IV Reestablishing a Credible Nominal Anchor After a Financial Crisis

G. Kincaid, and Charles Collyns
Published Date:
April 2003
  • ShareShare
Show Summary Details
Andrew Berg, Sean Hagan, Christopher Jarvis, Bernhard Steinki, Mark Stone and Alessandro Zanello 

This section examines how to achieve monetary policy credibility and price stability after a financial crisis.1 The experience of recent cases suggests that monetary policy by itself cannot restore stability following a crisis, but must be complemented by measures to close financing gaps and to solve banking sector problems without resorting to massive liquidity support. With respect to the choice of post-crisis exchange rate regime, stabilization can be achieved quite quickly under a floating exchange rate regime once the prerequisites have been met, and particularly when monetary policy is tightened early and sustained. In such cases, countries have typically moved progressively toward some form of inflation targeting, while attempts to use monetary aggregate budgeting have been less successful. Hard pegs have also had some success in reestablishing stability even in situations where the economy is in deep disarray, but disinflation has tended to be less complete, and countries often face a difficult exit problem.

The question addressed in this section is how to achieve monetary policy credibility and price stability after a financial crisis. We consider currency crises in which monetary policy credibility has been lost, focusing on the most severe episodes associated with broader banking and financial crises. This section draws on the evidence from ten recent crises: Brazil (1999); Bulgaria (1997); Ecuador (2000); Indonesia (1997); Korea (1997); Malaysia (1997); Mexico (1994), Russia (1998); Thailand (1997); and Turkey (2001).2

Prerequisites for a Credible Nominal Anchor

The countries that experienced currency crises generally went through two phases: an initial chaotic period of crisis containment, and a longer period during which the policy framework and institutions were more fully developed. The beginning of the crisis is defined as the month before the first large movement of the exchange rate (Table 4.1).3

Table 4.1.Crisis Countries: Indicators of Recovery and Stabilization
t0 – month/quarter before crisis hitDec-98Dec-96Dec-99Jun-97Sep-97Jul-97Nov-94Jul-98Jun-97Jan-01
Fall in output (percent)123292191110111812
Number of quarters to recover half of output loss from pre-crisis level2435333243
Average inflation, t– 24 to t02416950653811553
Peak inflation, post-crisis (percent)1322,85521515030139844716121
Average inflation, t0 to t + 1293,235946376501571071
Average inflation, t + 12 to t + 24622227122819−1193
Months for inflation to fall below 10 percent411181768264244417
Months for inflation to fall below 5 percent141618177944417
Months until the nominal exchange rate stopped depreciating2217464579
Months until the real exchange rate stopped depreciating2116253767
Months until exchange rate volatility returned to normal657178146815127
Source: IMF staff estimates.

Fall in output from quarterly peak in the year preceding (or following) the crisis to the lowest quarterly output level following the crisis.

Here and elsewhere in this table, inflation is measured as a three-month geometric moving average of the annualized monthly change in the CPI.

The latest observation is t + 17, since the crisis was relatively recent.

Inflation never fell below either 10 or 5 percent. Figures shown indicate inflation at t + 24.

On a nominal basis, the depreciation never stopped in Russia.

Volatility is measured as the standard deviation of daily changes in the log nominal exchange rate. Normal volatility is defined as the range observed in a number of developed and noncrisis emerging market countries with floating exchange rate regimes.

Volatility never returned to normal.

With major increase in volatility again in October/November 1995.

Source: IMF staff estimates.

Fall in output from quarterly peak in the year preceding (or following) the crisis to the lowest quarterly output level following the crisis.

Here and elsewhere in this table, inflation is measured as a three-month geometric moving average of the annualized monthly change in the CPI.

The latest observation is t + 17, since the crisis was relatively recent.

Inflation never fell below either 10 or 5 percent. Figures shown indicate inflation at t + 24.

On a nominal basis, the depreciation never stopped in Russia.

Volatility is measured as the standard deviation of daily changes in the log nominal exchange rate. Normal volatility is defined as the range observed in a number of developed and noncrisis emerging market countries with floating exchange rate regimes.

Volatility never returned to normal.

With major increase in volatility again in October/November 1995.

The first phase ended roughly when the free fall of the nominal exchange rate was arrested and exchange rate volatility declined markedly—which generally took a few months. The second phase has no well-defined end; this section focuses on the policies pursued and results achieved during the first two years after the beginning of a crisis.

The achievement of initial stability typically required meeting two conditions—in addition to a sound monetary policy.

  • Eliminating an ex ante dollar shortage. Sometimes, a credible monetary policy can only be achieved if it is supported by the provision of external finance or action on the debt. In particular, where shortage of foreign exchange was the key trigger for the currency and banking crisis, the excess demand for foreign exchange typically has to be eliminated—through default/rescheduling (Russia), provision of external support (Mexico), or a combination of external support and rescheduling/rollover of debt (Korea)—in order to reach nominal stability. Monetary policy alone (through the usual high-interest-rates, higher-capital-inflows channel) has generally been incapable of eliminating the ex ante gap in the midst of a crisis. At the height of the crisis, a tension exists between setting domestic currency interest rates high enough to compensate for risks of further depreciation and default and keeping them low enough to avoid raising the probability of default to unacceptable levels, given their effects on balance sheets and real activity. A similar logic applies to interest rates on dollar obligations: higher interest rates will not attract investors in the context of a panicky “rush for the exits.”4 Finally, the normal mechanisms to eliminate foreign exchange shortages, demand compression, and currency depreciation act over time but also do not serve this purpose effectively in the first few months of a crisis. Nor, arguably, should they: in a capital account crisis, the challenge is often to prevent an excessive contraction in domestic demand or a massive overshooting of the exchange rate. Thus, a strong monetary policy is usually an essential complement to external support, but it cannot substitute for it completely.

  • Solving related problems in the banking sector without resorting to massive liquidity support. The currency crises we studied were generally accompanied, and sometimes caused, by banking crises. Central banks in this situation often faced the dilemma of trying to manage monetary policy while also dealing with liquidity problems in the banking sector. Typically, this problem has been resolved by the government explicitly accepting responsibility for recapitalizing the banking system (see Section VI). As a result, rescuing the banking sector has led to large increases in (measured) public debt levels during these crises, often by 15 percent of GDP or more. Perhaps surprisingly, even countries that already had high levels of public debt were able to absorb this increase without compromising the achievement of initial stability. In some cases, however, the high debt load resulting from the banking crisis has reemerged as a problem a few years later and limited countries’ ability to conduct monetary policy (Brazil, Turkey) because of concerns about the effects of high interest rates on fiscal sustainability. In cases in which the sovereign defaulted on its obligations during the crisis (Russia, Ecuador), the government had to rely on mechanisms other than government-led recapitalization to resolve the banking crisis. In these cases, the government essentially had to eliminate its fiscal deficit in order to achieve nominal stability, since it could no longer borrow from the commercial banks, the public, or the central bank (i.e., print money). Figure 4.1 illustrates the complementary role of strong policy packages and adequate U.S. dollar financing in two important cases. Even after strong policies were put in place, in early December 1997, the Korean won continued to fall. Only the combination of an adequate financing package, through the coordinated rollover at the end of that month of external interbank debt, plus a further increase in interest rates, was sufficient to stabilize the exchange rate. Similarly, in Mexico in the beginning of 1995 quite high interest rates and substantial IMF financial support did not arrest the exchange rate collapse. The exchange rate stabilized only in mid-March, when interest rates were increased and the first disbursement of bilateral support eased doubts about the financing package.

Figure 4.1.Putting Together the Package

Experience with Post-Crisis Exchange Rate Regimes

Currency crises are characterized either by the forced abandonment of a fixed exchange rate regime or by a sharp depreciation in a floating one, often accompanied by a substantial loss of reserves. Given the loss of credibility, countries with a relatively open capital account have only two choices for the exchange rate regime in the immediate aftermath of a crisis: some variant of a float, or a very hard peg. Attempts to retain a soft peg after a controlled devaluation in the face of a major speculative attack are nonetheless surprisingly common and have generally ended in failure (Mexico, Russia, Brazil). It is hard to assess the cost of this additional loss of credibility in the first few days of the crisis, but it cannot help.5

Most countries studied succeeded in achieving nominal stability with a floating exchange rate regime. Although large depreciations and high exchange rate volatility have characterized the immediate post-crisis period, nominal stability has generally been restored quickly after the float was adopted, particularly in countries where inflation was low before the crisis. In most cases that floated, the nominal and real exchange rates ceased to depreciate within a few months, and exchange rate volatility also fell sharply (see Table 4.1). Where initial inflation was low, the period of “freely falling” exchange rates was fairly short, ranging from two months in Brazil to seven in Thailand and Indonesia.6 A fraction of the initial depreciation (i.e., the overshooting) was also reversed rapidly, generally within a year of the crisis; typically, the reversal occurred through nominal appreciation rather than through higher inflation. The depreciation did not unleash inflationary explosions. Most countries that floated achieved single-digit inflation (measured as annualized monthly price changes) within four to eight months, and lowered inflation further to 5 percent within two years. Furthermore, in most cases average inflation was below its pre-crisis level two years after the crisis. (Table 4.1 and Figure 4.2).7

Figure 4.2.Floaters and Fixers1

Source: IMF staff calculations.

