This annex presents the detailed assessment of the role of the IMF in Indonesia’s capital account crisis of 1997–98, which forms the basis for the analysis in the main report. It covers the role of the IMF in the precrisis surveillance phase and the crisis management phase. Issues related to the ongoing program with Indonesia, which began in February 2000, are outside the scope of our enquiry.


Annex 1 Indonesia


This annex presents the detailed assessment of the role of the IMF in Indonesia’s capital account crisis of 1997–98, which forms the basis for the analysis in the main report. It covers the role of the IMF in the precrisis surveillance phase and the crisis management phase. Issues related to the ongoing program with Indonesia, which began in February 2000, are outside the scope of our enquiry.

The Indonesian crisis was particularly severe and prolonged, compared with the other crisis cases reviewed in this report. GDP fell by 13 percent in 1998 and there was a substantial increase in the percentage of the population in poverty. Subsequent recovery was slow, with an average annual growth rate of just above 3 percent from 1999 through 2002, so that at the end of 2002, GDP remained about 2 percent below the 1997 level. It is useful to recall that the crisis, which largely started out as economic, became increasingly political. Particularly, between December 1997 and the spring of 1998, while it was apparent that the first program had failed, political issues related to the succession of President Suharto and growing social unrest made it difficult to design a credible alternative. Our evaluation suggests that the exceptional severity of the Indonesian crisis is in large part a reflection of the confluence of economic and political crises, which limited the ability of conventional policy tools to address economic problems.

This annex is organized as follows. It first evaluates the effectiveness of surveillance prior to the crisis. It then discusses issues of program design, including (1) fiscal policy, (2) interest rate policy and monetary targets, (3) exchange rate policy and capital controls, (4) official financing, (5) bank closure and restructuring, (6) deregulation, (7) corporate debt restructuring, and (8) the initial strategy and its adaptation. The following section discusses the IMF’s mode of operations, covering such issues as country ownership, the decision-making process, human resource management, and the role of major shareholders and collaboration with the World Bank and the ADB. The final section presents conclusions and an overall assessment.

Precrisis Surveillance

This section discusses the effectiveness of IMF surveillance in three areas of potential vulnerability: macroeconomic performance, banking sector weaknesses, and corruption and cronyism. The IMF broadly identified the potential vulnerabilities in all these areas, but it failed adequately to recognize their seriousness and adverse implications.1

Macroeconomic performance

Indonesia’s performance before the 1997 crisis was characterized by strong economic growth and apparently sound macroeconomic fundamentals (Figure A1.1). IMF surveillance, however, noted the risks associated with the large capital inflows, which averaged 6 percent of GDP during 1992–96. As a result, the stock of private foreign debt increased rapidly from about US$38 billion in 1995 to US$65 billion just before the crisis and US$82 billion at the end of 1997. Moreover, short-term private foreign debt was a large proportion of the total, reaching US$33 billion, just before the crisis in 1997, equivalent to 1.5 times the stock of gross international reserves. However, IMF surveillance grossly underestimated the magnitude of short-term debt, hence the vulnerability of capital flows to a shift in market sentiment.2

Figure A1.1.
Figure A1.1.

Indonesia: Selected Macroeconomic Indicators

Source: IMF.

Both the IMF and the Indonesian authorities recognized that the volume of capital inflows was uncomfortably large. This was a frequent subject of discussion at official meetings (e.g., the EMEAP Central Bank Governors’ meeting in early 1995) and in the academic literature (Radelet, 1995). As a counterpart of the increasing capital inflows, the current account deficit widened from 1.8 percent of GDP in 1992/93 to 3.3 percent in 1995/96, and to 3.5 percent in 1996/97.

The policy advice from the 1996 Article IV consultation mission, endorsed by the Executive Board, was that the authorities should follow tight fiscal and monetary policies, combined with faster external debt repayment. According to a former senior Indonesian official, Bank Indonesia (BI) made attempts to measure the capital inflows, an idea also endorsed by the Executive Board. This, however, sparked protests from the financial community, fearing that it was a precursor to imposing capital controls. Limitations were placed on the overseas borrowings of state enterprises, but the effectiveness of this initiative was uncertain.

The 1997 Article IV consultation report noted that the country was vulnerable to external shocks, and warned that excessive demand pressures were contributing to higher inflation and a wider current account deficit. The IMF advocated a tighter fiscal and monetary policy stance, greater exchange rate flexibility, and accelerated structural and banking sector reforms to maintain progress in reducing inflation, contain current account deficits, and minimize external risks. The IMF argued for a smaller current account deficit than the amount considered acceptable by the Indonesian authorities, who thought that a deficit of up to 4 percent of GDP was sustainable.

The authorities’ views were based on the following factors:

  • There were no strong indications of exchange rate overvaluation and non-oil exports were registering robust growth;3

  • The current account deficit remained smaller than those in most ASEAN countries and was no higher than in 1991/92 and was significantly lower than the levels in Thailand (5–8 percent of GDP) and Mexico (6–7 percent) in the three years prior to the crises in those countries;

  • The counterpart of the higher current account deficit was an increase in private sector investment to 27 percent of GDP in 1996 from 20 percent in 1992, likely contributing to faster economic growth; and

  • Although debt was high by regional standards, it was evolving favorably with a stable and relatively low debt-service ratio of just over 30 percent, accompanied by a reduction of total external debt to under 50 percent of GDP in 1996/97 from 56 percent in 1991/92.

In retrospect, the elements that were missed in the authorities’ analysis—and underemphasized by IMF surveillance—were the macroeconomic implications of short-term capital flows that were vulnerable to a sudden shift in market sentiment and the underlying weakness of seemingly buoyant private investment, much of which was in fact supported by imprudent lending and of questionable productivity.

Banking sector weaknesses

The risks from large and potentially volatile capital inflows were amplified by the poor quality of domestic financial intermediation and governance problems in the corporate and banking sectors. The fragile state of the banking system mainly resulted from the rapid deregulation following the so-called Pakto reform of 1988, which allowed a substantial increase in the number of banks without adequate prudential regulations.4 Entry to the banking industry was made possible with a small amount of capital, but there were no adequate provisions for weak banks to exit.

A reasonable structure of prudential regulations had been put in place, in part with extensive technical assistance received over the years from the World Bank, but this had little impact on the quality of banking, because enforcement was poor. Apart from the general problem of weak public administration, attempts to impose rules ran into stiff opposition from politically well-connected vested interests. This was demonstrated most clearly in the removal of the head of prudential supervision at BI in 1993, when he attempted to enforce connected lending limits on the largest of the private banks, which had close political connections. With this precedent, banks flouted prudential rules with impunity. The easy flow of financial resources to conglomerates through the banking system was facilitated by an international environment that encouraged flows of foreign capital into emerging markets.

Some academic researchers have argued that the Pakto reforms were designed to provide the well-connected with access to cheap money and created a process of financial flows closely approximating a “Ponzi” game (Cole, 2002). Indeed, banks affiliated with large conglomerates owned by the well-connected tapped the large pool of household savings and used the deposits to fund their own affiliated firms, often in risky or questionable ventures. Many of the loans were never repaid, while the owners paid themselves high interest rates on their deposits (Gie, 1993). BI dealt with the resulting insolvency by “nursing” the banks to health through long-term low-interest loans. The maturity of these loans could be as long as 30 years, with a grace period as long as 10 years and an interest rate as low as 1 percent.

The IMF correctly perceived that there were major problems in the state banking sector, an area where the World Bank was in the lead in the efforts to promote reform.5 In several surveillance reports, the IMF staff alerted the Executive Board to the serious governance issues in the state banks and encouraged the authorities to move forcefully in this area. It is understandable against this background that the staff perceived the shift from public- to private-sector banks as a positive contribution to dealing with the problems of the banking sector. However, the dangers of poor governance in private sector banks appear to have been underplayed.

There were serious governance problems in the private sector banks. These problems first came to the knowledge of the IMF in 1994, when a technical assistance mission from MAE visited Indonesia. Upon examining the supervision data provided by BI, the MAE mission identified serious solvency problems in a number of private banks and learned that the problem banks were being effectively recapitalized with subsidized loans provided by BI, creating enormous moral hazard. The mission also came to view the “losses” of the banking system as largely representing transfers to conglomerates run by the well-connected. Despite these suspicions of corruption, however, there were no hard data to make the link between balance sheet weaknesses in the banks and governance failures. The confidential nature of technical assistance work meant that it was never presented to the Executive Board or widely discussed within the staff. However, the area department also did not explore the implications of warnings made by RES during the interval review process that there would be serious macroeconomic consequences from these vulnerabilities if there were confidence shocks.

These concerns were noted in surveillance reports but they were not adequately addressed, for example, by stress testing or exploring their potential policy implications. Drawing on the work of MAE, the background paper for the 1997 Article IV consultation observed that the main problems of the Indonesian banking sector were reflected in a high share of NPLs, incomplete compliance with prudential requirements by some banks, concentrated bank ownership and connected lending, continued operation of problem banks, and large exposure of banks to property loans. While the paper offered precise technical measures to address these problems, the governance and moral hazard issues identified by the earlier MAE missions were understated. A deposit insurance scheme, an idea recommended by MAE as a measure to increase confidence in the banking system, was taken up by the Selected Issues paper but was not followed up in the staff report.

In short, the nature of the main problem was identified and signaled to the Executive Board, but in a muted fashion. In line with the prevailing convention of the time that corruption should not be directly discussed, Board papers did not present an explicit assessment of the cronyism and corruption that created moral hazard in the banking sector. They also failed adequately to analyze the potential macroeconomic impact of shifts in market sentiment.

IMF surveillance noted that a number of reforms were being initiated by the authorities. For example, in 1996, six private banks were merged into three, and the authorities were considering merging the seven state banks. BI was encouraging problem banks to address their NPLs and the President issued a decree in December 1996 on the procedures for revoking the business licenses of banks and their dissolution and liquidation. In February 1997, the President approved the closure of seven banks to be implemented after the elections, and BI strengthened the prudential regulations by requiring (1) a gradual increase of the capital-adequacy ratio to a minimum of 12 percent by 2001 and a minimum Rp 150 billion (around US$60 million) of paid-up capital for each foreign exchange bank; (2) rating of commercial paper issued and traded through banks; and (3) tougher selection standards for bank management positions. However, with the benefit of hindsight, the IMF appears to have been overly impressed by the initiatives that did not contribute substantively to addressing the underlying problems.

The weak banking system proved highly vulnerable to external shocks. Once the Thai crisis prompted a reassessment of potential risks throughout the region, foreign investors began to pull out of Indonesia, thereby drying up the previously plentiful source of low-cost financing to the corporate sector. The heavily indebted corporate sector found itself facing liquidity problems,6 which were then compounded by a sharp exchange rate depreciation that raised the cost of servicing foreign debt. Conglomerate after conglomerate stopped servicing their loans, as the value of foreign currency debt doubled and then quadrupled in value. Foreign lenders rushed to close their exposure to Indonesia. At the time of the crisis, the banking system thus faced a huge portfolio of potential NPLs. This risk was on top of the system’s own severe internal difficulties.

Of course, it is not possible to say with any certainty that the banking system would have been able to survive the massive exchange rate shocks of 1997–98, even if it had been stronger financially and with more robust governance. Nor is it possible to say that a more candid discussion of these issues as part of surveillance would have significantly affected domestic policies. Nevertheless, it is the case that the potential risks were not sufficiently flagged or analyzed. As a consequence, the knowledge of the underlying balance sheet vulnerabilities was relatively limited, when the crisis did hit.

Corruption and cronyism

Indonesia’s vulnerability to crisis was greatly increased by the increase in corruption and its changing nature (Pincus and Ramli, 1998; Kenward, 2000; Lee, 2000; Booth, 2001; Cole, 2002). In the 1990s, there emerged a creeping return to restrictive business practices and rent-creating opportunities for the President’s family and well-connected businessmen, with a corresponding weakening of regulatory and supervisory controls. For example, in 1996, the palm oil sector was closed to foreign investment, export bans and restrictions were introduced in a wide range of products, and impediments were placed on intraregional trade in livestock; in April 1997, the preshipment inspection system, a customs procedure designed to prevent corruption and managed by a foreign firm, was canceled although it had proved highly effective.

Originally, corruption in Indonesia was akin to a tax on the cost of a project, charged by and paid through established channels to maintain the stability of the political system (Charap and Harm, 1999). Even such corruption raises moral and equity concerns, but its impact on efficiency was said to be limited by the certainty and relatively low levels of the charge. In the early 1990s, however, the media began to see a change in the system of corruption, and to draw links with the empire building of the President’s children and well-connected businessmen.7 Corruption was being transformed into an ever-widening system of deliberate rent-creation for the well-connected, including the creation of monopolies and monopsonies, and exclusive rights to large industrial or infrastructure projects, such as the National Car Project.

These issues surfaced in discussions with the authorities in the precrisis period, and the staff consistently supported the World Bank’s view that slippages in structural areas were damaging Indonesia’s medium-term prospects. As noted, much of it involved favored treatment given to the First Family and close associates of the Palace, but some simply represented a continuation of the dirigiste tendencies that were still the way of doing business in Indonesia. The staff reports for the 1996 and 1997 Article IV consultations recommending renewed deregulation received broad support within the IMF, including from the Indonesian chair on the Executive Board. The 1997 report identified a list of structural reform measures that would later become the core elements of structural conditionality in the IMF-supported program (see Appendix A1.1).

It is difficult to determine the extent to which the staff was aware of the growing scale of corruption and its deleterious effects because it was customary at the time for governance issues to be dealt with only obliquely and indirectly in surveillance reports and Executive Board discussions. The staff took a technocratic approach of dealing with symptoms (i.e., creeping regulation) without explicitly addressing their underlying causes (i.e., cronyism), thereby blunting their analysis and Board discussion. An explicit focus and candid Board discussions might have brought out more clearly the changing nature of corruption in Indonesia, and the macroeconomic risks it posed. Whether it would have had an impact in Indonesia is an open question, but at least it would have better prepared the IMF to deal with the crisis when it broke out.

Program Design

This section reviews major elements of program design in the IMF-supported programs in 1997 and 1998, with a focus on how the emphasis in program design changed from November 1997 to January 1998. The initial program was designed on the assumption that the crisis was essentially a moderate case of contagion and the implementation of a relatively conventional IMF-supported program would bring the rupiah back into a reasonable range. These expectations were belied and, toward the end of December, it became clear that the crisis in Indonesia was much more severe than elsewhere in the region. The crisis at this stage had become intensely political and there were doubts about whether the government was committed to the program. This led to the renegotiation of the program in January and a new LOI. The emphasis in program design switched to the establishment of structural conditionality to signal a new way of doing business in the hope that this would restore confidence.

Fiscal policy

Prior to approaching the IMF, the Indonesian authorities had already responded to the crisis by cutting public spending on low-productivity projects. This was meant both to facilitate the required current account adjustment and, more important, to help rebuild international confidence by signaling the authorities’ determination to reduce dependence on capital inflows while improving governance.

The November 1997 program broadly endorsed this approach. In internal discussions, the First Deputy Managing Director moderated the fiscal targets proposed by staff and rejected proposals to increase the value-added tax (VAT), in order to avoid fiscal overkill at a time when output developments were uncertain. The program planned for a modest improvement in the fiscal position in fiscal year 1998/99 to cover partially the unknown carrying cost of bank restructuring (Table A1.1).8 Specifically, the initial program, based on growth assumptions of 5 percent for 1997/98 and 3 percent for 1998/99, targeted:

Table A1.1.

Indonesia: Fiscal Outomes and Targets

(In percent of GDP)

article image
Sources: IMF staff reports; and IEO staff estimates.
  • An overall budget surplus of 0.75 percent of GDP for 1997/98, compared with a surplus of 0.5 percent projected during the 1997 Article IV consultation, and a surplus of 1.3 percent during 1996/97;

  • An overall budget surplus of 1.3 percent of GDP for 1998/99, though this was to be reviewed later in the light of developments before being fixed as a performance criterion;

  • A reduction of capital spending amounting to 0.5 percent of GDP in 1997/98 and a further 0.5 percent cut in1 998/99 through postponing or canceling low-productivity projects (such as inter-island bridges);

  • Cuts in operations and maintenance expenditures amounting to 0.25 percent of GDP in 1997/98 and a reduction in fuel subsidies amounting to 0.5 percent of GDP in 1998/99; and

  • Various tax and expenditure measures, including higher excise taxes on tobacco and alcohol; lower transfers to state-owned enterprises and improved tax administration.

The Indonesian program has been extensively criticized for an overly contractionary fiscal and monetary stance which, according to some critics, actually made matters worse. As far as fiscal policy is concerned, the tightening proposed for 1997/98 was modest and reflected the basic assumption that Indonesia was suffering from a moderate case of contagion. The implementation of the program was expected to bring about a quick restoration of confidence and a recovery of the exchange rate, while growth would decelerate but still remain respectable.

The growth assumption on which the November program was based turned out to be far too optimistic and this was a fundamental weakness of the initial program design. While GDP growth in 1997/98 was 4.8 percent, only marginally lower than the 5 percent rate projected in the program, there was a collapse in 1998/99 with GDP declining by 13 percent instead of growing by 3 percent as projected. Some critics have attributed the collapse in output to the pursuit of tight fiscal and monetary policies in circumstances where these were not warranted, but the problem arguably lay elsewhere. The output collapse in 1998/99 was driven not by the stance of fiscal policy but by the near-collapse of private investment in the first and second quarters of 1998. Private investment is difficult to forecast over a business cycle and earlier studies have shown that IMF-supported programs tend to be overoptimistic about private investment (Goldsbrough and others, 1996). In Indonesia, the collapse of private investment was especially severe because of (1) the unexpectedly large exchange rate depreciation in a situation where corporations had borrowed heavily in foreign exchange, and (2) the impact of political developments—including especially rioting against the ethnic Chinese community—on business confidence.

The role of fiscal policy in the Indonesian crisis needs to be evaluated in this broader context of larger forces driving developments in the real economy. The November 1997 program implied modest tightening in 1997/98 and further tightening in 1998/99, but it also stated that the fiscal target for 1998/99 would be updated and converted to a performance criterion at the time of the first review in January 1998, taking into account, inter alia, output developments (see Appendix A1.1). Unfortunately, these provisions incorporating flexibility were not made public. The 1998/99 draft budget presented by the government on January 6, 1998, which proposed zero deficit, appeared to violate the terms of the agreement with the IMF and triggered speculation in the press that it might signal a possible withdrawal of IMF support.9 In fact, by the time the 1998/99 draft budget was put together in the latter part of December, it was known that the growth forecast for 1998/99 would need to be revised downward and internal documents and interviews make clear that a consensus had emerged within the IMF that a surplus was not appropriate under the conditions that were then prevailing or were likely to prevail in Indonesia.10 The IMF did issue a statement of support for the announced budget within two days. The confusion could have been avoided if the authorities had consulted with the IMF before they released the draft budget, explaining that the overall balance differed from that in the November program because the situation had changed and that this was done in full consultation with the IMF.

The second LOI agreed in mid-January 1998 reduced the earlier 3 percent growth projection to zero growth and provided for a relaxation of the fiscal stance to a deficit of 1 percent of GDP for 1998/99. The third LOI signed in April 1998, which was the operationally relevant one for the 1998/99 budget, further raised the programmed overall deficit to 4.7 percent of GDP, acknowledging the need for temporary subsidies to protect the poor, while proposing a further cut in low-priority projects in the development budget. As the sharper output decline became more evident in the following months, the subsequent LOI in June 1998 further relaxed the fiscal target to a deficit of 10.1 percent of GDP, the largest in any IMF-supported program.

The actual budget deficit in 1998/99, at 2.1 percent of GDP, was much smaller than programmed. Fiscal policy was therefore much more contractionary than allowed under the program. In part, this resulted from institutional inflexibilities in using fiscal policy in a countercyclical manner, in the absence of preexisting social safety nets that would automatically be activated in an economic downturn. The failure of the authorities to use all the fiscal room provided in the program also reflected the fiscal conservatism of the Ministry of Finance and the limited implementation capacity of the Indonesian government in general. The absence of a government bond market also limited the ability of the authorities to finance expenditures through noninflationary means, imposing another constraint in operating fiscal policy countercyclically. Thus, the main countercyclical element realized was on the revenue side, as the targeted increases in spending were not met (Table A1.1).

In retrospect, the IMF was slow to recognize that the decline in GDP was being driven in large part by the collapse in investment. In April 1998, when the sharp contraction in investment should have been clear, the staff report for the first review simply noted that economic activity had fallen off “markedly” during the second half of 1997/98, “especially in construction and services,” without mentioning the behavior of private investment. It is only in August 1998 that this feature was noted and the EFF request projected a remarkable decline of private investment from an estimated 22.5 percent of GDP in 1997/98 to 9.2 percent of GDP in 1998/99. Even then, there was no explanation of why investment had collapsed to this extent, suggesting that the IMF may not have focused sufficiently on one of the key forces driving the adverse macroeconomic outcome.

Interest rate policy and monetary targets

Contrary to the widespread image that the IMF mechanically pushed for high interest rates, internal documents make clear that there was in fact considerable debate among staff on the best way to deal with the situation. The staff was fully aware of the basic dilemma: a large exchange rate depreciation would bankrupt many firms (and thereby adversely impact the banking system), while any interest rate high enough to support the exchange rate was also likely to have similar adverse effects on balance sheets.

Interest rate policy

The Policy Development and Review Department (PDR) and MAE argued for tight monetary policy with high interest rates. PDR argued that the corporate and banking sectors could not bear the added costs from any further depreciation, and recommended foreign exchange market intervention supported by tight monetary policy. Interest rates were to be raised temporarily at the outset of the program to signal the commitment of the authorities to exchange rate stability and to encourage nominal appreciation of the exchange rate following the intervention. MAE supported high interest rate policy to achieve an early exchange rate appreciation, but expressed reservations on the benefit of extensive early foreign exchange market intervention.

On the other hand, RES and APD argued against further tightening monetary policy and raising interest rates. RES was concerned that an interest rate defense was not feasible with a weak banking system and a vulnerable corporate sector. It pointed out that if confidence remained low, the agreed intervention limits would be reached and higher interest rates would be required to defend the exchange rate. But higher interest rates would damage the corporate and banking sectors, thereby further eroding confidence.

During the program negotiations, the APD mission argued that it would not be desirable to support the exchange rate solely through monetary tightening, especially because monetary conditions were already tight. Instead, it advocated a policy of giving the authorities more flexibility to intervene when necessary, without further tightening monetary conditions. The mission also pointed out that, on a practical level, BI was reluctant to raise SBI rates, when it had already done so unsuccessfully in August 1997. The mission noted that, as early as September, the central bank Governor had begun to reduce interest rates and was still talking in terms of further reducing the rates.

The business community in Indonesia was calling for lower interest rates, and market participants were discussing the problems associated with maintaining high interest rates for a long period. By early September 1997, market commentary was suggesting that the balance sheets of Indonesian firms had been severely damaged by high interest rates and the weaker exchange rate. By the end of the month, tight liquidity was a serious concern for the banking sector, as the banks’ portfolios had deteriorated rapidly as a result of their exposure to corporate borrowers with a large amount of unhedged foreign currency–denominated debt.11

The differences between RES and APD, on the one hand, and PDR and MAE, on the other, reflect the dilemma of designing crisis management policies in the face of a twin crisis affecting both the external sector and the banking sector, with policies aimed at addressing one problem causing problems in the other. However, while the problem was posed, there was no satisfactory way of resolving the dilemma. The policy that finally emerged from the debate represented a compromise: to keep monetary policy tight without setting specific interest rate targets. BI would maintain the one-month SBI rate at 20 percent but would raise it if needed to support foreign exchange market intervention. In approving the program, no Executive Director explicitly opposed the strategy; several Directors, however, expressed strong dissatisfaction with the lack of specific and sufficiently tight monetary action.

Less than a week after the program was launched, the staff was alarmed by the apparent loosening of monetary policy reflected in a fall in interbank rates and urged BI not to lower interest rates prematurely. Initially, during the first week of November, the rupiah had appreciated from Rp 3,600 to Rp 3,250–3,300 per U.S. dollar, supported by coordinated foreign exchange market intervention (with Japan and Singapore), and the Jakarta interbank offered rate (JIBOR) began to rise. These gains, however, were not supported by sustained high interest rates, with the SBI rate remaining virtually constant (Figure A1.2).12

Figure A1.2.
Figure A1.2.

Indonesia: Monthly Nominal Interest Rates

(In percent a year, at the end of the month)

Sources: Bloomberg; and University of Indonesia, Institute for Economic and Social Research.

BI argued that JIBOR was not a good measure of the stance of monetary policy. The interbank market had become more segmented than usual between foreign and state banks with adequate liquidity positions, on the one hand, and private banks with increasingly difficult liquidity positions, on the other. BI was urging first-tier banks to lend to other banks with assurances that there would be no second round of bank closures. At the same time, BI was providing liquidity to second- and third-tier banks at a rate lower than JIBOR. Staff was concerned that the injection of liquidity might cause monetary targets to be breached. In the second half of November, a mission was dispatched to assess the situation.

