Journal Issue

Appraising the IMF’s Performance

International Monetary Fund. External Relations Dept.
Published Date:
March 2004
  • ShareShare
Show Summary Details

A review of the first three studies by the new Independent Evaluation Office

DOES THE IMF adopt a one-size-fits-all approach to fiscal policy in countries that seek its assistance, requiring retrenchment in every instance? Were the fiscal provisions of IMF programs responsible for the large output contractions in the countries beset by the Asian crisis of 1997–98? Why were those programs so slow to halt the capital outflows from those countries? Do IMF programs depress growth, and do they hurt the poor? And why have some countries been chronically dependent on IMF financing, which is supposed to be temporary? Fresh answers to these questions are now available in the first set of studies produced by the Independent Evaluation Office (IEO) of the IMF, and if they are typical of the studies it will produce in the future, the IMF and its member countries will benefit significantly. Its analyses are thorough, combining careful quantitative work, detailed country studies, and discussions with IMF staff, national officials, and others. Its conclusions are eminently sensible, and most of its recommendations should be adopted, although some of them run afoul of an intractable conflict between candor and transparency.

The IEO chose three subjects for its first year’s studies: fiscal adjustment in IMF-supported programs, the role of the IMF in three capital account crises (Indonesia, Korea, and Brazil), and the prolonged use of IMF resources.

Assessing fiscal strategy

The IEO report on fiscal adjustment poses two main questions: Have IMF programs required excessive fiscal adjustment? Has the fiscal adjustment been adequate in quality—in composition, sustainability, and incidence, especially in its impact on social spending and the vulnerable members of society?

A country obliged to improve its current account balance often has to make more resources available for the production of exports and import-competing goods and must then reduce aggregate domestic demand. For this reason, if no other, it is often necessary to tighten fiscal policy. But the amount of tightening cannot be chosen without first projecting aggregate expenditure, especially domestic investment. In some cases, moreover, a tightening is needed because of constraints on governments’ ability to borrow or the need to achieve or maintain debt sustainability.

Yet the IEO finds that the link between the fiscal and current account balances in IMF programs is weaker than might be expected. A cross-sectional analysis covering 143 recent programs does reveal a positive link between the sign of the targeted change in the primary fiscal balance and the sign of the projected change in the current account balance, but the two are quite different in size. When the current account is projected to improve by 1 percent of GDP, the fiscal balance is supposed to improve by only one-fifth as much. A far larger share of the targeted change in the fiscal balance is explained by the initial state of that balance. If a government has a big budget deficit, it is supposed to reduce it.

The same pattern emerges in Table 1, which classifies most of those same programs by the projected changes in the two balances. Although 70 percent of them call for an improvement in the fiscal balance, only two-thirds of that group call for improvement in the current account balance. The main feature of programs that call for an improvement in the fiscal balance is the initial size of the fiscal deficit. Conversely, half of the programs that contemplate a deterioration in the fiscal balance still call for an improvement in the current account balance.

Table 1Varying approaches

Not all IMF programs called for improvements in both the fiscal balance and the current account balance.

CategoryPercent of

Initial as percent of GDP
Current accountFiscal balance
Improvement in fiscal balance:
Improvement in current account43-6.6-5.7
Deterioration in current account27-1.9-4.4
Deterioration in fiscal balance:
Improvement in current account15-8.2-1.1
Deterioration in current account15-3.7-1.1
Source: IEO, Evaluation of Fiscal Adjustment, Table 2.3.
Source: IEO, Evaluation of Fiscal Adjustment, Table 2.3.

Invoking these and other findings, the IEO concludes convincingly that IMF programs do not display a one-size-fits-all strategy. Furthermore, its findings suggest that IMF programs do not depress economic growth. Table 2 presents some of the relevant evidence. Typically, GDP growth in the preprogram year was much slower than trend growth in the prior decade, save in low-income countries. And, in all but one country group, growth in the first program year was faster than in the preprogram year, and it was typically higher in the second program year. The table also shows that the exceptional case—slower growth in the first program year—was due chiefly to sharp economic contractions in the small number of countries beset by the capital account crises of the past few years, a subject discussed below.

Table 2Is growth depressed?

The IEO found that countries grew faster with an IMF program in place.


