Chapter 2 Country Policies and Programs
- International Monetary Fund. Independent Evaluation Office
- Published Date:
- April 2007
This chapter reports on the evaluation’s findings about aid-related issues in the design of PRGF-supported programs. It covers (1) the links between aid and current account and fiscal adjustment in PRGFs; (2) PRGFs’ analysis of aid; and (3) the PRGF’s pro-poor and pro-growth agenda.1 The chapter’s focus is on program design—both for the initial PRGF program period and for subsequent program periods following reviews—as it is at the design stage that Fund staff’s inputs and contributions are most clearly seen.
Figure 2.1 provides an overview of developments in SSA on aid, macroeconomic indicators, growth, and poverty reduction. Panel A summarizes recent aid trends. As illustrated, official development assistance (ODA) to the 29 SSA countries under study declined during the ESAF period, bottomed out in 1999, and recovered during the PRGF period. These developments reflect the changing aid environment for SSA since the adoption of the MDGs and the improving performance of many SSA countries, a factor in donor aid plans. Panel A also shows trends in debt relief grants, which surged starting in 2002.
Figure 2.1.Trends in Aid, Policies, and Outcomes in Sub-African Africa Figure 2.1.(concluded)
Three factors in the changing aid profile are worth noting. First, the ESAF period’s aid downswing affected almost all SSA countries, while the PRGF period’s upswing has mainly affected two groups of countries—post-conflict countries and good-performing countries. Second, aid volatility has remained high throughout the period2 (see panel B). Third, the aid shown in panel A includes grants and concessional loans with a grant element of at least 35 percent. Under Fund guidelines, all PRGFs strictly limit—and often totally preclude—government contracting or guaranteeing of nonconcessional foreign debt, with specific limits placed on the minimum degree of concessionality.3
Figure 2.1 also illustrates the improving macroeconomic policies and outcomes in the 29 SSA PRGF countries. As shown in panels C and D, both the government deficit and inflation have dropped sharply since the mid-1990s. Growth in per capita income, while still low, has become much more consistently positive, and per capita incomes have begun to recover from their lows of the mid-1990s. Going forward, enormous challenges clearly remain, especially in terms of poverty reduction, which has not yet seen a significant reduction of the proportion of people living on less than $1 a day.
Accommodation of Aid
This section reports on the evaluation’s findings on the design of PRGF-supported programs as a basis for addressing critics’ concerns that the IMF “blocks” or prevents the full use of available donor funding. To this end, the evaluation looked at how changes in the aid forecast mapped into changes in programmed levels of the fiscal and current account deficits.4 In the parlance of the IMF’s 2005 “spend and absorb” framework (see Box 2.1), this section of the report asks: (1) how much of increased aid was programmed to be absorbed (in higher net imports); and (2) how much of increased aid was programmed to be spent (in higher net public expenditures)? It also examines (3) how PRGFs analyzed aid absorptive capacity and (4) PRGF program “adjusters” to see whether and how much of aid surprises could be spent and absorbed.
Current account adjustment
The evaluation’s empirical analysis finds that country conditions, as proxied by the level of international reserves, are the main determinants of whether and to what extent PRGFs permit the absorption of incremental aid. It also finds that on average SSA PRGFs do not call for current account adjustment during the first program year. This represents a departure from SSA ESAFs, which typically called for significant current account adjustment in the initial program year. The evidence points to increased expectations regarding aid inflows for the initial program year as well as improved reserve levels as reasons for this shift in program stance. Abstracting from these two determinants of program design, there is no evidence of an independent shift over time in program design with respect to the programmed absorption of increased aid.
On average, across time and countries, SSA PRGFs programmed an immediate absorption of 63 percent of anticipated aid increases. As illustrated in Figure 2.2, anticipated aid increases in SSA PRGFs are on average correlated with a widening of the current account5 of 63 percent of the anticipated increase. The remaining 37 percent is programmed to increase international reserves.6 This relationship is significantly affected by the initial stock of reserves measured in months of imports.
Figure 2.2.Programmed Absorption of Aid Increases
Average programmed current account response to increases in aid in SSA PRGFs
(In percent of anticipated aid increase)
Note: The cutoff for low and high stocks of reserves is 2.5 months of imports. See Annex 2 for definitions, methodology, and a discussion of robustness.
Underpinning the average rate of programmed absorption out of incremental aid are country differences in net international reserves. As shown in Figure 2.2, for countries with reserves below a threshold of 2.5 months of imports, absorption of incremental aid is nearly zero—as those countries are programmed to build their reserve position. This result is consistent with the evidence from the evaluation’s desk reviews, which found programmed increases in international reserves—in cases where initial reserve positions are low—identified as a way to manage vulnerabilities to external shocks associated with variations in the terms of trade or aid volatility (see Annex 3). For countries with reserve levels above the threshold, programmed absorption averages 100 percent of incremental aid.7 Case study analysis indicates that debt-sustainability concerns may be an additional factor reducing the programmed level of absorption—and increasing the programmed buildup of reserves—in response to an increase in aid8 (see Annex 4).
