advanced search

<< Back To Home

DOES FUNDING FROM DONORS DISPLACE GOVERNMENT SPENDING FOR HEALTH IN DEVELOPING COUNTRIES?

Friday, 17th of June 2011

'[A]id needs to be structured in a way that better aligns donors’ and recipient governments’ incentives, using innovative approaches such as performance-based aid financing. Differing donor and government priorities lead to Does Funding From Donors Displace Government Spending For Health In Developing Countries? 1. Marwa Farag, 2. A.K. Nandakumar, 3. Stanley S. Wallack, 4. Gary Gaumer and 5. Dominic Hodgkin + Author Affildonor funds partly replacing government resources, especially in poor countries.'


iations 1. Schneider Institutes for Health Policy, Heller School for Social Policy and Management, Brandeis University, in Waltham, Massachusetts 1. Marwa Farag (marwa@brandeis.edu)

Best viewed at http://content.healthaffairs.org/content/28/4/1045.long
 

Abstract The notable increases in funding from various donors for health over the past several years have made examining the effectiveness of aid all the more important. We examine the extent to which donor funding for health substitutes for—rather than complements—health financing by recipient governments. We find evidence of a strong substitution effect. The proportionate decrease in government spending associated with an increase in donor funding is largest in low-income countries. The results suggest that aid needs to be structured in a way that better aligns donors’ and recipient governments’ incentives, using innovative approaches such as performance-based aid financing. Differing donor and government priorities lead to donor funds partly replacing government resources, especially in poor countries.

DONOR FUNDING FOR HEALTH AND POPULATION nearly quadrupled over the period 1992–2006, reaching $13.7 billion annually in 2006.1 As the volume of health aid to developing countries has increased, so has concern about effectiveness. One of the issues influencing effectiveness is the “fungibility” of aid.2 Aid is considered fungible “when government offsets donor spending for a particular purpose by reducing its own expenditures on the same purpose and therefore aid substitutes rather than supplements local spending.”3 We examine the issue of fungibility at the health-sector level. How much aid money intended for health is displaced from this sector? Estimating the extent of aid fungibility has many policy implications. An important example is whether the mode of aid financing should differ depending on a recipient country’s level of economic development. Especially after the Paris declaration of 2005, in which both donors and recipient governments agreed to take steps toward increasing aid effectiveness, it is important to understand the implications of increased and consolidated donor funding on national health spending in countries receiving coordinated aid. Using panel data for 144 countries for 1995–2006, we examined empirically whether donor aid for health care has contributed to reductions in the government health care budgets of low- and middle-income countries. Previous Research There are few empirical studies of the extent of fungibility at the health-sector level, and the results are inconclusive. Through personal experience, case studies, or assessments of the work of others, a number of authors believe that donor aid will result in substitution for government funding. Governments of developing countries faced with competing demands on their budgets are believed to be willing and able to shift resources away from an activity for which donors are providing funding.4 The ability of governments to circumvent donors’ intentions by changing their own expenditure patterns is an accepted reality.5 Catriona Waddington similarly warns that it should not be assumed that earmarked donor funding increases total funding for targeted health programs.6 Dominique van de Walle and Ren Mu add that “while most economists assume that aid is fungible, most aid donors behave as if it is not.”7 These conjectures frame the empirical issue: if government priorities differ from the target priorities of the donor, then some shifting of government spending may occur, making the contribution of a dollar of donor funding less than one dollar. On the other hand, if the donor funding is aimed at a high priority of the recipient government’s public sector, then the apparent impact of a dollar of donor funding may exceed one dollar. When expenditure stimulus from external aid exceeds that from an equivalent increase in local funding, the phenomenon is referred to as a flypaper effect—that is, donor money “sticks where it hits.” The empirical issue of aid fungibility has been examined in the literature at multiple levels: (1) at the macro level (how do changes in total foreign aid for development affect total public expenditures?); (2) at the sector or meso level (how does donor aid for one sector(s) affect public expenditures for this sector?); and (3) at the micro (within-sector) level (how does aid for an intervention within one sector affect public spending for this intervention?).8 The methods used vary widely as well, from time-series studies of single countries or single sectors within countries to multiple country cross-sectional studies. Macro-level substitution. Research evidence on macro-level