The strategy of intervening in foreign exchange markets presented a further complication, given the already tight liquidity situation caused by the monetary squeeze of August.13 Intervention of some US$5 billion in the last quarter of 1997 was equivalent to one-third of the stock of base money at the end of September 1997. As the intervention was to be only partially sterilized, this left a large segment of the banking sector short of liquidity when settlement came. BI claimed to have no alternative but to provide liquidity but, as a result, the rupiah only strengthened for two days before sliding, by the end of November, to its level of October.

With the exchange rate sliding almost continuously, it was clear that the original expectation of a quick recovery would not be realized. The IMF urged an immediate rise in the SBI rate by 5 percentage points as a first step and by more if needed, in accordance with the understanding on which the program was based. The IMF also urged that, as agreed in the program, liquidity support should only be offered at market rates and against collateral and that additional banks should be closed if necessary. However, President Suharto ordered an immediate reduction of 5 percentage points in the SBI rate (which the economic team did not implement). He also signaled that there should be no more bank closures. With conflicting demands on monetary policy coming from the IMF and the President, the economic team by this time had all but lost access to the President and could take no effective action.

Our evaluation suggests that the criticism that the high interest rate policy pushed by the IMF was responsible for the collapse in Indonesia is not well founded for the simple reason that the IMF’s recommendations in this respect were never implemented. Interest rates were not raised despite repeated IMF urging. Instead, liquidity was expanded and resulted in a loss of monetary control (Box A.1). As a result, real interest rates were substantially negative (Figure A1.3). It was only after March 23, 1998 that the new economic team was able to raise nominal interest rates, pushing up the one-month SBI rate to 45 percent from 22 percent. The exchange rate steadily appreciated from Rp 9,750 per U.S. dollar the previous week to Rp 7,500 by mid-April and remained below Rp 8,000 until the May troubles, which provoked a further depreciation (Figure A1.4).

Figure A1.3.
Figure A1.3.

Indonesia: Monthly Real Interest Rates

(In percent a year)

Source: University of Indonesia, Institute for Economic and Social Research.1 The lending rate is for working capital, deflated by actual two-month-ahead CPI inflation.
Figure A1.4.
Figure A1.4.

Indonesia: Daily Movements of the Rupiah–U.S. Dollar Exchange Rate

Source: University of Indonesia, Institute for Economic and Social Research.
Monetary targets

Performance criteria for base money were set for end-December 1997 and end-March 1998, and indicative targets for end-June 1998 and end-September 1998. Base money was to grow by 4 percent in the last quarter of 1997 and to remain more or less flat in the first quarter of 1998. In the event, unlimited liquidity support from BI to the banking sector led to a virtual explosion in base money, which grew by 14 percent in the last quarter of 1997 and a further 32 percent in the first quarter of 1998, before its growth slowed down to 12 percent in the second quarter and finally to 2 percent in the third quarter (Figure A1.5).

Figure A1.5.
Figure A1.5.

Indonesia: Base Money Outcomes and Targets Under IMF-Supported Programs

Source: IMF documents.

While central bank liquidity support expanded sharply during the IMF-supported program, BI was already providing lender of last resort (LOLR) support to several banks experiencing shortages of liquidity well before the program. As the crisis developed, LOLR support was provided under a variety of schemes, which were later consolidated under the general title of Bank Liquidity from Bank Indonesia (BLBI) early in 1998. With the greater segmentation of the interbank market in the final quarter of 1997, the LOLR role of BI became all the more important. By the end of January 1998, total support under BLBI had reached 5 percent of GDP, or close to 100 percent of base money.

BI operated under severe constraints. When a bank had a shortfall at clearing, BI had to either supply the needed liquidity, or else close down or take over the bank immediately (Djiwandono, 2002). In November 1997, the Cabinet had decided, in accordance with commitments under the program, to provide LOLR only to solvent banks, but both the BI Governor and the Minister of Finance were certain that the President did not want any more banks to close. Without willingness on the part of BI to intervene in some other way, these two objectives were mutually incompatible.

Indonesia: Was Monetary Policy Tight?

IMF staff has argued that monetary policy was never tight in Indonesia, because most standard measures of real interest rates were negative from the inception of the program to early 1999 (Lane and others, 1999; Boorman and others, 2000; Ghosh and others, 2002). It is true that, for the first five months of the program, the Indonesian authorities hardly raised the policy interest rate despite urging from the IMF. It was only in March 1998 that, for the first time under the program, BI substantially increased the SBI rate. The one-month rate rose from 22 percent to 45 percent and reached, after several rounds of increases, 70 percent in August 1998.

The assessment of monetary policy under the IMF-supported program is made difficult by several factors:

  • Before IMF assistance was requested, in August 1997, BI had already raised the one-month SBI rate from 10–12 percent to more than 30 percent. However, under pressure from the President, BI was forced to reduce the rate to around 20 percent in September 1997.

  • With a sharp depreciation of the currency, relative prices in the economy were rapidly changing and the impact of interest rates was different in tradable and nontradable sectors. Real interest rates faced by the nontradable sector likely remained positive—and substantially so—during this period, while they were substantially negative for the tradable sector.

  • The banking crisis led to a greater segmentation of the interbank market with a shift of deposits within the banking system. At least initially, 24 of the major institutions—the so-called JIBOR banks—had plentiful liquidity, while other banks found it difficult to raise funds at any interest rate. The high nominal interest rates faced by these banks reflected a large risk premium, not a particular stance of monetary policy.

  • BI provided liberal liquidity support to all banks experiencing liquidity problems, so that high interbank interest rates did not present an issue for these banks.

  • Continued pressure on the rupiah meant that Indonesian interest rates included a component reflecting the expected rate of depreciation.

It is fair to say that while high real interest rates were faced by some potential individual borrowers at different points in time, the stance of monetary policy as a tool of macroeconomic policy was never tight and, contrary to the wishes of the IMF, did not become any tighter as a result of the IMF-supported program. Moreover, market segmentation, always a feature of the Indonesian system, worsened markedly and intermediation spreads in the banking system became negative as banks attempted to keep payments current. There was, however, a period of tight monetary policy prior to the inception of the program which, according to the BI Governor and market observers, had adverse consequences for the corporate and banking sectors.

A related issue is whether or not high interest rate policy caused a credit crunch, a situation where existing demand for credit is not fully satisfied at a given interest rate. In the case of Indonesia, as banks experienced liquidity and then solvency problems, the supply of credit clearly fell. At the same time, as the balance sheets of many firms were adversely affected by the sharp depreciation of the rupiah, the number of creditworthy borrowers also declined. To identify a credit crunch is inherently a difficult exercise, because it requires the identification of both demand and supply. A study by IMF staff argues that there was a credit crunch in Indonesia as the banking crisis deepened, but that the crunch disappeared when the demand for credit fell (Ghosh and Ghosh, 1999). The aggregate picture, however, may not tell the whole story about potential individual borrowers, particularly small and medium-sized enterprises (SMEs) with no recourse to nonbank financing (Yoshitomi and Ohno, 1999). There was evidence of some unsatisfied credit need, mainly reflecting supply factors (Bank Indonesia, 2001). Given the likely impact on the ability of banks to provide financial intermediation, a strategy to deal with the financing needs of viable SMEs would have been helpful, although it is inherently difficult to design such a strategy.

In this climate, liquidity support served both legitimate LOLR and fraudulent purposes. Together with third-party depositors withdrawing funds in a “flight to safety,” some bank owners were stripping assets. In parallel, liquidity support also went to cover large off-balance-sheet exposures in foreign exchange. This was particularly evident in early 1998, when the exchange rate plummeted and the banks could no longer borrow foreign exchange in the interbank market. This led to an explosion in liquidity support during that period. The increasingly negative intermediation spreads, as banks tried to keep payments current, added to insolvency and illiquidity that contributed to a buildup in liquidity support.

By the time the situation stabilized in mid-1998, the volume of liquidity injected through BLBI amounted to around Rp 144 trillion (or 14 percent of GDP). In the initial phase, penalty rates were imposed on BLBI, which were then capitalized, leading to a steady rise in the outstanding volume of BLBI. When it was recognized that this was not serving any purpose, the rates were reduced. As BLBI was unsecured, the bank owners were required to provide personal guarantees, which later became the basis of the shareholder settlements administered by IBRA.14

Once BLBI support became routine, moral hazard became real. According to the official report of the Supreme Audit Agency (BPK), irregular practices dominated the administration of BLBI, with Rp 82 trillion out of total Rp 144 trillion judged to have been misused.15 It should be noted that the report took a legalistic approach and thus characterized any violation of central bank rules as fraudulent, which likely overestimated the economic cost of corruption. However, it is certain that BLBI not only raised the cost of saving the banking system, but also contributed to greater exchange rate depreciation by effectively funding capital flight.

Almost all of the BLBI went to private banks, except for the special case of the state-owned Bank EXIM. The liquidity support to Bank EXIM was not in response to deposit withdrawals but rather to fraudulent losses in the bank’s treasury operations. BLBI was concentrated in a handful of institutions, with EXIM and three private banks (BCA, Danamon, and BDNI) receiving 75 percent of the total. This concentration of BLBI implies that pressure was not necessarily on the overall financial system. The case of Bank EXIM is particularly noteworthy, as state banks benefited from the shift of deposits from the private banks (given the implicit deposit guarantee by the government).

Interviews with staff and a review of internal documents make clear that the staff was not fully aware of governance problems in the injection of liquidity until January or February of 1998. Thus, although the IMF staff was in daily contact with the authorities and monitored the amount of liquidity support, the IMF did not capture the extent of irregularities in the support operations during the crucial months of November and December, when monetary control was lost.

Exchange rate policy and capital controls

The rupiah was floated in August 1997 at the outset of the crisis before the IMF program was negotiated, and this decision was welcomed by the IMF. Nevertheless, in view of sustained downward pressure on the rupiah, the IMF staff, during the review of the brief for the October 1997 mission, discussed the idea of introducing capital controls. The idea was quickly dropped because of the likelihood that controls could not be administered effectively in a country with widespread corruption and weak administrative capacity. The Indonesian authorities told the evaluation team that they had never considered introducing capital controls, knowing that there was no infrastructure to administer such a system effectively. They also pointed out that one of the reasons for abolishing controls in the 1970s in the first place had been their ineffectiveness due to corruption.

By December 1997, the rupiah had depreciated substantially more than the currencies of the other crisis-hit economies of the region, and was continuing to depreciate, indicating that the Indonesian crisis was exceptional. In part, this reflected political developments. The illness of President Suharto in early December injected new sources of uncertainty as succession concerns surfaced prominently, and politically motivated attacks on the ethnic Chinese community also intensified.

With the currency in virtual free-fall from December through January, even after the signing of the revised LOI, both the IMF and President Suharto independently began to consider introducing a currency board arrangement (CBA). In Indonesia, business interests close to the President initiated the idea and invited an American academic expert to advise on the subject (Hanke, 1998b). The idea of formally introducing a CBA was declared by the President in February 1998. There was widespread though unsubstantiated concern, including within the IMF, that if the CBA were adopted, the rate would be Rp 5,000 per U.S. dollar, around half the going market rate, and that its supporters would use it to convert their rupiah holdings into U.S. dollars.

There were some advocates for the CBA within the IMF, but a consensus soon emerged that the existing conditions in Indonesia, including the weak banking system and the absence of respect for rule of law, were not appropriate for a CBA, at least over the short to medium term. On February 11, the IMF took a firm stance on the issue by sending a letter to the Indonesian authorities opposing the CBA and explaining why it was not appropriate for Indonesia at that time. A stalemate continued until the major IMF shareholder governments, including Germany, Japan, and the United States, stated their unequivocal opposition, through high-level contacts with President Suharto.

Official financing

As noted in the main report, determining the size of access in a program designed to build confidence is an inherently difficult exercise, because the residual financing need is endogenous to the effectiveness and speed with which confidence is restored. This also makes difficult our evaluation of the size of access in Indonesia, which was based on a projection of the likely balance of payments need under certain assumptions.

The IMF assumed that the current account deficit in 1997/98 would show a small improvement of about US$2 billion compared to the previous year, but this would be accompanied by a large deterioration in the capital account of about US$14 billion, reflecting failure to rollover short-term debt, withdrawal of portfolio investment, and lower net FDI flows (Table A1.2). The program also aimed to stabilize the level of gross foreign assets of BI at about US$26 billion.

Table A1.2.

Indonesia: Balance of Payments Projections and Outcomes

(In billions of U.S. dollars)

article image
Sources: IMF Staff Reports; and IEO staff estimates.

Given these assumptions, the IMF determined that an amount equal to one-third of the short-term debt of US$33 billion (i.e., US$11 billion) would need to be financed over the two years 1997/98 and 1998/99. In calculating access, however, it used the more conservative figure of US$22 billion (or two-thirds of the total short-term debt) as the amount that was required to meet short-term obligations over the first year of the program. Access from the IMF was thus set at US$10 billion (490 percent of quota), after taking account of additional multilateral financing of about US$8 billion from the World Bank (US$4.5 billion) and the ADB (US$3.5 billion), and the use of US$5 billion of BI’s own reserves if needed.16 Of the US$10 billion to be provided by the IMF, US$8.7 billion was to be disbursed over the first two years, with US$6.1 billion for 1997/98 and US$2.6 billion for 1998/99.

The program also incorporated a substantial foreign exchange market intervention of up to US$7.5 billion over the first three months of the program, with up to US$5 billion during the month of November alone. In the event, the improvement in the current account was much larger, at US$6 billion, and the reversal of capital flows was much worse than projected. Compared with the net inflow of some US$14 billion in 1996/97, the November program had projected a net outflow of US$0.5 billion for 1997/98. The actual outcome was a net outflow of some US$12 billion in 1997/98, including capital flight by domestic residents.

The working assumption that only one-third of the short-term debt would be rolled over was not unreasonable, as were the rest of the balance of payments assumptions. In retrospect, the projections were belied by large-scale capital flight by domestic residents, which became ever larger over time. As a result, what had seemed a reasonable package ex ante began to look inadequate as confidence collapsed.

In our view, the size of financing was not the cause of the failure of the November program. The origin of the failure was the inadequacy in program implementation and the associated rapid expansion of liquidity, and this technical failure was soon transformed into a political crisis, which undermined business confidence especially among the ethnic Chinese business community. At the technical level, the main oversight was the failure to take into account the unknown but large amount of short-term interbank lines of credit essential to finance imports. Trade credits were not rolled over and this exacerbated the crisis until the spring of 1998, when explicit efforts began to be made by the IMF and its major shareholder governments to encourage major commercial banks to do so.

Bank closure and restructuring

The need to reform the banking system had been identified in surveillance and measures to this effect were rightly included in the program. As noted in the main report, in October 1997, the MAE team, collaborating with teams from the World Bank and the ADB, examined the supervisory data provided by BI and concluded that at that time intervention was needed for only a limited number of private banks. This assessment turned out to be a serious underestimation of the true state of the banking sector. The reality at the time was that, except for foreign banks, state banks, and a few large private banks, much of the rest of the banking system was illiquid and possibly on the verge of insolvency.17

The IMF reached its assessment in the following manner. Using the June 1997 data, the World Bank reviewed all 7 state banks (accounting for 40 percent of total banking sector assets); the ADB, 13 out of 27 regional development banks (2 percent of total banking sector assets); and MAE, 72 out of 160 private banks (43 percent of total banking sector assets and 87 percent of total private banking sector assets). Taken together, the combined IFI team investigated 92 out of 238 banks, accounting for 85 percent of market share.

Exclusive reliance on BI data proved to be a major problem for two reasons. First, the June 1997 data were not the right basis for making solvency assessments, given the exchange rate depreciation that had occurred since then. Second, supervisory information from BI was flawed by the low level of supervisory skills and, according to some observers, suspicions of corruption. This was clear from a widely known academic work (Cole and Slade, 1996) as well as from the findings of the World Bank’s financial sector mission in 1996. The IMF staff did go beyond official data and asked the heads of large banks how the crisis had affected their balance sheets and also discussed the likely current balance sheets of banks with BI supervisors, bank by bank. However, these inquiries did not in most cases lead to a significantly more negative assessment.

The combined team identified 50 vulnerable banks, of which 34 banks were judged insolvent, including 26 private banks, 2 state banks, and 6 regional development banks. Another 3 private banks were on the borderline of solvency, requiring rehabilitation. The remaining 13 (out of the 50 vulnerable) banks were found to have diverse weaknesses, including capital adequacy ratios below the required minimum for some, and needed to be placed under intensified supervision. According to MAE, the 34 banks identified as insolvent accounted for about 15 percent of total banking sector assets, with the 26 private banks alone accounting for 5 percent.

The extent to which the IMF missed the scale of the problem is obviously crucial in making an ex post evaluation. Internal documents and interviews indicate that there was a considerable debate within the staff over the extent to which Indonesia faced a systemic banking problem. Some APD staff argued that the MAE analysis was too sanguine because it assumed that (1) there were a relatively few bad banks in an otherwise sound banking system, when the whole banking sector had become vulnerable as the exchange rate had depreciated and interest rates had risen; and (2) runs were caused by small and ignorant depositors, while it was in fact the high-wealth individuals with inside information who were withdrawing deposits.18 However, these concerns were downplayed and therefore not reported in the staff report accompanying the November SBA request to the Executive Board. MAE insisted until January 1998 that the banking system was sound except for the 50 banks identified, and that no data existed to support the contrary view. Even so, the MAE mission did note in its back-to-office report, dated November 11, 1997, that there might be other problem banks than the sample reviewed; NPLs might have been underestimated; and some banks not identified for action might have deteriorated since June 1997.

In any case, it is unlikely that identification of deeper sickness would have led to corrective action. BI argued that it could only close 16 of the 26 insolvent private banks (accounting for only 3 percent of total banking sector assets) because the other 10 had “nursing” agreements with BI, which legally prevented closure unless rehabilitation efforts failed.19 Among the banks to be closed were three connected with the President’s family: Bank Andromeda, in which one of his sons had a minority ownership; Bank Industri, with partial ownership by a daughter; and Bank Jakarta, with some ownership by his half-brother.

A critical program design decision was the nature of a guarantee for depositors of closed banks. There was a consensus between the authorities and the IMF staff that a blanket guarantee would not be desirable on grounds of both fiscal cost (emphasized by the Indonesians) and moral hazard (emphasized by the IMF). It was agreed that depositors of the closed banks would receive up to Rp 20 million (about US$6,000), covering 93 percent of the accounts and 20 percent of the deposits in the closed banks.

Initially, the closure of the 16 banks and the tough statement from the Minister of Finance that henceforth all banks allowed to become insolvent by their owners would be closed down was welcomed, as it seemed to imply a new way of doing business. However, several factors undermined the credibility of this policy. Most important, the President’s family challenged the closures. His son arranged for the business operations of Bank Andromeda to be shifted to another bank in which he had acquired an interest. The President’s half-brother initiated a legal challenge to the closure of his bank. The public also saw some inconsistency in the closure of 16 banks, when it was widely—and correctly—believed that many other banks were also in a similar condition. The authorities insisted on secrecy regarding the nursed banks and, as a result, the public had no idea of what was being done to address the wider problem.

BI also did not make an adequate effort to communicate its bank-closure policy to the public. There were flip-flops in announced government policy. Under pressure from the President, the Minister of Finance soon reversed his previously announced tough position, saying that there would be no more bank closures. Some private individuals told the evaluation team that uncertainty had been compounded by lack of clear information on how and how quickly depositors would have access to their funds. In the event, by the end of November 1997, two-thirds of the 222 banks had experienced runs. Rp 12 trillion (or about US$2.7 billion) of rupiah deposits shifted to large private banks, foreign banks, and state banks, and about US$2 billion of U.S. dollar funds left the banking system entirely.

It was not until the end of January 1998, in the face of continuing banking sector problems, that the authorities accepted the banking strategy proposed by the IMF, involving a comprehensive bank restructuring plan, a general guarantee scheme, and the creation of the IBRA as a combined bank-restructuring and centralized-public-asset-management agency. The new strategy initially succeeded in stemming the exit of deposits from the banking system, and the appreciation that followed the announcement of the end-January banking and corporate debt measures was not fully reversed for almost four months, until the ethnic riots in May 1998.

The negative experience of November 1997 can be contrasted with what happened in early April 1998, when 7 banks representing 16 percent of banking sector assets were taken over by the IBRA and another 7 smaller banks were closed. The April 1998 operation differed from the November 1997 action in the following ways: (1) the existence of better arrangements for meeting depositors’ claims and a professionally managed public relations campaign designed to calm the public; (2) an assurance that the interventions were based on uniform and transparent criteria and that no banks failing these criteria were excluded; (3) a full guarantee that covered all deposits, as well as all liabilities in other banks; and (4) the existence of a comprehensive banking sector strategy within which the operations were carried out. The failure to have all these elements in place in November 1997 was a major factor contributing to the deepening of the crisis. While the IMF alone was not responsible for this failure—since the unwillingness of the government at the highest level to back key parts of the strategy was also critical—it does point to important lessons (see also the discussion in the main report).

Many, including IMF staff, have increasingly come to accept the view that the decision not to install a blanket guarantee was the critical mistake of the November 1997 bank closure (Lindgren and others, 1999). However, the question of a blanket guarantee, particularly in the context of Indonesia, requires careful consideration.20 In November, bank runs were associated with a shift of rupiah deposits from weak private banks to foreign banks, state banks (with an implicit guarantee), and some large private banks, with no decline in the assets of the banking sector as a whole. Large withdrawals from the banking system from the start of the crisis reflected the running down of foreign currency deposits.21 It is only with the presidential succession crisis in May 1998 that the real value of rupiah deposits began to decline, owing to a loss of confidence in the banking system as a whole. At that time, the blanket guarantee could do little about the crisis of confidence in the entire economic and political system (Booth, 2001), let alone the ability of the government to honor that guarantee.


The need to reverse the creeping increase in rent-creating regulation over the past several years had been identified as a major issue by the World Bank and also in IMF surveillance. It was also on the agenda of the reformist economic team and had frequently been advocated by commentary in the local press. IMF management also viewed the program as an opportunity to assist the reformist team in pushing desirable reforms and the team viewed the program as providing leverage to do so.

Internal reports and interviews with staff indicate that, as the negotiations progressed in October 1997, the mission was under increasing pressure from Washington to include structural measures directed at dismantling the system that had given rise to extensive rent-seeking and cronyism in Indonesia. In part, this reflected the prevailing atmosphere of domestic politics in some of the major shareholder countries, where support was lacking for a large financing package without addressing the increasingly well-known governance issues in Indonesia.

Although several deregulation measures were included in the November program, a key feature of structural conditionality at this stage was the absence of both specificity and a clear timetable. Almost all agreed measures were general in nature and were to be implemented over the program’s three-year lifespan. This provided the reformists with the necessary leverage to pursue reform but gave them discretion to push when and where they felt they could achieve results. This feature of the November structural conditionality, however, was not well understood by the public because, as was customary at the time, the LOI was not published.22 Without access to the LOI, the public began to speculate on the content of structural conditionality in the November program. Given the press references to certain deregulation measures, this led to an excessive focus on governance-related measures in public debate.

The failure of the initial program, combined with frustration over the lack of progress in structural reform, led to increased emphasis on the need for reforms as a key element of the strategy to restore confidence. Some of the IMF’s major shareholders pressed for greater specificity in structural conditionality. At the time of the Executive Board meeting on November 5, 1997, several Executive Directors had expressed their unhappiness with what they regarded as the vague and general nature of the structural conditionality, arguing that no progress would be likely in needed reforms without specificity and a clear timetable. The lack of progress in structural reform under the initial program reinforced their sense of misgiving.

This led to a much more specific and time-bound approach to structural conditionality in the January 1998 program. The World Bank’s Jakarta-based staff took the lead role in drafting the structural conditionality for the January LOI, and the IMF team went out of its way to ensure that all concerns of the Bank were fully met. By this time, the Indonesian economic team had all but lost direct access to the President (Boediono, 2001). Negotiations were carried out directly with the President, at his own request. On the IMF side, the First Deputy Managing Director was personally engaged in finalizing the understandings with the President.

Contrary to what the IMF had expected, President Suharto did not openly oppose the expansion of structural conditionality or the inclusion of specific measures, including the cancellation of the National Car Project in which his son was involved. Indeed, President Suharto publicly signed the revised LOI in an attempt to indicate his commitment publicly. However, the President’s opposition was expressed in other ways. The President is reported to have said in a high-level meeting of his advisers that not all agreed measures needed to be respected, and that he would “wage a guerrilla war against the IMF.” Later, he expressed the view that some of the reforms violated the Constitution. In February 1998, the staff reported in a memo to management that “all of the deregulation and liberalization measures relating to wood, cloves, BULOG, palm oil, wholesale and retail trade, and interregional trade [were] being subverted by various groups close to the President.”