The empirical analysis finds that country macroeconomic conditions, as proxied by the inflation rate, are the main determinants of whether and to what extent PRGFs permit the spending of incremental aid.9 It also finds that the sustained decline in SSA inflation rates—coupled with the recovery (and more) of donor inflows—has reduced the average fiscal correction in the first program year of new programs by about 1 percentage point of GDP relative to the ESAF period. The results of staff interviews and desk reviews are consistent with these findings.
On average—that is across all countries experiencing aid increases during the PRGF period—SSA PRGFs programmed immediate spending of about 30 percent of anticipated aid increases. Figure 2.3 illustrates how much of the programmed full absorption that can be observed in SSA countries with sufficient reserve stocks is translated into fiscal expansion (spending). On average, anticipated aid increases in SSA PRGFs are correlated with a widening of the fiscal deficit10 amounting to 28 percent of the anticipated increase. The remaining 72 percent is programmed as public savings, often through the retirement of domestic public debt.
Figure 2.3.Programmed Spending of Aid Increases
Average programmed use of anticipated aid increases in SSA PRGFs with high stocks of reserves
(In percent of anticipated aid increase)
Note: See Annex 2 for definitions, methodology, and a discussion of robustness.
Within the average, the evidence points to inflation concerns as a major driver of cross-country differences in programmed spending of incremental aid.11 As shown in Figure 2.3, differences in programmed spending levels of anticipated aid increases are highly correlated with initial inflation levels. According to the estimation, countries with inflation rates below 5 percent12 get to spend 79 percent of anticipated aid increases, on average, whereas countries with higher inflation get to spend only 15 percent of such increases, on average.13 Consistent with this finding and as illustrated in Figure 2.4, cross-country analysis shows that on average SSA PRGFs have targeted inflation rates below 5 percent—with even lower program assumptions in countries that use the CFA franc. Recently, in 2004 and 2005, while the average remained around 5 percent, greater upside and downside variation in PRGF inflation targets has emerged, with a significant number of programs targeting inflation rates above 7 percent in non-CFA franc zone countries.
Figure 2.4.Inflation Targets in PRGFs and ESAFs in Sub-Saharan Africa
Sources: IMF, MONA and World Economic Outlook databases.
1 Data include ESAF and PRGF inflation targets for two periods forward (t + 2) for SSA countries with current programs.
2 Actual data are for the initial conditions when the targets were made (t0) weighted by average GDP for 2001–05.
Box 2.1.Spending and Absorbing Additional Aid
This section’s empirical analysis utilizes the “spend and absorb” terminology set out in “The Macroeconomics of Managing Increased Aid Inflows—Experiences of Low-Income Countries and Policy Implications.”1 The latter was a background paper for the Board’s 2005 PRGF review. Focusing on five countries with aid increases during 1998–2003—Ethiopia, Ghana, Mozambique, Tanzania, and Uganda—it compared how much was “absorbed” (as measured by changes in the current account deficit) with how much was “spent” (as measured by changes in the fiscal deficit). It focused on program outcomes, especially on the authorities’ performance in program implementation, which is not subject to this evaluation. This evaluation’s analysis focused on program design, as the primary conduit of the IMF’s influence.1 See IMF (2005h) and Annex 2 of the current report.
Critics argue that these inflation targets are unnecessarily low and prejudicial to country growth and poverty-reduction objectives. Drawing on the IMF’s own analysis, they argue that the targets could be raised without sacrificing stability and growth objectives, thereby providing an additional source of fiscal space for priority programs.14 Survey results illustrated elsewhere in this report suggest that Fund SSA mission chiefs are evenly divided on whether (or not) the Fund should tolerate higher inflation rates in good performers, with no support for relaxing inflation targets in weak performers.15 During interviews, staff said that the authorities—especially of countries where there has been considerable success on stabilization—tended to resist an easing of targets. Meanwhile, IMF policy staff acknowledge that the empirical literature on the inflation-growth relationship is inconclusive, but weighing benefits and costs of inflation argue for an inflation target range of 5 percent to 10 percent since in their view “the scope for creating more fiscal space through a higher inflation tax is likely limited, if it exists at all.”16 This is consistent with IMF Board policy of single-digit inflation.17
The evaluation’s desk reviews support the finding that inflation control and domestic debt management have been key drivers of programmed spending levels. Program documents frequently cite the control of inflation as a factor in explaining program design, especially the setting of monetary and fiscal targets. And interviews with staff confirm that inflation remains a key driver of program design. Desk reviews show that domestic debt considerations loom large in PRGFs—with most programs limiting domestic financing of the government amid concerns about inflation, debt sustainability, and private sector crowding out (see Annex 3). Among the evaluation’s case studies, Mozambique was a clear case where concerns about crowding-out motivated the limitations on domestic borrowing, while in Ghana the level and sustainability of domestic debt was the main concern behind the program’s constraint on domestic financing.
