Health care systems today account for about 9 percent of

Transcription

1 Health Care Financing And Delivery In Developing Countries Developing countries, which contain 84 percent of the world s population, claim only 11 percent of the world s health spending. by George Schieber and Akiko Maeda ABSTRACT: Developing countries account for 84 percent of world population and 93 percent of the worldwide burden of disease; however, they account for only 18 percent of global income and 11 percent of global health spending. Limited resources and administrative capacity coupled with strong underlying needs for services pose serious challenges to governments in the developing world. This paper analyzes health spending, health outcomes, and health delivery system characteristics for the six developing regions of the world as well as for low-, medium-, and high-income country groupings. Health care systems today account for about 9 percent of global production. While much of the debate on health spending and reform has focused on developed countries, these issues are of even greater importance to developing countries, whose significantly higher disease burdens, severe resource constraints, limited administrative capacity, and numerous competing critical priorities pose serious challenges. This DataWatch provides an overview of the health care financing and delivery systems in developing countries and compares the various developing regions of the world both among themselves and with the developed countries. It compares trends by low-, middle-, and high-income country groups as well as the six developing regions of the world: East Asia and the Pacific, Europe and Central Asia, Latin America and the Caribbean, Middle East and North Africa, South Asia, and Sub- Saharan Africa. 1 The analysis is based on information contained in the World Bank s Health, Nutrition, and Population database. 2 Global Income, Population, And Disease Burden Global spending on health totaled about $2.3 trillion in 1994, or about 9 percent of total global income (Exhibit 1). 3 High-income countries (those with per capita income above $8,500 in 1994) spent DATAWATCH 193 George Schieber is health sector leader and Akiko Maeda, health economist, in the World Bank s Middle East and North Africa Region ThePeople-to-People Health Foundation, Inc.

2 D a t a W a t c h EXHIBIT 1 Global Distribution Of Income, Health Spending, And Population, 1994 Distribution by income group High income Low and middle income Distribution among low- and middle-income countries East Asia and Pacific Europe and Central Asia Latin America and Caribbean Middle East and North Africa South Asia Sub-Saharan Africa SOURCE: The World Bank Health, Nutrition, and Population database. NOTE: Income groups are based on 1994 gross domestic product (GDP) per capita data: low income: $725 or below; middle income: $726 $8,500; and high income: above $8, % % % DEVELOPING COUNTRIES just over $2 trillion, amounting to 89 percent of total health expenditure, while their populations accounted for 16 percent of the global population. Developing countries, with 84 percent of the world s population, accounted for only 11 percent of all health spending. This disparity underscores the enormous difference between developed and developing countries in terms of capacities and types of health services that can be provided. This translates into large differences in health infrastructures and outcomes. The gap between rich and poor nations appears even more marked when the distribution of disease burden is included in the picture (Exhibit 2). Of the estimated 1.4 trillion disability-adjusted life years (DALYs) lost in 1990, industrialized countries accounted for just 7 percent, while developing countries accounted for 93 percent. 4 In industrialized countries 81 percent of DALYs are attributable to noncommunicable diseases, 12 percent to injuries, and only 7 percent to communicable diseases. In developing countries nearly half of DALYs result from communicable diseases (occurring mainly EXHIBIT 2 Global Disease Burden In Millions Of Disability-Adjusted Life Years (DALYs) Lost, 1990 Communicable diseases Noncommunicable diseases Injuries Total 605 (44%) 566 (41) 204 (15) 1,379 (100) 7 (7%) 80 (81) 12 (12) 99 (100) 602 (47%) 486 (38) 192 (15) 1,280 (100) SOURCE: Report of the Ad Hoc Committee on Health Research Relating to Future Intervention Options: Investing in Health Research and Development (Geneva: World Health Organization, 1996). a Organization for Economic Cooperation and Development (OECD) member countries, excluding Hungary, Mexico, and Turkey. b Developing countries are all countries outside of the established market economies. H E A L T H A F F A I R S ~ V o l u m e 1 8, N u m b e r 3

