Abstract

The central role of private capital flows in capital account crises fundamentally altered the nature of the policy response. Macroeconomic and structural policies needed to be directed at restoring confidence and stemming capital outflows. For fiscal policy, this meant adjusting public sector imbalances when they were the root of the vulnerability; for monetary policy, it meant maintaining interest rates sufficiently high to compensate investors for higher perceived risk; for structural policies, it meant underpinning the fiscal adjustment where appropriate, and addressing weaknesses in the banking and corporate sectors.

The central role of private capital flows in capital account crises fundamentally altered the nature of the policy response. Macroeconomic and structural policies needed to be directed at restoring confidence and stemming capital outflows. For fiscal policy, this meant adjusting public sector imbalances when they were the root of the vulnerability; for monetary policy, it meant maintaining interest rates sufficiently high to compensate investors for higher perceived risk; for structural policies, it meant underpinning the fiscal adjustment where appropriate, and addressing weaknesses in the banking and corporate sectors.

The extent of the macroeconomic disruption triggered by the crisis clearly depended on the magnitude of the shift in the capital account. Therefore, as previously pointed out, the immediate objective of the programs was to restore confidence, halt the hemorrhage from the capital account, and ensure orderly adjustment in the current account. To achieve this, the underlying vulnerabilities that had led to the crisis needed to be addressed. Given the different sources of these vulnerabilities, this implied that the focus of the policy packages differed across countries.

In some countries, such as Brazil, Mexico, and Turkey, dealing with the fragilities in public sector finances was the most pressing issue, while in others, notably in Asia, structural reforms to address the fragilities in the financial sector were required. In all countries, monetary policy needed to be sufficiently restrained to ensure, at a minimum, that domestic monetary conditions did not give impetus to additional capital outflows and put further pressure on the exchange rate. (All but two of the countries, Argentina and Brazil, had floating exchange rates by the time the programs were adopted.) Moreover, maintaining monetary stability and preventing a sustained pickup of inflation in the wake of the large exchange rate depreciations that had already occurred was essential for the restoration of macroeconomic stability as well as market confidence. Finally, fiscal policy, besides addressing the fragilities in public sector finances, had a potential role to play in the external adjustment process by contributing to the increase in net domestic saving necessitated by the shift in the capital account.

This chapter discusses the three elements of the policy programs: fiscal policy, monetary and exchange rate policy, and structural policies. It examines how their respective roles were defined in the original programs, how they were adapted as the crisis evolved and the macroeconomic outlook changed, and how they affected the evolution of the crisis. The chapter draws some tentative conclusions about the roles of the different components of the policy programs in capital account crises.

Fiscal Policy

The crises presented fiscal policy with a twin challenge. On the one hand, to the extent that countries’ vulnerabilities were rooted in the public sector, addressing these vulnerabilities was essential to creating a basis for confidence to return. At the same time, the public sector could contribute directly to the current account adjustment, easing the burden on private sector savings. On the other hand, a more relaxed fiscal stance could help offset weakening economic activity. Moreover, to the extent that such weakening activity was itself of concern to investors, too tight a fiscal stance might erode, rather than enhance, confidence. Thus, the two goals—dealing with underlying fragilities and facilitating macroeconomic adjustment—did not necessarily always point in the same direction. The potential conflict faced by fiscal policy was compounded by the uncertainty under which the original macroeconomic frameworks were formulated, given the inherent risks of the financing strategies. This section examines how fiscal policy approached these challenges.

Fiscal Adjustment in the Original Programs

Without exception, the original programs envisaged a strengthening of fiscal balances relative to the previous year. Both the rationale and the magnitudes of the planned adjustment varied across countries. Programmed improvements in overall fiscal balances52 ranged from 1–1½ percent of GDP in Mexico, Argentina, and Korea, to 2–½ percent of GDP in Philippines and Thailand. 3½ percent in Brazil, and over 5 percent of GDP in Turkey (Table 5.1).53 In Turkey, this was expected to be achieved through a large improvement in the primary balances, whereas in Brazil the primary balance was expected to strengthen less—by some 2½ percent of GDP—as lower interest rates were assumed to reduce financing cost.54 The implied fiscal measures were typically larger than the planned changes in overall balances as all countries, with the exception of the Philippines, were expected to experience some decline in output growth, albeit much less than the decline that eventually materialized.

Table 5.1.

Evolution of Fiscal Performance Criteria and Indicative Targets1

(As percent of GDP)

article image
Source: IMF staff reports.

In some cases, the precise definitions of performance criteria differed in coverage from the concept labeled “performance criterion” here.

Estimate as of time of original program approval.

For Argentina (1995), “original program” refers to the Ninth Review of the Extended Fund Facility (EFF) approved in March 1992.

Concept used for calculation of fiscal impulse measure, below.

Fourth-Fifth-Sixth Review refers to Original Program, Second, and Third Reviews under the EFF approved in August 1998.

There was also a performance criterion on banking system credit to the public sector and an indicative target on central government expenditure.

For Philippines (1997), “original program” refers to the Fourth Review of the EFF approved in June 1994; December targets were indicative.

A minus sign indicates a deficit.

First and Second Reviews were combined.

Performance criterion on overall public sector borrowing requirement and indicative target on primary balance of federal government, replaced by performance criterion on primary balance of consolidated public sector and indicative target on net debt of the consolidated public sector.

With starting conditions and definitions of fiscal positions varying across countries, the planned fiscal adjustments are difficult to compare. An alternative, and more meaningful, approach is to assess the programmed adjustment in each country against a well defined yardstick. A useful metric for such an assessment is the medium-term sustainability of the public sector. In general, gauging medium-term sustainability is not straightforward, but one operationally useful yardstick is the balance needed to stabilize the ratio of public debt to GDP.55 Naturally, the level at which the debt ratio is being stabilized is also an important consideration. In some of the countries reviewed here, the initial level of public debt was already high, so a reduction in the debt ratio may have been desirable to provide flexibility against future shocks, or to free up future budgetary resources from interest payments. In other countries, the initial level of debt was sufficiently low that stabilizing the debt ratio may have been less important.

