ANNEX 4 Country Case Studies: Program Change in Major Aid Recipients
- International Monetary Fund. Independent Evaluation Office
- Published Date:
- April 2007
This annex describes the context and evolution of program design in five major aid recipients: Burkina Faso, Ghana, Mozambique, Rwanda, and Tanzania.1 It complements the desk review analysis set out in Annex 3. It begins with a description of a framework for analyzing the evolution of program design in PRGFs and concludes with specific examples of program change from each of the five case study countries.
This section highlights three sources of change in program design. The changes all took place against the backdrop of improving macroeconomic policies and outcomes.
The first is a relaxation of fiscal policy in Tanzania and Rwanda to allow greater use of available aid. In both countries, the relaxation occurred at the same time as improving macroeconomic conditions and in the context of discussions with donors and the authorities. The programmed fiscal deficit increased both as a share of GDP (Figure A4.1) and as a share of total expected aid (Figure A4.2).
Figure A4.1.Programmed Fiscal Deficit
Source: IEO staff estimates based on IMF staff reports.
Note: Fiscal deficit defined as the difference between expenditures (excluding interest payments) and revenues (excluding grants). Dates indicate year for which program targets apply.
Figure A4.2.Programmed Ratio of Fiscal Deficit to Aid
Source: IEO staff estimates based on IMF staff reports.
Note: Fiscal deficit defined as the difference between expenditures (excluding interest payments) and revenues (excluding grants). Aid defined as the sum of grants, net foreign financing, financing gap, and the net change in external arrears, minus external interest payments. Dates indicate year for which program targets apply.
The second is a change in the medium-term forecasting of aid and the fiscal deficit.2 As discussed in the main text, throughout the early PRGF period the IMF generally forecast the tapering of aid beyond the program year, in line with experience with actual aid flows. But this has begun to change, with recent medium-term aid forecasts catching up with ongoing changes in the aid environment. Figure A4.3 shows the difference between the medium-term forecast of aid (t + 1) and the aid projection for the program year (t0). As seen, programs forecast a decline in aid flows over the medium term before 2005, but started to project less or no tapering by 2006. At the same time programs have also begun to project less tapering of the fiscal deficit. Figure A4.4 shows the trend for the medium-term forecast of the fiscal deficit. Before 2005, programs generally forecast a medium-term tightening of the fiscal deficit compared to the program year, but by 2006 they assume less tapering beyond the program year.3
Figure A4.3.Change in Medium-Term Aid Forecasting over Program Year
Source: IEO staff estimates based on IMF staff reports.
Note: Difference between aid forecasts (as percent of GDP) in t + 1 and t0 (the program year).
Figure A4.4.Change in Medium-Term Fiscal Deficit over Program Year
Source: IEO staff estimates based on IMF staff reports.
Note: Difference between forecasts for fiscal deficit (as percent of GDP) in t + 1 and t0 (the program year).
The third is a change in program adjusters to give countries more flexibility in responding to unanticipated changes in aid flows. Figure A4.5 shows the programmed reaction, through program target adjusters, to shortfalls in aid before and after changes in program design.4 Before the changes, three out of the five programs did not permit domestic financing of aid shortfalls. After the changes, all programs but Ghana’s provided flexibility to finance shortfalls, at least partially. Figure A4.5 also presents adjusters for aid windfalls, and again shows greater flexibility in more recent programs. Before the programs changed, none of the five countries could fully spend aid windfalls before the next review. But after the changes, three of the five countries could fully spend aid wind-falls—with Rwanda the only country that could not spend any.
Figure A4.5.Program Adjusters for Spending of Unanticipated Shortfalls or Windfalls of Aid
Note: Year of change: Tanzania (TZA): 2001/02; Rwanda (RWA): 2003; Burkina Faso (BFA): 2005; Ghana (GHA): 2005; and Mozambique (MOZ): 2006.
This section gives specific examples of how program design has changed in each of the five case study countries. (For each case, the year of program design change is in parentheses.)
Tanzania (2000 and 2005)
Prior to the 2000 PRGF, Tanzania’s programmed fiscal policy stance assumed a substantial reduction of net domestic debt of the government, which meant that a significant share of available aid could not be spent. Programs justified this fiscal stance as necessary to correct the fiscal slippages incurred at the end of the previous ESAF program and to build up reserves, and out of concerns about macroeconomic stability and Dutch disease. The continued compression of public expenditures, even after a degree of macro stability had been achieved in 2000–01, triggered a debate between the Fund and the authorities, donors, and civil society.5 As discussed in a previous IEO evaluation of the PRSP and PRGF process, these discussions took place in the context of the public expenditure reviews (PERs) initiated by the World Bank and were informed by donor-financed studies done by an outside academic.6
In the context of more predictable aid and sustained macroeconomic stability, PRGF programs began to relax the fiscal policy stance in 2001, allowing for greater programmed expenditure of projected aid.7 This is illustrated in Figure A4.2. By the end of 2001, IMF internal reviews were calling for more ambitious government expenditure, and programs started subsequently to project less tapering of expenditures beyond the first program year as well. Although the discussions with donors and the authorities that preceded the changes in Fund stance did not feature prominently in mission briefs or the internal review process, internal correspondence between IMF and World Bank staff shows that the debate on fiscal policy was very active.
