Assessment of Current Practice in Fund-Supported Programs

13. The case studies show that projections of next-year aid inflows have become more accurate—in particular, less cautious—over recent years (Figure 1). Considerable aid uncertainty during the Enhanced Structural Adjustment Facility (ESAF) period had led to cautious aid projections in the early PRGF-supported programs. Repeated underestimation, however, prompted more accurate forecasts in recent years. Underestimation on a small scale continued in the Nicaragua and Uganda programs during recent years, where broader macroeconomic concerns called into question the future support of donors. Projection errors remained sizeable even in the past few years in Ethiopia and Ghana because aid was exceptionally volatile. These results largely confirm the findings of the IEO report.¹⁵

View Full Size

Case Study Countries: Range of Forecast Errors in Projecting Aid Disbursements ^1/

(In percent of GDP, excluding Ethiopia and Ghana)

Citation: Policy Papers 2007, 023; 10.5089/9781498333504.007.A001

Source: IMF country reports.^1/ Grants and net external borrowing as recorded in the budget.

• Download Figure
• Download figure as PowerPoint slide

14. For the medium term, aid was often under projected, although projections have become more accurate since 2005.¹⁶ The IEO report noted that Fund forecasts for medium-term aid often proved pessimistic in the past, but had recently caught up as rising overall aid levels were taken into account. However, there is substantial variation in scaling up across countries, that authorities and Fund staff must take into account when doing country-level projections. Targeted donor aid policies led to sizeable increases in aid flows in selected countries where economic performance improved, including those discussed in the case studies. But a scaling up of aid has in fact not been widely observed across low-income countries (Box 3 and background paper, Annex III). Relative to GDP, a large majority of PRGF-eligible countries did not see aid flows from the OECD's Development Assistance Committee (DAC) members rise substantially in 2000-05.

Box 3.

Recent Developments and Initiatives for Scaling Up¹

Overall aid inflows have doubled in U.S. dollar terms since 2000. However, aid seems to be falling behind the commitment by the G-8 countries, at the 2005 Gleneagles summit, to double aid to Africa by 2010. Annual growth in aid continues to slow (excluding one-off disbursements for Tsunami relief and debt forgiveness for Afghanistan and Nigeria). Furthermore, no scaling up is evident when measuring movements relative to GDP. Net ODA flows have remained broadly unchanged as a share of PRGF-country GDP since 2002 (although this may partly reflect the depreciation of the U.S. dollar). Aid volatility has been significant with standard deviations of several percentage points of GDP, not uncommon.

Net ODA Disbursements to PRGF-Eligible Countries, 2000-05

Sources: OECD-DAC; and WEO.

^1/Excluding Afghanistan, Liberia, Somalia, and Timor Leste.

Net ODA Disbursements to PRGF-Eligible Countries, 2000-05

	2000	2001	2002	2003	2004	2005
In U.S.$ billions	23	26	31	36	39	48
excluding Afghanistan, Nigeria, and Tsunami relief	22	25	29	34	36	36
In percent of GDP 1/	24	27	31	32	30	31

Sources: OECD-DAC; and WEO.

^1/Excluding Afghanistan, Liberia, Somalia, and Timor Leste.

View Table

Net ODA Disbursements to PRGF-Eligible Countries, 2000-05

	2000	2001	2002	2003	2004	2005
In U.S.$ billions	23	26	31	36	39	48
excluding Afghanistan, Nigeria, and Tsunami relief	22	25	29	34	36	36
In percent of GDP 1/	24	27	31	32	30	31

Sources: OECD-DAC; and WEO.

^1/Excluding Afghanistan, Liberia, Somalia, and Timor Leste.

Aid movements vary significantly across countries. Relative to GDP, less than 20 percent of PRGF-eligible countries saw a scaling up of aid over the five-year period since 2000. About half the PRGF countries actually received less aid-to-GDP in 2005, than in 2000.

Most of the aid took the form of grants. Excluding debt forgiveness, grants accounted for about 70 percent of DAC member and multilateral assistance to PRGF-eligible countries in 2005, up from less than 60 percent in 2000.

The prospects for aid to low-income countries in the coming years will depend on the implementation of the HIPC and MDRI Initiatives and on commitments to help low-income countries achieve the MDGs.

Estimates by the World Bank and the UN suggest that extra ODA on the order of $40-60 billion a year could be needed to meet the MDGs. The OECD has estimated that ODA from the OECD-DAC countries should rise by $50 billion in real terms between 2004 and 2010 to meet the Gleneagles commitment.

¹ Based on OECD-DAC data for PRGF-eligible countries, which excludes significantly rising aid from non-DAC donors. A more extensive discussion is presented in Annex III of the background paper.