The inclusion of extensive governance-related structural measures in the IMF-supported programs with Indonesia has been widely criticized as having been counterproductive in dealing with a financial crisis (Feldstein, 1998). A former U.S. Federal Reserve Chairman, during his visit with the President in early January 1998, is reported to have criticized the structural conditionality as irrelevant to financial stabilization by facetiously calling the conditions on marketing deregulations in cloves, oranges, and other foodstuffs a “recipe” (Kenward, 2000; Blustein, 2001).23 Likewise, a high-ranking Indonesian official remarked that “things might have turned out differently” if the conditionality had been confined to the macrocritical areas more relevant to dealing with the crisis, including comprehensive bank restructuring (Boediono, 2001).

In assessing these criticisms, it is important to recognize that structural conditionality became a seriously contentious issue only in January 1998. It was not the cause of the failure of the November program, which had more to do with the nonperformance of conditionality relating to bank restructuring and monetary control. In the wake of the collapse of the November 1997 program and the accelerated currency collapse in December, the IMF and officials of some key shareholder governments came to believe that more extensive structural conditionality was the only hope of restoring confidence by signaling a decisive break with the past, a view shared by some members of the academic community (Frankel, 2000; Goldstein, 2002) and the press (Financial Times, January 14, 1998).

The problems with the structural conditionality in the January 1998 LOI concern the lack of focus and ownership of the reform program, rather than its intrinsic usefulness to the Indonesian economy or the capacity to implement it. First, a number of the structural measures were popular with the public and did have beneficial effects on the economy when they were implemented. According to recent academic research, for example, the dismantling of monopolies and monopsonies, implemented from late January, substantially raised the farm-gate prices of major agricultural crops, and, as the IMF had hoped, helped minimize the adverse impact of the crisis on poverty (Montgomery and others, 2002). However, the program clearly did not benefit from ownership at the time it was announced and the ready perception of this lacuna made it completely ineffective. Second, the government’s capacity certainly was not a binding constraint in the implementation of structural conditionality (Boediono, 2001). This is borne out by the fact that once the new Cabinet installed in March 1998 had convinced the President that there was no alternative to the IMF-supported program, the “50-point” program announced in January began to be implemented more fully.

The January LOI also failed to impress the markets because it did not simultaneously address the key macrocritical issues of bank and corporate debt restructuring. In this respect, the focus on extensive structural conditionality in areas outside the concern of the IMF can be said to have distracted attention from some core reforms that were indeed macrocritical.

Corporate debt restructuring

In early October 1997, before the negotiations began, PDR had expressed concern that uncertainty about the size of private sector short-term debt was not being addressed, and had suggested action on corporate debt, including the creation of a mechanism to identify firms needing assistance. However, because the IMF lacked expertise in this area, and given the optimism that the program would rapidly restore confidence, the IMF-supported program did not actively address the corporate debt issue until January 1998. The World Bank was also slow to get involved and it was only in the middle of 1998 that it began to assume a major role in supporting the dialogue between creditors and Indonesian conglomerates. The slow start on corporate debt restructuring partly stemmed from the authorities’ view that the issue should be left largely to the private sector.

Starting in January 1998, the IMF provided technical assistance to a Private External Debt Team (PEDT). This had been set up in late 1997 as a voluntary initiative with the encouragement of the Indonesian authorities to provide a framework for the negotiations between creditors and corporations unable to service their debts. The role of the government was only indirect in this framework, and was limited to strengthening the legal and regulatory mechanism to enforce contracts. The debtors set up a committee to work with the PEDT but made it clear that little progress could be made without stronger government involvement, including financial support.

In the second half of March, a consensus emerged between creditors and the PEDT that some limited government involvement was necessary in the form of an exchange rate guarantee similar to that used in Mexico’s so-called FICORCA scheme.24 This position was endorsed by the IMF, with the caveat that there should be no subsidies to the corporate sector, a position shared by the authorities. The proposed voluntary approach aimed to protect debtors and creditors against exchange rate risk and to give assurance that foreign exchange would be available for debt-service payments in return for the restructuring of debt on specified minimum terms. Negotiations would seek to limit the exposure of the government to exchange rate risk.

In June 1998, adapting the FICORCA-type scheme to the conditions of Indonesia, a framework for the voluntary restructuring of debt was agreed in Frankfurt, and the Indonesia Debt Restructuring Agency (INDRA) was set up in August. Several problems remained, however. First, there was a need to reform the regulatory and legal framework, including removing restrictions on debt-to-equity conversions, eliminating tax disincentives for restructuring, streamlining approval procedures for FDI, a new arbitration law, and measures to provide for the registration of collateral. Second, the insurance provided by INDRA against further exchange rate depreciation was not attractive to many market participants, given the extent of exchange rate depreciation that had already occurred, for which market participants wanted some compensation. Third, as debt restructuring would take time, firms would remain short of working capital. Fourth, given the financial condition of many enterprises belonging to conglomerates, there were strong incentives for asset stripping by shifting assets to those entities better sheltered from the creditors.

On September 9, 1998, a “Jakarta Initiative” was finalized and became operational a month or two later. The initiative provided a framework to promote voluntary restructuring of debt through INDRA and to complement the amendments to the bankruptcy law aimed at providing incentives for debtors and creditors to negotiate. It included provisions for creditors to provide interim financing to distressed companies. Government involvement, however, was limited to the role of facilitator, including serving as a forum for the one-stop approval of regulatory filings. Despite all these initiatives, however, delays in implementing regulatory changes and difficulties in obtaining redress through the Indonesian legal system limited the progress of private sector debt restructuring. Well-placed interlocutors saw the failure to tackle the corporate debt issue as an important deficiency, as these debtors brought political pressure to bear on other issues. In this process, the IMF played a relatively limited role.

Initial strategy and its adaptation

Because the Indonesian crisis went through several phases, it is necessary not only to assess its conventional program design issues, but also to evaluate how effectively the IMF responded to emerging signs of failure and revised the initial program accordingly.

The initial strategy reflected the assumption that the crisis was a moderate case of contagion in which the rupiah had overshot. This view, which appears overly sanguine in retrospect, was widely shared by major market players at that time.25 Market insiders interviewed told the evaluation team that some important hedge funds had in fact been betting in favor of the rupiah at the time the program was being negotiated, indicating their expectations that the IMF-supported program could work. The strategy, however, was a risky one and the staff recognized that if the basic assumption that the rupiah had overshot and could be nudged back to a more reasonable level was questioned, an entirely different approach would be necessary. However, the staff never explored what this alternative might imply.

In this regard, in the light of the Mexican experience, one Executive Director representing a major shareholder government encouraged the staff to have a fallback plan. There is no evidence, however, to suggest that the staff either prepared or discussed a contingency plan with the authorities. While it is not realistic to expect the IMF and the authorities to negotiate a comprehensive alternative strategy when time is short and the ability to take key political decisions is limited, it should have been possible to identify at an earlier stage more comprehensive measures to deal with a bankrupt corporate sector and a systemic banking crisis, both of which were quite likely outcomes.26 In responding to emerging signs of failure in mid-November, the IMF was handicapped by the absence of an agreed fallback plan.

When the original program failed to restore confidence, the underlying assumptions of the strategy needed to be reassessed. In the latter part of November, a mission was dispatched to assess the situation and to consult with the authorities. However, the mission’s brief was largely focused on implementation within the logic of the original program and blamed the failure on nonimplementation. While the lack of implementation was undoubtedly part of the story, the original premises of the strategy were rapidly overtaken by events and there was a need for a more fundamental shift of strategy. The IMF’s continued attempts to push the unwilling Indonesian economic team to raise interest rates led to a public display of disagreement, which was not helpful to building market confidence.

A critical oversight was the failure to follow up on the close monitoring of BLBI undertaken by staff in the field. IMF staff was monitoring liquidity support bank by bank on a daily basis and keeping senior staff at headquarters informed. However, the IMF did not immediately take a firm position on the issue. For example, it did not press the authorities on the staff’s suggestion that BI should take control of banks receiving excessive support so as to prevent asset stripping. Given the culture of forbearance at BI and the lack of political support, little was done to contain the explosion of liquidity support. The IMF staff was prevented from knowing what was taking place within the recipient banks, particularly when collusion of some BI staff with bank owners was involved. Remedial action likely would have included a comprehensive intervention mechanism to deal with insolvent or illiquid banks, relying on the existing regulatory framework. In the event, it took the IMF staff four or five months to find out that corrupt and abusive practices were involved in the allocation of BLBI.

At the root of these problems was the lack of a fallback strategy to be pursued if the original somewhat sanguine assumption about an easy recovery of the rupiah proved misplaced. The IMF did revise the fiscal policy aspects of the program, but there was no reassessment of the underlying strategy itself. In particular, there was no comprehensive strategy to deal with the fundamental issues driving the crisis, namely, the collapsing banking and corporate sectors. While the issues were under constant review and various “Plan B” options were considered internally, existing differences of view within the IMF were not resolved until late January 1998.

In part, this delay reflected the lack of internationally accepted best practice in bank restructuring and the onset of a major crisis in Korea in late 1997, which took part of the attention and resources away from Indonesia. As a result, the IMF made a premature announcement of a package in mid-January, which focused heavily on deregulation and nonfinancial structural reform, but without including a comprehensive strategy to deal with banking system problem. With the benefit of hindsight, the signing of the second LOI should have been postponed for two weeks, to coincide with the announcement of comprehensive banking reform and corporate debt restructuring initiatives.

The Mode of Operations

This section discusses issues related to the IMF’s mode of operations, including country ownership, the decision-making process, human resource management, and the role of major shareholders and collaboration with the World Bank and the ADB.

Country ownership

Indonesia poses a paradox regarding country ownership. Management took the view that the IMF should support the reformist economic team because they shared common views of economic policy. Moreover, most of the reform measures were almost universally applauded within Indonesia, except by a small number of powerful elites.27 Nevertheless, the program failed because the key political authority, the President, did not buy into the reform process.

The IMF misjudged the commitment of the President and underestimated the pressures likely to come from his family and some of his influential associates. On several previous occasions, the economic team had received the full backing of the President to deal with economic crises and often successfully implemented the required reforms against opposition from powerful vested interests. With the increasing presence of the First Family and other competing stakeholders among the Indonesian elites, however, the economic team had lost much of that influence by the time of the crisis in 1997 (Booth, 2001). At the time of the crisis, this was well known to close observers of Indonesia.

The Indonesian economic team was very aware of its own limited influence in the country’s decision-making process. In part, this was precisely the reason why the team needed the leverage of an IMF-supported program to implement the reforms. Knowing its limitations, the economic team also made sure to secure the personal commitment of the President to measures agreed in the IMF-supported program. One can only speculate what outcome would have resulted, had the President not received the kind of opposition from his children and their close associates that he did in the last weeks of 1997, particularly following his illness in early December.

As it was, the program implied that firms and banks should be allowed to fail if they were insolvent. However, the President, under pressure from his children and close associates, was unwilling to let this happen. He also faced difficulty in allowing the structural reforms to go too far because they could undermine the very basis of his regime. According to some political observers interviewed by the evaluation team, the President wrongly came to view the IMF-supported program as an instrument of foreign powers seeking to undermine him.

How to secure ownership in such circumstances and what to do in its absence remains one of the unresolved issues arising from the Indonesian experience. To enhance ownership, the IMF did begin to recognize the need both to engage the President and to engage in a wider dialogue with various stakeholders. In January 1998, as noted, the First Deputy Managing Director visited Jakarta to negotiate directly with the President. Following the signing of the second LOI in mid-January, in which he himself participated, the Managing Director requested a retired member of management to serve as his personal representative to the President on an ongoing basis. The Indonesian team initiated conscientious efforts to talk to a wider group of people, both inside and outside the government. By then, however, the crisis had become largely political, overshadowing any consideration of ownership of economic policy.

Could a different approach have produced a better result? It is, of course, impossible to say. It could well be that no strategy would have been successful in separating the political and economic dimensions of the crisis. Nevertheless, a number of lessons on the ownership dimension do suggest themselves. First, an earlier assessment of the broader political economy issues underlying key elements of the program would have been useful. Second, a smaller set of structural measures that were fully owned could have reduced the scope for immediate implementation problems that damaged market confidence. Third, whatever the final judgments on ownership and the scope of the structural reform package, the January program should have included all of the measures judged macrocritical in order to be credible.

Decision-making process

In retrospect, it was probably a mistake to ignore the advice of PDR and the Resident Representative, and to rush the negotiation process in October 1997.28 The decision to rush was understandable, given the prevailing perception of a major regional crisis in Southeast Asia. However, Indonesia still had sufficient foreign exchange reserves to last for several months, as indicated by the fact that the program included use of Indonesia’s own reserves. The rushed procedure compromised quality in program design, particularly relating to the formulation of a comprehensive banking strategy and even possibly the assessment of insolvent banks, and prevented the IMF from fully benefiting from the safeguards of the internal review process. It is not possible to say whether a materially different assessment would have emerged from the established procedures.29 With less pressure, however, the IMF could have given greater time to examine the full implications of each policy option being considered, including a fallback option.

The rushed procedure had additional consequences. Management often worked directly with the mission in the field, bypassing the safety mechanism inherent in a bureaucratic organization. Some senior review department officers told the evaluation team that they had often felt sidelined and excluded in the decision-making process. Moreover, the Executive Board became involved in day-to-day and very detailed aspects of the program negotiations through informal sessions. Along with communications especially from major shareholders, this subjected the staff to considerable political pressure.

By the end of November 1997, the IMF had an urgent need to make a fundamental reassessment of its strategy. However, the IMF’s modus operandi, namely, short and intense country interactions, often with a pre-set and tight agenda, made it difficult for the staff to undertake such reassessment. Under the conditions prevailing in Indonesia at that time, the more permanent presence of a high-level team on the ground may have been beneficial as a mechanism for closely monitoring developments, providing timely policy advice and, if required, rapidly and smoothly modifying the strategy.

Human resource management

The Indonesian crisis, occurring as it did along with the other Asian crises, inevitably placed great strains on IMF resources and key decision makers within the institution. In many respects, the IMF responded very rapidly and with considerable flexibility. However, some aspects of the internal managerial approach, compounded by the IMF’s modus operandi discussed above, did have an adverse impact on the effectiveness of the response. First, management took some time to reallocate human resources to APD, whose staff was overstretched by the simultaneous crises in Indonesia, Korea, and Thailand. When the Korean crisis erupted a few weeks after the Indonesian SBA was approved, more of management’s attention and the institution’s available human resources were shifted from Indonesia. Some senior staff members have indicated that the simultaneous pressures on resources probably contributed to the delay in the reformulation of the program from December 1997 to January 1998.

Second, APD took time to mobilize experts to the field. Even after a banking expert had been identified, it took months before he was formally assigned as a Resident Representative in Jakarta. This appointment was made in May 1998, over six months after the banking crisis had come into the open.

Third, available internal knowledge was not effectively used in formulating the program. Part of this was an unfortunate outcome of the reorganization of the Asia-Pacific operations of the IMF in early 1997.30 The mission chief for the just-concluded 1997 Article IV consultation was not included in the mission that negotiated the program in October 1997 and had little input into the subsequent discussions on program formulation. Moreover, only a limited number of staff members of the first and subsequent APD missions had previous experience with Indonesia; the few with previous experience had not worked on the country for many years. This reflected a broader problem with excessive turnover of country teams within the IMF, as also noted in the IEO’s evaluation of prolonged use of IMF resources (IEO, 2002).

Fourth, financial sector expertise was not fully shared within the missions. No one from MAE was a formal member of the negotiating mission, and the MAE technical assistance mission worked side-by-side with, but independently of, the APD mission. This arrangement was costly because the views of individual members of the MAE mission were not necessarily brought to the attention of the negotiating team.31

Fifth, there was little rationale for splitting responsibilities without defining clear lines of command in the staffing of the October 1997 mission, which was simultaneously headed by two mission chiefs. With a separate MAE mission, this meant the presence of three mission chiefs with different channels of communication with mission members and senior officers in Washington. Likewise, in February 1998, a decision was made to alternate two missions with two separate mission chiefs. This arrangement, which lasted only briefly from February to March 1998, was an understandable attempt to create a permanent high-level presence on the ground without creating the family and other personal pressure associated with permanent relocation at short notice. However, despite cooperation between the two teams, such an arrangement was not ideal in terms of maintaining continuity during a crisis. According to some of the mission members interviewed, the mission chiefs had slightly different points of emphasis, and the transfer of information from one team to the next was inevitably incomplete. Some Indonesian officials interviewed told the evaluation team that they had often needed to repeat the same information twice.

The role of major shareholders and collaboration with the World Bank and the ADB
Major shareholders and the Executive Board

Broad agreement existed on the strategy for Indonesia among most of the IMF’s major shareholders who played an active role in the design of the program. Working through numerous informal sessions of the Executive Board, Executive Directors representing the major shareholders generally advocated tight fiscal and monetary policies and urged the adoption of structural reform measures aimed at improving governance. If there were dissenting views, they were not expressed at the formal Board meetings.32 Once the depth of the recession became clear, however, the Board supported the loosening of fiscal policy.

Frequent informal sessions facilitated a flow of information between the staff and the Board. Executive Directors could not only receive information on rapidly changing developments at these meetings but also express their views relatively freely. While the dissemination of information may not have been perfect, the informal sessions nonetheless provided the Executive Directors with opportunities to voice their inputs into the program at different stages. However, detailed involvement by the Board in specific elements of program design probably went too far. Although it was appropriate for the Board to define the policies and principles to be applied to the IMF-supported program, the staff and management should have been given greater freedom to pursue a strategy based on their judgment of country ownership, technical merits, and political feasibility. Detailed involvement by the Board or a subgroup of major shareholders appears to have added to the pressures for an extensive list of detailed structural reform and deregulation measures in the January and April 1998 programs.

The World Bank

Management explicitly instructed staff to consult World Bank staff on program design, particularly regarding structural conditionality, and to cooperate closely in reviewing the financial condition of the banks. During the October 1997 mission, IMF staff was given a series of notes the Bank’s Jakarta-based staff had prepared for the authorities during August and September 1997, advising them on how to deal with the crisis. The IMF staff also formally requested the World Bank for comments on the proposed content of conditionality but received no written response. However, some of the World Bank staff, including a senior official of its Jakarta office, felt that the IMF was not fully drawing on their resources and expertise.

Early difficulties between the IMF and World Bank teams in Jakarta in part resulted from the differences in the way the two institutions operate. IMF staff members involved in the negotiations said that they had initially found it difficult to work with Bank staff when tasks needed to be performed with tight deadlines since, in their view, the operational approach of the Bank often did not fit with such a timetable. Bank staff felt excluded because it was not informed of or invited to policy discussions. By January 1998, however, the working relationship had improved markedly, and the Bank’s Jakarta team was fully involved in designing the structural conditionality of the revised program. Moreover, from late January 1998, the MAE team worked closely with its financial sector counterparts from the World Bank. World Bank staff participated fully, and was identified as co-authors in the series of reports prepared by the MAE staff during the crisis. As part of this close collaboration, the World Bank took the lead in the financing of the mid-1998 audits of the “IBRA banks.”

Despite the active involvement of World Bank staff in much of the program negotiations and design, dissenting voices were heard from the Bank’s Washington headquarters, and the Bank’s Chief Economist publicly criticized the IMF-supported program. To deal with precisely this type of situation, the IMF and the World Bank had earlier agreed, in the so-called Concordat on Fund-Bank Collaboration prepared in March 1989, on a general procedure to resolve differences of view on economic issues. The Concordat stipulates a five-tiered procedure, starting with working level staff and ending at the Executive Boards; each additional tier comes into play only after best efforts to resolve differences have failed at the previous level. On an ad hoc basis, moreover, it envisages the possibility of establishing a study group, under the direction of the IMF’s Director of Research and the Bank’s Vice President, Development Economics, to examine analytical issues that may arise in areas of shared interest.33 However, this procedure was not utilized to resolve the differences of view, in part because the differences did not follow a simple IMF-World Bank divide.


The relationship with the ADB was also difficult. Its participation was initially conceived in the context of a technical assistance mission, given its earlier work on regional development banks. As a consequence, once a decision to negotiate a program was taken, the ADB’s inputs, if any, were channeled through the MAE technical assistance mission. In addition to examining the balance sheets of regional development banks, the ADB was put in charge of looking at the nonbank financial institutions regulated by the Ministry of Finance, and not by BI.

Citing confidentiality, however, the IMF staff did not keep the ADB team fully informed of issues being discussed with the Indonesian authorities. The relationship was cool at best and continued to deteriorate until the end of January 1998, when the ADB temporarily pulled out of the collaborative relationship with the IMF over disagreement on the creation of the IBRA. The first ADB program loan, for US$1.4 billion, was not approved until June 1998. Subsequently, working relationships were established again. ADB staff was involved in financial sector work with MAE, and took the lead in the audits of the “non-IBRA banks.”


This section provides a summary of major findings and our assessment of the role of the IMF in the Indonesian crisis, as reviewed in this annex.

Precrisis surveillance

IMF surveillance of Indonesia in the precrisis period had limited effectiveness in terms of both diagnosis and impact. Although it identified the key issues, it did not emphasize the risks and assess comprehensively the impact if these risks were to materialize. The weaknesses of surveillance were particularly evident in the underestimation of governance problems in the banking sector, and the failure to analyze the implications of risks and corruption in an explicit and candid manner. Data weaknesses also hampered the effectiveness of surveillance, although a more systematic effort to analyze the potential vulnerabilities would have highlighted these weaknesses earlier.

Regarding the banking sector problems, the IMF identified the key issues but did not take a strong enough position, perhaps owing to the judgment that the weaknesses did not pose a systemic risk in an environment of strong macroeconomic growth. The IMF was not alone in this failure. In fact, even some of the closest observers had a generally positive assessment of the Indonesian banking system, while being well aware of pervasive corruption (Cole and Slade, 1996). The staff was handicapped by prevailing conventions that required it to approach governance issues with obliqueness. Moreover, banking sector issues were identified as part of technical assistance work, a voluntary process in which the IMF acts as the authorities’ confidential advisor for their exclusive benefit. There was thus tension over how much of what was uncovered could be used to raise difficult questions during surveillance. Nevertheless, a more candid discussion of these issues in the Executive Board would have been helpful in highlighting the dangers of poor supervision, the moral hazard inherent in Indonesia’s banking policy, and the urgency of dealing with insolvent banks while conditions remained favorable.

The lack of candor in discussing the implications of vulnerable balance sheets and pervasive corruption was another area of weakness in precrisis surveillance. As early as 1995, internal reviewers, especially those in RES, had pointed out that the adverse impact of a shift in market sentiment for the corporate sector and its macroeconomic consequences in an economy with a weak banking system, but these concerns were not pursued by exploring their implications.34 As a result, the staff made only a limited attempt to collect data on corporate balance sheets.35 While it is unlikely (and impossible to test) that greater candor would have led to a marked change in the authorities’ policies, such a candid discussion would have allowed the IMF and the authorities to consider worst case scenarios in an atmosphere free of crisis.

The failure to present a candid analysis of the extent and nature of corruption in Indonesia led to unrealistic expectations about the ease with which reforms could be implemented and misled the IMF on the potential adverse short-run impact of the drive to deregulate. Corruption had always existed in Indonesia, but it did not prevent the economy from growing at an impressive rate over many years. This may have caused the IMF to overlook the changing nature of corruption in the 1990s, when both foreign and domestic investors began to focus on links to the Palace, rather than on the intrinsic economic merits of projects, in their investment decisions. By not openly discussing this aspect of the buoyant capital inflows, the IMF failed to perceive that Indonesia was particularly vulnerable to a sudden shift in investor confidence that might result, for example, from presidential succession concerns.

These weaknesses in part reflected a failure to take account of the wide range of views that might affect policy options and to grasp the broader political economy context within which presidential decisions were made. The surveillance dialogue placed too much faith in the ability of reformists to deliver policies, and failed to explicitly consider the various political constraints on policymaking. A focus on the reformist economic team was understandable. They had, after all, delivered important policy corrections during earlier crises and the IMF clearly has to interact primarily with its official counterparts. Nevertheless, staff could have sought informal inputs from a much wider set of people in order to obtain a broader sense of the political constraints for economic reform. The Resident Representative, who had significant local knowledge, could have been better integrated into the surveillance process. In practice, surveillance was largely conducted, with short country visits, by IMF staff in Washington.

Program design and implementation

The November program was based on a critical assumption that the crisis was a moderate case of contagion and that a program of tight macroeconomic policies and banking reform, supported by foreign exchange market intervention, would succeed in restoring stability with only a temporary deceleration in growth. This proved grossly optimistic as the rupiah depreciated uncontrollably, owing initially to implementation failures and later to political developments. The initial assumption that the crisis would be easily controlled was at best fraught with risk, given the possibility of multiple equilibria. These risks were underestimated because the extent to which the crisis was a twin crisis, with severe weaknesses in the banking and corporate sectors, was not recognized early enough.