substitution of foreign aid is mixed. Country studies from the Dominican Republic and India show high levels of fungibility—that is, donor funding was diverted from the purpose intended by donors. Other individual-country studies, such as the one from Indonesia, show no evidence of fungibility.9 Aggregate-level fungibility. There is no consensus about the magnitude of fungibility in the aggregate-level cross-country studies, either. Using data from their main sample of fourteen countries (1971–1990), Tarhan Feyzioglu and colleagues found evidence that foreign aid was not fungible.10 On the other hand, Santanu Chatterjee and colleagues, using a panel of sixty-seven countries (1972–2000), found that about $0.70 of every dollar of foreign aid is fungible.11 National- and sector-level substitution. A study by Shantayanan Devarajan and colleagues found aid to be less fungible at national levels, but more fungible across sectors.12 Although total foreign aid to a developing country might not reduce total public spending, aid to a specific sector results in diverting public resources to other sectors of the economy. In an African multicountry study, fungibility of aid varied greatly from sector to sector. Health- and agriculture-related projects exhibited the most substitution against donor aid, while aid to education was the least fungible. The energy, transport, and communication sectors were in the middle.13 Sector-level fungibility. Research on fungibility at the health-sector level is scant and inconclusive. Authors of the oft-cited World Bank publication Assessing Aid argue that a country’s response to aid for health and education varies widely depending on the recipient government’s own priorities. They showed that although in Sri Lanka, 5.9 cents of an aid dollar went to education and health, the estimated effect was to reduce government spending on education and health by 1.9 cents. In contrast, in Indonesia, 7.2 cents of an aid dollar went to health and education, but health and education spending increased by 18.9 cents: aid money attracted funding from other sectors.14 A multicountry study by Pablo Gottret and George Schieber found that a $1.00 increase in off-budget donor funding for health was associated with a $1.65 decrease in government resources to health. They used this finding to recommend budget support as the preferred mode of aid financing.15 Subsector funding. At the subsector level, Jeremy Shiffman examined the effect of increased donor funding for HIV/AIDS on other health-sector development assistance and found that substitution probably occurred, but increases in overall health aid may have masked some of the effects.16 Summary. To summarize, there is no consensus in the literature on the magnitude of fungibility. There is some evidence of substitution (total spending goes up by less than the donor aid amount), and some evidence of the opposite result (total spending goes up by more than the amount of donor aid). We believe that this disparate evidence may result from using different data (countries and years) and research methods, and from variation in the match between donor and country priorities. Country-level studies by Howard Pack and Janet Rothenberg Pack show the extent to which fungibility varies by country.17 In light of the large increase in donor funding for health over the past decade, aid fungibility is of increased importance today. If the reduction in government spending attributable to aid is large, then the impact of donor funding on health care improvements is likely to be small. This is of greater concern for the poorest nations, for which donor funding is a sizable proportion of all health spending. Study Data And Methods Data source. This paper uses panel data from 144 countries for twelve years (1995–2006), constructed from the National Health Accounts (NHA) supplied by the World Health Organization (WHO).18 Primary data sources and estimation methods are explained in detail elsewhere.19 The NHA provides measures of health spending and the source of health funds. The quality of information varies considerably among countries.20 However, NHA estimates are the only available data for countries not in the Organization for Economic Cooperation and Development (OECD). Income classification of countries. We classified the countries into two groups according to their gross domestic product (GDP) per capita in international dollars. We assigned the lowest sixty-five countries to the low-income category and the remaining seventy-nine countries to the middle-income category. We conducted income-group assignments each year of the twelve-year period to allow for shifting between the low-income and middle-income categories according to changes in per capita income. (Only minor shifting actually took place.) To test whether or not our findings hold using a different country-classification method, we divided the sample into three groups—low-income, lower-middle-income, and upper-middle-income—using the World Bank country classification.21 Methods. We used a two-way, fixed-effect generalized least squares (GLS) model, which controls for both country- and year-specific effects. This is similar to the model estimated by Gottret and Schieber.22 All other studies of fungibility of health-specific aid used data from a single country. Model selection between