Aid absorptive capacity
The desk reviews also point to PRGF analysis of three aspects of aid absorptive capacity—including competitiveness risks, fiscal and debt sustainability, and fiscal governance—but almost no attention to sectors such as education, health, and infrastructure. The macroeconomic analysis of absorptive capacity typically focused on “Dutch disease” and competitiveness risks, which were generally found not to be a concern for the levels of aid inflows under discussion.18 Debt sustainability analysis was also common—given the preponderance of HIPCs in the evaluation sample countries—using the Fund’s standard framework for analyzing debt dynamics for low-income countries.19 On fiscal governance, PRGFs included structural conditionality on public financial management and accountability, often supported by technical assistance.20 But PRGF attention to aid absorptive capacity constraints in education, health, or infrastructure, where the Bank is the lead agency, were rare, as was the integration of the individual dimensions into an overall assessment that takes account of synergies and trade-offs across the individual dimensions.
These findings are generally consistent with the responses to the evaluation survey’s questions on absorptive capacity (see Figure 2.5). They show much lower scores for the coverage of sectoral and integrated approaches than for the core Fund areas of macroeconomic issues, fiscal governance, and debt sustainability. This pattern is in line with agreed division of labor between the IMF and the Bank, and the IMF’s comparative advantage vis-à-vis the World Bank and other partners. But it suggests a missed opportunity for considering synergies and trade-offs between areas where the Bank has the lead on one issue and the Fund on another—such as the Bank’s lead on infrastructure, with its obvious supply-side effects and the Fund’s lead on macroeconomic stability and sustainability, including exchange rate competitiveness. The evidence from the evaluation survey suggests that in appraising the feasibility and consistency of the underlying program and its financing—and in determining the corresponding levels for the programmed spending and absorption of aid—IMF staff looked at the macroeconomic aspects of absorptive capacity. There is no evidence that staff took into account possible trade-offs with sectoral constraints and opportunities.
Figure 2.5.Survey Views on PRGF Analysis of Absorptive Capacity
Proportions of surveyed respondents who agreed/strongly agreed that PRGF program design reflects a systematic analysis of absorptive capacity and supply-side effects of:
Almost all SSA PRGFs include automatic adjusters to deal with unanticipated aid shortfalls or windfalls.21 Such adjusters set out the preprogrammed response for targets on international reserves (affecting the current account) and domestic financing of the budget. They typically cover six-month periods; beyond that, program design is reconsidered at the next review.22
Limited domestic financing of shortfalls and full saving of windfalls is the most common practice to deal with fluctuations in budget support in SSA PRGFs. In practice, for aid shortfalls, most SSA PRGFs balance concerns about macroeconomic stability against cuts in priority spending and allow for limited domestic financing. For aid windfalls, most SSA PRGFs call for full saving, until the next review—often citing the need to reduce domestic debt.23 This approach is in line with Fund policy and guidance for staff, which call for full or partial adjustment (to financing shortfalls) depending on various factors including, inter alia, the level of international reserves and the rate of inflation. For windfalls, IMF policy calls for full savings “where desirable.”24
As the overall stance of policy has eased in good performers in recent years, the adjusters have also eased, allowing for both more spending of windfalls and/or more financing of shortfalls. As discussed more fully in Annex 4, in Burkina Faso, Ghana, Mozambique, and Tanzania, for example, aid windfalls have been allowed to be fully or partially spent; shortfalls have been allowed to be fully financed in Tanzania and partially financed in Mozambique and Rwanda.25
But the use of adjusters continues to breed controversy among civil society critics of the IMF, which has failed to get across the short time period to which the adjusters apply. For example, in Mozambique adjusters had been interpreted by critics as a way for the IMF to block the use of aid rather than as a way to manage the short-term macroeconomic implications of changing aid flows. Subsequently, the formulation of the adjuster was changed to allow for full spending of aid windfalls, linked to priority poverty-reducing expenditures. 26
Analysis of Aid
As a basis for considering critics’ concerns that the IMF has lacked ambition in projecting, analyzing, and identifying opportunities for the use of aid in SSA countries, this section looks at three issues: (1) the basis for the PRGF aid forecast; (2) how possible alternatives were taken into account; and (3) the transparency of the aid forecast.
Forecasting aid inflows
Concerns about PRGF aid forecasts have long been at the core of external criticisms of the IMF, because of their ties to medium-term macroeconomic and expenditure planning and donor aid plans.27 Empirical analysis carried out for the evaluation suggests that SSA PRGF aid projections were typically slightly optimistic for the program year and significantly pessimistic for the outer years.
Cross-country analysis indicates that PRGF aid forecasts are accurate for the program period one year ahead. If anything, they are on the optimistic side. This finding is not new to the evaluation, but the evaluation’s analysis does reaffirm it.28 What is new is the evaluation’s finding of a significant underprediction for the outer years of SSA PRGF program periods.
PRGF aid forecasts have typically tapered down over the medium term—reflecting long-standing experience with actual aid flows.29 But, in recent years, the pattern of actual aid flows to SSA countries has changed. As can be seen on the right-hand side of Figure 2.6 rather than tapering down, aid outcomes have remained fairly constant over the medium term—in line with current trends illustrated in Figure 2.1. This changing aid environment created gaps between the level of aid actually realized in given periods and what country PRGFs had projected several years previously, complicating medium-term expenditure planning.30
Figure 2.6.Programmed and Actual Aid Flows: PRGFs Underpredicted Medium-Term Inflows 1
Additional analysis suggests that PRGF medium-term aid projections have begun to catch up with the changing trends in aid tapering (see Annex 2). This trend seems to be driven by changed expectations regarding the tapering of aid levels. The volatility of aid about its mean level (whether tapered or not) has not come down (see panel B in Figure 2.1).
