Empirical evidence on protection mechanisms 33 development of health financing pilots (with user fees or other financing mechanisms) in public facilities, with the explicit purpose of "increasing financial resources to the sector and obtaining better value for money" (National Charter on Health Financing in the Kingdom of Cambodia, 1996:2). Three specific goals were implicit in this statement: to reduce unofficial charges and household out-of-pocket expenditures; to improve quality of care through increased and timely availability of medical supplies; and to motivale staff through performance-related payments funded by fees. In accordance with the third goal, the government defined a revenue allocation rule whereby 49 percent of user-fee revenue could be used to improve salaries, 50 percent could be devoted to pay for the facility's non-salary operating costs, and a nominal 1 percent must be transferred to the National Treasury. Under the HCFC, therefore, facilities retain and control locally 99 percent of all user fee revenue. (c) Protection mechanisms Exemptions from payment in health facilities are not common in Cambodia, although most public facilities do provide them according to some ad-hoc procedures. The World Bank (1999) reported that in 1997 only about 18 percent of users of health care services were exempted from fees around the country, and noted that individuals from higher income households were more likely to get exempted from user fees than the poor. A major obstacle to acceptable levels of exemptions, i.e., ones consistent with Cambodia's high levels of poverty, is the conflict that exists between health staff income and their awarding of exemption to patients. The average staff of a government health facility earns a monthly salary of between $10 and $15. This income level is below the poverty line. To subsist, government health workers depend heavily on revenue from user fees and from other remunerated activities. In interviews with facility health staff in three provinces (Takeo, Sotnikum, Kandal) and the capital city of Phnom Penh, Bitran (2002) found that owing to cost recovery from user fees, the average monthly income of a government health worker may be as much as $180, of which less than 10 percent comes from its official salary, and the rest from user fees. Wilkinson et al. (2001) note the following: There is an inherent tension in a facility seeking to operate a viable exemption scheme and a viable salary incentive scheme. The two systems are essentially in competition, especially in facilities which are operating at, or near to full capacity. In this situation, every exemption provided is effectively paid for by the staff themselves from their salary uplift. If, as in Takeo, the hospital is operating at full capacity, and is striving towards improving efficiency, then granting exemptions would be virtually intolerable Ironically, exemptions are more likely to be granted in facilities which are performing less well, and where serving a patient for free does not necessarily mean excluding a fee-paying patient. The competition, outlined above, between patients for more exemptions and staff for greater salary incentives is actually part of a broader systemic tension, inherent in the design of the health financing scheme, between equity and efficiency. There is a real danger that increasing equity, by lowering prices and providing more exemptions to the poor, will undermine efficiency, both at facility level and [Operation District] O.D. level. Conversely, as facilities strive towards greater efficiency, there is a real danger that the poor will become even more marginalized. If the tensions outlined above are to be relieved, it is clear that the mechanism for financing exemptions must be completely separated from the mechanism for financing salary supplements and operating costs. Exemption policies vary widely among provinces and districts in Cambodia (Espinosa and Bitran 2000). For example in 1999 one-fourth of the population under the responsibility of Rovieng health center was waived from fees. By contrast, at Pereang operational district the rule is that no waiver is granted, except in very special cases. The policy, there, is to keep fees low enough to make everybody pay while at the same time avoiding problems of financial accessibility for the poor. In Takeo Hospital, the rate of waivers was estimated as 2 percent while at Pursat Hospital the value of waivers accounted for 13 to 15 percent of cost recovery revenue (Pursat Report on user fee payments, January 2000). Aside from the system of informal waivers already discussed, two interesting formal models exist in Cambodia for promoting equitable access to health services by the poor: the Calmette Hospital model and 3. Empirical evidence on protection mechanisms 34 the equity fund (EF) model. Calmette Hospital, also known as the National Hospital, is a semi- autonomous, 250-bed public facility located in Phnom Penh that provides health services free of charge to the poor and that gets reimbursed for it by the government on the basis of a fixed payment of $26.00 per hospitalization. A special mechanism known as Chapter 31 (a budget line item reserved for social allowances for health staff), operating exclusively in Calmette, makes this protection system possible. Calmette has a formal fee schedule for non-poor patients and a formal means testing mechanism for classifying patients into the paying and exempt categories on arrival. Calmette devotes about one-third of its beds to indigent care. Occupancy rates are 65 percent for paying beds and 100 percent for the beds devoted to the poor. Calmette Hospital also receives financial and technical support from the French Cooperation, an agency that is currently seeking government approval for expanding the Calmette model to six other hospitals, of which two are in Phnom Penh and four in the provinces. Unfortunately, tight public budgets make this prospect unlikely and government officials seem unwilling to generalize or expand this system. Equity funds were conceived in Cambodia to finance the cost of health services provided at no charge or at reduced prices to the poor. Unlike the Calmette model, which so far remains a single exception, the EFs model is likely to become more widespread. An upcoming World Bank loan may promote and finance EF expansion. The basic mode of operation of an EF is illustrated in Figure 11. The process of setting up an EF begins with the submission by the health facility to the MOH of an application for approval of its health financing charter, including its fee schedule (Action 1). Once the charter is approved by the MOH (Action 2), the facility officially adopts its fee schedule (Action 3), including some criteria to waive the poor. The MOH, in turn, quantifies the budget for the health facility, in principle taking into account the provider's expected ability to generate complementary revenue from users (Action 4). Patients arriving in the facility (Action 5) and wishing to be waived from fees are subject to a means test to determine their eligibility (Action 6). Patients applying but found not eligible for waivers, along with patients not applying for waivers (Action 7), are subject to and must pay the provider's customary fees (Action 8). Exempt patients, instead, are offered care for free or at a reduced price (Action 9). Periodically the health facility reports to the EF on the level of waivers provided as well as on the monetary value equivalent of the subsidized services (Action 10). For example, the provider may keep a record of all services delivered for free and then, based on the user fee revenue foregone, at established prices, it bills the EF. The latter in turn reimburses the provider, after controlling and approving the statement submitted by the provider (Action 11). The EF requires a periodic refill of its fund (Action 12), which gets depleted with the reimbursements to the provider. EF financing has been, until now, the role of donors (e.g., the Swiss Red Cross in Takeo Hospital, the U.K. and Save the Children in Sotnikum, and the U.K. in the Phnom Penh Urban Health Project), but there is nothing in the design that would preclude the government from financing EFs. Not all EFs operate in exactly the same way; there are variations which may have important behavioral implications and consequences on the performance of the health system. They include the management of the waiver process; where lies the responsibility to establish eligibility; the method used to pay the provider; the insertion of EF in referral system; and the type and extent of financial protection. Waivers may be managed either by health staff, as in Takeo Hospital, or by the combined and coordinated work of health staff and EF clerks, as in Sotnikum and the Phnom Penh Urban Health Project (PPHUP). In Sotnikum, for example, health staff in the referral hospital first detects signs of indigence among patients, and then refer these patients to the EF for a means test. Likewise, the responsibility to determine eligibility may be in the hands of EF clerks, health staff, or both. In Sotnikum the ultimate decision about the granting of waivers rests with a specially trained professional who carries out a formal means test of applicants. Also in Sotnikum EF staff visit patients in their homes after discharge, to confirm the information provided during the means test. 3. Empirical evidence on protection mechanisms 35 Figure 11 Cambodia: Operation of equity funds (4) Budget for Health Facility (15)1% M iMlnistry of Health | (1! p Ministry of Finance 1(2) Approval of Health (1t ) | Financing Application C Charter Approval of F ( F -lealth I _F___ _ _e_ Financing X Schedule Charter, (14) 49% Salary Inclmding Supplement for Health Fcedue 7) Payin (8) user Staff Schedule ati ents Fee Government F [ ~ < Payments Health Facility I 0 t/ (13) 50% Revenue rimIllI~ ~ upleier for Non- I ~ ~~~~~~ _ _ ~~~~~Personnel Oper'ating Equal Health Costs ! _/(9) Exem pt __ Care for All ( 0) Monetary value of exemptions (user fee revenue forgone) I (11) Rembursement to health facility of amount (6) | equal to value of exemptions ((5) Patients Means Equity Fund Test (12) Funding of Equity Fund Source Bitran 2002. Providers may be given a fixed budget to finance waivers, as in Takeo or in the PPUHP, or they may get paid by the EF on a fee-for-service or per case basis, as in Sotnikum. The EF may promote patient flow through the referral system as in PPUHP, where patients seen in health rooms who need a referral get a waiver for the fees associated with health services required at the higher level. In Sotnikum, instead, as in Takeo, exempl ions are granted when patients show up in the hospital, because the EF there does not operate at, or in coordination with the primary care facilities. The extent of protection afforded by the EF varies as well, in terms of the types of services it covers and the proportion of the services' costs it pays for. Thus, an EF could in principle provide protection against catastrophic health problems (none does so far), or it could finance all or part of the costs of an episode of illness. All three EFs studied by Bitran (2002) provide financial protection for high-cost services in hospitals, but they have implicit or explicit limits in what they cover and therefore may not be providing true catastrophic coverage. Aside from the extent of coverage of health expenditures, EFs may or may not cover other health-related expenses, such as transportation to and from the health facility, food for the patient and family members, and the like. Hardeman (2001) examined this issue, among other aspects of EF, through a case study in Sotnikum. He examined a sample of 51 individuals who had been hospitalized in June-July 2001 to assess patient expenditures and level of patient financial support, if any, by Sotnikum's EF. He found that among the poor and the very poor, the highest expenditure associated with a hospitalization was food, on average varying in the range 38-42 percent (Figure 11). Hospitalization fees were the second most important patient expenditure, representing about 32-35 percent of all expenses. Transport costs were the third most important category and accounted for just below 20 percent. Other expenses represented 8 percent. Whereas Sotnikum's EF supported all four kinds of patient expenditure, the structure of its support did not match that of patient spending: one-half of the aid was directed toward hospitalization fees and one-third toward transport expenses. Only 5 percent of the average support to patients was devoted to food outlays. In absolute terms, the average support per patient amounted to $10.00, and covered the bulk of hospital fees and transport costs. Total 3. Empirical evidence on protection mechanisms 36 average expenditure by the poor and the very poor amounted to $25 and Figure 11 Cambodia: Mean patient expenditure per $18, respectively, and therefore hospitalization and average support by Sotnikum equity Sotnikum's EF covered 40 percent of fund, 2001 ($) average spending by the poor and 56 percent of spending by the very poor. 30 (d) Results 25 Knowles (2001), who conducted an 20 evaluation of the PPUHP EF, l 8% oOther concluded that the project effectively 32 l l Transport protected the poor against the high EFee costs of health care, and also 8Food prevented people from falling into poverty as a consequence of high E l health care costs. He also estimated 5 42%- that the EF was a fiscally efficient 5%' b"" policy, because the costs of running 0 the EF were smaller than those o Very poor Average support associated with poverty alleviation for those who, as a consequence of health care payments, would otherwise fall into poverty. Knowles also noted that the costs of running the EF in PPUHP were high because of the high incidence of expatriate managers costs. Hardeman (2001), who evaluated the Sotnikum EF, provided an equally upbeat vision. He found that the EF improved equity in access to health services by not discriminating in the provision of care between the poor and the non-poor. The cost of supporting each hospitalized patient was, on average $10, making it possible for poor and near poor patients to receive medical services worth $50 ($10 co-financing by the EF; $40 financed by the government on average, through the support of recurrent hospital expenditures). He also found that there was virtually no leakage of benefits to the non-poor. He concurred with Knowles' finding about the EF's ability to prevent poverty from high health-related expenditure households vulnerable to poverty. (e) Lessons learned Hardeman (2001) identified a number of limitations in the Sotnikum EF model. First, he noted that the EF was passive in the sense that it did not actively search for the poor in need of assistance, but instead waited for them to show up. This likely resulted in under-coverage of the poor, a phenomenon that may be overcome gradually as the EF becomes known to the local residents. He reported that an estimated 30- 40 percent of the population was poor or near-poor in Sotnikum district, but only 15-20 percent of the hospitalized patients fell in those two categories. He also concluded that there was a lack of awareness about the EF, noting that the total number of patients so far receiving EF support (309 people) was still small in relation to total beneficiary population of 220,000. He felt, however, that awareness would improve in response to current community involvement in EF promotion. He also remarked that the poor were still subject to financial uncertainty associated with health care demand, since the outcome of the means testing procedure was unknown to them prior to seeking care. He remained concerned about the EF's ability to improve accessibility and to have a noticeable impact on use by the poor. He also worried also about the replicability of the EF, particularly its management structure and its methods for identifying recipients and paying providers. A final concern was the ability of future EFs to control and maintain appropriate levels of quality of subsidized health services. 3. Empirical evidence on protection mechanisms 37 In his evaluation of the PPUHP EF, Knowles (2001) recommended that the existing caps in the size of EF support of patient expenditures be removed, that case-based be maintained as the reimbursement system for providers, that a proposal for the adoption of a flat fee reimbursement be rejected, and that the EF reimburse providers at full cost instead of the current 70 percent of cost. He also advised that the means testing procedures used in the EF to avoid type I errors (wrongly denying benefits to the truly poor) be discontinued. Other recommendations included the addition to EF benefits of vouchers for home-based obstetric care; the provision of patient partial assistance for referrals to all higher-level facilities; the continuation of provision by the EF of supplementary welfare payments to individuals to encourage the continuity of care in higher level facilities; the turning over of EF management to a local NGO; and the future opening up of the EF to the participation of the non poor in exchange for a monthly fee. Main lessons arising from the Cambodia case follow: * EF seem to be an effective mechanism for targeting assistance to poor individuals in need of health care in Cambodia * Private patient costs of care other than the health professional's fees and medicines can be quite substantial; to be effective, EFs should also contemplate paying for such costs as transportation and food for patients and accompanying relatives. e The initial operation of EF has been characterized by under-coverage. Further dissemination of e EF may lead to greater demand for their assistance. * The provision by EF of waivers at the time of care involves uncertainty for prospective beneficiaries who do not always know their chances of getting a waiver. This may limit demand for EF assistance and for health care by the poor. The possibility of distributing waivers ex-ante at the household level should be considered by those designing EFs. * Paying providers for the medical services delivered to EF beneficiaries seems a key factor in assuring access by the poor to timely and good quality care. Such a payment confers an economic benefit to providers which makes them indifferent between treating EF beneficiaries and regularly paying patients. Thus, they do not discriminate against patients receiving EF assistance. 3.4 Ghana15 (a) Backgiround Ghana's population of 20 million has a PPP-adjusted per capita income of $1,760. In 1999 per capita health expenditure was $54.00 (PPP-adjusted) and total health expenditure amounted to 3 percent of GDP, with 61 percent being privately financed mainly through out-of-pocket payments by households. According to the World Bank (1999), one-third of Ghana's population lives below the national poverty line. In recent years Ghana has undergone economic difficulties that have led to severe budget constrains and restricted resources for health financing. In the 1990s government health spending fell both in real Based on Garshong Ansah, et al., 2001; Nyonator and Kutzin 1997, Nyonator, Daimenu, Amedo, and Eleeza, 1996; and Coleman, 1997 3. Empirical evidence on protection mechanisms 38 terms and as a percentage of total public expenditure (from 11.1 percent in 199117 to 4.7 percent in 1998).18 (b) User-Fee policy In 1985 the government formalized the policy of user fees with the release of Law 1313 on fee regulations. The law specified fees for most services and called for full cost recovery for drugs. In the following years, drug fees were adjusted annually by inflation whereas all other fees remained nominally constant and therefore became increasingly outdated. In the time that has elapsed since the inception of fees, the real value of official non-drug fees has dropped by more than 90 percent (Nyonator, 1997). Prior to 1983 any revenue generated from user fees by government health institutions went into a central government account. Since 1983 health facilities have been allowed to retain and spend locally their user fee revenue. Proceeds from fees are kept in two separate bank accounts: one pays for drugs and another pays for non-drug staff expenses. Neither account can be used to supplement staff income, an explicit policy measure seeking to devote fees to the financing of non-personnel quality improvements. (c) Protection mechanisms Law 1313 provided for waivers for "paupers" but failed to give an explicit definition of such a term and to include guidelines for the identification of paupers. The law also called for waivers and exemptions for health workers, for some preventive and curative services for women and children (immunizations, pre- and postnatal care), and for patients with tuberculosis, leprosy, and mental disorders. In addition, the law contemplated partial exemptions for patients suffering from a wide range of communicable diseases. Since the publication of Law 1313 the target groups and qualifications for exemptions have been broadened by policymakers in response to emerging health issues and policy priorities. Groups added to the list of the waived are children under five and people over the age of 70, while conditions added to the exemptions list are snakebite, rabies, and buruli ulcer (Grashong et al., 2001). The law thus considered a mix of targeting mechanisms including targeting by type of service, group targeting, and individual targeting (the "pauper"). Individual assessment of ability to pay is to be undertaken by the social welfare officer at the hospital who is expected to carry out a means test upon admission. Facilities are reimbursed for waivers and exemptions. The national level transfers funds to the regional level from where they are allocated to districts, and from there to health facilities. (d) Results The combined effect of a drop in government funding and a real reduction of fees due to inflation led public health providers to adopt their own user fee policy as in Kenya (see above). In a sample of four government district hospitals (Table 8), actual fees were systematically higher than official fees and large variations in fee levels were observed across facilities.'9 17 Nyonator, 1997. 18 World Bank, 1999. Adapted from Coleman, 1997. 