random effect and fixed effects was based on the Hausman test. The dependent variable was the log of per capita government health spending, and the independent variables were the log of per capita donor funding for health and the log of GDP per capita.23 The model used all years (1995–2006) for each country. Given the log-log specification, the coefficient on donor funding estimated the elasticity: the percentage response of government health spending to a given percentage increase in donor funding. We expect a negative elasticity (donor funding will reduce government spending), with a smaller negative impact for wealthier countries. The two-way fixed-effect model controlled for unobserved country-level determinants of health spending and unobserved year-specific shocks or influences on health spending, so that any bias from time-invariant country-level factors was removed. However, there could still be bias from country-level factors that vary through time. The constant percentage effect implies different dollar impacts of donor health funding on government health spending at different income levels. We interpreted the results in dollar terms at the mean income of the sample. Study Results We present descriptive statistics for each group of countries. These statistics were not weighted for population, so each country (no matter how small or large) had the same influence on the country-group average. Government share of health spending. Exhibit 1⇓ shows the unweighted country-level summary statistics for the first, last, and mid-point years (1995, 2006, and 2000) of the twelve-year data set. Between 1995 and 2006, the average government share of total health spending fell. For low-income countries, the share fell from 24.3 percent in 1995 to 20.9 percent in 2006, and for middle-income countries it fell from 52.2 percent to 48.6 percent. The data also show that the percentage of GDP spent on health increased slightly in both low- and middle-income countries. View this table: • In this window • In a new window EXHIBIT 1 Summary Statistics: Per Capita Gross Domestic Product (GDP), Per Capita Health Spending, And Per Capita Government Health Spending, In 144 Countries, Unweighted, Selected Years 1995–2006 Donor funding for health. Per capita donor funding for health increased in low-income countries and slightly decreased or remained stable in middle-income countries. Exhibit 2⇓ shows that donor funding grew from 9.62 percent in 1995 to 19.79 percent in 2006 in low-income countries and remained at around 3.7 percent in middle-income countries. View this table: • In this window • In a new window EXHIBIT 2 Summary Statistics: Donor Funding For Health In 144 Developing Countries, Unweighted, Selected Years 1995–2006 Substitution of donor funding for government spending. The results show clear evidence of substitution of donor funding for government health spending: increases in one tend to be partly offset by decreases in the other. In low-income countries, a 1 percent increase in donor funding for health was associated with a 0.14 percent decrease in government spending on health, independent of changes in GDP per capita and country-specific and year-specific effects (Exhibit 3⇓). The effect was much smaller in middle-income countries, although still negative: a 1 percent increase in donor funding was associated with a 0.04 percent decrease in government funding for health.24 We expected a smaller effect in middle-income countries because government health spending in those countries is much larger in absolute terms and donor funding constitutes a much smaller share of the total. View this table: • In this window • In a new window EXHIBIT 3 Changes In Government Spending Associated With A 1 Percent Increase Or One-Dollar Increase In Donor Funding Per Capita, 1995–2006 Our statistical models estimating the fungibility of aid resources in low-income countries accounted for a relatively small amount of the overall variation in government health spending, which indicates that other factors influence that spending in these countries, such as aspects of governance. We found significant variation across countries. After considering different models, we ended by favoring the model including fixed effects for each country and for each year.25 Relationship of GDP to government health spending. As expected, GDP per capita is positively associated with government health spending. We estimated that for low- and middle-income countries, a 1 percent increase in income per capita results in increases of 0.66 percent and 0.96 percent, respectively, in government health spending. The results for the year-effect variables show that spending tended to increase over time, with the differences between one year and another sometimes being significant. (Results for the year and country fixed effects, not shown, are included in the analysis mainly to avoid bias in estimates of effect of donor spending.) Impact of donor financing on government health spending. In our analysis, the absolute impact of an increase in aid depends on the baseline levels of donor funding and government expenditures. With unweighted mean values of $24.50 per capita for government health expenditures and per capita donor funding at $12.70 in low-income countries, the results imply that a one-dollar increase in donor funding is associated