NBER Program on Exchange Rate Crises in Emerging Markets

John McHale
July 18, 2001

Project Directors: Martin Feldstein and Jeffrey Frankel
Program Organizers: Jeffrey Frankel and Dani Rodrik


The latest meeting in the NBER's series of country sessions on exchange rate crises in emerging markets took place in Cambridge on July 18th, 2001. Although the meeting had been planned since early in the year, the proceedings took on extra drama from the severe market turbulence still buffeting the Turkish economy at the time of the gathering. Many of the issues discussed were familiar from earlier country meetings, including the risk posed by exchange rate pegs, weaknesses in the banking system, and the dangers of explosive debt dynamics. Some of the issues at the center of the Turkish experience were more novel, however, including the fact that Turkey had earlier adopted an unusual sort of crawling peg exchange rate regime as part of an IMF inflation stabilization program.

As usual, the meeting drew a distinguished group of former high ranking government officials (Turkish and U.S.), officials from the international organizations, financial market participants, and academics working on international finance. The proceedings were divided into four sessions--the historical background to the 1999 stabilization program, the stabilization program and its controversial exchange rate regime, banking problems and the crisis of November 2000, and the crisis of February 2001 and outlook. Each session started with brief comments from four or five distinguished panelists, followed by a free-ranging general discussion. The following provides a summary of both the comments and the discussion.

Session 1 Background: Up to 1999

Chairman: Jeff Frankel, Harvard University and NBER

Rusdu Saracoglu, former governor of Turkey's central bank
Ted Truman, Institute for International Economics
Erik Nielsen, Goldman Sachs
Rudi Dornbusch, MIT and NBER

Rusdu Saracoglu began his presentation with slides showing strong co-movements between capital flows and growth (positive), credit to the private sector and growth (positive), and the current account balance and growth (negative). Outlining the forces driving the boom-bust cycles that have afflicted Turkey, he stressed the importance of the constraint international capital flows place on the economy, with capital flows affecting the availability of private sector credit and thus economic activity. When activity improves, financial markets think things are going well, reducing the financing constraint still further. Gradually, however, the bad news builds up as the current account deteriorates. The markets conclude that too much credit has been allocated to Turkey, and the constraint begins to tighten again.

Turning to fiscal policy, Saracoglu noted that budget deficits have been getting larger over time with an accommodating monetary policy. He said the cultural and institutional environment is conducive to monetary financing, with the result that inflation has been in the 50 to 60 percent range for most of the period since the mid 1980s. Interest payments are an important component of the deficit problem, but he cautioned analysts to be careful with interest burden calculations as real rates were negative over part of the period. With inflation so high the move to an exchange rate peg was central to lowering inflation expectations, he said. The government, however, faced the dilemma that banks could be driven to bankruptcy as a result of defending the exchange rate with a tight monetary policy. The weakness of the program was that when the "crunch came," the banking system was bailed out.

Ted Truman said that Turkey had been an accident waiting to happen. He noted that there had been increasing concern with the pattern of capital inflows that Saracoglu described. What was increasingly apparent was the need for a political consensus to break the inflationary cycle. In the 1990s, the high inflation environment led to very low growth. The resulting dissatisfaction with economic performance together with the desire for EU accession pushed the government to put a program in place. Truman raised the issue of the role of the IMF when there is not an external financing problem. In 1999 Turkey had an inflation problem not an external financing problem, and the IMF supported a program based on rapid disinflation.

Turning to the link between the public finances and inflation, Truman stressed the fiscal dominance of monetary policy, noting that the central bank had little choice. The result was high real interest rates. The overall public sector deficit in 1999--including the deficits of the banks and states enterprises--was an astonishing 23 percent of GDP.

Two tests were applied in judging whether the program would work. Were the authorities serious enough? And was the program comprehensive enough? Both tests failed with respect to domestic banking reform.

Continuing on the theme of the dominance of fiscal factors over the overall policy mix he observed that the Turkish financial system was opened up in the late 1980s because of difficulties financing the fiscal deficit--in effect, there was fiscal dominance of capital account liberalization.

Truman closed by asking if the underlying problem was political or economic. The authorities agreed on only one thing--the need to bring inflation down. The problem was that there wasn't much else in the way of tough policies reinforcing this desire.

Erik Nielsen sought to explain why there was such a skeptical view of the 1999 program. The key issue was overly deep involvement of the government, leading to large deficits and weak institutions. The deficits were largely financed through the banking system, which in turn borrowed abroad. High inflation also caused volatile real interest rates. The stage was thus set for a crisis.