Starting from the public sector’s flow budget constraint, and assuming a unit elastic demand for base money, yields the primary surplus required to stabilize the debt-to-GDP ratio at a given inflation rate.56 Table 5.2 reports the minimum required surplus on the eve of the programs, assuming historical growth, real interest, and inflation rates (and excluding financial sector restructuring costs).57 It bears emphasizing that these estimates embody a number of underlying assumptions and thus should be taken as broad indications rather than precise parameters. Moreover, the reported primary balances are those required to maintain a constant ratio of public debt to GDP over the medium term. Such surpluses are unlikely to stabilize the debt-to-GDP ratio in the short run if real interest rates are higher and real growth rates lower than the assumptions underlying the estimates—a situation typically encountered during financial crises.58 More generally, the “required” primary balance should be viewed as referring to the structural primary balance—that is, abstracting from cyclical movements of the budget balance.

Table 5.2.

Medium-Term Fiscal Sustainability

(As a percent of GDP, unless otherwise specified)

article image
Sources: IMF, World Economic Outlook; database and IMF staff estimates.

Primary balance required to stabilize ratio of public debt-to-GDP, estimate for year t as of year t–1; excludes financial sector restructuring costs.

Year t–1 debt data exclude financial sector restructuring bonds and quasi-fiscal losses.

Year t.

Includes only actual or imputed carry costs of financial sector restructuring.

Date refers to year t, not to program approval year.

The estimates suggest that all programs, with the exception of Brazil, initially envisaged primary surpluses that exceeded the balances required for medium-term debt stability. In Brazil, the targeted primary surplus was almost identical to the “required” surplus; in fact, the Brazil program explicitly targeted a fiscal balance consistent with medium-term stability. In the other countries, there were several reasons for targeting larger fiscal adjustment than suggested by the criterion of stabilizing the debt-to-GDP ratio over the medium term.

In Turkey, the program explicitly aimed at a reduction of the public debt ratio to reduce the vulnerability of the fiscal position to interest rate shocks and a funding crisis. The government’s decision to switch to money financing as it encountered increasing funding difficulties and rising real interest rates had been a trigger of the crisis.59 The program targeted a lowering of the public debt ratio by 5 percentage points (from 45 percent of GDP at end-1994) which, together with a more ambitious inflation target, would have necessitated the programmed primary surplus of 4 percent of GDP.60

Likewise, in the Philippines, there was a longer-term strategy of reducing public debt, which had exceeded 85 percent of GDP in 1993, and had been steadily declining to 60 percent of GDP by end 1996. Moreover, with the depreciation of the exchange rate after the floating of the peso in July 1997, debt servicing costs increased61 and there were doubts about the government’s ability to access international capital markets to refinance its maturing external obligations in the unsettled environment caused by the Thai financial crisis. Both factors argued for stronger fiscal adjustment than medium-term stability of the public debt ratio would have suggested.

Acute financing problems were the main reason for targeting substantial fiscal adjustment in Mexico and Argentina. Medium-term public debt sustainability was not a concern, but governments in both countries faced a short-term funding crisis as they were unable to refinance their maturing external obligations. In addition, in Mexico, fiscal plans sought to accommodate the carrying cost of financial sector restructuring as financial institutions were encountering growing difficulties in the wake of the crisis and the floating of the exchange rate.62

The prospective cost of financial sector restructuring also argued for stronger fiscal positions in Indonesia, Korea, and Thailand, Even after taking into account initial estimates of the carrying cost of financial sector restructuring, however, the targeted primary balances exceeded the balances required to stabilize the public debt-to-GDP ratio by 3 percentage points of GDP in Indonesia and by about one percentage point of GDP in Korea and Thailand. In Indonesia, some additional margin—albeit probably less than 3 percentage points of GDP—seemed consistent with the government’s policy in the pre-crisis years of running overall surpluses to lower the debt-to-GDP ratio, which was comparable to those in Argentina and Mexico and likely to rise as a result of the fiscal burden of financial sector restructuring. In Korea and Thailand, by contrast, public debt was very low and was viewed as likely to remain in a moderate range, even after the burden of financial sector restructuring was taken into account. In Korea, the objective was to keep the overall balance, including the estimated carrying cost of financial sector restructuring, at roughly the same level as that achieved in 1995–96; in Thailand, the program sought to reverse the weakening in the fiscal position that had occurred in the previous year. In terms of the underlying economic arguments, however, it is clear that the main rationale for fiscal adjustment in the Asian programs lay elsewhere.

Confidence Effects of Fiscal Adjustment

In all of the programs, the impact of fiscal adjustment plans on market confidence was an important consideration. Fiscal consolidation was expected to reassure markets. This seemed a reasonable assumption where fiscal issues were a problem and vulnerabilities rooted in public sector finances had contributed to the emergence of the crisis—either because public debt dynamics raised concerns about sustainability, as in Brazil and Turkey, or because the composition of public debt made the government vulnerable to a funding crisis, as in Argentina, Mexico, and the Philippines.63 It was also reasonable to assume that markets would seek assurances that fiscal issues would not become a problem as the costs of the crisis in the financial sector were surfacing.64 This suggested that fiscal plans in the countries facing serious problems in the financial sector—Indonesia, Korea, Mexico, and Thailand—needed to allow for the carry cost of financial sector restructuring, even though initial estimates of these costs would inevitably be rather crude.

The confidence argument was explicitly articulated in the Mexico program, which noted that “… these steps [among others, consolidation of public finances]… would help bolster foreign investors’ confidence…” and was echoed in several subsequent programs—most notably for Argentina, Thailand, Korea, and Indonesia. However, while there are strong reasons arguing for fiscal adjustment to address current or prospective vulnerabilities in public sector finances, it is less clear whether fiscal consolidation beyond what is required to achieve this helps strengthen confidence.