Guiding Principles for Program Design

15. In the face of promises for a substantial scaling up of aid that have yet to materialize across a large number of countries, how should low-income country authorities and Fund staff project aid—and thus frame the authorities' development financing. There are three key considerations for a Fund-supported program that promotes the spending of aid.

16. First, aid projections should represent staff's best estimate of the amount of aid that will materialize, based on all available information. The latter includes formal and informal donor indications, historical patterns, and information from the authorities. An important implication is that aid projections should not be restricted to firm donor commitments. Many country teams already use this approach.

17. second, deliberate over or underprojection of aid would require explicit justification. In the context of a forward-looking budget strategy that takes into account the level and pattern of expected future revenues, including aid, there are different opportunity costs and risks associated with over or underprojection. A downward bias in aid projections would systematically drag down current-year outlays, ultimately resulting in less than full spending of aid. Furthermore, in the short term, if the authorities consider that more optimistic aid projections could start a virtuous cycle of results and funding, staff should weigh the merits of an optimistic approach. On the other hand, an upward bias could result in a mismatch between revenues and financing, force future expenditure cuts, and disrupt needed expenditure continuity. This might not be sustainable over the medium term. Moreover, the absence of budgetary safeguards (see below) could argue for prudent projections, especially if aid is highly uncertain—for example, in post-conflict cases, where prudence may call for a projection based on donor commitments. Finally, estimates of budget support deserve careful scrutiny, as forecast errors may force difficult decisions, such as spending cuts or increased domestic borrowing, and could have a broader macroeconomic impact since they are not self-correcting. Staff reports and Memoranda of Economic and Financial Policies (MEFPs) should state clearly the basis for the aid projections being used both for the next year and the medium term. Projections that contain a significant element of staff judgment—in either direction—should be justified in program documents.¹⁷

18. Third, aid forecasts in a program context need to reflect debt sustainability concerns—through a two-stage approach. The first stage involves an assessment of aid availability as discussed above. In the second stage, for cases where risks of debt distress argue for restricting the use of available concessional loans, Fund staff should reiterate the need for donors to take into account the results of the joint Bank-Fund Debt Sustainability Analysis (DSA) and increase the concessionality of aid. However, if the overall grant element of the entire aid envelope is not raised sufficiently, the projected level of foreign borrowing and the related program targets should be set below the potential amount available.¹⁸

19. In some cases, aid projections may not close the fiscal or balance of payments financing gap. Realistic projections of aid combined with a fiscal program consistent with resource availability should generally reduce the need for a financing gap—which signals spending plans that have not been covered by adequate prospective financing. However, medium-term budget planning may not be synchronized fully with revisions of aid projections. On this basis, financing gaps could emerge, in particular, if: (i) sufficient aid has not yet been identified to cover recurrent costs in outer years—to signal the need to address these recurrent expenses; and (ii) domestic and external resources for the outer years are insufficient to cover essential or "ring-fenced" expenditure needs (see below).¹⁹

20. Fund staff should assist country authorities in preparing alternative macroeconomic scenarios based on higher aid, which would be presented in PRS and Article IV reports. Alternative scenarios can show how additional aid would be used consistent with macroeconomic stability and debt sustainability, and possibly start a virtuous cycle of results and funding. Scenarios that indicate room for the effective use of higher aid consistent with macroeconomic stability can be used by the authorities in their efforts toward aid mobilization. The most useful scenarios would focus on an ambitious but controlled acceleration in aid inflows, rather than on MDG- or needs-based scaling up, which may entail financing gaps that could not realistically be filled. Alternative scenarios can inform changes in program design over time, especially when aid projections are revised. At the same time, PRGF-supported programs should be based on a single, realistic baseline with an aid projection that incorporates the best available information, given that: (i) the costs of meeting the MDGs are hard to assess;²⁰ and (ii) deliberately optimistic projections might give rise to recurrent shortfalls in donor assistance, disrupting fiscal management and giving insufficient guidance on how to prioritize the spending of the lower level of aid actually received.

Assessment of Current Practice in Fund-Supported Programs

22. Fund policy advice and program design have shifted toward a spend-and-absorb approach to accommodate aid increases in countries where macroeconomic stability has been established and fiscal vulnerabilities have been addressed. This shift was noted in the recent IEO report, and is confirmed by the case studies.

23. In terms of program design, a spend approach to increased aid would be reflected in higher overall deficits before grants.²¹ An absorb approach would be visible in an increase in projected net imports that matches the higher aid, and a corresponding programmed change of international reserves that reflects the extent of aid absorption. The higher net imports result not only from the imported component of the aid-based spending, but also from the second round-effect as spending on domestic goods reduces the available resources for producing tradables.²² It follows that while the authorities control the spending of aid, aid absorption is also conditioned by private sector behavior.