Given the initial highly optimistic assumptions on growth, fiscal policy was not inappropriate. One can argue in retrospect that, given the low initial level of public debt, it was misguided to include in the budget the carrying cost of bank restructuring, as the cost could have been financed by a slightly higher stock of debt over the medium term. However, the banking sector presented large contingent liabilities for the government, so that there was in fact less room than the formal public debt figures might have suggested for a massively countercyclical fiscal policy. Indonesia also faced the financing constraints resulting from the absence of a government bond market and the inherent difficulty of financing expenditures with issuance of debt during a crisis. In the case of Indonesia, the only recourse the government had to financing expenditure was drawing down its deposits at the central bank and foreign borrowing. Use of central bank deposits would have been counterproductive when base money was already exploding with liquidity support to the banking sector. Foreign borrowing was not an option when foreign lenders were fleeing from the country. Thus, while initial tightening was not necessary—and should not have been part of the program if a more realistic estimate of short-term growth prospects had been incorporated—there was little feasibility for a markedly expansionary fiscal policy.

As the crisis evolved, fiscal policy was continuously relaxed and the targets were never operationally binding. The fiscal program in 1998 also included adequate social considerations, as subsidies were increased on essential goods, while price increases were targeted toward goods and services consumed by higher income groups.

Monetary policy was never tightened during the early months of the program, despite the urgings of the IMF to the contrary. Most reasonable measures of real interest rates became increasingly negative, because the monetary base was expanding out of control with the provision of unlimited liquidity support to the collapsing banking system. As part of this support was used to fund capital flight, it placed downward pressure on the rupiah. Exchange rate and price stability only returned when monetary policy was tightened and nominal interest rates raised in the spring of 1998. In this respect, the adoption of base money targets, rather than conventional NDA targets, was not helpful as it allowed intervention and liquidity to get out of hand.

More generally, quarterly targets for any quantitative measure of base money (or its NDA component, for that matter) proved to be of little operational use in monitoring the conduct of monetary policy on a day-to-day basis during the crisis. Base money, consisting largely of the public’s currency holdings, has a large endogenous component and is thus difficult to control in the short term, even under normal circumstances. During a banking crisis, base money is even more difficult to control, as there is a portfolio shift of unpredictable magnitude from deposits to currency. In the case of Indonesia, this difficulty was compounded by unlimited liquidity support, which caused base money to go out of control. A more direct discussion and explicit agreement on interest rate policy, as happened in the spring of 1998, along with a closer monitoring of the liquidity support operations, might have provided a better framework for monetary policy.

In this respect, a critical mistake in the initial strategy was to settle for an ill-defined “understanding” on interest rates without fully specifying what action would be required, given the unwillingness of the Indonesian economic team further to raise interest rates. This papering over of a fundamental disagreement about the appropriate approach subsequently led to a constant public display of disagreement between the IMF and the economic team, further damaging public confidence. The monetary policy the IMF advocated would have involved higher interest rates, and one can argue whether this would indeed have been appropriate, but the fact is that high interest rates were not applied.

The size of financing was based on conservative assumptions and may have appeared small in relation to the large capital outflows that took place. The IMF did not anticipate the magnitude of capital flight by local residents, but it is difficult to argue that the initial IMF-supported program should have been designed to take account of all such capital outflows. A number of staff members interviewed have argued that the relatively small amount of official financing available in the first few months of the program lowered the probability of success. However, in our view, shortage of financing was not the critical factor, especially since key aspects of the initial program were not implemented. Much of the capital flight that occurred can be attributed to political uncertainties, which were in turn exacerbated by the failure of the initial program. Additional official financing would not have helped to address any of the underlying issues and would have only allowed such flight to take place at a more appreciated exchange rate.

The initial design of structural conditionality in nonfinancial areas, mainly addressing governance issues, was reasonable, as almost all agreed measures were general in nature and were to be implemented over the three-year lifespan of the program. Structural reforms in nonfinancial areas became a contentious issue only in January 1998, when the initial program had failed and the crisis had turned political. By January 1998, key shareholders and the press no longer saw deregulation as just an issue of microeconomic inefficiency, but had begun to perceive the governance-related reforms as something necessary to restore confidence by signaling a clean break with the past. The extensive structural conditionality, a widely criticized feature of the IMF response, was not the cause of the failure of the initial program, but a response to it. While many of the measures were popular with the public and undoubtedly had beneficial effects on the economy, in retrospect, the extensive structural conditionality in the January 1998 program became a distraction from taking much needed action on bank and corporate debt restructuring, which was missing from the January program.

In bank closure and restructuring, there was no internationally accepted best practice at the onset of the Indonesian crisis. While the initial strategy of closing 16 banks was consistent with the program’s logic (including the expectation of an exchange rate appreciation), it was based on a gross underestimation of the systemic nature of the banking sector problems. The IMF concluded that no other private banks needed to be intervened beyond the 10 under rehabilitation and the 16 being closed whose deposits represented only 3 percent of total banking sector assets, believing that the private banking system was sound beyond the troubled banks in the initial sample.36 In retrospect, the mistake was not the closure of the 16 banks which was initially well received, but the absence of a comprehensive strategy to deal with insolvent or illiquid banks. Such a strategy was only introduced at the end of January 1998.

The question of the partial deposit guarantee in the November program requires careful consideration. Arguably, the amount of Rp 20 million was too small and should have been expanded to cover some legitimate institutional deposits. However, the concept of a partial guarantee was entirely reasonable in a corrupt banking system, where the well-connected insiders had benefited both from high deposit rates and from questionable lending practices. In the early months of the program, moreover, confidence was maintained in the banking sector, where state banks with an implicit government guarantee accounted for a large share. What was happening in November was a shift of deposits from those private banks that were perceived to be weak to state, foreign and larger private banks, so that the banking crisis was not yet systemic (in the sense of affecting the whole banking system).

In the end, the blanket guarantee enormously raised the fiscal cost of banking sector restructuring, which is now estimated at over 50 percent of GDP, and allowed the same insiders who had benefited from the system an additional way to profit from abusive and corrupt practices. Would the introduction of a blanket guarantee in November have halted the banking crisis? It is impossible to test such a counterfactual. However, the evidence discussed here suggests that the most damaging aspect of the November crisis was not the nature of the guarantee itself, but the lack a well-communicated, comprehensive strategy to deal with problem banks.

Finally, corporate debt restructuring was a missing element of the IMF-supported program. It started late and did not progress very far. Restructuring of corporate debt was a difficult process, particularly in a corrupt system lacking an adequate legal infrastructure. Even so, something could have been done early in the program, when Indonesia’s corporate debt compared favorably with that of Korea, Thailand, and Brazil (Ghosh and others, 2002). If debt restructuring had been enforced with strong support of the President—clearly, a very big “if”—it might have gone a long way toward an equitable sharing of losses among various stakeholders, including the well connected, their foreign financiers, and the tax-paying public. In the end, the burden was almost entirely passed on to future generations through an increased stock of public debt.

The mode of operations

The failure of surveillance and weaknesses in program design and implementation in part reflected the IMF’s mode of operations. The IMF overestimated the extent of country ownership, particularly in structural reforms. While most of the measures were endorsed by the economic team and popular with the general public, the program lacked the ownership of those who counted the most in the decision-making apparatus of Indonesia. Greater understanding of the political economy dynamics might have contributed to a different program design. Nevertheless, it must be recognized that separating the economic and political elements that made Indonesia’s crisis so toxic would have been very difficult with any program.

The quality of program design was affected by the rushed procedure. While such a procedure may be necessary in certain cases, and the decision to rush was understandable under the conditions of great concern about a regional meltdown, the case of Indonesia—which initially had substantial reserves—does not seem to fall in that category. The rushed procedure led to detailed involvement by the Executive Board, subjecting the staff to greater political pressure. Management often worked directly with the missions in the field, bypassing the normal review mechanisms inherent in a bureaucratic organization. These problems were compounded by some weaknesses in human-resource management practices, which resulted in the failure to utilize available skills and resources in an efficient manner. The IMF showed flexibility in responding with speed, but there was a significant cost in terms of quality, especially in terms of understanding the nature of the crisis and the degree to which the program was owned and hence would be implemented.

Appendix A1.1 Indonesia: Selected Conditionality Under IMF-Supported Programs: Evolution and Implementation, 1997–981

A. November 1997 Letter of Intent

article image

B. January 1998 MEFP and Letter of Intent

article image

C. April 1998 Supplementary MEFP

article image
article image
Note: Unless italicized, all the structural measures were included in the 1997 Article IV consultation report.

*** = subject to revision during subsequent reviews; ** = fully satisfied conditionality without delay; **d = fully satisfied conditionality with delay; */* = partially satisfied conditionality; and * = unsatisfied conditionality. When no mark is attached information was considered insufficient to judge.

PC for April and June 1998 were established.

Appendix A1.2 Indonesia: Timeline of Major Events1

article image
article image
Sources: Bloomberg, Reuters, IMF, and local newspapers.

Local time, unless noted otherwise.

U.S. eastern standard time.

Western European time.



At a meeting of the Indonesia Consultative Group held in Tokyo on July 16–17, 1997 for example, the IMF representative stated that “financial market confidence in Indonesia [remained] strong,” while noting the “need to guard against changes in market sentiment, weaknesses in the banking system, relatively high external debt and increased financial market turbulence in the region.


Although no precise figure is given, the staff report for the 1997 Article IV consultation noted that the stock of short-term debt was “low,” suggesting a range of US$10 billion.


According to IMF data, the annual average real effective exchange rate was 97 in 1994/95, 99 in 1995/96, and 105 in 1996/97, with the base of 100 for 1990.


In this context, Pincus and Ramli (1998) argue that Indonesia’s fundamental mistake was to deregulate the banking sector in “deeply entrenched patrimonial state structures.”


The World Bank became engaged in the restructuring of large state banks through a Financial Sector Development Project. While the purpose of the World Bank project was to recapitalize and improve the operations of the problem state banks, the Bank became aware of serious governance problems in the summer of 1996 and eventually decided to suspend the project.


Indonesia’s average debt to equity ratio was high at 250 percent (Ghosh and others, 2002).


See, for example, articles that appeared in the Asian Wall Street Journal, April 13 and October 24, 1994, and June 29 1995; and in the Far Eastern Economic Review, July 11, 1991, June 23, 1994, and February 9, 1995. The topic of changing business practices, particularly in Asia, also began to receive an increasing focus of attention in the academic literature (e.g., Fukuyama, 1995; Weidenbaum and Hughes, 1996).


Until fiscal year 2000 (April–December), Indonesia’s fiscal year ran from April 1 to March 31 of the following year. Thereafter, it corresponded to the calendar year.


A Washington Post article of January 7, 1998 emphasized the lack of commitment to the reform program and only mentioned in passing that analysts perceived that the budget unveiled by the authorities had made suspension of the program more likely.


In late December 1997 and early January 1998, the staff expected no growth in 1998/99 and did not yet anticipate collapse of output in the first and second quarters of 1998. The output collapse was in large part driven by political developments. There were also negative balance sheet effects on investment, resulting from the sharp depreciation of the rupiah.


See, for example, investors’ comments reported in the Bloomberg News on September 4, 1997 (quoting analysts at Peregrine in Jakarta and Hong Kong SAR) and September 26, 1997 (quoting an analyst at Merrill Lynch Asia Pacific).


In fact, BI took only a small interest rate action. What happened was that interbank interest rates rose sharply when BI only partially sterilized intervention. When BI found some banks failing to clear at settlement, it injected liquidity, causing interbank rates to decline.


According to the BI Governor, liquidity problems in the banking sector developed as a result of monetary and fiscal tightening in August 1997. Weak banks began to experience distress and bank runs emerged in the second half of the month. Interbank rates increased from an average of 22 percent to more than 80 percent (see Figure A1.2). By the end of August, “more than 50 banks had failed to comply with the minimum reserve requirement of 5 percent” (Djiwandono, 2000). In the technical files of MAE, however, there is nothing to indicate a systemic liquidity problem and it is not clear if the whole system became illiquid or if the problem was limited to weak banks and those subject to runs. Market segmentation, however, does seem to indicate that at least the first-tier banks (e.g., the JIBOR banks) were not short of liquidity.


While liquidity support in principle required collateral, for a variety of reasons, there was little collateral available in 1997–98, which necessitated the pledge of personal guarantees from the bank owners that their banks met the conditions for liquidity support. With the subsequent discovery that many of these pledges were in fact invalid, in most cases because the banks had breached the legal lending limits, the owners became liable for making the repayment. Under the so-called shareholder settlements, IBRA was to recover such funds from the respective owners, but the nonimplementation of commitments and manipulation of the process resulted in large LOLR losses.


The report was prepared at the request of Parliament, in cooperation with the Finance and Development Supervisory Body (BPKP), with Price Waterhouse serving as a consultant. BPK audited all allocations of BLBI to 48 troubled institutions as well as the use of funds by 5 “Take Over Banks” (BTOs) and 15 liquidated banks (BDLs). BPKP audited the use of BLBI by 10 “Frozen Operation Banks” (BBOs) and 18 “Frozen Trading Activities Banks” (BBKU).


In view of the high level of reserves, it was assumed that BI could temporarily cover delays in the disbursement of multilateral resources from the other IFIs.


Some on the IMF staff hold the view that most banks would have remained solvent if the exchange rate had recovered to the programmed target range of Rp 3,000 to Rp 3,500.


For example, Bank Danamon, a large retail bank, had experienced sporadic runs even before the IMF was called in and, by end-October 1997, had already received Rp 3.5 trillion of liquidity support.


BI had an understanding that the 10 banks being rehabilitated would be closed if they did not demonstrate the capacity to become viable within six months to a year.


Some representatives of the Indonesian authorities told the evaluation team that they had not been adequately informed on this issue by the staff, especially regarding the blanket guarantee that had been provided in Thailand. Within the Indonesian government, however, the Ministry of Finance was adamantly opposed to a blanket guarantee on grounds of both equity and cost. In Washington, following criticism of the blanket guarantee in Thailand, there was strong opposition to establishing a blanket guarantee in Indonesia. Some former Executive Directors and U.S. government officials interviewed told the evaluation team, as a matter of their personal opinion, that a program for Indonesia would not have been approved by the Executive Board if the program had included a blanket guarantee.


The balance of foreign currency deposits is estimated to have declined from about US$30 billion in August 1997 to about US$15 billion in June 1998.


PDR, however, explicitly recommended that the IMF should learn from the mistakes made in Thailand and publish the LOI. The IMF thus sent an annotated version of the LOI suitable for publication to the authorities, who in turn agreed to make it public. However, it was never published.


At the suggestion of Singapore’s Senior Minister, this former central banker was invited by President Suharto to provide an independent assessment of the IMF package. Kenward (2000) suspects that this negative assessment of the package may have influenced the President’s subsequent actions.


In the FICORCA scheme, creditors and debtors were provided a guarantee against further depreciation of the exchange rate from its value at the time the debt was restructured.


See, for example, Goldman Sachs, “Emerging Markets Currency Analysis,” November 1997.


Indeed, the quite prescient memorandum from PDR in October 1997, referred to earlier, did call for such action.


When the package of reforms was announced to the press in January 1998, Indonesian journalists spontaneously congratulated the IMF officials for their achievement.


PDR’s comments on the brief included a proposal that a two-step approach of fact-finding followed by program design should be pursued. Likewise, the Resident Representative also advised strongly against rushing into a program, as it would unnecessarily panic the markets.


For example, given the assumption that the exchange rate would quickly bounce back, use of BI’s September data (available in early November) may not have given a substantially different diagnosis of the banking sector than did the June data, particularly because the staff was not allowed in any case to examine the loan files of individual banks.


The Central Asia Department (CTA) and the South Asia and Pacific Department (SEA) were merged to form what is now APD, effective January 1, 1997. Staff coming from CTA, which previously had not covered the country, assumed the crisis management of Indonesia.


The banking strategy announced in January 1998 was based on a January 13, 1998 memo prepared by a member of the MAE technical mission while the second LOI was being drafted. This memo was circulated to the negotiating mission late in the process and almost by chance. Perhaps a broader dialogue on banking sector ideas in October could have provoked an earlier formulation of the key elements of that strategy.


Since the minutes of informal Board meetings are not kept, the evaluation team could only rely upon interviews with those present to ascertain what was said. There were also meetings of the Executive Directors for the G-7 countries, for which no minutes were kept.


When the Concordat was discussed in the Executive Boards in 1989, however, the Bank’s Executive Directors expressed serious reservations, so that the Bank did not consider it to be institutionally binding. More recently, in September 1998, the Managing Director of the IMF and the President of the World Bank issued a joint statement, reaffirming the principles underlying Fund-Bank collaboration as set out in the 1989 Concordat. See Boughton (2001), pp. 1003–05, 1055–61.


There is a striking parallel to what happened at the World Bank. According to the Country Assistance Note on Indonesia prepared by the Operations Evaluation Department (World Bank, 1999), in February 1997, the office of the Chief Economist “stressed that risk factors had been underestimated, that the Bank’s strategy should not be limited to the optimistic base-case scenario, and that a ‘downside analysis’ was needed in view of the high country risks.” According to this note, as late as August 20, 1997, Bank country staff and management downplayed these risks and communicated to the Executive Board that there was no cause for concern.


The staff was aware of the importance of corporate debt restructuring. However, the few attempts made at corporate data collection were not sustained because of the inherent difficulty of obtaining such data as well as the perception that the corporate sector was outside the IMF mandate and in the purview of the World Bank.


The staff knew that the state banks were in serious difficulty, but determined that they could more appropriately be dealt with separately.

Annex 2 Korea


This annex provides a detailed assessment of the role of the IMF in Korea’s capital account crisis of 1997–98, focusing on the role of the IMF in precrisis surveillance and in the process of crisis management.

The annex is organized as follows. First, it evaluates the effectiveness of IMF surveillance in identifying underlying vulnerabilities and the potential risk of crisis. It then discusses issues of program design, including monetary and exchange rate policy, fiscal policy, financial sector reform, and nonfinancial structural reforms. Next, it examines the appropriateness of program financing and the role of the IMF in the debt rollover agreement of late December 1997. The final section presents conclusions.

Precrisis Surveillance

With the benefit of hindsight, one can identify several weaknesses in the IMF’s surveillance of Korea during the period leading up to the crisis. This section discusses two areas in which these shortcomings proved to be most damaging: the analysis of the vulnerabilities introduced by the uneven process of capital account liberalization; and the initial assessment of the risk that the crisis spreading through Asia in the fall of 1997 would soon hit Korea.

Underlying vulnerabilities

Throughout the 1980s and the first half of the 1990s, the Korean authorities alternately liberalized and restricted both inward and outward capital account transactions in pursuit of their policy goals for the external sector.1 Thus, in the early 1980s, capital inflows were liberalized and capital outflows restricted to assist the financing of current account deficits. Later in the decade, when Korea began to run substantial current account surpluses, controls were reimposed on inflows and controls on outflows were eased. The environment of current account surpluses also contributed to the authorities’ decision, in 1988, to fully liberalize current account transactions and thereby accept the obligations of Article VIII of the IMF’s Articles of Agreement.

When current account deficits reappeared in the early 1990s as a consequence of the strong won and the global recession, the Korean government again imposed controls on purchases of foreign exchange by residents and removed controls on certain categories of capital inflows. The stock market was opened to foreign investors in 1992, though with ceilings on the fraction of a given company’s shares that could be held by any foreigner individually and by foreigners in aggregate. FDI was partially liberalized. Short-term borrowing by banks and certain nonbank financial institutions was liberalized in the mid-1990s. Merchant banks, which would later play a central role in the 1997 crisis, were at the forefront of institutions taking advantage of the easier rules on overseas borrowing (Box A2.1).

As a result, capital inflows surged, which led to upward pressure on the currency. Significantly, rather than attempting to restore balance by reimposing controls on inflows, as might have been done in the past, the authorities instead chose to liberalize outward portfolio investments by Korean residents. A Foreign Exchange System Reform Plan was issued in December 1994, which outlined a gradual, staged liberalization process for the capital account and the foreign exchange market.

In spite of the overall commitment to freeing capital flows, this process had not moved very far by 1997. Korea still maintained substantial controls on many capital account transactions, particularly on the external issuance of long-term bonds and long-term commercial loans by financial and nonfinancial entities. Limits also remained on foreign participation in domestic equity and bond markets. The decision to pursue liberalization of capital inflows had in part resulted from lobbying by the business community, which wanted to take advantage of relatively low short-term interest rates in global markets. Yet many reform-minded officials, while favoring the liberalization of financial markets as a general principle, resisted measures to allow firms to raise funds directly from foreign bond investors. It was feared that this would enhance the power of the large conglomerates (the chaebol) at the expense of small and medium-sized enterprises. As a result, in the mid-1990s, a new policy was initiated that deliberately steered capital inflows through domestic financial institutions.

Merchant Banks in Korea

The merchant banks, most of them owned by chaebol, had been created from short-term finance companies, which in turn had been established in 1972 to facilitate “curb market” transactions, that is, those not permitted to the established commercial banks. The policy of liberalizing short-term flows before long-term flows and restricting direct capital-raising by nonfinancial firms gave the merchant banks a profitable market niche. They acted as intermediaries for chaebol-affiliated firms, discounting commercial paper and reselling it to commercial banks. They also offered cash-management accounts and other instruments to investors, and dealt in corporate promissory notes. These opportunities proved to be so lucrative that 24 new merchant banks were established between 1994 and 1996. The merchant banks were required to keep their currency exposures in balance, but there were many loopholes in these rules and supervision was poor. For their part, commercial banks felt pressure to compete with the merchant banks, and began to borrow abroad at short maturities as well.

Even Korea’s accession to the Organization for Economic Cooperation and Development (OECD) in December 1996 did not lead to a substantial additional opening of capital markets. Joining the OECD was seen as an important political goal and as a way to reduce borrowing costs,2 but in the accession talks the authorities resisted efforts to bring Korea’s capital account regulations in line with those of other OECD members.3 In taking this stance, the authorities cited their concern about the consequences of a sharp increase in capital inflows, given prevailing interest rate differentials. The policy of permitting short-term borrowing and restricting long-term flows allowed the authorities additional flexibility vis-à-vis the OECD’s rules, which grant members the right to “roll back” previously adopted liberalization measures with respect to most short-term capital movements but not those regarding long-term movements.

The decision to liberalize short-term transactions before long-term ones had unintended consequences. Given the opportunity, the chaebol and the banks would probably have strived to secure long-term financing even at the expense of a small term premium. If a greater share of Korea’s external debt in 1997 had been in the form of long-term instruments, issued by a mix of financial and nonfinancial institutions, rather than in the form of short-term bank debt, the character of the December crisis would have been different and probably less damaging. For one thing, a diversity of financing channels might have made the system more resilient to a breakdown in one channel, in this case interbank loans to overseas branches and subsidiaries. If the international market for the long-term debt of Korean nonfinancial corporations had been deeper and possessed a lengthy, successful track record, then foreign investors might have been willing to continue financing investment by healthy borrowers, while avoiding troubled corporations and banks.4

Moreover, if more of Korea’s external debt had been at longer maturities, the sudden drop in the market’s confidence in the Korean financial system might have led to an explosion of spreads and a severe credit crunch, but not a liquidity crisis. This is because holders of maturing short-term debt can demand payment from the original issuers, forcing the latter to rush to obtain cash or liquid assets, while holders of long-term debt that has been downgraded but has not yet matured can only sell the obligations to other investors (or simply write down the loss).

The distinction is important because liquidity crises tend to spread more rapidly and have a broader impact than do incidents where perceived levels of credit risk merely rise sharply. In a foreign exchange liquidity crisis, there is the further risk that the authorities will impose a standstill on payments. As a result, the risk premium imposed by foreign investors on all borrowers increases, regardless of their creditworthiness. Creditors, concerned over whether any borrower will be able to honor their foreign exchange–denominated obligations, may demand repayment as soon as these obligations mature. Once some creditors start to take this approach, all creditors find themselves forced to do so, introducing dynamics that are strongly reminiscent of a bank run (Radelet and Sachs, 1998a). By contrast, in a credit crunch that is not a liquidity crisis, the market’s ability to distinguish between good and bad borrowers eventually returns, even if risk premiums may increase for a time on all borrowers. Creditors do not demand repayment from creditworthy borrowers simply because they fear that liquidity will run out. The extent of damage to the real economy is therefore likely to be less.

The IMF followed Korea’s capital account liberalization process closely and, through Article IV consultations, regularly urged the authorities to establish and follow a steady timetable for liberalization. However, staff papers and Board discussions were concerned primarily with the speed of liberalization (typically recommending a faster process) and with whether it should be contingent on the convergence of Korean interest rates to international levels (typically concluding that it should not). Issues of sequencing and supervision were inadequately addressed in the surveillance process, though these topics were attracting increasing attention elsewhere in the IMF.5 According to staff members interviewed by the evaluation team, the focus on capital account liberalization in Korea reflected the IMF’s belief that liberalization of its external accounts would encourage the authorities to pursue genuine reforms of the domestic financial sector, including improvements in supervision.