3. Empirical evidence on protection mechanisms 39 In addition, unofficial, or "under Table 8 Ghana: Official and actual fees charged in the table" payments became a government district hospitals, 1997 (cedis) common practice to cover resource Maximum shortages. A study undertaken in actual fee Ghana's Volta Region found that Actual fee over the median user payment in Actual fee over official minimum gos Price category Official fee (mean) fee actual fee government hospitals was 8 times Outpatient higher than the officially authorized department 50 350 7 0 2 5 fees. Although actual fees seem Hospitalizations 100 325 3 3 2 5 well above the official amounts, Deliveries 100 2,000 20.0 3 0 they are below official fees if these X-Rays 200 2,000 10 0 1.0 were adjusted by inflation. Multiple Hemoglobin I0 475 47.5 1 3 fees were charged for different Urine tests 40 575 14.4 1 6 services during a single patient visit, Stool tests 20 475 23.8 1 3 leading to administrative workload, Cesarean 1,000 55,000 55 0 20 0 especially for nurses. In hospitals, Appendectomy 1,000 55,000 550 220 deposits were asked of patients prior Hernia removal 500 28,333 56.7 10 0 to admission. Data from 15 Source Adapted from Coleman (1997) facilities located in the Volta Region showed that user-fee revenue accounted for about two-thirds of non-salary revenue in public health centers and for more than 80 percent of non-salary revenue in public hospitals. According to more recent information, user fees constitute a substantial part of revenues for many hospitals and, by 1999, represented about 12 percent of total funding for public hospitals, up from 8 percent two years earlier (Table 9). 2(1 These figures illustrate the current high dependence of government health facilities on user- fee revenue to finance non-salary costs. (e) Protection mechanisms Table 9 Ghana: Revenue by funding source (in million cedis) Even though no systematic Distribution moiosDistribution 1999 monitoring systems exist, studies Source of Funding 1997 1998 1999 1997 (percent) (percent) indicate that eKemptions were seldom MOH 139.5 195 0 238 2 43 57 awarded. According to MOH data User fees 27.7 330 50.7 8 12 (Nyanator etal., 1996), only the blind Health fund 64 12 6 32.9 2 8 or the mentally handicapped were Earmarked 6 8 78 13 9 2 3 deemed poor and were therefore Donor managed 43 5 34 9 43 1 13 10 exempted. Patient exemptions on the Donor total 566 552 1049 17 25 grounds of inability to pay were Credits 101 6 77 0 27 8 31 7 uncommon. A survey of the Volta Net total 3254 3602 4216 100 100 Region revealed that 5 out of 24 Source Adapted from Coleman (1997). health facilities provided no exemptions wlhatsoever. The total number of waivers and exemptions represented less than I percent of the total number of recorded patient contacts, and the most common reason for exemptions (71 percent of the total) was that the patient was a health worker. Patient exit interviews confirmed the low coverage of protection mechanisms: of 313 patients selected at random, fewer than 2 percent were fully exempted from payment. A more recent survey of selected areas in Ghana confirms this finding: "Household heads were asked whether they or any member had ever been treated free at any government health facility. In the 200 houselholds studied in AMA (name of the region), only six (3 percent) had someone treated free (i.e., waived or exempted). In the rural sample, only three (less than 2 percent) had ever received free 20 Reference provided by F Decaillet, World Bank 3. Empirical evidence on protection mechanisms 40 treatment. In the former case, 24 households have someone who had been treated free but for whom someone had paid (their own or their spouse's employer or)"(Saprin, 2001). Recent reviews by Garshong (2001) and Coleman (1997) show that health facilities are generally reluctant to waiver or exempt patients and charge fees for services that are officially and universally free of charge (Table 10 and Table 121). This once more illustrates the widespread disregard that exists among public facility managers for national policies on fees and exemptions. Table 10 Ghana: Percentage of patients paying full user fee by patient category and by region Patient category Pregnant Child Curative care Care for those women Welfare children under five aged 70 and Other specific Paupers (percent) Clinic (percent) over Tuberculosis diseases Region (N=20) (N=149) (N=35) (N=129) (N=103) (N=39) (N=28) Ashanti - 100 100 100 60 67 100 Brong - 50 38 100 63 0 40 Central - 50 100 75 13 33 - Eastern 0 82 100 50 73 0 100 Greater 25 93 100 94 63 43 100 Accra Northern 33 18 0 24 8 - 43 Upper 50 24 0 56 44 0 0 East Upper 50 42 0 31 9 33 20 West Volta 0 56 - 100 31 0 / 6 50 Western 100 40 60 100 50 33 - Total 40 56 51 71 41 28 36 Source: Garshong et a/., 2001 Seven factors seem to explain low coverage of protection Table 11 Ghana: Proportion of mechanisms in Ghana. First, exemption categories were not facilities charging for exempted well defined by health authorities or well understood by services health personnel and various local interpretations of the law Disease officially exempted District Hospitals existed. Below is a list of common definitions of the term Measles 10/11 "pauper" according to a recent survey by Garshong (2001): Typhoid 11/11 Hepatitis 11/11 * Someone who has no relatives Tetanus 11/11 * Someone who has no means of looking after Sickle cell 11/11 him/herself Source. Garshong et al, 2001 * A disabled person with no money and no family * Someone who cannot pay, who is poorly dressed, and who has a chronic illness * Someone who has no traceable relatives and is financially not capable of paying for services received * Someone who is classified by the priest or by management as unable to pay the required amount Second, many potential beneficiaries were unaware or misinformed about waivers and exemptions and even facility staff did not have a full understanding of the policy. The following statement collected by 21 Coleman, 1997. 3. Empirical evidence on protection mechanisms 41 Nyanator and his colleagues during focus group discussions gives a sense of the high degree of public's misinformation. "To be fair and frank, I am not aware of any such facility [that grants waivers and exemptions for the poor]; however, I am aware that hospital bills of government workers and their dependents are refundable by the government." (Nyanator, 1997) In a more recent study by Saprin (2001) based on a sample of 33 patients who visited a public facility for treatment, he asked whether some patients knew they were not expected to pay. Virtually all of them (94 percent) were unaware of their potential right to exemptions. Also, Garshong' eta!. 's 2001 study alluded to above shows that whereas facility staff seem to know patient waiver categories (for example the blind, the handicapped), they do not fully understand what type of services are to be exempted (for example immunizations). Focus group discussions with pregnant women held within a facility reveal major variation in the interpretation of this policy. Third, most facilities have not made any institutional arrangements to identify poor patients. A reason given by health staff for this was the lack of social welfare workers or of specialized staff for carrying out individual assessment. Fourth, there vvas no supervision or monitoring of the process of exemptions and waivers and there were no sanctions or penalties to those who did not follow the guidelines. Fifth, there are financial and administrative difficulties in the reimbursement process for those facilities granting waivers and exemptions. In 1999, the Govemment of Ghana was able to provide only about 22 percent of the amount need to reimburse public providers for free care (Coleman, 2000). Also, there are serious delays in the flow of free-care compensation funds at all levels of the system. As can be seen from Table 12, facilities may have to wait between I month and I year (on average 4 months) to get reimbursed for free care. Districts also had to wait up to year. Staff interviewed noted the following: 'The government asked us to give exemptions and they [told us] they will refund later. We found that the money was not coming, so we stopped. [...] It affects our purchasing power: consumables and drugs get out of stock." (Garshong, 2001) In addition, facility staff complained about increased workload and stationery costs related to exemption and waiver procedures. As one health provider put it: 'There is extra work of recording exemptions in the lab. It is extra work so someone should be employed to be doing it. We sometimes have to stay longer to work on the recording at the lab, no one pays us ior it and I haven't told anyone to pay for that." (Garshong et al, 2001) Sixth, this situation, together with the increasing Table 12 Ghana: Average duration of fund reliance of facilities on user fees for their financing transfer at various levels, Jan-Dec. 1999 seems to have created strong incentives to emphasize (months) revenue collection rather than the protection of the Region Districts Institutions poor. Ashanti 5 4 3 Greater Accra 12 12,75 4,5 Seventh and finally, there seem to be problems of Eastern 10 4 4 leakage. For instance, since the late 1980s facilities Western 6 7 6 have been exempting not only health facility staff but Brong 1 3 4 other, non-target people have been getting waivers or Volta NA 4 5 reimbursements for their health care payments, Central NA NA 2,4 including civil servants, employees of para-statal Upper East 3 2 3 firms, members of public boards, and the staff of large Upper West NA NA 3,1 corporations (Saprin, 2001). Northern 1 2 4 Source: Garshong et at., 2001 3. Empirical evidence on protection mechanisms 42 (f) Lessons learned Coverage of the mechanism to protect the poor under cost recovery remains low in Ghana. The following are the main factors explaining this situation, many of which are relevant to policy makers wishing to implement or strengthen mechanisms to protect the poor under user fees: * Insufficient funding for the reimbursement of providers granting waivers and exemptions. * Inconsistency of objectives at the facility level arising from a tradeoff between the objective of protecting the poor through exemptions and waivers and generating user fee revenue to pay for local costs of health services. * Excessive bureaucracy and untimely reimbursement of providers. * Inadequate explanation of the poor-protection policy by the central to local level implementers and lack of staff training on waiver and exemption procedures. * Negative attitude of staff towards policies for protecting the poor as waivers mean less income and more work. * Lack of knowledge about exemptions and waivers among potential beneficiaries in the population. * Erosion of user fees by inflation and associated spontaneous and unregulated attempts by providers to update fees. * High and uncompensated administrative costs for facility staff engaged in leading. 3. Empirical evidence on protection mechanisms 43 3.5 Zimbabwe22 (a) Background Although Zimbabwe is not the poorest of the countries in the case study sample, it is the one with the lowest life expectancy at birth (LEB). Zimbabwe's population of 11.4 million has a LEB of only 40.41 years. In 1997 per capita income was $2.680 (PPP-adjusted) and per capita health expenditure was $49.00, representing 6.6 percent of GDP. According to the World Development Report (2000), about one-fourth of the population lives below the national poverty line. One person out of three living in rural areas stands below the poverty line and one-third of all children exhibit second and third degree malnutrition. Moreover, currently Zimbabwe is among the world's hardest hit countries with the AIDS epidemic, leading to a surge in demand for AIDS- related treatment. Real government spending for health increased rapidly in the economic boom years following independence, and then grew only 4 percent annually from 1982 to 1988. In 1990/91, health spending peaked at 6 8 percent of the national budget. On a per capita basis, real health spending reached Z$49 (about US$20) in 1991, among the highest in Africa (Error! Reference source not found.). In 1991, the government launched its economic structural adjustment program just before the worst drought of the century struck. The combination of drought and inadequate implementation F of the adjustment program contributed to igure 11 Zimbabwe: Trends in government health a 30 percent decline in per capita health spending from 1990/91 to 1992/93.23 $50 . ' ' /-- 8 00 S $46 (b) User-fee policy 700 $42 * 0. At independence, government a $ eliminated user fees for all those earning . $38 - 600 less than Z$ 150 per month. Fees were s reintroduced in 1985 but inflation and $34 weak enforcerment eroded the real value of fees. In early 1994 fees were $30 5 00 m 0 T 0) CD CD 0 0) increased by an average of 2.5 times in da X 4 A 4 us z c) o ( 0) 0) CD CD CD 0) nominal terms, with the largest increases - X - _ - - at tertiary level care facilities and in _ _ __ _ -- Health expend per cap + Health as % expendrture urban areas (World Bank, 1998a). At the ___ ___ _ _____ primary level they increased from Z$ I per visit to Z$6.5. Previously free prenatal care was set at $Z 10. District hospital fees rose to Z$ 17 and, to streamline the referral system, a bypass fee of Z$24 was established for all those patients demanding care directly in district hospitals, and Z$38 for those doing so in provincial hospitals (Watkins, 1997). However, since accumulated inflation was more than 800 percent in the same time span, real fees were lower in 1994 than they had been in the eighties. Bank policy advice influenced the government's decision to strengthen enforcement and increase levels of fees for government health facilities (see Box 8). The actual design of the fee structure, however, was in many respects contrary to what the World 22 The Zimbabwe case covers facts taking place in the 1990s, as the authors were unable to obtain current information on the subject 23 This section is taken from World Bank, 1 998a 3. Empirical evidence on protection mechanisms 44 Bank had suggested. For example, the Bank had recommended providing certain preventive services for free or at highly subsidized rates, an advice that was not followed by policymakers. Also, the government did not implement other recommendations made by the Bank to increase revenue. A World Bank report (World Bank, 1 998a) gives the following explanation: "although user fees were not a major potential source of revenue, they were within the direct control of the MOH and could be increased by administrative decree. The other recommendations (see Box 8) were longer-term endeavors, requiring new legislation, institutional change, and the cooperation of other ministries and stakeholders". In 1995 the government ordered the abolishment of user fees in rural health centers following an MOH study (MOH, 1995 quoted in World Bank, 1998a) concluding that the high levels of fees and widespread confusion about fees and referral policies reduced attendance without improving referral efficiency. 3. Empirical evidence on protection mechanisms 45 Box 8 Health Financing Sector Review Recommendations The 1990 health financing study was the Bank's Table 13 Zimbabwe: Expected increase in MOH most comprehensive and influential analysis of health resources, 1990 (Z$ millions) funding and expenditure in Zimbabwe. The study was organized around three major topics: resource Annual mobilization for health; increasing the allocative and Revenue item Revenue technical efficiency of health care; and developing User charges 42-52 systems and skills to support stronger health Full-cost pricing for private patients (15-20) financing. Out-patient fees (10) To improve resource mobilization for health, the Drug charges (10-15) health financing study recommended increasing user Adjusting fees for inflation (7) fees in public: facilities (to restore them to real 1985 National Health Development Fund 80 levels); establishing a national health development Cost-sharing with local government 10-15 fund (financed through alcohol or tobacco taxes); and mission facilities increased 'cost-sharing" by local govemment and Expanded private insurance 80-100 church mission health providers; and expansion of the TOTAL 212-247 private health insurance system. The Bank suggested Source World Bank, 1998a that the recommended measures could generate up to The Bank's recommendations, however, were Z$250 million additional resources annually for the centered on cost recovery. The study appropriately health sector (Table 12). emphasized that improved billing at central hospitals Of this total, lhe study estimated that Z$130 million and reduced public subsidies to private care were could be available to the MOH (equal to more than priority actions relative to the increase in cost recovery half of the MOH's 1987-88 budget). Based on these at primary and secondary health facilities. Fees recommendations, the ESAP and the second Family should be set to encourage clients to seek care at the Health Project set a goal of recovering 10 percent of lowest level of the system. At lower levels, the main the MOH budget by 1995. goal of cost recovery was to enhance the efficiency of the referral system and deter unnecessary use. The The improve efficiency, the Bank also study called for charging separately for inpatient and recommended al reallocation of health expenditure to outpatient drugs to recover costs and deter give greater emiphasis to primary health care, unnecessary use. The Bank also recommended improved management practices, and contracting out shifting the burden of proof of income to the client and of services with the private sector. increasing the income threshold for free care. 3. Empirical evidence on protection mechanisms 46 The Ministry of Finance did not allow facilities to retain revenue from user fees until late 1997. Prior to then, when user fees were retained by the central treasury, there was little incentive for facilities to collect fees. By 1995 revenue collected from user fees represented about 5 percent of total revenue in government health facilities. (c) Protection mechanisms In 1991 the government introduced the Social Table 14 Zimbabwe: Eligibility criteria Development Fund (SDF) with two components, of which for waivers in real and nominal terms one sought to grant assistance to poor households to cover Year Nominal temms Real terms (1980=100) school and health care fees. Waivers were to apply to 1981 150 131 those with monthly family income below Z$150. This 1982 150 111 threshold was maintained until 1991 when it was 1983 150 95 increased to Z$400 per family. As can be seen from Table 1984 150 84 14 steady inflation during this period eroded the income 1985 150 77 criteria. Potential beneficiaries' real income had to drop 1986 150 67 progressively to remain eligible for waivers.24 Even when 1987 150 61 the income criterion was adjusted by inflation it was still 1988 150 57 below the threshold initially established in 1981. Also, the 1989 150 51 income cutoff point was permanently below the urban 1990 150 45 poverty line (Table 15). In addition, the income criterion 1991 400 110 did not take into account family size. Source To establish eligibility, potential beneficiaries had to go to social welfare offices with extensive and often hard-to-get documentation. In urban areas eligibility documents included pay slips, income-tax returns and letters from social welfare offices. This information was the basis for income determination. In rural areas this information had to be supplemented by information on household size, landholding, and type of dwelling. Unfortunately the criteria adopted proved rigid and hard to apply because only 20 percent of the population works in the formal sector and most applicants have in-kind or erratic incomes (Loewenson ,2000; World Bank, 1998a). There was some additional screening process as eligibility had to be confirmed by head teachers and village elders and potential beneficiaries had to pay administrative fees for this. High transport costs were involved and often several trips were necessary to complete this process. Eligible beneficiaries were given "free care letters" which they were to present at facilities to receive free care.25 Facilities were to be compensated for Table 15 Zimbabwe: Comparison of poverty lines and exemptions through the SDF in eligibility criteria Harare. However, reimbursement to Poverty line (month, per health units would take up to 8 Income cut off point household)* months since they required Year for waiver eligibility Urban Rural authorization from Harare. Funding 1990 150 302 193 was inadequate as SDF made smaller 1992 450 529 338 allowances than actually required. Source- * Poverty line established in World Bank 1995 poverty assessment report Level Finally, potential beneficiaries were is based on resources required to cover expenses from the consumption of a misinformed or uninformed about the minimum package of goods (for further details see World Bank, 1995). waiver system. Data on inflation was taken from information of the Central Statistical Office of Zimbabwe. 25 Data on poverty line for 1990 is taken from World Bank, 1995. Poverty data for 1992 are expressed in real 1990 Zimbabwe dollars 3. Empirical evidence on protection mechanisms 47 (d) Results There is substantial evidence suggesting that user fees have negatively affected the poor.26 For example, according to the 1994 Zimbabwe Demographic and Health Survey, 42 percent of the urban poor who reported an illness in the previous month gave "cannot afford" as the reason for not seeking treatment, compared to 14 percent for rural poor. 27 Also, attendance for antenatal care initially declined in some districts and urban hospitals serving poor populations following fee increases in 1994, but it subsequently recovered in most districts. The major negative consequences of the fee increases seem to be: (i) low- income womeni are waiting to begin antenatal care, either to save money or to ensure the pregnancy will reach full term; and (ii) the financial burden on poor households. This indicates that the analysis of the impact of user fees on the poor should not only consider its consequence on the quantity of services consumed but that is should also analyze its financial impact on households as well as the changes it might introduce in the health care seeking behavior (type of services consumed, timing of visits). The same survey indicates that the percentage of births in health facilities has not changed since 198[?]