Evidence from the evaluation’s case studies supports this finding (see Annex 4). PRGF programs in five major aid recipients—Burkina Faso, Ghana, Mozambique, Rwanda, and Tanzania—show a recent shift in 2005–06 with respect to the forecasting of aid. This shift is characterized by less tapering of projected aid over the medium term than in earlier years of the PRGF period.31 And in three instances (Burkina Faso, Ghana, and Tanzania) aid is assumed to remain constant or even increase over the medium term. These program shifts have occurred in the context of improved country macroeconomic conditions and aid prospects.
Assessing aid requirements
The larger debate over the IMF’s analysis of aid flows lies in the appropriate concept to be used for the program’s aid “requirements.” To provide a common vocabulary for the discussion that follows, Box 2.2 sets out four different concepts, as used by different members of the international financial and development communities.
The evaluation team’s review of documents confirms that PRGFs in the evaluation sample incorporated most-likely-scenario aid forecasts. Assessing program external financing requirements against availabilities and financing gaps has long been a feature of Fund-supported programs, with “satisfactory assurances” of any gaps being filled an essential prerequisite for Board consideration of a request for use of IMF resources. In the evaluation sample’s PRGFs, there were few cases of inadequate initial financing. In almost all cases, the forecast of available aid, as discussed above, was above the minimum required for the program.
PRGFs in the evaluation sample did not analyze potentially higher levels of aid than the aid forecast underpinning the program. As detailed in Annex 1, during the 2004 and 2005 Board discussions of PRSP implementation, Executive Directors considered how “alternative scenarios” could help to bridge gaps between realism and ambition in national Poverty Reduction Strategies and provide a possible basis for the scaling up of aid at the country level. They “concurred that Fund staff should help those countries that sought assistance in preparing such scenarios.” Subsequently, IMF staff undertook several exercises that covered a range of objectives and levels of complexity in the design of alternative scenarios. An MDG-costing scenario was carried out for Ethiopia at the request of the authorities, while several other country exercises have utilized a simpler methodology.32 But even these exercises have been delinked from the PRGF programs, which remain single-scenario processes, grounded in the most-likely-scenario aid forecast.
Box 2.2.Different Concepts of Aid Scenarios
Four concepts of aid requirements are used by different members of the development community. These are:
The minimum requirements for a viable macroeconomic program, in light of other sources of funding and taking account of macroeconomic and other relevant policy adjustments;
The most-likely-scenario aid forecast based on donor indications and past experience—basically, the aid forecast before any extraordinary measures to mobilize additional funding;
The maximum amount of aid consistent with country absorptive capacity, beyond which capacity constraints materially undermine the net benefits from further increases in aid; and
The normative financing requirements for achieving or pursuing the MDGs and/or other development goals.
IMF-supported programs deal with the first and second concepts. If actual aid (and other external resources) appear likely to fall short of the minimum deemed necessary for a viable program, there is a “financing gap.” That gap would need to be filled before any PRGF arrangement could be considered by the Board.
Many in the development community focus on the third and fourth concepts, which involve financing gaps. Jeffrey Sachs, for example, argues that the IMF should also utilize these concepts as a way of drawing shareholders’ and donors’ attention to SSAs’ vast needs for pursuing the MDGs.1 (In cases in which the level of aid considered necessary to reach the MDGs exceeds the maximum level currently considered absorbable, policy and institutional reform measures to improve country capacity are the priority.)1 See Sachs (2005 and 2006).
Transparency of aid forecast
PRGF documents say little about the derivation of the aid forecast and its underlying assumptions. This is true both for Board documents that are ultimately published and the internal working documents to which the evaluation team had access. PRGF documents allude to the degree of donor support, noting the importance of good country performance in sustaining donor flows and aid predictability, the need for improvements in donor coordination, and—during the early part of the PRGF period—the desirability of reducing aid dependence. But in-depth discussion of how the forecast is actually made, along with key assumptions—such as discount factors used in translating donor promises into program assumptions and how the current forecast relates to current donor undertakings and past donor performance—is rare. Some program documents note past forecast errors, but typically do not link such observations to the current forecast.33
During the evaluation interviews, staff said that they generally took the forecast of the authorities for the program year, validated through discussions with donors. They said that the finance ministries of most SSA countries receiving large volumes of aid are now tracking these flows well, building on the improvements in donor practices in recent years. Nevertheless, where necessary—for example, in post-conflict cases where government capacity was more constrained—Fund staff played a more active role in working with the authorities to aggregate donor plans in the context of the program’s macroeconomic framework. Interviewed staff said that the authorities were in many cases very conservative about future aid flows, and, for medium-term forecasts, staff often triangulated between the authorities’ forecast, to which they added a premium, and indications from donors.