The Fund came to Turkey advising the adoption of a currency board. But the authorities were not enthusiastic, largely because of problems in the banking system. The IMF went along with a "second best" program, but there were five main weaknesses.

· The exchange rate was overvalued in the crawling peg regime.

· The costs of bailing out the banking system were underestimated.

· Too little money from the 1999 IMF program was provided for financing the budget deficit.

· No monetary anchor outside the exchange rate anchor.

· No agreement with the labor unions on an incomes policy.

So, overall, it was a fragile program with little room for external shocks. To make matters worse, there was little political ownership, with heavy criticism from across the political spectrum.

Rudi Dornbusch said that the members of the OECD club that shouldn't be there are the ones to have had a crisis. In looking for an impending crisis he advised: check for trouble spots, "look to see if there is dynamite around," and consider what might provide the spark. Turning to the macroeconomic indicators for Turkey before the crisis, he noted that growth was strong until 1997, inflation was high (but, like Brazil, people had learned to live with it), and there wasn't a serious current account problem. More negatively, the budget deficit was large (although the primary balance was not a big problem), the debt ratio was high, and there was a large external debt. On the real exchange rate, it was not obvious that there was overvaluation. On the financial system, Dornbusch said that Turkey is a very corrupt country, adding that problems in the financial system are very hard to find out about. So trouble spots?--yes--though not an imminent problem.

What about the dynamite? In December 1999 short-term debt was greater than reserves.

Lastly, on the spark, he said that in emerging markets if you don't like one asset you leave--you don't move to another asset. All that is needed to induce exit is some event.

General Discussion

Dani Rodrik asked for the panelists' views on the sources of the high inflation. He said that it is hard to tell a fiscal story after 1994, although up until the early 1990s there was a close link between deficit financing and inflation. After 1994 there was a much healthier fiscal balance and little monetary financing. So there is a problem with the easy association of deficits and inflation. This explanation ignores the important inertial element of inflation. The absence of an easy association also has implications for the proper design of the program--for example, the need to emphasize incomes policy.

Dornbusch responded that, as in Brazil, indexation is important. But somewhere in the system the "nominals" get produced by the government. He agreed, however, that inertia is important. Truman reiterated that the fiscal deficit was the seed of the problem. He added that high inflation was consistent with growth for some time, but not in the late 1990s. Saracoglu noted that monetary policy was not as accommodating as time went on, and real interest rates rose.

Michael Dooley asked if the theory was that the taking away of the inflation tax would discipline the fiscal authorities. He asked: "Has this ever worked?"

Carlo Cottarelli asked why the Turkish case was so difficult to handle. He made a number of suggestions: the problem of establishing the credibility of policy makers given no real changes in political leadership; indexation was essentially backward looking creating substantial inertia; the banking system was very weak; and the fact that there was no "white knight" foreign country that was "structurally inclined" to invest in Turkey. Turning to Rodrik's question he said he had no doubt there was a link between deficits and inflation. The fiscal problems were there in the 1990s including in the second half, he countered, adding that we should look at the primary balance of the public sector as a whole, not just at the primary balance of the central government. Lastly, on the exchange rate, he said that it did not look significantly overvalued at the end of 1999. The current account deficit had reached 5 percent of GDP in 2000, but much of this was due to the rise in the price of energy. This terms of trade shock, together with higher-than-targeted inflation in 2000, was at the root of the overevaluation problem that had emerged during 2000.

Jorge Braga de Macedo asked why resistance to reform continued, not just on the macro side, but also in relation to privatization and regulation. Rudy Dornbusch laid the blame on coalition governments. Rusdu Saracoglu said resistance to privatization lies deeply in the constitution. The politicians wanted to do the privatization themselves--not necessarily for corrupt reasons, but because it was important politically.

Addressing Ted Truman, Peter Garber said that in all the crises the geopolitical importance of the country seemed to drive the programs. He asked if the Treasury sign-off on the program was economically or geopolitically driven. He also asked about the role of the National Security Council. Truman said that though Turkey is a strategically important country, his impression is that strategic elements were not the overriding consideration. Saracoglu countered hat the 2000 program was very much dominated by US security concerns. After a visit of the Turkish government to Washington, the degree of support from the Bank and Fund, the attitude towards the EU, and relations with Greece, all changed. Michael Mussa agreed with Truman that the 1999 program was not driven by external factors. Turkey wanted to make a major effort to bring inflation down through an exchange rate based stabilization backed by fiscal adjustment. Such programs lead to overvaluation, he added. There was some discussion of a currency board, but this was not the preferred strategy. The goal was to move to a fully flexible exchange rate. The view inside the fund was to get out of the exchange rate regime as soon as possible, but the government did not want to change the regime.