1 The floating-regime countries are Brazil, Korea, Malaysia, and Thailand. The fixed-regime countries are Bulgaria and Ecuador.

2 Average inflation was negative between t+16 and t+20, so it does not display on the logarithmic right scale.

The success of the 1990s crisis countries in reducing inflation contrasts favorably with the difficulty that countries hit by the debt crisis of the 1980s had in reducing inflation. However, the depth of the problem also differed. Whereas in the 1980s inflation was a chronic problem typically rooted in large fiscal imbalances, in most of the 1990s crisis countries inflation (and monetization of fiscal deficits) was not a problem before the crises.

The countries that opted for a float in the aftermath of their 1990s crisis modified their exchange rate regime over time, after some measure of stability had been restored. In some cases, the float became a de facto crawling peg (Russia) or an explicit peg (Malaysia). In many other cases, the exit was eventually to inflation targeting (Mexico, Brazil, Thailand, Korea). This choice is discussed further below, under “The Framework for Monetary Policy in a Floating Regime.”

Two countries in this sample stabilized with hard pegs in the aftermath of a crisis (Bulgaria, Ecuador). Both did so after suffering extreme collapses in the exchange rate and especially sharp and prolonged increases in inflation. Thus, prior to the peg these two countries were pursuing an unsuccessful floating exchange rate policy in the absence of adequately tight monetary policy or other preconditions for nominal stability (or both). In particular, they had intractable banking and fiscal problems that severely limited the effectiveness of monetary policy. The timing of the adoption of the hard peg varied in the two cases. In Bulgaria, the currency board was formally adopted after the situation had stabilized, although its anticipated introduction served to anchor expectations. This delay, which was due to the electoral cycle, allowed the institutions to be established and permitted inflation to greatly reduce the real value of bank deposits and hence the fiscal cost of the banking crisis (although at a cost of the steepest recession in the sample). Ecuador’s dollarization permitted stabilization with almost no prior preparation, although here too the high inflation prior to the adoption of dollarization eroded the value of bank deposits (and was associated with a steep recession as well).

Interest rates (nominal and real) fell rapidly after the hard pegs were adopted, although inflation remained at higher levels than in countries that floated (Table 4.1; Figure 4.2 also shows average levels of inflation, interest rates, and the exchange rate for the two hard peg cases). The incomplete disinflation was probably a consequence of the initial overshooting of the exchange rate. The fixing of the exchange rate at an overly depreciated level created pressures for a real appreciation, which could only be accommodated via higher inflation.8

Notwithstanding the faster decline in real interest rates, and the more rapid adoption of a firm nominal anchor for monetary policy, the output decline (and ensuing recovery) was similar in the hard pegs and the floats. This presumably reflects the various initial conditions and shocks that shaped both choices and outcomes, as well as the policy choices themselves. There is no evidence, however, that the higher real interest rates that the floating countries experienced for a period of time had an obvious and large output cost.

Hard pegs had both benefits and costs for the countries that adopted them. Adoption of the hard peg anchored expectations and therefore provided a context more conducive to the adoption of fiscal and banking reforms, although it did not in and of itself resolve the banking and fiscal problems. Hard pegs constrain future exchange rate choices, however, in that exit is costly. The long-run costs (and benefits) of this constraint depend on the usual considerations that have been widely analyzed in recent years. From the perspective of a post-crisis country, whether the benefits of establishing credibility early on through a hard peg are worth the potential long-run costs will depend on the appropriateness of a hard peg over the long run for the particular country and on how difficult it would otherwise be to restore credibility. With respect to the question of which sort of hard peg to choose, dollarization is a more natural choice than a currency board for those countries that are more confident of the long-term value of the hard peg, as well as for those in too much disarray to implement a currency board.

The Stance of Monetary Policy in a Post-Crisis Float

The countries that floated and were most successful at quickly ending the period of volatility were those that tightened monetary policy early and sharply and that did not ease monetary policy until stability had clearly been restored. This monetary policy response resulted in a period of very high real interbank interest rates and, later, exchange rate appreciation. However, the success most countries achieved in lowering inflation did not require prolonged periods of higher real interest rates. Typically after an initial spike, nominal interest rates returned to pre-crisis levels or below in only a few months.9

A key objective of monetary policy in the aftermath of a float was to contain the exchange rate depreciation. The main rationale was to limit the overshooting and, hence, the inflationary impact of the depreciation. On the whole, the cases reviewed provide support for this rationale, in that monetary policy did help to reverse overshooting and hence limit the subsequent inflation.10 Korea, for example, raised interest rates sharply only two months after floating and rapidly undid much of the exchange rate overshooting. Partly as a result, post-crisis inflation was low (Figure 4.3). It took Mexico somewhat longer to arrest the overshooting. This, combined with ongoing doubts about the resolution of the dollar liquidity problem and other elements of the policy framework, resulted in a much longer period of overshooting and thus higher inflation pass-through. Indonesia took much longer to isolate monetary policy from the banking system problem and tighten consistently and had a much more protracted period of instability (see Figure 4.3). Eventually, both countries tightened monetary policy as much or more than the others in order to stabilize, to judge by the level of nominal and ex post real interest rates.

Figure 4.3.Short-Term Dynamics of Monetary Policy

(In percent)

Source: IMF staff calculations.

The decision on how tight to set monetary policy and for how long—that is, how much to resist initial overshooting and ensuing inflationary pressures— depends on several factors. Exchange rate and price adjustments can play positive roles in adjusting to the banking crisis and associated disequilibria. Some degree of exchange rate adjustment was clearly necessary where overvaluation or excess absorption (or both) were part of the problem (Brazil, Mexico, Thailand). In some cases it was also useful in reducing the real value of government and bank liabilities. Where deposits were not highly dollarized, the depreciation helped in some cases to reduce the real value of nonindexed banking system deposits (Indonesia; also Bulgaria and Ecuador prior to the peg). In others, it reduced the real value of government fixed-rate domestic liabilities (Russia).

Nonetheless, the degree of initial exchange rate depreciation observed in these cases was generally more than could be justified by the above considerations. It was important to avoid overshooting and keep ongoing inflation as low as possible. Particularly for countries with a history of poor monetary credibility (Mexico, Brazil) and where political and structural disarray is the most extreme (Indonesia), the risk is that high inflation becomes embedded in expectations and therefore makes subsequent disinflation highly costly. Another risk is that inflationary pressures become uncontrollable and lead to a hyperinflation. For countries with substantial U.S. dollar liabilities (Brazil, Indonesia, Thailand), excessive exchange rate depreciation is dangerous in its effects on balance sheets and, when the government is the U.S. dollar borrower, on fiscal solvency.11

The relationship between the stringency of monetary policy and the size or duration of output loss is weak. The fall in quarterly output exceeded 10 percent in most of the crisis countries; in general, the greater the fall in output, the longer it took countries to recover. Nevertheless, all but one of the countries had recovered at least half of the output loss within a year of the crisis (see Table 4.1). The floating countries that most quickly regained monetary policy control tended to have the smallest output declines. The causality is unclear, however: the rapidity with which countries regained monetary control and the limited fall in output may both have reflected a less devastating initial crisis. The case of Indonesia discussed above and shown in Figure 4.3 suggests that tighter policy eventually was necessary to restore stability. In this case, a looser initial policy may serve only to prolong the period of instability. There is certainly no strong evidence that tighter monetary policy was associated with larger output declines.

The Framework for Monetary Policy in a Floating Regime

Restoring credibility in the aftermath of a financial crisis requires setting up a monetary policy framework that helps to anchor public expectations. (Appendix 4.1 reports on the monetary policy framework post-crisis for each of the countries in the sample.) A hard peg achieves this almost instantaneously, since it provides a highly visible rules-based policy with no scope for discretion. With a float, the task is more demanding. The authorities need to choose the goals, intermediate and operating targets for monetary policy, and a battery of instruments to deploy in order to obtain these targets.12 In the initial phase, the goal in the majority of cases was to halt the free fall of the nominal exchange rate, limit inflation pass-through to reasonable levels, and restore some minimal stability. The policy environment inevitably involved a substantial ad hoc component during this phase as the overall policy package was being assembled. However, even then, the question of how to organize thinking about and communicate monetary policy posed itself immediately. Subsequent to the initial basic stabilization phase, the goal was to achieve price stability while balancing competing goals such as output stability.

