Chapter 4 Program Design and Implementation

Kevin Barnes, Ali Mansoor, Benjamin Cohen, and Shinji Takagi
Published Date:
September 2003
  • ShareShare
Show Summary Details

Given the nature of capital account crises, the primary objective of crisis-management programs in such cases should be to restore confidence as quickly as possible in order to restore normalcy to the capital account. This was indeed the approach adopted in all three cases. In each case, the crisis-management strategy relied upon a mix of fiscal and monetary policies combined with a range of structural reform measures, supported by a large financing package. In this chapter, we present a summary assessment of the critical elements of program design and implementation in the three country cases.

Macroeconomic Framework and Projections

Adjustment programs are designed to achieve particular macroeconomic outcomes, and several policy measures are calibrated around these outcomes. However, the key determinants of macroeconomic outcomes are not always well understood and are in any case subject to large uncertainty. This can lead to macroeconomic outcomes that are very different from program projections. This was evident in both Indonesia and Korea, where the initial projections were overly optimistic, leading to the design of macroeconomic policies that turned out to be tighter than necessary (Table 4.1).1 In contrast, the initial projections for Brazil in 1999 were too pessimistic, which contributed to fiscal adjustments that turned out to be insufficient, in light of that country’s adverse public debt dynamics.

Table 4.1.Real GDP and Investment Projections and Outturn in Crisis Countries


Indonesia (1998/99)
Fixed investment–0.4–26.8–33.0
Korea (1998)
Fixed investment–14.2–21.2
Brazil (1999)
Fixed investment–9.518.2–3.2
Sources: Various IMF staff reports.

March 1999 for Brazil, April 1998 for Indonesia.

Sources: Various IMF staff reports.

March 1999 for Brazil, April 1998 for Indonesia.

In Indonesia, the November 1997 program projected GDP growth in 1998/99 at 3 percent. This was then revised downward to zero percent in January 1998 and to –5 percent in April, while the actual outcome was even worse at –13 percent. The original optimism was due to the assumption that the crisis was a moderate case of contagion in which the exchange rate had overshot. It was thought that, with a combination of tight macroeconomic policies and structural reform, the exchange rate would appreciate quickly. This did not happen, and the resulting currency collapse had severe negative effects on the balance sheets of corporations and banks. Such negative balance sheet feedback was further exacerbated by the political developments affecting the minority Chinese community, which had a dominant role in business. Fixed investment in Indonesia, which was expected to decline by only 0.4 percent in 1998/99 in the November program projection, actually declined by a massive 33 percent, explaining much of the turnaround in GDP performance.

In Korea, the IMF was of the view that the macroeconomic outcome would be worse than projected, but the government was reluctant to accept a lower figure for GDP growth. Growth in 1998 was therefore projected at 2.5 percent in the initial program, whereas it actually declined by 6.7 percent. Investment, which was projected to decline by 14.2 percent, actually fell by 21.2 percent, again indicating that the negative balance sheet impact was underestimated.

In the case of Brazil, the IMF staff correctly identified a number of the elements that proved critical in the country’s relatively strong growth performance after the exit from the exchange rate peg, such as a relatively strong financial sector, and a corporate sector with limited leverage and little foreign exchange exposure. In part reacting to the overoptimistic projections in East Asia, the projections for output were deliberately cautious, although in line with outside forecasts and considered by some to be on the optimistic side. It was felt that this would help persuade the markets that the targeted path of the primary surplus was consistent with sustainable debt dynamics even under relatively adverse developments in output.

Part of the problem arises because macroeconomic projections in an IMF-supported program are necessarily the outcome of a negotiation. In the case of Korea, the authorities were reluctant to accept a growth projection lower than 2.5 percent for 1998; in Brazil, the authorities deliberately wanted to be cautious. More important, forecasts were not derived from an analytical framework in which the key determinants of output and their likely behavior during the crisis could be dealt with adequately. In particular, there was insufficient appreciation of (1) the large currency depreciation which might occur in view of the possibility of multiple equilibria and (2) the severe balance sheet effects that might result, which would affect macroeconomic outcomes adversely. In retrospect, these can be called analytical weaknesses in light of the new type of crises. Balance sheet analysis was not yet in the tool kit of most macroeconomists in the economics profession, let alone in the IMF, at the time.2


In both Indonesia and, to a lesser extent, Korea, much attention has focused on whether the initial stance of fiscal policy was appropriate in view of the output collapse that subsequently occurred. Fiscal tightening was said to have been unnecessary and have damaged market confidence when output was beginning to fall, and we turn to this issue in the next section. However, this was the direct consequence of the overoptimistic projection of output for the reasons indicated above. Thus, the key questions in this respect are: (1) were the initial macroeconomic projections a good guide for judgments on the fiscal policy stance? (the answer is no in the case of Indonesia and Korea); and (2) was program design sufficiently flexible to respond reasonably quickly to a different macroeconomic situation? (in our view, the answer, as discussed further in the next section, is a qualified yes. However, the flexibility was not sufficiently transparent and gave mixed signals, especially in Indonesia). These problems did not arise in Brazil because the projections were deliberately pessimistic and the outcomes were actually better, which was probably less damaging to market confidence. However, routinely making pessimistic projections cannot be the answer, not least because the markets would then quickly learn to discount the pessimistic bias in IMF projections.

Growth projections that are overoptimistic not only call into question the credibility of the IMF, but they can also lead to macroeconomic policies that are either too tight or too loose. It is inherently difficult to forecast macroeconomic outcomes reliably, most of all in crisis situations. However, these problems could be reduced if there was a more explicit focus on the key factors that will have significant impact on aggregate demand, particularly private investment. It is well known that forecasting private investment over a business cycle is extremely difficult even under normal conditions. This difficulty is compounded by greater uncertainty during a capital account crisis, making accurate projections difficult even with best practice. It is thus important that quantitative targets and benchmarks in an IMF-supported program should incorporate that uncertainty. In particular, a more explicit discussion was needed in the program documents of the major risks to the macroeconomic framework, with a clear indication of how policies would respond if the risks materialized. This could have helped facilitate subsequent program reviews (which did show flexibility) and would also have sent a more transparent signal on the expected stance of policies.

Fiscal Policy

Some critics have accused the IMF of mechanically applying to East Asia the tight fiscal policies that it had traditionally recommended in Latin America. The three countries studied suggest that the approach adopted was more nuanced. In both Indonesia and Korea, the staff recognized that the underlying fiscal position was sound, and the fiscal tightening envisaged was therefore mild. The November 1997 program in Indonesia targeted an increase in the fiscal surplus from 0.5 percent in the budget for 1997/98 to 0.75 percent, with a further tightening to yield a surplus of 1.3 percent in 1998/99. The initial program therefore involved a turnaround of 0.8 percent of GDP over an 18-month period. For Korea, the program incorporated only a small fiscal surplus of 0.2 percent of GDP for 1998, compared with a deficit of 0.2 percent of GDP projected for 1997, that is, a fiscal turnaround of only 0.4 percent of GDP. In sharp contrast, the Brazilian program involved a turnaround of over 4 percentage points of GDP for 1999, relative to the fiscal position expected to prevail in the absence of adjustment measures.