24. Since the launch of the PRGF, programs have increasingly accommodated the spending of aid in the case study countries. Targeted fiscal deficits were adjusted in line with the availability of additional donor support in Mozambique (following debt and disaster relief), Tanzania (following debt relief), Uganda, and Zambia (Figure 2). Upward revisions of expected aid flows during fiscal years were largely incorporated into updated projections during program reviews. In other countries, the degree of spending was limited in earlier years: aid projections were uncertain (Rwanda) or domestic debt was too high or reserves too low (Ghana, Nicaragua, and Rwanda). Subsequently, the Rwanda program adopted a spend approach in 2004, followed more recently by Nicaragua (2005) and Ghana (2006).

View Full Size

Case Study Countries: Net Aid Inflows and Fiscal Deficit (Before Grants), 1998-2006

(In percent of GDP)

Citation: Policy Papers 2007, 023; 10.5089/9781498333504.007.A001

• Download Figure
• Download figure as PowerPoint slide

25. From a program standpoint, the absorption of aid is monitored through the monetary program. Most monetary programs target price stability and the build-up of international reserves. A high Net International Reserves (NIR) floor can block aid absorption, but a moderate floor cannot ensure absorption. Thus, by construction, NIR floors allow above-target reserve accumulation, which implies that aid absorption can fall short of its baseline projection, without jeopardizing program targets. Three-quarters of first generation programs adopted a NIR/Net Domestic Asset (NDA) framework that limited central bank domestic credit while setting a floor on the accumulation of international reserves. Two-thirds of second generation programs continued to do so.²³

26. Almost all country cases show that NIR floors were exceeded and thus actual aid absorption was substantially smaller than projected and permitted under the program.

As a result, absorption lagged behind spending. Of the country case studies, only in Zambia did aid absorption generally move in line with spending. In the other cases, a reluctance by the monetary authorities to allow their currencies to appreciate in nominal and/or real terms led to higher-than-programmed international reserves (see below). It is hard to assess, however, to what extent NIR targets were motivated by considerations related to aid absorption. While program documents explicitly assess the need for raising reserve coverage, reserve build-up reflects several sources, and program documents have not always explained on the degree to which the absorption of aid inflows should be accommodated.²⁴

Guiding Principles for Program Design

27. The Fund aims to bring all low-income members to the point where all aid can be fully and effectively spent and absorbed. Fund-supported programs should support the full use of aid whenever this approach does not jeopardize macroeconomic stability. Furthermore, if limits to microeconomic capacity to use the aid become evident, in particular in macroeconomic outcomes, staff should advise the authorities on ways to address these problems.²⁵ If these concerns are not remedied, a spend-and-absorb strategy may not be conducive to sustained growth and poverty reduction, but might instead allow unproductive spending and entail risks that would likely undermine progress down the road. Helping countries address macroeconomic vulnerabilities that might not allow the full use of aid in the short term is at the core of the Fund's mandate.

28. Staff should assess carefully whether the preconditions for the effective and prudent full use of aid are met based on a range of country-specific considerations.

Given the variety of arguments, country circumstances, and development strategies, the appropriate policy mix can only be determined on a case-by-case basis.²⁶ Program documents should explain the strategy for spending and absorbing aid. In particular, deviations from an approach of full spending and absorption would warrant explicit justification in program documents:

• A policy of partial spending and absorption is appropriate for countries with low reserves and/or high external debt. For countries emerging from instability and with low initial reserves, a prudent approach includes saving part of the aid inflows— through the build-up of a prudent reserve buffer and/or the early repayment of relatively expensive external debt. The appropriate measure of reserve adequacy should be assessed in the context of aid volatility and other shocks (see below).

• The authorities may also choose to smooth the use of aid over time. Aid that is not spent and absorbed immediately is still available for later use, when the need may be greater or the use more beneficial. For example, if time is needed to prepare for scaled-up spending, spending effectiveness could call for delaying the use of higher aid. Moreover, there is a strong case for expenditure smoothing (see below). It follows that a spend-and-absorb approach should not imply the simple matching of annual aid flows and their use, but should be implemented in a medium-term fiscal framework.

• An absorb but do not spend approach can be used to lower domestic public debt and/or reduce inflation. In this scenario, the public sector does not spend all the aid, thereby reducing reliance on domestic government financing. This approach also involves the purchase (or reduced emission) of outstanding domestic public debt or base money in exchange for the foreign currency counterpart of aid received by the central bank. This approach was used effectively in Rwanda before 2004 and in Ghana until 2006, preparing the ground for the subsequent switch to a spend-and-absorb approach. In evaluating inflation, available evidence broadly supports the use of single-digit inflation objectives. Higher inflation can undermine growth and also hurt the poor directly.²⁷ However, there is no simple rule: headline inflation above the single-digit level would be more worrisome in the absence of a strong central bank or as a reflection of high domestic government financing, but could be benign in the wake of supply shocks or with scaling up under an exchange rate peg (see below).²⁸