One reason why surveillance failed to highlight the potential vulnerabilities in Korea’s external accounts was that the IMF—along with many others at the time—thought of the capital account solely in terms of transactions between residents and nonresidents. For this reason, short-term borrowing by overseas bank branches and subsidiaries was not recognized as an important issue. For example, a study of capital account liberalization in Korea and three other countries conducted by MAE and published in November 1997 exhaustively catalogued the liberalization measures undertaken by each country and the associated developments in transaction volumes (Johnston and others, 1997). Yet this paper did not draw attention to the growth in borrowing by Korean overseas bank affiliates, except to mention that the establishment of overseas branches and subsidiaries had been permitted as part of the liberalization of outflows of direct investment. The authors did not treat borrowing by the affiliates as potentially equivalent to borrowing by their parent institutions.6

Assessment of the risk of crisis

The prevailing IMF view in the early months of 1997 was that, while Korea faced problems in its financial sector that were potentially very serious and that needed to be addressed promptly, there was no risk that this would lead to a loss of confidence and crisis-inducing capital account outflows. There was some concern at the widening current account deficit, but these concerns dissipated as the deficit narrowed in the first half of 1997. The failure of Hanbo Steel was treated as a political matter, because of its impact on the standing of the ruling party, rather than in terms of the impact of further failures of chaebol on the health of the banking sector. The IMF’s view, which was shared by many (though not all) other public and private sector observers at the time, was influenced by Korea’s strong macroeconomic record and its proven ability to raise foreign funds with little difficulty.

As the East Asian crisis spread in the summer and fall of 1997, there were grounds to reassess this view. Because of the activity of their overseas branches, Korean banks faced a maturity mismatch between their foreign currency assets and liabilities, while the chaebol to which the banks had lent in dollars faced a currency mismatch. Much of the Korean banks’ debt was at short maturities and was vulnerable to a decision by foreign lenders not to roll it over. Their situation was reminiscent of those of the financial sector in Thailand and the corporate sector in Indonesia. Market commentary and credit spreads indicated that international investors and bank lenders were reappraising the riskiness of their exposure to the East Asian region as a whole.

The IMF was aware of these issues. In internal memos circulated in August and September 1997, the staff criticized the support package put together by the authorities in response to growing financial sector problems, on the grounds that the package fell far short of what needed to be done to restructure the financial sector. The guarantee extended to the external liabilities of Korean banks in late August was especially troubling. In the staff’s view, the guarantee raised the risk of a spillover of domestic financial difficulties into the external sector, because to honor the guarantee the authorities would either have to borrow on international capital markets or dip into foreign exchange reserves.

The Article IV consultation mission that visited Korea in October 1997 included a staff member from MAE, who produced a detailed analysis of financial stability issues. Yet, while acknowledging the possibility of a spillover from the financial sector to the capital account, the mission concluded, in its back-to-office report, that Korea was “relatively well equipped” to handle further external pressures.7 Because of this assessment, which was heavily influenced by incomplete reporting on the part of the authorities about their reserve position (see below), there was no attempt to analyze rigorously Korea’s vulnerability to a cutoff of external short-term financing until after the country’s usable foreign exchange reserves were all but depleted. Had such an analysis been attempted earlier in 1997, important data gaps might have been recognized sooner, particularly in such areas as the nature of the BOK’s advances to commercial banks, the ability of the authorities to access these funds in a crisis, and the multiple strains on Korea’s dwindling stock of foreign exchange reserves.

The failure of IMF bilateral surveillance to identify Korea’s vulnerability to a crisis was not unique. Other observers in the private sector were also caught off guard. In retrospect, one can attribute the failure on the part of the IMF to five misconceptions, which were compounded by critical information gaps.

First was the misestimation of the degree of flexibility in the country’s exchange rate policy. The briefing paper for the October Article IV consultation mission lists, as one of the reasons for the staff’s view that Korea faced only a “moderate” risk of a foreign exchange crisis, “the relatively flexible exchange rate policy and absence of indications of exchange rate overvaluation.” It noted that the Korean won had depreciated almost 17 percent against the dollar since the beginning of 1996, reversing an earlier period of appreciation. Yet, depreciation up to that point in response to the Asian crisis had been very limited (Figure A2.1). At the time the paper was written, the won had depreciated barely 2 percent since July 1, 1997, after having weakened 8 percent from October 1996 to July 1997. The behavior of other Asian currencies at that time could have offered evidence that the won was being artificially supported. Singapore, which pursued a more flexible managed float from the beginning of the crisis, allowed its currency to depreciate 7 percent from the beginning of July 1997 to the end of September. Malaysia’s currency fell 29 percent over the same period.

Figure A2.1.
Figure A2.1.

Won–U.S. Dollar Exchange Rate and Korean Equity Prices

Source: IMF database.

With regard to exchange rate policy, the mission team misconstrued the authorities’ willingness to let the currency weaken further if foreign demand for Korean assets fell significantly. Internal documents suggest concern at the degree of foreign exchange market intervention, and particularly at the possibility that Korea might have adopted a large forward exposure, as had been the case for Thailand. At the end of the Article IV consultation mission, the staff advised the authorities to scale back such intervention. Yet, perhaps because of their judgment that the won was not overvalued, the staff did not put much emphasis on this issue. Instead, the IMF’s policy advice to the Korean authorities focused more on the need to accelerate structural reforms than on macroeconomic policy. The staff at that point did not view an excessive commitment to support the won as a factor hindering Korea’s ability to respond effectively to the crisis. The authorities’ failure to share critical information with the staff about the extent and nature of their intervention, and about the actual status of their reserves, was central to the staff’s mis-diagnosis of the situation. In the event, the authorities’ attempts to support the won during November through intervention would prove to be a critical drain on Korea’s foreign exchange reserves.8

There was also excessive optimism regarding Korea’s ability to prevent speculative pressure on the won. In September, the staff found reassurance in the fact that “the remaining capital controls [limited] the ability of international investors to take short positions in won.” Yet the Thai experience should have shown that capital controls of this type cannot protect a currency when domestic and foreign investors move decisively out of domestic assets through whatever channels are available. For example, as RES pointed out at the time, foreign investors could take short won positions in offshore derivatives markets. Downward pressure on the won would then be transmitted to the domestic market through hedging by the domestic Korean institutions that acted as market-makers in these instruments. In other words, pressure on the won, if it developed, would take whatever form it could.

Second, the staff underestimated the risk of a breakdown in funding the capital account. The staff recognized that such a risk was present, particularly in the crisis conditions then prevailing in East Asia, but concluded that the authorities could handle any pressures by making renewed efforts in the area of financial reform, by addressing financial sector weaknesses, and by loosening controls on long-term external borrowing. In part, this risk was underestimated because there was insufficient data on Korea’s short-term external obligations (though some relevant data sources were overlooked). While the staff was concerned at the level of short-term external debt and pressed the authorities to lengthen the maturity structure of this debt, efforts to clarify these concerns, for example by requesting the appropriate data more forcefully, do not seem to have been pursued until the crisis had already broken out.

More fundamentally, the staff (and most other observers at the time) did not foresee the degree to which market sentiment would swing against Korea, and the consequences this would have for the provision of credit of all kinds. This shift in sentiment rendered the recommendation for looser controls on long-term borrowing moot; surely, if Korea had difficulty rolling over its short-term external debt, it would have even more difficulty refinancing its short-term debt at longer maturities.

Third, the potential short-term impact on growth of problems in the financial sector was underestimated. The September 1997 briefing paper contained three scenarios for macroeconomic developments in Korea: a “baseline” scenario positing growth of about 6 percent in both 1997 and 1998; a scenario assuming the adoption of the IMF’s “preferred policies,” under which growth would fall to 5.3 percent in 1997, then rise to 6.7 percent in 1998 (after which the outlook would remain higher than in the “baseline”); and “disorderly adjustment,” a scenario supposedly incorporating a possible spillover of the domestic financial problems to external financing, resulting in growth of 4.0 percent in 1997 and 4.5 percent in 1998.9 Slower growth in this last scenario resulted, not from a breakdown in financial intermediation or a fall in investment reflecting a drop in confidence, but from tighter macroeconomic policies in response to downward pressure on the won. Yet the experiences of other economies in the 1990s, such as Japan, Sweden, and Finland, showed that broad-based financial sector restructuring can have a serious impact on growth rates over a period of several years. The narrow range of growth estimates across the three scenarios, a reflection of the remarkable stability of Korea’s growth rates over the previous decades, prevented the staff from exploring the possibility or, more importantly, the consequences of a more serious slowdown.

Fourth, not enough attention was paid to relevant market indicators, for example, the yield spread of Korean Development Bank (KDB) bonds (state-guaranteed obligations denominated in dollars) over U.S. treasuries, and the expected won depreciation implied by prices in the offshore nondeliverable forward market. As noted in Park and Rhee (1998), both of these began signaling profound market unease over events in Korea as early as August 1997 (Figure A2.2, top panel; and Figure A2.3, top panel). From August to October 1997, the bond spread widened and the nondeliverable forward rate indicated increased expectations of depreciation (though these movements would be dwarfed by developments during the crisis period).10 Nowhere in the briefs leading up to the November program-negotiation mission can one find a reference to the negative signals emanating from these sources.

Figure A2.2.
Figure A2.2.

Korean and Emerging Market Borrowing Spreads

(In basis points over comparable U.S. treasury rates)

Source: Datastream.
Figure A2.3.
Figure A2.3.

Won–U.S. Dollar Spot and Forward Rates1

Source: Reuters.1 The forward rates are implicitly derived from offshore nondeliverable forward prices.

Finally, and more generally, bilateral surveillance in the years preceding the crisis was not sufficiently sensitive to the short-term stability implications of financial sector liberalization.11 The prior experience of liberalization in other countries, such as the Nordic countries or the savings and loan crisis in the United States, was that liberalization tended to be followed by excessive lending and radical restructuring of the financial industry, with firms, consumers, and regulators learning the ins and outs of the new system through trial and error. The long-term benefits of such liberalization came only after a period of experimentation and instability. The advice offered to Korea in the late summer and early fall of 1997, to the effect that the solution to the immediate problems of the financial sector lay primarily in strengthening and accelerating the reform agenda, may have been valid from the perspective of the long-term health and efficiency of the system, but did not offer much guidance as to how the Korean authorities should secure the system against the external shocks that had already started to hit nearby countries.12 The Article IV consultation mission did urge the authorities to assess the extent of the banks’ NPLs and the scope for provisioning. Relatively little advice was offered, however, toward the formulation of a strategy for restructuring and recapitalizing the banking sector in the face of a possible crisis, beyond general principles such as avoiding regulatory forbearance, limiting public support to the minimum necessary, and broadening the role of the KAMCO. This reflected the IMF’s lack of experience at that time in the resolution of domestic financial sector crises.

While these factors were not adequately assessed in the IMF surveillance reports, there was recognition internally that they may pose serious problems. A team from the RES Capital Markets Division had visited Korea earlier in the year, as part of their preparation for the annual International Capital Markets report, and had identified several weaknesses in the Korean financial system. Expanding on these findings, a member of the team later prepared an internal note detailing some of the vulnerabilities, including the NPL problems. Commenting on the Article IV pre-mission brief, RES cited the authorities’ “widespread and unconditional” support for troubled financial institutions, the poor state of supervision and regulation, and the rapid rise in short-term debt as potential sources of risk. RES also expressed skepticism over the willingness of the authorities to allow the exchange rate to adjust in the way envisaged in the briefing paper, given their history of intervention.

In part, the shortcomings of surveillance in the precrisis period reflected a shortage of analytical resources. Because of Korea’s record of stability, relatively few staff members were following the country regularly during the time preceding the crisis.13 Korea was usually covered either by the division that also was responsible for following Japan, or by the one that covered China; in both cases, the bulk of analytical resources was devoted to the larger country. There was little in the way of structural analysis of Korea’s financial and corporate sectors available from the World Bank, because Korea had “graduated” from Bank lending programs in the early 1990s.14 Moreover, by the fall of 1997, APD was stretched thin by the crisis spreading throughout the region, so senior staff had to be transferred from other country assignments to lead the Article IV consultation mission.

However, several of the lapses identified above represented not so much a lack of familiarity with or knowledge of Korea, as a failure to draw the appropriate parallels with experiences in other economies. This was the case both for contemporaneous developments—contrasts with Taiwan Province of China, Singapore, Malaysia, and Thailand have already been noted—and for prior experience, as with bank reform and restructuring in the Nordic countries.

Surveillance also suffered from the poor quality of the available data, particularly on the vital topics of NPLs, external debt, and usable reserves. The lack of good data appears in part to have reflected the data provision policies chosen by the authorities. Until November 1997, there was little internal discussion of the need to press the authorities to improve the quality of statistics on their debt and reserves.15 At the same time, certain data sources appear to have been overlooked. For example, the consolidated and locational statistics compiled by BIS pointed to sharp increases in interbank debt, and particularly short-term debt, in the years immediately preceding the crisis.16 As noted above, another key reason for the poor quality of the data was the tendency for both the staff and the authorities to think about capital flows only in terms of a “residence” concept rather than “nationality.” As a result, they relied on prevailing statistical definitions that did not include the obligations of Korean banks’ overseas branches among the liabilities of the Korean financial sector, although nationality-based data were available, albeit in limited form, from BIS and national sources.17

Program Design

This section reviews the major elements of program design in the IMF-supported program for Korea, as agreed at the beginning of December 1997 and modified over the subsequent months (Box A2.2), including monetary and exchange rate policy, fiscal policy, financial restructuring, and nonfinancial structural reforms.

Monetary and Exchange Rate Policy
Evolution of the IMF’s policy advice

The monetary policy section of the briefing paper prepared for the November 1997 negotiating mission was the outcome of considerable internal debate. In commenting on an earlier draft of this brief, RES suggested that monetary policy should guide the exchange rate to a range close to its then prevailing level, while MAE suggested that it would not be possible to determine an appropriate exchange rate target at that time. MAE argued that there was a risk that targeting the exchange rate would prove unsustainable, because the high domestic interest rates needed to defend it would exacerbate the bad loan problem over time.

The IMF-Supported Program in Korea

The policy actions undertaken by Korea in connection with the SBA were detailed in a “Memorandum on the Economic Program” attached to the “Letter of Intent,” which was signed by the Minister of Finance and Economy and the BOK Governor, and in a letter on “Prior Actions,” signed by the Minister of Finance and Economy alone. The Memorandum detailed the actions that the authorities intended to undertake and identified performance criteria and structural benchmarks that they were committed to achieve. In fact, there were only two explicit performance criteria in the initial program, namely, targets for NIR and for NDA, though there was extensive discussion of policies that would be pursued in other areas. The prior actions letter detailed measures that had been taken or would be taken in short order upon approval of the program by the Executive Board. These two documents specified actions that had been or would soon be taken in the following areas:

Monetary and exchange rate policy

Fiscal policy

Financial sector restructuring

Corporate sector, trade, labor market, and information provision

The above list formed the basis for the policies that Korea would undertake over the next two years. A modified program was agreed on December 24, 1997, along with a faster disbursement of IMF resources. The revised program specified additional measures that would be undertaken (in most cases, this amounted to an accelerated timetable of agreed measures). Subsequent program reviews would convert some of the items listed into explicit structural conditions, and add new reform measures in the same spirit as those listed.

The brief that emerged represented a compromise. It envisaged “a tightening of monetary policy directed at containing the impact of recent won weakness on inflation and preventing a significant further weakening of the currency.” A target of 8.5 percent growth in M3 was set for 1998, significantly lower than the 15.8 percent M3 growth projected for 1997. The proposed program also involved “an (implicit) [parentheses in original] target range for the won’s nominal effective exchange rate with the understanding that monetary policy [would] be tightened if the rate [fell] to the bottom of this range,” though the exact range was not specified.

By the time the program was finalized in early December 1997, the role envisaged for monetary policy had shifted. A nominal effective exchange rate target was no longer contemplated. Instead, the objectives of monetary policy were defined to be to contain inflation to 5 percent and to limit downward pressure on the won. There would be an immediate increase in interest rates to demonstrate the government’s resolve in the face of the crisis and to calm the markets. Interest rates would later be brought down somewhat, but would remain high enough to limit downward pressure on the won and to ensure that inflation would be no higher than 5 percent in 1998. A target for broad money growth was set for the fourth quarter of 1997 but, unusually for IMF-supported programs at that time, monetary policy for the following year was to be guided by an inflation target. An inflation target, however, was not made part of formal conditionality in the program.

Two principal developments appear to have contributed to this change in focus. One was the sharp depreciation in the won, from W 987 per U.S. dollar on November 17 to W 1,249 on December 4. This made it virtually impossible to determine an exchange rate range that could be relied on as an anchor for policy. A second factor was that the BOK continued to provide won liquidity to troubled banks at favorable interest rates, even while the program negotiations were under way. As a result, the call money rate declined to 12.7 percent on December 2 from 15.0 percent on November 17 (Figure A2.4). The staff felt that strong action would now be necessary, in part to stem the drop in the exchange rate, but primarily to reestablish monetary control and to demonstrate the authorities’ resolve to regain exchange rate stability. Once the foreign exchange market had been stabilized, policy would be loosened, but would remain geared toward containing any inflationary effects of the weaker won and counteracting further depreciation.

Figure A2.4.
Figure A2.4.

Korea: Overnight Call Money Rate and Inflation

(In percent)

Source: Datastream.

Monetary policy, along with the closure of the merchant banks (discussed below), was one of the principal issues on which the authorities and the IMF disagreed most strongly during the first phase of the negotiations. The authorities feared that excessively high interest rates would cause an increase in bankruptcies in the highly leveraged corporate sector. Only with the intervention of the Managing Director in the final stages of the negotiations did the Korean authorities agree to raise interest rates to the levels thought necessary by the IMF.

The temporary nature of the rate increase was underscored in the letter on prior actions, signed by the Minister of Finance and Economy, that accompanied the request for an SBA. This letter specified that the call rate would be raised to 25 percent, and “maintained at that level until the time it [would] be judged—in consultation with the IMF staff—that it [could] be progressively brought down to a range of 18–20 percent.”

The call rate was duly raised to 25 percent in early December 1997, but confidence was not restored. Instead, the won remained extremely volatile and fell to record lows (Figure A2.1). The IMF urged a still tighter policy, but this could not be implemented immediately because of a 35-year-old usury law that set a ceiling of 25 percent on the call money rate. A law increasing the usury ceiling was passed on December 14, after which this rate was promptly raised to 30 percent. Further increases followed and the call rate peaked at 34 percent in early January 1998. However, by mid-December, it was clear to the authorities and the IMF (particularly the mission team in Korea) that this situation was not sustainable, given the impact it had begun to have on corporate balance sheets and given continued capital outflows. This spurred the search for another solution, namely the strengthened program and coordinated debt rollover announced on December 24.

The staff continued to endorse the maintenance of relatively high real interest rates in the early months of 1998, believing that the exchange rate had not yet fully stabilized and that there was still a risk of accelerating inflation. The call rate was maintained in the 20–25 percent range for the first three months of the year, and then was lowered gradually in the spring and summer. There was a strong concern that premature loosening of monetary policy would lead to a loss of monetary control and renewed depreciation of the exchange rate, as had happened in Indonesia.18 The staff acknowledged that the tight monetary policy (along with higher capital adequacy requirements) contributed to a credit squeeze, but contended that the best way to ameliorate the squeeze would be to implement the accelerated timetable for financial sector restructuring and to provide official liquidity to sound institutions against appropriate collateral. In the Executive Board reviews of the Korean program, comments tended to favor maintaining a tight monetary policy in support of the exchange rate. However, in the February discussion, one chair warned about “overkill” and suggested a more active willingness to ease policy once the exchange rate had stabilized.

Real interest rates during the first half of 1998 were very high by the standards of most industrial countries facing a recession, though not unusually high for emerging economies in crisis (see Table 4.2 in the main report). Using the actual 1998 CPI inflation rate of 7.5 percent, the overnight call money rate reached a high of 26.5 percent in real terms in early January, before falling to around 15 percent in February, 10 percent in early May, and single digits for the rest of the year.19 This represented a return to precrisis levels, which averaged about 8 percent in 1996–97.

Recent Studies on the Impact of High Interest Rate Policy in Korea

In the case of Korea, two recent empirical papers have yielded the result that higher interest rates (relative to their U.S. equivalents) had an appreciating effect on the won–U.S. dollar exchange rate during the 1997–98 crisis. Cho and West (2000) used daily data for the period December 17, 1997–June 30, 1999 to estimate regression and vector-autore-gression (VAR) models and found that, with appropriate control for risk, liquidity, and some external factors, a higher call rate was associated with exchange rate appreciation. Likewise, Chung and Kim (2002) applied a nonlinear econometric methodology (in order to take account of both levels and changes in interest rates) to daily data for the period January 4, 1995–September 30, 1998 and found that, in a bivariate VAR framework, a higher certificate of deposit (CD) rate led to an initial depreciation of the exchange rate for a few days, followed by an appreciation sustained over a few months, even during the crisis period (December 1, 1997–March 31, 1998) when the level of interest rates was high. It should be noted, however, that (1) a substantial portion of the sharp currency depreciation of the crisis period had already occurred by the beginning of the sample period in the Cho-West (2000) study, with a trough on December 24, and (2) the parameter estimates of the Chung-Kim (2002) study come from a sample that include a relatively long noncrisis period. Given these qualifications, the studies do not seem to present a strong case against the undeniable fact that the Korean won depreciated from W 1,163 to W 1,964 per U.S. dollar in December 1997 while the call rate was raised from 13 percent to 30 percent. More likely, these studies provide a confirmation of the conjecture that tight monetary policy maintained in the aftermath of the sharp depreciation helped to ensure that the subsequent real appreciation took the form of nominal appreciation rather than higher inflation.

The three-year corporate bond yield was 200–500 basis points less than the call rate from January through May, after which the call money rate fell substantially below the corporate bond yield. This implies, after appropriate allowance is made for the corporate credit risk premium, that the latent domestic currency term structure moved from being inverted to being upward-sloping either in May or soon afterward, offering another indicator that the monetary stance loosened around this time.


As in other crisis countries, monetary policy in Korea reflected a trade-off between, on the one hand, the need to reestablish external credibility, control inflation, and stabilize the exchange rate, and, on the other, the need to support domestic demand at a time of financial sector restructuring. As discussed in the main report, most economic policymakers at the time accepted the existence of a link between higher interest rates and a stronger exchange rate. While this view has been challenged since the Asian crisis, the large theoretical and empirical literature that has emerged has yet to settle the matter (Box A2.3). The literature, however, does suggest 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 the situation in the aftermath of an episode where the exchange rate has overshot its equilibrium level. In the latter case, the objective is to ensure that the required real appreciation occurs not through domestic price increases but through nominal appreciation (Goldfajn and Gupta, 1999).

For Korea, this suggests that there are in fact two distinct issues to consider:

  • Were high interest rates justified as a means to stabilize the won at the outset of the crisis in December 1997?

  • Were high interest rates justified in the early months of 1998, after the most critical stage of the crisis had passed but the exchange rate remained substantially weaker than its earlier levels?

It is difficult to answer these questions conclusively, given the lack of consensus in the academic and policy communities. However, we can look at which of the identified effects were operating in Korea at the time, and at whether the IMF took sufficient account of the relevant issues in formulating its policy advice to the Korean authorities.

Stabilizing the exchange rate

After the Korean authorities floated the currency, the immediate objectives of monetary policy were to arrest the sharp decline of the exchange rate and stabilize the foreign exchange market. However, the closed nature of Korean capital markets limited the channels through which monetary policy could achieve those goals. Higher interest rates could not have stabilized the won by increasing the cost of speculation against the currency. While the offshore market, mentioned earlier, was a potential source of speculative pressure, it was by no means the primary reason for the won’s weakness. Korea faced increased demand for liquidation of foreign currency claims rather than a conventional speculative currency attack. There was little risk of domestic capital flight, because of limits on the ability of residents to take funds out of the country.20 Though foreign currency deposits held by residents rose 267 percent in U.S. dollar terms from end-June to end-November 1997, even at the end of November they totaled only US$5.3 billion, so this trend had little impact.21 Conversely, foreign entities did not have many vehicles through which to invest in won-denominated assets. In practical terms, neither the long-term bond market nor the short-term money market was open to foreigners. It was possible to lend to Korean entities—but these were precisely the loans that foreign financial institutions were rushing to liquidate. Foreigners could also invest in the stock market, but higher interest rates would be likely to discourage foreign share purchases in conditions of panic, by lowering realized returns and thus depressing market sentiment.

In the view of IMF staff at the time, the main channel through which the interest rate defense would operate was that higher interest rates would raise the opportunity cost for Korean banks of not having their foreign currency loans rolled over. A Korean bank with an outstanding short-term dollar-denominated loan about to come due could either promise to pay its foreign lender a higher dollar interest rate, inducing the latter to roll over the loan, or it could borrow won on the domestic market––in effect, a loan from the central bank, given the guarantee mechanisms in place––and use the won to buy dollars in order to pay off the loan. The second of these two options, if pursued by enough institutions, would cause downward pressure on the won. A higher won interest rate might induce more Korean banks, at the margin, to choose the former course rather than the latter. In this sense, the high won interest rates were a complement to the policy of having the central bank charge a penalty rate for foreign exchange advances; both policies were intended to induce Korean banks to seek rollovers rather than drawing on central bank liquidity.