The IMF staff justified the mild tightening of fiscal policy in Indonesia and Korea on the grounds that countervailing measures were needed to lessen the burden of the private sector in external adjustment and to cover the carrying cost of the public-debt burden arising from recapitalizing the financial sector. Moreover, fiscal tightening has traditionally served as a signaling device, indicating the government’s resolve to take corrective action. The signaling role was particularly pertinent in Indonesia, where the tightening largely reflected the elimination or postponement of prestige projects linked to the family of the President. The need for a fiscal correction to cover the cost of bank restructuring cannot be disputed, because the potential quasi-fiscal costs of the banking crisis were very high. Nevertheless, with the benefit of hindsight, it can be argued that, certainly in Korea, this adjustment could have been deferred by accepting a slightly higher public debt profile in the medium term, which would not have been a problem given the relatively low initial debt position. There was less justification for deferring the adjustment in Indonesia, where the cost of bank restructuring was higher.

The real problem with the fiscal targets in Indonesia and Korea was the growth assumptions built into the program, which proved unrealistic because of the contractionary forces generated by the sharp exchange rate depreciation and the resulting balance sheet effects. In Indonesia, these were compounded by a developing political crisis. Failure to take these influences sufficiently into account led to unnecessary fiscal tightening. Better anticipation on this count would have called for a more countercyclical stance in fiscal policy.

The fiscal targets in both countries were quickly adjusted as the contractionary effects became evident.

  • In the case of Indonesia, the January 1988 LOI relaxed the fiscal policy target from the surplus of 1.3 percent of GDP initially envisaged to a deficit of 1 percent for 1998/99, and this was further relaxed in April (at the start of the fiscal year) to a deficit of 4.7 percent, on the assumption that GDP would decline by 5 percent. The actual deficit achieved in 1998/99 was only 2.1 percent of GDP, indicating that the fiscal target was not a binding constraint. The lack of automatic stabilizers, such as social safety nets, and the weak capacity of the government to achieve the increases in expenditure that were targeted in a number of social sectors made it difficult to use fiscal policy countercyclically even within the limit permitted by the revised program.

  • In Korea, as early as late December 1997, within a month of the approval of the program, the staff recommended that the authorities should not adhere to the fiscal targets but let automatic stabilizers work. However, the Korean authorities were reluctant to deviate from their balanced budget philosophy despite urging from the IMF staff, who favored a more expansionary fiscal policy once the extent of the economic downturn became apparent. In the event, government consumption expenditures fell by 0.4 percent in real terms in 1998, but Korea ended up running a budget deficit of 4.3 percent of GDP in 1998, because tax revenues fell even further.

Fiscal policy was much more restrictive in Brazil, where the fiscal adjustment of over 4 percent was programmed for 1999 relative to the outcome projected to prevail in the absence of adjustment measures.3 This was appropriate, as fiscal sustainability was a factor driving the evolution of the crisis. The main objective of the 1998 program was to stabilize the ratio of net public debt to GDP, in order to ensure medium-term debt sustainability. To achieve this, a performance criterion was set for the public sector borrowing requirement (PSBR), with an indicative target for the primary surplus that involved an increase of 2.5 percentage points over the previous year. The depreciation of the real following the collapse of the program in early 1999 raised the debt-to-GDP ratio from 43 percent at the end of 1998 to 52 percent in February 1999 because of the revaluation of external debt and high levels of foreign exchange–indexed domestic debt.

The revised March 1999 program set a performance criterion on the primary surplus, with an indicative target for the net debt of the public sector, and an informal target for the proportion of domestic debt indexed to the U.S. dollar that would be rolled over. Moreover, it raised the primary surplus to 3.1 percent of GDP in 1999, 3.25 percent in 2000, and 3.35 percent in 2001. While all the primary balance targets were achieved, the targeted debt-to-GDP ratios were not achieved, in large part owing to the greater-than-expected depreciation of the currency, which raised the local currency value of external and foreign currency–linked domestic debt.


The three country experiences studied for the report suggest that the fiscal policies recommended by the IMF did differ depending on the initial position, but the real reason for the inappropriateness of the fiscal policy in Indonesia and Korea was the failure to take account of the key factors that would affect aggregate demand during a crisis, notably the impact of balance sheet effects and confidence factors on private investment. The fiscal stance in Korea, given the low initial stock of public debt, can be said in retrospect to be too contractionary. The government could have drawn on its spare borrowing capacity to offer its obligations in exchange for those of the troubled financial sector—as eventually happened. In contrast, the similarly low outstanding stock of debt in Indonesia probably did not present a strong case for an ambitious countercyclical fiscal policy because the banking sector was much weaker than in Korea, with serious solvency rather than mainly liquidity problems, and posed large contingent liabilities for the government. The absence of a bond market also limited the ability of the government to finance expenditures without resorting to inflationary means. There was little scope for a substantially expansionary fiscal policy.

The Indonesian and Korean programs have been criticized for pursuing tight fiscal policy in Indonesia and Korea, on the grounds that this was unnecessary and may have been partly responsible for the severe output contraction that followed (Furman and Stiglitz, 1998; Sachs, 1998). Our evaluation suggests that, while the initial fiscal tightening may have been misguided, the severe output contraction experienced by these countries was not due to the fiscal stance but to the operation of other contractionary forces, linked to the impact of balance sheet effects and confidence factors on private aggregate demand, which were clearly underestimated.

The fiscal correction in the Brazilian program was much stronger, but this was appropriate under the circumstances, since fiscal weakness and debt sustainability were critical issues driving the evolution of the crisis. A balance sheet perspective, however, suggests a weakness in another area of the program. In Brazil, from late 1997, the government was effectively providing the private sector with a hedge for exchange rate risk by issuing foreign currency– linked debt, intervening in the foreign exchange futures market and, latterly, by selling foreign exchange reserves. While the exchange rate policy maintained in the 1998 program thus helped mitigate any adverse balance sheet impact of exchange rate depreciation, it was a form of expansionary fiscal policy in the face of an impending currency crisis. Unlike the case of Korea, however, this policy had serious consequences for Brazil’s medium-term debt sustainability.

Monetary Policy

Some of the strongest criticisms of the role of the IMF in the capital account crises of the 1990s have been in the field of monetary policy. The IMF has been criticized for requiring countries to pursue an excessively tight monetary policy, thereby damaging the balance sheets of banks and corporations, disrupting the flow of credit to small and medium-sized enterprises, and constraining aggregate demand unduly at a time of recession (Furman and Stiglitz, 1998; Sachs, 1998). The IMF and its defenders have responded that a tight monetary policy was necessary in the crisis countries in order to support the exchange rate (at least in part through a signaling effect), combat inflationary pressure from depreciation, and limit the external financing gap through a combination of reduced capital outflows and a lower current account deficit (Lane and others, 1999; Corsetti and others, 1999).

Internal documents reveal that, in all three cases, monetary policy targets were set on the basis of an explicit consideration of the trade-off between higher interest rates and a weaker exchange rate. The cases differed, however, in the emphasis placed on monetary policy in program strategy and the perceived impact of high interest rates on the private financial and nonfinancial sectors (see Table 4.2 for the comparative level of real interest rates in these countries).