• By contrast, a policy of sustained spending but not absorbing would be harder to justify. This approach is essentially equivalent to one of higher domestically-financed fiscal spending, combined with increased (aid-based) foreign exchange reserves. Nonetheless, this expansionary policy could help exploit possible spare domestic production capacity, with the reserves providing the option of absorbing the spending pressure through higher imports rather than inflation in case the expected supply response does not materialize. Furthermore, other considerations may play a role when deciding on the application of the spend-and-absorb principle on a country-by-country basis.²⁹

29. Attempts to smooth spending of project support is challenging, as projects are "lumpy," and donors and countries have often agreed on a fixed schedule for disbursements and implementation. Thus, country authorities are likely to find it hard to delay donor-funded projects or to advance them with domestic funds. Should the authorities find benefit to such an adjustment, it would require direct negotiation with the donor involved.

30. Strengthening revenue mobilization remains critical in the wake of scaled-up aid. Foreign aid could undermine efforts to mobilize domestic resources, which could lead to long-term aid dependency and aggravate resource volatility. Accordingly, in most cases, domestic revenue mobilization should not be weakened, especially if domestic revenues are below, about 15 percent of GDP.³⁰ However, in case of relatively distortive taxation, there can be a case for using part of the aid to lower tax rates.

Assessment of Current Practice in Fund-Supported Programs

32. Attempts by the monetary authorities to stabilize the nominal exchange rate by accumulating foreign exchange have often led to foreign aid not being fully absorbed. In almost all country cases, the monetary authorities at times resisted nominal appreciation pressures resulting from aid inflows.³⁵ In Ethiopia, Ghana, and Rwanda, the exchange rate was kept stable relative to the U.S. dollar, and in mid-2006, their exchange rate regimes were reclassified by the Fund from managed floating with no predetermined path to a conventional fixed peg.³⁶ The resulting inflation pressures were offset through sterilization using domestic instruments in Mozambique (2000-03), Tanzania (2000), and Uganda (1999-2004).³⁷ In some cases, including in Rwanda (2005), the monetary authorities refrained from sterilizing excess liquidity using domestic instruments despite rising inflation, because of the high sterilization costs. Incomplete absorption of foreign aid may be the result of diverging priorities between monetary and fiscal authorities, when the fiscal authorities decide on rapid increases in aid-based expenditures while the monetary authorities seek to maintain exchange rate stability, or build up a reserve buffer.

33. In several cases, the design of the monetary program was modified to allow or promote the sale of foreign exchange by the central bank. The Nicaragua and Ghana programs lowered international reserves targets during subsequent program reviews—even when aid flows were revised upward. Program design was revised in two countries (Tanzania and Uganda) when limited aid absorption as a result of exchange-rate targeting, led to significantly higher inflation. For both countries, the PRGF-supported programs switched to reserve money targets (to curtail NIR-driven money growth) and lowered their international reserves targets repeatedly. In both cases, inflation was finally reined in, although at the cost of higher and more volatile domestic interest rates and some crowding out of private investment.

Guiding Principles for Program Design

34. A prerequisite for managing these challenges is a clear understanding of the exchange rate regime and the objectives of monetary policy. Effective program design requires that the authorities and the staff share an understanding of the regime, and actual monetary management to be in line with it. Scaling up heightens the importance of avoiding ambiguities in this matter. De facto exchange-rate regimes are identified in staff reports and in the Fund's annual and quarterly reports on exchange arrangements.³⁸

35. In principle, a strategy of spending and absorbing aid can be implemented in the context of each of the standard exchange rate regimes. At the same time, each regime poses different challenges for reconciling the full absorption of aid with the need to avoid high inflation or excessive exchange rate volatility.

36. A floating exchange rate regime implies full absorption of aid and supports price stability; but there can be a case for mitigating short-term exchange rate volatility. In the context of a spend-and-absorb approach, a floating exchange rate regime allows for combining real (and nominal) appreciation with price stability. By insulating the monetary program from the impact of real appreciation pressures, the regime supports the central bank's focus on maintaining price stability (and ensuring an adequate level of reserves). Accordingly, for countries with large and increasing aid inflows, staff has generally advocated continued adherence to prudent monetary objectives and a market-based exchange rate.³⁹ However, with lumpy and volatile aid, the result could be sharp exchange rate movements. This could give reason for stabilizing interventions, without altering real exchange rate trends. In particular, the authorities could adapt the timing of foreign exchange sales, but without relaxing trend monetary growth or resorting to domestic sterilization.⁴⁰

37. The opposite strategy, of fixing the nominal exchange rate also provides a viable framework for scaling up; but aid absorption would lead temporarily to higher inflation. With reserves responding endogenously to higher (spending-induced) import demand, aid absorption is built in, but the monetary expansion would lead to an increase in the inflation rate, and efforts should then focus on minimizing the associated costs. This applies to both a simple peg, as in the franc zone in Africa, or to a crawling peg as in Nicaragua. However, the likely higher inflation under a peg in the presence of large aid inflows should be a criterion in the choice of exchange rate regime.