3 among children), about one-third, noncommunicable diseases, and the remainder, injuries. Developing countries in Europe and Central Asia, as an exception, exhibit a demographic and epidemiological profile similar to that of industrialized nations. 5 Over the next three decades developing countries are expected to undergo a major demographic and epidemiological transition, with significant increases in their injury and noncommunicable disease burden (that is, noncommunicable diseases and injuries will account for some 80 percent of the disease burden, compared with just under 50 percent today). This transition will alter the nature of these countries demand for health services and will increase pressure for new investments. Developing countries will need to focus on the prevention and treatment of noncommunicable diseases instead of the more cost-effective and lower-cost interventions targeted at communicable diseases. Per Capita Income, Outcomes, And Health Inputs Developing countries have much lower per capita incomes (gross domestic product, or GDP), much poorer health status indicators, and many fewer health-sector inputs than industrialized countries have (Exhibit 3). Per capita GDPs in developing countries are on average less than 10 percent of those in developed countries, where per capita GDPs averaged over $18,000 in Low-income developing countries have per capita GDPs that are only some 3 percent of those in developed countries. In terms of regional groupings, South Asia has the lowest (country-weighted) per capita GDP, while the Latin American/Caribbean region has the highest (Exhibit 3). Latin America/ Caribbean, Europe/Central Asia, Middle East/North Africa, and East Asia/Pacific are middle-income regions, while South Asia and Sub-Saharan Africa are low-income regions. Ten-year projections of growth in real per capita GDP paint a bleak picture for the Middle East/North Africa and for Sub-Saharan Africa, where annual growth on the order of 1 percent is projected. This is especially problematic for Africa, which has a very low income base. Given the well-known strong positive relationship between per capita GDP and per capita health spending, both the low absolute levels and the pessimistic future growth projections raise serious questions about many developing countries ability to provide their populations with access to basic health services. With respect to health outcomes, mortality under age five (the probability of a child s dying before reaching his or her fifth year) in developing countries is almost ten times the level found in developed countries. Under-five mortality in Sub-Saharan Africa, the lowest- DATAWATCH 195

4 D a t a W a t c h EXHIBIT 3 Key Economic And Health-Sector Indicators, By Region And Income Level, 1994 Region East Asia and Pacific Europe and Central Asia Latin America and Caribbean Middle East and North Africa South Asia Sub-Saharan Africa Income g Low Middle Low and middle High $ 1,214 1,792 3,138 2, ,707 1,774 18, % h h h h SOURCES: See below. a World Bank (Washington, D.C., 1998), country-weighted averages, exchange rate based dollars; in 1994 U.S. dollars. b Global Economic Prospects and the Developing Countries, 1998/1999 (Washington: World Bank, 1998). Real annualized gross domestic product (GDP) growth rates. c Probability of dying before age 5, based on current World Bank life tables. d Probability of dying between ages 15 and 60, based on current World Bank life tables. e Per 1,000 population. World Bank estimates, based on latest available data, f Per 1,000 population. World Development Indicators (Washington: World Bank, 1997). g Income groups are based on 1994 GDP per capita data. Low income: $725 or below; middle income: $726 $8,500; and high income: $8,500 and above. h Not available h h income region, is more than fifteen times higher than in developed countries. However, interestingly, under-five mortality is not strictly related to income among the middle-income regions, as shown by the higher levels in the Middle East and North Africa relative to East Asia and the Pacific, despite the higher income levels in the Middle East and North Africa. Worldwide there is an inverse relationship between per capita income and under-five mortality; however, a great deal of variability exists among countries, which indicates the strong influences of social and cultural factors as well as the differences in the availability and efficacy of targeted maternal and child health (MCH) programs. Differences in women s education is yet another important factor affecting these differentials. 6 Differences in adult mortality (probability of dying between ages fifteen and sixty) show similar patterns. For developing countries as a group, adult mortality is more than double that found in developed countries. Among regions, the poorest region (Sub-Saharan Africa) has an adult mortality rate more than four times the developedcountry average and two and one-half times that found in Latin H E A L T H A F F A I R S ~ V o l u m e 1 8, N u m b e r 3