CategoryTrend In

prior decade

Program years
All programs1.
Low-income countries11.
Transition countries2-2.1-
All other countries33.
Capital account crises4.82.9-5.04.7
Other programs3.
Source: IEO, Evaluation of Fiscal Adjustment, Table 5.1.

Average for 64 programs using the Enhanced Structural Adjustment Facility or the Poverty Reduction and Growth Facility.

Average for 34 programs using a Stand-by Arrangement or the Extended Fund Facility.

Average for 61 programs using a Stand-by Arrangement or the Extended Fund Facility, of which 10 involved the capital account crises for which the outcomes are shown separately.

Source: IEO, Evaluation of Fiscal Adjustment, Table 5.1.

Average for 64 programs using the Enhanced Structural Adjustment Facility or the Poverty Reduction and Growth Facility.

Average for 34 programs using a Stand-by Arrangement or the Extended Fund Facility.

Average for 61 programs using a Stand-by Arrangement or the Extended Fund Facility, of which 10 involved the capital account crises for which the outcomes are shown separately.

The IEO evaluation also addresses the common criticism of the IMF that it is not sufficiently sensitive to the effect of fiscal adjustment on vital social programs, such as health care and education. To this end, the IEO compares the relevant levels of government spending by individual countries in years when they had IMF programs with levels of spending in other years. In the vast majority of cases, there was no significant difference, no matter how spending was measured. Where there were significant differences, moreover, the number of cases in which spending rose typically exceeded the number in which it fell. The IEO points out, however, that the vulnerable members of society are not always well protected from the effects of fiscal tightening, and it cites a distressing example:

In 1999, Ecuador suffered a macroeconomic crisis. Inflation rose to 60 percent, public sector wages rose by 34 percent, and the currency depreciated sharply. But the health budget rose by only 12 percent. Therefore, six public hospitals were visited to see how they coped with higher wage costs and the much higher costs of medical inputs caused by the depreciation. As the hospitals had to cut back patient care, outpatient care fell by more than 25 percent, the number of prescriptions dispensed dropped sharply in three hospitals, and patients’ food rations were cut in at least one hospital although staff rations stayed relatively constant.

The IEO concludes that the protection of small but critical nonwage budgetary items poses a major challenge for the design and monitoring of fiscal adjustment—and one that the IMF cannot meet by itself. Help from the World Bank is needed to design the requisite expenditure safeguards.

Design and implementation of fiscal adjustment

Although the IEO rejects the most common criticisms of the fiscal measures in IMF programs, it raises three concerns about the design and implementation of those measures.

First, excessive optimism about the recovery of output and, especially, investment tends to impart a contractionary bias to those fiscal measures. As a result, the current account balance adjusts more strongly than projected. Reviewing 52 conventional programs (those that did not involve the transition countries or the low-income countries eligible for concessional financing), the IEO finds that the projected change in the current account during the second program year was typically close to zero, yet the actual improvement was typically larger than 1 percent of GDP. And this result obtained even though the actual improvement in the fiscal balance was typically smaller than the projected improvement. Although the IMF does not require fiscal tightening in every program, it may call for too much tightening in too many programs.

Second, the IEO is dissatisfied with the rationale provided for the fiscal aspects of IMF programs. Having examined the documents pertaining to 15 programs, it finds that they often fail to justify the size and pace of the fiscal measures and do not explain how the fiscal targets mesh with the assumptions about economic recovery or with the rest of the program. These matters, it notes, are covered more thoroughly in subsequent reports on the progress of programs, but not in the initial briefs. It urges the staff to undertake more intensive brainstorming in the initial briefs—to articulate clearly the rationale for the fiscal measures, as well as the risks to programs and the revisions that might be needed if those risks materialize. In its response to the IEO, the staff points out that the key assumptions of IMF programs have to be negotiated with the governments involved and that the excessive optimism mentioned earlier may reflect the governments’ desire to achieve a rapid revival of confidence and to generate domestic support for the programs. When that is true, however, the viability of the programs may well be at risk.

These two concerns, taken together, are reinforced by another finding. When programs are reviewed for the first time, most of the revisions in the fiscal measure are minor, and the few major revisions are distributed rather evenly between upward and downward changes in the fiscal targets. But when programs are reviewed for the second time, more fiscal targets have to be relaxed than tightened, suggesting that the initial fiscal targets were too demanding.