The lack of transparency about the aid forecast means that readers cannot understand (or challenge) the basis for key program assumptions. Nor can they use IMF documents to track donor actions against promises—although there are exceptions, where quarterly aid projections and actuals are included in the PRGF documents.34 In responding to this point, some staff questioned whether it is the job of the Fund to provide such information for outside partners and observers. But the basis for Fund aid forecasts and the specifics on donor commitments and disbursements are matters of increasing public interest, recognized by the Managing Director and others,35 so greater transparency about their underpinnings could be a useful and cost-effective investment.36
Key Features Agenda
The “Key Features of PRGF-Supported Programs” were issued by management as guidance for staff in 2000. They remain central to the staff PRGF Handbook37 (see Box 2.3). They were discussed by Executive Directors at an informal seminar in 2000 and re-endorsed as a useful summary and guidance document during the 2002 Board review of the PRGF.38 They were considered in a 2004 IEO evaluation of the IMF’s role in the PRSP and the PRGF.39 Building on these efforts, the current evaluation has focused on those aspects of the key features with particular relevance for the Fund’s role in aid to SSA.40
Box 2.3.Key Features of PRGF-Supported Programs
A Handbook for the Staff on PRGF Arrangements, updated on May 24, 2006, sets out key features that PRGF-supported programs share.1 As summarized in the Handbook, these include:
“Budgets should be pro-poor and pro-growth.”
“Appropriate flexibility in fiscal targets should be ensured by presenting in PRSPs normative macro-projections to signal financing needs and, where warranted, seeking higher aid flow commitments that can be built into the program.”
“The social impact of major macroeconomic adjustments and structural reforms are to be analyzed and taken into account in the formulation of the program.”
“There is strong emphasis on measures to improve public resource management and accountability by opening fiscal policies and objectives to public debate, developing transparent monitoring systems, and considering selective conditionality on fiscal governance measures.”
“Structural conditionality should be selective.”
This evaluation considers all of these features except the last, as it is the subject of an ongoing IEO evaluation on structural conditionality that covers PRGFs as well as other IMF instruments.2
The Handbook notes that the key features were included in a paper discussed with Executive Directors in 2000, and that progress on them was assessed in another paper discussed in early 2002, when Executive Directors “arrived at a broadly favorable assessment but saw the need for an increased focus on the sources of growth in PRGF-supported programs and structural reforms to develop the private sector, increase foreign direct investment, enhance external competitiveness, and increase labor productivity where these goals are critical to the success of the Fund-supported program.”1 See IMF (2000a and 2006g).2 See IEO (2005a).
Pro-poor and pro-growth budgets
The key features agenda has long included pro-poor and pro-growth budgets. At the outset, the expectation was that PRGFs would support “a reorientation of government spending towards the social sectors, basic infrastructure or other activities that demonstrably benefit the poor.” Subsequently, with the Board’s increasing focus on the sources of growth in PRGF-supported programs—as noted in Box 2.3—the relative importance of infrastructure in the pro-poor and pro-growth budget equation has implicitly risen.
Public expenditures for education and health have generally kept pace with overall public spending relative to GDP—while spending for poverty-reducing expenditure (PRE) programs tracked under the HIPC Initiative have expanded more rapidly41 (see Figure 2.7). Though clearly a sectoral area of Bank leadership, IMF staff have played a role in promoting these increases, albeit indirectly. They vet the PRE spending numbers as part of their work with the authorities on budget forecasts and outturns, and on which they are widely perceived by SSA authorities as playing an important role. Most PRGF documents include a table on PREs, which is subsequently reflected in the Bank-Fund annual HIPC implementation reports.42 In addition, a number of SSA PRGFs have included conditionality on such spending, expressed in terms of floors.43 More recently, program adjusters for the spending of unexpected changes in aid flows have been linked to PREs in a number of the evaluation’s desk-review country cases,44 in contrast to the more common approach to adjusters, which typically calls for the saving of such inflows until the next program review.45
Figure 2.7.Public Spending on Education, Health, and Poverty-Reducing Expenditure (PRE)
Average Social Spending in SSA PRGFs1
(In percent of GDP)
Sources: IMF, Fiscal Affairs Department, and WETA and World Economic Outlook databases.