. The 1994 DHS found that 91 percent of urban women and 61 percent of rural women delivered at a health facility, virtually the same as found in the 1988 DHS (CSO 1988, 1995). Yet a study by Lennock (1994) shows that following stricter enforcement of charges, maternity admissions at Harare Central Hospital Maternity unit showed a 21 percent increase in the number of babies 'born before arrival' and later required admission due to problems; among these babies mortality rose 156 percent. According to Lennock cost was one of the major reasons for giving birth at home and for requesting early discharge after delivery. Many people do not seek care because they are unaware that they are exempt from fees. Moreover, prescription costs lead to discontinuation of treatment. District outpatient attendance has declined following increased fee enforcement, with the greatest declines among children. For children under 5 it fell by 30 percent from 1990 to 1992, which coincided with increased fees in 1991 and decreased incomes from the 1992 drought. Attendance rose sharply following the removing of rural fees in 1995, but average under-5 outpatient attendance rates are still below 1990 levels. In contrast, adult attendance fell only 20 percent over those two years and by 1996 was 20 percent above 1990 levels. In hospitals serving low-income populations, inpatient attendance dropped following fee increases, then recovered. A study by Hongoro and the MOH (Hongoro, 1994) shows that in rural facilities health care demand increased after fees were abolished in 1995. Loewenson (2000) found that 6 months into the elimination of fees, demand fell again as the inventories of drugs and medical supplies dwindled leading patients to bypass primary care facilities and seek care directly from the fee-charging district-level hospitals. There is evidence of people's willingness to pay fees when they felt that payment would give them access to quality services. Communities in Zimbabwe have in fact mobilized resources in many different ways such as in kind contribution of labor and food donation. In the case of rural Zimbabwe, exempting the rural poor from payments was an ineffective way of protecting the poor, as the exemption resulted in a drop in quality of care. Loewenson (2000) noted: 'People no longer accepted political messages of 'free health services' when services were collapsing [and] weie willing to make fair contributions." Total cost recovery from fees was low but increased substantially when facilities were allowed to retain fees. Since then, it has stagnated at below 5 percent of the recurrent health budget (Loewenson, 2000). Increased fees were criticized for not being associated with better quality services, especially when patients where charged irrespective of whether drugs or other supplies were provided to them. In the urban area28 where protection mechanisms were supposedly in place to protect the poor against user fees, 26 A sumrmary of this evidence can be found in World Bank, 1 998a 27, Reference is made to the urban poor as fees had been abolished in rural facilities at this time. 28 User fees were abolished in rural areas in the mid 1990s. 3. Empirical evidence on protection mechanisms 48 about 40 percent of the urban poor stated that the reason for not seeking health care at government facilities was the high price of services (Loewenson, 2000). Also, many poor people did not apply to program waivers because either they had not heard of the SDF or they did not know how to apply for waivers. By late 1993 only around one-half of the population had heard about the possibility of getting waivers through the SDF (Watkins, 1997). Many eligible persons seemed unwilling to apply because of the stigma associated with welfare benefits. Overall, only 20 percent of the urban poor and 10 percent of the rural poor received assistance with health fees (World Bank, 1 998a). There is anecdotal evidence suggesting that some waivers were granted on the basis of political preference (e.g., depending on who was the political candidate favored by the applicant in the latest election). Finally, health facilities were often reluctant to grant waivers as reimbursement mechanisms for waivers and exemptions did not work adequately. (e) Lessons learned The following are main lessons emerging from Zimbabwe's experience with protection of the poor under user fees: * The setting up of a fund to compensate facilities for exemptions and waivers does not suffice to confer incentives to the providers for exempting the poor. Among the most important reasons are: lack of funding; delay in reimbursements of revenue loss due to waivers and exemptions and high costs associated with labor-intensive bureaucratic procedures. * High participation costs discourage the poor from applying for a waiver. * Limited information about the program and its benefits arnong prospective beneficiaries hampers the effectiveness of the exemption system. * The abolishment of user fees may not boost demand among the poor when it is accompanied by a drop in quality. * The income criterion is difficult to apply in a country where most work in the informal sector, have irregular income, and there are no infornation systems that might help to establish applicant eligibility. * The income criteria have to be adjusted to changing economic conditions. 3.6 Indonesia: The Kartu Sehat Program29 (a) Background With a population of 206 million in 2001, Indonesia is the world's fifth most populated nation, after China, Russia, the United States and India. The country enjoyed remarkable economic growth in the 1980s and much of the 1990s. As a result, Indonesia's middle class grew considerably, but poverty remained widespread. Indonesia plunged into an economic crisis in 1997 that led to significant political changes and increased the vulnerability of the poor. 29 Unless otherwAse noted, the information for this case comes from Gibbons 1995, Saadah et al, 2001, Sumarto et al., 2001, Ausaid v2000, Saddah, 2001, and SMERU, 2000 3. Empirical evidence on protection mechanisms 49 About one quarter of Indonesia's population lives below the national poverty line. In an effort to expand coverage of health services for the poor, since the early 1970s the government has promoted the expansion of public health centers and sub-centers. In the late 1 970s, the probability of seeking modern treatment by the poor when ill was less than 40 percent; by 1987 it had increased to 52 percent. These gains, however, have been reverted by Indonesia's sharp drop in income during the current economic crisis. (b) User-fiee policy A major expansion and strengthening of the government's primary health system in the 1980s strained the public budget for health and led to the government's decision in the early 1990s to increase reliance on user fee revenue for health financing. Facilities can only retain part of the revenue collected from user fees. In some places, health centers can keep 25 percent of the revenue collected; the balance is distributed with the budget allocation from the province (Newbrander, 2001). Fees seem to be quite low, they vary according to the level of care and across the country, and they are generally higher at higher levels of care. A RAND study analyzing the impact of user fees on access concluded that the higher prices led to decreased utilization of modern health care facilities, especially by low-income families. It also found that many of the poor would exit the formal medical care market in response to higher fees. Although to date fees remain low compared to cost, they are unaffordable to many of the poor, particularly for inpatient care. Private expenditure represents almost 65 percent of total health expenditure and out-of-pocket payments by patients finance nearly one-half of all spending in the public sector. (c) Protection policy Indonesia combines different mechanisms to protect the poor under user fees. Surat Miskin or "poor letters" are given to persons who cannot afford to pay the full cost of services. To obtain this waiver, the poor have to go to the village head and sub-district officials for signed approvals. Government health facilities grant: total or partial waivers to patients holding a Surat Miskin. Very few people seem to be using Surat Miskin, however, owing to the complicated administrative requirements and high travel costs associated with them. Also, government officials often charge additional fees in exchange for their signature of the necessary documents. Free care is traditionally given to people who work as volunteers in the village posts that deliver child health and prenatal care services. Since 1994, in an effort to mitigate the adverse effect of user fees on the poor, Indonesia adopted the Kartu Sehat Program, an initiative that continues to function today and that has received additional impetus during the recent economic crisis of South East Asia.30 The program is intended to cover all poor families in the: country. It distributes cards which entitle the recipients to obtain free services in all government health facilities. A single card is handed out to each recipient family, and up to eight family members can be listed on the card. Cardholders needing health care must first go to a government health center. For further, higher-level care, the health facility must issue a referral letter entitling the patient to guaranteed free access in public hospitals. Until recently public facilities had to absorb the revenue foregone fronri this free-care policy. Since 1998, however, service providers are compensated for the additional woirkload through a governmental budget transfer that is established on the basis of the number of card holders living in the district (Cameron, 2000). Kartu Sehot cards follow a long bureaucratic journey until they reach the individual recipients. They originate al the provincial level of government and move down the decentralization path.3' Each 30 Kartu Sehat means 'exemption card" in Indonesian 31 There are 32 provinces and 357 districts (regencies). 3. Empincal evidence on protection mechanisms 50 provincial MOH distributes cards to the head of the district government, which in turns forwards the cards to the head of its poor sub-districts. Within each sub-district, priority is given to those villages identified as being the poorest (also known as IDT villages or "left behind" villages), based on information from the Central Bureau of Statistics. From there, cards are allocated to the heads of low-income villages, and then to the head of the sub-villages where cards are finally handed out to the poorest households (village heads or health center staff selects recipient households from a list containing all poor households in the poor villages). Originally, different villages used different types of information to select the poor. Some use the so-called "Dinas Social list" of all poor households in the IDT villages; others select households according to their socioeconomic status (Gibbons, 1995). The process of selecting the poorest is made by the village or neighborhood head (World Bank, 2000). Kartu Sehat has received a major impetus with the Social Safety Net Program that was introduced by the government in the late 1990s to combat the impact of the economic crisis. Still it was recognized that the scheme was not functioning as envisaged. At most, card holders obtained free services at Puskesmas (health centers), but rarely did the services extend to hospitals or to delivery services for pregnant women. With the adoption of mechanism to compensate providers for waivers given in the late 1 980s access to health services by the poor improved. Further, Kartu Sehat now puts greater pressure on providers to play an active role in administering the scheme, including the process of identification of poor families. Providers participating in the program are also expected to recording details about beneficiaries and to issue the program cards (Ausaid, 2000). Also, eligibility criteria have been modified. The poverty measure currently in use to determine eligibility is the so called household "prosperity status". Under this new policy a household is deemed eligible for the health card if it fails to meet any of the following criteria (Sumarto S, Suryahadi, A, 2001): * All household members are able to practice their religious principles. * Al household members are able to eat basic food twice a day. * All household members have different sets of clothing for home, work, school and home. * The largest floor area of the house is not made of dirt. * The household is able to seek modem medical assistance for sick children and family planning services. The National Family Planning Board collects information on this issue. The task of selecting beneficiaries is now handled by local teams that must operate within program guidelines but which still have substantial discretion in defining eligibility (d) Results The Kartu Sehat program is rapidly increasing its coverage. Data from a nationwide household survey (Susena, 1999) indicates that 10.6 percent of Indonesian households report ownership of a health card (Saadah and Pradhan, 2001).32 Data from the Central Independent Monitoring Unit (CIMU) is even more promising as it found that by July 2000, in 5 provinces being closely monitored, 89 percent of all poor families were Kartu Sehat beneficiaries; only about 10 percent of all covered were non-poor (leakage); and over 70 percent of pregnant women in poor households had received ante-natal care and assistance during delivery (World Bank, 2000). However, there is information that indicates much higher levels of leakage than indicated by CIMU. According to the 1999 National Household Survey, about 39 percent of the health cards were owned by household from the wealthiest three quintiles (Saadah and Pradhan, 2001). 32 Each card covers an entire household 3. Empirical evidence on protection mechanisms 51 The presumption about high leakage has been confirmed in a study by Sumarto et al. (2001). These authors found that the proportion of those belonging to the poorest quintile using health cards to obtain program benefits was 10.6 percent while the equivalent proportion among the non-poor was 5.3 percent (Table 17). Leakage may partly be explained by the fact that the income criteria used by Kartu Sehat may not adequately capture poverty. Sumarto (2001) used household survey data to simulate the target population with the current income criteria and compared it with the population that would be eligible if the national consumption based poverty line would be used instead. They also found that 38 percent of those who are according to the national poverty line are not eligible for Kartu Sehat according to the official eligibility criteria. On the other hand, 46 percent of the non-poor households were eligible and 38 percent of ihe poor were not eligible (Sumarto, 2001) Several studies question the effectiveness Table 16 Indonesia: Relationship between Kartu of Kartu Sehat as many health card Sehat eligibility criteria and income-based poverty owners do not seem to use them when criterion, 2000 seeking care from a public provider Consumption-based measure of poverty (Saadah, Pradhan , 2001 and Cameron, Poor Non-poor Total 2000). Soelaksono et al. (1999, quoted in Kartu Eligible 15% (62%) 85% (46%) 100% Saadah and Pradhan, 2001) mention Sehatl Non- 25% (38%) 75% (64%) 100% eligibility eligible among the reasons for this behavior a criteria Total (100%) (100%) lower quality of service for health card users (for example less time allocated to Source Sumarto, 2001 patients). According to the same authors, health card holders confirm this situation: they perceive the care received through the health card to be of lower quality than services obtained when not using the card. This situation may be explained by the fact that providers are paid on a lump sum basis rather than on a case-based or other volume-related form of provider payment. Table 117 Indonesia: Coverage of social safety net programs across expenditure quintiles Program Coverage ([percent?] Ratio Poor Non-Poor Non- Eligible Total poor to Program Recipients Q0 Q2 Q3 Q4 Q5 Total Q. - Qs Poor Subsidized Rice 50,385,444 52 64 46.24 41 71 35 76 24 33 36 90 40.09 0 70 Employment Creation 50,385,444 8 31 6 89 5 79 4 58 2 53 4 94 5.61 0 59 Primary School Scholarships 29,745,369 5 80 4 84 4 02 3 52 2 04 3 60 4.03 0 62 Lower Secondary School Scholarships 10,394,621 12 15 10 31 8 34 6 73 4 85 4 53 8 42 0 62 Upper Secondary School Scholarships 6,430,146 5 40 5 06 3 32 3 04 1 96 3.32 3.71 0 62 Medical Services 27,567,138 10.60 7.24 6.3 4.52 3.09 5.28 6.33 0.50 Nutrition 19,970,948 16 54 16.64 16 38 15 94 14 24 15.94 15 94 0 95 Source Sumarto 2000. Kartu Sehat has experienced several implementation problems that might be of interest to other countries wishing to imlplement similar poor-protection mechanisms. One of the problems has been widespread lack of knowledge about the program by both health facility personnel and patients. A newsletter evaluating the program states: "Too many people remain confused or simply do not know how the Health Card scheme is intended to work Who is entitled to receive a card? What benefits does it provide? Which members are covered and for how long?" (SMERU, 2000) 3. Empircal evidence on protection mechanisms 52 Also, in the first years of implementation knowledge about the program among public officials varied widely from region to region.33 For example, in a 1995 field study from district health officers in the province of Daerah Istimewah Yogyakarta (DIY) received clear implementation instructions and criteria for the selection of poor households from local lists. In contrast, in the province of Nusa Tenggara Barat (NTB) the implementation procedures were unclear and program staff was unable to implement Kartu Sehat (Gibbons, 1995). According to the same study, health service providers also had limited knowledge about Kartu Sehat. Most health staff participating in a survey about the program were not aware that beneficiaries referred to hospitals were also entitled to free care in those facilities. (e) Lessons learned The following are identified as the most critical findings in Program implementation of Kartu Sehat: * Lack of information by consumers, providers, and government officials and insufficient training have been two major problems encountered in the implementation of Kartu Sehat. * Cumbersome distribution of Kartu Sehat cards from higher to lower levels of government not only means high administrative costs but also leads to delay and retention of cards. This issue illustrates that national program implementation can be very challenging in very large and decentralized countries. * Eligibility criteria fail to adequately capture the poor as many non-poor (as measured by the national poverty line) are eligible and many poor are not eligible for Kartu Sehat. * The Indonesia case illustrates the importance of reimbursing facilities for revenue foregone from waivers. Program coverage has increased substantially when facilities started to get compensated for waivers given under Kartu Sehat. * Coverage is an insufficient measure of protection of the poor. In Indonesia many poor do have a health card but they do not use it as perceived quality of services is sometimes low. 3.7 Thailand's Low Income Card Scheme (LICS)34 (a) Context Thailand went through a period of remarkable economic growth in the 1980s and 1990s that made it one of the fastest growing economies of South East Asia and the world, with an average annual GDP growth of 9.1 percent from 1986 to 1996. Thailand's population of 62.4 million has a life expectancy at birth (LEB) of 68.6 years -three more years than Indonesia and about 1.5 years less than China. As most regional economies, Thailand started to experience a severe economic crisis in the late nineties. Poverty has increased as a consequence although not to the extent seen in other countries in the region, such as Indonesia and Korea The economic crisis also drastically changed the health financing scenario. The per capita government budget for the Ministry of Public Health (MOPH) declined by 25 percent in real terms between 1997 and 1999. The MOPH responded by cutting capital expenditure, rationalizing the use and procurement of drugs, and reducing expenditure for utilities and travel. Benefits under the Civil Servants Medical No up to date information was available on this issue when this report was being written 34 Based mainly on Gilson etal., 1998, Donaldson, 1999 and Somchai 1998 3. Empirical evidence on protection mechanisms 53 Benefits Scheme were restricted to care from public sector facilities. In addition, between 1996 and 1998 household annual health expenditures on health facility-based care declined sharply -by 36 percent in real terms (to 2,3 1 6 baht).35 At the same time, household spending on self-medication increased by 12 percent (to 552 baht). This change in patient behavior reversed a long-term trend away from self-treatment towards treatment by trained health providers (NESDB, 2000). A wide range of insurance or benefit schemes are available to the Thai population. The five major systems are: the Civil Servants Medical Benefits Scheme (CSMBS), a Social Security Scheme (SSS), the Voluntary HeaLlth Card Scheme (VHCS), the Low Income Card Scheme (LICS) and commercial private health insurance (Table 18). Together, they provide protection to about 76 percent of the population. The poor and near poor are protected by LICS (see details below) and VHCS (Table 19). VHCS covers near- poor households as well, mostly in rural areas. Beneficiaries can voluntarily buy a card which attracts a matching tax subsidy and which gives them access to free care at public facilities. Table 18 Thailand: Characteristics of health insurance schemes, circa 2000 Insurance Nature of Coverage Coverage Population Financing Program Scheme (millions) (percent) Characteristics Source of Funds Body CSMBS ErBenefit 6,6 11 Civil Servants MOPH Fund MOF Employees in Firms Tripartite contributions Social SSS Compulsory 4,8 8 Larger than 10 (MOPH, employer, Sci Persons employee, 1,50% of Scrt Persons ~wagesg Organization VHCS Voluntary 6,0 10 Near Poor MOPH Fund Mlnistry of Indigent, Children < Social 12, Elderly, Veterans, Ministry of LICS Welfare 27 45 Handicapped, MOPH Fund Health Religious & Community Leaders Private \Voluntary 1,2 2 Richest segment of Premium Households the population Total 50,4 76 Source Donalclson et a, 1999 Table 19 Thailand: Coverage of LICS and VHCS Services Insurance Provider Prevention not Program Ambulatory Inpatient Choice Matemity Promotion Covered VHCS Public Public Requires Yes Possible Pvt Bed Referral Requires ~~~~~~Special LICS Public Public Referral Yes Limited RN P Referral ~~~~Bed Source: Donaldson et al, 1999 (b) User Fee Policy 37 Thailand has a long standing user fee policy in the public sector implemented with clear policy guidelines since 1976-77; it raises a significant amount of revenue. A rough estimate indicates that, on average, about 60 percent of Ministry of Public Health (MPH) hospital's total revenue comes from the 35 [Provide! exchange rate ] Due to the economic crisis, this percentage was reduced to 1.0% after 1998 (see Donaldson et al 1999) Taken from: Health Care Reform Project Thailand on: http l/www moph go th/ops/hcrp/mainhcrp php7select=Finance html 3. Empirical evidence on protection mechanisms 54 government's allocation and 40 percent from user charges. Of this user fee revenue, approximately one- third is collected from patients on a fee-for-service basis (mostly from the sales of drugs), and the remainder comes from reimbursements from insurance plans. Often a major portion of hospital payments comes from the civil servant benefit schemes which cover many of the hospitalized patients and which pay hospitals their full fee. In health centers, user fees may account for as much as 70 percent of total revenue. Health centers provide an average of 3,644 visits per year, half of which for patients covered by the free care program. User fees are collected mainly from the sale of drugs. Hospitals and health centers set their charges on a sliding scale, and have the discretion to decide how much to charge patients based on the provider's assessment of their ability to pay. Patients covered under insurance plans, such as civil servants and their dependents, are often charged a higher price -two to three times the lowest price. (c) Protection mechanism To protect the poor against the financial burden of user fees, the government developed the Low Income Card Scheme (LICS) under which the poor have free access to public facilities. Thailand has over twenty years of experience with LICS as operations started in 1975. At the beginning of LICS, the directors of Table 20 Thailand: National poverty line and cut off public hospitals had the authority to screen points for LICS the beneficiaries and grant exemptions to 1988-89 1993-94 1997-98 the poor on their own criteria. Different National per capita poverty Imen cash incomes cutoff points have been used per monthl 473 636 911 to determine applicant eligibility. Since Cut off point for a single/month 1500 2000 2000 1993, eligible beneficiaries are single Source: Donaldson, 1999, and based on Somcha, persons with a monthly income of less than 2,000 baht ($47) and households with a combined monthly income of less than 2,800 baht ($66). 