On the real exchange rate question, Erik Nielsen said the 1999 program did lead
to real appreciation, and that even if the exchange rate had been fairly valued initially, overvaluation would have resulted. Gazi Ercel said that account must be taken of developments in unit labor costs in judging real appreciation. Rudi Dornbusch observed that every time an overvaluation develops, the central bank governor points to unit labor costs--you must look at prices, he said. Jeff Frankel asked what was wrong with unit labor costs, and Dornbusch responded that it was a question of data reliability.

Roberto Chang asked what role financial sector liberalization had played in the crisis. Ted Truman responded that most of the liberalization took place at the end of the 1980s, with the pressure to open up coming as a result of the need to finance budget deficits. Dani Rodrik said that financial liberalization gave Turkey an extra decade, allowing it to move from money financing to debt financing.

Shang-Jin Wei noted that the boom-bust cycles reminded him of the Chinese cycles. But there was obviously a difference in the political systems. In the Chinese case it was not necessary to "buy out the coalition," he said. On the stabilization issue, he said that it is wishful thinking that you can stabilize an exchange rate regime without fiscal adjustment.

Rusdu Saracoglu said that when Turkey liberalized the capital account no one thought of it in terms of financing the fiscal deficit. The main concern was that it would lead to excessive capital outflows.

Rudi Dornbusch said that gradual real appreciation was a problem, but it got preempted by banking sector problems. The exchange rate would have caused problems anyway, he said. Looking to the future, he added that inflation targeting with a flexible exchange rate is not a panacea, and he cautioned that there will be problems again without financial restructuring.

Session 2 The stabilization program and the exchange rate regime

Chairman: Dani Rodrik, Harvard University and NBER

Mahfi Egilmez, former undersecretary of Turkey's Treasury
Caroline Atkinson, former US Treasury
George Hoguet, State Street Global Advisors
Kasper Bartholdy, Creidt Suisse First Boston
Carlo Cottarellie, IMF

Mahfi Egilmez began with a list of the pillars of the exchange rate based stabilization program: fiscal discipline, a pre-announced crawling peg, and implementation of structural reforms. He then described how as of September/Octorber of 2000 the current account was rising and was forecast to hit $10 billion by the end of 2000. However, the government announced that it would be $5 billion, leading financial markets to believe the government did not understand what was driving the current account balance. The government also delayed key structural reforms in the banking sector despite clear weaknesses. The result was that banks lost faith in each other, hampering inter bank lending. With the current account deficit rising, foreign investors sought to withdraw their funds and foreign banks cut credit lines. But the situation was not a typical foreign exchange crisis in which the public loses confidence in the domestic currency. In this case the public was willing to hold lira. The crisis was misdiagnosed in November 2000, however, making it almost inevitable that a second crisis would occur.

Caroline Atkinson recalled that Turkey had a long history of boom-bust cycles and foreign exchange crises. In this case, the crisis did not begin with a foreign exchange crisis but with an IMF program designed to address an inflation, rather than a balance of payments, problem. The design of the program was innovative but risky. It centered around a pseudo currency board in which the exchange rate would crawl while in a true currency board, monetary policy would be dictated by foreign demand for Turkish lira. However, this led to the worst of both worlds. Interest rates went up rather than down as Turkey did not get the boost to credibility that comes with tying your hands, and the exchange rate became overvalued.

Atkinson then turned to the lessons learned. First, the ability to make such an exchange rate-based stabilization program credible is now in doubt. Second, having an explicit exit strategy--something that was supposed to be an innovative part of the Turkish program--was of little help. And third, banking crises have loomed large in all the crises of the 1990s, Turkey included.

Finally, Atkinson turned to issues of politics, reiterating Truman's contention from the first session that the program was not driven by external pressure. She also noted that more successful exchange rate-based stabilization programs have come after a long period of hyperinflation and thus there was a strong political will to get rid of inflation. It is not clear that such a will existed in Turkey.

George Hoguet observed that equity investors in emerging market funds typically want the funds to be fully invested in emerging markets, but to make decisions between countries. To help put in perspective how Turkey was seen by fund managers leading up to the crisis, he noted that the Turkish stock market had performed relatively well over the longer term, and that a hedged strategy would have underperformed an unhedged one. He noted that the Turkish market was quite volatile, creating opportunities as well as challenges for the active manager. Hoguet pointed to several structural weakness in the Turkish economy, including undercapitalized banks and low levels of foreign direct investment, which he ascribed to the absence of adequate arbitration law.