Did fiscal adjustment in fact bolster market confidence? This question is difficult to answer because confidence cannot be measured directly and changes in confidence, however defined, cannot necessarily be attributed to a specific policy, but reflect the impact of the whole policy and financing package. Bearing in mind these caveats, the behavior of private capital flows can be considered a useful indicator of market confidence in a capital account crisis and there appears to be some episodic evidence suggesting that fiscal policy did affect confidence in those countries where fiscal issues were widely viewed as the main source of vulnerabilities. This is particularly borne out by the experience of Brazil, where news about weaknesses in the provincial finances raised concerns about the viability of the program and prompted further capital outflows, which led to the abandonment of the exchange rate peg. Following the devaluation, and a more resolute fiscal stance, capital outflows were stemmed and a massive swing in the current account was avoided. Likewise, in Turkey, fiscal adjustment under the program appears to have been instrumental in restoring confidence in capital markets, with capital flows reversing soon after the adoption of the program.

There are also indications that determined efforts to strengthen the fiscal position helped restore confidence in Argentina and Mexico (although, of course, many other factors were simultaneously at play). In response to the Mexican financial crisis, Argentina experienced large withdrawals of bank deposits and sizable capital outflows during the first quarter of 1995. At the same time, it became increasingly apparent that, in the absence of policy changes, the overall fiscal deficit would widen significantly. Concerns about the size of the prospective deficit were underscored when the government had to cancel plans to issue Arg$2 billion of treasury bills as interest rates jumped sharply at initial auctions. The IMF-supported program approved in March 1995 aimed at achieving a fiscal surplus of 0.7 percent of GDP. which, together with official financing, would have been sufficient to amortize public sector debt falling due and fund two Trust Funds for financial system restructuring without recourse to the international capital market. The program was largely successful in restoring confidence. Capital flows were reversed relatively quickly and the government was able to tap the market sooner than expected. In the event, instead of having to generate a surplus of 0.7 percent of GDP, the program was able to finance a deficit of about 1.4 percent of GDP.

In Mexico, the link between fiscal policy and the restoration of market confidence was less clear. The original program, which envisaged an improvement in the overall fiscal balance of 1 percentage point of GDP, failed to restore market confidence and capital outflows continued in the first quarter of 1995. With capital outflows significantly larger than projected, at the first review the target for the primary balance was raised by a further 1 percent of GDP to offset higher interest payments and keep the overall deficit target unchanged. At the same lime, monetary policy was lightened and the official financing package was strengthened. These measures stemmed private capital outflows. Clearly, the reversal of capital flows in the next quarter cannot be attributed to the tightening of the fiscal target alone. In particular, the clarification of the financing arrangement appears to have played an important role. Nevertheless, given Mexico’s vulnerabilities, it is questionable whether an augmentation in the financing package in the absence of credible fiscal adjustment would have been sufficient to turn around capital flows.

Why did fiscal adjustment appear to help restore confidence in Brazil, Turkey, Argentina, and Mexico? In Brazil and Turkey, unsustainable public debt dynamics were widely perceived as the countries’ main vulnerability, calling for a significant reduction of the fiscal deficit. In Argentina and Mexico, two factors may have been at play. First, even though the fiscal deficits in both countries were modest, the governments were facing, or were close to facing, a funding crisis, with potential difficulties in rolling over maturing obligations. As such, fiscal adjustment would reduce the vulnerability of the government to a funding crisis, and its announcement could be expected to bolster confidence. Second, both Argentina and Mexico, despite the modest imbalances at the time of the programs, had a long history of fiscal deficits and high inflation. Accordingly, markets may have welcomed the reassurance that these countries were not reverting to the inflationary policies of the past.

In Asia, the experience was quite different. In all three countries, the programs failed to restore market confidence and private capital outflows continued for over a year after the onset of the crisis. In Indonesia, Korea, and Thailand, neither the initial tightening of fiscal policy nor the subsequent easing had a perceptible influence on market confidence. These countries’ vulnerabilities were viewed as rooted primarily in the private sector and fiscal policy did not have much of a role to play in addressing them, other than ensuring that the fiscal cost of financial sector restructuring would be dealt with in a responsible manner. Fiscal policy neither faced unsustainable debt dynamics, as in Brazil and Turkey, nor an imminent funding crisis, as in Argentina and Mexico. Against this backdrop, the negative market reaction to the January 1998 draft budget in Indonesia, which failed to secure the originally programmed fiscal adjustment, probably needs to be seen as an expression of concern that support for the whole program was lacking rather than a verdict about the importance of fiscal adjustment in Indonesia.

In addition to the effect of the fiscal stance itself, the composition of adjustment and the accompanying structural measures can play a critical role in underpinning the sustainability of fiscal adjustment and thereby contributing to a restoration of confidence.

Fiscal Policy, External Adjustment, and Output Declines

Confidence effects aside, there is another, quite distinct, argument for undertaking fiscal adjustment in the face of a capital account crisis. For a given capital outflow, the larger the increase in public saving, the smaller is the increase in net private saving that is required to achieve the necessary adjustment in the current account. Fiscal tightening can thus ease the burden of adjustment falling on the private sector. Although the argument is, by definition, correct in terms of private saving, it does not necessarily follow that greater public sector adjustment eases the burden of adjustment in the private sector in terms of absorption. Fiscal adjustment, by reducing aggregate demand, can cause output to slow or even contract. Thus, while greater public sector adjustment reduces the required increase in private saving, it may do so at the cost of a loss of output and private absorption.

The need for an increase in public saving to assist the external adjustment process evidently depends on the position of the economy at the beginning of the crisis—whether it is close to full employment or well below—and how supply and private demand are expected to respond to the crisis. As noted in the previous section, the original programs typically assumed a moderate adjustment in the current account accompanied by some slowing of output growth. The sharp declines in output that occurred were not foreseen. Moreover, in most countries, with the possible exception of Brazil, it seemed plausible to assume that output on the eve of the crisis was close to potential. In Argentina, Mexico, the Philippines, and Turkey, growth had remained strong or accelerated in the year preceding the crisis. In Indonesia, Korea, and Thailand, growth was already slowing, but these countries had experienced several years of strong expansions. Against this background, it was not unreasonable to argue that an increase in public saving will reduce the burden of adjustment on the private sector, a view that was expressed in several program documents.