Program adjusters were also changed in 2001 to allow for the full use of aid windfalls and the full financing of aid shortfalls (see Figure A4.5). Prior to this change, Tanzania had been unable to finance any aid shortfalls and allowed to partially spend windfalls.
PRGF programs started to forecast less tapering of aid over the medium term in the third PRGF review of 2005 (see Figure A4.3 above). This change was underpinned by Tanzania’s strong performance in core structural areas in 2003 and 2004, the finalization of the second generation PRSP in 2005, and the continued stability of aid inflows between 2003 and 2004. Program concerns about Dutch disease, which had been present in previous years, were no longer considered an issue, despite the sustained aid.
The 2002 PRGF request and subsequent reviews were guided by concerns over external debt sustainability, which led to a decrease in programmed spending of aid in 2003 and 2004 (see Figure A4.2). As in Tanzania, this fiscal stance generated significant debate between the IMF and the authorities, donors, and other multilateral organizations. A donor-financed PSIA was conducted in 2003 that focused on the sustainability of substantially higher fiscal deficits financed by additional external borrowing.8 According to that assessment, Rwanda could expand fiscal spending and the deficit, if financed on concessional terms. The influence of the PSIA in the discussions about fiscal stance is itself a matter of controversy. IMF staff contend that it played no role in the assessment of underlying conditions (as they considered the quality of the analysis to be subpar),9 while many donors considered the PSIA a relevant and influential analysis.10
Rwanda’s PRGF began programming greater absorption and expenditure of aid in the fourth review of 2005, amid reduced concerns over debt sustainability.11 The program change coincided with discussions to top up debt relief under HIPC in 2004 in advance of the country’s reaching the completion point in early 2005. At the same time, programs started to forecast less tapering of aid, the fiscal deficit, and absorption beyond the first program year.12 (These changes were preceded by a change in adjusters in the first review in 2003 to allow partial domestic financing of aid shortfalls.)13
With debt sustainability less of a concern, PRGF program assessments in 2005 focused on the underutilization of aid. In the 2005 program, IMF staff’s concerns of previous years gave way to concerns about the underutilization of aid. But, in practice, aid absorption was limited by the Central Bank, because of its concerns about exchange rate appreciation. In the program documentation, IMF staff argued for limiting reserve accumulation to allow for greater aid absorption (but stopped short of using conditionality). This new policy stance was supported by the IMF internal review process.
Burkina Faso (2005)
Driven by a long record of macroeconomic stability, PRGFs had long allowed the use of anticipated aid in Burkina Faso. This was noted by the 2006 EPA, which highlighted the strong economic performance and record of program implementation.14
By the combined second and third review of the PRGF in 2005, program design changed to forecasting less tapering of aid and the fiscal deficit over the medium term. The program also started to correct for past overestimation of aid at the first program year.15 With the change in medium-term forecasts of aid, projected spending also became more stable with less programmed tapering off.
Adjusters also changed in 2005 to allow greater program flexibility in the spending of aid windfalls. The switch to more accurate aid projections for the initial program year was accompanied by a change in program adjusters. In contrast to previous programs where all aid windfalls had to be saved until the next review, adjusters now allowed for limited use of such windfalls for priority spending. As before, adjusters continued to allow for limited financing of aid shortfalls.
The 2003 PRGF program request was concerned with macroeconomic stability, through containing both domestic debt and the rapid expansion of public expenditures. The program stance was influenced by the serious fiscal and quasifiscal slippages in 2002 that precluded the completion of the fifth and final review of the 1999–2002 PRGF arrangement. The budget slippages related to, inter alia, large public sector wage bill overruns, nonimplementation of revenue measures, delays with the divestiture program of public enterprises, and shortfalls in donor financing related to country performance.16
The third PRGF review in 2005 started to forecast less tapering of aid and the fiscal deficit over the medium term. This program change stemmed from improved macroeconomic performance during 2003/04, including contained domestic government borrowing.
Program adjusters were also changed in 2005 to allow for the full use of aid windfalls. But, as before, financing of aid shortfalls was not allowed, reflecting lingering concerns over domestic debt.