1 Weighted by average GDP for 2001–05.
2 Excludes the Democratic Republic of the Congo, Guinea, Rwanda, and Zambia.
Many critics see the IMF as undermining the social sectors—especially through PRGF conditionality on the public sector wage bill. Such conditionality is said to adversely affect health- and education-sector salaries and staffing (sometimes donor financed) and in turn the quality of service delivery.46 After much dialogue and debate between IMF staff and their critics, a consensus is emerging that, with few exceptions,47 PRGF wage bill ceilings in SSA have typically been designed to be sector-neutral—basically macro-focused—but not pro-poor.48 They have generally been included in programs because of concerns about macroeconomic stability and administrative capacity constraints on keeping the wage bill within budget (see Annex 3). But they are not first-best solutions and clearly have sometimes had unintended consequences. Possible side-effects range from the limited ability to immediately absorb and spend unanticipated aid inflows for the hiring of teachers and nurses to the proliferation of fringe benefits and other nontransparent forms of remuneration designed to circumvent the ceilings. In either case, the IMF has generally reacted to negative publicity and controversy on wage bill ceilings with program modifications at the next review, as for example in Mozambique in 2006 and Zambia in 2004. More proactively—and in an important innovation—program adjusters in the 2005 Malawi PRGF allowed for both increased PRE spending and exemption from the wage bill ceiling in the event of larger-than-programmed disbursements from the multi-donor AIDS SWAp.49
By all accounts, public spending for infrastructure fared less well than education and health during the evaluation period.50 Despite weak data, the increasingly widespread view is that SSA’s public-expenditure “pendulum” has gone too far in the direction of pro-poor spending for safety net programs, at the expense of pro-growth spending for infrastructure. This is especially so, given the latter’s importance for private sector development, productivity growth, external competitiveness, and employment creation—and in turn for durable poverty reduction.51 As noted earlier, the IMF has done little to take into account spending composition issues—including between the social sectors and infrastructure—in considering country aid absorptive capacity, despite the implications for the supply-side response over the medium term, and the more immediate implications for the optimal absorption and spending response to additional aid.
Government officials in most countries visited by the evaluation team complained about what they characterized as the Fund’s overemphasis on pro-poor spending, which they saw as prejudicial to the needed spending on infrastructure, which they saw as pro-growth. Of course, the Bank is the lead agency on infrastructure, just as it is on health and education. Even so, the indications are that IMF staff could have been more proactive in the dialogue (with the authorities and the Bank) in querying the infrastructure constraints to growth, especially given the relevance to aid-related exchange rate and competitiveness issues, which are at the core of the Fund’s mandate. Indeed, during the Board’s 2005 review of the PRGF, “Executive Directors also encouraged countries in which higher aid-based spending would pose a serious threat to competitiveness to consider using the aid for enhancing productivity and/or removing domestic supply constraints.”52
Financing needs for pro-poor and pro-growth budgets
Where IMF staff deserve more credit for their work on the Key Features is on the increase in fiscal space that opened the way for the expansion of the spending programs discussed above. On average in the 29 SSA PRGF countries, public expenditures rose by about 2.5 percent of GDP over the period.53 In the aggregate, the financing came from increases in aid and domestic revenues and a decline in external debt service, partly offset by the retirement of domestic debt and other transactions.54
Through the PRGF, Fund staff played an important role in the increased fiscal space, especially with respect to their support for ambitious measures to mobilize domestic resources. As spelled out in Annex 3, most PRGFs called for improving domestic resource mobilization, with their program intent evolving from the avoidance of aid dependency in early PRGF programs to the widening of fiscal space for priority expenditures more recently. In addition, as shown in Annex 2, revenue mobilization targets have both increased in PRGFs relative to ESAFs, and more frequently been met or exceeded.
But Fund staff were less ambitious externally—in signaling the incremental aid needs for financing larger pro-poor and pro-growth spending, as called for in the Key Features and the PRGF Handbook.55 As discussed earlier in this report—and because of the policy cautions discussed there—Fund staff have generally not been proactive in analyzing alternative aid scenarios or normative aid requirements for meeting national growth and development objectives, or in discussing with donors additional aid opportunities where country absorptive capacity exceeded projected aid flows.56 Similar reservations clearly apply to the consideration of possibly higher aid commitments for pro-poor and pro-growth programs in education, health, and infrastructure. But there are also other complications, namely that (1) the Bank is the lead agency in these sectoral areas and (2) Fund-Bank collaboration is not working particularly well in these areas, because of problems with resources and delivery modalities (see the last paragraph of the section “World Bank staff” on Fund-Bank collaboration).
Poverty and social impact analysis
From the launch of the PRGF, social impact analysis was to inform the consideration of distributional impacts of program design and the identification of countervailing measures to offset adverse impacts.57 IMF staff were generally not expected to do the PSIA analysis themselves, but rather to integrate the analysis of partners, especially of World Bank staff, into program design.58 As summarized in Annex 1, Board discussions have repeatedly emphasized the importance of PSIA for PRGF program design and called for systematic treatment of impacts and countervailing measures in PRGF documents.59 They also have highlighted the fact that the World Bank was the lead agency on PSIA, given its role as the lead agency on poverty reduction.60
PSIAs carried out by World Bank staff, DFID, and other agencies have not systematically informed PRGF program design. During interviews, IMF staff said that most PSIAs prepared by other agencies generally lacked the necessary timeliness, relevance, and/or quality to underpin PRGF design. It was for this reason that the Fund’s PSIA Group, set up primarily to help staff integrate PSIAs done by others into PRGF-supported programs, had become a producer of PSIAs. World Bank staff working on PSIAs indicated that they generally lacked incentives and resources to meet the specific needs of IMF-supported programs. However, there have been exceptions when the collaboration worked well, such as in the PRGF for Ethiopia, where Fund staff relied on a World Bank PSIA on petroleum pricing.