From the point of view of the original macroeconomic frameworks, the two principal goals of fiscal policy—addressing weaknesses in public finances to restore market confidence and contributing to macroeconomic adjustment—coincided. On the basis of the original projections, all programs, with the exception of the Philippines, envisaged a withdrawal of fiscal stimulus, ranging from half a percent of GDP in Indonesia to over 8 percent of GDP in Turkey (Table 5.3).65

Table 5.3.

Fiscal Balances and Fiscal Impulse Ratios: Programs versus Outcomes

article image
Sources: IMF, World Economic Outlook; IMF, MONA database; and IMF staff estimates.

Dates refer to year t (or closest fiscal year), rather than to program approval year.

The picture changed as it became increasingly evident that output was declining sharply, reflecting both adjustments on the supply side and a contraction of private demand. Given the prospect of a significant contraction of output, adhering to the original fiscal targets would have implied a more substantial withdrawal of stimulus, exacerbating the economic downturn. A tension between the two goals of fiscal policy was emerging.

The response to this tension differed. In countries where fragilities in public sector finances were widely viewed as the core of their vulnerabilities, the need to address these fragilities generally took precedence over concerns about the contraction of economic activity. These countries tried to adhere to, or even strengthen, their original fiscal targets. As noted above, in Mexico, the planned primary surplus was raised by 1 percentage point of GDP at the time of the first program review to achieve the original target for the overall balance in the face of higher interest cost (Table 5.1). At the same time, growth projections were reduced by 3½ percentage points.

Similarly, in Turkey the original fiscal target was maintained as growth projections were revised downward. In the event, the target was achieved with a smaller adjustment in the primary balance than envisaged in the initial program, as interest rates came down sharply. In Brazil, with growth turning out somewhat stronger than originally projected, unanticipated output shocks were less of an issue.66 The target for the primary surplus was raised at the first review to offset part of a sharp increase in interest cost, which nevertheless implied a significantly larger overall deficit than envisaged in the original program.67

In Argentina and the Asian program countries, in contrast, the unexpected contraction of output prompted a revision of the original fiscal targets. In Argentina, this was facilitated by a swift recovery of confidence in response to the initial adjustment measures: by the time of the first review under the extension of the EFF arrangement (September 1995), half of the deposits withdrawn during the spring crisis had returned. By contrast, in Asia, capital outflows were not quickly reversed and the deepening recession was increasingly seen as contributing to weak confidence. Since public finances were not perceived as a major source of vulnerability in these countries, there were no strong reasons to adhere to the original fiscal plans in a macroeconomic environment that was radically different from that envisaged in the initial programs. Beginning in early 1998, fiscal targets were thus eased successively as growth projections became more pessimistic.

As a result of the changing macroeconomic outlook and the ensuing modifications of fiscal targets, fiscal outcomes and the impact of fiscal policy on the adjustment process turned out significantly different from what had been originally envisaged. In Mexico, the fiscal adjustment that was undertaken entailed a withdrawal of stimulus well in excess of what had been planned; in Turkey, the contractionary impact of fiscal policy was also stronger than originally programmed. In all other countries, the withdrawal of fiscal stimulus was either smaller than initially planned (Argentina, Indonesia, and Thailand),68 or fiscal policy provided a positive stimulus (Brazil, Korea, and the Philippines), 69

This suggests that in most crisis countries (notably in Asia), fiscal policy did not contribute to the output declines, and, in fact, offset part of it—albeit a relatively small part. Nevertheless, the question remains whether the initial fiscal tightening contributed to the negative output dynamics. Figure 5.1 graphs quarterly fiscal impulses and seasonally adjusted quarterly real GDP growth during the first two program years when the largest output declines occurred. In most cases, the decline in real GDP precedes the negative fiscal impulse. Indeed, rather than tight fiscal policy causing the output decline, the output decline “drives” the measured negative impulse. In Korea, the fiscal impulse was actually strongly positive in the last quarter of 1997, when real GDP growth (on a seasonally adjusted basis) was already negative, 70 Similarly, in Indonesia, the fiscal impulse is strongly positive in the last quarter of 1997 and the first quarter of 1998, before turning negative as discretionary expansionary measures did not keep up with the output decline. Similar patterns may be seen in Thailand and Mexico. Only in Turkey does the strongly negative fiscal impulse in the second quarter of 1994 coincide with the sharp output decline.71 Sample correlations (during the crisis periods) between fiscal impulses and changes in output growth range from 0.82 and 0.55 in Turkey and Mexico, respectively, to 0.11 in Indonesia and Korea, and –0.16 and –0.30 in Thailand and the Philippines.

Figure 5.1.
Figure 5.1.

Quarterly Fiscal Impulses and Real GDP Growth

(Seasonally adjusted, in percent)

Source: IMF staff estimates.

The Role of Fiscal Adjustment in Capital Account Crises

Although the fiscal plans in the original programs bore certain similarities, the outcomes in the first, full program year—the critical year in the short-term adjustment process—differed significantly. Two factors appear to account for the divergence: the importance of fiscal consolidation for the restoration of confidence and the unpredictability of the macroeconomic adjustment process triggered by the crisis. Fiscal adjustment was sustained where it was needed to alleviate vulnerabilities in public sector finances. In these countries, the underlying need for fiscal consolidation took precedence over concerns about its impact on aggregate demand in the face of an unexpectedly large decline in output and. indeed, appears to have contributed to the restoration of confidence. In contrast, in countries where fiscal problems were not seen as a factor contributing to the emergence of the crisis, fiscal targets were revised substantially as concerns about the impact of tighter budgets on aggregate demand ultimately prevailed.