Turning to the lessons, Hoguet concluded that the crisis was largely one of political economy. The costs of delay in addressing structural problems were substantial. He highlighted three areas: 1) Fund program design, and the real appreciation of the Turkish Lira, 2) Delays in implementing some aspects of the IMF program and 3) A deterioration in the external environment in the last half of 2000 and early 2001. He also troubled by the fact that the IMF is now Turkey's largest external creditor. Finally, the Turkish case is a further demonstration of the large amount of resources required to clean up the banking system.

Kasper Bartholdy said that there was general satisfaction with what the central bank was doing between 1995 and 1998. There was a fiscal problem, but this was not the fault of the central bank. He does not believe that the main concern leading up to the crisis was the exchange rate or the current account. The crisis was mainly a banking crisis. Bartholdy added that the role of the state banks in putting pressure on the inter bank market was also underestimated. Problems in the banking sector got worse as high interest rates caused further deterioration in balance sheets.

Turning to monetary and exchange rate policy, he said the new framework for monetary policy is a black box to most observers. The base money target is really an indicative target, and there are doubts that the exchange rate is really floating.

Carlo Cottarelli disagreed that the overnight borrowing of the state banks was a major issue in November 2000, though it did become a problem later on. In evaluating the 2000 stabilization program he began by recounting the situation as he saw it in 1999. Output was falling; inflation was rising; nominal and real interest rates were unsustainable; the fiscal accounts were on an unsustainable path. If this was not a crisis, "it is not clear what is," he observed. At the root of the crisis was the unsustainable level of real interest rates. The latter in turn reflected not only the weakness of the primary fiscal position, but also high inflation and exchange rate uncertainty. So, in order to bring interest rates down it was necessary not only to strengthen the fiscal primary position but also to reduce exchange rate uncertainty. The crawling peg was introduced to manage expectations in order to bring interest rates down more rapidly. At this time the bipolar view was certainly not a novelty, and the risks of "soft pegs" were known. So why did the fund support such an approach? It was a solution that reflected the preferences of the authorities. It also reflected the view that, in an open economy, exchange rate uncertainty is a key impediment to the rapid decline in inflation and interest rates. Was there an alternative? One possibility to bring at least inflation, and possibly interest rates, down more rapidly was incomes policy, Cottarelli said. But there were constraints on what the authorities could deliver on this front.

General Discussion

Jeff Frankel suggested that what Turkey did fell into the category of an intermediate exchange rate regime. He asked if what happened in Turkey was evidence for the bipolar view of exchange rate regimes (the corners hypothesis). George Hoguet answered that it is. Caroline Atkinson also saw it is as evidence of the risks of intermediate regimes.

Frankel further asked if it was the case that the IMF had initially wanted a currency board but accepted the crawling peg as a second best. Gazi Ercel said the currency board was resisted partly because of Argentina's experience. Peter Garber said that the IMF is now concluding that currency boards are not sustainable regimes. They lead to volatile real interest rates, which can go high enough to destroy the banking sector and lead to an unsustainable fiscal position. Jorge Braga do Macedo said that no exchange rate regime is right for all times. He wondered if the bipolar view was something of a fad. It takes time to build credibility, he said, and suggested that the heart of the problem for Turkey was that it could not build that credibility. Michael Mussa reported that the IMF had not accepted the bipolar view. He said he would not exclude the middle ground, and suggested that some countries can operate currency boards quite well. Steve Radelet said it is possible to have a middle ground when there isn't full capital account liberalization and volatile capital flows. He also noted that there had not been much discussion about how the high inflation environment complicates your choices.

Frankel also asked whether Turkey's built-in exit strategy (a pre-scheduled widening of the band) made the sustainability of the intermediate regime worse.

Erik Nielsen observed that a planned exit strategy is typically seen as a positive feature, but admitted that he became increasingly worried about it. Gazi Ercel said that the combination of the crawling peg strategy and an exit strategy did help credibility. Caroline Atkinson said that the exit strategy did not help, but it probably did not hurt either, since at the time of the crisis the switchover date (July 2001) was still relatively far out in the future.

Refet Gurkaynak pointed out that exchange rate stabilizations are associated with booms, real appreciation, and falling real interest rates on foreign borrowing. He added that capital controls can be used to limit the exposure of domestic banks to exchange rate risk, and wondered why limits on capital inflows were not considered.