The major issue that confronted the authorities was how to pursue monetary policy without relying on a single clear and operational nominal anchor. Countries can in principle choose to target a money aggregate or inflation. However, countries in this sample rarely followed a money anchor in the aftermath of a crisis. In a context where inflation is impossible to predict with any confidence, money targeting would seem to offer the promise of setting a money target as a clear nominal anchor—its achievement assures that there is at least some anchor to the price level. It rarely worked that way in practice, for several reasons:

  • Because of the unpredictability and instability of money demand, monetary aggregate targets rarely served to guide monetary policy execution. Monetary targets were rarely binding, since they were usually widely missed (Mexico) or overachieved (Korea, Thailand, Brazil). These errors were mostly due to surprises in money demand or net international reserves, the latter often the result of large errors in predicting capital flows (Mexico), and did not serve to indicate the adequacy of the monetary stance.13

  • Even if a money target is met, the exchange rate may still be subject to wide swings. These fluctuations, particularly during the panic-prone post-crisis period, risk feeding rapidly into expectations and being validated by balance sheet effects and wage and price-setting dynamics. Monitoring of monetary policy therefore needed to rely on indicators that were observable at high frequencies and bore a direct relationship to market conditions.14

  • The low interest elasticity of money demand in the short run implies that any attempt to strictly control the money supply in the short run tends to result in unbearably high or volatile interest rates (e.g., Turkey, for the first few days after its float). Nonetheless, monetary aggregates can still play a useful supportive role, particularly as objective “trip wires” for cases of egregious failure to conduct an appropriate monetary policy, as has been highlighted by Ghosh and others (2002).15

Inflation targeting has become a popular policy choice for floating exchange rate countries, including many emerging markets.16 However, after a crisis, full-fledged inflation targeting can rarely be put in place quickly. An exception was Brazil, where full-fledged inflation targeting could be implemented after only a few months.17 In other cases, particularly early in the crisis, it is difficult to forecast inflation with any confidence, in part because of residual doubts about the rest of the policy package. Investing the credibility of post-crisis institutions and policymakers in achievement of an inflation target was therefore seen as risky.

In most cases, the authorities exercised a fair amount of discretion in the conduct of monetary policy, taking into account the high-frequency behavior of a variety of indicators, such as expected inflation, the exchange rate, the level of activity, wage developments, and monetary aggregates. Some of these cases can usefully be characterized as having followed informal inflation targeting. In these cases the authorities’ monetary policy actions were largely guided by their stated inflation objective, although they did not have in place the full-fledged inflation-targeting apparatus of central bank accountability, transparency, and independence (Brazil until June 1999; Mexico, at least after 1995; Korea; Turkey). Others for many months after the initial stage maintained an eclectic monetary policy, with no clear nominal anchor (Thailand).

Countries varied in the emphasis they placed on the exchange rate in the conduct of monetary policy. Malaysia adopted a formal peg in mid-1998, several months after stability had largely been restored and when pressures had shifted toward appreciation. Other countries also gave substantial weight to the exchange rate as an indicator of the stance of monetary policy, particularly in the initial turbulent period, since it was the highest-frequency and most visible manifestation of the state of nominal stability and monetary policy credibility. When a large degree of de facto dollarization exists, it may make sense to attach a special importance to the exchange rate, given high pass-through to inflation and potential balance sheet effects. Nonetheless, de facto exchange rate targeting is rarely possible or advisable after a crisis, given the vulnerability to speculative attack it presents.18

The most important instruments of monetary policy in a floating exchange rate regime are open market operations that influence the level of the domestic interest rate. In cases where domestic money markets were not well developed or were seriously disrupted (Russia, Indonesia, Malaysia), other instruments were necessary, such as unsterilized foreign exchange intervention, manipulation of reserve requirements on bank deposits, and direct changes in the central bank’s discount rate.

Sterilized foreign exchange market intervention has also been an important instrument, particularly in the immediate post-crisis period. The closing of the U.S. dollar financing gap typically required not just an adequate supply of dollars “on paper” but substantial sterilized foreign exchange market intervention as well, particularly in the immediate post-crisis period, after the complete policy package had been put in place but before it had become fully credible. Although many countries lost large amounts of reserves both in defending the peg and in the immediate aftermath of the float in ineffectual but costly bouts of sterilized intervention, some limited sterilized intervention may usefully complement an appropriate policy package. In the initial phase, before confidence has returned but after appropriate policies have been put in place, sterilized intervention has helped to accommodate capital outflow until confidence returned.

Foreign exchange has also effectively been provided by indirect means, acting as the equivalent of sterilized intervention. In several cases (Mexico, Korea, also Brazil in 2002), the central bank provided U.S. dollar loans to local banks at a predetermined U.S. dollar interest rate. This lending, and the redemption of dollar-indexed government liabilities directly in U.S. dollars (Mexico), is similar in its effects on the foreign exchange market to sterilized interventions through the foreign exchange market: the central bank accommodates a demand for U.S. dollars in a way that avoids pressures on the foreign exchange market and does not directly change the money supply or domestic interest rates.19

Figure 4.1 (bottom panel) illustrates the role played by U.S. dollar lending during the stabilization phase of the Mexico crisis. The line labeled “central bank lending to banks” shows the stock of direct U.S. dollar-denominated lending by the central bank to the banking system. It illustrates several points. First, the quantity of U.S. dollars provided was substantial, reaching $3.5 billion in early April. Second, net dollar lending continued in substantial quantities for several weeks beyond the critical mid-March point at which stability had begun to return. Finally, the U.S. dollars lent through this window were recovered quickly; this was not a sustained outflow.20

After the initial period during which nominal stability is first established, large-scale interventions to accommodate capital outflows do represent a sign of failure. Indeed, a more typical experience has been for countries to intervene substantially on the buying side later in the post-crisis period, buying U.S. dollars to rebuild international reserves as foreign capital begins to return rapidly, or the current account swings strongly into surplus (Korea, Russia), or both. Prolonged large reserve outflows suggest an inherently futile attempt to substitute provision of U.S. dollars for an adequate overall policy stance. However, even several months or years into the still highly uncertain post-crisis environment, relatively small-scale and intermittent intervention can be a useful tool, particularly in rare moments of panic (Brazil, Mexico in late 1995) and when accompanied by appropriately high interest rates (as discussed in the preceding section).

The monetary authorities in post-crisis countries should be encouraged to quickly devote attention to solidifying and clarifying their monetary policy framework. Most countries that chose to float had trouble articulating and implementing clear strategies and tactics for monetary policy in the aftermath of crises. Some delay in choosing a clear nominal anchor in the aftermath of a crisis is understandable and perhaps inevitable, given the uncertainties surrounding the overall policy framework in the first few months after the crisis.

Nevertheless, the situations where the authorities either have no clear framework (Thailand) or claim that they are money targeting when they are not (Mexico) cannot be conducive to the fastest possible return to monetary policy credibility. The eclectic approach may be sufficient for countries with a strong history of monetary policy credibility, such as Thailand. Countries in Latin America are more likely to benefit from a more explicit strategy, owing to past bouts of high inflation and hence relatively low central bank credibility. For example, it may be helpful to recognize that while aggregates are useful guides to monitoring monetary policy, they are not generally useful in describing or conducting monetary policy. The examples in the sample suggest that, for most countries in Latin America that float, informal inflation targeting moving to full-fledged inflation targeting would appear to be the best choice.

Institutional Reforms to Promote Monetary Policy Credibility

The reforms that can most directly enhance monetary policy credibility after a crisis are changes in the central bank law (or the constitution) and organizational restructuring of key agencies involved in financial sector supervision. (Appendix 4.2 reviews institutional reforms enacted post-crisis to enhance monetary policy credibility in some of the cases examined here.) Monetary policy must be separated from the needs of the private sector or the government for U.S. dollars and from the problems resulting from widespread insolvency in the banking sector. This normally requires concerted action by the government and the legislature, not just (or even necessarily) the central bank, and it is often accompanied by institutional reforms.

  • Many countries undertook institutional reform to handle banking system insolvency by creating new institutions, either in the central bank or as separate agencies capitalized or funded by the government.

  • Some countries off-loaded banking supervision from the central bank. Others did not, on the reasonable grounds that the expertise and resources still lay with the central bank.

  • Some countries enacted measures to limit central bank financing of the government.

It is possible in many instances to stabilize monetary and exchange rate policies effectively without solving institutional weaknesses, even those that contributed to the crisis. Indeed, after a crisis, when institutions have typically just failed dramatically, actual performance may be more important than institutional and especially legal reform in cementing positive expectations.