Finally, the IEO criticizes the sorts of fiscal measures found in many programs. They stress the revenue side more than the expenditure side, and the revenue measures themselves are too narrowly focused. On the revenue side, the IEO says, the IMF rightly views the value-added tax as the cornerstone of a modern tax system, but it has not paid enough attention to income and property taxes and to the need to combat tax evasion. On the expenditure side, the IEO says, the IMF has focused too heavily on cutting public employment or capping public sector wages, measures that tend to be short lived because they are easily reversed.

The IEO concedes that these defects reflect an unavoidable mismatch of time frames; IMF programs do not last long enough to see deeper reforms to completion. Therefore, the IEO rightly urges the closer integration of programs and surveillance. Programs should exploit the previous findings of surveillance, and surveillance should monitor the completion of reforms introduced by programs.

The IMF and recent capital account crises

Concerns about fiscal adjustment in IMF programs also appear in the IEO study of three capital account crises: the Indonesian and Korean crises of 1997–98 and the Brazilian crisis of 1998–99. In the first two crises, it says, the IMF called for too much fiscal tightening because it failed to foresee the collapse of investment and resulting fall in output. In the Brazilian crisis, it says, the IMF called for too little tightening, given the need to achieve debt sustainability. Like IMF staff studies of these crises (the 1999 account of the Asian crisis by Timothy Lane and others, and the 2002 study of capital account crises by Atish Ghosh and others), the IEO blames the collapse of investment in Indonesia and Korea on the balance sheet effects of the currency depreciations, effects that were not foreseen at the time. Like those staff studies, moreover, the IEO notes that the IMF modified its fiscal targets quickly and thus absolves it of blame for the large fall in output. In Brazil, by contrast, the IMF went in the opposite direction. It called for more fiscal tightening after Brazil had to let its currency float and the subsequent depreciation raised the government’s debt burden, as much of its debt was dollar denominated or dollar linked.

The IEO is less firmly supportive of the monetary policies prescribed by the IMF, apart from noting that Indonesia and Brazil failed to follow those policies during the early stages of their crises. Recent research, it says, has not dispelled doubts about the net benefit of raising interest rates when a country confronts a currency crisis. Theoretical work has shown that higher interest rates can strengthen or weaken a currency, empirical work has not settled the matter, and the issue is even harder to resolve when a country faces a banking crisis as well as a currency crisis. In light of these considerations, the IEO concludes, it is difficult to pronounce definitively on the appropriateness of monetary conditionality in the three crisis countries.

“The IEO should contemplate follow-up studies to ask whether its recommendations have had the expected effect.”

The IEO is less circumspect when assessing the results of the IMF programs. The initial programs failed in that they did not stabilize the three countries’ currencies. In Indonesia, the IEO blames the government for refusing to take ownership of the program and, indeed, subverting it. In Korea, it blames the IMF’s major shareholders for attaching vague conditions to the use of the bilateral financing they provided and for taking too long to involve Korea’s private sector creditors. In Brazil, it divides blame between the government and the IMF for trying to defend an overvalued currency and failing to address the government’s adverse debt dynamics. But it also faults the IMF for failing to detect some of the vulnerabilities that explain the severity of the Korean and Indonesian crises and for failing to flag the vulnerabilities it did, in fact, detect.

The detailed accounts of the crises in the IEO evaluation are well worth reading for the light they shed on relations between the IMF and its major shareholders. It is widely believed that the IMF and the U.S. Treasury worked closely together to resolve the crises under study. There were marked differences of view, however, between the IMF and its major shareholders. In Korea, the IMF quickly concluded that the available financing was insufficient and immediately pressed its major shareholder governments to achieve a rollover of bank credit lines, but to no avail. In Indonesia, a long list of structural reforms was attached to the January 1998 program at the urging of major shareholders in the belief that confidence could be restored only by signaling a clean break with the past. The IEO finds that this strategy deflected attention from vital reforms in the banking sector and reduced the willingness of the Suharto government to take ownership of the program.