Carlo Cottarelli said that the exchange rate peg did create the risk of overheating and real exchange rate overshooting. He also said that exchange rate based stabilizations had never been tried in Turkey, so it was difficult to assess what the impact would be particularly on aggregate demand and inflation. The key was to use fiscal policy to limit the growth of aggregate demand, should the latter appear to be excessive as it indeed turned out to be. Sterilized intervention could, in principle, also have been used to keep interest rates high, but this was precluded by the currency board rules regulating monetary policy. Later he said that the November 2000 crisis was close to a plain vanilla current account crisis. While it was complicated and accelerated by the banking problems, it was essentially a fairly standard story of overheating, overvaluation and a deteriorating current account balance.

Turning to the issue of why interest rates were so high in the second half of the year, Gazi Ercel contended that interest rates could not come down without adversely affecting expectations relating to the exchange rate regime.

Mahfi Egilmez argued that the exchange rate regime was the right approach to reducing inflation. The real problem was that no one knew how serious the problems in the banking sector were; the banking sector was what was mishandled, not the exchange rate regime, he said.

Michael Mussa said that the view that the problems in the banking system proved the undoing of the program is essentially correct. These problems were critical in triggering the crisis. But he said that it is also important to understand that the banking sector problems interacted with the exchange rate regime and monetary policy. He disagreed with Cottarelli that credibility could have been obtained with the quasi-currency board feature. Finally, he said that an IMF program is a kind of battle plan. The key is how forcefully to respond to adverse developments. In the end, he said, the capacity to respond was not there, and the problems in the banking system led to a violation of the currency board rules.

Session 3 Banking problems, and the crisis of Fall 2000

Chairman: Michael Dooley, UC, Santa Cruz, and NBER

Selcuk Demiralp, former Undersecretary of Turkey's Treasury
Joyce Chang, Chase Bank
Warren Coats, IMF
Steve Kamin, Federal Reserve Board

Selcuk Demiralp outlined a number of factors that led to weaknesses in the Turkish banking sector.

· Macroeconomic instability, which led to declining maturities and increasing foreign exchange liabilities. The disinflation program also had a major impact on bank balance sheets. Lower cost funding was available from foreign sources and there was a sharp increase in maturity mismatch.

· Large public sector debts, which led to a decrease in the resources being allocated to the private sector.

· Structural problems, including inefficient management, an un-level playing field between state sector and private sector banks, and inadequate capital bases and control systems.

The outflow of foreign funds led to the erosion of the capital base and revealed the underlying financial fragilities.

Demiralp went on to note what he sees as the drivers of the November 2000 crisis, listing concerns about the health of the financial system, the growing current account deficit, and concerns about political stability. Given these factors, market participants questioned the sustainability of the program. Moreover, problems in some banks initiated losses of funding throughout the system.

Joyce Chang asked rhetorically what the key triggers were for the November crisis. Her answer: capital account weakness--notably the high dependence on short-term flows--and poor asset management in the banking system--notably the severe maturity mismatch. Chang saw the classic components of a banking crisis, with sharp increases in interest rates, concerns about criminal investigations, and efforts at selective intervention. She also noted some key structural features of the Turkish banking system from an international perspective, highlighting that the banking sector is relatively large as a share of GDP, foreign borrowing is relatively small, and the largest banks tend to dominate.

Looking forward, Chang observed that although deposits have recovered recently--especially lira deposits in response to high interest rates--the medium term outlook is for pressures towards riskier loans. Stress points going forward include remaining foreign exchange risk (though it has been reduced), the fact that high interest rates have eroded capital, ongoing liquidity risk, and asset quality risk. She expressed particular concern about the outlook for private sector banks. Turning finally to the fiscal outlook, she sees a vulnerable situation over the near term. There is a need for accelerated expenditure cuts given the dangerous debt dynamics (and a debt level close to 90 percent of GDP). She cautioned that government financing requirements will be large in 2001.

Warren Coats recounted how the IMF recognized both the importance of a strong banking system to the growth potential of the Turkish economy, and the weakness of the system in practice. A special problem in Turkey was that the banking sector was a major lender to the government, so that a substantial effort was made from the beginning to focus on the banking system. Coats highlighted two facts that contributed to the banking sector weakness: inappropriate and overly liberal licensing, and the close association of banks with industrial groups leading to a high level of insider lending.

Coats said that what Turkey set out to do--and indeed what it has accomplished--has been striking. The focus has been on disinflation, but Turkey was attempting a major restructuring of its entire economy. The accomplishments in the financial sector included a new banking law and supervisory authority in the summer of 1999, new limits on insider trading, and a substantial strengthening of prudential regulations.