The experience suggests two lessons regarding the role of fiscal policy in capital account crises. First, fiscal consolidation is critical for the restoration of confidence when fragilities in public sector finances are perceived as a source of a country’s vulnerability. But conversely, it seems doubtful that fiscal adjustment as such—when it is not motivated by underlying weaknesses—helps strengthen confidence. A more tricky case concerns the possibility of prospective fiscal costs, such as those associated with bank recapitalization costs (beyond initial estimates of the carry costs). This may call, not so much for a short-term fiscal squeeze in the midst of a crisis, but rather for an articulation of a clear set of policy rules that will restore a viable debt path over the medium term, and whose credibility might be buttressed by conditionality.72

Second, the extraordinarily complex dynamics of supply and demand shocks—and uncertainty about the private sector’s response more generally—means that, beyond confidence effects, the scope for short-term fine tuning may be limited. Thus, to the extent that fiscal adjustment is not dictated by medium-term considerations, it may be preferable to aim for fiscal targets that are broadly neutral from a cyclical point of view and allow automatic stabilizers to work if macroeconomic developments differ from the original program framework.

Aside from the aggregate fiscal stance, the composition of the targets may play an important role in mitigating the effects of the crisis, in particular by expanding and strengthening social safety nets (see Box 5.1).

Monetary and Exchange Rate Policy

Monetary policy had a critical role to play in restoring confidence and stabilizing capital flows. First, monetary policy had a direct bearing on the incentives for investors to hold domestic currency assets and could thus help mitigate the adverse shift in the capital account. Second, by ensuring nominal stability and preventing a sustained increase in inflation that would eventually feed into further exchange rate depreciation, monetary policy could make an important contribution to orderly macro-economic adjustment, thereby helping to restore confidence. The latter role is particularly salient in countries with a history of high inflation, but it is nonetheless important in countries with no preexisting inflation problem, given the magnitude of the shocks involved.

The importance of a credible monetary policy stance was underscored by the fact that five of the eight countries considered had been forced to abandon their exchange rate pegs after the first waves of capital outflows and had thus lost previous nominal anchors. Brazil initially sought to maintain the peg but had to follow the same route soon after the program was approved, then establishing inflation targets as a new anchor. Only Argentina managed to maintain its exchange rate peg in the context of the existing currency board arrangement. Against this background, the programs invariably placed strong emphasis on maintaining a monetary policy stance that would keep inflation under control.

The two main objectives of monetary policy—ensuring appropriate returns on domestic currency assets to limit capital outflows and preventing a spiral of inflation and depreciation—called for monetary restraint. There were also important considerations arguing against excessive monetary contraction, however. Pursuing the two main objectives in a single-minded manner and allowing interest rates to rise to arbitrarily high levels risked worsening public debt dynamics (in Turkey and Latin America). Another risk of aggressive tightening was that it could aggravate weaknesses in the financial sector and highly leveraged firms, especially in Asia, where the corporate sector was highly leveraged with excessive reliance on short-term debt (Table 5.4). Indeed, “second-generation” models of balance of payments crises suggest that a perceived reluctance on the part of the authorities to raise interest rates (for these or other reasons) may have triggered the crises in the first place, as markets challenged the parity in a self-fulfilling speculative attack.73

Table 5.4.

Firm-Level Risk Measures: Country Medians, 1995–96

article image
Source: Claessens and others, 1999.

Excluding crisis countries.

Social Safety Nets

The recent crises triggered by large capital outflows underscored the need to cushion the impact of adverse economic developments. Sound macroeconomic and structural policies may create foundations of long-term sustainable growth, but they may also hurt some of the poor in the short-term. Examples are the removal of generalized price subsidies, an exchange rate devaluation that increases the price of tradable goods, the reduction of budgetary support to state enterprises, and so on. Most of the countries in our sample recorded increases in poverty, loss of physical assets among the poor, rising rates of malnutrition, and a shrinking middle class.

To mitigate the adverse short-term effects on the vulnerable groups, most IMF-supported programs incorporate outlays on temporary social safety nets or increased allocation to existing social programs to transfer income or protect consumption (Chu and Gupta, 1998). Safety nets—in addition to their redistributive effects—can help to garner the necessary political support for the adjustment and structural reforms.1 Traditionally, safety nets included subsidies or cash compensations targeting particular social groups, improved distribution channels for or temporary price controls on essential commodities, severance pay and retraining for redundant public sector employees, and employment through public works programs. Most programs in countries in our sample included one or several instruments outlined above. Although it would have been preferable to have these instruments in place before the crises erupted, some countries had to install or expand safety nets only during the crisis. The pre-crisis transfer programs were very small, particularly in Asia.

In Indonesia, no explicit safety net was in place prior to the crisis and generalized price subsidies were used instead. Although the initial cost of these subsidies was relatively modest (about 3 percent of GDP in 1997—98), following the sharp depreciation of the domestic currency, this cost rose to more than 4 percent in 1998—99, mostly on account of higher fuel prices. The generalized food and fuel subsidies were eventually replaced by a targeted subsidy on lower quality rice, and through a combination of public works programs and cash assistance, respectively. In addition, microcredit programs were initiated and the government launched a “stay-in-school” campaign providing block grants to schools to help fund lunches for eligible children.

Korea had the most developed social safety net system in Asia, but the system had to be expanded substantially to cope with the crisis. The relatively small pre-crisis Livelihood Protection Program (LPP) was expanded to include cash benefits (up to $70 per month), tuition fee waivers, lunch subsidies for students, and a 50 percent reduction in family medical insurance contributions. Eligibility was based on a minimum income and asset tests. In October 2000, the LPP was replaced with a program linking cash and in-kind transfer to participation in labor programs. The formal unemployment insurance scheme remains relatively small and the jobless have benefited more from the public works program. Social protection spending increased threefold from 1997 to 1999.

In Thailand, the safety net was decentralized, with a range of inadequately linked programs. It included, among others, cash transfers to families and the elderly poor, in-kind transfers (mostly subsidized health care), and public works programs (Tambon Development Program and Poverty Alleviation Program). Most of these programs were expanded in 1998–99, with funding coming both from domestic resources and donors (Social Investment Project). In 1998, the Labor Protection Act increased the minimum severance pay, extended some social security benefits to small firms, and added old-age and child allowance schemes.

