IMF comments on "Does the IMF Constrain Health Spending in Poor Countries"June 29, 2007
The views expressed in this paper are those of the staff and do not necessarily reflect the views of the Executive Board of the IMF.
The report of the Working Group on IMF Programs and Health Spending is well researched and thoughtful. IMF staff discussed many aspects of the report with David Goldsbrough and other members of the Working Group during the preparation stage. While we have a more nuanced view on some issues raised in the report, it is a significant input to enhance our understanding of the policy choices facing low-income countries (LICs). Indeed, one of the aims of the IMF's Medium-Term Strategy is to provide a framework for a more focused and effective Fund engagement in LICs to help these countries making progress towards the UN Millennium Development Goals (MDGs).
The report comes at an opportune time. On July 6, the IMF Executive Board will discuss a series of papers that address some of the very issues the report deals with: the appropriate design of Fund-supported programs and of fiscal policies in an environment of scaled-up aid. This discussion will help further clarify the Fund's position on these issues, thus responding to the report's call that the IMF be `crystal clear' about its role in low-income countries.
Positive aspects of Fund work in LICs are highlighted in the report. For example, it finds that the average increase in health spending as a share of GDP was larger for countries with Fund-supported programs than in LICs without such programs. Also, the report makes clear that IMF program conditionality has not included any wage-bill ceilings (or hiring freezes) specifically on the health sector. In addition, program design has not been `one-size-fits-all', but has been flexible to accommodate higher spending, depending on the circumstances.
The report of the Working Group rightfully emphasizes a number of critical issues. These include the importance of expenditure smoothing when designing medium-term fiscal programs, the role of wage-bill ceilings and the need to protect priority sector spending, as well as the importance of considering alternative scenarios in designing macroeconomic policies. The Fund has also been thinking about these issues for some time; this thinking has culminated in a set of papers which will shortly be discussed by the Fund's Executive Board.
There is merit in countries smoothing expenditures over a period of time while ensuring that all programs undertaken are adequately funded. This is critical because aid is volatile and uncertain. One important implication of the expenditure smoothing policy is that countries would respond symmetrically to aid increases and decreases. When aid is less-than-expected, they would draw down reserves. When it is above projections, or absorptive capacity prevents its full spending in the short-run, it would likewise be saved in the form of an increase in reserves to be spent in future years. This, of course, presumes, that reserves are above a certain minimum level to permit a reduction in case of an aid shortfall. This has not always been the case in the past in sub-Saharan countries but is increasingly feasible because countries have made progress in building up reserves. In this context, expenditure plans financed by scaled-up aid flows should take a medium-term perspective, with multi-year aid commitments by donors an essential element of expenditure smoothing and planning. Such a medium-term perspective would also provide an anchor for fiscal policy.
To ensure that critical programs are not starved of funds in face of aid volatility, priority spending would need to be safeguarded. However, the report fails to present evidence on whether or not health spending under Poverty Reduction and Growth Facility (PRGF) arrangements has actually been affected by fiscal consolidation.1 Moreover, PRGF-supported programs often allow for some accommodation of shortfalls in aid financing through drawing on other financing sources, and typically protect priority sectors from spending cuts.
Recent work within the Fund on wage-bill ceilings concludes that these were introduced as a short-term measure when wage-bill dynamics threatened macroeconomic stability and other first-best mechanisms to control wage spending were not available.2 The ceilings allowed additional hiring in the health sector if concessional external financing was available. However, they tended to persist and were not always efficient in achieving their objectives. Going forward, their use should be further limited to select cases and only for macroeconomic reasons, such as in post-conflict countries where the capacity of governments to control such spending is particularly weak. The latest data on wage bill ceilings indicates that we are moving in this direction, with the share of PRGF-supported programs with wage bill ceilings declining by 25 percent over the last 4 years. Wage bill ceilings were recently dropped from programs with Burkina Faso, Kenya, and Niger.3 Moreover, only 3 programs under the PRGF (in the Central African Republic, Chad, and Malawi)out of a total number of 29 PRGFs currently include them as quantitative performance criteria, the primary indicators for assessing program outcomes at the time of reviews; in the other PRGFs they are indicative targets.4
The report emphasizes the need to develop alternative scenarios to inform decision making on macroeconomic choices. As you know, scaling up scenarios have been prepared for a number of countries in SSA and more are under preparation-often at the request of the authorities, and in the context of the PRSP process. This work would be facilitated if costing of sectoral needs was available, a responsibility that lies beyond the Fund's area of expertise or mandate. It should be noted, however, that the momentum toward developing these scenarios has slowed down somewhat because the envisaged increase in aid flows after the Gleneagles Summit has not yet materialized. Also, in presenting alternative scenarios one cannot overlook the necessity of ensuring macroeconomic stability and debt sustainability. We agree with the report's conclusion that the increase in fiscal space that can be created with higher inflation is likely to be small or negligible. In this regard, the suggestion that there is scope for raising inflation in countries that have managed to bring inflation down to low levels risks rekindling inflationary expectation without any benefit. Also, inflation hurts the poorest segments in the population.. In this context, advocating scenarios of higher inflation could undermine the high levels of growth that sub-Saharan Africa has achieved since 2004 and work against the poor.