38. Experience indicates that scaling up in the context of managed floating is relatively hard to manage well. First, this regime tends to be less transparent, which complicates program design. Second, in the absence of a clear rule for ensuring absorption, monetary authorities have frequently refrained from selling aid-based foreign exchange, with the aim of stabilizing the nominal exchange rate. This in turn, resulted in higher (reserve) money growth, thus undermining efforts to reduce inflation to single-digit levels or endanger hard-won price stability.⁴¹ By itself, this impact on (reserve) money can be offset through domestic sterilization. However, such sterilization can not undo the underlying problem of increased (fiscal) spending without an actual resource transfer. Furthermore: (i) sterilization entails significant quasi-fiscal costs—especially in case of thin financial markets, leading to sharply rising interest rates; and (ii) crowding out private investment has adverse growth effects.⁴² Finally, the scope for discretionary exchange rate and monetary policies raises the importance of monetary-fiscal policy coordination in order to match spending and absorption.

39. Therefore, program design in the context of managed floating requires special attention. Targeting a certain exchange rate path or band need not be problematic provided that the regime, including subordination of inflation objectives to exchange rate stability, is clearly spelled out. This case would be akin to a formal peg, in which case no specific conditionality should be set to control money creation. However, if the intervention strategy is not clear, the monetary program may be of little value for guiding monetary management, and NIR targets would be exceeded as long as the monetary authorities give priority to resisting aid-induced appreciation pressures. In order to address these complications, and effectively promote aid absorption, it is important to have a clear understanding of the monetary authorities' objectives and reaction functions regarding inflation, real and nominal exchange rate stabilization, and reserve accumulation.⁴³ Policy coordination between the fiscal and monetary authorities should be ensured in the context of the program discussions. In case of persistent problems, the central bank's formal objectives could be reevaluated, since the pursuit of multiple monetary policy objectives can complicate policy making in the event of a conflict.⁴⁴

40. The NIR/NDA conditionality framework for monetary policy is most conducive to supporting scaling up. Fund conditionality has supported macroeconomic stability, by ensuring reserve adequacy (through the floor on NIR) and a prudent monetary stance (through a ceiling on NDA or reserve money) aimed at price stability. The choice between an NDA or a reserve money ceiling merits careful consideration.

• A reserve money ceiling provides a monetary anchor that can help curtail money growth. It can also promote foreign exchange sales, and hence absorption.

• However, aid volatility may complicate adherence to a reserve money ceiling. As aid is received and deposited, NIR rises and NDA falls, with no change in its sum (i.e., reserve money). The subsequent spending, however, may involve a temporary rise in reserve money, as the fiscal authorities draw down their aid-based deposits (bringing NDA back up) while net imports—to be accommodated by foreign exchange sales (and, thus, a fall in NIR)—have not yet moved up in response. In this situation, selling foreign exchange early in order to adhere to a reserve money ceiling could entail sharp exchange rate swings. ⁴⁵

• On balance, in most cases, the optimal design may involve NDA and NIR performance criteria complemented by an indicative ceiling for reserve money.

• The importance of maintaining an adequate reserve buffer for handling shocks— including temporary aid shortfalls—supports the need for a floor on NIR.

• In the context of large scaling up, however, problems have often involved excessive rather than too little reserve accumulation. NIR-based conditionality cannot by itself address the latter concerns, given the unpredictability of foreign exchange inflows from many sources beyond aid. These concerns should, instead, be addressed through an understanding with the monetary authorities on absorption and exchange rate flexibility, or with the fiscal authorities on matching limited absorption with spending restraint. More generally, these issues highlight the importance of coordination concerning the policy mix between the fiscal and monetary authorities.

Assessment of Current Practice in Fund-Supported Programs

43. The case studies show that large aid volatility during the first years of the decade complicated aid forecasting in Ethiopia and Ghana, resulting in cautious baseline projections and a build-up of a prudent reserve buffer. Both programs continued to underestimate aid flows even recently. They also limited the spending and absorption of aid in order to rebuild reserves, which had fallen to two months of imports or less. Early on, the Ghana program also adopted a domestic debt repayment strategy for the use of windfall revenue from incremental aid. As aid flows became more predictable, Ghana increasingly spent scaled-up aid.