In Mexico, elements of a social safety net were in place for a number of years, but very few of these specifically targeted the poor. Recently, the authorities began shifting the programs from pure income transfers to transfers conditional on human capital investment. Examples of such programs include PRO-GRESA. which is conditional on children’s school attendance and regular health care visits. In response to the 1994 crisis, a program of small-scale public works (PET) was instituted, providing up to three months of employment at 90 percent of minimum wage.

In Brazil, social security benefits and labor protection programs are the main safety nets. Although most programs are at the national level, some of them are run by local governments. Examples of the latter are Bolsa Escola, targeting cash transfers to low-income families on the condition of school attendance, and child labor eradication programs (PETI). As an innovation in IMF programs for Brazil, 22 core assistance programs, chosen in collaboration with the World Bank and the Inter-American Development Bank, have been protected from fiscal adjustment in 1999 and 2000.

In Argentina, the safety net comprises the public pension system, free education up to the secondary level, and a free public health system for the indigent and uninsured. The major programs targeting the poor and the unemployed include health care for the elderly (PAM1), housing programs (FONAVI), and a public works program (TRABAJAR). Similar to Brazil, social spending in the IMF-supported programs was protected in real terms.

1 Of course, the economic distortions that come with redistributive policies ought to be recognized in the design of safety nets.

However, in cases where (as in Asia) the balance sheets of financial institutions and corporations had substantial foreign currency exposures, allowing the exchange rate to depreciate further would entail a further deterioration in these balance sheets. Adverse effects on the public sector balance sheets would also occur in cases where (as in Brazil) the government had issued substantial foreign currency debt. Moreover, an accommodative monetary stance could have entailed a spiral of depreciation of the nominal exchange rate, as against a temporary spike in interest rates.

In practice, the choice between high interest rates and the exchange rate depreciation was one of degree: given the capital outflows from the economy and corresponding decrease in real savings, it would have been difficult, if not impossible, to avoid some combination of higher real interest rates and a more depreciated real exchange rate (Figure 5.2). Certain combinations of higher real interest rates and real exchange rate depreciations may be preferable to others, however, depending on the relative importance of vulnerabilities to exchange rate and interest rate risk. Still, assessing the tradeoffs among these combinations was often difficult: while information on the public sector’s exposure to these risks was largely available, that on the private sector’s exposure was much more limited. In these circumstances, to the extent that both exchange rate and interest rate exposure mattered, neither a staunch defense of the exchange rate at any cost nor a complete abandonment of the exchange rate was likely to be optimal. In most countries, monetary policy thus tried to strike a balance between exchange rate and interest rate adjustment.

Was monetary policy under the programs too tight or too loose in light of these considerations? In order to address this question, three issues are addressed in the following sections: How was monetary policy implemented? How successful was it in achieving its main objectives—stabilizing capital flows and controlling inflation? And how large was the negative impact of monetary tightening on output?

Figure 5.2.
Figure 5.2.

Changes in Real Interest Rates and Real Exchange Rates

Sources: IMF, International Notice System; and IMF staff estimates.1 Average real effective exchange rate during 12 months following program approval minus average real effective exchange rate during 12 months prior to program approval (percent change).2 Average real overnight interest rate during 12 months following program approval minus average real overnight interest rate during 12 months prior to program approval. Real interest rates based on estimates of contemporaneous inflation, a three-month moving average of monthly consumer price index changes (previous month, current month, and one month ahead, as available).

Monetary Policy Implementation Under the Programs

To monitor monetary policy implementation, all of the programs—with the exception of Indonesia and the Philippines—included the traditional quantitative performance criteria: a floor on net international reserves (NIR) and a ceiling on net domestic assets (NDA) of the central bank.74 Implementation of the monetary programs was mixed, however. In most of the cases—with the notable exceptions of Indonesia, Mexico, and Turkey (Table 5.5)—the program ceilings were respected, 75 but these quantitative ceilings provide only a rough check to help ensure that reserves are not squandered in futile sterilized foreign exchange market interventions and to prevent monetary expansion from fueling a sharp acceleration of inflation. Particularly in the tumultuous environment prevailing in financial markets in these countries at the height of the crisis, these quantitative limits were insufficient to guide the response of monetary policy to changing market conditions on a day-to-day basis. Monitoring of monetary policy under the programs therefore needed to rely on indicators that were observable at high frequencies and bore a direct relationship to market conditions.

Table 5.5.

Monetary Conditionality1

article image
article image

Initial performance criteria or indicative targets related to monetary sector or net international reserves.

The performance criterion also required no new exposure in forex futures or forward markets. Fourth Review introduced consultation mechanism on inflation.

Performance criteria introduced at first review.

LOI of December 24, 1997, required that call money rates be raised to 30 percent, or above if needed, to stabilize the exchange rate and sterilize activated amounts of 11.3 trillion won worth of liquidity support to keep overall liquidity sufficiently tight to maintain interest rates at adequate levels.

LOI of First Quarterly Review (Feb-98) required that “Monetary Policy will be conducted flexibly with the aim of maintaining stability in the foreign exchange market… call rates will be cautiously allowed to ease, in line with continued exchange rate stabilization. A further lowering of interest rates is envisaged only after the foreign exchange market has durably stabilized.”

Targets and criteria for March 1998 and June 1998 were set during the First Quarterly Review.

December 1997 figures pertain to the authorities’ indicative targets.

The LOI required that “monetary policy will be geared to preventing the nominal exchange rate from depreciating beyond the following path.”

In this context, interest rates came to play an increasingly important—if often nonspecific 76—role in program monitoring. The use of nominal interest rates for program monitoring has well known limitations, however, particularly in cases where rising nominal rates primarily reflect exchange rate premia or of inflationary expectations rather than monetary tightening. In Indonesia, for instance, at end-1997 nominal interest rates rose to 60 percent—by far the highest in the region—at a time when the money supply was expanding at a monthly rate of 30 percent. Thus, while the increased focus on interest rate developments helped guide monetary policy with regard to the programs’ external objectives, they did not provide an alternative to the nominal exchange rate anchor that most of the countries had lost when they abandoned their exchange rate pegs. Accordingly, quantitative limits may still have served as an important check on monetary expansion.