38 Cutoff points were neither differentiated for urban and rural settings nor updated on a regular basis. Table 20 compares cutoff points of eligibility for LICS with national poverty lines for several years. As the income level for being entitled to the LICS systematically exceeds the poverty line, the program potentially covers not only the poor but also the near poor. Figure 11 Thailand: Official procedures for screening the poor, 1981 During the 1990s LICS eligibility was expanded to include not only the income DOstnct offirer application forms, which poor but also other groups, including the ' show all household eamings elderbutals children belowu, includin the 4 These are retumed to village elderly, children below 12, veterans, the Tambon ommittee heads handicapped, and monks. With this 2 The village head, In expansion, ~~~~~~~~~~~~~~~3 5 consultation with the village expansion, LICS changed its name from V committees, selects and lists the medical welfare scheme for the low- Village Committee those believed eligible income to the medical welfare scheme 2 3 The list is debated and I amended where relevant by for underprivileged groups. VIllage Head the Tambon committee 4 In the district the cards are The other major set of changes 1 produced and stamped The other major set of changes 1 ~~~~~~~~~6 5 The cards are sent back to introduced in the implementation of the village head LICS concerns the dissemination of -Applicants Successful 6 The village head distnbutes Applicants - the card to the successful information on the LICS allocation Applicants applicants process, as lack of information was Source:IGilson._1998 considered a major obstacle to coverage of the poorest. Accordingly the village head was given a more active role in the application procedure: he was made responsible for announcing the scheme one month Exchange rate June 2002, 42 31 baht/dollar 3. Empirical evidence on protection mechanisms 55 before the deadline for application and for conducting a house to house survey to ask people to complete their applications. The identifical ion of the poor is conducted at the community level, primarily by village and Tambon (sub district) leaders. The official procedure for screening the poor that was introduced in 1981 is depicted in Figure 11 Y'A committee headed by the village leader examines the application forms that are submitted by applicants. Then a preliminary list of beneficiaries is produced by the village head in conjunction with the village committee. This list is then screened and amended by the sub district level: Since its implementation in 1981, the participation of the community in the screening procedures has been gradually sirengthened by adding new members to the village committee. Most importantly, health workers were included in the screening process as it was argued that committee members from outside of the health sector did not understand well enough LICS procedures. Also, there was much concern that the village leaders might allocate cards according to their own criteria or needs. Finally, the power of the village head was further counterbalanced by a regulation introduced in 1987 whereby the person responsible for the Tambon health center checked the list of eligible candidates drawn up by each village in the Tambon and by adding new local actors to improve judgments about eligibility. People who qualify are given a beneficiary card valid for three years. The card specifies one or two designated heaLlth facilities -normally, they are local health centers or district hospitals- for beneficiaries to visit in case of illness or injury. The program requires beneficiaries to seek medical care from the Tambon health center as a first point of entry into the system. Health center staff will then determine whether to refer patients for which they will be provided with free care. For emergencies, eligible cardholders can go to any public facility (Somchai, [year?]). Under the free care program, poor individuals whio do not hold a card are classified by an ad hoc procedure on arrival at the government facility and exempted totally or partially from fees.40 Special funds are set aside to compensate facilities for waived services. A budget is allocated to the provincial level and is financed through general revenue. The Government of Thailand (GOT) has used a number of criteria for allocating the LICS budget to provinces, including population size, number of health facilities, and number of card holders, standardized mortality ratios, and workload. LICS has undergone modifications that reflect difficulties experienced with targeting. Several revisions have focused on strengthening community based screening processes involving a wider range of key participants and in particular village volunteers (Gilson et al., 1998). More recently Thailand has introduced radical changes in health policy, with a shift towards universal health insurance coverage. VHCS and LICS are to be substituted gradually by the so called "30 Baht health policy" launched in February 2001. With the new policy all uninsured citizens would receive a universal health card. Individuals must produce this new card when seeking care in either public or private facilities registered with the government for this initiative, along with some other individual identification. The accessing health service has to follow the referral system from the primary health center or the nearby hospital. For emergencies and accidents, the insured can access any government health services. All beneficiaries must contribute a copayment of 30 Baht ($0.71) per visit. 4' The 30 Baht Universal Coverage Policy, the insured will receive a predefined package of health services, which includes most health services except cosmetic care, obstetric delivery beyond two pregnancies, drug addiction treatment, hemodialysis, organ transplantation, infertility treatment, and other high cost interventions. From the government side, the funding of the system is paid by capitation. The annual per capita payment, which comes from general tax revenue is 1,404 Baht, part of which is paid to the health care facilities according to the number of Gilson et al. 1998. 40 They are called Type B" patients, distinguishing them from Type 'A' patients. 41, Exchange rate in June 2002: 42.31 baht = $1 00. 3. Empincal evidence on protection mechanisms 56 cardholders registered with them (APHEN, 2001; The Nation 2002). Since this new scheme was just started evaluation is still lacking. Main concerns raised by analysts are insufficient funding to finance the scheme, incentives for facilities to avoid the provision of high cost and incentives to switch to the 30 baht health card for people insured under other systems (The Nation 2002). It is worth mentioning that more recently Thailand has opted away from mechanisms specifically focused on exempting the poor from user fees and has instead started to promote a scheme under which all the non insured Thai are eligible for coverage against paying a moderate copayment (30 baht) per episode. The financial viability of this scheme remains to be seen as it might create incentives for the insured to switch to this system. (d) Results42 In 1998, the numbers of beneficiaries of all categories (low income, elderly, handicapped etc.) was approximately 17.6 million or 29 percent of the population. Low income card holders represented approximately 5.8 million of total beneficiaries (Table 21). According to a more recent source, by the year 2000 LICS covered as much as 37 percent of the total population (Tangcharoensathien 2001). Data on coverage of the target groups and leakage to non eligible beneficiaries varies. According to studies by Supachutikul (1996) and Mongkolsmai (1993), in the early 1990s coverage of the Ministry of Health target group (people below income cut off points) was equal to about 76 percent. Also coverage of the poor living below the national poverty line appears to have increased from the range 30 percent-40 percent in the late 1980s to around 80 percent in the early 1990s. This sharp increase in coverage is not confirmed by all studies. According to research by the National Institute of Development Assistance, NIDA, serious problems exist with respect to both coverage and leakage. A survey showed that about one-third of responding households were poor, and only 32 percent of them had the low-income cards. Furthermore, there seem to be problems of leakage as only 55 percent were poor according to the family income criteria (Sriracha Charoenparij, et al., 1999). Leakage of benefits to the non-poor may reflect the fact that official eligibility income criteria are above the official poverty line and therefore many of the non-poor are included in LICS's target population. Additionally, it seems that LICS beneficiaries are not always sufficiently well protected from health care payments. A 1993 study found that that those holding the LICS spent 6.1 percentof annual income on health as compared to the 0.6 percent to 2.3 percent spent by other insured groups (Pannarunothai and Mills, 1993). Various studies explain this situation with data indicating that LICS cardholders often do not use their card to access services and prefer to pay rather than to receive free services under LIC. As has been already illustrated by Kartu Sehat in Indonesia, this is a situation that seems to apply not only to Thailand. High coverage of documents certifying eligibility for exemptions does not seem to be a proof of effectiveness of protection mechanisms. Table 21 Thailand: Number of low-income cardholders by region, 1981-1998 (millions) 1981 1984 1987 1990 1993 1998' 1998 2 Central 1.84 166 129 182 182 0 75 3.24 Northeast 4 99 4.52 3.50 5.57 5 57 2 97 8 14 North 2.97 2.72 1.85 2.39 2 39 1.40 3.94 South 1.10 1.23 978 1 64 1 64 0,69 2 33 Bangkok 0.00 0,03 0,022 0,08 0,08 0,02 0 03 Total 10.89 1016 7 64 11.50 11 82 5 79 17 67 Source. Charoenparij et al, 1999 based on data from the Rural Health Division, MOPH, Thailand. Notes: ' Reported Figures for the low-income cards issued in 1998 2 Reported figures for all types of cards issued to the underprivileged (low income and other groups) 42 Discussion is limited to LIC as the '30 baht policy" is still in its first months of implementation. 3. Empirical evidence on protection mechanisms 57 The budget set aside to cover exemptions for card holders and other poor patients has increased five fold in real terms between 1988 and 1997 (Table 22). It is however estimated that it is below the amount of revenue foregone due to exemptions. According to a study by Donaldson (1999), in 1997, the budget was about 30 percent below expenditure for exempted patients. The remainder was cross-subsidized from each hospital's own revenue generated from user charges and insurance plans (Table 23). The problem of inadequate funding is Table 22 Thailand: Resources for LICS, 1988-97 worsened by a problem of inequitable Number of geographic allocations where wealthier people covered Budget in current regions seem to receive higher per Year by the program prices (million Budget in 1993 prices regions seem to ~ ~ ~ ~ ~~ear(million) baht) (million baht) capita allocations than poorer regions 1988 7 65 706 901 (Table 24) even though the situation has 1989 7 65 800 936 improved as the allocation has come 1990 10 73 1,500 1,736 closer to a capitation based formula. 1991 10 73 2,000 2,205 1992 11.70 2,500 2,625 Gilson et al. (1998) carried out 1993 11 70 2,750 2,750 community based field research which 1994 11.80 4,273 3,876 showed that thre card is greatly valued by 1995 11.80 4,475 4,059 the people because "it protects the 1996 14 00 5,706 4,929 poorest against the cost of both minor 1997 15 00 6,703 5,515 and more serious illness and for higher Source Somchai Suksir'serekul, 1998 based on database of office of health income groups it provides a safety net: insurance free access to government care in cases of expensive chronic or serious illnesses, or illnesses not effectively treated by other providers." Despite these successes Gilson and her colleagues identified various obstacles to further success of LIC among which: * The multiple vulnerabilities and marginal situation of the poor which makes it difficult to reach them: they tend to live further off and work longer hours, making it more difficult to disseminate the iniformation and to persuade them to apply for the health card. 'I've never been issued an LICS card. I've never come across any messages on it. Anyway, it could be possible that my house is isolated from the rest of the community and I don't have much time to chat with neighbors. I've been working as an aide for an old, long-term patient from 7 o'clock in the moming till 6 or 7 o'clock in the afternoon.. I might be too busy to get messages." (Gilson et al., 1998) * The difficulty in using formal Table 23 Thailand: Budget and expenditure of the LIC, income criteria, when most of 1987-1997 applicants belong to the Resources allocated to informal sector, are highly Year finance exemptions Loss due to exemptions mobile, and receive some in- kind income. 1987 705,839,500 2,051,856,237 kind imrcome. 1991 2,000,000,000 2,345,067,875 1997 43 ~~6,370,524,000 9,018,341,515 * Use of the power to allocate S997rc Donaldson 189, Source. Donaldson 1999 cards for personal favors and Note. Expenditures are not true measures of costs of service, rather of the political purposes. charges not paid by LICS patients * A negative perception of public service quality by the poor and many cardholders prefer not to use LICS 43 Note that the numbers quoted by Somchai Suksiriserekul, 1998, do not coincide with the information contained in the report by Donaldson 1999 3. Empirical evidence on protection mechanisms 58 Table 24. Thailand: LICS per capita budget allocation by region (baht), 1992-1999 1994 1996 1998 1999 Northeast 132 140 205 264 North 194 193 263 306 South 323 160 239 273 Central 539 183 258 316 Central Northeast 4 08. 1 00 1 38: 1 00 1 16: 1 00 1.20. 1 00 Source: * Insufficient funding (as confirmed by the data contained in Table 23). Poor experience with cost and quality of service provided under LIC discourages some low income people to apply for the card and use it to obtain health care. 'I never got free drugs. I had to pay every visit' "Since I got the card, I have never used it. I did not know when and how to use the card. I hear that clients using the card would obtain only low quality drugs and had to wait longer. Clients who pay for treatment would get the service first. I prefer buying drugs. Having LIC was good but it also has disadvantages. My sister in law used it at the hospital They did not pay attention to us. They thought we did not have money, they paid less attention to us.. "(Gilson et al., 1998). * Stigmatization of the poor, which deters some card holders from using their rights under LICS. Low income card holders feel that health workers discouraged them from using their cards making them feel inferior and embarrassed. 'Sometimes I self-treat because I do not want [the health worker] to complain that I often get free drugs from the health centre" (Ibid.) (e) Lessons learned Thailand's overall record is quite outstanding when compared to most other countries. First of all, Thailand's LICs covers possibly as much as 80 percent of its target population and, as has been shown by the study of Gilson and colleagues (1998), the card is greatly valued by many beneficiaries. Also, the extensive experience with the program has allowed the govemment to gradually strengthen LICS by increasing community participation, changing income cutoff points, increasing funds to reimburse institutions, adjusting formulas to allocate resources meant to cover expenses and, more recently, by reconsidering the criteria to identify the poor. As was noted, only about 13 percent of the Thai population, or 7.9 million people, live below the national poverty line (Somchai 1998). Yet by the year 2000 LICS had reached about 37 percent of the total population. Thus, the number of program beneficiaries greatly exceeds the number of poor people in the country. Coverage reaches not only part of the poor but also many non poor. This is mainly explained by the adoption of an eligibility threshold that stands above the poverty line and by the inclusion of other groups (among which children below 12 and monks) many of which might not be poor. It remains an open question whether Thailand should try to limit LICS to the poorest by redefining its cut off points or even replacing them by other poverty proxies. Better targeting would allow the program to provide more funding for exemptions and might thereby improve quality of services given to LICS patients. As regards the implementation process the following are some lessons: * The effectiveness of a protection system cannot be measured in terms of leakage and coverage only. It is crucial that beneficiaries actually access the free services. * Users may prefer to pay for health services rather than receive low quality services. 