Other areas in the report would benefit from a more nuanced view that would better reflect the prevailing reality:
• A key objective of IMF-supported programs in LICs is to bring countries to a situation where they can effectively utilize all aid. But, in the short term, important macroeconomic questions may arise. Does the economy have the capacity to absorb scaled-up spending financed from abroad? Can the economy release complementary resources to support expanding health sector without causing pressures on other sectors? Are there any risks to the economy from a real exchange rate appreciation? What challenges arise from foreign inflows for management of monetary and fiscal policies? The answers to these questions are necessarily country-specific and policymakers need to continuously monitor the responsiveness of an economy to external inflows. We welcome the report's finding that the design of Fund-supported programs has been flexible, and increasingly so in the last few years, to accommodate higher spending depending on country-specific circumstances. However, a few clarifications are in order:
- The report could have acknowledged more the need for macroeconomic stability as a prerequisite for sustainable and poverty-reducing growth. This complementarity is well understood by the donor community, which values the Fund's macroeconomic assessments and, in many cases, uses them as an important input for disbursement decisions.
- The report's emphasis on the overall fiscal targets appears to largely ignore the improvement in the composition of spending that may have taken place within unchanged overall targets. There is a growing evidence that such spending is becoming increasingly propoor.
- The report seems to attribute caution in the design of macroeconomic programs solely to the conservative stance of Fund staff, without taking into account views of the authorities, which often argue against optimistic aid projections in order to prevent expenditure cuts in the future, or who may strongly favor low inflation.
- The report could have taken better account of the changed macroeconomic environment and the resulting fiscal space in many LICs after debt relief, as well as the corresponding evolution of Fund policy advice. In many LICs, increased fiscal space after the delivery of debt relief has provided scope for accommodating more ambitious spending plans for priority sectors. In response, and supported by an updated analytical toolbox, Fund staff has been actively engaged in assessing the feasibility of more ambitious fiscal paths without compromising macroeconomic stability. Examples of the latter are the new IMF framework for debt sustainability analysis in LICs which allows, on a case-by-case basis, for the consideration of less concessional financing depending on the debt sustainability outlook and the quality of investment; the papers on managing aid inflows that will shortly be discussed by the Executive Board will provide IMF staff with enhanced guidance on how to accommodate aid surges in program design; and country-level scaling-up studies which have become increasingly common.
- The report's assessment of spending and absorption of aid is based on estimates by the IMF's Internal Evaluation Office (IEO), which have serious shortcomings. 5
• We do not share the report's assessment that the proportion of total government spending devoted to health has not increased as much as it should have. The report recognizes that data on government health spending underestimate the level of resources for health. But available data only provide a partial picture of the resources flowing into the health sector. In particular, the data exclude private household spending (out of pocket expenditures), which is estimated at almost 50 percent for the world average, but can top 96 percent for low-income countries.6 In addition, a recent survey found that on average only 37 percent of external aid is channeled through national budgets.7 A large proportion of aid for health, such as under the PEPFAR, is channeled through the private sector. In some cases, such resources are much larger than budget allocations. In Rwanda, for example, government accounted for only 14 percent of spending in the health sector in 2006 while the share of NGOs was 55 percent. In addition, a significant share of the official aid flows for health is also not reported in national budgets. As a result, a key challenge facing governments is to align their priorities with those of various donors and private sector initiatives. And as more funding is channeled outside of the budget, governments seem to have responded by reallocating budget resources to other priorities, for example education, which is estimated to receive twice as much resources (on average 4.7 percent of GDP) than health (2.5 percent of GDP) in LIC government budgets. Reflecting this fungibility of budget resources, a study of the IMF's Fiscal Affairs Department found that an increase in aid by $100 results typically in an increase in budget health spending of only $2.8 Also, care should be exercised with spending targets, such as the commitment African leaders made at the 2001 Abuja Summit to spend 15 percent of their national budgets on health care (see also page 19 of the report). Similar norms have been proposed for other types of spending. 9 A proliferation of such targets would fragment the budget and constrain the flexibility of authorities to allocate resources toward their most optimal use.