  • Thailand did not make most of the changes in the structure of its central bank law that would have formally increased its central bank’s autonomy. It nonetheless was able to restore monetary policy credibility. (A long history of inflation credibility and a new government certainly helped here.)

In some cases, full-scale and up-front institutional reform of the monetary policy framework can be necessary. Where the failure of the previous monetary regime was most dramatic, full-scale institutional reform of the monetary policy framework and the central bank in the immediate aftermath of the crises was warranted (Bulgaria, Ecuador). In other cases, where monetary policy has lost credibility (e.g., because of a failed defense of an explicit peg), reforms to help focus expectations on future performance rather than past failures have been helpful.

It is important not to rush institutional and legal reform in a crisis or to reform institutions myopically. Legal and institutional changes are for the long run and can take time to design and implement properly; it is vital to get the new institutions right the first time.21 Moreover, it is possible to move too soon on institutional reform. In the post-crisis turmoil, monetary policies may be pursued that would not be appropriate once normality has been restored. It may not be helpful to the credibility of new institutions to ask them to carry out these unusual policies. For example, where there is a necessary adjustment in the exchange rate (for reasons of overvaluation, for instance) or even the price level (in the context of the need to resolve a banking crisis, for example), the usually desired single focus of monetary policy on price stability must temporarily change (Bulgaria).

However, if the necessary legal and institutional changes are delayed too long or abandoned, this can come back to haunt the country in the future. Thus, the post-crisis period is one that can usefully be taken advantage of to pursue good long-run reforms. For example, the Central Bank of Brazil enjoyed de facto independence in the conduct of monetary policy. However, seating certain aspects of this independence in the law, in particular by staggering terms for the board, would have reduced uncertainty during the presidential election campaign about the future actions of the central bank.

Concluding Remarks

The key conclusions of this analysis, based on the review of experience we have undertaken, are as follows:

  • Monetary policy alone cannot close the financing gap.

  • Floats bring nominal stability quickly in countries with low pre-crisis inflation; hard pegs have been at least narrowly successful for countries in deeper disarray.

  • Targeting of money aggregates will rarely serve as a coherent framework for floats; informal or formal inflation targeting offers more promise.

  • Early and determined tightening brings nominal stability and does not appear more costly for output.

Appendix 4.1. Frameworks of Monetary Policy in Crisis Countries

Table 4.2 reports on the post-crisis framework of monetary policy in the countries of the sample. The table presents information—for the various phases of the crises examined—on duration, final and intermediate targets or goals of monetary polilcy, policy instruments, and IMF program targets.