3. Empirical evidence on protection mechanisms 59 * The multiple dimensions of vulnerability of the poor require careful design of the protection mechanisms. * It is difficult to assess an applicant's income level as they usually work in the informal sector, has e irregular and mobile jobs, and receive some income in-kind. * When designing screening bodies to identify the beneficiary population, careful consideration has to be given to skills and incentives of its members. The Thai experience shows that community leaders had incentives to allocate exemptions to pay personal and political favors and lacked the skil ls ito identify the poor by themselves. * Uncornpensated free care created incentives to provide low quality services. * Geographic allocation formulas of budget influence effectiveness of protection mechanisms. * Thailand has made irregular and sporadic adjustments to its income cut off points. Adjusting them according to changes (for example inflation) seems to be highly desirable. * Lack of consistency between eligibility criteria and the national poverty lines can create serious problems. In the case of Thailand it has led to substantial leakage of program benefits as many card holders are not poor but are eligible for a health card. 3.8 Chile's Nacional Health Fund" (a) Context Chile is an upper middle-income country that exhibits some of the best health status indicators both among Latin American countries and among countries in its income category. With a life expectancy at birth of 75 yeaLrs, infant mortality of 10 per thousand life births, and annual per capita income of $8.440 (PPP-adjusted), Chile is the richest country in the case study sample, and the one with the best health indicators. Chile has a mixed health care system, with both financing and production in public and private hands. Health insurance coverage is mandatory for formnal sector workers who, according to a law passed in 1981, have to devote 7 percent oftheir payroll either for public or private health insurance. There is a single, large public insurer known as the National Health Fund (Fondo Nacional de Salud, FONASA), and there are rnultiple, competing private health insurers known as Instituciones de Salud Previsional (ISAPREs), as; well as traditional commercial indemnity insurance firms. In principle, all Chileans are free to choose between FONASA and the ISAPREs. In practice, however, it is a person's level of income and age that determines whether he or she becomes affiliated with the public insurer or with one of the several ISAPREs. FONASA covers middle, lower-middle, and low income people, whereas ISAPREs cover middle, upper-middle, and high income individuals. Also, the active or retired elderly are often unable to enter the ISAPRE system owing to the high premiums that they face; many choose lo enroll in FONASA. Approximately three fourths of the population is covered by FONASA and one-fourth by the ISAPREs. FONASA receives about 40 percent of its funding through subsidies from the national treasury and 60 percent through contributions by beneficiaries. Contributions consist mainly of the 7 percent social This case is based on Batran (2000) and on information provided directly by current and forrner FONASA officers 3. Empirical evidence on protection mechanisms 60 insurance payroll withholdings and to a much lesser extent of co-payments made by patients in government health facilities. The ISAPREs are self-financed through the 7 percent payroll deductions and additional contributions made by voluntary members. Forty percent of FONASA's beneficiaries have been classified as "indigent individuals"; these make neither payroll contributions to FONASA nor payments in public facilities. Health services to them are free of charge but only when obtained from public health facilities. The other 60 percent of those covered by FONASA are known as the contributing affiliates and their dependents. They are those who make their payroll contributions and who, when using public or private facilities, make co-payments that increase with their income. (b) User fee policy User fees have been in place in Chile's public hospitals for over two decades now, but they represent only a marginal part of public health facilities' revenue. Bitran et al. (2000) estimated that in 1995 user fees revenue in government hospitals accounted for only 11 percent of total revenue. User fee revenue is retained at facilities and can be spent locally according to certain official guidelines. FONASA pays hospitals for services delivered to its beneficiaries (the indigent and the contributing affiliates) according to a mixed payment system that consists primarily of prospective budgets and to a lesser extent of payment per case resolved (a simplified version of the U.S. Federal Government's Diagnotic Related Groups, DRGs). Budgets are calculated according to expected volume of services and historical expenditures. Revenue from user fees supplements hospital income in public hospitals, and therefore these facilities have an incentive to collect fees. This seems to be especially true for tertiary hospitals and for complex procedures where bills can be quite high. Public hospitals have been plagued by chronic deficits and are using resources from user fees to cover part of their financial gap. It would seem, however, that FONASA may be conferring a perverse incentive for fee collection with a recent policy that bridges the financial gap of hospitals with the largest deficits. This situation, together with the administrative costs involved in cost recovery, weakens incentives to collect user fees. A former FONASA official explains: 'Why should they [public hospital's management] bother if their expenditures get covered [by FONASA] anyway?" 3. Empirical evidence on protection mechanisms 61 (c) Protection mechanisms FONASA classifies its beneficiaries into groups, from A to D, according to income and socioeconomic status (Table 25). Group "A" comprises the indigent. These are the unemployed poor or people who, despite having some informal employment or occupation, are deemed poor and thus are entitled to full public subsidies for health care services. Groups B, C, and D are active workers who fall into these groups on the basis of the income. As income goes up, people move from group B, to C, and to D. Their payroll health tax remains a constant 7 percent. There is a limit to the income that is subject to the 7 percent, however, currently equal to about $1,470 per worker per month.45 Thus, those in group B do not pay any fees in public hospitals but do make the regular payroll tax contribution. Affiliates in groups C and D make thieir payroll tax contributions and pay user fees in public facilities. In any case, payments made by benefliciaries in groups C and D represent only a small part (20 percent at the most) of the actual cost of care. Table 25: Chile: Classification of affiliates of the National Health Fund (FONASA) Obligations as FONASA beneficiaries User fee in Group Eligibility criteria Payroll tax public hospital "A" Persons with 'lack of resources", beneficiaries of subsidized pensions, No No beneficiaries of family subsidies (subsidies for children in low income families) "B" Monthly income below national minimum salary and families with more than Yes No three dependant household members and monthly income above minimum salary and below 1 46 minimum salaries .C. Monthly income above minimum salary and below 1.46 minimum salaries and Yes Yes, 10 percent less than 3 dependents Monthly income above 1 46 minimum salaries and 3 of cost of or more dependants service "D" Monthly income above 1 46 minimum salaries and less than 3 dependents Yes Yes, 20 percent of cost of service Source. Fonasa, 2002 Identification of the poor is decentralized and is carried out both by municipalities and health facilities. In municipalities clerks are in charge of selecting the poor; in health facilities this responsibility usually belongs to a social worker who subjects potential beneficiaries to an interview for eligibility determination. Interviews are often followed by a home visit, for verification, or to supplement the information already collected from the person. A "free health card" is handed out once indigence status is determined. Floor patients who do not possess a "free health card" are classified on arrival by health facilities. "Lack of resources" is the main eligibility criterion to get a free health card (Group A). There are no specific guidelines for establishing what constitutes "lack of resources" and it is very much up to the judgment of the officer in charge. There is no supervision of this eligibility determination process. (d) Results" Table 26 presents "actual" and "imputed" number of affiliates according to information from the 2000 CASEN national household survey. Actual status is determined based on the respondent's self- declaration (respondents were required to produce the FONASA card for the survey). A person is considered to be an actual FONASA affiliate if he or she declares to hold a FONASA card. Imputed status was defined as the one the person should have (i.e., the FONASA group where he or she should be) based on poverty status and cash income level. There is almost total coverage of the poor as only I 1 45 This means that the maximum, legally-required payroll contribution is $103 FONASA does not receive more than this amount from anyone ISAPREs, however, are allowed to charge an additional amount above this maximum mandated contribution, if that maximum does not suffice to pay for the premium of a given ISAPRE health plan. 46 Mainly based on Bitran et al, 2000a 3. Empircal evidence on protection mechanisms 62 percent (15.064) of individuals that should belong to FONASA A group are not covered by any health insurance. Further, even those poor that do not hold a free health card usually obtain free health services quite easily through a means test conducted at the facility, as mentioned earlier. Table 26: Chile: Self-declared and imputed FONASA group, 2000 Declared FONASA Imputed FONASA group according to CASEN47 Group A B C D Total A 89,390 529,729 140,514 242,715 1,002,348 B 22,550 647,805 384,794 448,835 1,503,984 C 4,412 139,864 183,343 279,140 606,759 D 3,500 104,790 114,538 520,511 743,339 Unknown 1,856 30,398 22,999 43,231 98,484 Military 184 15,790 12,173 155,513 183,660 ISAPRE 2,283 76,715 92,926 1,182,015 1,353,939 Ninguno 15,064 196,286 71,463 317,603 600,416 Otro 190 4,382 1,996 12,523 19,091 No sabe 610 18,475 10,194 20.834 50,113 Total 140,039 1,764,234 1,034,940 3,222,920 6,162,133 Source. Bitran and Associates based on Casen 2000 household survey Note FONASA imputed groups are obtained by applying groups defined by FONASA to income information from Casen FONASA A group is imputed by applying the national poverty line to the income data provided in Casen. Table excludes dependents. Also, FONASA is highly progressive in the delivery of its health benefits. Table 27 shows per capita benefits, contributions, and subsidies for the four FONASA groups from a study based on data from the early nineties (Bitran et al. 2000). Per capita benefits for group A and B are substantially higher than benefits of groups C and D, reflecting the more intensive use of FONASA'S network of health facilities by the poorer groups. In other words, there is a progressive inequality in the distribution of benefits in favor of the poor -one that may help to bridge an equity gap in delivery of health services in the country (note that beneficiaries of groups B, C, and D may make use of private, unsubsidized health care, and that this information is not shown in the table). Health financing also seems progressive. As can be seen from the table, per capita contributions increase with income, going from Ch $7,461 for group A to Ch. $70,479 for group D. Progressivity would appear even greater if non-health taxes, such as income and property taxes, were also accounted for in the exercise. 47 Imputed FONASA group refers to the FONASA group should belong to according to their income level Declared FONASA group refers to the FONASA group individuals actually belong to according to Casen 2000 3. Empirical evidence on protection mechanisms 63 Table 27 Chile: Per capita benefits and contributions in FONASA (Ch$ of December 1995) FONASA Group B, C and A B C D Total A-D D only Estimated nurnber of beneficiaries 3,664,733 2,806,016 1,050,634 946,484 8,467,867 4,803,134 Benefits Primary Care 6,247 4,691 3,388 2,249 4,930 3,925 Secondary Care 7,513 8,134 2,806 2,883 6,617 5,934 Tertiary Care 38,289 31,346 9,261 10,377 29,267 22,383 Subtotal Primary, secondary and tertiary 52,049 44,170 15,455 15,509 40,814 32,241 care Free-Choice Modality 794 11,608 17,927 19,053 8,544 14,457 Pregnancy ancl Illness Subsidies 932 4,381 5,833 7,401 3,406 5,294 DFL 36 (additional private sector care) 1,507 1,283 438 447 1,182 933 Total benefits 55,282 61,442 39,654 42,409 53,945 52,926 Contributions 7 Percent Payroll Contributions 6,987 32,851 43,739 55,495 25,539 39,695 Co-payments in Public Facilities 0 0 3,515 3,586 837 1,475 Free Choice Modality Co-payments 475 6,944 10,725 11,398 5,111 8,649 Total contnbutions 7,461 39,794 57,979 70,479 31,487 49,819 Source Bitran and Associates, 2000. Values are Ch$ at December 1995. More recently, however, there has been much turmoil when FONASA discovered in 2001 that as many as half a million of the presumed indigent in group A were actually taxpayers who on average made, annual income tax payments above Ch. $3,000,000 (about US$ 4,500). This finding was made possible through an administrative breakthrough whereby FONASA was for the first time granted permission to cross check information on its affiliates with information on income tax contributions from the internal revenue service. This situation is confirmed by results of the 2000 CASEN survey. Table 26 shows that of a total of 1,002,348 self-declared Group A affiliates, only 9 percent are poor according to the national poverty line. F urther, almost 40 percent of group A affiliates have income levels that would put them in FONASA' groups C and D. Overall, the number of FONASA beneficiaries in Group A exceeds by a factor of seven the country's population falling below the poverty line. The problem in Chile is clearly related to leakage and not so much to coverage. Some of the factors explaining this situation are the following: * There are no clear rules regarding what criteria to apply in health facilities when determining poverly. "Lack of resources" is a very vague term. When asking a former official on how they decided on whether a person would be given the "free health card" or not he said: 'When we [at the hospital] decided on whether someone was to be given the free health card" or not it depended very much on how much a patient would cry .' * There are incentives to cheat as Group A patients receive in practice the same benefits as Group B patients even though those in Group A make no contributions of any kind to FONASA while Group B beneficiaries make a monthly contribution equivalent to 7 percent of their monthly income. * There are no clear incentives to identify the poor correctly. In Chile, municipalities' income is not linked to cost recovery. Also, municipalities are not responsible for the financial performance of public hospitals. Under this condition, municipalities do not have strong incentives to avoid 3. Empirical evidence on protection mechanisms 64 leakage when identifying the poor and possibly do have incentives to gain political favors by giving "free health cards". (e) Lessons learned The implementation of a protection mechanism for the poor in Chile reveals several challenges that are of interest for countries willing to engage in similar processes. Chile has reached almost all of the poor through FONASA and the poor seem to be accessing health services much the same as the richer segments of the population. Problems are associated with leakage and not so much with coverage: When designing protection mechanisms careful consideration must be given to who determines eligibility. The case of Chile illustrates that countries have to choose entities in charge of determining eligibility in a way that strikes a balance between health facilities interest in collecting revenue and administrative or political entities possible interest to pay favors by allocating cards to the non poor. When designing protection mechanisms for the poor under user fees, careful attention has to be given to possible incentives created for the contributing richer population. In Chile such incentives clearly exist as the benefits for the indigent and the non poor are the same in practice. Group "A" beneficiaries are completely exempted from payments and get basically the same benefits as Group "B" "C" and "D" beneficiaries which make payroll contributions and pay user fees (with the exception of group "B"). Under this circumstance, strong incentives exist for potential beneficiaries to get classified as Group "A" beneficiaries. The case of FONASA in Chile illustrates once again the importance of clearly defining the eligibility criteria. In Chile no consensus seems to exist as to how to interpret "lack of resources". It is crucial to implement effective monitoring mechanisms capable of measuring coverage and tracking problems of leakage. Chile recently made a great leap forward when deciding to cross check its data base of FONASA beneficiaries with the data base of the Tax Administration. It was the integration of data bases that allowed Chile to detect a massive problem of leakage. 3. Empirical evidence on protection mechanisms 65 4 Lessons and best practice This final chapter summarizes the findings from case study countries and draws lessons about best practices. Section 4.1 reviews the performance of waiver and exemption systems. Section 4.2 comments on the financinig of protection mechanisms and the kinds of health benefits they covered. It also discusses general design and implementation features, particularly those regarding beneficiary eligibility. Sections 4.3 and 4.4 examine design and implementation aspects of waivers and exemptions that influence, respectivelv, provider behavior (or the supply of waivers and exemptions) and consumer behavior (the demand for waivers and exemptions). Finally, Section 4.5 offers concluding remarks about best practices. 4.1 Performance of waiver and exemption systems (a) Performance monitoring and evaluation Performance assessment for all cases studied is hampered by a lack of evaluation. None of the cases reviewed systematically monitors performance in terms of who the beneficiaries of waivers and exemptions are and what is the influence of these protection mechanisms on access and out-of-pocket health spending by the poor. The lack of monitoring is a major weakness of waiver and exemptions systems, as the consequences of these cannot be regularly assessed and the policies thereby evaluated. In all cases performance indicators come from ad-hoc studies. Regular monitoring of pro-poor protection systems should at a minimum, through routine facility recording and via periodic household surveys * Record exemptions and waivers granted. * When using individual targeting, establish a data base containing basic information on beneficiaries such as identity number, name, age, sex and geographic location * Compare actual exemption and waiver levels with targets. * Estimate coverage and leakage of protection mechanisms. (b) Targeting methods used Table 28 describes the targeting, or beneficiary identification method used, and presents a summary of performance indicators, including coverage of the target group, leakage, administrative cost, access to health services by beneficiaries of the protection mechanism, and effect of protection on out-of-pocket health spending ("financial burden"). Targeting methods to provide waivers were in some cases a combination of geographic targeting and individual means testing (Kenya, Ghana, Indonesia, Thailand), while in others they were only based on individual means tests (Cambodia, Zimbabwe, and Chile). By combining geographic with individual targeting, some countries, such as Ghana and Zimbabwe, attempted to improve equity in the allocation of public subsidies for health by compensating providers in a system that linked incremental public subsidies to the level of waivers provided. Others, such as Kenya, also combined group and individual targeting but did not accompany the policy of granting waivers by a stream of compensating subsidies. Countries which only use individual targeting, such as Cambodia, could eventually improve equity in the allocation of public subsidies for health if they incorporated geographic targeting criteria. 4. Lessons and best practice 66 C LL~~~~~~U C 0 0 W~~~~~> C~3 co a.- - ~~~~~U Ca) U) 16 gi Q) = .0 M n n .0 >. :5 'D0 (a m m co ~ ~ ~~~~~~~~~- U) C/) coC i -U )- 0 0 CU)rEu. uC ir >W > z z > U) 0. U) 0 C co tn M 0~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~0 Q C E~~~~~ C.r M Ut 00>. U~~~~)C G.Cj- 0 00 0 U) ~~~~~~~~~~~~~~~~~~~~~~~ ~~~~~ ~ ~ ~~o ~~~~~~~~0 U) U m.U). U (U -~~~~25 t 0 .0>~~~~~~~0 -C .0 ~~~E0 U) 0r U U- 0 0 U U)U) U Oh >.~~~~~U) U)~- U) UE U) U) )CoC M ) 40 ( ~ .00( > > > c > * 0 CD C 06 1 ~ a 0. 00~~~~ zm4)() u a) o CU~~~~~~~4 0 o :30c ~~~~~~~~C ~ ~ ~ ~ ~ CU 0 ,,U) C U) .> ~~LL (UC) ~ 3 d ~~~~ jC -~~~~~~~~~~~~-0C0 o 0-U) o~~~~~~~~~~~~~~-.v ~~~~~~~~0c co U .0 0-c ma C.)~~~~~~~~~~~~~~~~~~~~~~~~~~C z < CU.C O)L L : -. O)W U) U) _ >m U ~ ~ -~ C to 0 _ C7 OC(U - 'E > a) mu CL0 U) >-U) >o > Li 9 CU- ~-E>. a) > r m .0... 0C w 0~~~~~~~~~~~~~~~~c D(0 coC05 0o C(U <00 c - - -0 OE U-. C C C_ . E CU ~~~~:93 5 _ a) U)~~% L 2 ) % )=CLMC U) 0 CUOCOin ~~~~~~.0) cmo ) t-.0 - )- 0.0 _ 0 L C.) C" eI 0 (UCU!: 2-1 U) -- _~~~~m C a) p-U)...C(U C.> cm ~ ~ 'L .0~ (D 0U X. 0 0U( 0.0 0C OmnU)C.C~~~~~~~~~ Eo CU8EC ~ L coCO CU(U z a - 0 EL U, in c.Ln ~~~U)C *~~~~~~~O % U) 8 U r --0 ~ 0 M g2~~(CU LO (.CD CDZ C 04 7 0 (1-40 o 'E co E cmr- CL 0 -2 c, cj ~ ~ ~ I- Cambodia has been attempting to reallocate the government's budget for health care on the basis of regional poverty. Success, so far, has been limited, as shown in Error! Not a valid bookmark self- reference.. The figure shows the per capita government health budget allocated to each province in 1999, normalized between 0 and 1, and the equally normalized illiteracy rate (as a proxy for poverty) in the same year. To be equitable, the allocation of public health budgets should at least be correlated with illiteracy, or follow the same pattern as illiteracy. Both measures are indeed positively correlated, with a Pearson correlation coefficient of 0.39. This means that the government has allocated its recurrent health budget more or less according to illiteracy, generally sending more money per capita where illiteracy (and thus poverty) is high, and less where illiteracy is low. Yet the correlation is weak and there are grave exceptions. It can be seen from the figure that many provinces where illiteracy is relatively high - Kampong Speu and Siam Reap are examples- the per capita recurrent public budget is relatively low. Cambodia rnay be a special case because of the country's high reliance on donor assistance in the health sector, although user fees constitute the bulk of health financing, and in these high-illiteracy provinces the public budget is inequitably too low. Increasing the per capita allocation of public resources in the poorest provinces of Cambodia would enable the health facilities substantially to lower their user fees, or to remove thern completely. Such a reallocation would be equitable on a national scale and it would greatly improve access to health services by the poor living in those provinces. Figlure 12 Cambodia: Per capita allocation of government health budget, 1999 1 20 1 00- Normalized per I ~~~~~~~~~~capita recurrent Normnalized E 080 -ealth per capita _ expernditure ilte11racy rate IL 0860 0 40 ,i.ii I I e1 20 C In the poorest countries reviewed, particularly in Kenya and Ghana, coverage of the poor by waiver mechanisms was low. This is not surprising since in Kenya health care providers were not reimbursed for the value of services given to waived patients. In fact, in Kenya members of the health staff were reluctant to grant waivers. In Ghana there was financial compensation to providers for free care (see Table 30), but the flow of funds was uneven and late. In countries such as Cambodia and Indonesia coverage of the poor was low either because of a lack of funds (Indonesia) or because the system was new and there was limited awareness about it among the target population. Countries with high coverage, Thailand and Chile, were also the countries richest in the sample. Both those countries, however, had high levels of leakage. This is not surprising since both defined income levels for eligibility way above their official poverty line. 4. Lessors and best practice 68 Virtually no information was available about the administrative cost of waiver systems. Cambodia's EF kept track of administrative costs or had had their costs evaluated. Such costs were high in projects which had a high component of (relatively very expensive) expatriate labor -Sotnikum and Phnom Penh Urban Health Project. Likewise the is no systematic effort in place in any of the case countries to assess the impact that protection systems have had on rates of utilization of health services and on out-of-pocket expenditure by the poor. The ad-hoc evaluations conducted in Cambodia suggest that EF do promote greater utilization of services by the poor. In rural areas (Sotnikum) this requires that the protection mechanism not only waive poor patients from use fees in health facilities, but also that they reimburse these patients for their health-seeking related costs, such as travel and food. Evidence from Chile (Bitran 1995) also shows that the beneficiaries of waivers enjoy utilization rates for some services that are as high as those of higher- income, self-financed affiliates of the public insurer. 