Given the nature of the Korean crisis, however, very high interest rates were necessary for this channel to operate effectively. For many creditor banks, the rollover decision depended more on their assessment of credit risk—including the suddenly heightened risk, not merely of their Korean positions, but of East Asian exposure in general—than on the interest rate their Korean counterparties offered them. It is not clear if any level of interest rates offered by Korean borrowers would have been high enough to induce such banks to roll over their loans. This was borne out by events, since capital outflows only stopped when the high interest rates were complemented by the coordinated rollover agreement.

Thus, a tighter monetary policy may have been necessary to slow the leakage of foreign exchange and to prevent a full-scale collapse of the exchange rate, but it was not sufficient as a means to reverse capital outflows and resolve the crisis. If the staff had come into the crisis with a better understanding of the nature of Korean capital markets, then it is possible that less emphasis would have been placed on monetary policy in the initial formulation of the program, and more on finding an alternative solution to the worsening liquidity crisis.

The recovery and the transition to lower rates

According to the objectives set out in the IMF-supported program, monetary policy during the first half of 1998 had two goals: to stabilize the foreign exchange market and to counteract the inflationary effects of the depreciation. As regards the first objective, one can argue, with the benefit of hindsight, that monetary policy was guided by an excess of caution rather than deliberate overkill. The won strengthened from W 1,964 to the U.S. dollar on December 24, 1997 to around W 1,400 at the end of March 1998, and remained at or near that level for the next three months (see Figure A2.1). The volatility of the exchange rate declined steadily in the first half of 1998 (Figure A2.5). The volatility of the won–U.S. dollar rate (measured as the standard deviation of daily logarithmic changes) fell to 0.7 percent in June 1998 from 8.1 percent in December 1997. For comparison, the monthly volatility levels of the Japanese yen–U.S. dollar and deutsche mark– U.S. dollar exchange rates during this time ranged between 0.4 percent and 0.8 percent.

Figure A2.5.
Figure A2.5.

Daily Volatility of U.S. Dollar Exchange Rates Against Korean Won, Japanese Yen, and Deutsche Mark1

Source: Datastream.1 Measured as standard deviations of daily logarithmic changes.

Did high real interest rates contribute to this stabilization? It is difficult to answer this question without being able to test the alternative hypothesis. Certainly it was important to maintain high real interest rates in order to prevent a flight from won-denominated assets by Korean institutions and individuals, though, as noted above, there were few channels through which this could occur. But the government’s prompt actions in starting to address the problems in the corporate and financial sectors are likely to have done more to rebuild the market’s confidence in the Korean economy.

The second principal motivation for the level of interest rates, namely the need for a tight monetary policy to contain inflation, was open to question. With unemployment at 7 percent, the gap between actual and potential GDP was probably quite large. The experience of such countries as the Philippines and Thailand in the late 1990s and Finland and Sweden in the early 1990s shows that there is no reason to assume a large sudden depreciation will necessarily lead to a correspondingly large acceleration of inflation. Academic work produced after the crisis has investigated the reasons why “pass-through” tends to be weaker than expected in such situations. Burstein and others (2001) cite two countervailing effects: first, consumers tend to substitute domestic for foreign goods; and second, the component of the final price of a “tradable” good that is sensitive to the exchange rate is often quite small relative to domestic cost components such as transportation and distribution. Of course, these experiences, and the lessons that have been drawn from them, were not fully available or understood at the time of the Korean crisis.

High interest rates undoubtedly imposed costs on the domestic economy, but these are difficult to quantify. Given the short-term structure of corporate finance, the transmission of high interest rates to the real economy was rapid. At the time of the crisis, some 35 percent of domestic corporate debt had an average maturity of less than three months, and about 70 percent had a maturity less than one year. One reason for this was the extensive use of three-month promissory notes as a means of payment among enterprises, especially among small and medium-sized ones (Baliño and Ubide, 1999). In the case of Korea, given the high leverage and export orientation of the corporate sector, the adverse balance sheet consequences of a lower exchange rate may well have been much smaller than the cost of higher interest rates (Krueger and Yoo, 2002).

As the crisis developed, the IMF staff became more aware of these vulnerabilities and often mentioned the impact of high interest rates on the corporate sector in program reviews and communications with management. However, the collapse in business and consumer confidence and the sudden, sharp contraction in financial intermediation, which was due to the need to clean up balance sheets and rebuild capital levels, probably played a more important role in creating the recession than did the level of interest rates per se. As the experience of Japan has shown, banks that are burdened with weak balance sheets are usually reluctant to finance business investment, even when their cost of funds is very low. An alternative financing channel, the corporate bond market, began to grow rapidly during this period, but it took time for the necessary market infrastructure to be developed (Oh and Rhee, 2002).22

While the move toward a gradual reduction in policy rates in the aftermath of the crisis was justified, the rates themselves remained above levels that would have been more appropriate to helping the country get out of recession. In this context, the lack of a clearly defined and well-announced framework to guide monetary policy was not helpful.23 The IMF was aware that there was scope for further easing and understood the effect the high rates were having on the corporate sector but, for fear of the crisis returning, was reluctant to allow rates to fall more quickly. In retrospect, an earlier easing of rates would have been justified. However, it must be recognized that the IMF was faced with making a very difficult judgment, based on incomplete information, as to whether an earlier easing of rates would have triggered renewed exchange rate pressures, particularly given the unsettled currency markets in the East Asian region. Moreover, the period of unusually high real rates was only a few months. Given other weaknesses of the economy, particularly the breakdown in financial intermediation, it is doubtful that an earlier loosening of monetary policy would, by itself, have prevented the recession, although hindsight now suggests that some earlier loosening would have been warranted.

Fiscal policy

Korea’s public sector budget was essentially in balance at the onset of the crisis. The authorities projected a deficit of 0.2 percent of GDP in 1997 and a surplus of 0.25 percent in 1998, after surpluses of 0.3 percent in the two previous years. Public debt was only 6 percent of GDP.

Despite the very healthy position of public finances at the start of the crisis, the staff’s initial approach was to favor a tight fiscal policy. In a draft briefing paper prepared before the program mission in late November 1997, APD proposed fiscal measures that would not only pay for the carrying cost of debt issued for financial sector restructuring, but also result in a surplus of 1.5 percent of GDP in 1998. The tighter fiscal policy was meant to secure the needed current account adjustment without increasing the burden on monetary policy and the exchange rate. Some review department comments urged a still greater fiscal adjustment, in order to signal the government’s resolve and to be prepared if restructuring costs were larger than expected. It was pointed out that, given the experience of Thailand and Indonesia, the authorities could not always be trusted to maintain tight monetary policies under conditions of severe financial sector weakness, so that a greater burden of adjustment should be placed on fiscal policy.24 However, this latter approach was rejected by IMF management in recognition of the fact that the precrisis fiscal situation was largely in balance and the need to avoid fiscal “overkill.” As a result of management’s intervention, the proposed surplus for 1998 in the premission brief was reduced to 1 percent from 1.5 percent of GDP.

Fiscal policy was not a major area of disagreement between the IMF and the authorities in negotiating the IMF-supported program. The eventual program envisaged a surplus of about 0.15 percent of GDP in 1998, which was still smaller than in the premission briefing paper. This figure incorporated the surplus of 0.25 percent of GDP projected before the crisis; a 0.8 percent shortfall because of slower growth; 0.8 percent in carrying costs for the financial sector cleanup, based on the assumption that the cleanup would eventually cost a total of 5.5 percent of GDP;25 and offsetting measures of 1.5 percent. As a demonstration of the authorities’ resolve to undertake these offsetting measures, an increase in the transportation tax and an excise tax were among the “prior actions.” This fiscal stance was broadly supported by the Executive Board when the program was discussed, with only one Executive Director expressing concerns and suggesting a more expansionary fiscal stance.

Even the 0.15 percent projected surplus could be said to incorporate an implicit assumption that Korea would run a deficit under the policies stated in the program document. This is because the staff believed that the program’s growth assumption of 2.5 percent for 1998, as agreed with the authorities, was overstated. According to staff members interviewed by the evaluation team, the staff expected growth in 1998 to be zero or even negative. It nevertheless agreed to include a positive growth forecast at the urging of the Korean authorities, who wanted this for political reasons.26

In the early months of 1998, as growth projections worsened, the program assumed an ever greater deficit. The staff recommended that the authorities “let automatic stabilizers work,” in other words, that they take no action to offset the projected deficits. Internal documents show that IMF management was apparently more keen than the staff to allow for flexibility in fiscal policy. An early draft of the staff report accompanying Korea’s late-December request for accelerated disbursements stated that fiscal policy would “need to remain tight.” This conformed to the authorities’ commitment, in the December 24 LOI, that “the initial fiscal adjustment of the program [would] be maintained despite higher costs to the government associated with the larger depreciation of the won and with financial sector restructuring.” However, at the urging of management, the language in the staff report (but not in the LOI) was replaced by a statement that “the staff’s preliminary assessment [was] that … automatic stabilizers should be allowed to operate.”

The Korean authorities were reluctant to do so, in accordance with their traditional inclination toward fiscal conservatism. Government consumption expenditures fell by 0.4 percent in real terms in 1998. Nevertheless, because tax revenues fell even further than did government spending, Korea ended up running a budget deficit of 4.3 percent of GDP in 1998, or 1.5 percent in cyclically adjusted terms. The public sector deficit was further augmented by the activities of off-balance-sheet quasi-public entities such as the KDB (Cho and Rhee, 1999).


In terms of the role of fiscal policy, the Korean situation in November 1997 differed from the typical situation in which IMF assistance is sought. The level of outstanding public debt was very low. The government faced potentially very large contingent liabilities through its de facto guarantee of foreign currency bank debt.27 The bailout of Kia raised suspicions that the government’s de facto domestic-currency contingent liabilities were also large. However, domestic and foreign investors did not doubt either the capacity of the public sector to maintain control of its fiscal processes, or the political commitment of the authorities to maintain a sustainable level of sovereign debt. When international credit-rating agencies lowered Korea’s debt ratings sharply at the end of 1997, they were careful to assert that this reflected the country’s dire liquidity situation, rather than the underlying strength of its economy or its overall debt position.

Under these conditions, there was scope for a “debt for debt” swap, in which the government would draw on its extensive spare domestic and international borrowing capacity to offer its obligations in exchange for those of the country’s troubled financial sector. Indeed, this is what eventually happened. Leaving aside the admittedly important moral-hazard and burden-sharing issues of such an approach, the question for fiscal policy then comes down to whether the carrying cost of the government debt issued in this process should have been paid for through current receipts, as the IMF initially proposed, or through issuing additional debt. A good argument could be made that there was scope for the carrying costs, too, to be financed temporarily by public borrowing, since Korea’s past record was sound enough to convince the market that such borrowing was unlikely to spiral out of control.

Another possible role for fiscal policy would have been to counteract the contractionary effects of the restructuring of the financial sector. In this sense, the emphasis on “reducing the burden on monetary policy” was misplaced. The drive for the banks to write off bad loans and to rebuild their capital adequacy as quickly as possible was exerting a deflationary pull far stronger than monetary policy could have provided.

The initial recommendation for a relatively tight fiscal policy was, in part, the result of an excessively optimistic growth projection and in part a reflection of the IMF’s traditional preference for fiscal tightening in crisis situations. However, within a month or two from the outbreak of the crisis, once it became clear that output would be well below program targets, the IMF showed flexibility in recognizing the need for a looser fiscal policy and transmitting this advice to the Korean authorities. The latter, however, were reluctant to act upon it.

The idea that a country engaged in financial sector reform should pursue a loose fiscal policy in order to support aggregate demand was not unknown at the IMF; the 1998 Article IV consultation report for Japan, produced a few months after the Korean crisis, urged just such a policy. Of course, Korea’s ability to borrow during the crisis was limited by the need to rebuild the confidence of foreign investors, while Japan could finance its deficits at home, and the role of fiscal policy in Japan was itself a subject of considerable debate. There was also a limit to how aggressively the IMF could have pushed for a looser fiscal policy, given the authorities’ preference for fiscal conservatism and the damage to credibility that might have come from any public criticism. Nevertheless, it is striking that such an approach was not considered more seriously at an earlier stage in the crisis.

Financial sector reform

In the years preceding the crisis, Korean policymakers pursued a slow but deliberate policy of financial sector reform. The authorities announced a blueprint for financial liberalization in 1993. This led to deregulating interest rates, liberalizing the issuance of corporate bonds and commercial paper, and sharply reducing subsidized “policy lending” through state-owned institutions.

However, progress on many issues in 1997 remained incomplete, partly because of conflicts between different interest groups in the public and private sectors, and partly because of a reluctance to take bold policy measures in the lead-up to the presidential elections in December. Many observers viewed apparent reform initiatives, such as the establishment of a Presidential Commission on Financial Reform, as attempts to deflect criticism by postponing concrete action until after the elections. IMF staff members involved in surveillance before the crisis told the evaluation team that, despite the many reform initiatives, they were never sure how much genuine reform had actually taken place in Korea. Announced reforms often did not seem to have much practical effect on the behavior of financial institutions. Formal controls on transactions or activities were sometimes replaced by less transparent controls, or by informal channels of influence.

After the crisis hit, the Presidential Commission’s recommendations suddenly assumed much greater relevance. The Commission had recognized that reform was needed, not just in the content of financial sector regulation, but also in the organizational structure of the bureaucracy responsible for regulation. For example, while commercial banks were supervised by the Office of Bank Supervision at the BOK, responsibility for supervising the merchant banks lay with the MOFE. This contributed to the uneven quality of financial sector supervision across different types of institutions.

There were three important differences between the Commission’s recommendations and the restructuring program that was ultimately followed. First, the Commission did not specify the sequencing of the reforms that it recommended. Second, the Commission did not offer recommendations on the resolution of the NPL problem, partly because it was charged with offering a “big picture” vision of reform, but also because its members, like most outside observers, were not aware of the depth of the problem. Third, while it recommended the establishment of a consolidated supervisor, the Commission did not fully address issues relating to the bureaucratic structure of supervision, with the result that political infighting stalled the reform process at a crucial time.

The structural program for the financial sector had two distinct goals: to restore the health of the financial sector through the disposal of bad loans and closing or rehabilitating insolvent institutions; and to institute reforms that would improve the sector’s efficiency and stability and enable it to contribute to Korea’s growth in the longer term. Each of these aspects of the program are considered separately below, although the measures taken in each area were closely related.

Rehabilitating the financial system

With regard to cleaning up bank balance sheets, the strategy followed was similar (though not in all respects identical) to that pursued by other countries facing banking crises in the middle and late 1990s. The key elements were the prompt closure of the most troubled institutions; the extension of the deposit insurance system, funded by government-guaranteed bonds, to protect depositors and prevent bank runs; the utilization of an asset management company, also funded by government-guaranteed bonds, to buy and dispose of bad loans; and the requirement that weak but solvent institutions submit a restructuring and recapitalization plan for approval by supervisors or face closure.28 A bridge bank was set up to buy and dispose of nonperforming assets held by the merchant banks. The asset-management and deposit insurance agencies had been set up prior to the crisis but were given expanded responsibilities.

During the negotiations, the authorities initially resisted some aspects of the IMF’s strategy for cleaning up the financial sector, particularly the proposed closure of insolvent commercial and merchant banks. Such action was unprecedented in recent Korean history and the authorities were worried about the consequences for systemic stability. IMF staff members had the impression that this official reluctance to confront Korea’s financial sector problems influenced other aspects of their interactions with the Korean authorities, for example, the provision of data on NPLs. After the intervention of the Managing Director in the final stage of the negotiations, the authorities agreed as part of the first program to close nine merchant banks and to restructure two large commercial banks. Subsequently, and particularly after the election, the authorities demonstrated a greater commitment to reform, closing additional banks and accepting a more rapid pace of liberalization. What emerged was a politically realistic, yet bold program of financial sector restructuring under a team of competent administrators.

Some of the authorities’ actions did not meet the ambitious timetable set by the two December programs. Rather than being closed or sold off, Seoul Bank and Korea First Bank were nationalized. The government also became a major shareholder in several other commercial banks. Five years later, the privatization of these banks was still not complete.29 Five smaller banks were closed in June 1998, the first such closures in Korea’s recent history, but the rehabilitation plans for other undercapitalized banks were not finalized until September 1998. Legislation allowing supervisory authorities to write down the equity of failed banks without restriction was not passed until August 1998.

Despite these delays, the IMF and international investors remained confident about the Korean authorities’ commitment to reform. This was because, even if measures were delayed or revised, the authorities were careful neither to backtrack from earlier pledges nor to take actions that ran counter to the spirit of reform. Between December 1997 and March 1998, 6 of the 26 commercial banks and 16 of the 30 merchant banks were closed or merged (Baliño and Ubide, 1999). As already mentioned, public funds totaling over 20 percent of GDP would eventually be committed to cleaning up the banking sector. This included equity injections, purchases of subordinated debt, purchases of NPLs, restitution to insured depositors, and funds for recapitalization.

Reforming the financial system

Many of the financial sector conditions in the IMF-supported program called for carrying out recommendations that had been made by the Presidential Commission during 1997. These included creating an independent, consolidated financial regulator; liberalizing the market for ownership rights in financial institutions; removing restrictions on the activities of financial institutions; modernizing monetary policy operations; and completing the deregulation of interest rates. According to interviews with Korean officials, the fact that the financial sector elements of the program were based upon a homegrown policy meant that the authorities were generally more willing to implement these measures than they would have been if they had been entirely imposed from outside. Essentially, what the IMF did in terms of financial sector restructuring was to tip the balance of power in Korea in favor of advancing the homegrown agenda. Starting in early 1998, the World Bank played an important role in advising the Korean authorities on reform policies and in outlining operational measures.

Following the first IMF agreement in December, the previously failed legislation passed the National Assembly, establishing the independence of the BOK and consolidating financial sector supervision in a single agency. However, the institutional arrangement of financial supervision continued to be a major area of dispute among the Korean authorities. The IMF-supported program, like the Presidential Commission, envisaged an independent regulator, but did not resolve the question of its relationship with the BOK and the MOFE. When the authorities began to draw up plans to implement this provision, the IMF at first took a neutral stance over the disposition of the new regulator, while insisting that it should remain independent of the MOFE. Ultimately, though, the IMF felt compelled to take a position in order to speed up the restructuring program. With the IMF’s backing, the program ended up adopting the MOFE’s vision of subordinating the Financial Supervisory Service (FSS) to a government agency, the FSC. This setup had some virtues. The new supervisory system was not formally part of the MOFE or the BOK, and the bureaucracy charged with managing and resolving the crisis was separate from the ongoing supervision function. However, the new framework had the disadvantage of allowing the MOFE to exercise influence over supervision through the participation of its officials in the FSC, a situation that was not entirely in keeping with the preferences of the IMF and World Bank.

As with the cleanup, the actual implementation of the promised reform measures was somewhat slower than had been specified in the IMF-supported program. New loan-classification standards and provisioning rules were put in place in June 1998, but the FSS, charged with carrying out the supervisory function, did not formally begin operations until January 1999. Rules imposing stronger risk management for banks’ foreign exchange operations also did not become effective until 1999. Limits on large credit exposures, intended to insulate the banking system from the failure of a small number of large companies, as happened in 1997, have been phased in only gradually, in order not to disrupt credit flows to the companies concerned. At the same time, the cumulative progress of reform over the past five years has been impressive, and there have been no attempts to roll back previously implemented reforms.


Over time, the financial sector restructuring program achieved its goals of facilitating the relatively prompt removal of bad loans from bank balance sheets and reducing the system’s vulnerability to external shocks. The rapid restructuring of the banks and the short timetable for attaining high capital adequacy levels contributed to the severity of the economic slowdown in 1998.30 Yet, a less rapid cleanup would not necessarily have resulted in a better outcome. Other liberalization measures, such as those that made it easier for corporations to issue bonds directly to investors, fostered the development of alternative financing channels. This strengthened the Korean economy by reducing the dependence of investment on the health of the banking sector.

There were, however, gaps in the new supervisory framework. A prominent example was the investment trust company (ITC) sector. The pressure on the banks to recapitalize and restructure led them to reduce their corporate lending and the returns they could offer to savers. This provided a window of opportunity to the ITCs, which channeled funds from small investors into the rapidly growing corporate bond market (Oh and Rhee, 2002). ITC accounts were intended to behave like mutual fund holdings, but in practice many ITCs offered guaranteed yields to investors. In 1999, when corporate bond prices fell following a large corporate bankruptcy, the ITCs could not meet the guarantees and the result was widespread panic selling. This episode, while causing losses to many investors and disruption to Korean capital markets, did not have a substantial impact on the country’s overall investment and growth trends—a sign that the system had become more resilient, even though clearly more effort was needed in the area of improved supervision.

Nonfinancial structural reforms

As was the case with the financial sector, there had been an ongoing debate within Korean society before the crisis on the optimum design of the corporate sector. Some reformers opposed the concentration of economic power in chaebol. Influenced by these ideas, the authorities, in a dramatic reversal of policy, had begun to allow large corporate bankruptcies (such as that of Hanbo) to take place even before the onset of the crisis. But others in Korea advocated the preservation of the chaebol system in light of its track record in facilitating rapid economic growth.

Comprehensive reforms in the nonfinancial corporate sector envisaged in the December 4, 1997 program included provisions on accounting standards, bankruptcy procedures, and governance mechanisms. In the late December program and subsequent reviews, provisions were added mandating the appointment of outside directors, liberalizing the market for corporate control, and enhancing labor market flexibility. In the World Bank’s structural adjustment loans, more specific measures were identified, such as curtailing emergency loans, facilitating use of debt-equity conversions to address excessive leverage among chaebol affiliates, reducing cross-guarantees, providing additional encouragement for corporate mergers and acquisitions, debt restructuring and asset dispositions, and improving procedures and coordination for court-supervised insolvency (Mako, 2002; Joh, 2002). Thanks in part to these efforts, the debt-equity ratio of the manufacturing sector was gradually reduced from nearly 400 percent in 1997 to 211 percent in 2000 (Im, 2002).

A common criticism of the Korean program, echoed at the time in the Korean press, was that certain measures were included in the program at the insistence of major shareholder governments to serve their particular national interests. For example, the requirement that Korea eliminate its “import diversification” program was said to be a response to Japanese pressure,31 while the measures to allow increased participation of foreign institutions in the Korean financial system were alleged to reflect pressure by the U.S. authorities on behalf of U.S.-based institutions.

The IMF’s largest shareholder governments made no secret of their view that IMF assistance should be accompanied by strong reforms. The U.S. authorities, in particular, insisted that strong reform measures should be a condition for IMF support. However, internal IMF documents do not support the allegation that the specific policy measures mentioned were included solely because large IMF shareholder governments demanded them.32 These governments may indeed have had an interest in these measures, but they were also on the agenda of policy reforms which had surfaced in the course of IMF surveillance and had been discussed by the staff with the authorities. For example, increased participation by foreign financial institutions in the Korean financial system had long been on the list of IMF recommendations made in the surveillance process and was among the measures recommended by the Article IV consultation mission two months earlier. It was in the briefing paper prepared by the staff on the eve of the negotiations, and is a policy recommended by the IMF for virtually all emerging market economies. After the crisis started, takeovers and other asset purchases by foreign institutions were seen as a way to improve bank governance and to reduce the amount of public money needed to recapitalize the banking system. Similarly, the elimination of the import diversification program had been included in the recommendations of the earlier Article IV consultation mission and was incorporated in the prenegotiation brief. The staff saw this as a vital trade liberalization measure that would demonstrate the authorities’ commitment to reform.

As with the financial reforms, the Korean authorities initially were eager to demonstrate their commitment to the program by moving forward rapidly on implementation. However, after the economy survived the initial phase of the crisis and began a quick recovery, the government reduced its efforts to pursue painful and costly restructuring. At the end of 2000, more than a quarter of manufacturing firms still had earnings that were below their interest costs. There were also signs of persistent official favoritism toward chaebol. For example, when a large conglomerate experienced financial difficulties in 2000, at a time when illiquidity in the corporate bond market made it difficult for companies to raise new capital, the Korean government mobilized such means as “fast-track underwriting” (in which the KDB refinanced maturing corporate bonds at a penalty interest rate) to prevent the company from going bankrupt.


Some observers have argued that nonfinancial structural reform measures were not crucial to the resolution of the crisis, and have cited the fact that output recovery began well before many of the key reforms were implemented (Feldstein, 1998; Park, 2001). In particular, they argue that labor market and trade measures were a distraction from the core program requirements, although they may well have proven helpful to the long-term efficiency of the Korean economy.33

The effectiveness of some of the structural measures in the IMF-supported program can also be questioned. Some of them appear to have been rushed into implementation because of the short time horizon. Some staff members told the evaluation team that they had been under pressure to show quick results and had known that they would need to reduce intensive monitoring once the crisis had passed. This led to a focus on measurable benchmarks that could be achieved in the first six months or so of the program, at the expense of more lasting but less visible actions. For example, companies listed on the Korea Stock Exchange were required to appoint at least one outsider to their boards of directors. Some have questioned whether the newly appointed outside directors were truly independent of management or able to exert influence on corporate decisions (Joh, 2002).

