Journal Issue

Interview with Shinji Takagi: IEO urges more focus on “what ifs” in IMF handling of financial crises

International Monetary Fund. External Relations Dept.
Published Date:
August 2003
  • ShareShare
Show Summary Details

The resulting Independent Evaluation Office (IEO) initiated, in early 2002, a review of the IMF’s handling of capital account crises in Indonesia, Korea, and Brazil. Its report, released on July 28, chronicles vulnerabilities that were in some cases identified but were in others undetected or underappreciated; it also examines weaknesses in IMF-supported programs. Shinji Takagi, Advisor in the IEO and team leader for the capital account crises project, talks with the IMF Survey about honest efforts and costly missteps. He compliments the IMF for taking measures to enhance the effectiveness of surveillance in recent years but emphasizes that there is still more work to be done.

IMF Survey: The crises of the latter part of the 1990s were notable for their ferocity and the very visible and often highly criticized role of the IMF. That made them likely candidates for an IEO review, but why were only Korea, Indonesia, and Brazil chosen?

Takagi: We obviously talked about including Thailand and Russia. We had two overriding considerations in deciding which crises to cover: our own resource constraints and what we were likely to learn. Russia was obviously special—it’s too nuclear to fail and there won’t be another Russia. And Thailand was distinctive in that it was the first of these crises and not a product of contagion, and we knew that IMF surveillance had long identified its problems and discussed these vulnerabilities with the authorities. We felt there was less to learn there also.

IMF Survey: The three countries have remarkably different stories to tell. But how are they similar?

Takagi: They all experienced massive outflows of capital and had IMF-supported programs that involved large amounts of IMF resources. But their crises were distinctive, and the differences are instructive. Brazil’s vulnerabilities were largely macroeconomic, and the IMF’s management and staff were very aware of them. The initial IMF-supported program tried to support the existing crawling peg exchange rate, and that didn’t work. Korea had something very close to a pure liquidity crisis. The macroeconomic impact was severe, but short-lived. Once sufficient money was available, the crisis was resolved. And Indonesia, like Korea, had a balance of payments crisis and a banking crisis. But Indonesia’s twin crisis was compounded by a political crisis.

IMF Survey: IMF surveillance is meant to alert authorities to impending difficulties. Did it do so in these instances?

Takagi: In Brazil, IMF surveillance was very effective. In Korea and Indonesia, it wasn’t. In Asia, the IMF and most others saw sound economies that were growing fast. There was complacency, and the IMF missed the buildup of vulnerabilities in areas that were not its traditional focus of analysis. In Indonesia, the IMF identified weaknesses but thought them small compared with the bigger picture.

IMF Survey: When the IMF did spot vulnerabilities, was it able to convince the authorities to take action?

Takagi: Here, you first have to acknowledge that there can always be differences of view, between the IMF staff, the authorities, and the markets. This was the case with Brazil on the extent of overvaluation. The Brazilians were optimistic, the IMF staff was less so, and some in the markets were pessimistic. Our report argues that the IMF progressively downplayed the overvaluation of the exchange rate in Brazil. The staff argued that there were technical grounds for its view, but from an outsider’s perspective, it looks like the IMF downplayed the overvaluation issue because of the authorities’ view that the peg should be maintained. I can understand the staff’s point of view: the choice of an exchange rate regime is each country’s prerogative, a good relationship with the authorities is very important, and the staff can’t do its job without it. Exchange rate policy was a big political issue in Brazil, and it’s very difficult for the IMF to have real influence in political decision making.

IMF Survey: These crises all seemed to have a political component. Did the IMF have sufficient knowledge of, and sensitivity to, these political aspects?

Takagi: In Brazil, the staff spoke Portuguese and had worked on the country for quite a long time, but we know now that they still were not fully aware of the political dynamics. In Korea, the IMF’s management was aware of the political dimension and asked for a commitment to the IMF-supported program from the presidential candidates ahead of the elections in late 1997. In Indonesia, some on the staff, including the resident representative and some old-timers like Jack Boorman, knew a lot about the country, but the people who worked on the program didn’t have the same level of country experience. And, given the political dimensions of capital account crises and the speed with which they move, they didn’t have time to acquire those skills on the spot. A lesson for the IMF is that you want some continuity in staff assignments so that sufficient country knowledge is maintained.