The IMF has already addressed some of the issues raised in the IEO evaluation—a fact that the IEO readily acknowledges. The IMF has decided, for example, that programs should not include structural reforms unless they are required to achieve the macroeconomic objectives of the programs. We should not expect to see again long lists of structural reforms like those in the Indonesian and Korean programs. It may be harder, however, to implement some of the IEO’s other recommendations, notably those that pertain to IMF surveillance. At times, the IEO suggests that candor and transparency are complementary—that the publication of the IMF’s views can stimulate public debate about national policies and bring market pressure to bear on a recalcitrant government. Elsewhere, it concedes that they can conflict. When candor is combined with transparency, markets may react abruptly—faster than governments can react constructively.

In an effort to marry candor and transparency, the IEO suggests that the IMF find ways of engaging in escalated signaling when, in the course of surveillance, the staff has identified key vulnerabilities and they have not been cured after several more rounds of surveillance. This approach, it says, would strike a balance between the role of the IMF as confidential advisor to governments and its role in providing information to markets. But this approach assumes implicitly that markets will respond in a graduated way to the escalating signals from the IMF, and there is little reason to expect that outcome. It may thus be better to rely on confidential warnings that a government will not qualify for large-scale IMF financing if, by failing to cure key vulnerabilities, it succumbs eventually to a capital account crisis.

What should the staff tell the Board, however, when the risk to an IMF program does not derive from unforeseen shocks or perceived vulnerabilities but is instead political—the risk that a government will be unwilling or unable to implement an IMF program? The IEO tackled this question in its report on prolonged use.

Prolonged use of IMF resources

The IEO treats a country as being a prolonged user if it has had IMF programs for 7 or more years in any 10-year period. Under this definition, a country is not likely to be a prolonged user unless it has had at least two 3-year programs within a 10-year period. Nevertheless, the IEO finds that prolonged use has risen since the 1970s, whether measured by the number of countries involved or the total financial exposure of the IMF. To a significant extent, prolonged use reflects the adaptation of IMF policies and financial facilities to meet the special needs of low-income countries. Most prolonged users belong to that group. But much of the increase in financial exposure resulting from prolonged use reflects the involvement of middle-income countries, such as Argentina, Mexico, and Turkey. The IEO finds, moreover, that both sorts of prolonged use are due partly to flaws in IMF policies and programs.

Though the findings in this report derive from studying prolonged use, some have wider applicability. The IEO’s case-studies of prolonged use, for example, yield valuable findings about the design of conditionality:

• The specific structure of conditionality is less important than an underlying domestic political commitment to core policy adjustment.

• Excessively detailed conditionality—whether used because of a weak track record, doubts about ownership, or to support reform-minded groups within a government—does not appear to have been effective.

• Conditionality focused on policy rules or procedures, rather than discretionary onetime actions, was ultimately more effective.

The IEO also discusses the political economy of conditionality and the problem raised above: How and where should political feasibility enter into the IMF’s decision-making processes? The IEO finds that judgments about political feasibility influence IMF mission chiefs working with national governments on the design of programs and that staff members often understand the political risks to a program. It also finds, however, that these judgments do not surface in staff reports to the Board. But its conclusion suggests a different solution to this crucial problem. Judgments about political risks should be clearly distinguished from judgments about other risks and should be made in a transparent manner at the level of the Managing Director and the Executive Board, who are accountable for them. But this surely implies that the staff should be strongly encouraged to advise the Managing Director, if not the Board itself, of the concerns that it may have about the political risks to a particular program.

Returning to the narrower problem of prolonged use, the IEO finds that aid donors and others, including the Paris Club of official creditors, often key their own decisions to those of the IMF—that is, an IMF program is seen as a seal of approval. This practice, it says, may lead the IMF to tolerate prolonged use, and it should therefore develop other ways to signal its approval of a country’s policies, including more frequent use of enhanced surveillance, shadow programs, and precautionary arrangements.

Unfortunately, the IEO does not explore sufficiently another explanation for prolonged use. The IMF may tolerate prolonged use because it is loath to face the obvious consequence of refusing to approve an additional program—the failure of a large middle-income country to repay what it already owes. When tracing the involvement of the IMF in concessional lending to the low-income countries, IMF historian James Boughton notes that it began when the IMF started to worry about the ability of those countries to repay concessional loans from the trust fund set up in the wake of the first oil crisis. Similar concerns may help to explain the recent growth of prolonged use by some middle-income countries. It is hard to believe that those concerns did not influence the decision to approve a new program for Argentina in 2003.