Table 4.2.Monetary Policy Frameworks in Crisis Countries
Brazil, January 1999
Initial phaseDatesJanuary 15, 1999 to March, 1999
Goal/final target of monetary policyFloating exchange rate; reduce pass-through from exchange rate depreciation to inflation
Intermediate targetsOperating target is the overnight interest rate (the average overnight interest rate in the repo market on government securities, called the SELIC)
InstrumentsOpen market operations in the form of outright sales and purchases and swaps of the central bank (and later Treasury) securities
Substantial intervention
IMF program targetsDecember 1998: Net domestic assets (NDA) based on fixed exchange rate rule; money and NDA breached1
Second phaseDatesMarch, 1999 to June, 1999
Goal/final target of monetary policyInformal inflation targeting, underpinned by a quantity-based framework (staff report acknowledges high uncertainty surrounding money demand estimates)
Intermediate targetsOperating target is still the SELIC
Intermediate target is inflation expectations and the exchange rate
InstrumentsOpen market operations in the form of outright sales and purchases and swaps of central bank securities
Substantial intervention, particularly in early March as interest rates were increased
IMF program targetsMarch 1999: NDA based on money demand. Substantial overshooting of NDA by June (because of the net international reserves [NIR] of the central bank overperformance)
Third phaseDatesJune 1999 to current
Goal/final target of monetary policyInflation targeting with floating exchange rate
Intermediate targetsIntermediate target is inflation forecasts (both internal and market). Operating target is the SELIC
InstrumentsOpen market operations; also, periodically frequent interventions in foreign exchange market, to counteract disorderly conditions and, at times, to resist trends
IMF program targetsConsultation bands on inflation
Bulgaria, February 1997
Initial phaseDatesFebruary 1997 to June 1997
Goal/final target of monetary policyPrepare for introduction of currency board
Intermediate targetsLimits on NDA and money; as operating target, the authorities apparently pegged the interest rate on 28-day government securities (at just under 20 percent monthly Jan.–April, until overperformance on inflation, exchange rate, and NDA allowed a reduction)
InstrumentsStanding lending facilities to banks and purchases of government paper (open market operations)
Other notesStabilized early, without currency board
Second phaseDatesJuly 1997 to current
Goal/final target of monetary policyCurrency board
Intermediate targetsn.a.
InstrumentsIncomes policy based on wages of state-owned enterprises; tight and flexible fiscal policy
Other notesExchange rate was chosen to balance competitiveness concerns with desire to avoid additional inflationary burst at a level close to prevailing spot market rates and at a round number
A special and separate account was established for lender-of-last-resort credits, financed by well-defined external and fiscal resources
Ecuador, January 2000
Initial phaseDatesFebruary 2000 to current
Goal/final target of monetary policyn.a. (full dollarization).
Intermediate targetsn.a. (full dollarization)
InstrumentsInterest rate controls eliminated as part of stabilization
Central bank liquidity recycling by issuing short-term dollar notes in auctions to absorb liquidity and by repos of government securities to inject
Separate facilities to manage banking crisis, including lender-of-last-resort, established with remaining hard currency reserves after dollarization
Indonesia, July 1997
Initial phaseDatesAugust 14, 1997 to April 1998
Goal/final target of monetary policyNo announcement was made at the time of the float; in practice, contain the goal was to limit the impact of the devaluation on banks and to provide necessary liquidity to banks
Intermediate targetsOn October 31, 1997 a target for 12-month base money was set; in practice, authorities monitored interest rates closely, relying on various indicators of stability, including attempting to achieve positive real interest rates
InstrumentsThe authorities set a policy interest rate through a variety of mechanisms including direct control, at least until a central bank securities market was created in July 1998. In August 1997, state enterprise deposits were transferred from banks to the Central Bank of Indonesia (BI), resulting in huge tightening of liquidity (this took most of the reserves of the banking system and sharply reduced base money). BI also intervened in the interbank market to redistribute liquidity from strong to weak banks. Liquidity support was provided through a variety of instruments, with capitalization of high interest rates removing any deterrent effect thereof. In March, new procedures put in place to provide liquidity at small premium above market, with non-market sanctions for excessive borrowing.
The initial program allowed substantial sterilized intervention; in fact, there was foreign exchange intervention of $7.47 billion between September and December 1997
IMF program targetsBase money, hugely breached through excessive liquidity provision not fully sterilized
Other notesPost-crisis exchange rate regime was a fairly free, though still managed, float, with substantial foreign exchange intervention and also active use of monetary policy (the interest rate) to counter exchange rate movements
Capital controls put in place in August 1997, including restriction on forward rupiah transactions between banks and nonresidents
Second phaseDatesMay 1998 to May 1999
Goal/final target of monetary policyPrice and exchange rate stability, controlling liquidity effect of support to banking system
Intermediate targetsBase money targeting, as well as quantitative targets on other aspects of BI’s balance sheet; between reviews, monetary policy was oriented by exchange rates and interest rates
InstrumentsOpen market operations were not effective, because of thin SBI markets; so where prior to this, the interest rate was targeted in the auction, this was changed to quantities on July 29, 1998
Some unsterilized foreign exchange intervention from time to time to mop up liquidity to meet targets
IMF program targetsSwitch to performance criteria on NDA; April plan (constant NDA) was breached with 18 percent growth in one month; June 25 plan’s constant NDA ceiling was met for several months
Other notesStrict quantitative targets were essentially prudential
Third phaseDatesMay 1999 to current
Goal/final target of monetary policyInflation targeting; new central bank law of May 1999 specified the maintenance of the value of the rupiah as the overriding goal, the announcement of an inflation target, and the granting of instrument independence to the central bank
IMF program targetsBase money
Korea, 1997
Initial phaseDatesJuly 1997 through December 1997
Goal/final target of monetary policyStabilize the exchange rate
Intermediate targetsThe operating target was an interest rate (overnight interbank rate)
InstrumentsRepos and outright transactions with government-guaranteed and Bank of Korea bonds; rediscount facility for policy purposes, standing facility for banks to meet settlement obligations. Substantial (sterilized) intervention to provide dollars to meet withdrawals of foreign credit lines, largely through state-owned financial institutions and other indirect methods, through foreign exchange deposits in overseas branches and direct market foreign exchange intervention. Reserves severely depleted, from $30 billion end-September to about $6 billion usable by early December, but run continues and won weakens further.
Second phaseDatesJanuary 1998 through December 2000
Goal/final target of monetary policyAvoid inflation/depreciation spiral, stabilize the won, and accumulate reserves; floating exchange rate regime, but with substantial intervention and important role for the exchange rate in conduct of monetary policy
Intermediate targetsOperating target: overnight call rate; intermediate target was largely the exchange rate, informally2
InstrumentsSame as above
Intervention to accumulate reserves and stem appreciation in 1998, partially sterilized
Other notesLegal ceilings on interest rates had to be removed in December 1997; measures to redistribute liquidity among banks in December 1997
Third phaseDatesJanuary 2001 to current
Goal/final target of monetary policyInflation target
Intermediate targetsOvernight call rate
InstrumentsSame as above
Malaysia, July 1997
Initial phaseDatesJuly 1997 through September 1998
Goal/final target of monetary policyManaged floating exchange rate; goal was to stabilize the exchange rate, but avoid increases in interest rate that would damage highly leveraged economy
Intermediate targetsOperating target: three-month interbank rate (basis for lending rates of commercial banks); also controlled the overnight interest rate, which the authorities allowed to move (and increase) more; finally, also controlled credit quantities directly
InstrumentsDirect deposit and loan operations with commercial banks, government deposits, outright sales of central bank bills, overnight credit facility to facilitate clearing and settlement. During periods of pressure, unremunerated reserve requirements; after the rate spike in July 1997, the authorities let rates come down but put more emphasis on direct instruments such as credit plans for financial institutions (limiting overall credit growth) and a ban on new lending to the property sector.
Second phaseDatesSeptember 1998 through current
Goal/final target of monetary policyMalaysia adopted a fixed exchange rate against the U.S. dollar in September 1998
Intermediate targetsOperating target is still the policy interest rate; intermediate target is the exchange rate
InstrumentsMalaysia imposed limits on noncommercial bank swap ringgit offer-side swap transactions in August 1997; much more comprehensive capital controls imposed in September 1998, with the elimination of the offshore ringgit market
The central bank also engages in substantial sterilized intervention in defense of the peg (with interest rate volatility fairly low but reserve volatility high)
Mexico, December 1994
Initial phaseDatesJanuary 1995 to March 1995
Goal/final target of monetary policyOffset the inflationary effects of the devaluation, to reduce inflation volatility and prevent further excessive depreciation; an inflation objective of 17 percent was announced in January, changed to 42 percent in March
Intermediate targetsInitially, a monthly target on NDA, along with assumption of no foreign exchange intervention, supposed to imply a money path. Some public emphasis was placed on an annual target for NDA and money base. By early 1995, this failed to perform as expected, due to unstable velocity and the fact that the money rule did not prevent exchange rate fluctuations from feeding quickly into inflation, and the fact that the central bank had little control on the monetary base in the short run. In late March, authorities moved to target interest rates directly (through floors in open market operations), spiking the overnight rate up to 100 percent
InstrumentsOpen market operations using fixed-interest-rate auctions (or announcing maximum or minimum rates). Also, substantial off-market (sterilized) intervention in form of government bond amortization and dollar loans from central bank to banking system at a given (though changing through time) interest rate. No explicit foreign exchange market intervention after January during this period.
Other notesSubstantial increase in all aspects of information provision, including daily data on money, accounts of banks with central banks, etc.
Second phaseDatesMarch 1995 through December 1996 (and beyond)
Goal/final target of monetary policySame; consistent inflation objective through year; from 1996 on, more public emphasis is placed on the annual inflation target
Intermediate targetsPublic targets for the path of the monetary base, along with commitments on NDA and NIR (to assure the market that the Central Bank of Mexico [BOM] will not create the most basic source of inflation: excess supply of primary money). Operationally, the BOM establishes a target on the average borrowed reserves that it changes from time to time. An increase in the borrowed reserve target (the corto) tends to increase interest rates. Importantly, the announced path for money is not a formal policy objective, given uncertainties about the relationship between base money and inflation and the basic assumptions about GDP growth, interest rates, etc. Thus, the BOM observes the exchange rate, available measures of inflation expectation, wages, and the output gap and tightens or loosens its monetary borrowed reserve target depending on whether it sees inflation as being on track.
InstrumentsThe BOM estimates the demand for liquidity daily, and through open market operations provides enough liquidity to meet that demand, less the target for the size of borrowed reserves
The BOM automatically sterilizes any changes in NIR, which occur frequently due to government debt operations and changes in dollar lending to the banking system; high volatility and panic in foreign exchange markets in October–November 1995 led to $500 million market interventions by the central bank
In August1996, the BOM began to auction the right to sell dollar to the central bank; the options were structured so that they are only executed “against the wind” and in predetermined amounts, so that no level objective for the exchange rate is implied
IMF program targetsThe IMF’s NDA targets were overshot early on, as dollar outflows were larger than anticipated and were sterilized, and for the year as a whole
Other notesMore recently, the BOM has moved to a more systematic inflation targeting
Russia, August 1998
Initial phaseDatesAugust 17, 1998 to December 1998
Goal/final target of monetary policyPolicy was reactive, financing the government and providing credit to ailing banks (resulting in a large increase in net credit)
Intermediate targetsNo single intermediate target, though arresting the exchange rate depreciation was one priority
InstrumentsWith banking crisis, rehabilitation loans at negative real interest rates, collateralized with bank equity, were extended in often nontransparent fashion
Substantial foreign exchange interventions, both through the market and with government to repay foreign credits. Starting in early August, run on foreign exchange reserves fueled by bank liquidity support. Later, unsterilized foreign exchange purchases to partially offset impact on reserves of debt-service payments on Russian-era debt (i.e., debt payments came partially from market).
Other notesThe introduction of capital controls, a complex usage of the main savings bank (Sberbank) to limit deposit outflows, and the deflationary effect of the banking crisis somewhat limited the inflationary and depreciative effects of the liquidity injection
Key initial measures included a forced restructuring of domestic treasury bills (GKOs) and a 90-day freeze on private external debt service (including hedge sold to foreigners by banks
Second phaseDatesJanuary 1999 to current
Goal/final target of monetary policyInflation control in the context of a floating exchange rate, though with heavy implicit exchange rate targeting; by December 1999, de facto crawling peg (with occasional deviations of up to ± 2 percent)
Intermediate targetsThe official intermediate target was reserve money. Given the uncertainties, foreign exchange market developments would provide early indication of unexpected changes in monetary conditions. Thus, according to IMF staff, the Central Bank of Russia (CBR) intended to lower reserve money below the projected path in the event that foreign exchange market pressures were larger than expected, and vice versa. In practice, the CBR seems to have had implicit exchange rate targets: between April and September 1999, interventions were such as to keep the ruble mostly constant against the dollar; between September 1999 and January 2000, the ruble depreciated by 1–2 percent per month; between January and May 2000, the ruble was again largely constant.
InstrumentsReserve requirements were unified in January and subsequently raised several times. Open-market operations hampered by GKO default and legal issues with respect to CBR bills. That left deposit taking from commercial banks, which is nontradable and hence inflexible. Foreign exchange interventions both through the foreign exchange market to sterilize injections of liquidity and directly to the government for foreign debt service. Intervention would be both at accumulating reserves and smoothing exchange rate fluctuations.
IMF program targetsBase money was 10 percent above programmed level by June 1999 (program was agreed in March), but with overperformance on NIR and hence NDA, there was substantial slack in the targets
Other notesCBR government financing had little monetary impact, as it was largely limited to foreign exchange credit to service Russian-era foreign debt. Fiscal improved somewhat by 1999:Q1 and markedly by Q2, as the overall cash deficit fell from 5.4 percent of GDP in 1998:Q4 to 4.7 percent in Q1 to 2.4 percent in April. The government had no resort to domestic or external financing in 1998–99. New GKOs were issued in December 1999 to foreigners and in February 2000 to residents.
Thailand, July 1997
Initial phaseDatesJuly 1997 to April 2000
Goal/final target of monetary policyInitial objective was to stabilize the exchange rate; The authorities adopted a float, with exchange rate volatility going up relative to interest rate and reserve volatility. Still, there was substantial effort to influence the bilateral exchange rates. Over time, and with stability, there was a subtle firming of understandings to defend the exchange rate within some (implicit) band.
Intermediate targetsThe operating target is a money market interest rate; as in Korea, there were NDA and base money targets, but program monitoring put a special, less formal, focus on interest rates. Monetary policy between program reviews was oriented by exchange rates and the nominal interest rate. An eclectic approach evolved, with pragmatic considerations determining the setting of the Bank of Thailand (BOT) policy interest rate. However, as exchange rate stability was achieved and maintained, the focus of monetary policy shifted to supporting economic recovery, with the BOT guiding money market rates to as low a level as possible without undermining confidence. By early 1999, overnight repurchase rates had fallen to a range of 1 to 2 percent (annual rates), and have generally remained around that level.
InstrumentsOpen market operations through repos with public sector securities, in addition to a loan window for a lender-of-last resort facility, an intraday liquidity facility, and an overnight facility
Foreign exchange intervention through foreign exchange swaps, particularly right after the crisis. Over time, this has abated. The BOT continues to auction variable quantities of foreign exchange daily. As pressure built, in May–June 1997, Thailand limited baht lending to nonresidents, exempting “genuine underlying business transactions.” This led to a two-tier market, though spreads between the two exchange rates were narrow
IMF program targetsLike Korea, Thailand was always substantially under the reserve money floor, somewhat over on NIR, and slightly under on NDA
Other notesThailand moved to full-fledged inflation targeting in April 2000; there is an institutional tension between the role of the BOT in monetary policy and its role in providing ongoing financing of bank recapitalization
DatesFebruary 22, 2001 to present
Goal/final target of monetary policyPrice stability within the context of a floating exchange rate regime. The authorities announced an inflation objective of 52 percent in March 2001 but avoided calling it a target. The authorities have intended to move to inflation targeting when possible3
Intermediate targetsAn initial attempt was made to freeze domestic liquidity, then, after a few days, the Central Bank of Turkey (CBT) publicly committed to providing liquidity at a maximum interest rate of 150 percent (simple) (maximum).4 Subsequently, an interest rate has been the operating target. By May 2001, the CBT was to focus on the control of monetary aggregates, with a target for base money. Because of a large margin for error, it was acknowledged that the CBT would follow other inflation indicators, so it would raise interest rates even if base money were close to target if developments threaten to jeopardize the disinflation process. In practice, the CBT looked increasingly at expected inflation and indictors thereof, mostly the exchange rate but to a lesser extent money, in setting the policy interest rate. There is some suggestion that the aggregate targets were asymmetric, with overshooting of the money base supposed to result in tightening, while undershooting or hitting the target meant the other indicators including expected inflation were what mattered. Over time, particularly in 2002, the policy became more clearly one of informal inflation targeting
InstrumentsThe authorities set the level of a policy interest rate through open market operations. Initially, the authorities also intervened in an attempt to avoid overshooting and allow banks and residents to honor external liabilities (losing $4 billion more in reserves by April). They moved to predetermined foreign exchange auctions in March. Since September 11, 2001, they have periodically intervened, typically on a predetermined basis. In 2002, they have tended to intervene in a preannounced fashion to buy dollars. Discretionary intervention since early 2001 has been minimal
IMF program targetsThe CBT met all NDA and money targets and indicative ceilings, in some cases by a small amount and in some cases with a substantial margin
Other notesThere was some market confusion, particularly early in the program, about the apparent absence of nominal anchor and lack of CBT clarity about what it was doing. By 2002, it was becoming clearer to market participants that the CBT was engaged in a sort of informal inflation targeting, with expected inflation the main intermediate target and with the monetary aggregates as checks against going off track. Dollarization greatly complicated base money targeting in 2001, as a shift into dollars lowered base money demand (thus, base money was met, but inflation was not). Fiscal dominance has at times constrained monetary policy (raising rates to hit money or inflation would cause fiscal problem)
Sources: IMF staff reports, IMF Staff Country Reports; government Letters of Intent; IMF technical assistance reports; and government central bank reports, as well as Reinhart and Rogoff (2002) and Hernandez and Montiel (2001) for exchange rate arrangements; Edwards and Savastano (1998) and Carstens and Werner (2000) for Mexico; Gulde (1999) and Enoch, Gulde, and Hardy (2002) for Bulgaria; Enoch and others (2001) for Indonesia; and Boorman and others (2000), Lane and others (1999), Lindgren and others (1999), and Ghosh and other (2002) for various cases.