Weaknesses remained, however, at the time of the crises - November 2000 and February 2001. The banking law amendments of mid 1999 still had the weakness of the Council of Ministers' involvement in several areas including licensing and delicensing banks. However, this was fixed by a further amendment in December 1999 as a prior action for the IMF supported program. Resolving the intervened banks was too slow, which was part of the problem in November 2000 and February 2001. At the time none of these banks had been closed, thereby distorting the banking market - good progress is finally being made in this area as well. Coats asked why there has been such a dismal performance in the area of bank resolution. One reason is that it took a year before the new supervisory agency was operating. Also previous experience with interventions in banks had shown it to be a messy affair, and there was confusion in the public mind between bailing out depositors and bailing out owners.

Steve Kamin stressed the importance of the sequence of events that led to a large increase in public liabilities--contingent and otherwise. He noted that the factors that brought down the December 1999 program were not the usual ones that bring down an exchange rate based stabilization program--overvaluation, overheating, a large current account deficit. The problem was the ballooning of actual and contingent liabilities of the public sector associated with the interlocking of the fiscal and financial sectors, including direct loans to the government and loans from viable to weak banks that were guaranteed by the government. This made the system vulnerable to shocks. Kamin expressed the view that these problems were not caused by the stabilization program, and the crisis could have happened without it. He agreed with Carlo Cottarelli that one of the key elements of the program was to prevent unstable debt dynamics. However, recapitalization has added to the debt dynamics problem by placing a large number of additional government bonds paying high market interest rates into the hands of the private sector. Asking where Turkey stands now, Kamin stressed the need for continued fiscal adjustment and the need to get interest rates down. He also said a debt swap would be helpful, but the numbers being discussed are relatively small in proportion to the government's total debt.

General Discussion

Michael Mussa said that the most recent Fund program does provide financing to the budget that can be used for the financial sector. He recalled, however, that the IMF is supposed to lend for balance of payments purposes not to finance the public sector. Ted Truman said there was no precedent for World Bank money being used for financial sector recapitalization. So why did the Turkish authorities ask for it, Truman asked. The Turkish political system could not generate the money for banking system bailouts. But he noted that the money for such bailouts eventually comes from the domestic fiscal authorities in most counties.

Linda Goldberg asked if there had been any discussion about bringing in foreign banks. Warren Coats answered that there are no restrictions on foreign banks operating in Turkey, and the number of them is still growing. Joyce Chang added that uncertainty about the environment makes Turkey unattractive for foreign banks.

Rusdu Saracoglu said there was difficulty in implementing the provisions of the banking reform, including political infighting about who would lead the board. The appointees had very little experience. Also, although they had formal independence, they could be fired. In countries like Turkey, long tenures and independence are not part of the culture, he said. Turning to the question of the timing of the crisis, Saracoglu asked why the crisis occurred in the fall of 2000. He accepted that there were problems, but asked "why then?" He said that Turkish politicians made mistakes, but so did the IMF.

Michael Dooley observed that you don't set up a non-currency board framework and then take the central bank out of the role of lender of last resort. Michael Mussa responded that in the crisis the central bank violated the terms by providing liquidity to the system. He said there was another way to deal with the situation--provide a blanket guarantee to depositors--and close institutions that you believe are illiquid or insolvent. Carlo Cottarelli recalled that liquidity was provided during the crisis and in massive amounts. Thus, the currency board rules can hardly be blamed for the crisis as they were rapidly abandoned. The problem was that all the liquidity provided fueled foreign exchange purchases. This is because at the root of the problem as the increasing perception that the exchange rate framework could not be sustained because of the economic overheating. There was a need to respond to this overheating with fiscal tightening. But the government said the IMF's advice was wrong. When fiscal policy was finally tightened it was too late.

Refet Gurkaynak pointed out that Turkey already had deposit insurance and asked why the Treasury extended this to a guarantee on all bank liabilities. He also said, given the earlier statements that lower interest rates were desired by the program's designers and implementers, he was confused about what the IMF program was actually about--government finances or disinflation.

Sebastian Edwards commented that although it is true that when a crisis hits there is little time to respond, you can anticipate and prepare for it. Referring to difficulties in Argentina that seemed to be reaching a critical phase during the week of the conference, he wondered if Argentina had a "Plan B."