5 America and the Caribbean. Adult mortality in Europe and Central Asia is well above that found in East Asia and the Pacific, much of this resulting from the transition from planned to market economies. This transition, accompanied by large declines in GDP and the breakdown of social systems, has manifested itself in large increases in the incidence of cardiovascular disease and accidents, violence, and suicides as well as the reemergence of virtually eradicated infectious diseases such as tuberculosis. 7 Similarly, the high under-five mortality rates in the Middle East and North Africa, well above those found in the much poorer East Asia/Pacific region, raise questions concerning both the effectiveness of MCH programs as well as the low educational attainments among girls in the region. 8 As to health-sector inputs, developed countries have six times as many hospital beds and three times as many physicians per capita as developing countries have. There are also large regional differences, with Sub-Saharan Africa having only 4 percent as many physicians as developed countries and South Asia having only one-twelfth the number of hospital beds per capita. These figures reflect many factors, including income, population sizes and age distributions, and differences in underlying epidemiology. On the other hand, the number of beds and physicians per capita in Europe and Central Asia far exceeds that in developed countries. These large quantities of inputs are often of poor quality and reflect excessive use of services and inappropriate clinical interventions that raise serious questions about these systems efficiency and effectiveness. 9 Clearly, quality, service intensity, and efficiency are important dimensions of health-sector inputs for which valid international comparative data are almost always lacking, even for developed countries. Closing the gaps between developed and developing countries will be a major challenge for many developing countries in light of pessimistic income-growth projections. The Middle East/North Africa and Sub-Saharan Africa regions, in particular, are likely to face resource constraints in expanding health services. The recent economic slowdown in the East Asia/Pacific region also will curtail growth in health spending and could eradicate previous gains. Health Spending Patterns Here we compare per capita health spending and per capita GDP in U.S. dollars based on both exchange rates and purchasing power parities (PPPs). 10 Health-to-GDP ratios and the public shares of total health spending are also described and analyzed. Finally, we discuss the revenue-raising capacity of developing versus developed countries. Obtaining health spending information for more than 200 countries is problematic. No data source comparable to the Organization DATAWATCH 197

6 D a t a W a t c h 198 DEVELOPING COUNTRIES for Economic Cooperation and Development s (OECD s) ECO- SANTÉ database exists for developing countries, and most do not collect information on private health spending. 11 The data presented here reflect an effort by the World Bank to develop health spending information for some 200 countries. The effort relied on using existing data sources, not on information derived from using a standardized national health accounts data collection instrument for each country. As a result, the data should be interpreted with caution. 12 Exhibit 4 presents per capita health spending in both exchange rate and PPP-based U.S. dollars, health-to-gdp ratio, and the public share of total health spending for the six regional and three incomeclass groupings. Average per capita health spending in exchanged rate based U.S. dollars varies greatly by country income class. OECD countries spend more than 100 times the amounts spent in low-income countries and ten times that spent in middle-income countries. Among regions, South Asia spends the least. Although systematic data on the types of services purchased are lacking, it is obvious that providing even basic public health services is a challenge in many low- and some middle-income countries. Data from certain countries indicate that disproportionate shares of limited public health spending are being allocated to tertiary care, while primary care services are inefficiently provided and of poor quality. 13 Nevertheless, it appears that if countries did spend rationally and EXHIBIT 4 Health Spending, By Region And Income Level (Country-Weighted Averages), 1994 Region a East Asia and Pacific Europe and Central Asia Latin America and Caribbean Middle East and North Africa South Asia Sub-Saharan Africa $ $ % Established market economies b 1,827 1, Income c Low Middle High , ,521 SOURCE: World Bank Health, Nutrition, and Population database. NOTES: US$ is U.S. dollars, exchange rate based. PPP$ is purchasing power parity dollars. a Low- and middle-income countries. b Organization for Economic Cooperation and Development (OECD) member countries, excluding Hungary, Mexico, and Turkey. c Income groups are based on 1994 gross domestic product (GDP) per capita data. Low income: $725 or below; middle income: $726-$8,500; and high income: above $8, % H E A L T H A F F A I R S ~ V o l u m e 1 8, N u m b e r 3

7 efficiently, they could afford a cost-effective package of basic primary care and emergency services. 14 As expected, PPP-based comparisons, which adjust for price differences across countries, narrow these disparities. The spending gap between low-income and OECD countries narrows from 100 to one to twenty-five to one, while the middle-income country gap narrows to five to one. Regional differences also narrow. In PPPbased terms, South Asia is the only region with per capita health spending below $100, and expenditures in the Europe/Central Asia, Latin America/Caribbean, and Middle East/North Africa regions exceed $300, about one-sixth of the OECD levels. There is a strong direct relationship worldwide between health spending and GDP per capita (Exhibit 5). The global income elasticity of per capita health spending relative to per capita GDP is estimated at This signifies that for every 10 percent difference in per capita income (GDP), there is an 11.3 percent difference in per capita health spending (that is, 10 percent higher per capita GDP is associated with 13 percent higher health spending). The income elasticity for the public component of health spending is estimated at 1.21, greater than the estimated private health expenditure elasticity of This suggests that public health spending is more responsive to income differences than private health spending is. This is consistent with higher-income countries having larger public shares of total health expenditures (Exhibit 6). We also estimated income elasticities for countries grouped according to income levels. Income elasticities for total per capita health spending relative to per capita GDP are the highest for highincome countries (1.47) followed by middle-income (1.19) and low- DATAWATCH 199