Table 4.2.Real Interest Rates in Selected Countries1
CountryAverageHighMonth of

January 1990–June 2002 (except where indicated)
United States1.93.7Nov. 97
United Kingdom3.88Aug. 92
Japan1.13.6Aug. 91
Italy4.613.6Sep. 92
Germany2.87.7Aug. 90
France4.29.8Jan. 93
Canada3.79.3Apr. 90
Sweden24.615.2Sep. 92
Indonesia4.949.1Aug. 97
Korea6.218.1Jan. 98
Brazil31840.6May. 95
Philippines5.517Oct. 97
Malaysia2.68.6Jul. 97
Thailand3.915Sep. 97
Mexico5.529.6Mar. 95
In the first six months after the adoption of an IMF-supported program4
Mexico1/95–6/9511.529.6Mar. 95
Philippines7/97–12/979.417Oct. 97
Thailand8/97–1/988.315Sep. 97
Indonesia11/97–4/98–8.40.5Jan. 98
Korea12/97–5/9814.818.1Jan. 98
Brazil11/98–4/9933.737.5Mar. 99
Source: IMF database.

Interest rates are 3-month treasury bill rates for G-7 (except for Japan) and Sweden; 60-day government securities rate for Japan; 3-month interbank rates for the Philippines, Malaysia, Korea, and Thailand; overnight interbank rate for Indonesia; overnight Selic rate for Brazil; and Cetes 90-day rate for Mexico. Real interest rates are calculated as the difference between the average daily nominal interest rate during a given month and the rolling 12-month CPI inflation rate centered on that month.

Until December 2001.

From January 1995.

For each country, the starting month of the program is the month in which the letter of intent was signed by the authorities. For the Philippines, this represented the extension and augmentation of an existing arrangement.

Source: IMF database.

Interest rates are 3-month treasury bill rates for G-7 (except for Japan) and Sweden; 60-day government securities rate for Japan; 3-month interbank rates for the Philippines, Malaysia, Korea, and Thailand; overnight interbank rate for Indonesia; overnight Selic rate for Brazil; and Cetes 90-day rate for Mexico. Real interest rates are calculated as the difference between the average daily nominal interest rate during a given month and the rolling 12-month CPI inflation rate centered on that month.

Until December 2001.

From January 1995.

For each country, the starting month of the program is the month in which the letter of intent was signed by the authorities. For the Philippines, this represented the extension and augmentation of an existing arrangement.

In Indonesia, the November 1997 program did not call for a substantial monetary tightening, mainly because monetary policy had already been tightened prior to the program. Internal documents and staff interviews make clear that there were considerable differences of view on this issue within the IMF, with some arguing for a further tightening of monetary policy, and some arguing that the initial tightening was sufficient to send the necessary signal, taking into account the potential impact on leveraged balance sheets. In the event, and given the political constraints faced by the authorities, the strategy adopted in the program was to maintain the relatively tight monetary stance, with the understanding that it would be tightened further if necessary. No explicit target was specified for interest rates. To allow the authorities to intervene in the foreign exchange market without affecting the overall liquidity position, the November program had the unusual feature of including a base money target as a performance criterion, instead of a more conventional NDA ceiling combined with a floor for net international reserves (NIR).

In practice, the monetary policy envisaged in the program was never implemented. A significant loosening of monetary policy took place almost immediately, with extensive unsterilized liquidity assistance to troubled banks, leading to increasingly negative real interest rates. The IMF staff objected strenuously to this loosening of monetary policy, with little effect. While this calls into question the quality of the IMF’s dialogue with the government, it cannot be said that the overall stance of monetary policy was tight through the early months of the program.4 Monetary control and exchange rate stability were only reestablished after March 1998 when a sharp interest rate increase was specified under the new program, base money targets were replaced by NDA targets as performance criteria, and the new cabinet acted decisively to end the central bank’s liberal liquidity support to the financial sector. At that stage the rupiah had depreciated to Rp 10,000 per U.S. dollar, arguably a sufficiently overshot level at which the restoration of monetary control was likely to yield the results that it did in terms of exchange rate stability. Although the economy undoubtedly suffered enormous damage in November and December 1997, the blame cannot be put on the tight monetary policy advocated by the IMF since this was not implemented.

The Korean experience with monetary policy is very different. In this case, a substantial increase in the central bank’s main policy rate was a key component of the IMF-supported program approved in early December 1997. Despite initial resistance by the authorities, significant increases in interest rates were implemented, though with a delay at one point because of the need to repeal an interest ceiling set by an anti-usury law. A penalty rate was also set on central bank advances of foreign exchange to the banking sector. While the monetary targets included an NDA ceiling, it was the specification of interest rate increases that had the central role to play in the Korean program. An inflation target was also included, but it was not part of formal conditionality.

The application of higher interest rates did not initially produce the desired results in terms of halting the capital outflow and easing pressure on the exchange rate. Foreign banks continued to reduce credit lines to Korean institutions and the exchange rate remained weak and volatile. The authorities expressed concerns at this time about the impact of high interest rates on heavily indebted corporations and, through them, on the banking sector, but the IMF staff assigned a higher priority to the immediate need to stabilize the exchange rate. In the months after the revised program was adopted in late December 1997, the policy rate was slowly but steadily lowered, as currency market conditions stabilized and inflation proved quiescent.

In retrospect, it would appear that, while high rates were necessary in December 1997 to prevent a complete collapse of the exchange rate, they were certainly not sufficient to resolve the crisis, as stability did not begin to be restored until after the rollover agreement was reached. Hindsight also suggests that, in the early months of 1998, interest rates were maintained too long at high levels, at a time when corporate sector balance sheets were fragile and a looser policy might have supported a faster recovery in domestic demand. However, the period of time when real interest rates may have been higher than they needed to be was at most a few months, and it is difficult to believe that this delay contributed significantly to the recession. Besides, the speed with which markets stabilized in early 1998 came as a surprise, and some caution was therefore understandable, given the unsettled market situation in East Asia and the need to ensure that price and exchange rate stability would not be put at risk from lower interest rates.

In Brazil, the December 1998 program prescribed a tight monetary policy to support the crawling peg regime, but the prescribed policy was not followed initially. Instead, interest rates were reduced toward the end of 1998—excessively and prematurely in the view of the staff—and the programmed target for central bank credit was substantially exceeded. This is not to say that pursuit of the prescribed policy would have succeeded in maintaining a peg that was widely seen to be overvalued.

Interest rates were increased again after the exchange rate peg was abandoned in early 1999—tentatively at first but later more decisively—in an effort to stabilize the exchange rate and prevent the exchange rate depreciation from sparking reindexation and a takeoff in inflation. As in Korea, rates were eventually brought down again (though at a somewhat quicker pace) as it became evident that the exchange rate had stabilized and the pass-through to inflation was modest. In contrast to Korea, the impact of high interest rates on investment through their effect on corporate balance sheets turned out to be limited, because of the low degree of leverage in the corporate sector. However, the public sector, which had issued increasing amounts of floating rate debt, was exposed to an excessive degree of interest rate risk.