4.2 Financing of waivers and exemptions and design and implementation features (a) Funding With the exception of Chile and Cambodia's EFs, all other case study countries were characterized by a lack of public funding to pay for waivers and exemptions. Kenya simply did not contemplate the creation of a fund to pay for the incremental cost of waivers in public health facilities. Ghana, Zimbabwe, Indonesia, and Thailand had such a fund but the level of resources allocated to it was insufficient to finance waivers for the entire target group. Thailand's case is peculiar because although that country had not been able to cover all of the poor with its free health cards (it covered only 80 percent of them), it had a high level of leakage. Indeed, it is estimated that 45 percent of card holders were above the national poverty line. Thus, Thailand did not lack funds to finance its waiver policy, but rather was poorly using its available funding. Cutting some of the leakage to non-poor beneficiaries would free up enough revenue to finance free cards for the uncovered poor. Chile's situation was in a way similar to Thailand's: coverage of the poor was even higher, but according to recent estimates, leakage was substantial. A reduction in leakage, if at all possible, would enable Chile's government to provide more and better care for the poor, or to reduce general taxes. (b) Health benefits covered Most countries in this review failed to define explicitly the set of benefits subject to waivers and exemptions particularly as it pertains to curative services. Cambodia is making at attempt at defining two basic packages, the minimum and the complementary package of health services, the first for health centers and the second for hospitals. However, the prospect of granting waivers that would cover the entire set of services in these packages has not been contemplated to date. Chile's National Health Fund explicitly states that all primary and preventive health services provided in ambulatory health facilities are to be given at no charge to its beneficiaries. In the case of hospital-based services all care is supposedly to be given to the indigent, yet limited funding often results in rationing through waiting periods which can last several months. (c) Existence and clarity of national policy of waivers and exemptions With the exception of Cambodia, all other countries in the review had an explicit national waiver policy and all had an explicit policy for exemptions (all exempted certain categories of preventive services for all citizens). At the same time, most of the case study countries included in this paper have experienced problems related to their eligibility criteria. Lack of clear identification criteria seems to be one major problem. Often, telling the poor from the non-poor depends to a large extent on the subjective criteria of 4. Lessons and best practice 69 the person in charge of determining eligibility. In Kenya, for example, staff in government health facilities used 10 different poverty proxies to establish eligibility for waivers. In Ghana health facilities seem to have several different definitions of the term "pauper", or the subjects of waivers. Likewise, in Chile there exist multiple criteria for determining eligibility and this has been a source of conflict between different government entities over some patients. In Chile, municipalities and hospitals often disagree on which of the patients are to be waived. Other countries experiencing similar programs have gradually improved clarity of eligibility criteria. Such is the case of Indonesia which has now established four criteria to idlentify the poor. In Thailand, the identification of the poor was initially up to the discretion of directors of public facilities but was later substituted by an income cutoff point. The lack of clear guidelines for granting waivers for user fees diminishes hampers the success of the waiver policy, and makes the monitoring of success a difficult task, and paves the way for a misallocation of subsidies through leakage. Formnulating clear identification criteria is a necessary but insufficient condition to make a system of waivers work. It is also crucial that identification criteria be applicable and easily verifiable. In Zimbabwe, for example, establishing eligibility proved cumbersome if not impossible since target beneficiaries were required to produce information sometimes not available to them. Thailand established an income cutoff point for eligibility, but since the target population belongs mostly to the informal sector, is highly mobile, and receives some in kind income, determining actual income is difficult and subject to arbilrariness. The foregoing discussion leads to a questioning of the appropriateness of using the income criiterion alone for eligibility determination. Case information points to a need to complement to combine the income criterion with other information, or to drop it as use instead other, more observable poverty proxies. In this respect there is general such thing as a general prescription except that the definition of poverty ought to respond to local circumstances and be adapted to the specific cultural context. (d) Taking into account the multiple dimensions of vulnerability of the poor In most cases r eviewed here, possibly with the exception of Chile, the poor often are deterred from claiming waivers as they feel ashamed of being poor. In Cambodia the staff of a large public clinic in Phnom Penh requires that waiver applicants be subject to a public means test in front of all other persons in the waiting room. Shame commonly leads to prospective applicants to forego their right to request a waiver. In 'rhailand, stigma also seems to be a problem limiting access by the poor, as the following statement illustrates: "Sometimnes I self-treat because I do not want [the health worker] to complain that I often get free drugs from the health centre" [a poor from Thailand] (Gilson, 1998) A similar situation was reported by a health official from Ghana: "On one occasion we had 11 patients who claimed they were paupers we infonned that we would publish them as paupers by the following day. After all of them were photographed 4 of them came to settle their bill in full. This is an indication there are those who can afford to pay and yet will refuse to pay." (Garshong, 2001) (e) Assigning responsibility for determination of eligibility Eligibility may be determined by persons or entities within or outside of health facilities. Examples described in this paper show a variety of situations. In Kenya and Ghana eligibility is determined by health facility staff; in Zimbabwe by social welfare offices; in Thailand, Indonesia, and in certain provinces of C'ambodia by the combined and coordinated work of health staff and other government officials (for example village leaders) and clerks; and in Chile by the separate and at times conflicting action of facility staff and municipal authorities. There is no single answer to who should be responsible for the exempl ion process but those engaged in eligibility determination should be aware of the selection 4. Lessons and best practice 70 criteria, adequately trained to carry out their job, and informed about the financial and other constraints governing the protection awarding process. (f) Updating fees and income eligibility thresholds Most of the countries have had problems with adjusting fees and income thresholds to changing circumstances. In Ghana and Kenya, for example, the real value of user fees real has gradually eroded due to inflation, leading to spontaneous and unregulated attempts by providers to update fees in ways that may defeat the policy objectives. In Zimbabwe income cutoff points were not adjusted for inflation making it gradually more difficult for the poor to be eligible, and likewise in Thailand. When fees and income eligibility thresholds are not adjusted with circumstances, de facto protection policies may drift away from their original goal: those in the originally-defined target group are no longer captured by the income thresholds and facilities may set fee levels in way that may hurt the poor. It is therefore crucial that countries wishing to implement protection mechanisms design, from the very beginning, mechanisms for regular adjustment of fees and eligibility thresholds. (g) Institutional aspects Clear guidance to implementers and the availability of institutions, resources, and appropriately trained are essential for the effective application of pro-poor policies. No such clarity existed in most cases reviewed nor was there adequate staffing and supplies to carry out the tasks of waiving patients. In Ghana, for example, facility staff complained about increased workload and cost of stationery associated with the exemption and waiver procedures. In Kenya, the administration of waivers is cumbersome, on average lasting about 1-2 hours. The process of assessing and exempting patients is thereby often delayed or postponed. Countries wishing to implement protection mechanisms have to be aware that the granting waivers and exemptions is a complex process that requires institutional investments. Those designing the kinds of protection mechanisms for the poor discussed in this report should envision those needs for resources and should therefore contemplate mechanisms for creating and making them available on a sustainable basis. This means that there should be written guidelines for waivers and exemptions (with enough flexibility to allow for regional or local variation if necessary) and that health facilities or other agencies granting this protection should know and be reimbursed for their administrative costs. 4. Lessons and best practice 71 Table 29 Waivers and exemptions in case study countries: Funding, design and implementation features Funding Reach General design and implementation Relation- Adjustment ship of fees and between income Feature-s Public exemption Responsive- criteria to resources Existenc criteria and ness of changing available to Defined e of national eligibility economic firnance benefits national Clarity of poverty Who criteria to local circumstan Country exemnptions package policy policy criteria decides? circumstances ces Initially, Not local No, Implied applicable, community Facilities replace lack of Kenya exemriptions No Yes No No leaders, income criteria control on not then, by other poverty fee policy of compensated health proxies each facility facility staff In EFs decision made by Generous, No. But basic No somtimes donor-funded, packages for No Only Eligibilityei sustarnability primary health national t e blise cinae Cambodia with public care and user fee avalabesonanlisad- noowinth o Yse funding hospital care policy hvailable oc basisd in hth currently not being exists ach baseting selta possible implemented ote r facilities health staff decide Funding Unclear covers only because of No Implied aboul 22% of ambiguous Health lack of Ghana reSOLIrceS No Yes esf initioeNo facility Not available control on required to nfistaff fee policy of cover beseficiorp each facility exemptions "paupers" No- Income Eligibility Almost criteria not Smaller criteria Social impossible to adjusted for allowances increasinggSlty determine inflation, Zimbabwe than actually No Yes Yes outdated wy Wetfare income threshold increasingly required due to based on below inflation existing data national poverty___ _ _ _ _ _ __I_ poverty line No: Some Insuflicient of the Indonesia No subseqlNo Yes Yes eligibility Local Not available Y e da 'ta maiccriteria are community Noavibl Ntaalbe non poverty related No ~~~~Almost Nogiilt impossible to criteria way Communitdermn Thailand Insuflicient No Yes Yes income threshold No above y based on poverty available lines information No Initially not, but Eligibility Chile Yies subsequently Yes Yes criteria way Municipalit NoaviblYe yes ~~~~~~~~above y or facility Noavibl Ye y.Lessonsandbestpries 72poverty lines III 4. Lessons an-d best practice 72 4.3 Design and implementation features influencing the supply of waivers and exemptions (a) Provider compensation Wilkinson noted that "there is a systemic conflict between a viable exemption scheme and a viable salary incentive scheme," in his report on the user fee system in Cambodia. By this he meant that it is unreasonable to expect that underpaid health staff that are responsible for, and have the ability to charge user fees, will act in accordance with general equity principles by providing appropriate levels of exemptions. Improving their income through fees will remain in obvious tension with the social aim of exempting the poor as long as budgets are inappropriate and accountability systems are absent. More generally when health facilities are viewed as a whole, there exists an equivalent tradeoff between the extent of exemptions and the aim to bridge any gaps in expenses or in revenue objectives through user fees. The exemption component of Kenya's cost recovery policy is likely doomed to failure as long as the government does not restore to health facilities the revenue foregone from waivers and exemptions. Up until recently, Indonesia faced a similar situation, but currently health care providers do get from the government funds that supplement their budgets to defray waivers and exemptions. Zimbabwe, Ghana, Thailand, Chile and, more recently Indonesia, all have in common the existence of compensated funds for partial and full exemptions provided by government health facilities to patients. The conclusion from the forgoing analysis is that a well-performing system of waivers and exemptions in government health facilities must be in harmony with institutional and individual staff objectives. More specifically, government funds or external funding from donors or lenders, are required to grant providers with the appropriate and minimum financial incentive to exempt the poor. (b) Timeliness of compensation A corollary of the preceding point is that compensated user-fee revenue should reach health facilities promptly. Only Thailand and Chile meet this requirement. In Ghana compensation to health facilities from SDF reportedly takes as long on 8 months to arrive. There, the need by facilities to meet expenses with concurrent funding is at odds with delayed reimbursement by the government. Where compensation exists, it must be timely; otherwise the cost of delayed reimbursement (for example the financial or opportunity cost) may be transferred by the provider to the poor, in the form of higher fees or lower (or fewer) exemptions. Policies seeking to improve the protection of the poor should therefore seek to streamline any bureaucracy involved in the reimbursement of facilities for exemptions granted. Reimbursement procedures may be timelier in various ways. For example, the regular allocation of compensation funds from the central level to regional health authorities, or to regional funds, may make compensation more opportune and predictable. Or, in the absence of a decentralization framework, monthly budgets sent from the central level to facilities may include an "exemptions allowance" equal to the monthly target for that facility, with any (relatively smaller) adjustments for differences between actual and budgeted exemptions being made later. (c) Harmonizing the incentives created by different payment mechanisms Some case studies show that quality of services given to exempted patients is lower than quality for paying patients. Such is the case in Thailand. Even though providers are compensated there, "quality discrimination" may well be related to the incentives conferred by payment methods and levels which 4. Lessons and best practice 73 differ across patient groups. Health facilities are paid mostly on a capitation basis, and the total capitation payment is below the revenue foregone by providers through waivers. At the same time, facilities get paid on a fee-for-service basis by paying patients or by the patients' insurer. Under this circumstance facilities have a strong motivation to select the "best paying patients", not only because reimbursement levels are higher but also because with a capitation-based payment system, facilities have an incentive to under-prov de services, especially in the absence of a tight monitoring system. As the Thai situation illustrates, payment mechanisms and payment levels have an impact on the effectiveness of protection mechanisms. Countries adopting poor protection systems where providers are compensated for their provision of free or subsidized care to the poor must give serious consideration to the payment method and payment level. Adopting the wrong payment or the wrong level might result in some "crowdinig out of the poor". 4. Lessons and best practice 74 Table 30 Waivers and exemptions in case study countries: Summary of Supply-side design and implementation features Supply-side design and implementation features Financial Feature-> Relationship significance of between staff compensation Existence and Compensation of User fee remuneration for exempt level administrative revenue kept at and user fee patients, relative Time frame of Country of compensation cost facility revenue to full price compensation No: Granting of waivers is Kenya No resource intensive Yes Staff reluctant to Not available Not available but no give waivers compensation available 50% of revenue Yes: But it does can be used to Cambodia Yes not include the 9% pay for salaries Compensation Monthly cost of expatriate and accounts for equal to full fee staff up to 95% of staff income Yes- But Revenue only to Evidence that cumbersome flow No Evidence that i be used to pay for Eidement Ghana from central level facmplties and staf subsequentty yes drugs and other Not available lasts on average 4 to local level complain on this non personnel months facilities expenditures Yes: Albeit Reimbursement cumbersome flow Inbally not, but takes up to 8 Ztmbabwe from central level No subsequently yes Not available Not available months as it to local level ~~~~~~~~~~~~~~~~needs approval facilities from Harare. Health centers can keep 25% of Initially not, but revenue; balance Delays in Indonesia subsequently yes No distributed with Not available Not available compensation the budget allocation from the province Compensation is lower than fees Thailand Yes No Yes Yes paying patients Not available (insured and uninsured) Fees paid for exempted patients Yes. But equal to fees paid Yes. Fees paid for Yes.Btrodny for patients exempted patients bdextralordinary insured in the Chile equal to fees paid Not explicit btu ent allout No public system. Not available for insured taker into aclcunt Provider patients use fcollecte indifferent between serving poor or non poor patient. 4. Lessons and best practice 75 4.4 Design and implementation features influencing the demand for waivers and exemptions (a) Disseminating pro-poor protection policies and mechanisms Lack of knowledge on protection mechanisms by both health staff and potential beneficiaries is a recurrent is,ue in most countries with the exception of Chile, where the poor are screened by public facilities by an ad-hoc procedure if they have not been previously identified. Under-coverage will be a constant problem when the poor do not know they are eligible for free or subsidized care and when health facilities are not aware of whom to exempt In Ghana, for example, most potential beneficiaries were unaware or misinformed about waivers and exemptions. Also, health facilities were not interested in disseminating information on waivers and exemptions. In Kenya, most inpatients and outpatients were unaware of waivers and exemptions. In Zimbabwe many of the poor had not applied for waivers because they did not know of the possibility of getting wai vers through the SDF. In Indonesia, neither the poor nor health facilities knew that the health card officially entitled the holder to free primary care services and to free referrals. The problem of information dissemination seems to be especially challenging in Indonesia and Thailand. In Indonesia information on protection mechanisms originates at the provincial level and has to move down to districts, sub districts, villages, sub villages and finally to health facilities and to the target population. In Thailand, lack of information was considered a major obstacle to coverage of the poorest, leading to a series of major changes in the dissemination policy. Accordingly the village head was given a more proactive role in the application procedure. Also, in many countries (e.g., Indonesia) health staff was seldom trained to grant exemptions and waivers. The following statement from Gibbons (1995) illustrates this point: The staff at the health centers and health sub centers did not receive formal training to explain the uses and implementation. In Lombok Barat village, there was one meeting with the health center leaders and the district health official. The information given oat the meeting was minimal. After the meeting, the healih center leaders were responsible for training the health sub center workers. This training of the health sub center workers consisted of brief conversations with the head of the corresponding health center. Little training actually occurred. Countries wishing to implement mechanisms to mitigate the impact of user fees on the poor have to be aware that dissemination of policies is not done by circulating a few government leaflets. A dissemination policy is effective only when the poor and health staff know about protection mechanisms. For this to take place, some general recommendations can be made: * Use different media (newspaper, radio, TV, house by house information, village meetings, schools to disseminate information on protection mechanisms * Tailor dissemination mechanisms to special characteristics of the poor. For example, take into account that the poor tend to live further off, have less access to formal media, tend to be less educa,ted, and have longer working hours. * Introduce accountability in dissemination policy. For example make village heads accountable for disseminating information and evaluate performance of officials on this issue. 