Defenders of the IMF-supported program respond that, aside from the intrinsic merits of the policies followed, a demonstration of the authorities’ commitment to reforms in both the financial and nonfinancial fields was needed to restore international confidence and promote rapid recovery. There were also some cases where nonfinancial structural measures were intended to facilitate a rapid recovery from the crisis, and thus formed a vital element of the IMF-supported response. In particular, smoother bankruptcy procedures and labor market reforms were designed to promote the reallocation of industrial assets and reduce the consequences of the reforms for employment.

It is difficult to evaluate these arguments because the objectives for many of these reforms were never fully spelled out. While the weak governance and high leverage of the Korean industrial sector certainly contributed to the crisis, the immediate need for action in these areas was not as clear as the need to address solvency issues in the financial sector.

To the extent that the reforms in the nonfinancial area were intended to facilitate other policies more directly linked to resolving the crisis, the argument for them would be much stronger. Even in this case, however, the IMF might have been better advised to confine its advice and conditionality to a narrower range of issues, and then to let the Korean authorities define their own agenda for implementing this more focused set of policy measures. This is particularly true of many of the trade and other external liberalization measures that were already mandated by agreements with the OECD and the World Trade Organization (WTO). The role of formulating and facilitating the needed reforms would then fall to institutions that are better placed to do so.

Program Financing and the Debt Rollover

This section reviews the process by which the financing package associated with the December 4, 1997 program was determined and whether the final size of the package was appropriate for the circumstances. Our overall assessment is that there was considerable ambiguity surrounding the publicly announced bilateral “second line of defense” and this damaged the program’s credibility. It forced the staff to adopt unrealistic assumptions in formulating the December 4 program, which led to underfinancing. Korea’s underlying liquidity shortfall was not resolved until the coordinated rollover agreement at the end of December.

The December 4 package

At the start of the negotiations with Korea in late November 1997, the staff estimated the country’s financing gap during the years 1998 and 1999 at US$25 billion, of which US$20 billion was for the first year (Table A2.1, columns 1 and 2). Funds were needed, it was thought, to finance the current account deficit (estimated to be US$2 billion for 1998), portfolio outflows (another US$3 billion), and a US$13 billion increase in reserves. These figures were based on two crucial assumptions. First, Korea’s short-term external debt, then estimated to total US$66 billion, was projected to be fully refinanced, though it was assumed that there would be little or no new short-term borrowing. Second, Korea’s reserves of US$30 billion were thought to be enough to cover the country’s obligations until the program funding was disbursed. No financing need was envisioned for 1997.

Table A2.1.

Korea: Balance of Payments and Financing Requirements

(In billions of U.S. dollars)

article image
Source: IMF database and documents.

Adjusted to include the impact of foreign currency liquidity support by the BOK to overseas branches of Korean banks.

Adjusted to exclude the impact of foreign currency liquidity support by the BOK to overseas branches of Korean banks.

An additional financing gap of US$5 billion was projected for 1999.

These assumptions had to be revised radically almost as soon as the IMF team arrived in Korea, because of a combination of new information and revised assumptions about the behavior of external creditors. It was discovered that Korea’s usable reserves—that is, official reserves, minus the amount that had been deposited at overseas bank branches to cover short-term debt repayments—were around US$11 billion, and falling very fast. This pointed to the fact that the major drain on the capital account was likely to arise, not from a reversal of portfolio investment, but from bank debt repayments.

The debt, in turn, was far larger than initially thought, because it comprised obligations of overseas borrowing by Korean institutions, which were not included in residence-based debt data used by the IMF. The most important component of this additional debt was some US$22 billion in offshore borrowing by overseas branches of domestic banks. After correcting for double-counting and including offshore borrowing and the debt of Korean banks’ foreign branches and subsidiaries, short-term external debt (bank and nonbank) was estimated at around US$86 billion at end-September 1997, of which banks owed US$62 billion. It was this component of the debt that triggered the crisis. By the end of November, short-term external bank debt had fallen from US$62 billion at end-September to US$49 billion, and further to US$33 billion at the end of December, representing an outflow of US$16 billion in a month.

Estimating the amount of financing needed in the fast deteriorating situation that prevailed in November 1997 was no easy task. In the event the staff report supporting the SBA request contained two different estimates of the financing needed for the rest of 1997 and 1998. In the text of the report, the total financing needed was indicated as US$55 billion but the detailed estimates in Table 6 of the same report (reproduced in columns 3 and 4 of Table A2.1) showed a smaller figure of US$33 billion. The difference between the two estimates was not reconciled in the staff report but interviews with the staff indicate that the larger estimate resulted from the initial expectation that the rollover rate on short-term bank credit would be only 20 percent. It was recognized that providing large volumes of financing from the IMF to Korea would be difficult because of the IMF’s resource constraints and also because Korea’s quota was unusually small relative to the size of the economy.34 However, the staff worked on the assumption that IMF financing could be supplemented by additional amounts from other IFIs (the World Bank and the ADB) and bilateral sources (i.e., the second line of defense).

The second line of defense

The incorporation of bilateral financing to supplement IMF and other IFI resources was in line with the principles of the so-called Manila Framework, endorsed by the APEC summit meeting only a few days earlier on November 24, 1997, which envisaged the provision of bilateral financing to support IMF-supported programs when necessary. However, the availability of these resources turned out to be uncertain. There also appear to have been miscommunications on the second line’s conditions and timing (in relation to the disbursement of the IMF’s own resources) that compounded the problem.

The staff had initially incorporated a specific level of bilateral financing in the proposed IMF-supported program. At virtually the last minute, headquarters informed the mission that it could no longer count on the second line of defense being available as part of the financing for the “baseline” program. Additional decisions would be needed before any part of the financing could be released, and the financing could in any case not be made available for several weeks.

IMF management and staff recognized that, without the assured availability of official bilateral financing, the program would be underfinanced. They accordingly approached the major shareholder governments to explore the possibility of concerted action to involve the private sector in some form of rollover. The major shareholders, however, were reluctant to use nonmarket instruments to influence the behavior of private sector institutions, given the lack of clearly defined regulatory authority and the fear that such action might precipitate an exodus of capital from emerging markets.

Faced with these circumstances, the staff presented a financing scenario in which the availability of bilateral financing was not essential. This was achieved by modifying one of the key assumptions which determined financing need. Specifically, the fraction of short-term interbank loans from external creditors that was assumed to be rolled over was raised from the 20 percent assumed initially to 80 percent.35 This arbitrary adjustment ensured that the amount of financing provided in the December 4 package would meet Korea’s ex ante needs as projected in the program documents.

Although the program financing requirement was reduced in this way, the package publicly announced was US$55 billion, including a second line of defense in excess of $20 billion. The press notice released on December 4 announcing the Executive Board’s approval of the program specifically stated:

[A] number of countries (Australia, Belgium, Canada, France, Germany, Italy, Japan, the Netherlands, Sweden, Switzerland, the United Kingdom, and the United States) have informed the IMF that they are prepared, in the event that unanticipated adverse external circumstances create the need for additional resources to supplement Korea’s reserves and resources made available by the IMF and other international institutions, to consider—while Korea remains in compliance with the IMF credit arrangement—making available supplemental financing in support of Korea’s program with the IMF. This second line of defense is expected to be in excess of US$20 billion.

Market participants were highly skeptical as to whether the second line of defense would truly be available. Political opposition to “bailouts” of crisis countries was running high in several of these contributing countries and, since there was no clarity on the circumstances under which the amounts would be released, their availability was widely discounted.

The unrealistic rollover assumption implicitly contained in the December 4 program lowered the package’s probability of success. It meant that, without a radical turnaround in confidence, the program was likely to be underfinanced. When this turnaround did not materialize, the credibility of the program (and of the IMF more generally) was damaged, making the task of formulating a revised response to the crisis more difficult.

Negotiations on a second line of defense between the Korean authorities and those of the contributing countries eventually took place in the early months of 1998. These negotiations did not lead anywhere, however. Those close to the talks have advanced differing reasons as to why this was the case. There were differences of view regarding the appropriate pricing and technical conditions for the facility. However, the most likely explanation for the absence of agreement is that private sector financing conditions by that time had improved substantially, so that setting up another official financing facility did not seem necessary.

The coordinated rollover

With the failure of the December 4 program, usable foreign exchange reserves dwindled rapidly and the won fell sharply in the first weeks of December. The staff projected that usable reserves, which had been temporarily boosted by the IMF disbursement in early December, would fall from US$8.5 billion on December 14 to US$4.5 billion at year-end.36 Further official financing was neither politically nor practically feasible. An attempted international bond offering by the state-owned KDB failed and was hastily withdrawn. A more vigorous and sustained increase in interest rates might have attracted some capital back into the country, but it could also have caused so much damage to Korea’s highly leveraged corporate sector that its impact on market confidence could very well have been negative. The IMF mission in the field advised management that a restructuring of Korea’s short-term debt would be necessary for resolving the liquidity problem.

It was in this context that the idea of pressing Korea’s creditors to agree to a coordinated rollover and a maturity extension of their short-term claims gained renewed prominence.37 The idea had been circulating among IMF officials and the large shareholder governments, as well as in the private sector, virtually from the start of the crisis and, as noted, had been raised by IMF management in its consultation with the major shareholder governments. It was also raised by foreign bankers in Korea with the authorities and IMF staff in early December, but did not receive support at that time from the banks’ head offices.

The decision to urge the rollover on creditor banks appears to have arisen from discussions among Korean, U.S., and IMF officials immediately after the Korean presidential election on December 18. The President-elect’s statements in support of the IMF-supported program had boosted its credibility in the markets. The incoming administration began to cooperate with the outgoing administration in vigorously implementing the program. Officials from large shareholder governments put aside their earlier concerns about excessive intervention, because of the gravity of the situation and the evident failure of the approach that had formed the December 4 program. Once the decision to pursue the rollover was made, it was arranged relatively quickly and announced on December 24, 1997. Central banks and finance ministries in the industrial countries contacted large banks based in their jurisdictions, which in turn contacted other lenders. The banks agreed to maintain their existing credit lines while they negotiated to extend the maturities of their claims on Korean banks. A system of daily monitoring of rollovers by individual banks, established with substantial IMF inputs, proved crucial in ensuring compliance.

Negotiations between the Korean government and the banks over the maturity extension began in early January 1998.38 A tentative maturity-extension agreement was concluded on January 28, and the final terms were settled in February. While these negotiations were under way, a second rollover announcement was made on January 16, which committed the banks to maintain existing credit lines through the end of March 1998.

The pricing on the extended bonds shows that the market’s confidence in the Korean economy had already started to revive. In April 1998, some US$22 billion of eligible bank debt maturing during 1998 was exchanged for government-guaranteed loans with from one to three years’ maturity and interest rates between 225 and 275 basis points over LIBOR. In early April, the Korean government issued US$4 billion in 5- and 10-year global bonds, respectively at 345 and 355 basis points over U.S. treasuries. The spreads on both transactions were well below that on the JP Morgan EMBI+ index, which was never lower than 464 basis points in April 1998. Even after differences in maturity are taken into account, the more favorable borrowing terms offered to Korea suggest that by that point the market already assigned Korea a lower credit risk than most other emerging market borrowers.

Subsequent events would justify the confidence that international creditors showed in the Korean financial system in early 1998. All of the extended loans would be repaid by the original borrowers; the government guarantee was never exercised. As Korea’s external financing conditions improved, most of the borrowers took advantage of prepayment options to refinance the debt at lower interest rates. Although only 63 percent of the debt was scheduled to mature by April 2000, 90 percent of it would end up being repaid by that date.39


It is of course easier to draw lessons on matters of program financing after the fact than at the time, when information was incomplete, market reactions could not be anticipated, and decisions needed to be taken rapidly. Nevertheless, three aspects of the approach to financing and the role of the private sector in the Korean case are worthy of note.

First, to the extent that the Korean economy in late 1997 faced a shortage of liquidity rather than a long-term debt-sustainability problem, the successful resolution of the crisis depended as much on how and how fast new financing was to be provided as on whether it would be provided. A delayed or highly conditional commitment of funds would do nothing to reverse the drive by creditors to liquidate their investments while they still could. An immediate commitment of liquid funds, from whatever source, would convince lenders that their chances of repayment were reasonably high and that it would be worthwhile rolling over existing credit lines, though perhaps at a higher risk premium than before the crisis.

In this respect, the ambiguity over the second line of defense was clearly counterproductive. The IMF and the national authorities of the contributing countries may have hoped that the mere announcement of broad international support, in conjunction with strong IMF endorsement of the Korean authorities’ policies, would be enough to restore market confidence and make any actual payout unnecessary. Given the absence of deeper solvency concerns, the announcement of official financing could have had the intended catalytic effect, and one can argue that this approach ex ante was worth the gamble. However, staff calculations suggest that the assumed increase in the rollover rate was unrealistic, especially in a very short time. Had the second line of defense been firmly committed, with clear indications to the markets that the funds would be automatically released if needed, the large “headline” figure might have produced a catalytic effect. In the event, there was too much uncertainty about their availability and the effort to influence the subtle dynamics of market confidence backfired. By including some US$20 billion that was not backed by actual commitments, the December 4 package only emphasized the extent of Korea’s cash shortfall. The market became skeptical, and the announcement of the IMF package failed to provide the boost to confidence that had been hoped for.

The second lesson to be drawn is that, in the end, the coordinating role of the IMF in the context of the rollover agreement proved to be at least as useful in resolving the crisis as its ability to provide or mobilize financial resources. The success of the coordinated rollover and private sector debt restructuring would ultimately render the second line of defense irrelevant. While the authorities of the IMF’s large shareholder governments made the key decision to pursue the rollover plan and to exert the necessary moral suasion on banks, the IMF played a useful role in facilitating communication among the different actors, in providing information, and in certifying that the policies to be pursued by the Korean authorities were appropriate. No single national government, nor any private sector institution, could have played this role as effectively.

A third lesson is that, for the success of a large financing package, the IMF’s coordinating role must be complemented by strong engagement on the part of its large shareholders. The role of the United States in pressing for vigorous reforms has already been noted. As part of this process, officials of the U.S. and other large shareholders were in regular communication with IMF staff and management during and after the program negotiations. However, the public face of this involvement must be managed carefully. The presence of a U.S. Treasury official in close proximity of the negotiations caused some in the public to have a wrong perception of the IMF involvement in Korea.


A definitive statement on the “success” or “failure” of the IMF-supported program in Korea would depend on one’s criteria for success. In terms of stabilizing markets and reversing capital inflows, the program announced in early December was clearly a failure, while the late December package can be called a success. However, one should also acknowledge that the key features of the second program became acceptable to the international community only after the strategy of the first was tried and proven to have failed. The depth of the 1998 recession may, in part, be attributed to the stringency of the financial sector restructuring measures required in the program—but significant restructuring would have been necessary whether or not a crisis occurred, and in any case the economy’s subsequent strong recovery suggests that these effects were temporary. The program induced the Korean authorities to take the necessary decisive steps toward reforming the economy and, in this sense, made a contribution to building the foundation for Korea’s impressive recovery. However, some needed reforms were later delayed or scaled back.

This annex has identified specific missteps in surveillance before the crisis, in the formulation of the adjustment program, and in the provision of financing that suggest lessons for the future. These are summarized below. More specific recommendations, including a discussion of the extent to which these lessons have already been identified and acted upon within the IMF, are discussed in the main report.


Partly because of Korea’s consistently strong economic performance, IMF surveillance did not fully anticipate many of the elements that would contribute to the Korean crisis. With hindsight, shortcomings can be detected at two levels: the analysis of the implications of Korea’s capital account liberalization policies in the 1990s, and the analysis of the vulnerabilities facing Korea in the months immediately preceding the crisis in 1997. Specifically, the IMF focused too much on the degree of capital market liberalization, and not enough on sequencing, thereby underestimating the systemic vulnerabilities introduced by a policy that combined liberalization of short-term flows, controls limiting long-term flows, and poor supervision of some of the institutions that borrowed externally. This was in keeping with the IMF’s standard approach at the time, which viewed financial sector issues in terms of their impact on microeconomic efficiency rather than in terms of whether they might increase the risk of an external crisis.

IMF surveillance in the months preceding the crisis did identify many of the relevant vulnerabilities. However, it paid insufficient attention to issues that would prove central to the onset and evolution of the crisis, and the overall assessment proved to be excessively optimistic. In retrospect, five misconceptions hindered the ability of the staff to offer a more accurate assessment:

  • An overestimation of the flexibility in Korea’s exchange rate policies.

  • An underestimation of the risk of a breakdown in funding the capital account. While recognizing that such a risk was present, particularly in the crisis conditions then prevailing in East Asia, the staff concluded that the authorities could handle any pressures by making renewed efforts in the area of financial reform, by addressing financial sector weaknesses, and by loosening controls on long-term external borrowing.

  • Excessive optimism about the short-term impact on growth of rehabilitating and reforming the financial sector. The narrow range of growth estimates considered, based on the remarkable stability of Korea’s growth rates over the previous decades, prevented the staff from exploring the consequences of a more serious slowdown.

  • Insufficient attention to relevant market indicators, some of which (such as spreads on KDB issues) showed mounting wariness among investors well before the crisis began in earnest.

  • Advice in the area of financial sector reform that was primarily oriented toward improving the long-term health and efficiency of the system. While this advice was generally well thought out, in the conditions of the summer and fall of 1997 when investors had become significantly more risk-averse, advice on securing the system against a possible crisis and preparing for the consequences of such a crisis might have been more helpful.

Several of these misconceptions had their origin in, or were exacerbated by, incomplete information and poor data availability. As discussed in the main report, this is an area in which substantial progress has been made since the Asian crisis, through the various initiatives on standards and codes.

Program design
Monetary policy

Some increase in interest rates was justified at the time of the crisis, given the need to prevent a collapse of the exchange rate and to maintain positive real rates in the face of high inflation expected to result from the depreciation of the won. The authorities also needed to demonstrate their determination to respond forcefully to the crisis. But, given the nature of the crisis, too much reliance was placed on high interest rates to stabilize the won. The key immediate issue in resolving the crisis was Korea’s lack of liquidity, and there were too few channels through which high interest rates could remedy this shortfall. The lack of ownership of monetary policy on the part of the authorities no doubt further weakened its credibility and hence its signaling effect. Hindsight suggests that rates were maintained at a high level in early 1998 somewhat longer than necessary, although an excess of caution was understandable under the circumstances. However, the stance of monetary policy was not the major cause of the steep output decline.

Fiscal policy

Given the low stock of public debt, the IMF could have urged Korea to use fiscal policy to counteract the likely contractionary effects of financial sector restructuring from the beginning of the crisis, rather than waiting until the early months of 1998 to start giving this advice. In any event, the Korean authorities, reflecting their tradition of fiscal conservatism, were not very receptive to this advice and cut government expenditures. Fiscal policy nevertheless ended up being countercyclical because tax receipts fell even further and because of off-budget activities.

Structural reforms

The financial and nonfinancial structural reforms were extensive, and had a positive effect in improving the efficiency and stability of the Korean economy. In retrospect, however, while the IMF was justified in using its leverage to insist on such change as a condition for financial support, too much attention was paid to producing visible results quickly. More emphasis should have been placed on the overall strategy, not on specific short-term measures, with the authorities being given greater freedom in setting their own agenda. In the case of the financial sector, a home-grown agenda was already available in the form of the reports of the Presidential Commission, and efforts in this area benefited from the sense of country ownership. In the case of nonfinancial reforms, the extent of ownership was less clear. The immediate need for action in areas such as corporate governance, while potentially important in the long term, was not as apparent as the need to reform bankruptcy laws or address solvency issues in the financial sector.

As confidence rapidly returned to Korea, some aspects of financial sector reform were delayed, while many nonfinancial structural measures were in the end never fully implemented. Particularly given the negative backlash some of these measures created on the public perception of their origin, the IMF might have been better advised to confine its advice and conditionality to a narrower range of issues, and then let the Korean authorities define their own agenda to implement this more focused set of policy measures. This is particularly true of trade and other external liberalization measures, which were already mandated by Korea’s agreements with the OECD and the WTO.

Financing and the debt rollover

The strategy adopted in the first program was predicated on the hope that tough monetary and financial sector policy measures would be sufficient to bring about a spontaneous rebound in confidence. In support of this strategy, the announced package kept a large “headline figure” that included a component whose availability was uncertain and was discounted by the markets. The attempt to present the financing package in as favorable light as possible proved damaging on two levels: in the short term, to the market’s confidence in Korea’s ability to overcome the crisis and, in the longer term, to the credibility of IMF-led financing arrangements generally. If the IMF and the authorities of the major shareholder governments had acknowledged the limited availability of these funds from the beginning, there might have been an earlier effort to seek alternative solutions, including a coordinated rollover of short-term debt.

Admittedly, it is difficult to be certain whether the private sector rollover and maturity extension of short-term bank debt could have been arranged earlier than they were. Some creditors wanted to eliminate their exposure to Korea under any circumstances. Others might have been more receptive to a coordinated rollover, but would not have wanted to make any further commitments in advance of the elections on December 18, 1997. Yet, according to individuals in the private sector interviewed by the evaluation team, from the very beginning of the crisis, some—if not many—creditors expressed an interest in finding a collective solution of some kind.

If the interest of some private creditors in concerned action had been recognized from the start of the IMF’s involvement, there might have been a greater effort to establish contacts, think through the broad outline of such an agreement, and follow up on private sector initiatives, such as those of the Seoul Foreign Bankers’ Association. IMF management appears to have understood that some concerted action might be necessary from the outset and communicated this message to the major shareholders, but the authorities of the IMF’s large shareholder governments were initially reluctant, fearing that such action might set undesirable precedents and adversely affect the flows to other emerging economies. Given the domestic political uncertainties prevailing before the elections, and the constraints faced by the major shareholders, it may well be that the first strategy needed to be proven to have failed before concerted action could be attempted.

Appendix 2.1 Korea: Timeline of Major Events1

article image
article image
Sources: Bloomberg, Korean government official homepage, Moody’s, Standard & Poor’s, IMF, and local Korean newspapers.

Local time unless noted.

Eastern standard time in the United States.



Much of the background material in this section is drawn from Kim and others (2001) and Johnston and others (1997).


The capital accords agreed by the Basel Committee on Banking Supervision in 1988 allowed a lower capital charge for obligations of (or guaranteed by) OECD member governments and for short-term loans to banks based in OECD member countries. However, the accords only prescribed a minimum charge. Regulators were free to set a higher charge for specific borrowing countries.


Members of the OECD agree to adopt the organization’s legal instruments, including the Code of Liberalization of Capital Movements and the Code of Liberalization of Current Invisible Operations (covering cross-border financial services). These codes incorporate a commitment to move toward full liberalization and not to introduce new restrictions. The existing members of the organization make the final decision on accepting new members, based on the recommendation of the OECD secretariat and committees.


This indeed occurred to some extent at the domestic level in the first half of 1998, but could not happen at the international level because Korean borrowers were not well enough established on international capital markets.


See for example Folkerts-Landau and Lindgren (1998), a draft of which had begun circulating internally in late 1997.


The same report noted that increased net private inflows had been associated with increased domestic credit growth, inflation, and current account deficits in Korea, Thailand, and Indonesia, but not in Chile, where capital inflows seemed to substitute for domestic credit growth. One reason for this, the authors suggested, was that Chile had done more to improve prudential standards before starting to liberalize its capital account. At the time the paper was written, Thailand had already been hit by a crisis, and Indonesia had started to experience its own difficulties. But the appropriate parallel to Korea’s vulnerability was not drawn.


The staff report for the 1997 Article IV consultation was never presented to the Executive Board because its relevance was overtaken by subsequent events.


As an example of an alternative response to the regional crisis, Taiwan Province of China successfully fended off a potential crisis by moving to a more flexible exchange rate policy in mid-October. The New Taiwan dollar weakened roughly 8 percent in the three days following this policy shift.


Given the strong growth that had already occurred in the first three quarters of 1997, this represented a prediction that there would be little or no growth in the fourth quarter, and growth below potential in 1998.


Under conditions of full capital mobility and liquid money and bond markets, the forward won-dollar exchange rate would simply correspond to interest rate differentials between Korea and the United States, but these conditions were not present for Korea at that time. The forward rates in the nondeliverable forward market were more depreciated than those in the relatively thin onshore won-dollar forward market during this time, implying that the onshore rates were being artificially supported by official intervention (Park and Rhee, 1998).


However, many of the relevant issues were well known to the IMF staff, having been addressed in studies such as Lindgren and others (1996) and Alexander and others (1997).


The briefing paper prepared for the 1997 Article IV consultation identified four priorities for reform of the Korean financial sector: removing nonprudential controls on balance sheets, which had served as vehicles for political involvement; removing the implicit and explicit guarantees against bank failures; ensuring “a well-targeted safety net,” an apparent reference to depositinsurance; and facilitating merger and acquisition activity in the financial sector.