8 D a t a W a t c h 200 DEVELOPING COUNTRIES income (1.00) countries. This also is consistent with higher-income countries devoting a relatively larger share of their income to the health sector. The foregoing analyses are based on simple descriptive relationships, which clearly need to be more fully explored using more sophisticated behavioral models. Low-income countries devote some 4 percent of their GDP to the health sector, compared with more than 8 percent in OECD countries (Exhibit 4). East Asia and the Pacific, South Asia, and Sub- Saharan Africa have health-to-gdp ratios in the 4 percent range, while the Middle East/North Africa, Latin America/Caribbean, and Europe/Central Asia regions spend 5 7 percent. There is a positive relationship between health-to-gdp ratios and per capita GDP (higher-income countries devote absolutely and relatively more resources to their health sectors). However, there is a great deal more variability in this relationship (as shown in Exhibit 7) than in the relationship between per capita GDP and per capita health spending shown in Exhibit For example, whereas Europe/Central Asia countries have the third-highest income level among developing regions, they spend more than 7 percent of their GDP on health, the highest among all of the regional developing country groups. As countries incomes increase, a larger share of total health spending derives from public sources (Exhibit 6). This appears to reflect the relatively greater revenue-raising capacity of higherincome countries coupled with governments choosing to counteract health-sector and health insurance market failures through pub- H E A L T H A F F A I R S ~ V o l u m e 1 8, N u m b e r 3

9 lic rather than private financing. 16 The public share of total health spending is some 50 percent in low-income countries, compared with 70 percent in high-income and OECD countries. The trend is clearly upward, although there is considerable variability at any given income level. These observations have two important policy ramifications. First, as countries incomes increase, larger shares of overall health spending are publicly financed. The underlying reasons behind this trend and whether the higher public share is justified with respect to the performance of the health care systems are important empirical questions for future research. Second, since half of health spending in developing countries is privately financed, policymakers must focus on making effective use of both public and the often neglected private sources of funding. The large observed regional differences in public spending shares reflect the wide variability around the upward trend in public shares relative to per capita GDP shown in Exhibits 4 and The Europe/ Central Asia region still shows strong vestiges of the completely state-controlled systems of the former Soviet era, having the highest public share of all regions. Similarly, Sub-Saharan Africa s public share is higher than those in the Middle East/North Africa and Latin America/Caribbean, two regions with much higher incomes. Thus, as in the case of the health-to-gdp ratios, public shares display an upward relation with per capita income, although there is a great

10 D a t a W a t c h 202 DEVELOPING COUNTRIES deal of variance among the countries of the world. As discussed above, health spending and public shares are directly related to countries per capita incomes. As incomes increase, total spending and public spending also increase, with public spending being more responsive to income changes than private spending is. This greater responsiveness may be due in part to governments increased ability to raise revenues as incomes rise. There are many reasons for this, including population shifts to urban areas; growth in the formal employment sector; growth of large, easier-totax enterprises; improved tax administration capacity; and so on. 18 Data on government revenues as a percentage of GDP for low-, middle-, and high-income countries bear this out. Low-income countries can only raise revenues equivalent to 20 percent of their GDP, less than half of the 42 percent figure in high-income countries and one-third less than the 31 percent figure in middle-income countries. 19 This has very important implications for health-sector priorities. For example, a country with a per capita GDP of $300 can only raise $60 per person in government revenues. These revenues must be used for all government functions, including security, roads, airports, seaports, education, Social Security, public enterprises, and so on. If a basic package of primary care and preventive services were to cost somewhere around $15 $20, then a low-income country must devote one-quarter to one-third of its government budget to the health sector just for these basic services. 20 Low-income countries cannot devote such large shares of their budgets to health. In short, there are real limits on what low-income countries can afford, and difficult choices need to be made. Moreover, the fact that half of all health spending is privately financed has important implications for equity as well as efficiency, since governments should focus on efficient allocation of combined public and private resources. Basic Issues In Health Care Reform In pursuing health-sector reform, developing countries face many of the same problems regarding access, efficiency, and quality that developed countries face, although these problems are much more severe. They also face greater challenges in providing basic public health services, especially in rural areas. Moreover, intersectoral linkages to education, water, food, housing, energy, and other relevant sectors are critical. Given their limited incomes, revenueraising capabilities, and administrative capacity, developing countries as a group face serious constraints in financing basic health services, providing physical access to care, and regulating both public- and private-sector entities. Here we review these issues for income subgroups of developing countries. 21 H E A L T H A F F A I R S ~ V o l u m e 1 8, N u m b e r 3