Ricardo Hausmann raised related two policy issues. First, he asked if blanket guarantees combined with recapitalization with governments bonds hinders rather than helps. He asked how does the issuance of more public debt when the pre-existing stock is seen as excessive restores confidence. Second, he asked if lengthening maturities is a good thing. If banks are the ones holding these securities and they fund them with shorter term deposits this could make things worse, he said. Warren Coats said that nobody liked blanket guarantees, but consider them essential so that the government would take over the insolvent banks without causing runs. On recapitalization, he suggested the term can mean a lot of different things. It should not mean dumping money into insolvent banks that will continue to have losses. What matters, he said, is that steps are taken to prevent new losses. Recapitalization was needed from banks taken over by the government to honor the pledge to depositors, allow them to operate within the regulations and to be resold. Steve Kamin agreed with Coats that recapitalization was necessary to restore credibility. But he asked if the method chosen was the least cost way of doing this.

Session 4 The crisis of February 2001, and where do we go from here?

Chairman: Martin Feldstein

Gazi Ercel, former head of Turkey's central bank
Steve Radelet, US Treasury
Michael Deppler, IMF
Peter Garber, Deutschebank

Gazi Ercel described what he called a "matrix of vulnerability" that included the exchange rate peg, the current account balance, weaknesses in the banking sector, high levels of short-term debt, capital flight, and political uncertainty. He noted that the December program, though short on capital flows, provided initial credibility gains and front-loaded actions that "gave more comfort." But he said that banking sector did not adapt itself for a disinflation program. Other, deeper weaknesses were an inability of the monetary policy rules to substitute for credibility, a relaxation of government determination after the first six months, the impact of the current account deficit on market sentiment, and an overly rapid decline in interest rates in the early stages of the program. He said confidence was also dented by a perception of overvaluation that built despite an increase in competitiveness, a continuous stream of "unwarranted criticism," misperceptions about the banking sector's foreign exchange position, non-compliance with the September stand-by review, and the fact that the "markets were open for all gossip."

On the causes of the February crisis, he listed the weakening of confidence in basic policy corrections, the growing vulnerability to currency attack, worsening maturity mismatches, and increased market concern about the risks of contagion. The trigger for the crisis, however, was the conflict between president and prime minister on February, 19.

Turning to the outlook for the Turkish economy, he pointed to a number of downside risks as well as to a number of reasons for confidence. On the downside, he noted the hard task of regaining confidence once lost, the uncertainty surrounding political developments, high domestic interest rates, the continuing inflation problem, and the challenge of managing a large domestic debt service. On the upside, he pointed to the low probability of balance of payments problems, progress in bank restructuring, low external debt pressures, and the achievement of reasonable growth.

Steve Radelet opened by agreeing with those who said that the Turkish crisis differed from the Asian variety. In Turkey problems started with the fiscal deficits, spilled into the banking sector, and then led to exchange rate problems. (Fiscal deficits did not play a leading role in the Asian crisis.) Radelet said that there were enormous risks in the IMF program from the beginning, notably the rapid increase in the short-term debt leading up to the November crisis. The shortening of maturities after the November crisis made the situation even worse, Radelet said, as "large amounts of money can leave very quickly."

In the six to eight weeks after the February crisis there was an effort to push through a number of fiscal, structural, and banking reforms (including getting the state banks out of the overnight market). One area where difficulties remain is monetary and exchange rate policy. There has been significant intervention to keep the exchange rate from moving and this has kept interest rates high. But he said that the Turkish government has "done a lot," and needs to hire a PR firm to get the word out.

Looking forward, Radelet said the markets are not yet convinced, and he worried that reform efforts had gotten off track. Political uncertainty remains a problem, and the debt dynamics continue to be extremely fragile. However, the majority of the debt is held by state institutions--including state banks--which gives some flexibility. There is thus a window to concentrate on the macro issues.

Michael Deppler said that the December program was beginning to work prior to the February crisis. Interest rates, for example, had declined from about 100 percent to 50 percent in mid-January. This was followed by a series of policy decisions that back-tracked on the spirit if not always the letter of the program. Together with increasing political controversy, interest rates started to rise once again, setting the stage for the explosion in outflows triggered by the Prime Minister's announcement of February 21. He stressed that the program had a chance of succeeding, but the crawling peg regime required a greater degree of policy and political discipline than was on offer.

Turning to the impact of the crisis, Deppler finds some aspects of the Turkish case hard to understand, however. Why did interest rates explode as they did? And why, after the initial correction in the exchange rate with the onset of floating, was the FX market so thin and trendless for some six weeks? Collusive behavior also appears to have been at play. Be this as it may, the interest rates were huge shocks to the banking system and to the debt situation. He said that the program had led to a sea change in the way banking is conducted in Turkey.