with a reduction in government funding of twenty-seven cents (95 percent confidence interval [CI]: $0.15–$0.40). In middle-income countries, a one-dollar increase in donor funding is associated with a reduction in government funding of sixty-three cents (evaluated at the mean values of $222.60 per capita for government health expenditures and $13.90 per capita in donor funding). The dollar estimates for middle-income countries range widely, from thirty-two cents to ninety-six cents. However, we should be more concerned with low-income countries, because although the sixty-three-cent decrease in government spending is from a base of more than $200 per capita government health spending in middle-income countries, the twenty-seven-cent decrease in low-income countries is from a base of only $24 per capita, so the impact of the reduction is likely to be more serious in low-income countries. We further examined the tendency of fungibility to increase as GDP per capita declines, by dividing the data into three income groups (low, lower-middle, and upper-middle) using the World Bank classification. We found the same pattern: low-income countries had the highest fungibility (elasticity of 0.19), followed by lower-middle-income countries (elasticity of 0.09), and the upper-middle-income group had the lowest fungibility (elasticity of 0.027), as shown in Exhibit 3⇑. Discussion We found evidence for partial fungibility of aid funding in the health sector in both low- and middle-income countries. Other things being equal, additional donor funding is associated with reduced government spending on health. Recipient governments appear to differ with donors about financing priorities; donors seem to regard the health sector as more important than governments do. The significant difference in fungibility between low- and middle-income countries may be explained by the relative importance of donor funding. We found that countries with higher donor shares of total health spending have higher health-specific aid fungibility. The literature points to several other factors that influence fungibility. Rune Jansen Hagen argued that a certain donor’s influence increases with the proportion of the available funding it controls, which is particularly relevant for low-income countries, where donor funding constitutes a much larger share of the total health budget.26 Sajal Lahiri and Pascalis Raimondos-Møller also suggested that a donor can affect fungibility by choosing both the amount of aid and its timing.27 Long-term consequences of aid fungibility. The long-term consequences of aid fungibility need to be considered as well. Gottret and Schieber argue that fungibility coupled with volatility in donor funding is likely to have negative effects on the allocation of resources. As donor funding for a donor’s priority activity is reduced or stopped, it is difficult for the government to shift resources back to this activity after committing the funds for other purposes. Donor funding may end up disadvantaging donors’ priority activities in the long run.28 Impact of volatility in donor funding. For middle-income countries, the reduction in government spending relative to the total is not large; for low-income countries, it is much greater. Volatility in donor funding in lower-income countries has greater consequences, and donors need to be more aware of its potential negative impact. To the extent that donors reduce funding because the intended goal was met, targeted funding is less problematic. However, when a reduction of funding reflects a change in donors’ priorities, the health consequences for lower-income countries are greater. Impact of fungibility with volatility. There is very little empirical evidence on the combined effect of aid fungibility and volatility. Gottret and Schieber used evidence from Lesotho and Ethiopia to show that spending on hospitals exhibited “downward stickiness”—“resistance to decreasing below a certain level in response to declining revenues.” This was happening in a period during which donors focused on primary health care. It is very likely that this downward stickiness persists after donors’ attention has shifted away from primary health care in these countries simply because of the difficulty of reducing resources to the hospital sector after the funds had been committed. Gottret and Schieber argue that the negative impact on donors’ priority interventions as a result of the combined effect of fungibility with volatility makes off-budget support a less desirable approach to aid financing because it could result in disfavoring the very interventions in which donors are investing in the long run.29 After the Paris declaration of 2005, in which donor and recipient countries pledged to take steps to increase aid effectiveness, the International Health Partnership (IHP) was launched in September 2007 to respond to the Millennium Development Goals’ health-related challenges of focusing aid on priority problems and working in concert with countries and other donors to improve aid effectiveness.30 The IHP built on work already started through the Harmonization for Health in Africa (HHA) mechanism, launched in 2006, which proposed that donors should adopt interagency coordination and a common work plan. The substantial aid fungibility in the health sector requires the agencies responsible