11 Low-income group. Government budgets finance most public health expenditures in this group, but these are limited by narrow tax bases and inefficient collection capacity of the government. They are sometimes supplemented by user fees, but these constitute a very small percentage of public revenues. External assistance continues to be a significant revenue source, especially in Sub-Saharan Africa, South Asia, and the island economies of the Pacific. Apart from the government budget, options for formal insurance schemes (public or private) are extremely limited, as a result of (1) the small size of the formal employment sector; (2) limited savings and underdeveloped financial sectors; and (3) the weak institutional environment to support such schemes. Instead, households rely on informal arrangements (such as extended families, traditional community support systems, nongovernmental organizations, or NGOs, charitable organizations, or rural cooperatives) to provide protection in the event of catastrophic illnesses. 22 Key issues include innovative use of informal risk-pooling mechanisms, which are likely to be important in expanding coverage to the poor, and more effective use of external assistance to supplement public financing. 23 Middle-income group. Countries in this group show a tendency toward the expansion of multiple financing sources, including the growth of Social Security schemes for civil servants and other categories of formally employed workers. Although some countries show preferences for expanding coverage through Social Security systems, others show movement toward a national health service model that relies on general revenues. There usually is growth in private spending with the expansion of modern private health services; private insurance begins to expand, although it and privatesector providers remain largely unregulated. Expansion in formal employment, capital markets, and financial sectors; improved institutional contexts for formal insurance; plus urbanization and other social changes also help households and enterprises to make the transition from informal to formal risk pooling. However, countries in this group (with the exception of most of the Europe and Central Asia countries) still generally fall short of universal coverage. Urban/rural disparities and persistent poverty contribute to these problems, and the multiplicity of financing sources, often with overlapping or inconsistent policies, contributes to the inefficiencies and inequities in the system. The key issues are development of actuarially sound social insurance schemes or other forms of riskpooling arrangements, expanding insurance coverage to rural or informal sectors, and regulation of the private sector. High-income group. With the exception of the United States, all of the industrialized countries have achieved universal coverage, DATAWATCH 203

12 D a t a W a t c h largely through public financing that is either publicly managed or publicly mandated. Again, with the exception of the United States, private insurance is used mainly to supplement the core services covered by public financing. Korea and a number of newly industrialized countries also have attained, or are close to attaining, universal coverage. Key issues are cost containment, efficiency, quality of service, aging populations, patient choice, and patient satisfaction. Research on global healt h spending is still in its infancy. A major obstacle continues to be the lack of reliable and consistent health spending data on which to base international comparisons. International efforts at establishing consistent national health accounts will provide valuable tools for policymakers in developing countries to evaluate the effectiveness of their health care systems. It also will open up wide areas for future research that addresses some of the broader questions about the evolution of health care systems that emerge as countries develop socially and economically. 204 DEVELOPING COUNTRIES The views expressed are those of the authors and do not necessarily reflect those of the World Bank. An earlier version of this paper is contained in G. Schieber, ed., Innovations in Health Care Financing (Washington: World Bank, 1997). NOTES 1. Low-income countries are defined as those with per capita gross domestic product (GDP) below $725; middle-income countries have per capita incomes of $726 $8,500; while high-income countries have per capita incomes exceeding $8,500. Income groups are based on 1994 GDP data. Developing countries are defined as those in the low- and middle-income groups. For a listing of countries by region, contact the authors at the World Bank, 1818 H Street, NW, Washington, DC Sector Strategy Health, Nutrition, and Population (Washington: World Bank, 1997); World Development Indicators (Washington: World Bank, 1998); World Bank, World Development Report 1998/99: Knowledge for Development (New York: Oxford University Press, 1998); and World Bank, World Development Report 1993: Investing in Health (New York: Oxford University Press, 1993). 3. This 9 percent figure reflects global health spending divided by global GDP. It is not a country-weighted figure. It obviously is heavily influenced by the high percentage of worldwide health spending and GDP in the United States. 4. The DALY is an indicator of the time lived with a disability and the time lost as a result of premature mortality. Years lost from premature mortality are estimated with respect to a standard expectation of life at each age. Years lived with a disability are translated into an equivalent time loss through multiplication by a set of weights that reflect reduction in functional capacity. As such, the DALY represents an attempt to combine in one indicator the impact of disease on mortality (through a calculation of the duration of life lost due to premature death) and morbidity (through an assessment of time lived with a disability). See World Bank, World Development Report For the analysis of DALY burdens among countries, see Report of the Ad Hoc Committee on Health Research Relating to Future Intervention Options: Investing in Health Research and Devel- H E A L T H A F F A I R S ~ V o l u m e 1 8, N u m b e r 3