The findings of PSIAs carried out by IMF staff are now typically reported in PRGF documents, although there is less evidence of material influence on PRGF program design. The results of the nine SSA PSIAs have been presented in program documents—often in freestanding boxes—with the results actually figuring in staff appraisals in fewer cases (Burkina Faso and Djibouti).61 Program documents indicated no specific countervailing measures linked to the PSIAs, in some cases because the recommendations were not adopted (Malawi and Uganda). The Ghana and Mali programs noted that the fiscal space created by the subsidy reduction would be used by the authorities to increase priority expenditures. Going forward, close management of PSIAs is needed to prevent them from becoming a bureaucratic requirement with little impact on program design and outcomes.
In the meantime, PSIA is clearly an area of continuing debate about what the IMF has actually committed to do and what is feasible both analytically and with available data and resources. Civil society critics are looking for greater attention to social impact analysis of “macroeconomic” issues, such as a reduction in inflation or the fiscal deficit.62 But Fund staff argue that distributional analysis of such high-order aggregates is not particularly tractable or cost effective; they contend that PSIAs are best done on narrower questions, such as a reduction in energy subsidies or an increase in value-added tax (VAT) rates, for which clear counterfactuals can be constructed.63 The IMF staff position on the coverage of PSIAs is consistent with the language of the Fund’s initial Key Features document submitted to Executive Directors in 2000, and subsequently posted on the IMF website. The undertaking clearly indicated that “to be feasible, this type of analysis would need to be restricted to substantial macroeconomic adjustments (e.g., a big tax increase, subsidy reform, or exchange rate realignment) or major structural reforms (e.g., civil service downsizing or price liberalization).”64 It also is consistent with the view emerging from PSIA practitioners in DFID and the World Bank, who highlight the importance of clear and narrow questions for successful and cost-effective analysis.65
Of all the Key Features agenda items, IMF staff have pursued improvements in the accountability and transparency for the management of public resources the most aggressively. This focus is in line with the priority attached to it by the Board, which has repeatedly stressed the importance of such work for growth, poverty reduction, and aid effectiveness—in the context of the HIPC Initiative as well as the PRGF. Executive Directors identified it from the outset as an area where conditionality might be expanded, while recognizing it as an area of shared responsibility with the World Bank. Subsequent feedback—in the context of Board discussions of PRSP and PRGF reviews, HIPC implementation reports, and individual country programs under the PRGF and/or program reviews—has served to further sharpen the focus of staff efforts. Systematic monitoring and reporting by Fund and Bank staff point to progress on fiscal governance but with major challenges remaining.66
IMF staff have made a major effort to support the strengthening of public financial management and accountability systems in SSA countries. Their increasing efforts in recent years reflect the confluence of (1) traditional concerns about macroeconomic stability and the underlying processes and systems for ensuring budget execution and reporting; (2) shareholder concerns about governance and the need to ensure the proper disposition of debt service savings from the HIPC Initiative and the MDRI more recently; (3) donor interest in improving country fiduciary systems as a quid pro quo for their own shift to budget support instruments; and (4) effective Fund-Bank collaboration on the issues, with country teams supported by technical specialists in both institutions.
Fiscal transparency and accountability has been a substantial area of PRGF focus, with extensive discussions of budgetary control and transparency issues in program documents and structural conditionality framed in a variety of ways—from prior actions and performance criteria to indicative targets and benchmarks. Strengthening the capacity of the ministry of finance for the monitoring of line ministries’ budget planning and execution has been common,67 including the monitoring of public sector employment and remunerations.68 Closely linked to these programs, the IMF has provided extensive technical assistance on budget execution issues—for expenditure monitoring and control—and information systems for the tracking of expenditures.69 Fiscal transparency and accountability issues are a shared responsibility with the Bank, and program documents frequently explain the scope of institutional collaboration and the division of labor (including on the provision of technical assistance), with the Fund generally focusing on budget execution issues—especially expenditure controls and fiscal reporting.
The end-of-period increase shown in Figure 2.1’s volatility chart (panel B) reflects the step-up in aid to SSA discussed in the preceding paragraph in the text.
See IMF (2006g). These debt limitations have stemmed from concerns about debt sustainability and free-rider issues in connection with debt relief initiatives. Until recently, they were generally not binding as market conditions limited creditor interest. But in a post-HIPC and post-MDRI world, the situation has changed, with major issues related to controlling the accumulation of new debt going forward.
More specifically, it looked at the correlation between changes in net aid inflows and changes in the net current account deficit before grants and interests (absorption of aid) and changes in the net primary fiscal deficit before grants (spending of aid) for those program years, in which the Fund anticipated increases in aid compared to one year before. The analysis relied on data from an internal IMF database—Monitoring of Fund Arrangements (MONA)—which contains macroeconomic variables for about 600 ESAF and PRGF requests and reviews, covering the period between 1993 and 2005. See Annex 2 for further details.
Excluding official transfers and interest payments.
Assuming no private net capital outflows. The increased reserves do not necessarily have to be accumulated at the central bank but could also be held by the private sector.
For aid decreases, the estimated empirical relationship between programmed absorption levels and reserve stocks is smooth—with higher reserves associated with greater reserve financing of aid shortfalls. This is in contrast to the estimated relationship for aid increases, where the threshold of 2.5 months of reserves applies (see Annex 2).