4. Lessons and best practice 76 Table 31 Waivers and exemptions in case study countries: Summary of Demand-side design and implementation features and of monitoring and evaluation efforts Demand-side design and implementation features Feature-s Accessibility to subsidized Access costs Social Discrimination services without Existence of Dissemination of to exemption importance of of the formal alternative Country information system stigma exempted exemption providers Majority of potential Kenya beneficiaries not Not available Not available Not available No aware of exemptions It varies Sotnikum's Important EF currently according to Cambodia promoting scheme, No high anccdiotal No No Notgood but cautiously, to infotal keep EF sustainable infommabon Majority of potential beneficiaries not Ghana aware of exemptions, Not available Not available Not available health staff not knowledgeable of exemption categories About 50% of the Evidence that Zimbabwe population have participation Anecdotal Zimbabwe never heard of the costs are high evidence Not available waiver policy and may deter demand Evidence that Yes: Anecdotal potential beneficiaries evidence that Indonesia exemptions and that participation Not available Not available health staff does not cost may deter know policy well demand Problems revealed that the poor have Yes, Anecdotal Yes: Anecdotal Thailand difficulties accessing Not available evidence evidence information on health card Chile Not available No Not available No 4. Lessons and best practice 77 * Make sure no major incentives exist to conceal information on protection mechanisms. As has been shown earlier, negative impact of protection mechanisms on staff's and facilities income may constitute a major obstacle to even the most sophisticated dissemination policy. This point clearly illustrates the strong relationship that exists among many of the major implementation issues related with the implementation of protection mechanisms for the poor. * When the process is implemented in a top down manner (from national level to local level) oppontunity should be given for discussion and clarification between official at different levels on how to implement the process and should not be limited to physically passing on government circulars. 4.5 Concliusion Different countries have tried different approaches regarding waivers and exemptions for health services. Those that have carefully designed and implemented waiver systems (e.g., Thailand and Indonesia) have had much greaLter success in terms of benefits incidence than countries that have improvised such systems (Ghana, Kenya, Zimbabwe). Key to the success of a waiver system is its financing. Systems that compensate providers for the revenue forgone from granting exemptions (Thailand, Indonesia, and Cambodia) have been more successful than those who expect the provider to absorb the cost of exemptions (Kenya). Where waiver systems exist, performance will improve with the timeliness of the reimbursement. Other success factors include the widespread dissemination of information among potential beneFiciaries about waiver availability and procedures; the awarding of financial support to poor patients for non-fee costs of care, such as food and transportation (as in Cambodia); and the existence of clear criteria for the granting of waivers, thereby reducing confusion and ambiguity among those responsible for managing the system and among potential recipients. Those facing the task of adopting a system of waivers face multiple design options. These include the following, among others: should exemptions be granted to whole groups or on the basis of individual targeting (the review finds that most systems are based on the latter)? Should waivers or exemptions be permanent or temporary? How frequently should eligibility be reassessed? Should waiver eligibility be determined ex-ante, in the household, or when individuals seek care in the facility? The review examines various approaches taken by countries, but assessing their relative practical merits is difficult, as the evidence is scattered and mixed. 4. Lessons and best practice 78 Bibliography Abel Smith, B. 1993. "Financing Health Services in Developing Countries: The Options." NU News on Health Care in Developing Countries, Vol. 7. APHEN (Asia-Pacific Health Economics Network), 2001, The Universal Coverage Policy of Thailand: An Introduction, download at: http://www.unescap.org/aphen/thailand_universal_coverage.htm Appleton S, 2001, User fees, expenditure rationalization and voucher systems in education in G.Mwabu, C Ugaz and G. White (ed.) New Patterns of Social Provision in Low Income Countries, Oxford University Press, forthcoming. Appleton S, Education and Health at the Household Level in Sub-Saharan Africa, 2000, CID Working Paper 33, January 2000, Harvard University Center for International Development at Harvard University. Ausaid (Australian Government Overseas Aid Program), 2001, The Impact of the Asian Financial Crisis on the Health Sector in Indonesia. Download at: http://www.ausaid.gov.au/publications/pubout Atkinson, A., and J. Stiglitz. 1980. Lectures on Public Economics. McGraw-Hill Book Company, New York, New York. Barnum, H., J.Kutzin, and H. Saxenian. 1995. "Incentives and Provider Payment Methods." HRO Working Papers No. 51. The World Bank. Washington, D.C. Bennet S. and L. Gilson. 2000. "Which Financing mechanisms are pro-poor?", DFID Issue Paper Health Systems Resource Centre. London. Bitran R, and C. Mufioz, 2000, Targeting Methodologies: Conceptual Approach and Analysis of Experiences, LAC-HSR No. 45, Washington. Bitran R, et al., 2000, Equity in the financing of social security for health in Chile, Health Policy 50, 171 - 196, Elevier Science Ireland Ltda. Bitran, R. 2002. "Protecting the Poor under Cost Recovery for Health Care in Cambodia." Bitran Associates for the World Bank. Cameron L, 2000, The Impact of the Indonesian Financial Crisis on Children: An Analysis Using the 100 Villages Data,, Dept of Economics University of Melbourne, Australia. Download at: www.ecom.unimelb.edu.au/ecowww/lcameron/papers/unicefvO2.pdf Charoenparij Sriracha, et al., 1999, Health Financing in Thailand: Final Integrated Report Thailand: Health Management and Financing Study Project, ADB No. 2997-THA, Management Sciences for Health (MSH), Boston. Coleman N.A, 1997, The Uneven implementation of user fee policy in Ghana, Research Paper, Harvard School of Public Health, Boston. Download at: http://www.hsph.harvard.edu/takemi/Reseaarch%20papers.htm Dercon Stefan Ruttens Christel, 1998, Cost recovery in health care in Africa: a Review of the Principles and Effect on the Poor. Dayl Donaldson,, Supasit Pannarunothai, Viroj Tangcharoensathien, 1999, Health Financing In Thailand Summary Review And Proposed Reforms, Management Sciences for Health, Health Systems Research Institute, Ministry of Public Health. Diop F, Yazbeck A, Bitran R. 1995. The impact of alternative cost recovery schemes on access and equity in Niger. Health Policy and Planning, 10 (3): 223-240. Dror, D. and A.S. Preker. 2002. Eds. Social Re-Insurance: A new A[pproach to Sustainable Community Health Financing. Washington: World Bank/ILO. Bibliography 79 EDI (Economjic Development Institute of The World Bank), 1996, Sustainable Health care Financing in Southern Africa, Papers from an EDI Health Policy Seminar Held in Johannesburg, South Africa, June 1996, Download at: http://www.worldbank.org/healthreform/library/sa/shaw.pdf Ensor T, 1996, Access and Payment for Health Care: The Poor of Northern Vietnam, The International Journal of Health Planning and management, Vol 11, issue 1, 1996. Garshong 13, Ansah E, Dakpallah G, Hujits 1, Adjei S, 2001, A study on factors affecting the imrlernentation of the exemptions policy in Ghana,, Health Research Unit, Moh, Danida, Ghana. Gertler, P. and J. Hammer. 1997. "Financing and Allocating Public Expenditure in the Health Sector." Draft (January 1). Haas School of Business and School of Public Health, University of California at Berkeley, and Public Economics Division of the World Bank. Gertler P, Locay Luis, Sanderson W, July 1987, Are user fees regressive? The welfare implication of health care financing proposals in Peru. NBER working papers 2299. Gibbons Donna M, October 1995, Equity and coverage of Health Care Provision in Indonesia, The Kartu Sehat Program, Basics, USAID. Giedion, U, It. Bitran, R. Munioz. 2002. Evaluation of the Surinamese MSA Health Card System. Bitian Associates and DAH Associates, for Suriname's Ministry of Health and the Inter- American Development Bank. Gilson L. el al, 1998, Exempting the Poor: A Review and Evaluation of the Low Income Card Scheme in Thailand, PHP Departmental Publication No. 30, ISSN 0926-6115, Department of Public Health & Policy, London School of Tropical Medicine. Gilson, L. and S. Russell. 1994. "Can Fees Recover Costs?" Health Action, Issue 9. Grosh, M. (ed). 1992a. "From Platitudes to Practice: Targeting Social Programs in Latin America: Volumle 11, Case Studies," Latin America and the Caribbean Technical Department Report No. 21. The World Bank. Washington. D.C. Grosh, M. 1995. "Toward Quantifying the Trade-Off: Administrative Costs and Incidence of Targeted Programs in Latin America." In Van de Walle, D. and K. Nead (eds.). 1995. Public Spending and the Poor. Theory and Evidence. The Johns Hopkins University Press for the World Bank. Gwatkin D, 2000, The Current State of Knowledge about Targeting Health Programs to Reach the Poor. Hongoro C, Chandiwana S (1994) Effects of User Fees on Health Care in Zimbabwe, Blair Laboratory / MoHCW, Mimeo Harare, 1994 (quoted in Loewinson, 2000) Illiff, Peter, 1995, Health for Whom? Mother and Child Care in Times of Aids, Poverty and ESAP, Mambo Press: Gweru, Zimbabwe (quoted in World Bank, 1998a) IPAR Policy Brief No 1, 1999, Effectiveness of Waivers and Exemptions in Addressing the Equity Objective: Lessons From Selected Facilities at: www.ipar.or.ke/healthl.htm. Hardeman, W 2001. "Considering equity in health sector reform. Case study of a New Deal in Sotnikum, Cambodia. Manuscript submitted for final thesis to obtain the degree of Master of Arts in Politics anof Alternative Development Strategies, The Hague. Leighton Cli, I)iop F, Abt Associates Inc, 1995, , Protecting the Poor in Africa, Impact of Means testing on equity in the health sector in Burkina Faso, Niger and Senegal. Technical Note 40, PHR projeci Leighton, Abt, 1995, 22 Policy Questions about Health Care Financing in Africa, USAID. At: http://vww.usaid.gov/regions/afr/hhraa/questions/engl .pdf Lennock, J., 1994, Paying for Health, Poverty and Structural Adjustment in Zimbabwe, Oxfam (UK & Irelanel), Oxford. Bibliography 80 Lewis, M. 2001. "Excerpts from Who is Paying for Health Care in Eastern Europe and Central Asia?." Human Development Sector Unit, Europe and Central Asia Region, The World Bank, Washington, D.C. Litvack, J.I., Bodart, C. 1993. "User Fees Plus Quality Equals Improved Access to Health Care: Results of a Field Experiment in Cameroon", Social Science and Medicine, Vol. 37, No. 3, 369-83. Loewenson Rene, 2000, Putting your money where your mouth is: participation in mobilizing and allocating health resources, Training and Research Support Centre, Harare, Zimbabwe. Download at: www.tarsc.org/docs/respart.pdf Ministry of Health of Zimbabwe, 1995, Study on the Effects and Impact of Hospital User Charges on Health Care Delivery in Zimbabwe. Ministry of Health and Child Welfare. Mimeo. Harare quoted in World Bank 1998a. Moses S et al. 1992. Impact of user fees on attendance at a referral centre for sexually transmitted diseases in Kenya. Lancet 340: 463-466 Newbrander W, Collins D, Gilson L, 2000, Ensuring Equal Access to Health Services, User fee systems and the poor, MSH.. Newbrander W, Njau M, Auma C, 1995, Equity and Coverage of Health Care Provision in Kenya, Basics-USAID at http://basics.org/Publications/Kenya/ Nolan B., Turbat V, 1995, Cost Recovery in Public Health Services in Sub-Saharan Africa, EDI Technical Material, World Bank National Economic Social Development Board (NESDB), 2000, Social Sector Program Loan, Health Finance and Management Technical Assistance, Health Management and Financing Final Integrated Report. At; Sohttp://www.nesdb.go.th/Main_menu/Hum_soc/data/human_soc- e/hfmopen/HFM-VIEW.htm Nyonator F, Diamenu S, Amedo E, Eleeza J, 1996, Caring for the Health of the Poor-Policy versus Implementation. A baseline Evaluation of Exemption practices within Health Facilities in the Volta Region of Ghana, Volta Regional Health Administration , Ghana. Download at: http://www.danida-health- Nyonator, , Joseph Kutzin, 1997:Health for Some? The Effects of User Fees in the Volta Region of Ghana", Volta Regional Health Administration Volta Ghana. Download at: http://www.danida- health- Osuga, B. and E. Nordberg. 1993. Effects of new service charges on attendance at rural health facilities in Kenya. East African Medical Journal. 70(10):627-3 1. Owino W, Abagi 0, 2000, Cost sharing in education and health in Kenya, Draft, Desk study commissioned by the Department for International Development (DFID)-Eastern Africa, Institute of Policy Analysis and Research Nairobi, Kenya. Owino W, Were M, 1999, Enhancing Access to Health Care Among Vulnerable Groups: The Questions of Waivers and Exemptions, Discussion Paper No. 14/99. Institute of Policy Analysis and Research, Nairobi, Kenya. Download summary at http://www.ipar.or.ke Owino, W, 1998, Public Health Sector Pricing Practices, Discussion Paper No. 13/98 Institute of Policy Analysis and Research, Nairobi, Kenya. Summary at http://www.ipar.or.ke Pannarunothai S, Mills A. 1997, The poor pay more: health related inequality in Thailand. Soc Sci Med 1997;44:1781-90. Preker, A.S. Forthcoming. Ed. Health Care Financing for Rural and Low-Income Populations: The Role of Communities in Resource Mobilization and Risk Sharing. Washington: World Bank. Bibliography 81 Reddy Sanjay, Vandemoortele, 1996, User financing of Basic Social Services, Unicef Staff Working Papers, Evaluation, Policy and Planning Series, Unicef,. Saddah F, Me nno Pradhan, Robert Sparrow, 2001, The effectiveness of the Health Card as an instrument to c nsure access to medical care for the poor during the crisis, World Bank. Saprin, 2001, Impact of SAP on availability of and access to health care. Draft. Download at. http://Hwww.saprin.org/ghana/research/gha_health.pdf Sedlacek C,, llahi N, Gustafsson-Wright E, 2000, Targeted Conditional Transfer Programs in Latin America: An Early Survey, Paper prepared for the Regional Study: Securing our Future, Office of the Chtief economist, Latin America and Caribbean Region. The World Bank. Sen, A. 1995. "The Political Economy of Targeting." In van de Walle, D. and K. Nead (eds.) 1995. Public Spending and the Poor. Theory and Evidence. A World Bank Book. SMERU newsletter 2000, Health services during the crises, semru newlesser, no 09, January-April 2000, SMERU research institute. Download at http://www.smeru.or.id/ Sriracha Charoenparij, Somsak Chunharas, Dayl Donaldson, Daniel Kraushaar, Supasit Pannorunothai, Suthamn Pinjaroen, Supattra Srivanichakorn, Paibul Suriyawongpaisal, Viroj Tangcharoensathien, Arce Valyasevi, 1999, Health Financing in Thailand: Final Integrated Report. MSH Suksiriserekul Somchai, 1998, Free medical care for the poor in Thailand: a hit or miss poverty reduction scheme, Number 8, Poverty Alleviation, EDAP JOINT POLICY STUDIE S, Korea Development Institute, Korea. Suksiriserekul Somchai, year?, Thailand, Chapter 3 in: The Poor at Risk: Surviving the Economic Crisis in Southeast Asia, Final Report of the Project Social Safety Net Programs in Selected Southeast Asian Countries, 1997-2000 Prepared by the joint Canada-Southeast Asia Project Team. Centre for Southeast Asian Research, Institute of Asian Research, University of British Columbia. Downlioad at: http://www.iar.ubc.ca/centres/csear/SSN/TOC .htm Sumarto S, Su ryahadi A, 2001, Principles and Approaches to Targeting. Smeru Working Paper Tangcharoensathien V, 2001, Universal Coverage: Experience from Thailand, Regional Consultation and Technical Workshop on Health Systems Performance Assessment, New Delhi, Inidca, 18-21 June, WHO. At: http://w3.whosea.org/hspa/presDrViroj20.htm The Nation, 2002, Baht-30 Healthcare, Is It Succeeding?, Special edition, April 17,18,19, download at: http://www.nationmultimedia.com/specials/30bt/ 170402 1.shtml Terence H. Hull Paper Alleviating Poverty: Conundrums of Planning, Administration and Governance, 1999, Demography Program, Australian National University USAID, 1998, Latin America and the Caribbean Selected Economic and Social Data, download at: http://lanic.utexas.edu/la/region/aid/aid98/povertv/tab5 .html Watkins K. 1997, Cost-Recovery and equity in the health sector: issues for developing countries, Working paper, Oxfam.. Willis C, 1993, Means Testing in Cost Recovery of Health Services in Developing Countries MAR No. 7, PHS Project, 1993. World Bank, 1995, Best Practice example of overall assessment, Annex 3.a. of Malawi: primary Education Project, at. World Bank. 1997. Health, Nutrition, and Population Sector Strategy Paper. Washington, D.C. World Bank, 1998a, The Impact of World Bank Support to the HNP Sector in Zimbabwe, Sector and Thematic Evaluations Group Operations Evaluation Department Bibliography 82 World Bank. 1998b. HNP Poverty Thematic Group. Report on the seminar protecting the poor from increased user charges in government health facilities. World Bank. World Bank. 2000. Health Strategy in a Post-Crisis. The World Bank Human Development Sector Unit East Asia and Pacific Region November 17, 2000, Report No. 21318-IND. World Bank. 2001 a. http://www.worldbank.org/html/extdr/pb/pbpoverty.htm). World Bank. 200 lb. Morocco Health Financing Brief (Draft). World Health Organization Study Group. 1993. Evaluation of Recent Changes in the Financing of Health Services. WHO Technical Report Series 829. WHO, Geneva. Bibliography 83 0~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~0 0~~~~~0 o w~~~~~~~~~~~~~ E (U 01 01 01 (U (U~~~~E C ID ID U)D C ol 0i 0 0 0 0 ) 0 0 E 0 D CD ID> E CD00 _ -2 - 0 E >~~~ 'a 0* 5 0) (U> 0 (U .2 o~~~ ~~ -c CMI to~ cn' DU)- 0. C 0 * C.) 0k-c~~~C >, Eu ~I>- H 0 '~~~~ ~ 0 t~ CE In * ID-- IDm-.2C 0 o .C0 E0 =3 U U 02.~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~0 )U U) a)~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~0 ~~ ______~~~~~~ 0 0 0 -0 L. -~~~~~~~~~~~ -~~~~~~ > I -aD C LLJ 0 cts~~I ID0 I 0 S~~~~ u J 0)2) U, 0 E 0~~~~~00 6 c ~ c- 0 CC D C 0 ~~~ -~~~ , 0Y 0 . I 4-~~~ ID ID... ID- I - L. .0 0 > IDD. I >E (U .0E ~~ -2E ~ 2 0 .C ~ 2 E 0 >,ID~~~~ 0) - 0 ID0 0 C. " I.' C 0 ID' C0 0 CAU) ~~~. -) a). 0. . ~ e ID . >D' >, E- co E - C .0 E ~ ,,0( ~~~~~~~~~~~ In~~~~~~~~~~~~~~~~~~~~~~~~~~~> C 0 ID -0. >.. 0 n C I~' 0 IDO C 0.L U IA ~~ ~ Ea ~ . .2~ I't - rA ~ D I C, .- D ~~ 0) cE-.~~1 0 InE0 noID 0 E 0 E~ ~ D 2 I 2 00 C0m _ 0~~~~4 W L --- ID 0 I ci) -E E C E ~~ o.. ~~~ -8 CI U) '-( , U E > 00.O 0 2 I 00 o 2 841 01 I E 0 -0C,I , 0C0 ID IAO 0 C~~~~~~~~~~~~~~ 00 ~ ~ CZ I0 (U( IDO0E >,,Z E D10 0 - I (UD 00 EI cL C) 01. IC. ~~)0 0 .0~~~~~~~ .CC ~~~~~EC C~~~: _ ID a) 0 ID~~~~~ ai ID 01 IDIco O _ ID . 0 b 0 0 U -8 0 ..N C 0C (IDU _ C>'~ ~~~~~ IDL v> 1o ) 2 0 < tol .(UC 2 E cmC )0 C ID E m 0 C.) ~~~~~ID01 E o 3 (U 01g UC 0 - C Cc CD~~ 0 00 00 m0 ) C.) o~~~~~~~~~~~~~~~~~~~~)Z : ID U n ID 0 In>-00 In 2 0 2 0)D0 u~~~~~~~H 2 C~~~~~~~2 C aC 4 FZ m 0 OCE01- Hc~~ ~ ~ Ifl (U U- W I a.. C.) -~~~U) to di in a3 o .0 o 4) r- in C -o c a). C E) 4) a ) 4) C ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~~ ~ ~~~~~~~~~~~~~~~~~~~~~~~C o * co >~~ C c~~~o > 4) c o i to ~~~~~ 0~~ )t o 0) i= co m 34 0 E 2 to- . 0) ~~~~~C0a 4)4 10d)25I 4) h _)r d) C M> 0 Z-c4) 0) co0 2.0 _' )0 d o E~~~~~~~o2 ;-~- jto4)L U :5 d) M M >o t CL 4) E S m E n ) E C ~ ~ 8 O )gEE 4) 4) W (D CL 4) CD CO -0 4) - 0 .~~~~~~~~~~~~~~ E- o o .I C a) ~~~~~~~~~~~~~~~t -o 0)~~~~~~~~~~~~~~~~~~~~~a E *~~~ ~~O 0 . -E C 4) ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~~4 E CD4rr _D c rr 0 C~~~~~~4 4IDto C 4) a) 4) U2.0~~~~~ cc.Eu2EO C O.000toi~~4) ) 'a C I g ~ CD -0 4 co 0 m 0~~~~~~~~~~~~to . a) ~~~~~~~~~ ~~~~~~~~~~~~~~~~~~~~~~0 0 ,2C _ L- 0~~)0~- u o to~~ ~~4)-.-'0)~~2) M 0~~~~fl t 0 S >5 5S S5 S M m~~~~ E~ ~~~~~4~) 4) 0 to~~~~~~~~~~~~~~~~~~~~60 D _C S S 7a 5M 5, S, 2 OM C E0))m2t 4) 0.~~~~~~~ -~~~~i~~~j~~~ ~ ~~ o-~~~~~- v) E E R 4)0 C... ~ oCt S 2g..00c. - 4) t0.>EC O> ~ ~ ~ ~ ) ) ID. 0 8 d)U) > t 4 )OL~ to~~~~~i~~~ ~~~~' 4)8 C ~~~~~~~~~~~-~ 04 - L > .= Cd )2.-o 0 4 vt 5tOv - ~~~~-. ...~~~~~ C02. ~ O o c - 4 o .~~'E~ - ai E_ _ _ __cm_ _ _ __C r- 00 0 0 E CL w :3 4) CL 11 06T 0 0 'r- :.s 0 a 4) Z5 :E 0 CL 0 u .0 9. m m m FO -s -a CL m =3 0 U) CME0 co 9 a)c -s9> > r- tn > E =-a coo o 0 0 w'O (D co c LL LL .9-- m m m to m CP 0>,2 mCM -L- a) r cm r- .- M0 00) >1 co c .2 >, CL E0 CL co cx ;= C. 2 r- co (D 0 :3 C 'a -0 0) 0 En E M 0 -2 m V5 E Oj T E 0co m 0 0= 0 0 > Ul) (t :cEa. 42 0CL 0 Hi L I? CL 0 m Lm CL > 0 a? ica .2 oom -o o ti (3) > m m3n 4511 CD 0 r- E 'O c (L) U) w c-D 0> r- m 2cm 0 co L 0 = .