13 opment (Geneva: World Health Organization, 1996). 5. Report of the Ad Hoc Committee on Health Research. 6. Sector Strategy Health, Nutrition, and Population. 7. J. Klugman and G. Schieber, Health Reform in Russia and Central Asia, in National Research Council, Transforming Post-Communist Political Economies: Task Force on Economies in Transition, Commission on Behavioral and Social Sciences and Education (Washington: National Academy Press, 1997), A. Maeda et al., Health, Nutrition, and Population in Middle East and North Africa Region (Washington: World Bank, forthcoming). 9. Klugman and Schieber, Health Reform in Russia and Central Asia. 10. Comparisons based on purchasing power parities (PPPs) provide a somewhat better international comparative metric, since PPPs adjust for price-level differences among countries. However, PPPs are not available at all for many developing countries, and in other countries they are estimated using regression analysis. Thus, while we present some of the data in both PPP and exchange rate based U.S. dollars, we use exchange rate based per capita GDP and health spending in the more detailed analyses since there are more observations and less reliance on imputations. 11. OECD Health Data 98: A Comparative Analysis of Twenty-nine Countries (Paris: Organization for Economic Cooperation and Development, 1998). 12. G. Schieber and A. Maeda, A Curmudgeon s Guide to Financing Health Care in Developing Countries, in Innovations in Health Care Financing, ed. G. Schieber (Washington: World Bank, 1997), 1 40; Sector Strategy Health, Nutrition, and Population; and E. Bos et al., Health, Nutrition, and Population Indicators: A Statistica l Handbook (Washington: World Bank, 1999). 13. Better Health in Africa: Experience and Lessons Learned (Washington: World Bank, 1994). 14. World Bank, World Development Report Per capita GDP accounts for 95 percent of the variation in per capita health spending but only about 22 percent of the variation in the health-to-gdp ratio. This is obvious visually in Exhibits 5 7 from the much larger dispersion of countries around the trend line in the case of the health-to-gdp ratio relationship. Nevertheless, the trend is upward and statistically significant in both cases. 16. Schieber and Maeda, A Curmudgeon s Guide to Financing Health Care; and P. Musgrove, Public and Private Roles in Health, World Bank Discussion Paper 339 (Washington: World Bank, 1996). 17. Compared with the relationships found between per capita GDP and both per capita health spending and the health-to-gdp ratio, there is even more variability in the public share relationship. Although the relationship is upward and statistically significant, there is even greater dispersion among countries in the public share as evidenced from both Exhibit 6 and the fact that only 12 percent of the variability is accounted for. 18. Schieber and Maeda, A Curmudgeon s Guide to Financing Health Care, Ibid., World Bank, World Development Report 1993, See Schieber and Maeda, A Curmudgeon s Guide to Financing Health Care, 31 36, for a regional discussion of basic health care reform issues. 22. In the field of international development, NGOs are nonprofit private institutions, usually engaged in the provision of social services. 23. See A. Creese and S. Bennett, Rural Risk-Sharing Strategies, in Innovations in Health Care Financing. DATAWATCH 205