Photo credits: Denio Zara, Padraic Hughes, and Michael Spilotro for the IMF, pages 217-18, 225, and 230-31; Str for Reuters, page 217; Kate Newman for Reuters, page 222; George Mulala for Reuters, page 223.

IMF Survey: In each of these crises, the IMF had a false start and then adjusted.

Takagi: It had no choice. In Korea, the United States took the lead. And only sovereign countries could do what it did to twist the arms of the creditor banks. Nevertheless, the IMF played an effective role as a facilitator, doing the necessary technical work. But Korea was unique: almost all of Korea’s external debt was held by banks. Brazil was different, too. It didn’t want to take any formal steps or do any arm twisting. Even now, it doesn’t want to be perceived as a country that would harm investors.

IMF Survey: In terms of crisis management, the most difficult case was Indonesia. What went wrong?

Takagi: The principal weakness in the first IMF-supported program was the lack of a comprehensive banking strategy. That was a mistake, in some sense, but why was it made? Two reasons. One, the staff honestly thought that the banking sector was sound, except for a handful of bad apples. And, second, the staff misjudged President Suharto. It closed three banks connected with his family. The president didn’t oppose the closure, but then what happened? His family publicly challenged the closures; his half-brother took the minister of finance to court; and his son acquired another bank, transferred all the assets from the closed bank to the new bank, and started business on the same premises. Everyone, of course, saw this as a reopening of the old bank. Public confidence plummeted. The central bank provided liquidity, which skyrocketed while the IMF was advocating a tight monetary policy. In reality, credit just exploded.

The IMF’s strategy was clearly not working, but it went on for two more months pushing the reluctant Indonesians to do what they had already proved incapable of doing. The appearance of persisting with a failed program undermined market confidence even more. And then noneconomic factors came into play.

IMF Survey: Your report acknowledges that the IMF learned from these crises and has taken corrective steps, but it also suggests there’s more to be done. What contribution could the first recommendation—to take a stress-testing approach to surveillance—make to crisis prevention?

Takagi: Look at Korea and Indonesia. The IMF recognized weaknesses in the banking sector but thought high growth would take care of these problems. Stress testing asks “What if … ?” What if growth slows down? You want to consider a range of possibilities and be better prepared in case you have to deal with them. IMF staff traditionally used a baseline scenario that presented the average view to the Executive Board. We make the case for itemizing risks rather than simply saying, “OK, most likely this will happen.” The IMF’s new surveillance guidelines already incorporate this concept of stress testing.

IMF Survey: You also urge the IMF to be more candid in its surveillance.

Takagi: This is really a controversial issue in the IMF. What do we mean by being candid? Well, we assume the staff already holds candid discussions with the authorities. But the staff sometimes fails to convey this candidness to the Executive Board. Brazil’s exchange rate policy, for example, was for a long time a taboo subject. The staff hardly discussed policy options in the papers it prepared for the Board.

Staff argue that if it is candid with the Board, authorities may not be candid with staff. Indeed, some of us here in the IEO sympathized with the view that the confidential advisor role of the IMF has to be respected. But it didn’t work in Brazil, and we have to ask how surveillance can be made more effective. When the authorities won’t listen, the only responsible options may be to take the issue to the Executive Board or to make the debate public.

Critics say this will alarm the markets, but there are different stages of country development and different degrees of transparency. In the United States, nobody says that the government knows more about the U.S. economy than people on Wall Street. Economic issues and policy options are openly debated. And the IMF’s views are an input into this public debate. Where transparency and openness are the rule, the IMF should follow suit. In a developing country, where the IMF may be the only source of information, well, the IMF’s role may be different.

Staff response

When the IMF’s Executive Board discussed the Independent Evaluation Office’s (IEO’s) report “The Role of the IMF in Recent Capital Account Crises,” it also took into account the response of IMF staff. Broadly, the staff welcomed the important lessons that the report drew from Indonesia, Korea, and Brazil. Staff also broadly supported the IEO’s recommendations, which go in the direction of strengthening the institution’s ability to help countries head off financial crises and deal with crises when they do occur. But the staff also cautioned that the report, in focusing on the early stages of the crises, did not tell the “whole story,” as it overlooked later successes in restoring confidence, stemming capital outflows, and putting in place needed structural reforms, as a basis for economic recovery.