The contrasting cases of Korea and Brazil point to the importance of having a clear framework to guide monetary policy in the poststabilization period. In Korea, the high interest rate policy was subject to public criticism in early 1998 because the criteria for maintaining it—exchange rate and price stability—were not clearly defined. In Brazil, by contrast, the guiding principles of monetary policy were clearly communicated by the Central Bank. Once the formal inflation targeting framework was put in place, it provided a measurable benchmark that could be used both to guide monetary policy and to explain it to the market and to public opinion. These experiences illustrate the value of straightforward, publicly stated frameworks guiding the return to a less restrictive monetary stance in helping to clarify expectations and improve public acceptance.


Most economic policymakers at the time of the 1997–99 crises accepted the existence of a positive link between interest rates and exchange rates. This approach conformed to the practice in other countries that faced currency crises in the 1990s, notably those affected by the European exchange rate mechanism (ERM) crisis of 1992. During the Asian crisis, economies with IMF-supported programs, such as Indonesia, Korea, the Philippines, and Thailand, and those without IMF-supported programs, such as Malaysia and Taiwan Province of China, used high interest rates to try to reduce downward pressure on their currencies. Interest rates in Hong Kong SAR rose sharply on several occasions in 1997 and 1998, owing to both deliberate policy actions and the automatic provisions of its currency board arrangement.5

Since the Asian crisis, a large theoretical and empirical literature has reexamined the question as to when, and under what conditions, high interest rates can be effective in defending the exchange rate. Theoretical work has tended to show that effects in both directions are plausible.6 Empirical research has been unable to settle the matter.7 However, researchers have established that the relevant issues and relationships differ depending on whether one is defending an exchange rate in the midst of a crisis, or attempting to manage real appreciation in the aftermath of an episode where the exchange rate has overshot its equilibrium level. If it is judged that there has been an excessive real depreciation, one function of monetary policy is to ensure that the subsequent real appreciation occurs through nominal appreciation rather than through inflation (Goldfajn and Gupta, 1999). This would argue for maintaining a tight monetary policy. Yet the resolution of a crisis in the financial sector would call for a loose monetary policy.

This highlights the fact that interest rate policy poses special problems in situations of “twin crises,” in which a balance of payments crisis triggered by capital outflows takes place simultaneously with a banking crisis. As Krueger (2002) put it: “To confront a balance of payments crisis, the appropriate policy responses entail an exchange rate change, tightening of monetary policy, and tightened fiscal policy. To stem a financial crisis, by contrast, entails loosening of monetary policy, maintenance (or even appreciation) of the nominal exchange rate, and financial restructuring…. To a significant degree, in the presence of twin crises, whatever is done to address one will, in the short run, make the other worse.” [parentheses in original]. In the light of these considerations, it is difficult to pronounce definitively on the appropriateness of monetary conditionality in the three crisis countries. The IMF was aware that tight monetary policy designed to stabilize exchange rates could have an adverse impact on the corporate and banking sectors, if they were highly leveraged. However, it was also concerned about the adverse impact on the economy of excessive exchange rate depreciation if the corporate sector had a large unhedged debt position in foreign currency. In a twin crisis, it remains an unresolved issue how to reconcile the two conflicting objectives of monetary policy.

Official Financing and Private Sector Involvement

The size of financing needed in a capital account crisis is inherently difficult to determine for two reasons. First, the ex ante estimate of the financing gap depends upon the speed with which confidence is restored and capital flows return to normalcy, which is difficult to predict. Confidence is a psychological phenomenon and depends on both the technical soundness of the adjustment program and also on whether the markets believe it will be implemented and be effective. Second, the financing requirement in a capital account crisis is typically very large, exceeding what the IMF can provide from its own resources, given the role of quotas in limiting access and also the constraints on total resources available to the IMF. Fischer (1999) has pointed out that the IMF, therefore, has to perform two functions: to act as a “crisis lender” providing financing from its own resources, and also to act as a “crisis manager” arranging supplementary resources from other sources, for example, multilateral and bilateral official financing, and encouraging private sector involvement to the extent possible. This is indeed the approach it adopted in all three cases.

The scale of IMF financing

In all three cases, the IMF was able to provide a large volume of its own financing combined with a substantial recourse to official financing from other international financial institutions (IFIs) and bilateral sources (Table 4.3). The scale of total official financing in each case was comparable in terms of GDP to the financing provided to Mexico in 1994. All three programs involved highly front-loaded disbursements, reflecting the need to make resources available quickly.8 As a proportion of quota, IMF assistance to Korea was exceptionally large, made possible by the introduction of the SRF at that time. Nevertheless, all three programs failed to restore confidence initially.9

Table 4.3.Official Financing Assumed Initial IMF-Supported Programs(In millions of U.S. dollars)
Date of

IMFWorld Bank

and Other

IndonesiaNovember 199710,0838,00018,000136,083
KoreaDecember 199720,99014,20023,100258,290
BrazilDecember 199818,2629,00014,538341,800
Memorandum item: MexicoFebruary 199517,843033,95751,800
Source: Ghosh and others (2002).

Not included in the financing assumptions.

Including US$20 billion in the second line of defense, which was included in the press release, but was not part of the programmed package.

BIS-coordinated bilateral financing and Japanese assistance.

Source: Ghosh and others (2002).

Not included in the financing assumptions.

Including US$20 billion in the second line of defense, which was included in the press release, but was not part of the programmed package.

BIS-coordinated bilateral financing and Japanese assistance.

In Indonesia and Brazil, it is difficult to argue that the failure of the initial program was due to the financing package. The failure in Indonesia resulted largely from the evident lack of commitment of the government to implement the program and the rapid emergence of a major political dimension to the crisis, which accelerated not only the reversals in capital flows but also capital flight by domestic residents. The first Brazilian program failed because the initial objective of maintaining the crawling peg was not perceived as credible, particularly given the lack of sufficiently supportive policies and the overvaluation of the real.

In Korea, however, the initial failure of the program was more directly related to deficiencies on the financing side. The package as announced in the press note included US$20 billion of bilateral assistance as a second line of defense, but there was considerable lack of clarity as to whether this amount was really available. The program was originally based on the assumption that this amount would be needed to fill the estimated residual financing gap, but it was communicated to the staff at a fairly late stage that it should not count on this amount being available. The estimated financing gap was, therefore, reduced by arbitrarily increasing the assumed rollover rate of short-term debt.

There was lack of transparency in dealing with the problem, since details of the residual financing gap, and the rollover assumptions on which it was based, were not made public, and the second line of defense was included in the press announcements to give the impression that the actual resources being made available were larger than they were. However, the markets doubted the availability of the second line of defense and perceived the program to be underfinanced. The IMF recognized this fact and immediately pressed its major shareholder governments to achieve a rollover of bank credit lines, but to no avail (see “Private sector involvement” below). Outflows continued unchecked, and it was only when a rollover agreement with the banks was reached that the financing problem was effectively resolved. The conclusion is that if a rollover was not feasible, the amounts included in the second line of defense should have been made more readily available.