According to IMF staff, demand for base money in February 1999 may have been boosted by a flight to liquidity at the end of January, prompted by rumors of a possible asset freeze (which would not have applied to demand deposits), as well as by seasonal factors, notably the carnival holidays.

For example, there was an understanding in early 1998 that the authorities would not reduce interest rates until the exchange rate had substantially appreciated back to W 1,400 per dollar, though there was no explicit commitment to raise rates until that could be achieved. Money and credit aggregates were not useful given shifting market conditions, particularly for day-to-day policymaking, because of lags in measurement and uncertainty about money demand. There was also substantial uncertainty about required real exchange rate adjustment, so money and NDA may have served to warn if the program were well off-track. In the event, reserve money was well below program levels at end-March and end-June 1998. Meanwhile, Korea substantially overperformed on NIR (and thus NDA), which was thus also not a binding constraint on monetary policy.

The authorities have not moved to inflation targeting earlier due to a belief that high inflation and ongoing fiscal problems made a clear commitment to hit a particular inflation target too risky, and because of a need to improve inflation forecasting techniques; set up procedures for implementation, transparency, and accountability; and prepare public opinion.

The effort to freeze the money base resulted in extremely high interest rates (some 10,000 percent annualized) in an almost totally frozen market. Rolling overnight claims were resulting in huge transfers to creditors and away from state banks. The exchange rate continued to depreciate anyway.

Sources: IMF staff reports, IMF Staff Country Reports; government Letters of Intent; IMF technical assistance reports; and government central bank reports, as well as Reinhart and Rogoff (2002) and Hernandez and Montiel (2001) for exchange rate arrangements; Edwards and Savastano (1998) and Carstens and Werner (2000) for Mexico; Gulde (1999) and Enoch, Gulde, and Hardy (2002) for Bulgaria; Enoch and others (2001) for Indonesia; and Boorman and others (2000), Lane and others (1999), Lindgren and others (1999), and Ghosh and other (2002) for various cases.

According to IMF staff, demand for base money in February 1999 may have been boosted by a flight to liquidity at the end of January, prompted by rumors of a possible asset freeze (which would not have applied to demand deposits), as well as by seasonal factors, notably the carnival holidays.

For example, there was an understanding in early 1998 that the authorities would not reduce interest rates until the exchange rate had substantially appreciated back to W 1,400 per dollar, though there was no explicit commitment to raise rates until that could be achieved. Money and credit aggregates were not useful given shifting market conditions, particularly for day-to-day policymaking, because of lags in measurement and uncertainty about money demand. There was also substantial uncertainty about required real exchange rate adjustment, so money and NDA may have served to warn if the program were well off-track. In the event, reserve money was well below program levels at end-March and end-June 1998. Meanwhile, Korea substantially overperformed on NIR (and thus NDA), which was thus also not a binding constraint on monetary policy.

The authorities have not moved to inflation targeting earlier due to a belief that high inflation and ongoing fiscal problems made a clear commitment to hit a particular inflation target too risky, and because of a need to improve inflation forecasting techniques; set up procedures for implementation, transparency, and accountability; and prepare public opinion.

The effort to freeze the money base resulted in extremely high interest rates (some 10,000 percent annualized) in an almost totally frozen market. Rolling overnight claims were resulting in huge transfers to creditors and away from state banks. The exchange rate continued to depreciate anyway.

Appendix 4.2. Institutional Reform for Monetary Policy Credibility in Selected Crisis Cases

This appendix reviews experience with the role of central bank credibility—or lack thereof—in shaping crisis response, and of post-crisis institutional reform in enhancing credibility.22

Inadequacies in the legal and institutional frameworks governing central banks may not have caused the recent financial crises in Asia and elsewhere, but these weaknesses have—to varying degrees—facilitated and amplified the problems. Factors that negatively affected central bank performance include: (1) insufficient political and financial autonomy; (2) numerous and conflicting objectives; (3) insufficient authority and powers to achieve these objectives; and (4) inadequate central bank accountability.

Asian Crisis

It is generally accepted that inadequate financial supervision was in part to blame for the weak financial systems of the countries affected by the Asian crisis. Other crisis factors involving the central bank were the defense of unsustainable exchange rate pegs and extensive financial support to insolvent financial institutions. The problems created by central bank actions reflected in part that the central banks had to pursue conflicting objectives, which they failed to achieve simultaneously. Some of these failures were the direct result of undue government interference with central bank decisions. The success of efforts to reform the legal and institutional framework of these central banks so as to avoid recurrence of such problems has been mixed.