- (D r m 0 E = r m > r(D a -EU gn U) C co -2E 'So V m 0 CO Z, 8 U) a) c 3: 0 -0 1.e > wr-mmM s2 E n a) E -2> > .2 r- N cn ID U) (D 0 rr M0 (D m CL L)a It 0 L)o m m 0 HLmll jc 11 (D E m ;_j co 9 cn E ID CD> CL c a 0(D 2 (D CD a. w> m Jle .2 c, E 0Co E 7r '; 0 0) IL) > CL 0 m2 022 E E cL a. U) (D -.Hi 0- im2 'O U) r 'L. C) -he 0 75 0 .- CL(D E>, 2., Z, o CL !.U) co U)2 >A! 4) 0 CL m CL (D - co Ml E E 'D E Lu 4- 0 co (D 0 ?.. m U) 00 E la M.S 0) r- S2 r- ,g co 0 .0 4) (D 7i 0 0 co E >Co C" Co Co 0 m c w .2 - CD E t5 '6 U) c'mE 2 -2I ff L. - 0) w Elm CA m MA r CL m 0 -E0 m 7F co V; C: 0 0 cx co FM or :3 M be D= -ii -cim a - < 0 co m m 2p F 2 co 0) -0c CL In C,O 0 -0c m - -2:- 2'o :6 r-m a 0r, El El 32 C Z0 S o co m n 0 OE 1 - :E [ I a- 0 'R s ccm, .2 0 3: (Dz p- co ?; 3: Social Protection Discussion Paper Series Titles No. Title 0307 Ageing and Pensions in the Euro Area Survey and Projection Results by P. C. Rother, M. Catenaro and G. Schwab 0306 Safety Nets in Transition Economies: A Primer by Louise Fox 0305 Child Work: An Expository Framework of Altruistic and Non-Altruistic Models by Furio Camillo Rosati and Zafiris Tzannatos 0304 Pension Reform in Croatia by Zoran Anusic, Philip O'Keefe, and Sanja Madzarevic-Sujster 0303 Joblessness and Precarious Work in Bulgaria: Addressing the Multiple Aspects of Vulnerability in the Labour Market by Alexandre Kolev 0302 Systemic Shocks and Social Protection: Role and Effectiveness of Public Works Programs by Kalanidhi Subbarao 0301 What Role for Safety Net Transfers in Very Low Income Countries? by W. James Smith and Kalanidhi Subbarao 0233 Public Attitudes Matter: A Conceptual Frame for Accounting for Political Economy in Safety Nets and Social Assistance Policies by Carol Graham 0232 Ensuring Access to Essential Services: Demand-Side Housing Subsidies by Harold M. Katsura and Clare T. Romanik 0231 Gender and Risk in the Design of Social Protection Interventions by Kene Ezemenari, Nazmul Chaudhury, and Janet Owens 0230 Vulnerability and Poverty Measurement Issues for Public Policy by Jean-Yves Duclos 0229 Emerging from Ethnic Conflict: Challenges for Social Protection Design in Transition Countries by Christian Bodewig 0228 Child Work in Zambia: A Comparative Study of Survey Instruments by Niels-Hugo Blunch, Amit Dar, Lorenzo Guarcello, Scott Lyon, Amy Ritualo and Furio Rosati Social Protection Discussion Paper Series Titles continued No. Title 0227 Strengthening Public Safety Nets from the Bottom Up by Jonathan Morduch and Manohar Sharma 0226 Incentives and the Role of Institutions in Provision of Social Safety Nets by Chris de Neubourg 0225 Food-Based Safety Nets and Related Programs by Beatrice Lorge Rogers, Ph.D. and Jennifer Coates, M.S. 0224 Subsidies as a Social Safety Net: Effectiveness and Challenges by Harold Alderman 0223 Assisting the Poor with Cash: Design and Implementation of Social Transfer Programs by Steven R. Tabor 0222 Effectiveness of Lending for Vocational Education and Training: Lessons from World Bank Experience by S. Canagarajah, A. Dar, R. Nording and D. Raju 0221 Participation of Children in Schooling and Labor Activities: A Review of Empirical Studies by Amit Dar, Niels-Hugo Blunch, Bona Kim and Masaru Sasaki 0220 The World Bank and Children: A Review of Activities by Iqbal Kaur and Zafiris Tzannatos 0219 Managing Public Pension Reserves Part II: Lessons from Five Recent OECD Initiatives by Robert Palacios 0218 Guidelines for Assessing the Sources of Risk and Vulnerability by Karin Heitzmann, R. Sudharshan Canagarajah and Paul B. Siegel 0217 Czech Pension System: Challenges and Reform Options by Esperanza Lasagabaster, Roberto Rocha and Patrick Wiese 0216 Extending Social Protection to Informal Workers in the Horticulture Global Value Chain by Armando Barrientos and Stephanie Ware Barrientos 0215 Social Fund Support of Microfinance: A Review of Implementation Experience by Alexandra Gross and Samantha de Silva Social Protection Discussion Paper Series Titles continued No. Title 0214 Income Support Systems for the Unemployed: Issues and Options by Milan Vodopivec and Dhushyanth Raju 0213 Social Protection @ Your Fingertips. Using Information & Communications Technologies in Social Protection by Knut Leipold 0212 Short-and Long-term Impacts of Economic Policies on Child Labor and Schooling in Ghana by Niels-Hugo Blunch, Sudharshan Canagarajah and Sangeeta Goyal 0211 Supporting and Expanding Community-Based HIV/AIDS Prevention and Care Responses: A Report on Save the Children (US) Malawi COPE Project by Susan S. Hunter 0210 World Vision's Experience Working with HIV/AIDS Orphans in Uganda - 1990-1995 by Joe Muwonge 0209 The Reformed Pension Systems in Latin America by Jose E. Devesa-Carpio and Carlos Vidal-Melia 0208 Mandatory Annuity Design in Developing Economies by Suzanne Doyle and John Piggott 0207 Long-Term Welfare and Investment Impact of AIDS-Related Changes in Family Composition: Evidence from Uganda by Klaus Deininger, Anja Crommelynck and Gloria Kempaka 0206 Child Labor Handbook by Alessandro Cigno, Furio C. Rosati and Zafiris Tzannatos 0205 An Overview of Labor Markets World-Wide: Key Trends and Major Policy Issues by Gordon Betcherman 0204 Options of Public Income Support for the Unemployed in the Philippines and Social Protection by Jude H. Esguerra, Makoto Ogawa, and Milan Vodopivec 0203 Unemployment Insurance and Unemployment Assistance: A Comparison by Wayne Vroman Social Protection Discussion Paper Series Titles continued No. Title 0202 Does Eurosclerosis Matter? Institutional Reform and Labor Market Performance in Central and Eastern European Countries in the 1990s. by Michelle Riboud, Carolina Sanchez-Paramo and Carlos Silva-Jauregui 0201 Pension Reform and Capital Markets: Are There Any (Hard) Links? by Eduardo Walker and Fernando Lefort 0131 Child Labor, Nutrition and Education in Rural India: An Economic Analysis of Parental Choice and Policy Options by Alessandro Cigno, Furio Camillo Rosati and Zafiris Tzannatos 0130 Social Protection and the Informal Sector in Developing Countries: Challenges and Opportunities by Sudharshan Canagarajah and S.V. Sethuraman 0129 Chile's Pension Reform After 20 Years by Rodrigo Acunia R. and Augusto Iglesias P. 0128 Labor Market Regulation: International Experience in Promoting Employment and Social Protection by Gordon Betcherman, Amy Luinstra, and Makoto Ogawa 0127 Generational Accounting and Hungarian Pension Reform by R6bert I. Gal, Andras Simonovits and Geza Tarcali 0126 Orphans and Other Vulnerable Children: What role for social protection? edited by Anthony Levine 0125 Child Farm Labour: The Wealth Paradox by Sonia Bhalotra and Christopher Heady 0124 What Can Be Done about Child Labor? An Overview of Recent Research and Its Implications for Designing Programs to Reduce Child Labor by Bj0rne Grimsrud 0123 Measuring and Analyzing Child Labor: Methodological Issues by Bj0rne Grimsrud 0122 Family-Controlled Child Labor in Sub-Saharan Africa-A Survey of Research by Jens Christopher Andvig 0121 Is Child Work Necessary? by Sonia Bhalotra Social Protection Discussion Paper Series Titles continued No. Title 0120 The Cost and Benefits of Collective Bargaining: A Survey by Toke Aidt and Zafiris Tzannatos 0119 The Informal Sector Revisited: A Synthesis Across Space and Time by Niels-Hugo Blunch, Sudharshan Canagarajah and Dhushyanth Raju 0118 Social Services Delivery through Community-Based Projects by Dinah McLeod and Maurizia Tovo (available in Spanish) 0117 Earnings Inequality in Transition Economies of Central Europe Trends and Patterns During the 1990s by Jan J. Rutkowski 0116 Viewing Microinsurance as a Social Risk Management Instrument by Paul B. Siegel, Jeffrey Alwang and Sudharshan Canagarajah 0115 Vulnerability: A View from Different Disciplines by Jeffrey Alwang, Paul B. Siegel and Steen L. Jorgensen 0114 Individual Accounts as Social Insurance: A World Bank Perspective by Robert Holzmann and Robert Palacios 0113 Regulating Private Pension Funds' Structure, Performance and Investments: Cross-country Evidence by P.S. Srinivas, Edward Whitehouse and Juan Yermo 0112 The World Bank and the Provision of Assistance to Redundant Workers: Experience with Enterprise Restructuring and Future Directions by Yi Chen 0111 Labor Markets in Transition Economies: Recent Developments and Future Challenges by Mansoora Rashid and Jan Rutkowski 0110 Operating Instruction Included: A Review of Social Investment Fund Operations Manuals by Juliana Weissman 0109 Risk and Vulnerability: The Forward Looking Role of Social Protection in a Globalizing World by Robert Holzmann Social Protection Discussion Paper Series Titles continued No. Title 0108 Australia's Mandatory Retirement Saving Policy: A View from the New Millennium by Hazel Bateman and John Piggott 0107 Annuity Markets and Benefit Design in Multipillar Pension Schemes: Experience and Lessons from Four Latin American Countries by Robert Palacios and Rafael Rofman 0106 Guide for Task Teams on Procurement Procedures Used in Social Funds by Jorge A. Cavero Uriona 0105 Programmes Actifs Pour Le Marche Du Travail: Un Apercu General Des Evidences Resultant Des Evaluations by Zafiris Tzannatos and Amit Dar 0104 Kazakhstan: An Ambitious Pension Reform by Emily S. Andrews 0103 Long-term Consequences of an Innovative Redundancy-retraining Project: The Austrian Steel Foundation by Rudolf Winter-Ebmer 0102 Community Based Targeting Mechanisms for Social Safety Nets by Jonathan Conning and Michael Kevane (available in Spanish) 0101 Disability and Work in Poland by Tom Hoopengardner 0024 Do Market Wages Influence Child Labor and Child Schooling? by Jackline Wahba 0023 Including the Most Vulnerable: Social Funds and People with Disabilities by Pamela Dudzik and Dinah McLeod 0022 Promoting Good Local Governance through Social Funds and Decentralization by Andrew Parker and Rodrigo Serrano 0021 Creating Partnerships with Working Children and Youth by Per Miljeteig 0020 Contractual Savings or Stock Market Development. Which Leads? by Mario Catalan, Gregorio Impavido and Alberto R. Musalem Social Protection Discussion Paper Series Titles continued No. Title 0019 Pension Reform and Public Information in Poland by Agnieszka Chlon 0018 Worker Reallocation During Estonia's Transition to Market: How Efficient and How Equitable? by Milan Vodopivec 0017 How Poor are the Old? A Survey of Evidence from 44 Countries by Edward Whitehouse 0016 Administrative Charges for Funded Pensions: An International Comparison and Assessment by Edward Whitehouse 0015 The Pension System in Argentina: Six years after the Reform by Rafael Rofman 0014 Pension Systems in East Asia and the Pacific: Challenges and Opportunities by Robert Holzmann, Ian W. Mac Arthur and Yvonne Sin 0013 Survey of Disability Projects. The Experience of SHIA, Swedish International Aid for Solidarity and Humanity by Kaj Nordquist 0012 The Swedish Pension Reform Model: Framework and Issues by Edward Palmer 0011 Ratcheting Labor Standards: Regulation for continuous Improvement in the Global Workplace by Charles Sabel, Dara O'Rourke and Archon Fung 0010 Can Investments in Emerging Markets Help to Solve the Aging problem? by Robert Holzmann 0009 International Patterns of Pension Provision by Robert Palacios and Montserrat Pallares-Miralles 0008 Regulation of Withdrawals in Individual Account Systems by Jan Walliser 0007 Disability Issues, Trends and Recommendations for the World Bank by Robert L. Metts Social Protection Discussion Paper Series Titles continued No. Title 0006 Social Risk Management: A New Conceptual Framework for Social Protection and Beyond by Robert Holzmann and Steen J0rgensen 0005 Active Labor Market Programs: Policy Issues for East Asia by Gordon Betcherman, Amit Dar, Amy Luinstra, and Makoto Ogawa 0004 Pension Reform, Financial Literacy and Public Information: A Case Study of the United Kingdom by Edward Whitehouse 0003 Managing Public Pension Reserves Part 1. Evidence from the International Experience by Augusto Iglesias and Robert J. Palacios 0002 Extending Coverage in Multi-Pillar Pension Systems: Constraints and Hypotheses, Preliminary Evidence and Future Research Agenda by Robert Holzmann, Truman Packard and Jose Cuesta 0001 Contribution pour une Strategie de Protection Sociale au Benin by Maurizia Tovo and Regina Bendokat * The papers below (No. 9801-9818 and 9901-9934) are no longer being printed, but are available for download from our website at www.worldbank.org/sp 9934 Helping the Poor Manage Risk Better: The Role of Social Funds by Steen J0rgensen and Julie Van Domelen 9933 Coordinating Poverty Alleviation Programs with Regional and Local Governments: The Experience of the Chilean Social Fund - FOSIS by Jorge C. Barrientos 9932 Poverty and Disability: A Survey of the Literature by Ann Elwan 9931 Uncertainty About Children's Survival and Fertility: A Test Using Indian Microdata by Vincenzo Atella and Furio Camillo Rosati 9930 Beneficiary Assessment Manual for Social Funds by Lawrence F. Salmen Social Protection Discussion Paper Series Titles continued No. Title 9929 Improving the Regulation and Supervision of Pension Funds: Are There Lessons from the Banking Sector? by Roberto Rocha, Richard Hinz, and Joaquin Gutierrez 9928 Notional Accounts as a Pension Reform Strategy: An Evaluation By Richard Disney 9927 Reform Options for Pay-As-You-Go Public Pension Systems by Sheetal K. Chand and Albert Jaeger 9926 An Asset-Based Approach to Social Risk Management: A Conceptual Framework by Paul Siegel and Jeffrey Alwang 9925 Migration from the Russian North During the Transition Period by Timothy Heleniak 9924 Pension Plans and Retirement Incentives by Richard Disney and Edward Whitehouse 9923 Shaping Pension Reform in Poland: Security Through Diversity by Agnieszka Chlon, Marek G6ra and Michal Rutkowski 9922 Latvian Pension Reform by Louise Fox and Edward Palmer 9921 OECD Public Pension Programmes in Crisis: An Evaluation of the Reform Options by Richard Disney 9920 A Social Protection Strategy for Togo by Regina Bendokat and Maurizia Tovo 9919 The Pension System in Singapore by Mukul G. Asher 9918 Labor Markets and Poverty in Bulgaria by Jan J. Rutkowski 9917 Taking Stock of Pension Reforms Around the World by Anita M. Schwarz and Asli Demirguc-Kunt Social Protection Discussion Paper Series Titles continued No. Title 9916 Child Labor and Schooling in Africa: A Comparative Study by Sudharshan Canagarajah and Helena Skyt Nielsen 9915 Evaluating the Impact of Active Labor Programs: Results of Cross Country Studies in Europe and Central Asia by David H. Fretwell, Jacob Benus, and Christopher J. O'Leary 9914 Safety Nets in Transition Economies: Toward a Reform Strategy by Emily S. Andrews and Dena Ringold 9913 Public Service Employment: A Review of Programs in Selected OECD Countries and Transition Economies by Sandra Wilson and David Fretwell 9912 The Role of NPOs in Policies to Combat Social Exclusion by Christoph Badelt 9911 Unemployment and Unemployment Protection in Three Groups of Countries by Wayne Vroman 9910 The Tax Treatment of Funded Pensions by Edward Whitehouse 9909 Russia's Social Protection Malaise: Key Reform Priorities as a Response to the Present Crisis edited by Michal Rutkowski 9908 Causalities Between Social Capital and Social Funds by Jesper Kammersgaard 9907 Collecting and Transferring Pension Contributions by Rafael Rofman and Gustavo Demarco 9906 Optimal Unemployment Insurance: A Guide to the Literature by Edi Kami 9905 The Effects of Legislative Change on Female Labour Supply: Marriage and Divorce, Child and Spousal Support, Property Division and Pension Splitting by Antony Dnes Social Protection Discussion Paper Series Titles continued No. Title 9904 Social Protection as Social Risk Management: Conceptual Underpinnings for the Social Protection Sector Strategy Paper by Robert Holzmann and Steen Jorgensen (available in Russian) 9903 A Bundle of Joy or an Expensive Luxury: A Comparative Analysis of the Economic Environment for Family Formation in Western Europe by Pierella Paci 9902 World Bank Lending for Labor Markets: 1991 to 1998 by Amit Dar and Zafiris Tzannatos 9901 Active Labor Market Programs: A Review of the Evidence from Evaluations by Amit Dar and Zafiris Tzannatos 9818 Child Labor and School Enrollment in Thailand in the 1990s By Zafiris Tzannatos 9817 Supervising Mandatory Funded Pension Systems: Issues and Challenges by Gustavo Demarco and Rafael Rofman 9816 Getting an Earful: A Review of Beneficiary Assessments of Social Funds by Daniel Owen and Julie Van Domelen 98-44 This paper has been revised, see Discussion Paper No. 9923 9814 Family Allowances by Suzanne Roddis and Zafiris Tzannatos 9813 Unemployment Benefits by Zafiris Tzannatos and Suzanne Roddis 9812 The Role of Choice in the Transition to a Funded Pension System by Robert Palacios and Edward Whitehouse 9811 An Alternative Technical Education System: A Case Study of Mexico by Kye Woo Lee 9810 Pension Reform in Britain by Edward Whitehouse 9809 Financing the Transition to Multipillar by Robert Holzmann Social Protection Discussion Paper Series Titles continued No. Title 9808 Women and Labor Market Changes in the Global Economy: Growth Helps, Inequalities Hurt and Public Policy Matters by Zafiris Tzannatos 9807 The World Bank Approach to Pension Reform by Robert Holzmann 9806 Government Guarantees on Pension Fund Returns by George Pennacchi 9805 The Hungarian Pension System in Transition by Robert Palacios and Roberto Rocha 9804 Risks in Pensions and Annuities: Efficient Designs by Salvador Valdes-Prieto 9803 Building an Environment for Pension Reform in Developing Countries by Olivia S. Mitchell 9802 Export Processing Zones: A Review in Need of Update by Takayoshi Kusago and Zafiris Tzannatos 9801 World Bank Lending for Labor Markets: 1991 to 1996 by Amit Dar and Zafiris Tzannatos Summary Findings In response to shortages in public budgets for government health services, many developing countries around the world have adopted formal or informal systems of user fees for health care. In most countries user fee proceeds seldom represent more than 15 percent of total costs in hospitals and health centers, but they tend to account for a significant share of the resources required to pay for no -i-personnel costs. The problem with user fees is that the lack of provisions to confer partial or ful waivers to the poor often results in inequity in access to medical care. The dilemma, then, is how to make a much needed system of user fees compatible with the goal of preserving equitable access to services. Different countries have tried different approaches. Those which have carefully designed and implemented waiver systems (e.g., Thailand and Indonesia) have had much greater success in terms of benefits incidence than countries that have improvised such systems (Ghana, Kenya, Zimbabwe). Key to the success of a waiver system is its financing. Systems that compensate providers for the revenue forgone from granting exemptions (Thailand, Indonesia, and Cambodia) have been more successful than those who expect the provider to absorb the cost of exemptions (Kenya). Where waiver systems exist, performance will improve with the timeliness of the reimbursement. Other success factors include the widespread dissemination of information among potential beneficiaries about waiver availability and procedures; the awarding of financial support to poor patients for non-fee costs of care, such as food and transportation (as in Cambodia); and the existence of clear criteria for the granting of waivers, thereby reducing confusion and ambiguity among those responsible for managing the system and among potential recipients. Those facing the task of adopting a system of waivers face multiple design options. These include the following, among others: should exemptions be granted to whole groups or on the basis of individual targeting (the review finds that mcist systems are based on the latter)? Should waivers or exemptions be permanent or temporary? Hcw frequently should eligibility be reassessed? Should waiver eligibility be determined ex-ante, in the household, or when individuals seek care in the facility? The review examines various approaches taken by countries, but assessing their relative practical merits is difficult, as the evidence is scattered and mixed About this series... The World Bank Social Safety Nets Primer is intended to provide a practical resource for those engaged in the design and implementation of safety net programs around the world. Readers will find information on good practices for a variety of types of intervenitions, country contexts, themes and target groups, as well as current thinking of specialists and practitioners on the role of social safety nets in the broader development agenda. Primer papers are designed to reflect a high standard of quality as well as a degree of consensus among the World Bank safety nets team and general practitioners on good practice and policy. Primer topics are initially reviewed by a steering committee composed of both World Bank and outside specialists, and draft papers are subject to peer review for quality control. Yet the format of the series is flexible enough to reflect important developments in the field in a timely fashion. The primer series contributes to the teaching materials covered in the annual Social Safety Nets course offered in Waohirigton, DC, as well dS various other Bank-sponsored courses. The Social Safety Nets Primer and the annual course are jointly supported by the Social Protection unit of the Human Development Network and by the World Bank Institute. The World Bank Institute also offers customized regional courses through Distance Learning on a regular basis. For more information on the primer paper series and papers on other safety nets topics, please contact the Social Protection Advisory Servi e; telephone (202) 458-5267; fax (202) 614-0471; email: socialprotection@worldbank.org. Copies of related safety nets papers, including the Social Safety Nets Primer series, are available in electronic form at www.worldbank.org/safetynets. The website also contains translated versions of the papers as they become available An ambitious translation plan is underway (especially for Spanish and French, some in Russian). For more information about WBI courses on social safety nets, please visit the website www.worldbank.org/wbi/socialsafetynets.