The staff also saw the findings confirming its own experience that each crisis is indeed unique in the problems it poses. Anticipating crises, the staff observed, will always require difficult judgments amid great uncertainty, and the capacity to prevent crises will rely on the actions of member countries. In the specific crises that the IEO examined, the authorities and the IMF staff faced enormous analytical and practical challenges. Much of the economic trauma was unavoidable, as many of the underlying problems were, at least in the short run, beyond the countries’ and the IMF’s control.

Future crises are likely to be no less challenging, but the staff recognized that the organization benefited significantly from incorporating the lessons gained from these experiences. It would also continue efforts to strengthen the effectiveness of surveillance and of IMF-supported programs in light of these lessons.

IMF Survey: Aren’t these situations particularly tricky in emerging markets?

Takagi: Investment houses have billions of dollars invested in these countries and gather a lot of relevant information. It is arrogant of the IMF to think that what it says counts more and could therefore alarm the markets. With Korea, the IMF was telling the market, “we think Korea is sound; don’t pull out. But there is some information we can’t share with you.” You can’t deal with markets like that. You have to say, “this is the information we have; we think everything will be fine if you don’t pull out.” Our report is not suggesting that in all situations the IMF has to say everything it knows. But in a capital account crisis in an emerging market where the information is readily available and the IMF does not necessarily have an information advantage, transparency has to be the rule.

IMF Survey: The report has a number of recommendations on how the IMF designs adjustment programs. Chief among them is greater attention to balance-sheet interaction effects. What’s the concern here?

Takagi: Well, it’s not a new concern for the IMF. In a standard macroeconomic model, if you devalue the currency, it should be a good thing. Exports are cheaper and the economy is boosted. But if the country has debt denominated in foreign currency, devaluation can be damaging. So don’t just look at relative prices; look at the impact working through stocks, especially debt denominated in foreign currency. There has been new appreciation of this interaction since the Asian crisis, and the IMF is already paying more attention to it. There was a degree of learning from Asia that the IMF applied to Brazil. In that instance, the staff did try to figure out what the stock implications would be. But it’s difficult to do because you have to gather data on the private sector.

There has to be some flexibility in the original program, and the public has to understand the logic behind it. If something does go wrong, the public then knows how the program will respond.

IMF Survey: The report also argues for built-in flexibility in IMF-supported programs.

Takagi: An IMF program is usually based on an assumption that next year’s growth is going to be, say, 3 percent. Now, if this goes wrong, the whole program is destroyed and confidence is undermined. There has to be some flexibility in the original program, and the public has to understand the logic behind it. If something does go wrong, the public then knows how the program will respond.

The staff says that the IMF’s three- or six-month program reviews are meant to monitor and adjust, but the public and the markets don’t perceive this as flexibility. The markets see wimpiness. They see the IMF getting it wrong each time and having to make changes.

IMF Survey: You also encourage the IMF to stay focused in a crisis and avoid the temptation to try to fix problems outside its mandate.

Findings and recommendations


In Indonesia, the IMF identified banking sector vulnerabilities but underestimated their severity and the potential dangers for macroeconomic stability. The financial and economic crisis also quickly became intensely political, which greatly complicated crisis management. Policy reversals, which undermined market confidence, exacerbated the crisis, but the IMF’s response was inadequate in some respects.

In Korea, the IMF failed to pick up vulnerabilities in the economy before the onset of the crisis and then proceeded with a program that the market strongly suspected was underfinanced. When major industrial country governments, however, took concerted action to address what was, in effect, a liquidity crisis, the chief obstacle to the crisis’s resolution was removed. The report credits the IMF with playing a useful role as a crisis coordinator but also notes it underestimated the impact of negative balance-sheet effects and initially advised fiscal tightening that, as the IMF’s own evaluation conceded, was unnecessary.

In Brazil, IMF surveillance successfully identified the macro-economic vulnerabilities at the core of the crisis but progressively downplayed the scale of the overvaluation and was unable to persuade the authorities to take early corrective action. In hindsight, the IMF was unduly concerned with the systemic efforts of an early exit from Brazil’s crawling peg exchange regime. It did, however, play a useful role in facilitating the transition to an inflation-targeting regime and in helping Brazil develop a more disciplined fiscal policy regime.


IMF surveillance had varying success in identifying key vulnerabilities that led to these crises. Where it did identify weaknesses, it did not appreciate the full danger (Indonesia) or had little success in persuading the authorities to take timely corrective action (Brazil). It was most effective in anticipating macroeconomic vulnerabilities (Brazil) and had considerably less success in foreseeing financial and corporate sector weaknesses (Indonesia and Korea) that were outside its traditional areas of expertise.