Critics have argued that large front-loaded packages of the sort used in these crises are subject to moral hazard, in that future investors may consequently lend imprudently in the expectation that they will be bailed out by the public sector in the event of adverse developments. This is possible in principle, but the empirical evidence is mixed.10 Certainly, private capital flows to emerging market economies have been very subdued since these capital account crises, a trend that may partially reflect the perception that the official sector will be less amenable to large packages and more insistent on private sector burden sharing in the future. This suggests that the moral hazard impact of official support in these cases was at best very limited.

Private sector involvement

The three country experiences provide some indication of the potential role for private sector involvement (PSI) in different circumstances. In Korea, the effort to encourage PSI in the second program was highly successful, because the short-term interbank credits covered by the agreement accounted for a large proportion of potential outflows. The direct involvement of the authorities of the major industrialized countries made it possible to orchestrate the rollover. The IMF was involved in consultations with the authorities and played a useful role in establishing quickly the comprehensive reporting system that enabled compliance with the rollover agreement to be monitored.

In Indonesia, the scope for PSI was more limited because the predominant form of capital inflows was foreign exchange borrowing by private nonfinancial firms. The need for an initiative in this area to establish a framework for negotiations and workout of such debts by the private sector was noted by the staff at an early stage but no action was taken. At a later stage, the authorities, with IMF technical assistance, tried to facilitate restructuring by establishing a voluntary framework for negotiations between creditors and corporations that could not service their debts, but progress was hampered by the absence of an effective bankruptcy system and other weaknesses in the legal system. Dealing with the external debt of nonfinancial firms is understandably much more difficult, but earlier attempts could have been made, at a minimum, to initiate the collection of data. Efforts should also have been made to protect the financing of exports and essential imports through official guarantees and other schemes for key trade credits, as was done in the summer of 1998 with Japanese bilateral assistance.

By the time of the Brazilian program, the potential role of coordinated private sector action in mitigating the impact of capital account crises was widely recognized. The Brazilian authorities, however, were extremely reluctant to appear to coerce the private sector, fearing that such action might accelerate the capital outflows and have adverse consequences on Brazil’s future access to international capital markets. The IMF made clear that its support would depend in part on the private sector response, but limited its role to helping to develop information systems and presenting the program to private creditors. Coordinated action was kept “voluntary,” and only informal pressure was exerted on international banks to maintain credit lines. The response from private creditors under the original program was only moderate but a renewed effort in the context of the more credible revised program proved much more effective. This suggests that a program with a high degree of credibility is necessary for the “voluntary” approach to PSI to work.


Despite initial failures, the large official packages were helpful in easing the adjustment to normalcy in both Korea and Brazil. In Indonesia, on the other hand, the depth of the collapse makes it difficult to argue that things would have been worse without the IMF, but the evolving circumstances made the size of access immediately irrelevant. In Korea and Brazil, official support was quickly repaid, in part ahead of schedule.

The role of the IMF in promoting PSI was fairly limited in all three cases. In Korea, the rollover agreement was a decisive factor, but this was only possible when initiated by the major shareholders. Under the circumstances, there was probably little alternative to the case-by-case approach to PSI actually adopted. Establishment of clear rules in this context might encourage an exit of capital in the early stages of the crisis. It may be useful for the IMF to have a menu of several well-defined options to use in a way most appropriate to the circumstances of each crisis, but some constructive ambiguity about the action to be followed in each case is desirable.

The three country cases thus suggest the following lessons:

  • The IMF can play a critical coordinating role in capital account crises, including vis-à-vis other providers of official and private financing. The ability of the IMF to perform this task, however, is limited by the reluctance of major shareholder governments to provide large bilateral financing and to use nonmarket instruments to influence the behavior of private investors in the absence of well-established rules. In other words, the lack of a clear mandate or framework for how the IMF should operate in such circumstances forced an ad hoc response. While a case-by-case approach may be to some extent inevitable, the lack of clear rules of the game create uncertainty.

  • Large access is difficult to justify when the program being supported lacks credibility in the markets in terms of policy sustainability. The decision to support Brazil’s unsustainable crawling peg, justified on the basis of global systemic considerations, is one example.

  • Markets tend to discount the availability and additionality of official financing from other IFIs and bilateral sources during the time of crisis, particularly if the non-IMF resources are subject to separate and vague conditionality and the country concerned already maintains ongoing financial relationships with the IFIs and the additionality is difficult to establish.11 Use of non-IMF resources in these circumstances to boost the “headline” size of the official financing package can damage the credibility of the program and distract attention from addressing the issue of involving the private sector, if necessary.

  • A dialogue with the private sector is necessary for the IMF to serve its facilitating role in involving the private sector. The Korean case illustrates that a more concerted approach to overcome “collective action” can work in some circumstances (e.g., when the relevant obligations are relatively concentrated), but it is not possible to say, within the context of the evaluation, how far such a conclusion can be generalized to other cases. Even when full-scale PSI is not feasible or necessary, concerted efforts should be made at the outset to make sure that trade credits for creditworthy firms are protected through official guarantee and other schemes.

Bank Closure and Restructuring

In both Indonesia and Korea, a weak banking system greatly contributed to the onset as well as the severity of the crises. Problems in the banking sector in these countries were further compounded by the distress of the highly leveraged corporate sectors brought about by sharp currency depreciations and the associated interest rate hikes.

Lessons from the East Asian experience

An important difference in how the banking crises were handled in Korea and Indonesia was the speed and decisiveness with which a comprehensive strategy began to be implemented. In Korea, a full guarantee for deposits and other bank liabilities was introduced before the IMF agreement, which was then immediately followed by the announcement of a comprehensive strategy, with appropriate enabling legislation. The functions of the Korea Asset Management Corporation (KAMCO) were enhanced, and a new consolidated system of supervision was established under the new Financial Supervisory Commission (FSC), which included a unit specially charged with bank restructuring. Even with best efforts, bank restructuring was a complex and prolonged process. It took Korea three months to establish the FSC and a full year to complete the setting up of the new regulatory framework. Bank restructuring is still an ongoing process. Nevertheless, the existence of a comprehensive strategy that was implemented, albeit with slippages in the timetable, helped ensure that there was no loss of monetary control and probably helped contain the magnitude of the crisis.

The restructuring effort in Indonesia was much less effective. A partial deposit guarantee was initially introduced for deposits of the closed banks, covering most of the accounts but only 20 percent of total deposits; this was followed three months later by a blanket guarantee for all bank liabilities, covering both depositors and creditors. The failure to introduce a full guarantee has been much discussed (and we return to this subject below), but the more important lacuna was the failure to adopt a comprehensive strategy for bank restructuring that was well-defined and well-communicated, and to apply consistently uniform and transparent intervention criteria to deal with problem banks. In the absence of such a strategy, the public saw inconsistency in the November 1997 closure of 16 banks (representing 3 percent of total banking sector assets), correctly believing that there were other banks in similar difficulty. Indeed, the IMF itself had identified 10 more banks that needed to be closed. The authorities’ insistence on secrecy, particularly regarding the 10 banks under BI-supervised rehabilitation that were not closed, prevented the public from understanding the whole picture.