Of course, debt sustainability is also a factor affecting the assessment of country aid absorptive capacity with respect to grants versus concessional credits.
Lack of MONA data on public domestic debt precluded the statistical analysis of its implications for program design, similar to inflation. Desk reviews, however, identified domestic debt as a key program driver (see page 10, third paragraph).
Excluding grants and interest payments.
The statistical analysis suggests that the programmed fiscal adjustment to aid reductions is determined not by the initial inflation rate, but by the initial stock of reserves. In other words, even when inflation is 5 percent or less, IMF-supported programs do not allow domestic financing to offset reductions in external aid. If there are sufficient reserves, they can be utilized. But if not, programs on average require the mobilization of increased domestic revenues and/or expenditure cuts to compensate for the reduction in aid.
Controlling for other initial conditions such as domestic financing and growth, similar results are also found for higher thresholds up to 7 percent. See Annex 2.
The statistical tests suggest that the 79 percent (associated with inflation rates below 5 percent) is not significantly different from 100 percent, while the 15 percent (associated with inflation rates above 5 percent) is not significantly different from zero.
See, for example, Action Aid International (2005).
See the middle panel of Figure 4.3. Of the 22 mission chief respondents to the survey, only one said that he/she agreed/strongly agreed that higher inflation should be tolerated in all countries, that is, regardless of performance.
See IMF (2005g).
See also IMF (2005g).
See the discussion in the section “Fiscal governance.”
See Annex 2 for a comparison with adjusters in non-SSA countries.
See IMF (2006g).
See Annex 2.
See IMF (2006g).
See Annex 4.
See, for example, Oxfam (2003).
See OECD-DAC (2005).
This is true of experience in SSA and elsewhere, both including and excluding debt relief. See Annex 2.
Medium-term underprediction has a less detrimental effect than underprediction for the initial program year. But it still carries the risk of distorting investment, savings, and employment decisions, which need to be formulated with a medium-term perspective in mind.
Measured by the change in aid projections in t + 1 and the fore-cast for t0, the immediate program year.
See IMF (2006f), and also Mattina (2006). In addition to Ethiopia, alternative scenarios have been prepared for the Central African Republic (IMF, 2005n), Ghana (IMF, 2006n), Mali (IMF, 2005p), Rwanda (IMF, 2004f), and Zambia (IMF, 2005q). See Gupta, Powell, and Yang (2005) for a discussion of the macroeconomic challenges of scaling up.
See Annex 3.
See, for example, the recent Tanzania program in IMF (2006d).
See, for example, the Managing Director’s statement to the Development Committee—IMF (2006l).
Especially as the basic PRGF documents for the 29 countries being evaluated are already available on the Fund’s website at www.imf.org.
See IMF (2002b).
See IEO (2004).
As noted in Box 2.3, the current evaluation does not address the selectivity of structural conditionality, since it is the subject of an ongoing IEO evaluation.
PREs are defined in country-specific terms as part of the PRS process; hence, common definitions do not apply across countries. Also, changing definitions of some countries’ PRE “baskets” reduce the measure’s value as a yardstick. To correct for this distortion, the dotted line in Figure 2.7 shows the relationship excluding those countries with changing definitions.
See, for example, IMF (2006i).
With performance criteria in Chad, Guinea, Rwanda, Uganda, and benchmarks and/or indicative targets in Benin, Ghana, Malawi, Mauritania, and Sierra Leone.
Both for using windfalls (in Burkina Faso, the Central African Republic, the Democratic Republic of the Congo, Ethiopia, and Mozambique) and for financing shortfalls (in the Democratic Republic of the Congo, Mozambique, Senegal, and Uganda).
Of course, increased social spending does not automatically translate into better outcomes. For example, studying PRGFs in four SSA countries, AFRODAD found only mixed effects on services and human welfare. Despite higher social sector spending in Ethiopia and Tanzania, poverty reduction and improvements in social service delivery were modest. AFRODAD also criticized the IMF for tight fiscal controls in Malawi and Zambia that inhibited teacher training and recruitment. See AFRODAD (2005) and 2006a–d).
See IMF (2005e).
See IMF (2005k).
Compared with an increase of 1 percentage point over the previous six years (during the ESAF era).
See Annex 2.
Or in the accompanying Staff Report Checklist. See IMF (2006h).
See the section “Aid absorptive capacity” above and Figure 2.5.
See IMF (2004e).
Of the nine PSIAs carried out by FAD, six focused on the removal or reduction of subsidies (such as those on electricity, petrol, agriculture, and fertilizers) and the others on devaluation, external shocks, and taxation.
See Gillingham (2005).
See, for example, IMF (2005c).
For example in Cameroon, Mozambique, Rwanda, Senegal, Tanzania, Uganda, and Zambia.
As in Mozambique, Tanzania, and Zambia. See Annex 3.
For example in Burkina Faso, the Democratic Republic of the Congo, Ghana, Mozambique, Rwanda, Senegal, Tanzania, Uganda, and Zambia. The Fund’s work on technical assistance for fiscal governance was evaluated as part of the IEO evaluation of IMF technical assistance. See IEO (2005b).