44. The broader examination of program adjusters reveals that unanticipated aid disbursements are subject to various combinations of conditionality. A review of fiscal deficit targets, net domestic financing ceilings and adjusters in 26 first- and 34 second-generation PRGF-supported programs shows that different types of aid were treated differently in between two (usually semi-annual) program reviews (Table 1 and Figure 3). In this context, it is important to note that if program adjusters do not accommodate higher spending in the case of higher aid, the program is not necessarily a binding constraint. Higher spending may not be feasible in the short run, for example given the need for a supplementary budget or prepare expenditure plans. Moreover, the scope for higher spending is always reassessed at the next program review.

Table 1.

Treatment of Unanticipated Aid: First PRGF Compared to Latest PRGF/PSI Programs ^1/

(Number of programs)

Source: IMF country reports.

^1/The review includes 60 programs: all 34 current or recent PRGF-supported programs (second-generation programs) and the first annual programs of the 26 existing predecessor arrrangements since 1999, when the PRGF was established (first-generation programs).

Table 1.

Treatment of Unanticipated Aid: First PRGF Compared to Latest PRGF/PSI Programs ^1/

(Number of programs)

Additional aid	Spend all		Spend some		Do not spend
	first	latest	first	latest	first	latest
Program grants	12	13	1	5	13	16
Project grants	22	30	0	1	4	3
Program loans	2	5	1	5	23	24
Project loans	12	18	0	1	14	15
Aid shortfall	Cut fully		Cut some		Do not cut
	first	latest	first	latest	first	latest
Program grants	12	11	12	12	2	11
Project grants	24	32	0	1	2	1
Program loans	5	12	16	12	5	10
Project loans	20	30	2	1	4	3

Source: IMF country reports.

^1/The review includes 60 programs: all 34 current or recent PRGF-supported programs (second-generation programs) and the first annual programs of the 26 existing predecessor arrrangements since 1999, when the PRGF was established (first-generation programs).

View Table

Table 1.

Treatment of Unanticipated Aid: First PRGF Compared to Latest PRGF/PSI Programs ^1/

(Number of programs)

Additional aid	Spend all		Spend some		Do not spend
	first	latest	first	latest	first	latest
Program grants	12	13	1	5	13	16
Project grants	22	30	0	1	4	3
Program loans	2	5	1	5	23	24
Project loans	12	18	0	1	14	15
Aid shortfall	Cut fully		Cut some		Do not cut
	first	latest	first	latest	first	latest
Program grants	12	11	12	12	2	11
Project grants	24	32	0	1	2	1
Program loans	5	12	16	12	5	10
Project loans	20	30	2	1	4	3

Source: IMF country reports.

^1/The review includes 60 programs: all 34 current or recent PRGF-supported programs (second-generation programs) and the first annual programs of the 26 existing predecessor arrrangements since 1999, when the PRGF was established (first-generation programs).

View Full Size

Treatment of Unanticipated Aid

Citation: Policy Papers 2007, 023; 10.5089/9781498333504.007.A001

• Download Figure
• Download figure as PowerPoint slide

• Fifty-two out of the 60 programs in the sample accommodated unlimited spending of additional project support grants in between program reviews.⁴⁷ This included 22 first-generation and 30 second-generation programs.

• Half of the programs—22 first-generation and 30 second-generation programs— accommodated unlimited spending of additional project support loans in between program reviews (provided borrowing was concessional).

• Almost half of the programs—12 first-generation and 13 second-generation programs—accommodated unlimited spending of additional budget support grants. In addition, one first-generation and five second-generation programs allowed the spending of additional funds up to a limited amount.

• Several programs with mature stabilizers accommodated unlimited spending of additional budget support loans—two first-generation and five second-generation programs.⁴⁸ Furthermore, one first-generation and five second-generation programs allowed the spending of additional funds up to a limited amount.

45. Three-quarters of PRGF-supported programs allowed at least some spending to continue in the face of shortfalls of budget support. But only a few programs, including more recently Afghanistan, Guyana, and the Kyrgyz Republic, allowed project spending to continue on the same scale when foreign funding fell short of programmed amounts.

46. Since 1999, PRGF-supported programs have become more accommodative of spending unanticipated inflows of program aid, and offsetting shortfalls in program support (Table 1 and Figure 3). More programs now attempt to protect spending levels from shortfalls in budget support than before—which was the result of a gradual evolution rather than any discrete shift. Twenty-one out of 34 second-generation programs permit domestic borrowing up to the full amount of an unexpected shortfall in budget support compared to seven out of 26 first-generation programs. In addition, the share of programs that do not accommodate additional aid has fallen. ⁴⁹ Furthermore, second-generation programs seek to safeguard priority spending. Six recent programs set a floor on poverty-related spending levels compared to two earlier programs. The prevalence of program ceilings on the public sector wage bill—in an effort to prevent a rapid rise in wage costs from crowding out other expenditures—decreased from late 2006 to March 2007, from eight programs to three, compared with two first-generation programs.