for donor coordination to pay special attention to what happens to the government share of health funding and to understand the influence of their efforts on national health systems. Budget support versus off-budget funding. Prior evidence on fungibility prompted some health policy researchers to suggest that budget support is superior to off-budget/project financing, on the assumption that recipient-country governments will divert fewer resources. However, there is no consensus on this issue. Elliot Berg argues that the “available evidence is too sparse and weak to support the argument that aid fungibility justifies a strong move away from the project mode of financing to budget support” and adds that the budget-support approach overlooks the conclusion in the World Bank’s Assessing Aid that shifting from projects to budget support will not work in countries with weak budget management.31 We estimate an elasticity of 0.084 for total health aid using data from all low-and middle-income countries, which is similar to the elasticity of 0.087 for off-budget support estimated by Gottret and Schieber.32 Because we did not explicitly examine the comparative fungibility associated with off-budget funding, we cannot interpret our results as indicating that fungibility is the same for off-budget and budget-support types of aid financing. However, our results raise the question of whether the mode of funding (budget versus off-budget support) does, in fact, matter in terms of fungibility of aid. Significant variation exists among low- and middle-income countries in aid fungibility, so the mode of financing must be studied, controlling for the level of government health spending and other factors that may interact with it. This discussion parallels the debate about central governments’ using block grants as opposed to specific or matched grants to support local governments. Economic theory and empirical evidence confirm that a local government will treat block-grant money as its own and apply its preferences in allocating the increased resources. A matching grant tilts the preferences of local government in the direction of the grant’s purpose because it must invest more of its own resources for the same purpose.33 Therefore, at least theoretically, there is no reason to expect that budget support for the recipient country’s health sector will result in less diversion of domestic health funding than off-budget (project financing) because neither mode of financing influences the country’s own spending preferences. The only difference is that off-budget/project financing combined with aid volatility is likely to undermine the target areas of aid financing in the long term. If donors want to minimize fungibility, they need to structure aid, using economic or political incentives, to influence recipient countries’ spending preferences to maintain or shift more local resources toward health. Aligning donors’ and recipient-governments’ preferences. There are several examples of aid-financing mechanisms that attempt to influence the preferences of recipient governments by providing incentives to align the preferences of donors and recipient governments such as the World Bank’s credit “buy-down” program for polio eradication, the Global Fund’s performance-based approach to grant making, and the GAVI Alliance’s performance-based payments for increasing immunizations.34 These methods raise implementation and technical challenges, particularly concerning the information and information technology needs of such methods and the implications of using these methods to fund particular health interventions on other parts of the health system. However, the level of fungibility associated with health aid and the potential implications of such fungibility point to the importance of using innovative health aid payment methods, which align the incentives of donors and recipient governments. Footnotes • Marwa Farag (marwa@brandeis.edu) is a research associate at the Schneider Institutes for Health Policy, Heller School for Social Policy and Management, Brandeis University, in Waltham, Massachusetts, and a research fellow at the Dubai Initiative, Kennedy School at Harvard University in Cambridge. A.K. Nandakumar is a senior program officer, Global Health Delivery, at the Bill and Melinda Gates Foundation in Seattle, Washington. Stanley Wallack is a professor at the Heller School and executive director of the Schneider Institutes at Brandeis. Gary Gaumer is an associate professor at Simmons College in Boston and a visiting associate professor at the Heller School. Dominic Hodgkin is an associate professor at the Heller School and the Schneider Institutes. • This analysis was funded by a grant from the Bill and Melinda Gates Foundation. The views expressed are those of the authors, and any errors or omissions are theirs. The authors thank Joanne Beswick for her support and assistance and Paroma Sanyal for her technical advice on the econometric methods used in this paper. NOTES 1. ↵ J. Kates et al., “Donor Funding for Health in Low- and Middle-Income Countries, 2001–2006” (Menlo Park, Calif.: Henry J. Kaiser Family Foundation, 2008) ; and J. Shiffman, “Has Donor Prioritization of HIV/AIDS Displaced Aid for Other Health Issues?” Health Policy and Planning 23, no. 2 (2007): 95–100. CrossRefMedline 2. ↵ World Bank, Assessing Aid: What Works, What Doesn’t, and Why (New York: Oxford University Press, 1998) ; and D. Walle and R. Mu, “Fungibility and the Flypaper Effect of Project Aid: Micro-Evidence for Vietnam,” Journal of Development Economics 84 (2007): 667–685. CrossRef 3. ↵ M. Foster and J. Leavy, “The Choice of Financial Aid Instruments,” Working Paper no. 158 (London: Overseas Development Institute, October 2001). 