Program design. In all three crises, program projections were sharply out of step with eventual outcomes (too optimistic in Indonesia and Korea; too pessimistic in Brazil), with the result that policy advice and actions were, with hindsight, inappropriate.

Transparent and adequate financing. Inadequate financing—most specifically unanswered questions about the conditions under which bilateral assistance would be provided—undercut the credibility of the first IMF-supported program in Korea.


Stress-test crisis vulnerabilities. IMF surveillance should build on and systematize the steps it has already taken to incorporate analysis of potential shocks. The report recommends itemizing major potential shocks—including balance-sheet effects—and discussing the authorities’ planned responses to such shocks. It also urges greater understanding of the political constraints that will likely shape policy decisions.

Make surveillance assessments more candid and more public. In addition to new surveillance guidelines that call for more systematic assessments of what happened under previous policy advice, the report urges the IMF to escalate signaling procedures when key vulnerabilities are repeatedly unaddressed and suggests it might be useful to seek an outside opinion when the IMF and the authorities are at odds on how best to address an issue. It also recommends publishing country-level analytical work to generate more informed public debate.

Revisit the design of IMF-supported programs to give greater attention to the interaction of balance-sheet weaknesses and key macroeconomic variables and to ensure that restoring confidence is the overriding objective of crisis management. The report suggests that there should be flexibility, particularly in capital account crises, to allow for adjustments if actual outcomes are at variance with expected ones. It asks the IMF to keep an eye on whether conditionality based on financial programming is appropriate for capital account crises and also stresses the importance of communicating the logic of the program to the public and the value of transparency. It also sees adequate financing as a key to restoring confidence. If the IMF does not itself supply adequate financing, it needs to ensure that the terms under which supplementary bilateral assistance would be made available are clear and credible.

Do more as crisis coordinator. The IMF should play a central role in identifying the circumstances in which concerted efforts can help overcome “collective action” constraints that can delay the resolution of crises. This should be based on a meaningful dialogue with the private sector and build on the new mechanisms for such a dialogue that have been developed in recent years.

Takagi: In Indonesia, the major industrial country members of the IMF thought that market confidence couldn’t be restored until Indonesia put an end to corruption and cronyism. They had a point, but the detailed structural conditionality that this view engendered didn’t work. Also, all this conditionality took on a life of its own and detracted from the more critical issues of banking and corporate debt. And in any case, people didn’t believe Suharto would reform, so that undermined confidence in another way.

IMF Survey: You also stress the role of communication in the design of a program.

Takagi: You need a communications strategy during a crisis. From the very beginning, there should have been an agreement between the IMF and the authorities that whenever program-related information was released, the IMF had to know about it beforehand.

IMF Survey: Finally, you see restoring confidence as a key ingredient in resolving these crises. What does the IMF need to do?

Takagi: In a capital account crisis, private capital is flowing out. Somehow, the IMF has to help stop that; it has to help the country build up market confidence. In Korea, a number of countries pledged additional money, but no one knew how this would be made available. It was just a number, and the market discounted it. We say that if the IMF is going to play a role in restoring confidence, it must either have large financing of its own or alternative financing on clear, credible terms subject to the same terms as IMF financing.

IMF Survey: Did the report address the oft-cited concern about moral hazard?

Takagi: We looked at whether the availability of large-scale financial support from the IMF created moral hazard, and we say it probably did not.

IMF Survey: You also suggest that confidence can be restored more quickly if the IMF strengthens its coordination role in crisis resolution.

Takagi: That, too, comes from the IMF’s experience in Korea. We argued that the IMF should have been more forceful in telling the major industrial countries that the financing was inadequate. We would like to see a more proactive IMF, a more proactive coordinator. If it had been so in Korea, it would have said, “this is an underfinanced program; it’s going to damage credibility. You either give us additional money or we aren’t going to do this.” If that had been the IMF’s position, I think the United States and other countries would have had to provide additional money or, as they ultimately did, twist the arms of the banks.

If the IMF is going to play a role in restoring confidence, it must either have large financing of its own or alternative financing on clear, credible terms subject to the same terms as IMF financing.—Shinji Takagi

The full text of the report, “IMF and Recent Capital Account Crises: Indonesia, Korea, Brazil,” related statements, the Board summing up, and the staff response are available on the IEO’s website:

Other Resources Citing This Publication