47. Monetary adjusters for above- and below-program budget support mirrored the fiscal ones:

• About half of the first- and second-generation programs allowed additional budget support grants to be at least partially absorbed (i.e., did not require a matching accumulation of international reserves).

• Twelve percent of first-generation and 30 percent of second-generation programs allowed additional budget support loans to be at least partially absorbed.

• About two-thirds of second-generation PRGF-supported programs relaxed international reserves conditionality for some or all of a shortfall in budget support grants or loans. This compared to 54 percent of first-generation programs in case of a shortfall in grants and 81 percent in case of a shortfall in loans.

Guiding Principles for Program Design

48. Fund-supported programs should promote a smooth path of fiscal spending, in the context of a medium-term budget framework (MTBF).⁵⁰ Temporary deviations from the predefined medium-term path of foreign financing are best absorbed through domestic borrowing, financed through reserve drawdowns. In that way, spending plans would not be interrupted, while the reserve response implies that the change in domestic borrowing by the government would not result in inflation or crowding out. However, persistent shortfalls will eventually require (gradual and well designed) expenditure adjustments, as reserves cannot close a lasting financing gap. Indeed, a spend-and-absorb approach calls for an eventual adjustment to both scaled-up and to lower aid, even if there is expenditure smoothing in the short to medium term. Staff advice on the design of the medium-term expenditure path and on the adjustment strategy should incorporate the following elements:

• Identification of aid shocks. The possibility that shortfalls may be temporary supports the case for short-term financing of deviations from the predefined fiscal path, at least until its persistence can be determined. However, it is often difficult to distinguish temporary from permanent aid shocks. Thus, it is important for the authorities and Fund teams to stay in close contact with major donors to understand the nature of aid fluctuations.

• Self-insurance. Expenditure smoothing using central bank reserves calls for self-insurance through the prior build-up of sufficient international and fiscal reserves. This, in turn, may require some initial deferral in the spending and absorption of scaled-up aid. The use of a target level of reserves from a fiscal viewpoint, in addition to the traditional external viewpoint (in months of imports) can guide these efforts.⁵¹

• Expenditure flexibility. Given the risk that scaled up spending paths may need to be revised down, it is important to examine the realism and flexibility of expenditure paths for broad spending categories (such as wages, entitlements programs, and major investments). Staff should also discuss the recurrent costs of capital projects proposed for scaling up, such as operations and maintenance, which may curtail budgetary flexibility over the longer term.

49. In terms of program conditionality, the case for expenditure smoothing suggests that program adjusters should allow shortfalls in program aid to be financed domestically once reserve adequacy has been achieved. At the time of the next review, the mission and the authorities should then assess the need for changes in the spending path. Project spending tends to be relatively flexible and is often donor controlled. Accordingly, in principle, program adjusters may assume project spending to be tied to actual disbursements, with no impact on domestic financing and NIR.

50. In most cases, best practice would be to allow for short-term spending increases in case of higher-than-expected program grants. In principle, the logic behind expenditure smoothing would call for saving positive aid surprises until their permanency can be properly assessed. This is indeed how the adjusters in the Nicaragua, Rwanda, and Uganda programs work. However, the existence of urgent spending needs and donor expectations likely justify a more flexible approach. The authorities and the staff could discuss in advance plans for supplementary expenditure guided by the expenditure priorities in the authorities' PRSPs and MTBFs. On that basis, programs could incorporate adjusters to allow higher spending (possibly with a cap) covered by higher-than-expected program grants (and loans—in the absence of debt sustainability concerns). This approach should be considered best practice and has been followed in the Mozambique and Tanzania programs. For countries where aid volatility is high and international reserves are low, however, a case can be made for a more symmetric approach to aid surprises. Adjusters could also be more liberal in combination with cautious (commitment-based) aid forecasts (as discussed in paragraph 17).

51. In seeking to protect essential expenditures against the impact of large and sustained shortfalls in aid, program design should draw on the prioritization of expenditures in the authorities' medium-term framework. If aid shortfalls are forcing serious spending cuts, it will be important to avoid disrupting necessary expenditures for selected sectors. A key requirement for facilitating expenditure-switching to more critical sectors is the existence of an operational framework for expenditure prioritization. Moreover, the MTBF could usefully identify ring-fenced priority spending that should be protected from spending cuts.⁵² This exercise could also inform program design. In case resources are projected to fall below the level required for safeguarding essential expenditures, staff could alert donors to the need for additional financing, and the program could accommodate domestic borrowing and/or a drop in reserves to lower levels.