Given weak implementation capacity and the rushed process, the logic and content of the bank closure were not well communicated to the public, and execution was less than satisfactory. As discussed, public confidence in the banking strategy was undermined by conflicting signals from the government. In contrast, the April 1998 action was competently executed by the IBRA, which took over the assets of 7 banks (representing 16 percent of total) and closed 7 smaller banks without causing any disruption. This was done under a comprehensive strategy in which uniform and transparent criteria were applied, and was accompanied by a professionally managed public relations campaign, better arrangements for meeting depositors’ claims, and a blanket guarantee. The failure to implement such an approach effectively in November 1997 proved to be one of the major weaknesses of crisis management.

The blanket guarantee

The issue of whether a blanket guarantee should have been offered in Indonesia in November 1997 deserves careful consideration. The lesson drawn by the IMF staff from the Indonesian experience is that “a blanket guarantee, rather than a limited deposit guarantee, is needed to restore confidence in the financial system” (Lindgren and others, 1999). Elsewhere in the same report, however, the staff recognizes that a blanket guarantee involves large contingent liabilities of uncertain value for the government, and that it can have regressive implications for wealth distribution—as taxpayers’ money is used to protect large depositors and even foreign creditors. The report concludes that the benefits of the blanket guarantee must be weighed against its potential costs.

In the case of Indonesia, the partial guarantee did not lead to a general loss of confidence in the banking sector. A large share of the banking system was accounted for by foreign banks as well as by state banks that enjoyed an implicit government guarantee, and the flight to quality in late 1997 took the form of a shift of deposits from private banks to foreign and state banks within the banking system (Enoch and others, 2001). The banking crisis was, therefore, not yet systemic (in the sense of affecting the whole banking system), and a blanket guarantee was, therefore, not essential. Under these circumstances, a partial guarantee was reasonable, though arguably the amount of the guarantee could have been increased, particularly to cover some institutional deposits, and extended to all banks at that time. Besides, in a corrupt banking system, where well-connected insiders had benefited both from high deposit rates and from questionable lending practices, a blanket guarantee would have given the same insiders an additional means of benefiting from abusive and corruptive practices. This is exactly what eventually happened with unlimited liquidity support.

In the end, the blanket guarantee was subject to abuse and consequently raised the fiscal cost of bank restructuring, which is now estimated at over 50 percent of GDP. The blanket guarantee in Indonesia was introduced as an act of desperation when the banking crisis seemed to be going out of control. Given the lack of adequate preparation, the guarantee was ill-conceived and was even made to cover some insider claims and interbank credits extended with full professional judgment and risk taking, including exposure in derivatives. It can be argued that the initial partial guarantee was too low. However, a higher guarantee introduced within the context of a well-communicated comprehensive strategy could have yielded a similar outcome without the fiscal cost and regressive distributional implications of the blanket guarantee.

The institutional setup for bank restructuring

The Asian experience also offers no clear lessons on the appropriate modality of government involvement in bank restructuring. Different institutional approaches were taken in Korea and Indonesia. In Korea, responsibility for bank restructuring (given to the FSC) was separated from that for asset management (given to the KAMCO). In Indonesia, the functions of bank restructuring and asset management were consolidated in a new agency.

In establishing the IBRA, the IMF staff believed that (1) BI needed to be protected from the fiscal cost of bank restructuring and the associated political pressure, in order not to impair its ability to conduct monetary policy, and (2) the new agency needed to be protected from the allegations of corruption plaguing BI. As a centralized public asset-management company, moreover, the IBRA offered the advantage of consolidating scarce financial expertise and the prospect of giving special legal powers to expedite loan recovery (Lindgren and others, 1999). As it turned out, however, the IBRA was plagued by problems from the outset. As a new agency, it was not given a clear mandate and was initially handicapped by lack of legal and regulatory powers. Moreover, the centralization of bank restructuring and asset management functions in one agency subjected the IBRA to tremendous political pressure and accusations of corruption; as a characteristic of a centralized public asset management company, there was also little incentive to maximize recovery values for the acquired impaired assets. On the other hand, the KAMCO was made to operate on commercial principles and, as a specialized agency, it could focus its sole attention on that function and was effective in rapidly selling the impaired assets.

Given the weak legal system and prevailing corruption in Indonesia, it may well be that no alternative could have worked better than the IBRA. In the light of the Korean experience, however, the fact that a better outcome was achieved after the establishment of the IBRA than previously cannot be used to conclude that the IBRA solution was the best strategy, something that should be adopted in all similar situations.


When bank restructuring was launched with the immediate closure of the least viable institutions in Indonesia and Korea in the fall of 1997, there was no internationally accepted best practice for handling bank restructuring in emerging market economies. The IMF staff (and others for that matter) had only limited experience in dealing with a banking crisis, particularly within the context of an IMF-supported program designed to deal with a capital account crisis. The contrasting outcomes of the Indonesian and Korean experiences have since formed an important basis for the IMF staff’s emerging views of best practice in dealing with a systemic banking crisis, as articulated in a recent policy paper by MAE.12 As this paper clearly states, the experience of East Asia suggests that a successful bank closure and restructuring program must include a comprehensive and well-communicated strategy in which uniform and transparent intervention criteria are consistently applied.

The experience of Indonesia and Korea, however, is less clear on the exact modality of public sector involvement in the restructuring process (i.e., consolidated versus nonconsolidated restructuring supervision), nor is it definitive in suggesting that a blanket guarantee, rather than a limited deposit guarantee, must be introduced at the outset of a banking crisis. A blanket guarantee may not stop runs motivated by wider confidence concerns than just banking sector problems, while it involves large contingent liabilities for the government with serious regressive implications for burden sharing. Its benefits must therefore be carefully weighed against its potential costs, within the specific context of the economy in question. In either case, the coverage of any guarantee scheme must be well designed and, particularly in a weak legal and supervisory system, early steps to preserve and correctly value assets are essential.

Structural Conditionality

Structural conditionality was present in all three cases, and has been the subject of much controversy (see Box 4.1 for how structural conditionality is typically included in an IMF-supported program). One view holds that the structural reform measures in the IMF-supported programs with Indonesia and Korea were unrelated to the immediate problem of crisis resolution; they distracted attention from the core macroeconomic and financial issues; and they were widely felt to be an encroachment into domestic decision making, creating an unnecessary opposition (Feldstein, 1998). Some have even argued that the extensive structural adjustment agenda had a perverse effect on confidence by signaling to the markets that the situation was much worse than they had feared (Radelet and Sachs, 1998a and 1998b). However, there is an alternative view, which holds that restoring market confidence required addressing the structural cause of the problem (Summers, 1999; Goldstein, 2002).

In the case of Indonesia, structural conditionality was linked primarily to governance-related objectives. It has been argued that this was essential to signal a clean break with the past, namely, that a new way of doing business was being established (Khan and Sharma, 2001). A guidance note issued by the IMF Executive Board in July 1997 indicated that IMF involvement in governance issues was justified when “poor governance [would] have significant current or potential impact on macroeconomic performance…. and on the ability of the government to credibly pursue policies aimed at external viability and sustainable growth.” 13 This certainly provided a somewhat open-ended mandate to pursue governance reforms if they had a significant impact on “potential” macroeconomic performance or on the credibility of policies aimed at external viability. The critical question is whether the scope of conditionality prescribed for Indonesia was indeed necessary.