Inflation Control

How successful was monetary policy in keeping inflation under control? In general, remarkably so given the magnitude of the shocks experienced.

Inflation in the crisis countries averaged some 13 percent per year in the year prior to the crisis, rising to 30 percent in the year of the crisis, before declining to 20 percent by the following year (Table 5.6). This inflation performance is all the more remarkable given the very sharp nominal exchange rate movements that occurred (Figure 2.1).

Table 5.6.

Inflation Program Projections and Outcomes

(Period average, in percent)

article image
Sources: IMF, World Economic Outlook; MONA database; and IMF staff estimates.

Date refers to year t, not to year of program approval.

These averages, however, mask some important differences among the programs. In general, inflation was in line with what might be expected, given monetary and exchange rate developments, with some exceptions. In Argentina, the only program here in which the fixed exchange rate (currency board) was maintained, inflation actually declined slightly in 1995, while in Korea and Thailand, despite the large nominal depreciation, inflation rose only modestly, from about 5 percent per year in 1997, to 8 percent in 1998, before declining to below 1 percent by 1999. Turkey made little headway in reducing inflation. In Brazil, where inflation stabilization had succeeded only four years prior to the program after numerous unsuccessful attempts, there was a serious concern that the currency devaluation might reignite inflationary expectations. In the event, the move to an inflation targeting framework to reestablish a nominal anchor helped to underscore the authorities’ commitment to keeping inflation under control, and inflation rose by less than 1 percentage point (to about 4.6 percent). Mexico saw large increases in inflation, partly reflecting exchange rate pass-through, but also, as noted above, overruns in monetary policy, while Indonesia provides a vivid illustration of how a previously stable economy can lurch into a vicious cycle of inflation, depreciation, and loss of monetary control.

Expected Returns and Capital Outflows

To what extent did monetary policy succeed in stabilizing capital outflows and nominal exchange rates? Although the view that monetary restraint is needed to counter an adverse shift in the capital account has long been widely accepted, it has recently been challenged in the context of the three programs in Indonesia, Korea, and Thailand. Some commentators have argued that higher domestic interest rates, far from reversing capital flows, had a perverse effect in these countries, exacerbating the capital outflows and resulting in further exchange rate depreciation. In this view, higher interest rates, by bankrupting domestic firms, resulted in a widening of the country (default) risk premium.77 This criticism has stimulated an extensive debate on the relationship between interest rate increases and capital flows and the exchange rate.

As noted in Chapter II. the dynamics of the capital flows differed markedly between the Latin American countries (especially Argentina and Brazil) and Turkey, on the one hand, and the East Asian countries (especially Thailand, Indonesia, and Korea), on the other. Whereas capital outflows were relatively short-lived in the Latin American programs (typically lasting one quarter following the onset of the currency crisis), they were more prolonged (and larger in magnitude) in East Asia. Can this contrasting experience be related to the monetary policy stance?

A precise correlation between capital flows and domestic interest rates is difficult to establish—not least because other factors, such as fiscal policy, political developments, and the credibility of the programs, were at play. Moreover, the response of capital flows to interest rate differentials may be expected to depend on the nature of those flows. To the extent that capital flows take the form of foreign direct investment or equity flows, they may be relatively unresponsive to higher interest rates.78 Likewise, if outflows consist mostly of the cutting of banks’ foreign currency lines of credit, higher domestic interest rates may be of limited use and, as discussed in Chapter III, more direct forms of private sector involvement may be called for.

Although there may be limitations to using higher interest rates to attract capital (depending, among other things, on the nature of the capital flows), at least some components of capital flows should be responsive to interest rate differentials. In this regard, Figure 5.3, which compares private capital flows and ex post monthly dollar rates of return on domestic assets relative to the London interbank offered rate (LIBOR), is revealing. In Indonesia, the Philippines, and Thailand, these interest differentials were negative for several quarters during the period mid-1997 to mid-1998. Put in this context, capital outflows of the magnitude and duration experienced in these countries were scarcely surprising: investors were not even being compensated for currency depreciation, let alone the risk of holding assets in a financial system that was in the midst of a crisis. By contrast, in Argentina and in Turkey, interest rate differentials were generally positive and capital outflows less persistent than in Asia. Comparing ex-ante dollar rates of return (based on surveys of exchange rate expectations), it is noteworthy that dollar interest rates (relative to LIBOR) were often negative—especially in Korea (fourth quarter of 1999) and Thailand (third quarter of 1997)—whereas they were consistently positive in Brazil (Figure 5.4).

Figure 5.3.
Figure 5.3.

Private Capital Flows and Ex Post Dollar Rates of Return1

Sources: IMF, World Economic Outlook; and IMF staff estimates.1 Capital flows are in percent of GDP. Ex post uncovered interest differentials in U.S. dollar terms relative to the London interbank offered rate (LIBOR) in percent per month.
Figure 5.4.
Figure 5.4.

Private Capital Flows and Ex Ante Dollar Rates of Return1

Sources: IMF, World Economic Outlook; and IMF staff estimates.1 Capital flows are in percent of GDP. Ex ante uncovered interest differentials in U.S. dollar terms relative to the London interbank offered rate (LIBOR) in percent per month.

The available empirical evidence on the relationship between ex post relative returns and private capital flows in the eight countries is clearly somewhat tenuous. In a number of instances, ex post positive returns on domestic assets were not sufficient to prevent capital outflows.79 Nevertheless, there were few episodes where negative interest differentials actually coincided with capital inflows, suggesting that a lax monetary policy stance is unlikely to be helpful in stabilizing capital flows.

