A frequently voiced concern of countries that receive development aid is that assistance flows are not predictable. In most years, the amount of aid disbursed differs widely from the amounts expected, and because most aid recipients lack access to international capital markets, they cannot borrow externally when expected aid fails to arrive. As a result, recipient governments are forced to adjust spending plans at short notice when promised aid is not provided or when additional aid is disbursed unexpectedly. Enhancing aid predictability has therefore been a key objective of the international agenda enshrined in the 2005 Paris Declaration on Aid Effectiveness.
A government’s inability to predict aid flows affects not only the level of government spending but also its composition and effectiveness. Unexpected aid shortfalls could force governments to disproportionately cut investments in physical and human capital, while aid windfalls could disproportionately boost government consumption—which, unlike investment spending, can be adjusted without much delay and planning. Thus, unpredictable aid may not only be more difficult to manage, but also affects how the money is spent, thereby reducing its intended impact. Such short-term distortionary responses to unexpected aid shortfalls and windfalls are more likely for budget aid—the kind of aid that flows directly into a government’s budget—because recipients have full discretion on where to spend such aid.
Aid predictability and aid volatility are distinct concepts, although they are often used interchangeably. Aid is predictable if recipients can be confident about the amount and timing of aid disbursements. Aid is volatile if it moves up and down significantly between two time periods. Although measuring predictability requires very detailed data, it is the more relevant concept in studying aid effectiveness issues.
Yet, little systematic empirical work is available about aid predictability. This article summarizes the results of our study (2008), which provides comprehensive empirical evidence on the predictability of aid.
Does aid arrive on schedule? We use two sources of data to examine this question. A first data set, comprising aid disbursements and commitments reported by donor agencies to the Development Assistance Committee of the Organization for Economic Cooperation and Development (OECD-DAC), has comprehensive time and country coverage. But it includes neither separate data on project and budget aid nor a direct measure of aid expectations, because aid commitments reported by donors do not necessarily correspond to the amounts of aid expected by recipients. Also, OECD-DAC data do not include detailed fiscal data to evaluate the impact of unpredictable aid on government spending.
A second, new data set—derived from IMF-supported programs—provides detailed information on joint macroeconomic programming exercises by IMF staff and recipient governments. It includes projections and outturns of aid and a large set of other fiscal and macroeconomic variables. It can thus be used to identify aid expectations of recipient countries and differentiate between budget and project aid. However, it has limited country and time coverage. Both data sets have advantages and disadvantages in addressing different research questions (see Table 2 in Celasun and Walliser, 2008).
Both data sets show that aid is highly unpredictable. According to OECD-DAC data, during 1990–2005, on average, annual aid disbursements in sub-Saharan Africa deviated from aid commitments by 3.4 percent of GDP. Other regions also show deviations of disbursements and commitments in the range of 1.7–2.4 percent of GDP during 1990–2005. But, contrary to the common belief that donors systematically disburse less aid than they commit, low aid predictability in both data sets is a result of disbursements falling short as well as exceeding expectations and commitments, in particular in sub-Saharan Africa. This shows that managing unpredictable aid flows involves both aid shortfalls and windfalls.
Predictability of budget aid is strikingly low even for better-performing recipient countries. In data on IMF-supported programs, budget aid disbursements deviate from projections by about 1 percent of GDP, which represents about 30 percent of disbursed budget aid, on average (see Chart 1). The degree of predictability varies considerably. For example, Sierra Leone, a postconflict country, received 6 percent of GDP in budget aid—and 50 percent of this aid arrived unexpectedly, implying that half of each year’s budget aid was either cut or added while the budget was under implementation. By contrast, in Ghana less than 25 percent of budget aid was unexpected.
The donor side
Most previous research has assumed that lack of aid predictability results mostly from unjustified bureaucratic and administrative delays by the donors. However, donors may also have valid reasons for not being able to provide fully predictable aid (see table). These valid reasons need to be distinguished from reasons not justified by aid effectiveness concerns to understand when lack of predictability negatively affects the potential development impact of aid.
Donors may have several reasons for adjusting the flow of aid unpredictably.
|Is donor behavior undermining aid effectiveness?|
|Reason for difference between expected/announced and disbursed aid||Budget aid||Project aid|
|Slow project implementation speed||n.a.||No|
|Difficulties meeting donor-specific project disbursement procedures||n.a.||Possibly|
|Country circumstances and conditions|
|Major shift in policy or country circumstances, including emergencies||No||No|
|Specific conditions not met||Possibly||Possibly|
|Administrative delays and slow response by donors||Yes||Yes|
|Aid reallocation or additions to aid envelopes for political or donor-related reasons||Yes||Yes|
As a purely technical matter, project aid disbursements may be in chunks (for example, for major infrastructure) and unexpected delays in project implementation by the recipients would lead to unexpected shortfalls in disbursements. Such shortfalls, however, would not be of concern from the aid effectiveness perspective. Delays in project disbursements may also result from recipients not meeting specific procedural requirements for safeguarding aid resources (such as procurement rules for project aid). Whether such delays are justified by aid effectiveness concerns largely depends on how relevant the procedures are in preventing aid from being misspent.