Looking ahead

Reflecting on the IEO’s work to date, the topics of its next three studies, and the published list of topics on which it may work thereafter, I offer two general observations:

The IEO’s initial studies dealt with a large subject, although two of them also contained detailed country studies. As a result, some of its recommendations have been cast too broadly to elicit firm responses from the Executive Board. Almost all of the recommendations made in the study of prolonged use are applicable generally to the work of the IMF and have thus far elicited rather vague responses. The IEO has made operational recommendations, one of which the Board rejected—the proposal that prolonged users be charged higher interest rates. Yet concrete recommendations may stand the best chance of producing focused debate in the Board and, when they find favor, subsequent implementation. The Board, for example, has agreed to adopt an operational definition of prolonged use and to use that definition to assess the effectiveness of IMF programs involving prolonged use.

Broad subjects, moreover, tend to spawn long documents, and the first three from the IEO were rather long indeed. To be sure, they were well organized, with executive summaries at the start and technical annexes at the end. Nevertheless, they may be too long to serve effectively one of the main purposes of the IEO—promoting wider understanding of the IMF. Shorter studies of narrower subjects might be read more widely.

Looking further ahead, the IEO should contemplate follow-up studies to ask whether its recommendations have had the expected effect. It will not run out of new subjects to study but should not wait long to revisit old ones.

Peter B. Kenen is Walker Professor of Economics and International Finance at Princeton University

Statement of Ownership, Management, and Circulation required by 39 USC 3685.

1. Title: Finance & Development. 2. Publication No. 123-250. 3. Date of filing: December 12, 2003. 4. Frequency: Quarterly. 5. Number of issues published annually: four. 6. Annual subscription price: NA. 7/8. Complete mailing address of known office of publication/publisher: Finance & Development, International Monetary Fund, 700 19th Street, N.W., Washington, DC 20431. 9. Full names and complete mailing address of the headquarters of general business offices of the publisher and editor: International Monetary Fund, Washington, DC 20431; Editor-in-Chief: Laura Wallace, same address. 10. Owner: International Monetary Fund. 700 19th Street, N.W., Washington, DC 20431. 11. Known bondholders, mortgagees, and other security holders owning or holding 1 percent or more of the total amount of bonds, mortgages, or other securities: None. 12. Tax status: has not changed during preceding 12 months. 13. Publication title: Finance & Development. 14. issue date for circulation data below: December 2003.

15. Extent and nature of circulationAverage no. of copies each

issue in preceding 12 months
Actual no. of copies of single issue

published nearest to filing date

(December 2003)
A. Total number of copies52,62552,200
B. Paid and/or requested circulation43,12042,675
C. Total paid and/or requested circulation43,12042,675
D. Free distribution outside the mail (carrier or other means)6,3206,409
E. Total distribution (sum of C and D)49,44049,084
F. Copies not distributed3,1873,116
G. Total (sum of E and F)52,62552,200
Percent paid and/or requested circulation87.2286.94
I certify that the statements made by me above are correct and complete.Laura Wallace, Editor-in-Chief
I certify that the statements made by me above are correct and complete.Laura Wallace, Editor-in-Chief

    BoughtonJames M.2001Silent Revolution: The International Monetary Fund 1979–1989 (Washington: International Monetary Fund)

    GhoshAtish and others2002IMF-Supported Programs in Capital Account Crises Occasional Paper 210 (Washington: International Monetary Fund).

    Independent Evaluation Office2002Evaluation of Prolonged Use of IMF Resources (Washington: International Monetary Fund).

    Independent Evaluation Office2003The IMF and Recent Capital Account Crises: Indonesia Korea Brazil (WashingtonInternational Monetary Fund).

    Independent Evaluation Office Fiscal Adjustment in IMF-Supported Programs (Washington: International Monetary Fund).

    LaneTimothy and others1999IMF-Supported Programs in Indonesia Korea and Thailand: A Preliminary Assessment Occasional Paper 178 (Washington: International Monetary Fund).

Other Resources Citing This Publication