Critical versus noncritical measures

One way of determining whether structural conditionality was excessive is to distinguish those structural measures that were critical to crisis resolution from other measures that, while potentially useful in eliminating distortions, were not critical to crisis resolution. In both Indonesia and Korea, as already discussed, deficiencies in the financial sector were central to the crises, and tackling these was crucial to regaining market confidence. They were correctly a major focus of the programs, though in Indonesia implementation was flawed and there were also design deficiencies, particularly, the absence of a comprehensive strategy for bank restructuring.

Box 4.1.Conditionality for Structural Reforms in an IMF-Supported Program

IMF-supported programs treat structural reform measures in one of four ways. We use the Indonesian program of November 1997 to illustrate how structural measures are included in a program. Some conditions are short term in nature (i.e., they must be met before the next review, while others are longer term (i.e., they should be completed by the end of the program).

  • Measures are targets with no conditionality attached. For example, the program envisaged a broad range of structural reforms, many linked to issues of governance, including elimination of export taxes and restrictions, dismantling of domestic monopolies, and greater private sector participation in the provision of infrastructure.

  • Structural benchmarks do not directly govern disbursement but trigger discussion on corrective action if not met. These included the introduction of full tax-deductibility of loan loss provisions, completion of a public expenditure review and audits of state-owned banks by internationally recognized accounting firms, and the reduction of tariffs.

  • Performance criteria govern disbursement (i.e., if they are not met, disbursements are automatically interrupted). These included the closure of certain unviable banks under central bank–supervised rehabilitation, establishment of quantitative performance targets for state-owned banks together with monitoring mechanisms, issuance of implementation regulations on procurement and contracting procedures, and elimination of subsidies by raising electricity and petroleum prices.

  • Prior actions are measures required before a program request or review can be considered by the Executive Board. The Indonesian program included the closure of 16 banks as a prior action.

Instead of limiting conditionality to these critical areas, the Indonesian programs, especially the revised January 1998 program, included a large number of additional structural reforms. The rationale for adopting extensive structural conditionality in the January program was that it was necessary to restore confidence—the problems of cronyism and corruption, which had not been explicitly dealt with thus far, were brought to the forefront both by extensive press commentary and by major shareholder governments. It was an atmosphere in which it came to be believed that confidence could only be restored if the Suharto regime demonstrated a radical change in its way of doing business.

It is difficult to establish the counterfactual as to whether confidence would indeed have been restored had all the reforms identified been implemented. What is known is that there was no positive announcement effect. Despite affirmation by President Suharto in the form of a public signing ceremony, the markets remained unconvinced about his personal commitment. Besides, the January program did not address the macrocritical areas of bank and corporate debt restructuring. In retrospect, the basic approach of loading the programs with an overly large agenda of structural reforms, however desirable they may have been on merit, seems ill-advised from a standpoint of restoring confidence. The elaboration of such an extensive agenda, much of which did not seem critical for stabilization, may have hurt confidence, once it became clear that the measures were not owned at the highest political level. It would have been better to concentrate on macrocritical areas, along with greater insistence on credible upfront action in those core areas.

In Korea, too, the agenda of reform was broader than seemed necessary, covering not only financial sector reforms but also trade liberalization, corporate governance, and labor market reform. Stabilization was achieved well before the reforms could be implemented and indeed the pace of structural reform in nonfinancial areas slowed when the economy rebounded from the crisis. It is difficult to say whether the authorities’ initial commitment to the broad reform agenda helped to restore market confidence, but certainly immediate progress in reform in some areas was not perceived by the markets to be necessary. This is not to say that these reforms did not have a significant longer-term beneficial effect on the economy. They may well have done so. But they were not critical to resolving the crisis.

The program in Brazil did not suffer from these problems. The focus of structural conditionality was on macrocritical reform, particularly covering structural fiscal reform and prudential supervision. The paucity of extensive structural measures in other areas reflected the fact that many of the distortions relevant in Asia did not exist in Brazil, at least to the same extent. There was also strong ownership by the authorities. The Fiscal Responsibility Law was particularly helpful in establishing a general framework to guide budgetary planning and execution, with disciplinary mechanisms for any failure to observe its targets and procedures, and contributed to the greater credibility of fiscal policymaking in that country.


Two important lessons to be drawn from these cases are now well recognized within the IMF:

  • First, ownership defined as broadly as possible (but especially at the highest political level) is key to the successful implementation of a structural reform program. But assessments of ownership can be very complex, requiring a good understanding of the political economy context. Even highly symbolic acts—such as the President signing the LOI—may be misleading.

  • Second, detailed and extensive structural conditionality, particularly in areas that are not macrocritical, is not helpful to crisis resolution. This is so because it is more difficult to demonstrate commitment in the short term to an extensive agenda and because the risks of subsequent disputes on implementation, which blur the message of commitment to a coherent strategy, are greater. Perhaps more important, a detailed structural program also tends to distract attention from the immediate macroeconomic issues. This conclusion supports the recent initiatives by IMF management to streamline conditionality and enhance ownership by applying conditionality more sparingly to “structural measures that are relevant but not critical, particularly when they are not clearly within the IMF’s core areas of responsibility and expertise.”14

The evaluation also suggests the following additional messages:

  • When action in areas that are not macrocritical is nevertheless deemed to be important, a “second-best” policy package that is strongly owned may be more likely to help restore confidence than a “first-best” package that is painfully negotiated and over which there are substantial domestic reservations. The possibility of such trade-offs needs to be recognized.

  • The crisis should not be used as an opportunity to seek a long agenda of reforms just because leverage is high, irrespective of how justifiable they may be on merits. This should be the approach even if reformist groups within the government are keen to use the leverage of the program to push reforms. When significant distortions are known to exist, and the government is committed to reform, laying out a road map for these reforms as an indicative direction by the government is appropriate, but these measures do not need to be the focus of IMF conditionality. The principle of parsimony should guide IMF conditionality in such situations. In large part, this was the approach taken in the Brazilian program.

Communications Strategy to Enhance Ownership and Credibility

Restoring confidence involves more than just program design. It is also necessary to have an effective communications strategy to enhance country ownership (with the public) and credibility (with the markets). All three programs initially suffered from the failure to communicate their logic to the public and the markets.

Building country ownership

Country ownership generates domestic political support for an agreed program, hence making it more likely to be implemented. Ownership, however, is a broad concept. While program negotiations must necessarily be conducted with a small group of senior officials in the finance ministry and the central bank, successful implementation depends on the support from other stakeholders, including the head of government, key officials from other ministries, the bureaucracy that must implement the program, the parliament that must approve the necessary legislation, and civil society at large (Khan and Sharma, 2001; Boughton and Mourmouras, 2002). An effective public communications strategy is needed to build broader public support, hence stronger country ownership, during a crisis, when speed is of the essence and wider consultation is therefore not feasible.