Major shifts in recipient country circumstances would clearly justify changing disbursement patterns. Fundamental shifts in a country’s policies or governance that put in doubt a recipient’s commitment to use aid for the intended purposes could result in donors withdrawing announced aid to prevent resources from being misspent. In some circumstances, aid must be disbursed unexpectedly to be effective. Emergency aid, by nature, is hard to predict, and such unexpected additions to disbursements in response to natural disasters and major economic shocks enhance rather than hinder aid effectiveness.
A more controversial and complicated question is whether specific conditions meant to ensure that country objectives are aligned with donor objectives justify lack of predictability. Such conditions, which are typically applied to budget aid, can include specific policy actions (for example, structural changes to the economy) or indicators (for example, increases in school enrollment rates). If recipients do not comply with such conditions, aid may be reduced or delayed.
In recent years, many budget-support donors have adopted measures to reduce the impact of specific conditions on annual predictability. They have done this by making financing decisions early in the budget cycle and by downplaying the importance of any one action or indicator as a condition for disbursement, instead relying on broader sets of performance measures.
Excessive administrative delays in donors’ aid bureaucracies, cumbersome approval and disbursement processes, and intra-year aid reallocations that prevent the timely disbursement of announced and expected aid for a recipient country clearly present a problem for effective assistance. Donors may also add to or subtract from their originally planned aid to a recipient country during the year in response to political developments in, or based on the aid needs of, other recipient countries. Such intrayear reallocations also hamper aid planning.
Stable program relationships matter
Statistical analysis helps determine whether some of the observable characteristics of recipient economies are responsible for the low predictability of aid flows. OECD-DAC data, which have the required comprehensive coverage, show that recipient countries that have more stable relationships with donors—as signaled by a sustained track record of implementing IMF-supported programs—receive more predictable aid. This relative stability of donor-recipient relations could reflect a higher degree of trust or sound macroeconomic policy implementation by the recipient country. Also, statistical analysis shows that a longer, continuous engagement with the IMF reduces aid windfalls (“surprise disbursements”) but not aid shortfalls. More stable country-donor relationships appear to result in better aid projections and less need for donors to step in unexpectedly with higher aid. The mere existence of an IMF-supported program, by contrast, does not matter for predictability.
Emergencies in recipient countries also explain some of the measured lack of predictability, because donors do not live up to their aid commitments in years when there are large disbursements of emergency aid. This finding indicates that donors sharply increase their commitments during emergencies without necessarily delivering on these promises, a possible indication that aid envelopes are being shifted toward emergency responses and away from other aid activities in the same country.
Our study does not find any other factors—such as governance or terms-of-trade shocks—to be strongly linked with predictability. In addition, a significant part of the lack of predictability cannot be directly linked to long-term donor relations or emergencies. The unexplained part of low predictability may reflect both technical factors in the case of project aid, but also specific conditionality and administrative delays by donors (as outlined in the table).
Adjusting to shortfalls
By using this IMF-based data set—which covers 13 countries with long-term program relations during 1992–2005—adjustments to budget aid surprises can be broken down into changes in tax revenue, current spending, domestically financed investment spending (total public investment spending minus investment spending funded by project aid), domestic bank financing (financing by the central bank and commercial banks), net debt service, and other categories. The other category mostly reflects nontax-revenue and non-bank-financing items.
How prevalent are budget aid shortfalls? Budget aid disbursements fall short of projections in about 60 percent of the fiscal years covered in the sample. The average budget aid shortfall is 1.1 percent of GDP (see Chart 2, top panel). The management of these aid shortfalls is often made more difficult by simultaneous tax revenue shortfalls (0.3 percent of GDP) and current expenditure overruns (0.3 percent of GDP). Recipients therefore typically need to address simultaneously aid shortfalls, tax revenue shortfalls, and current expenditure overruns, amounting to 1.7 percent of GDP. They do so largely, in order of magnitude, through higher domestic bank financing (0.7 percent of GDP), reductions in debt service or increases in arrears (0.4 percent of GDP), cuts in domestically financed investment spending (0.3 percent of GDP), and other financing sources outside regular channels—such as privatization or nontax revenue (0.3 percent of GDP).
Chart 2Adjustments to aid surprises
Countries adjust to aid shortfalls by raising additional domestic financing and by cutting investment spending …
Citation: 45, 3; 10.5089/9781451922394.022.A010
Source: Celasun and Walliser (2008), using data in IMF staff reports from 1992 to 2007 for a set of 13 countries.