• An exclusive focus on increasing health spending is not enough to achieve better health outcomes. Health outcomes are driven by a complex set of factors, in which the level of public health spending can play a significant but not always a decisive role. For example, better education outcomes, particularly education of women, have an important impact on health outcomes. Improvements in water and sanitation are also critical for better health outcomes. The report is largely silent on this issue.
• Achieving the MDGs for health will require more government spending but also a better targeting of the poor and strengthening of institutions. The report recognizes that the relationship between health spending and health outcomes is weak, but still emphasizes raising public health spending as the central issue in improving health outcomes. In fact, public health spending often does not reach the poor who most need it. Studies for the 1990s show that in countries of sub-Sahara Africa, the poorest quintile received on average only 13 percent of the benefits from government health spending while the richest received 29 percent.10 Other studies that have compared the level of government spending with outcomes have found large differences in countries' efficiency in the use of health resources, with countries that have better governance and fiscal institutions doing better. This brings us to a key issue, mentioned in the report, which is the weakness of public financial management (PFM) systems in many LICs. Unless PFM systems are up to the task, increasing health allocations in the budget is not likely to translate into better outcomes.
The report of the Working Group includes extensive and high-quality case studies for Mozambique, Rwanda, and Zambia. However, some aspects of the case studies could be nuanced.
• The statement that the medium-term projections for aid underlying the original program were too pessimistic is misleading (Box 3). The original program targeted a fiscal consolidation in the face of an expected decline in aid inflows after the aid surge and high inflation (2000-03) related to the flooding in 2000. In fact, aid inflows were lower in 2004-05 than in previous years and below what was projected. Fiscal consolidation limited recourse to monetary financing and helped reduce inflation to single-digit levels, relieving pressure on domestic interest rates. There was a substantial increase in credit to the private sector as a result in line with the authorities and program objectives. As a scaling-up of aid became apparent in 2006, aid projections were revised upwards (during the third and fourth review) and a number of illustrative scaling-up scenarios were considered for the 2007 Article IV consultations.
• In Mozambique, aid volatility (lack of predictability) stems from the donors' inability to commit on a firm basis and variability in disbursements. Aid projections are undertaken by the authorities based on firm commitments by donors. The Fund-supported program encourages a close collaboration between the authorities and donors but includes the projections made by the authorities. No discount of aid projections is made.
• Box 5 of the report inaccurately notes that no reason was given at the time of the fourth review when the wage bill ceiling was dropped. The staff report notes that "the indicative target on the wage bill will be discontinued as it was introduced in a post-conflict situation when there were concerns regarding a loss of fiscal control through a ballooning of the wage bill, which is not the case anymore. Moreover, the complexities of determining the optimal size of the public sector (e.g., hiring of teachers and nurses) to meet the MDGs calls for a more flexible approach but the Government will closely monitor the evolution of the wage bill to contain inflationary pressures and maintain fiscal sustainability."11
• The report could have placed more emphasis on the increased flexibility in Fund-supported programs since 2004. The report could also have noted that discussions on the donor-sponsored Poverty and Social Impact Analysis (PSIA) took place during the civil war in the Democratic Republic of Congo, when most donors were not inclined to step up financing. Moreover, though not successful in triggering a change in fiscal policies, the PSIA was very useful in generating agreement that an export promotion strategy was needed (now a key element of the authorities' medium-term strategy).