Assessment of Current Practice in Fund-Supported Programs

53. While risks of Dutch disease are a concern, the case studies suggest that they have not posed a generalized problem. There is some cross-country econometric evidence that higher aid has undermined growth in labor-intensive export-oriented sectors.⁵⁵ However, there have been no clear country case studies demonstrating aid-induced Dutch disease.⁵⁶

54. In case of a sustained real appreciation, staff reports generally discuss risks to competitiveness. However, programs have not restricted the use of aid because of a need to avoid or cure Dutch disease. Rather, they have included targeted measures to minimize the risk to export industries and diversification through enhancing their competitiveness. For example, the PRGF-supported program for Rwanda—which faced Dutch disease risks— included the development and execution of action plans for promoting exports in 2004 and 2005, while accommodating a large increase in aid inflows. For Zambia, the need to allow firms to finance temporary losses due to real appreciation motivated the emphasis in the June 2006 program on the authorities' financial sector development plan, parts of which were supported by conditionality.

Guiding Principles for Program Design

55. Given the scant empirical evidence, generally Fund-supported programs should not constrain aid-based spending on the grounds of risks to competitiveness. Only in exceptional cases would such risks provide a compelling argument against scaling up.⁵⁷Concerns about possible Dutch disease effects should be incorporated in the authorities' development strategy (as reflected in the PRSP) which should discuss how initial adverse effects on competitiveness and exports are to be surmounted over time. The Fund should help the country assess risks to export competitiveness, and closely monitor the related evidence in the aftermath of scaling up (Box 4). At the same time, Fund staff would not be required to predict the magnitude of a possible real appreciation as a result of programmed higher aid.⁵⁸

56. Possible adverse effects of aid on competitiveness underscore the importance of using aid well. Even with an initial contraction of exports, the net effect of scaling up on long-term growth could well be positive, provided aid is used effectively. In particular, aid can be used to address bottlenecks from infrastructure, for example, and thus contribute to lowering production costs and promoting exports over the medium term.

Assessment of Current Practice in Fund-Supported Programs

58. Controlling the pace of debt accumulation has always been a component of program design in LICs. However, until the introduction of the Debt Sustainability Framework (DSF) in 2005, programs did not explicitly consider medium-term debt sustainability, instead focusing on overcoming short-term financing constraints.⁶⁰ All programs included a performance criterion (PC) limiting new nonconcessional external debt, in some cases to a zero ceiling. However, for most members, there was no ceiling on medium- and long-term concessional borrowing and thus on overall debt. Also, only about half of the pre-2005 PRGF-supported programs had fiscal conditionality that extended to the overall deficit, which would have limited recourse to external financing.

59. Against this background, the Fund and Bank Boards in September 2004 called for a strengthening of control over excessive borrowing—including on concessional terms—guided by structured debt sustainability analysis. Specific instruments included more systematic use of limits on the overall fiscal deficit (including grants) for countries where debt sustainability is a concern and conditionality related to the net present value (NPV) of external debt.⁶¹ The immediate impact of the new approach has been limited so far. In particular, NPV-based conditionality, theoretically the most attractive measure, is difficult to use. Only the PRGF arrangements for Guyana and Rwanda have an indicative ceiling on the NPV of external public and publicly guaranteed debt, introduced in July 2004 and January 2007, respectively. The 2005 arrangement for the Kyrgyz Republic has a separate ceiling on the contracting or guaranteeing of concessional external debt in addition to a zero ceiling on nonconcessional borrowing.

60. Quantitative conditionality to contain risks of debt distress has been complemented by structural measures, including:

• Putting in place adequate procedures for approving and monitoring external debt, as in Uganda (2006).

• Strengthening investment project selection and prioritization, with independent feasibility studies, as in Albania (2004) and Guyana (2006).

• Developing sustainable medium-term debt strategies, which may require technical assistance.

Guiding Principles for Program Design

61. Guided by the DSF as updated in 2006, programs can use a variety of tools to limit the risks of debt distress (Box 5). Especially close scrutiny of scaling-up scenarios is suggested when: (i) current or projected risks of debt distress are above or close to the high end of the moderate range or the country moves to a higher risk category; (ii) scaling-up scenarios imply a need for sharp shifts in fiscal policy, the investment rate, the financing mix, productivity growth, or other key policy variables; and (iii) key assumptions from the previous two or three DSAs, and hence debt projections, have proven significantly too optimistic; or (iv) debt has grown rapidly—defined as an annual change in the NPV of debt of 5-7 percent of GDP or more.⁶²

62. In case of scaling up using borrowed resources, scenario analysis can help assess associated risks. There is some tension between the conservative projections recommended for DSF medium-term scenarios, and the desire not to under-project growth in outer years, so as not to unduly constrain borrowing. This tension underscores the importance of including an alternative "high investment-low growth" scenario in case the baseline assumes that ambitious scaling up will lead to sizeable growth dividends, given the mixed evidence on the growth effect of higher aid.