What emerges as a key adjustment pattern for aid shortfalls is a mix of additional domestic financing and cuts in investment spending, while current spending is, on average, higher than projected. The data thus confirm that recipient governments would normally not be able to reduce current spending (mostly salaries) but largely concentrate expenditure adjustments on budgetary investment spending. Governments operating in an environment of uncertain budget aid may restrain their budgetary investment expenditures if they do not receive aid early in the budget cycle. Persistent uncertainty about budget aid disbursements also undercuts simple budget management responses to shortfalls, such as the delay of investment spending from one year to the next.
Structural differences in countries’ policy frameworks can result in different adjustment patterns for similar aid shortfalls. For instance, member countries of the West African Economic and Monetary Union (WAEMU), which do not have access to monetary policy instruments and have limited or no ability to borrow from the central bank, had to cut investment spending more deeply than did other countries. On average, WAEMU countries compensated for half of the aid shortfall with a cut in investment spending and financed less than a third of the shortfall through the domestic banking system. Non-WAEMU countries cut investment spending by one-sixth of the aid shortfall, financing three-quarters by borrowing from domestic banks.
Adjusting to windfalls
Additional budget aid finances the repayment of debt or additional government consumption. Higher-than-expected disbursements of budget aid take place about 40 percent of the time, and average 1 percent of GDP for the IMF data set (see Chart 2, bottom panel). On average, none of the excess aid and revenue goes toward additional domestic investment spending. Instead, recipients reduce domestic bank debt (0.9 percent of GDP) and increase current expenditure (0.6 percent of GDP), benefiting from the fact that countries collect more nontax revenues in periods of aid windfalls. Aid windfalls typically come too late in the budget year and thus cannot be spent on items other than current expenditures.
Saving aid windfalls allows building up space for future aid shortfalls and could be part of a strategy to manage unpredictable aid. But, surprisingly, even countries that have received excess aid for several consecutive years appear to use most, if not all, of the extra aid for reducing debt rather than additional expenditure.
With almost identical budget aid and revenue windfalls, WAEMU countries expanded current expenditure by much less (0.4 percent of GDP) and saved more (0.5 percent of GDP) by paying down bank debt as compared with their non-WAEMU counterparts. These policies reflect a larger degree of self-insurance by WAEMU countries, given the tighter domestic borrowing limits faced by governments. In both WAEMU and non-WAEMU countries, little, if any, additional investment spending took place in response to aid windfalls.
How to improve predictability
One of the key results of our study is that low predictability of budget aid can hurt aid effectiveness (see box). We also conclude that a number of areas in the debate on aid effectiveness and improving donor practices need further consideration. First, the predictability debate should be linked more closely to the original question of aid effectiveness. In some cases, donors are justified in being unpredictable. Laying out upfront the circumstances under which donors are not expected to be predictable—for example, in cases of major emergencies—would help implement the aid effectiveness targets of the Paris Declaration.
The analysis shows that lack of predictability hurts investment outlays, which are cut in periods of aid shortfalls but not raised during aid windfalls. By contrast, government consumption rises in response to aid windfalls. This finding is further illustrated by a review of the bilateral relationship between key variables:
A 1 percent of GDP aid shortfall is associated with a statistically significant downward adjustment of investment spending of 0.1–0.2 percent of GDP, whereas investment spending does not rise with aid windfalls.
Government consumption does not fall during aid shortfalls, but a 1 percent of GDP aid windfall is associated with a 0.6 percent of GDP rise in consumption.
Domestic bank financing is used to absorb both aid shortfalls and windfalls, but to a different degree. A 1 percent of GDP aid shortfall is associated with additional domestic bank financing of 0.5 percent of GDP. A 1 percent of GDP aid windfall is associated with a reduction of domestic financing (domestic debt repayment) that is larger (0.8 percent of GDP) than the additional bank financing during aid shortfalls.
Second, data collection should be improved to measure more accurately the impact of low predictability. It is critical to record the mutual expectations of donors and recipients alike to capture aid flows expected by recipients. Better data would help explain low predictability caused by conditionality, administrative delays, and sudden adjustments by donors.
Third, the persistence of the predictability problem, especially for budget support, would suggest reconsidering some of the mechanisms of aid delivery to these countries. One possible way is to lengthen aid allocation periods and tie them to slower-moving country indicators rather than reconsidering fast-disbursing aid volumes annually within annual conditionality frameworks (Eifert and Gelb, 2006). That would remove discretion over aid disbursements, but it would still allow donors to rapidly cut aid if policies and/or governance in a country deteriorate sharply. The implication for the international aid architecture would be important. Currently, many aid budgets are set annually, and multilateral institutions need to replenish their resources for low-income countries every three years. Longer-term commitments to budget aid—say, over a 10-year horizon—would imply that aid funding mechanisms, including for multilateral institutions, would have to be reconsidered.
Oya Celasun is an Economist in the IMF’s Research Department and Jan Walliser is a Sector Manager for Poverty Reduction and Economic Management in the World Bank’s Africa Region.