Thailand. The 1997 crisis, which was primarily caused by an unsustainable exchange rate that led to the depletion of central bank reserves in support of the baht, also revealed weak prudential supervision by the Bank of Thailand. Main shortcomings involved the practice of providing financing to insolvent financial institutions while delaying closure or restructuring decisions. Substantial liquidity support was provided through the Financial Institutions Development Fund (FIDF), which was financed primarily from government-guaranteed bonds and Bank of Thailand resources; such financing was found to be a major contributor to Thailand’s exchange market pressures. In the context of the StandBy Arrangement approved by the IMF on August 20, 1997, the authorities committed to implement measures to initiate a credible and up-front restructuring of the financial sector, isolate and liquidate insolvent financial institutions, and conduct strict control over monetary expansion. Amendments to central bank legislation were not a performance criterion under the IMF program, but were policy commitments assumed by Thailand’s Letters of Intent and Memoranda of Economic and Financial Policies.23

Despite several IMF missions to review the ongoing preparations of amendments to the Bank of Thailand Act and to the Currency Act called for under the authority’s program, no amendments had been passed as of this writing.24

Indonesia. The central bank’s failure to pursue price stability was exacerbated by the financing it provided to the financial sector. Government pressure in this regard was a particularly important factor: under the 1968 central bank legislation, all Bank of Indonesia decisions were subject to the approval of a Monetary Council chaired by the Ministry of Finance and formed by other ministers of the government. Amendments to the 1968 legislation with a view to institutionalize the central bank’s autonomy formed a core element of the 1997 IMF Stand-By Arrangement (see Indonesia’s October 31, 1997 LOI, available on the Internet at; and IMF, “IMF Approves Stand-By Credit for Indonesia,” Press Release 97/50, November 5, 1997, at As a result, a new Bank of Indonesia Act was enacted on May 17, 1999, that established the central bank as an independent entity free from government interference and solely responsible for the formulation and implementation of monetary policy. Under the 1999 law, the Bank of Indonesia’s main objective is to achieve and maintain the stability of the rupiah, and other responsibilities, including banking regulation and supervision, are spelled out as tasks for achieving Bank of Indonesia’s main objective. The law provides for the transfer of responsibilities in banking supervision to a separate entity to be established no later than December 2002 (now extended to December 2003). After the passage of the 1999 Bank of Indonesia Act, the need for strengthening the Bank of Indonesia accountability regime without compromising the central bank’s independence was identified, and proposals were made for amending the law (but no such amendments had been passed as of this writing).

Korea. The financial system was undermined by excessive government interference in the economy, inadequate sequencing of capital account liberalization, and the lack of prudential regulation that should accompany liberalization. At the time, the government had considerable influence over the Bank of Korea, in particular through the Minister of Finance as the Chairman of the Bank of Korea’s Monetary Board, its main decision-making body. As part of its IMF-supported program (see Korea’s December 3 and 24, 1997 LOIs, available on the Internet at and, respectively; see also IMF, “IMF Approves SDR 15.5 Billion Stand-by Credit for Korea,” Press Release 97/55, December 4, 1997, at, and IMF, “Korea Strengthens Economic Program; IMF to Activate Additional Financial Support,” News Brief 97/32, December 24, 1997, at, the government committed to seek early passage of legislation to make the Bank of Korea independent and with price stability as its overriding mandate. It further committed to consolidate all supervisory functions under one agency with operational independence and adequate resources. The December 31, 1997 amendments to the Bank of Korea Act fell short of important IMF recommendations but constituted an overall improvement of Bank of Korea’s independence. Effective April 1, 1998, the responsibilities for all financial sector supervision were moved to a new agency, the Financial Supervisory Commission/Financial Supervisory Service, which, for the first time, ensured that all depositary institutions, including those over which the Bank of Korea had not had authority, were supervised effectively and on the same basis.

Other Countries

Turkey. Inflation in Turkey averaged close to 80 percent per year in the 1990s, after 25 percent in the 1970s and 50 percent in the 1980s. Although monetary financing of government deficits by the Central Bank of Turkey was identified as the ultimate cause of inflation, no commitments for reform of the legal framework for the Central Bank of Turkey, in particular a prohibition on credit to the government, were included as program conditionality. The creation of base money was contained through a quantitative performance criteria on net domestic assets. However, the December 18, 2000 LOI (available on the Internet at—in the context of the combined third and fourth reviews and the first augmentation of the Stand-By Arrangement—provided for the adoption of a new central bank law by end-April 2001 “with the goal of designating price stability as the primary monetary policy objective of the Central Bank of Turkey, enhancing its operational independence in the pursuit of that objective, increasing its degree of accountability for monetary policy decisions, and enhancing the transparency of monetary policy formulation” as a condition (i.e., prior action) for the eighth review under the program. The new central bank law was adopted in April 2001. It prohibits Central Bank of Turkey credit to the government, provides for price stability as the primary objective for the Central Bank of Turkey, and establishes operational independence for the Central Bank of Turkey (including the establishment of a monetary policy committee). A new Banking Act had been passed in 1999, and the supervisory powers of the Central Bank of Turkey and the Treasury had been shifted to the newly created Banking Regulation and Supervision Agency, which became operational on September 1, 2000.


    BergAndrewChrisJarvisMarkStone and AlessandroZanello2003Re-Establishing Credible Nominal Anchors After a Financial Crisis: A Review of Recent Experience,IMF Working Paper (Washington: International Monetary Fund, forthcoming).

    BoormanJack and others 2000Managing Financial Crises—the Experience in East Asia,IMF Working Paper00/107 (Washington: International Monetary Fund).

    BorenszteinEduardo and Jose DeGregorio1999Devaluation and Inflation,Working Paper (unpublished; Washington: International Monetary Fund).

    CaballeroRicardo and ArvindKrishnamurthy2002A ‘Vertical’ Analysis of Monetary Policy in Emerging Markets,Working Paper IMF Seminar Series No. 2002-69 (unpublished; Washington: International Monetary Fund,March8).

    CarareAlina and others 2002Establishing Initial Conditions in Support of Inflation Targeting,IMF Working Paper 02/102 (Washington: International Monetary Fund).

    CarstensAgustin and AlejandroWerner2000Mexico’s Monetary Policy Framework Under a Floating Exchange Rate Regime,Money Affairs Vol. 13 (July-December) pp. 11365.

    ChoudhriEhsan and DaliaHakura2001Exchange Rate Pass-through to Domestic Prices: Does the Inflationary Environment Matter?IMF Working Paper 01/194 (Washington: International Monetary Fund).

    ChristianoLawrence J.ChristopherGust and JorgeRoldos2002Monetary Policy in a Financial Crisis” Working Paper 2002–05 (Chicago: Federal Reserve Bank of Chicago).

    EdwardsSebastian and MiguelA. Savastano1998The Morning After: The Mexican Peso in the Aftermath of the 1994 Currency Crisis,NBER Working Paper 6516 (Cambridge, Massachusetts: National Bureau of Economic Research).

    EnochCharles A. and others 2001Indonesia: Anatomy of a Banking Crisis—Two Years of Living Dangerously, 1997–99,IMF Working Paper 01/52 (Washington: International Monetary Fund).

    EnochCharles A.Anne-MarieGulde and Daniel C.Hardy2002Banking Crises and Bank Resolution: Experiences in Some Transition Economies,IMF Working Paper 02/56 (Washington: International Monetary Fund).

    GhoshAtish and others 2002IMF-Supported Programs in Capital Account Crises,Occasional Paper 210 (Washington: International Monetary Fund).

    GoldfajnIlan and PoonamGupta1999Does Monetary Policy Stabilize the Exchange Rate Following a Currency Crisis?IMF Working Paper 99/42 (Washington: International Monetary Fund).

    GoldfajnIlan and SergioWerlang2000The Pass-Through from Depreciation to Inflation: A Panel Study,Working Paper 423(Rio de Janeiro: PUC-RIO).

    GuldeAnne-Marie1999The Role of the Currency Board in Bulgaria’s Stabilization,IMF Policy Discussion Paper 99/03 (Washington: International Monetary Fund).

    HernandezLeonardo F. and PeterJ. Montiel2001Post-Crisis Exchange Rate Policy in Five Asian Countries: Filling in the ‘Hollow Middle’?IMF Working Paper 01/170 (Washington: International Monetary Fund).

    HonohanPatrick and AnqingShi2001Deposit Dollarization and the Financial Sector in Emerging Economies,Policy Research Working Paper 2748 (Washington: World Bank, December).

    KellMichael and AxelSchimmelpfennig2003Fiscal Vulnerability and Financial Crises,IMF Working Paper (Washington: International Monetary Fund, forthcoming).

    LaneTimothy1999IMF-Supported Programs in Indonesia, Korea, and Thailand: A Preliminary Assessment,Occasional Paper 178 (Washington: International Monetary Fund).

    LindgrenCarl-Johan1999Financial Sector Crisis and Restructuring: Lessons from Asia,Occasional Paper 188 (Washington: International Monetary Fund).

    LybekTonny1998Elements of Central Bank Accountability and Autonomy,MAE Operational Paper 98/1 (unpublished; Washington: International Monetary Fund, Monetary and Exchange Affairs Department).