• The report should note that debt was unsustainable before the Heavily Indebted Poor Country (HIPC) Initiative completion point in April 2005. Even after full delivery of the assistance committed at the decision point, the net present value of Rwanda's external debt was estimated at 326 percent of exports at end-2003 (more than twice the HIPC threshold of 150 percent). Thus, it is surprising that the report states "....projections were ... avoiding borrowing even on concessional terms (Rwanda) without strong macroeconomic arguments in favor of the approach taken," (page 7, executive summary; also Box 2, paragraph 4).
• In its analysis of aid absorption and spending in IMF programs, the report states that "The IMF is right to take account of the level of reserves and domestic macro conditions when considering how additional aid should be used, but the degree to which these factors are influencing the allocation seems excessive." Drawing on the analysis in a Independent Evaluation Office report, the report of the Working Group suggests that this may have been the case in Zambia. What the report could have usefully noted is that Zambia's foreign exchange reserves stood at the very low level of 1.2 months of imports in 2004. It was thus important to achieve a modest build up of the country's reserves that could serve as a buffer to smoothen expenditure in the face of volatile aid flows that the report recommends. The buildup in reserves has been quite gradual, reaching only 2.2 months of imports at end-2006. Furthermore, the program strategy of containing domestic financing of the budget to support a decline in the high level of inflation was reflected in a close alignment between projected aid and the programmed overall deficit before grants. Aid was thus, on the whole, allowed to be spent and absorbed-which contrasts with the IEO's claim that with low reserves the Fund requires virtually all aid to be saved.
• The view that the fiscal strategy underlying Zambia's program in 2004 initially suffered from the lack of exploration of alternative aid and expenditure scenarios, is hard to support. It should be recalled that at the time, the immediate prospect was not one of scaling up of external assistance. Indeed, aid flows in 2003 had fallen by half compared with 2002. Once conditions changed, and a potential scaling up of aid became a possibility, the staff did (in 2005), as the report notes, explore alternative options for higher spending.
• As the report notes, the initial classification of "poverty" and "nonpoverty" spending was inadequate. However, it could also note that in collaboration with Zambia's development partners and the Fund, a more appropriate classification was developed.
• We agree that the need for wage ceilings in Zambia is no longer pressing, but the report's characterization of the rationale underlying the ceilings could be more nuanced. The report states that the rationale underlying wage ceilings has changed over time as the threat to macroeconomic stability had receded. Not all, including the Zambian authorities, would agree that the threat to macroeconomic stability had completely receded by 2005.
1 Financing under the IMF's PRGF is on concessional terms and targeted to LICs.
2 See Gerd Schwartz, Annalisa Fedelino, and Marijn Verhoeven, 2006, "Aid Scaling Up: Do Wage Bill Ceilings Stand in the Way?" IMF Working Paper No. 06/106, available at http://www.imf.org/external/pubs/cat/longres.cfm?sk=19144.0.
3 We have difficulty in agreeing with the conclusion derived from the discussion on the Mozambique and Zambia case studies that the wage-bill ceilings were harmful to health spending. The report does not present convincing evidence that "in practice, there was no way to enforce such protection for priority sectors or even monitor what actually happened" (page 50). See also the discussion of case studies below.
4 The non-observance of indicative targets under the PRGF arrangements does not automatically disrupt disbursements.
5 These were noted in the staff response to the IEO report on the IMF and Aid to Sub-Saharan Africa (available at http://www.ieo-imf.org/news/pr/pr0701.html), and include unreliable source data, an overly simplified model, and various specification problems. One particular shortcoming is that the IEO's static methodology ignores the multi-year context of program design.
6 World Development Indicators, 2005. World Bank (actual data for 2002).
7 Organization for Economic Cooperation and Development, 2006, "DAC Guidelines and Reference Series: Harmonizing Donor Practices for Effective Aid Delivery," Vol. 2 (Paris: OECD-DAC).
8 Mishra, Prachi and David Newhouse, 2007, "Health Aid and Infant Mortality," IMF Working Paper No. 07/100, available at http://www.imf.org/external/pubs/cat/longres.cfm?sk=20607.0.
9 For example, African heads of state have committed in the Maputo Declaration to raising agriculture spending to 10 percent of overall budget expenditure by 2008.
10 Hamid Davoodi, Erwin Tiongson, and Sawitree Asawanuchit, 2003 "How Useful are Benefit Incidence Analyses of Public Education and Health Spending," IMF Working Paper No. 03/227, available at http://www.imf.org/external/pubs/cat/longres.cfm?sk=16940.0.