The 10 Article IV consultation missions from 1988 to 1997 were headed by six different individuals, though the same person headed the consultation missions in 1995, 1996, and 1997. Only one other staff member participated in more than 2 of the 10 consultation missions.


On the other hand, with Korea’s accession in December 1996, the OECD was developing considerable expertise on Korea. In view of this, the IMF later invited the OECD to provide inputs into the structural conditionality of the December 1997 program in the area of corporate governance.


The staff report for the 1995 Article IV consultation remarked that “Korea’s economic statistics [were] of high quality and [were] reported to the IMF on a timely basis.”


In December 1997, data were available from the BIS consolidated banking statistics through the end of 1996. This put Korea’s liabilities to reporting banks at US$100 billion, of which US$67.5 billion was short-term (by resident maturity). This represented an increase in bank debt of US$22.5 billion over the previous year, including an increase of about US$13 billion in short-term debt. More up-to-date data were available from the “locational” series, published in November 1997, with data covering up to end-June 1997. This suggested that borrowing from international banks had continued to grow in the first half of 1997.


For example, some of this information was available from the U.K. and U.S. national supervisory data.


RES was opposed even to the gradual lowering of rates that occurred in early 1998, and continued to urge that policy be oriented toward a target range for the exchange rate.


Independent inflation forecasts for 1998 were around 11 percent in the early months of that year, suggesting that real interest rates measured using expected inflation were only a few percentage points below the levels cited here. See Bloomberg News, “Korea’s Consumer Prices Fall 0.2 percent in March,” March 31, 1998. In terms of realized monthly inflation rates, the real call rate was briefly negative in December because of the one-time jump in prices resulting from the depreciation.


However, residents could still circumvent these restrictions by accelerating or postponing cross-border payments.


Residents’ foreign currency deposits fell 31 percent in U.S. dollar terms in December 1997. This suggests that local residents did not try to flee the currency, but instead participated in the global run on the dollar holdings of the Korean banking system. Foreign currency deposits were stable as a fraction of total deposits during December, as these withdrawals were balanced by the won’s depreciation (which increased the value of dollar deposits relative to won deposits).


Later events would show that the growth of domestic corporate bond issuance had been partially supported by lax supervision of the investment trust companies that were a primary vehicle for retail investors (see section on “Financial sector reform,” below).


The program contained a broad inflation target, but no mechanism was specified to achieve that objective.


While there was no explicit discussion of fiscal sustainability, it was implicitly recognized that, given Korea’s low precrisis level of sovereign debt, sustainability was not an issue, provided that the crisis was resolved quickly.


The net fiscal costs of bank restructuring, now estimated at 23 percent of GDP, have turned out to be still greater than had been feared in 1997.


However, if there was staff pessimism about near-term growth prospects, there was no evidence of this even in internal program documents before the late spring of 1998. The confidential premission brief prepared on November 21, 1997, also projected 2.5 percent growth for 1998. The staff report accompanying the second biweekly program review, prepared on January 8, 1998, stated that, “[the] downturn in growth [was] likely to be sharper than previously expected, particularly during the first quarter.” Yet, that review included a positive growth forecast for 1998, as did every subsequent program review until that of May 19, 1998.


Ambiguity about the amount of liabilities covered by this guarantee was a further source of difficulty. At end-September 1997, short-term external debt of domestic commercial banks resident in Korea was roughly US$24 billion, but a broader definition that included medium- and long-term debt, borrowing by foreign branches of Korean banks, and borrowing by nonbank financial institutions would raise this figure to roughly US$106 billion.


Dookyung Kim (1999) and Baliño and Ubide (1999) review the early stages of this process.


Average government ownership stakes in commercial banks, weighted by bank assets, rose from 17 percent at the end of 1996 to 58 percent at the end of 1998, then fell to 34 percent at the end of 2001. In January 2002, the authorities announced a plan to complete the privatization process over the next three to four years. A majority stake in Seoul Bank was sold to another Korean bank in September 2002, with the government planning to sell the remaining 31 percent by March 2004.


Domaç and Ferri (1998) find evidence that the contraction of bank credit and increase in bank lending spreads in Korea contributed to the fall in activity after the crisis. Hyun Kim (1999) finds that the decline in bank loans resulted from a contraction in supply (the willingness of banks to lend) rather than demand (that is, a fall in investment), though Ghosh and Ghosh (1999) find the opposite result. Borensztein and Lee (2000) find that there was a reallocation of credit from less efficient (including chaebol-connected) borrowers to more efficient ones.


This program, which restricted imports from countries with which Korea had a large bilateral trade deficit, was designed to reduce certain categories of imports from Japan.


According to former senior U.S. officials interviewed by the evaluation team, their only direct input was to introduce a penalty rate for the BOK’s foreign exchange advances to Korean financial institutions.


There were 21 “structural performance criteria” specified in the course of the three-year IMF-supported program in Korea, that is, an average of seven a year (Chopra and others, 2002). The LOI attached to the December 4, 1997 program identified five “prior actions” to be taken by the authorities before the Executive Board approved the SBA. The total number of structural conditions was close to the median for SBAs during this time period, and somewhat less than that for longer-term IMF arrangements such as the Enhanced Structural Adjustment Facility and the Poverty Reduction and Growth Facility. See “Structural Conditionality in Fund-Supported Programs,” SM/01/60, Supplement 2, February 2001.


The commitments made in Thailand, Indonesia, and elsewhere had already stretched resources thin. As of May 31, 1997, the IMF had approved financing arrangements totaling slightly less than SDR 18 billion (or US$25 billion); six months later, before the Korea package, this figure had risen to almost SDR 28 billion (or US$38 billion).


These percentages were provided to the evaluation team by staff members and do not appear in the program documents. It is thus difficult to cross-check them and to determine which of the various possible aggregates the numbers refer to. Nevertheless, it is undoubtedly the case that a substantial change in the assumed rollover path took place at the last moment.


For comparison, Korea’s average monthly imports in 1997 were US$12 billion.


The written record on the evolution of this idea is thin. Most of the information in this and the following paragraphs is from interviews with IMF staff and former U.S. and Korean officials, as well as Kim and Byeon (2002). See also Blustein (2001).


Two approaches were on the table. One involved a new bond issue that would simultaneously finance the maturity extension and raise new money. The second proposal, which would ultimately be adopted, called for a sequential approach, with the extension of the maturity of existing bank debts under a government guarantee to be followed by a new sovereign bond issue when market conditions were more favorable. When asked by the Korean authorities for their opinion, IMF management declined to favor one approach or the other, and urged the authorities not to reject any reasonable proposal for the time being. Kim and Byeon (2002) recount the negotiating process.


The BOK’s facility providing advances of foreign exchange at a penalty rate, the other vehicle by which the banks’ huge overseas obligations were refinanced, was also repaid in full, and in fact made a profit of some W 6 trillion over the period from December 1997 to June 1998 (Baliño and Ubide, 1999, p. 42n).

Annex 3 Brazil


This annex provides detailed background for assessing the role of the IMF in anticipating and resolving Brazil’s capital account crisis of 1998–99. It first investigates the effectiveness of IMF surveillance in the precrisis period. The following section discusses program design issues, including (1) support for the crawling peg, (2) fiscal policy and debt sustainability, (3) monetary policy, (4) structural measures, (5) official financing and private sector involvement, and (6) program projections. It examines both the initial IMF-supported program that was agreed in the fall of 1998, and the program as revised in March 1999. It also touches upon the successor program agreed in August 2001, which was canceled in September 2002. The current program, beginning in September 2002, is outside the scope of our enquiry. The final section presents conclusions.

Precrisis Surveillance

This section assesses the role of the IMF in major areas of precrisis surveillance, including fiscal policy, exchange rate policy, banking sector issues, capital account developments and vulnerability indicators, and the impact of surveillance.


Many of the central issues of the precrisis period had their origins in the policies adopted in the aftermath of the Real Plan, the stabilization program launched in 1994 (Box A3.1). The IMF chose not to support the Real Plan with a program because, in its view, the proposed fiscal adjustment was insufficient to secure disinflation in a durable way.1 Instead, a monitoring arrangement, involving twice-yearly staff visits, was established, in part as a face-saving measure.2 Not agreeing on a program adversely affected the relationship between the IMF and the Brazilian authorities, and weakened the impact of IMF advice on Brazil’s policy formulation during the precrisis period.

The IMF’s skepticism about the ability of the Real Plan to reduce inflation appears ill founded in retrospect. The anti-inflationary gains were achieved and sustained over an extended period, albeit with a much greater fiscal deterioration than the IMF had feared. However, the IMF was correct in recognizing that weaknesses in the plan would pose challenges in consolidating these gains. As early as the first half of 1995, concerns emerged over the widening current account deficit, which prompted the authorities to tighten monetary policy in an attempt to contain the surge in consumption. Over a longer time horizon, there were questions about the eventual exit strategy from an appreciated real exchange rate, and the risk that the exit would reignite inflation.

Unlike East Asia, where the crisis took the IMF by surprise, the vulnerabilities of Brazil were well identified by surveillance, perhaps because they were mainly macroeconomic in nature. As early as September 1995, a briefing paper expressed concern that the external current account deficit did not seem sustainable and that its financing was highly vulnerable to shifts in market confidence. A prescient management comment on a briefing paper in October 1997 noted that “the current strategy [was] a risky one, and one thing far worse than a fiscal contraction a year before an election [was] a foreign exchange crisis a week before an election.” Staff reports for Article IV consultations were typically less frank, but carried a reminder that a relatively large current account deficit and heavy amortization commitments left Brazil vulnerable to shifts in investor sentiment. Even so, the IMF was generally more optimistic than the private sector. For example, in mid-1997, management instructed the imminent mission “to consider why some in the markets appear[ed] more skeptical about the Brazilian economy” than the IMF’s own analysis.”3

Brazil: The Real Plan

The Real Plan was a two-stage procedure of substituting the old currency, first by a unit of real value (URV) and second by a means of payment, the real, which was initially set equal to one U.S. dollar. The URV was a device designed to eliminate the backward-looking indexation by virtue of the fact that the URV itself was a price index. It was only after all contracts had been converted into multiples of the URV that the new unit of account was issued. All the steps were announced to the public, with no surprises or shocks (Franco, 2000; Bacha, 2001). Unlike some of the previous stabilization plans, no price or wage freeze was attempted; thus the Real Plan generated wide popular support. On March 1, 1994, the URV was introduced and, after four months of contract conversion, the real was issued by the Central Bank of Brazil (BCB) on July 1, 1994, with 30-days advance notice.

The IMF, however, was reluctant to support the Real Plan (and the 1994 Brady debt restructuring program) with a financing arrangement. According to internal documents, there was skepticism in the IMF that the Real Plan would succeed in reducing inflation in a durable way. The IMF did not believe that the proposed fiscal stance was sufficient to produce the envisaged reduction in inflation in 1994, and doubted its sustainability after 1994. The response of government revenues to lower inflation was highly uncertain, and there were doubts over planned structural fiscal reforms, which depended in part on approval by Congress in a constitutional review.

Following its introduction in July 1994, the real was allowed to appreciate by about 15 percent in nominal terms. Inflation fell from a monthly rate of over 40 percent in the first half of 1994 to between 1 percent and 3 percent a month by the end of the year, but was still greater than in Brazil’s major trading partners. According to contemporary IMF staff estimates, the real effective exchange rate appreciated 33 percent in terms of the general price index between June 1994 and February 1995.

Public warnings of these vulnerabilities were rare, although the 1997 World Economic Outlook noted, in a likely veiled reference to Brazil, that countries with insufficient fiscal consolidation, and therefore with “excessive reliance on short-term interest rates to restrain domestic demand,” might be “more vulnerable to changes in market sentiment.” In June 1998, the International Capital Markets report noted a risk that “the re-evaluation of emerging market vulnerabilities [had] not run its course” and that the terms and conditions of external financing could worsen further, leading to a broadening of the crisis to emerging markets outside Asia. Nevertheless, the staff appraisal in the capital markets report implicitly downplayed the risks to Brazil, noting that “many Latin American economies” had strengthened their financial systems, permitting the use of aggressive and credible interest rate defenses against contagion from Asia.

Fiscal policy

Brazil’s fiscal position weakened substantially in 1995, owing in part to large increases in public sector wages, public sector price freezes, and the loss of control mechanisms that had previously relied on high inflation to erode the real value of budgeted expenditures (Table A3.1). The staff consistently called for efforts to strengthen the fiscal stance, primarily in order to reduce the burden on monetary policy and permit a decline in interest rates and, as a consequence, some real depreciation of the currency. Given the high overall tax burden, staff consistently took the position that fiscal adjustment should be carried out mainly through expenditure restraint.

Table A3.1.

Brazil: Fiscal Developments

(In percent of GDP)

article image
Source: Data provided by the Central Bank of Brazil.

The coverage of the consolidated public sector in Brazil is very comprehensive.

The 1994 PSBR is high because of the very high nominal interest payments in the first half of the year, reflecting an inflation rate of over 40 percent a month.

The operational fiscal balance is defined as the primary balance less the “real” component of interest payments.

Comprises central administration, the Central Bank, and the social security system.

From 1996, concerns about public debt sustainability were also cited as reasons for a tighter fiscal stance. Staff projections in successive reports nevertheless consistently showed public debt on a downward path from progressively higher bases, implying that the debt at each stage was “sustainable,” if an adequate primary surplus could be achieved in the future. For example, projections in the staff report for the 1995 Article IV consultation showed net debt declining to 15 percent of GDP by 2000. For 1996, these projections assumed a primary surplus amounting to 3.3 percent of GDP and a reduction in real interest payments equivalent to 2 percent of GDP, whereas the actual outcome was a primary deficit of 0.1 percent of GDP, and real interest payments lower by just 1 percent of GDP.

The persistently weak fiscal position and high real interest rates led instead to a rapid expansion in the ratio of public debt to GDP, despite the start of a far-reaching program of privatizations and sales of other assets. The stock of net public debt rose to 33 percent of GDP at the end of 1996, from 30 percent in 1995, as neither of the assumptions (on the primary surplus and real interest payments) was fulfilled. The projections in the staff report for the 1996 Article IV consultation were not so optimistic, but they still showed the ratio declining to 28.3 percent of GDP by 2001, on the assumption of a medium-term primary surplus of 2 percent of GDP and substantially lower real interest rates.4 The report did not directly analyze why earlier projections had not been realized.

The authorities typically accepted in principle the IMF’s advice that fiscal adjustment was necessary but they were generally less ambitious in their efforts than the IMF recommended. Even the modest fiscal adjustment targeted by the authorities was rarely achieved and little progress was made in practice on fiscal consolidation between 1995 and 1998, with the fiscal accounts at best in primary balance. The authorities faced strong constitutional and institutional constraints in implementing such a consolidation, in part because of heavy earmarking of tax revenues and political pressures, including competing priorities for the congressional agenda.5

From time to time, the IMF identified specific policy measures to achieve adjustment, or to bring the fiscal balance back on track.6 However, instead of addressing immediate fiscal adjustment, the authorities accorded a higher priority to overcoming fiscal constraints in the medium term by establishing mechanisms to increase the flexibility of public expenditure, exercise control over state and local finances, and reform the pension and social security systems. The authorities were reluctant to seek congressional approval of revenue measures when constitutional reforms were under consideration by Congress. This does not mean that the Brazilian authorities never took tough fiscal measures. For example, faced with a major international turbulence in November 1997, they announced a package of fiscal measures (“the Package of 51”), estimated to yield over 2.5 percentage points of GDP.7 Its implementation, however, faltered in the face of electoral pressures in 1998.

The IMF was generally realistic about the political constraints, including risks to implementing agreed measures. In internal papers, for example, staff judged that the November 1997 package would deliver the projected savings if implemented in full, but correctly pointed out the risk that spending pressures would build during the election year 1998 “particularly if external constraints ease”; the limitations faced by the federal government in controlling the states; and possible slippage in securing congressional approval for fiscal measures. The Article IV consultation report in January 1998 presented the implementation risks less starkly than did internal documents, although it did note that steady implementation of the fiscal package would be essential to reduce Brazil’s vulnerability.

While progress was slow, enough groundwork on fiscal reform appears to have been done by the Brazilian authorities to facilitate fiscal adjustment under the 1999 program. Importantly, this included measures to control fiscal relations between the federal government and states and municipalities, which were linked to debt restructuring agreements. In this context, several Brazilian officials interviewed noted the useful contribution of IMF technical assistance in this area, including public debt policy and management. The experience of 1999 shows that it was indeed possible to tighten fiscal policy, given sufficient political will.

Exchange rate policy

The case of Brazil posed a number of challenges to the IMF’s approach to exchange rate policy. The Articles of Agreement have been interpreted as mandating that the IMF should take the exchange rate regime preferred by the authorities as given and try to ensure that other macroeconomic policies are consistent with it. Surveillance guidelines, however, state that Article IV consultation discussions and reports should include an accurate description of a country’s exchange rate regime, a candid appraisal of its appropriateness and consistency with underlying policies, as well as a forthright assessment of the exchange rate level.

Throughout the precrisis period, the IMF remained concerned about substantial real exchange rate appreciation and its adverse impact on Brazil’s external competitiveness, especially given the country’s poor export performance. However, implicit in its policy advice was the judgment that a gradual real depreciation could resolve the overvaluations, as long as this was supported by fiscal adjustment. Earlier in 1995, particularly in the aftermath of the Mexican crisis, there was greater skepticism—and much internal debate—as to whether the exchange rate system could be sustained (Box A3.2). Even then, internal papers reveal that the IMF favored a gradual exchange rate adjustment, combined with a major tightening of both fiscal and credit policy, rather than a step devaluation to counter current account problems.

Typically, the IMF’s policy advice was to accelerate the rate of depreciation within the de facto crawling peg system. Staff feared that floating the currency carried a substantial risk of overshooting. As time went on, concerns about overvaluation were downplayed, and the staff increasingly accepted the authorities’ arguments minimizing the size of any overvaluation, particularly in view of buoyant capital inflows that more than financed the current account deficit. Although at times the authorities indicated that they were open to accelerating the rate of crawl, they generally took the position that this would not noticeably benefit the current account and risked destabilizing market expectations and confidence.

The IMF was prepared to advise more drastic action in extremis, including a step devaluation or floating the currency. For example, at the time of the Asian crisis in mid-November 1997, IMF staff proposed that if there was a strong attack on the real, the exchange rate regime should not be defended. Management, however, advised against recommending a free float unless the band became totally untenable. The authorities reiterated their opposition to a discrete devaluation or a float, because of likely overshooting. In the event, Brazil weathered strong market pressure by raising interest rates sharply, announcing fiscal measures, and intervening heavily in the foreign exchange market.

Executive Directors generally supported the staff’s advice for a gradual acceleration of the crawl, though some believed even such an acceleration would be unnecessary or inadvisable. But there were exceptions. For example, in the discussion of the Article IV consultation in March 1997, one Executive Director argued that consideration should be given to allowing the exchange rate to float, so as to avoid an exchange rate crisis if investor confidence were to weaken. Most Directors, however, were of the view that, in the prevailing unsettled market conditions, any significant change in policy could lead to a loss of confidence. Some Directors encouraged the authorities to introduce greater flexibility into exchange rate policy, once market conditions had stabilized. At a meeting of the Executive Board on the Article IV consultation report in February 1998, staff orally disclosed that they had discussed a number of options with the authorities, including a discrete currency devaluation, more flexible exchange rate management, an acceleration of the rate of crawl, and the possibility of using a currency basket as a reference currency. One Executive Director, however, expressed displeasure over the absence of a clear discussion of exchange rate options in the papers prepared for the Board by staff.

While it was generally agreed that the currency was overvalued, there was considerable disagreement about the extent of the overvaluation (Figure A3.1). The IMF staff initially noted a strong real appreciation when the Real Plan was launched. In February 1995, the staff put the real appreciation since June 1994 at 33 percent in terms of the general price index. By late 1996, however, there was a tendency to downplay these figures, possibly in response to the views of the Brazilian authorities who used a wide range of arguments to suggest that any overvaluation was at most moderate (see Franco, 2000).8 The staff argued that the currency had been undervalued at the outset of the Real Plan and the subsequent significant real appreciation had been offset to some extent by productivity increases which were not fully reflected in prices. It also indicated that a pickup in export growth during 1997 weakened their arguments in favor of accelerating the rate of depreciation, although export volume growth of 10 percent in 1997 in fact was no greater than the increase in world trade volume.9

Figure A3.1.
Figure A3.1.

Brazil: Real Effective Exchange Rate1

(June 1994 = 100)

Sources: Central Bank of Brazil; and IEO staff calculations.1 Based on INPC consumer price index.

Brazil: The Evolution of Exchange Rate Policy

In the period immediately preceding the introduction of the real, the exchange rate was allowed to depreciate in line with contemporaneous inflation.

The real was introduced on July 1, 1994, with the unit of real value (URV) converted into the real at parity with the U.S. dollar. An exchange rate band of R$0.93 to R$1 = US$1 was initially established, but the exchange rate was ultimately allowed to appreciate in nominal terms to R$0.83 to US$1 in late 1994, with the floor of parity with the U.S. dollar maintained. The exchange rate was maintained at about R$0.84 per U.S. dollar until March 1995.

Exchange rate policy was altered on March 6, 1995 to a system of more depreciated adjustable bands, with a preannounced substantial further widening of the band from May 1995. This followed a rapid loss of reserves, reflecting a sharply widening current account deficit and weaker capital inflows following the Mexican crisis. However, in part owing to a lack of clarity about how the arrangement would operate and renewed pressures on the exchange rate, the policy was altered again on March 8, 1995, only two days later, with the adoption of a band of R$0.88 to R$0.93 per U.S. dollar. An inner band was established within this framework, which began to be operated as a de facto crawling peg, depreciating at an (unannounced) rate of about 0.6 percent a month. The outer band was periodically adjusted (approximately annually) to accommodate this rate of crawl, but had little operational significance.

In April 1998, the authorities introduced a marginal change in exchange rate policy by announcing a progressive widening of the inner band, initially just 0.4 percent wide, by 0.1 percent a month for the following three years.

On January 13, 1999, the inner band was abolished and the outer band became the operational band. It was initially established at R$1.20 to R$1.32 per U.S. dollar. It was announced that the band would evolve under a complex “endogenous diagonal band system” under which the rate of depreciation of the upper limit of the band would be faster when the actual rate was close to the lower limit of the band and vice versa. As the exchange rate fell immediately to the more depreciated boundary of the band, this involved a depreciation of about 9 percent.

On January 15, 1999, the exchange rate was allowed to float freely. This decision was confirmed on January 18, 1999, after consultations in Washington with IMF management. The program left some scope for Central Bank intervention in foreign exchange markets, although it was understood that this would not involve defending any specific exchange rate.

The net result was that, in late 1998, staff believed that the exchange rate was overvalued by 15–20 percent. The behavior of the exchange rate, after it was allowed to float in 1999, as well as comparison with outside assessments, suggests that staff likely underestimated the degree of overvaluation. Recent Central Bank estimates indicates that, measured by relative consumer prices, the real effective exchange rate had appreciated by 45 percent between June 1994 and December 1997, although this had fallen to 33 percent by December 1998. Contemporary analyses by the World Bank estimated the real to be overvalued by about 30 percent. The real effective exchange rate depreciated by some 35 percent in 1999 after the currency was floated. The exchange rate has fluctuated since then—and no doubt overshot at times—but peak levels of the real effective exchange rate have remained some 27 percent below the level of December 1998.

In retrospect, the IMF should have encouraged an earlier exit from the crawling peg regime at an opportune moment. This would have been consistent with the messages emerging from the IMF’s own crosscountry policy analysis of exit strategies from exchange rate–based stabilizations.10 Indeed, there were windows of opportunity to exit from a position of strength in late 1996 or early 1997, and again in the first half of 1998. The limited pass-through to inflation of the eventual float in 1999 suggests that, if well handled, carefully timed and supported by appropriate policies, floating the currency would have been possible without reigniting rapid inflation. Of course, at that time it was not clear that such a step would not lead to high inflation, although by then price stability had been established for some time.11 The immediate output impact of a float would likely have been greater than occurred in 1999 because the private sector was less hedged at that time. By the same token, the adverse impact on public debt would have been correspondingly smaller.

Banking sector issues

In the post–Real Plan period, some private sector institutions encountered difficulties and three major banks failed as they lost income from the “float” after inflation fell. The Brazilian authorities established two restructuring programs, which incorporated incentives to encourage the acquisition of weak private banks and privatization of weak state banks, resulting in a consolidation in the banking system. With strong encouragement from management, staff closely monitored these banking sector issues. The staff report for the 1996 Article IV consultation noted that the risk of a systemic problem had been effectively reduced through improvements in supervision, recapitalization, mergers, and the entry of foreign banks. The banking system, however, remained vulnerable to macroeconomic shocks and staff noted the desirability of further strengthening bank supervision.

By the time of the crisis, the IMF had analyzed in detail the risks to the financial system and rightly concluded