He said the program is compatible with debt sustainability. Admitedly, interest rates are strongly influenced by political uncertainty. However, the assumptions on real interest rates built-in to the program are conservative--17 to 20 percent in real terms over the period to end 2002. Moreover, this is within a program where Fund resources are providing about a third of the domestic debt coming due this year. The Fund lends to the central bank, but this money is on-lent to the Treasury. Thus the central bank is acting as an intermediary for the Treasury. One implication is that it is hard to distinguish the resulting foreign exchange sales from those being used to support the currency. This has led to some confusion about the degree of currency intervention, though the central bank is now being clearer about what the foreign exchange sales are for.

On monetary policy, he said the nominal anchor is meant to be based on inflation targeting. The government is reasonably on track for achieving two percent monthly inflation by the end of the year.

Summing up, Deppler said that the shift to a flexible exchange rate is a positive development. The remaining problem is debt sustainability. He cautioned that there is a "multiple equilibrium situation" with the economic fundamentals pushing to a good equilibrium, but with the dynamics dependent on the politics.

Peter Garber said that the February crisis was the tail end of the November crisis. His view now is that a "mortal wound" was incurred in November/December, and he's surprised that the "body did not fall over before then." As far back as summer 2000 Deutschebank's early warning model was suggesting a problem. They were asked to justify the model by the central bank, and were doing so as the crisis broke out. Garber noted that the strategists were generally upbeat before the November crisis. When it came, they believed it would end in devaluation. The new Fund program led to a fall in interest rates, however, and it was hoped it would thus lead to benign debt dynamics. The debt ratio could have been stabilized at 90 to 100 percent of GDP. Before the February devaluation Deutschebank was actually quite optimistic, and did not expect a "currency event." But on February 22nd the exchange rate was floated and it ended 40 percent lower. Now, a few months later, there is another program. It was assumed that interest rates would fall, but this has not happened, leading to "diverging debt dynamics." Something has to give, Garber concluded, "or there has to be very rapid inflation."

General Discussion

Refet Gurkaynak said that the number one requirement for successful inflation targeting is credibility, something that the Turkish central bank does not have and added that implementing inflation targeting in an environment of high and volatile inflation will be very difficult. Michael Deppler disputed the idea that you only use inflation targeting when things are going well. He agreed that it would be difficult to achieve in Turkey, but wondered what the alternative might be, noting that "nobody is enthusiastic about money" targeting, and that targeting the exchange rate is not a good idea.

Ratna Sahay asked why, given all the Turkish authorities have done, interest rates are not coming down. Erik Nielsen said that confidence is the key issue--investors have to become confident that the debt dynamics will work. He said that the IMF should have explained how it thought interest rates in the teens are sustainable. On the question of what should be done, he said sometimes there are no solutions, adding "you put yourself in a box when all depositors are guaranteed." Carlo Cottarelli said that at 6 percent of GDP, the primary fiscal surplus is large enough to service the stock of public debt even at high domestic real interest rates, taking into account the lower interest rate paid by Turkey on its external debt. Sebastian Edwards asked how you get 5 percent growth with 20 percent real interest rates.

Steve Kamin said that the jury is still out on whether Turkey is insolvent or illiquid. Comparing Turkey with Argentina, he noted the latter had done a lot to manage its debt, including a large-scale debt exchange, and thereby to improve its debt sustainability, at least in the near term. Turkey does not seem to have pursued this option, he said. Michael Dooley asked how a debt exchange helps if it is an insolvency issue. Michael Deppler said there are some opportunities for debt restructuring, but warned that the banking system--which holds much of the debt--is highly collusive. It tends to be very expensive when you try to restructure on a voluntary basis.

Ricardo Hausmann asked if the underlying problem was one of multiple equilibria. If yes, "how do you do away with the bad equilibrium?" he asked, pointing to the lack of instruments. Peter Garber noted that the accomplishments of the authorities are impressive, and wondered why the markets are not taking this into account. Responding to Deppler and Hausmann he said they characterized the situation as one of multiple equilibria where politics determined the outcome. But looking at the response to the accomplishments, he said it looks like a single equilibrium situation to him. Michael Deppler said the economics of the program does instill confidence, but this confidence is not carried forward because of problems on the political side. Steve Radelet observed that uncertainties about politics can lead to big changes in interest rates when the economics do not change at all. He also called for the authorities to look into other options (e.g., swaps) for debt management, and to clear up any uncertainties about exchange rate intervention.

Addressing Michael Deppler, Rusdu Saracoglu noted that the central bank was taken out of the money market with the 2000 program, and now is back in the money market but out of the foreign exchange market. He asked if this was a smart thing to do. Michael Deppler said that it is clear that exchange rate intervention is taking place, and noted any attempt to restrain the exchange rate feeds back on interest rates.