The restriction on nonconcessional borrowing has been the subject of an ongoing debate in Ghana between the authorities and the Fund and donors. The program has maintained throughout this period limits on nonconcessional borrowing, which the authorities describe as overly restrictive of their infrastructure investment plans. The discussions between the authorities and donors and the Fund on nonconcessional borrowing have taken place for example in the context of the Consultative Group meetings. A waiver on the related performance criteria was granted during the third review in 2005 when conditionality was breached for unintended reasons. The authorities were expecting concessional financing that did not materialize, and to avoid a sizable penalty under previous investment commitments by the government, alternative financing arrangements were made.17
In line with earlier programs, the 2004 PRGF request was concerned with macroeconomic stability, fiscal consolidation, and growth-enhancing structural reforms.18 The program placed special emphasis on strengthening government revenues and improving public expenditure management. These concerns, combined with an expected reduction in aid flows, led to a programmed reduction of the domestic primary fiscal deficit (moving to a surplus over the medium term). The program rationale for the fiscal policy stance included the reduction of pressures on domestic interest rates.19
In the fourth PRGF review in 2006, the program projected higher aid and expenditure for the program year, and forecast less tapering of aid and spending beyond the program horizon. These changes arose from the explicit program recognition that macroeconomic and fiscal performance had been better than expected, even in the face of revenue shortfalls. In addition, the program also noted the recent increase in aid and the expectation that it would be sustained. The latter was reflected in the program appraisal, with aid flows no longer considered a main risk to the program as in previous years.
Program adjusters were also changed in 2006 to allow for the full use of aid windfalls and the partial financing of aid shortfalls. Prior to these changes, the program design was a matter of controversy in 2005, with nongovernmental organizations arguing that program targets and adjusters restricted the use of additional aid.20 The Fund responded publicly in 2006 on the use of adjusters and program reviews with respect to aid.21 In due course, the fourth PRGF review in 2006 adjusted the definition of fiscal targets in the program to focus on domestic financing rather than the primary deficit, as was the case in previous programs and reviews.
Each of the five case studies included a country visit by the evaluation team.
This aspect of change in program design was not observed in other desk review cases, including those with more pressing macroeconomic performance issues (e.g., Cameroon, the Central African Republic, and Zambia) and in mature cases where programs had already been forecasting more stable aid and spending (e.g., Senegal and Uganda).
This change in both figures is reflected by a forecast difference between t+1 and t0 that begins negative and moves toward zero.
The change in program design did not occur at the same time in all five countries. The year of program change for each country is listed at the bottom of Figure A4.5.
See IEO (2004).
The 2006 EPA for Tanzania was silent on the discussions about fiscal stance in the early programs and subsequent program design changes (see IMF, 2006c). It noted that the design of programs had been broadly appropriate: programs were appropriately anchored on strengthening fiscal performance, sufficiently flexible to accommodate large aid inflows, and adapted to facilitate greater absorption.
Staff did inform management of the results of the PSIA and their disagreement with donors, and internal review comments supported the prudent policy stance taken by staff, highlighting the debt sustainability concerns.
A recent ODI review of DFID’s PSIA notes that “The immediate outcome was not sufficiently robust for the IMF to change their policy on the limit for the fiscal deficit of Rwanda. . . . However, subsequently it appears that PSIA has had an impact on IMF thinking, at least in terms of their rhetoric.” See Bird and others (2005).
Rwanda’s 2006 EPA did not cover the debate about tight fiscal policy and the subsequent program design changes (see IMF, 2006b). Noting that the design was largely appropriate to achieve macroeconomic stability, with appropriate flexibility to aid in conditionality through adjusters and target definitions, it found that implementation (reviews) accommodated a fiscal stance that was “probably looser than necessary to increase priority spending.” Moreover, programs did not adequately address the continuous deterioration in debt sustainability.
The use of aid windfalls—unlike in Tanzania—continued to be disallowed.
However the EPA was silent about the program design changes that ensued, highlighting strong program implementation and “exemplary” observance of conditionality. Program design was found to be broadly adequate, including pointing out limited absorptive capacity (but with greater attention needed on tax collection). See IMF (2006j).
Due to limited documentation, it is not possible to identify why actual aid flows have constantly fallen behind aid projections until this change beyond the observed basic pattern of programs usually projecting aid increases for the initial program year while actuals remain flat.
A background paper to the 2005 PRGF review examined the macroeconomics of managing increased aid inflows, focusing on the actual use of incremental aid for the period 2001–03. It found for Ghana that in practice there was neither spending nor absorption of additional aid for that period as a whole. Moreover, program design allowed absorption but only partial spending of expected incremental aid. See IMF (2005h).
Ghana obtained a $40 million loan from Nigeria to finance its participation in the West Africa Gas Pipeline.
Mozambique’s EPA took place in December 2003 (IMF, 2003h). It called for sustaining the efforts to consolidating macroeconomic stability and deepening structural reforms. The aforementioned background paper to the 2005 PRGF review found that Mozambique in practice mostly spent and absorbed additional aid for that period as a whole. IMF (2005h).
But the program made also a general reference that to achieve poverty and development goals a significant scaling up of aid would be required, noting also that sectoral absorptive capacity needed to be improved.