    MassonPaul R.MiguelA. Savastano and SunilSharma1997The Scope for Inflation Targeting in Developing Countries,IMF Working Paper 97/130 (Washington: International Monetary Fund).

    MeesookKanitta2001Malaysia: From Crisis to Recovery,Occasional Paper 207 (Washington: International Monetary Fund).

    MishkinFrederic S.2000Inflation Targeting in Emerging-Market Countries,American Economic Review Vol. 90 No. 2 (May) pp. 10509.

    ReinhartCarmen and KennethRogoff2002The Modern History of Exchange Rate Arrangements: A Reinterpretation,NBER Working Paper 8963 (Cambridge, Massachusetts: National Bureau of Economic Research, June).

Inputs to this section were also provided by Tonny Lybek.

Dates in parentheses refer to the year in which the crisis began.

Table 4.1 gives the starting dates for the crisis cases considered, as well as how many months it took for the nominal exchange rate to stop depreciating and for volatility to reach levels typical of stable floating exchange rates. The dating of the beginning of the crisis is somewhat arbitrary, particularly for those countries in which the crisis involved a more gradual loss of nominal control (Bulgaria, Ecuador). See Berg and others (2003) for a summary of the key developments and economic indicators for each of the cases.

In cases where dollar-denominated liabilities are a high proportion of the total, even substantial depreciation of the currency and accompanying inflation—one possible way of resolving the interest rate trade-off on domestic-currency-denominated debt— will not work because it also raises the probability of default.

All the cases reviewed here had generally open capital accounts. The implementation of new capital controls during the crisis was highly problematic. Capital controls have proven largely ineffective in situations when nominal stability had not already largely been restored (Thailand, Russia). They did not stop the exchange rate collapse and indeed may have promoted further capital outflow, at least in the short run. Malaysia introduced selective capital controls on a temporary basis in September 1998, after initial stabilization had been achieved, accompanied by a currency peg set at a significantly undervalued level. See Meesook and others (2001).

Reinhart and Rogoff (2002) characterize uncontrolled depreciations as “freely falling” exchange rate regimes to distinguish them from more functional floats.

Figure 4.2 demonstrates these conclusions by showing levels of the exchange rate, interest rates, and inflation for an average of Brazil, Korea, Malaysia, and Thailand. Mexico and Indonesia are excluded because the timing of their exchange rate trajectories is sufficiently different to obscure the implications of the average, but their outcomes are qualitatively similar. Berg and others (2003) provide figures showing real and nominal exchange rates, inflation, and interest rates for each of the countries in the sample.

Russia also had an incomplete disinflation, reflecting its decision to maintain a highly depreciated level of the nominal exchange rate after initial stabilization. Turkey, a country that like Bulgaria and Ecuador had a recent history of high inflation, also benefited from a relatively small nominal appreciation despite having floated its currency. This suggests that that history, as well as the type of exchange rate regime chosen, shapes the path of disinflation.

Berg and others (2003) give data on these variables for each country in the sample.

This is consistent with evidence in Goldfajn and Gupta (1999) on monetary policy after crises, and Borensztein and De Gregorio (1999), Goldfajn and Werlang (2000), and Choudhri and Hakura (2001) on inflation pass-through in developing countries. One important conclusion of this research is that exchange rate depreciation beyond levels that are consistent with some definition of long-run equilibrium (for example, defined simply as the value associated with the long-run trend real exchange rate) is particularly inflationary. Lindgren and others (1999) and Ghosh and others (2002) contain fuller discussions of the evidence on the relationship between exchange rates and monetary policy in the post-crisis environment. Newer work by Christiano, Gust, and Roldos (2002) and Caballero and Krishnamurthy (2002) tends to support the view, expressed in Ghosh and others (2002), that tight monetary policy is likely to be necessary after a crisis.

Where financial system vulnerabilities result mostly from excess domestic leverage rather than from liability dollarization, it may be more appropriate to keep interest rates lower and allow a larger depreciation (Malaysia).

The goal is the ultimate objective of policy, such as stable prices and output close to its potential level. Intermediate targets are more immediately observable indicators of whether policy is adequate, such as the inflation forecast or monetary aggregates. Operating targets, such as interest rates or the exchange rate, are directly achievable by the central bank on a regular basis. Instruments, such as open market operations and foreign exchange market intervention, are means to achieve operating targets.

Increasing dollarization was an important factor making money demand hard to predict in Turkey, although the issue is more general.

Ghosh and others (2002) emphasize this point. Carstens and Werner (2000) are revealing about the futility of Mexico’s short-lived experience with money aggregate targeting in early 1995. See also Edwards and Savastano (1998).

Indonesia may represent something of an exception to the rule that monetary aggregates did not help to guide policy. For a few key months in 1998, at least, base money did actually track targets quite closely. Two special factors may have been important here. First, Indonesian monetary policy credibility, and the level of the rupiah, had fallen to an extremely low point by April/May 1998, even relative to the other cases considered here, when the monetary aggregate ceilings started to bind effectively. In this context, even a crude policy of keeping aggregates constant was a major improvement. Second, the shocks that called for a contractionary monetary policy during this period tended to cause flight from bank deposits into rupiah cash, hence increases in money demand. Hence, a monetary aggregate target tended to at least give the correct sign to the policy response. In more typical cases, negative shocks may sometimes reduce cash demand, in which case a money aggregate target might well give the wrong sign for the policy response.

In its full-fledged form, inflation targeting involves (1) the public announcement of medium-term numerical targets for inflation; (2) an institutional commitment to price stability as the primary goal of monetary policy, to which other goals are subordinated; (3) an information-inclusive strategy in which many variables, and not just monetary aggregates or the exchange rate, are used for deciding the setting of policy instruments; (4) increased transparency of the monetary policy strategy through communication with the public and the markets; and (5) increased accountability of the central bank for attaining its inflation objectives. For discussions of inflation targeting in emerging markets, see Mishkin (2000), which contains this definition, as well as Masson, Savastano, and Sharma (1997), and Carare and others (2002), and the many references cited therein.

Brazil also stands out as the only country in the sample that did not suffer a banking crisis along with the currency crisis. This was surely an important factor in permitting the authorities to create the new monetary policy framework, and indeed to stabilize, as quickly as they did.

Russia represents an intermediate case. It heavily managed its float in 1999 with months of exchange rate stability interrupted by adjustments of the level, achieved in part through substantial intervention. Toward the end of the year, it moved to a de facto crawling peg. The emphasis on the exchange rate was facilitated by (1) its decision to maintain a highly depreciated level of the exchange rate; (2) the fact that its prior default had rendered it somewhat less vulnerable to further attack; and (3) strong fiscal performance, greatly abetted by the sharp rise in the price of oil in the post-crisis period.

A difference is that unlike sterilized intervention, the central bank incurs no foreign exchange risk.

As the Korean and Mexican examples illustrate, monetary policy is often tightened simultaneously with the intervention. We could, as a manner of terminology, call these interventions partially sterilized. It is useful to distinguish the monetary policy and intervention choices, however, for several reasons. First, in practice, the decision about what to do with monetary policy was a separate one from the decision to intervene (and sterilize). In the Mexico example, the dollars were simply lent directly; there was no automatic domestic monetary impact. More generally, monetary policy was typically conducted in terms of interest rate rules, so that foreign exchange interventions were automatically sterilized. Second, the local currency value of foreign exchange interventions often dwarfed the reductions in the money base associated with a monetary policy contraction. Thus, the measured share of the intervention that was sterilized was usually very large, even when the associated monetary contraction was important when measured in terms of interest rates.

A fuller overview of what have now become standard recommendations can be found in Lybek (1998).

The main authors of this appendix were Bernhard Steinki and Gabriela Rosenberg.

Starting with the November 25, 1997 LOI/MEFP (available on the Internet at and repeated in connection with program reviews, Thailand’s authorities expressed their intentions to “amend the Currency Act in order to modernize the regulatory framework for central banking operations.” In the LOI/MEFP dated May 26, 1998 (available on the Internet at the authorities set out their plan to review the central bank, commercial bank, and finance company laws by October 1998, and to obtain cabinet approval of the necessary amendments by December 31, 1998. They also noted their intention to amend the Currency Act by October 31, 1998.

The IMF’s recent July 12, 2002 Article IV consultation discussions underscored that an early passage of a revised Bank of Thailand Act and the Currency Act to enhance the operational independence of the central bank would strengthen the effectiveness of Thailand’s monetary policy. See IMF, “IMF Concludes 2002 Article IV Consultation and Post-Program Monitoring Discussions with Thailand,” Public Information Notice 02/94, August 29, 2002 (available on the Internet at

    Other Resources Citing This Publication