Building credibility

Given the need to restore market confidence, the communications strategy must also address the need to build the credibility of a crisis management program with the markets. In designing a program to restore confidence, the IMF must understand what the markets are looking for in a program and to explain the logic of the program. Particularly in a capital account crisis, the IMF may not necessarily have more information on critical issues than the markets, necessitating some dialogue with the markets (Cottarelli and Giannini, 2002). For example, the markets may become nervous if there is a perception that concerted action may be taken to involve the private sector, including a restructuring of sovereign debt. In such cases, it is important to disclose the financing assumptions when explaining the logic of the program. When concerted action is taken, of course, communication with the markets is the crucial ingredient.

At the time of the East Asian crises, the publication of LOIs was not yet customary. The failure to publish the LOI in a timely fashion in Indonesia in late 1997 undermined the potential impact of the program in restoring confidence, as private investors began to speculate on the details of the program. This lesson was quickly learned, and subsequent LOIs were published in all three cases. However, the staff reports supporting the requests for use of IMF resources were not published. The publication of such reports could have been particularly effective in communicating the logic of programs to the markets, hence helping to build credibility.

In building credibility, transparency can be a powerful tool. In the repeated game in which the IMF is engaged, relevant information should be disclosed even if it may cause negative shifts in market sentiment because, in the long run, the IMF cannot expect to be effective if it is perceived as willing to go along with hiding information from the markets. In Korea, a confidential staff report was leaked to the Korean press a few days after the program was approved, revealing that the level of usable reserves was very low and that the stock of short-term external debt was substantially higher than generally believed. Although this undermined the initially positive market response, it would have been better publicly to acknowledge these facts at the outset and to design the program accordingly.15


Given the high degree of uncertainty regarding both economic and political developments during a crisis, events often do not develop as planned. The right communications strategy can ensure that this does not cause damage to credibility. For example, an effective communications strategy is necessary to make sure that the markets do not misinterpret the degree to which the authorities’ policy actually conforms to their commitments under the program. In Indonesia, the January 1998 announcement of a 1998/99 budget confused the markets, because it appeared to violate the programmed fiscal target (see the Indonesia country annex). Such confusion could have been avoided, if the content of the program had been explained to the investors, and if the IMF and the authorities had agreed on a public communications strategy to be followed when program-related information would be announced.

As discussed earlier, such a communications strategy would be facilitated if Board papers were to spell out the major risks to a program and the broad direction in which policies would respond under different scenarios. It is sometimes argued that explicit discussion of the risks could itself undermine confidence. We do not find this argument convincing since (as the experience of the three country cases shows) financial market participants will usually be well aware of them. To the contrary, a communications strategy that explains how policies would respond to key risks is likely to enhance credibility.

Since the crises, the IMF has come to recognize the importance of public communications in its role as crisis coordinator. Important steps have been taken in recent years by the IMF, particularly through its External Relations Department, to improve the effectiveness of its “external” communications strategy, designed to enhance country ownership and transparency.16 While these steps are valuable, it is also necessary to emphasize the need to design an effective communications strategy to be followed in a capital account crisis, including appropriate ways in which public communications expertise—especially with financial markets—can be integrated quickly into the program negotiation and implementation process.


Overoptimism appears to be a feature of most large IMF-supported programs. Musso and Phillips (2001) find a significant optimistic bias in real GDP projections for the first year of adjustment programs for which access is large or where the economy is large. This bias, however, is not present in their sample of IMF-supported programs as a whole.

Balance sheet analysis began to figure more prominently in the thinking of the economics profession after the East Asian crises, with the emergence of the so-called third-generation model of currency crisis (Allen and others, 2002). However, the idea that devaluation could have contractionary output effect when there is net external debt denominated in foreign currency was well-known in the academic literature for at least 35 years, most frequently associated with the works of Carlos Diaz-Alejandro (1963, 1965). Similar balance sheet issues, such as unhedged foreign currency exposure and their effects on private aggregate demand, were raised following the Mexican crisis of 1994–95.

According to the November 1998 program document, the fiscal balance for 1999 was expected to deteriorate on account of several factors, including the “disappearance of once-off tax revenues,” “retroactive wage increases,” and “the effects on the social security finances of the acceleration of early retirements.”

Higher nominal interest rates, however, affected different sectors of the economy differently, because sharp changes were taking place in relative prices, even though real interest rates measured using average inflation were negative. These issues of monetary policy in Indonesia are explored in greater detail in the Indonesia country annex.

Subsequently, Hong Kong SAR and Malaysia resorted to less conventional measures: purchases of stocks in the secondary market and controls on capital outflows, respectively.

Lahiri and Végh (2002), for example, show that high domestic interest rates can induce a portfolio shift towards the domestic currency under the right circumstances but there is a range in which sufficiently high interest rates can also weaken the currency by contracting domestic output and by raising the government’s debt-servicing costs.

For example, Kraay (1998) finds that tighter monetary policy does not have a statistically significant impact on whether speculative currency attacks succeed or fail, even when one controls for the endogeneity of the policy response. Goldfajn and Gupta (1999) find some evidence that tighter monetary policy in the aftermath of currency crises helps to ensure that an undervalued real exchange rate returns to its equilibrium level through nominal appreciation rather than higher inflation. But their results are not robust to different specifications and do not hold when a currency collapse is accompanied by a banking crisis.

In the fast-moving crises of Indonesia and Korea, the procedures under the Emergency Financing Mechanism were invoked to allow the IMF to agree on a program quickly.

These experiences confirm the conclusion of earlier studies that the “catalytic” effect of IMF programs on private capital flows is typically small (Cottarelli and Giannini, 2002).

See Ghosh and others (2002) for a brief summary of the literature. Essentially, empirical work has focused on the presence or absence of significant market reactions (typically measured by bond spreads) to actions or decisions that are expected to affect the expectations of private investors that they will be “bailed out,” including the announcement of a large IMF-supported financing package, a large-scale default, and a sovereign debt restructuring.

In the case of Indonesia, while the ADB agreed to provide US$2.8 billion in quick-disbursing loans, it also canceled existing loans amounting to about US$900 million in 1998 and about US$660 million in 1999–2000 in view of “the reduced availability of counterpart funds and the changed priorities after the crisis” (ADB, 2001a). In Brazil, the emergency loans to be provided by the IDB included a loan of US$1.2 billion that had already been approved in September 1998 but had not yet been disbursed (IDB, 2001).

“A Framework for Managing Systemic Banking Crises,” SM/03/50, February 2003. Also see Andrews and Josefsson (2003).

“The Role of the Fund in Governance Issues,” EBS/97/125, July 1997. According to Goldstein (2002), some IMF staff interpreted this guidance note to imply that the Executive Board would not support programs that did not address serious and widespread governance and corruption problems.

“Managing Director’s Report to the International Monetary and Financial Committee—Streamlining Conditionality and Enhancing Ownership,” IMFC/Doc/4/01/6, November 6, 2001.

In this context, the former First Deputy Managing Director of the IMF has acknowledged the need for transparency, citing the loss of credibility that occurred in a similar situation in Thailand (Fischer, 2001).

See, for example, “A Review of the Fund’s External Communications Strategy,” SM/03/69, February 2003.

    Other Resources Citing This Publication