4. ↵ E. Berg, “Increasing the Effectiveness of Aid: A Critique of Some Current Views” (Paper prepared for Expert Group Meeting, Department of Economic and Social Affairs, United Nations, 24–25 January 2002). 5. ↵ P.T. Bauer, Dissent on Development (Cambridge, Mass.: Harvard University Press, 1972). 6. ↵ C. Waddington, “Does Earmarked Donor Funding Make It More or Less Likely that Developing Countries Will Allocate Their Resources Towards Programmes That Yield the Greatest Health Benefits?” Bulletin of the World Health Organization 82, no. 9 (2004): 703–706. Medline 7. ↵ Walle and Mu, “Fungibility and the Flypaper Effect.” 8. ↵ K. Jones, “Moving Money: Aid Fungibility in Africa,” SAIS Review 25, no. 2 (2005): 167–180. 9. ↵ H. Pack and J.R. Pack, “Foreign Aid and the Question of Fungibility,” Review of Economics and Statistics 75, no. 2 (1993): 258–265 CrossRef ; V. Swaroop et al., “Fiscal Effects of Foreign Aid in Federal System of Governance: The Case of India,” Journal of Public Economics 77, no. 3 (2000): 307–330 CrossRef ; and H. Pack and J.R. Pack, “Is Foreign Aid Fungible? The Case of Indonesia,” Economic Journal 100, no. 3 (1990): 188–194. CrossRef 10. ↵ T. Feyzioglu et al., “A Panel Data Analysis of the Fungibility of Foreign Aid,” World Bank Economic Review 12, no. 1 (1998): 29–58. Abstract/FREE Full Text 11. ↵ S. Chatterjee, P. Giuliano, and I. Kaya, “Where Has All the Money Gone? Foreign Aid and the Quest for Growth,” Discussion Paper no. 2858, June 2007, http://ftp.iza.org/dp2858.pdf (accessed 24 April 2009). 12. ↵ S. Devarajan et al., “What Does Aid to Africa Finance?” (Washington: African Economic Research Consortium and Overseas Development Council Project on Managing a Smooth Transition from Aid Dependence in Africa, 1998). 13. ↵ C. Lancaster, “Aid Effectiveness in Africa: The Unfinished Agenda,” Journal of African Economics 8, no. 4 (1999): 487–503 Abstract ; and J. Loxley and H.A. Sackey, “Aid Effectiveness in Africa,” Journal compilation from African Development Bank (Oxford, England, and Malden, Mass.: Blackwell Publishing, 2008). 14. ↵ World Bank, Assessing Aid. 15. ↵ P. Gottret and G. Schieber, “Health Financing Revisited—A Practitioner’s Guide” (Washington: World Bank, 2006). 16. ↵ Shiffman, “Has Donor Prioritization of HIV/AIDS Displaced Aid for Other Health Issues?” 17. ↵ Pack and Pack, “Is Foreign Aid Fungible?” ; and Pack and Pack, “Foreign Aid and the Question of Fungibility.” 18. ↵ The data set is composed of all 193 WHO member states. We excluded forty-nine high-income countries to form the core data set used for these analyses. WHO, National Health Accounts, http://www.who.int/nha/en (accessed 20 July 2008). 19. ↵ WHO, Guide to Producing National Health Accounts with Special Applications for Low-Income and Middle-Income Countries, 2003, http://www.who.int/nha/docs/English_PG.pdf (accessed 22 April 2009). 20. ↵ P. Musgrove, R. Zeramdini, and G. Carrin, “Basic Patterns in National Health Expenditure,” Bulletin of the World Health Organization 80, no. 2 (2002): 134–142. Medline 21. ↵ See Appendix 1, online at http://content.healthaffairs.org/cgi/content/full/28/4/1045/DC1. 22. ↵ Gottret and Schieber, “Health Financing Revisited.” 23. ↵ The data source permitted two different estimates of government health spending. We used both in the modeling work, and the results are quite similar. The first, which is the one we used, is total health spending (private health spending + donor spending). The second estimate is total health spending (out-of-pocket spending + donor spending). Results of these models are available from the author; send e-mail to marwa@brandeis.edu. 24. ↵ Both results are statistically highly significant despite the relatively wide confidence intervals shown in Exhibit 3. 25. ↵ Although the R2 was higher for the random-effects model (33 percent), we selected the fixed-effects model based on the Hausman test results. 26. ↵ R.J. Hagen, “Buying Influence: Aid Fungibility in a Strategic Perspective,” Review of Development Economics 10, no. 2 (2006): 267–284. CrossRef 27. ↵ S. Lahiri and P. Raimondos-Møller, “Donor Strategy under the Fungibility of Foreign Aid,” Economics and Politics 16, no. 2 (2004): 213–231. CrossRef 28. ↵ Gottret and Schieber, “Health Financing Revisited.” 29. ↵ Ibid. 30. ↵ Organization for Economic Cooperation and Development, “Paris Declaration on Aid Effectiveness: Ownership, Harmonisation, Alignment, Results, and Mutual Accountability,” 28 February–2 March 2005, http://www.oecd.org/dataoecd/11/41/34428351.pdf (accessed 22 April 2009). 31. ↵ World Bank, Assessing Aid ; and Berg, “Increasing the Effectiveness of Aid.” 32. ↵ Gottret and Schieber, “Health Financing Revisited.” 33. ↵ R.A. Musgrave and P.B. Musgrave, Public Finance in Theory and Practice (New York: McGraw-Hill, 1976). 34. ↵ J. Clift, “Health and Development: A Compilation of Articles from Finance and Development” (Washington: International Monetary Fund, 2004).