Some commentators, particularly in the Asian context, have turned this argument on its head, suggesting that higher interest rates—by leading to widespread bankruptcies—actually resulted in further capital outflows and exchange rate depreciation. Existing empirical evidence does not give much support for this view (Box 5.2). In fact, the spread between onshore U.S. dollar and rupiah interest rates in Indonesia suggests that high domestic interest rates reflected expected exchange rate depreciation rather than a higher default risk premium, as the critics of a tight monetary policy would argue.80 If higher interest rates were indeed contributing to domestic bankruptcies and, therefore, reflected greater bankruptcy risk, then both local and foreign currency interest rates should rise in tandem. In fact, however, onshore U.S. dollar interest rates in Indonesia rose only slightly. Dollar interest rates did increase to about 15 percent in April-June 1998 at the height of the political and security concerns, but even then, the increase in rupiah interest rates was significantly higher.

To summarize, in capital account crises, monetary policy—in the form of higher expected rates of return—can play a crucial role in stemming capital outflows and thus avoiding massive swings in the current account (with the attendant consequences for economic activity). In Argentina (and, to a lesser extent, the other Latin American countries), such interest rate increases were indeed successful in stemming capital outflows. In Asia, there is little evidence of a perverse effect (whereby higher interest rates contributed to capital outflows and greater exchange rate depreciation), and relative U.S. dollar rates of return on domestic assets were not positive for several months after the onset of the crises.

Impact on Economic Activity

Tightening monetary policy in the midst of a crisis, even to the extent that it was necessary to stabilize capital flows and keep inflation under control is likely to have had a negative impact on economic activity. This impact, of course, needs to be compared with the relevant counterfactual: if failing to tighten adequately had resulted in sustained large capital outflows and a vicious circle of exchange rate depreciation and inflation, these developments likely would have had more devastating consequences. Nevertheless, it is legitimate to ask whether monetary and credit conditions contributed significantly to the observed output declines.

Figure 5.5 graphs nominal and real (overnight and lending) interest rates.81 Although some caution is required in comparing the levels of interest rates across countries (particularly lending rates, where definitions and applicability differ widely), judging by the developments in real lending rates in the first year after program approval, credit conditions do not appear to have been excessively tight. In Indonesia, real lending rates were strongly negative for most of the year; in Brazil, they declined steadily after a moderate initial increase and fell well below the pre-crisis 1997 average from mid-1999 onward. In Korea, Mexico, and Thailand, real lending rates at first turned negative as inflation rose in the wake of the large initial exchange rate depreciations. In Korea, this movement was followed by a sharp but relatively brief rise in real interest rates and then a significant decline from mid-1998 onward. In Thailand, in contrast, the rise in real lending rates was more gradual and sustained. Real lending rates also rose in Mexico after the initial decline. In all three countries, as well as the Philippines, real lending rates in the first year after program approval were, on average, barely higher than in the preceding year. By contrast, in Argentina real lending rates rose sharply at the beginning of the program and stayed well above pre-program levels throughout the year.

Figure 5.5.
Figure 5.5.

Nominal and Real Overnight and Lending Rates

Sources: Data provided by the national authorities; IMF, International Financial Statistics; and IMF staff estimates.1 Nominal overnight rate in March 1994 was 3,119 percent per year.

In the Asian countries and Mexico, real interest rates were not closely correlated with the periods of the steepest output declines in late 1997 and early 1998: indeed, these declines occurred well before the increase in real interest rates. If light monetary conditions resulted in increased quantity rationing, however, developments in real lending rates may not tell the full story. There is evidence that relatively low (or negative) real interest rates in these countries were associated with credit rationing (see Box 5.3). This appears to have reflected mainly a lack of sufficient bank capital and an unwillingness to lend in a period of heightened uncertainty, rather than a lack of sufficient liquidity, however. As such, there may have been little scope to address this problem through expansionary monetary policy.

The Interest Rate-Exchange Rate Nexus in Currency Crises: A Review of the Literature

A number of recent studies have tried to assess empirically whether higher interest rates are useful in supporting the exchange rate (that is, the “traditional” effect) or whether they instead have an opposite, “perverse” effect. Rather than examining the long-run relationship between monetary policy and the exchange rate, these studies focus on patterns inside selected short episodes.

The results of these studies are inconclusive and indeed quite mixed. In general, they fail to find overwhelming evidence of the traditional effect. This is not surprising, however, given the inherent policy endogeneity problem: that is, interest rates are likely to be raised precisely during episodes of currency depreciation, as both variables respond to shifts in market sentiment. On the other hand, there is also no clear pattern of evidence across studies of a perverse effect of interest rate policy.

Furman and Stiglitz (1998) identify a set of 13 episodes, in nine emerging markets, of “temporarily high” interest rates (episodes during which interest rates rose by more than 10 percentage points for at least five days, then fell back). Using a simple regression analysis, they find that both the magnitude and duration of such interest rate hikes are associated with exchange rate depreciation. While Furman and Stiglitz note that this evidence is not definitive, and that its interpretation is fraught with difficulties concerning endogeneity, they conclude that it at least questions the usefulness of raising interest rates.

Kraay (2000) focuses instead on episodes of speculative attacks on currencies and uses a more sophisticated and complex methodology. He identifies a set of 121 attacks that were successful, in the sense that there was an uncharacteristically large monthly depreciation; he also identifies (with greater inherent difficulty) a set of 192 unsuccessful attacks. The essential finding is that increases in central bank discount rates are neither necessary nor sufficient for staving off a speculative attack. Indeed, no relationship is found between central bank discount policy and the success or failure of speculative attacks. When Kraay tries to control for the endogeneity of interest rate policy, the results are similar, although, as he notes, they are preliminary and could reflect the difficulty of specifying appropriate instrumental variables to control for policy endogeneity.

Goldfajn and Gupta (1999) ask a somewhat different question, one probably more relevant for the East Asian countries during their IMF-supported programs. They consider cases following an exchange rate crisis in which the real exchange rate has become clearly undervalued, so that considerable real appreciation is likely to follow. They then study whether tighter monetary policies—in terms of higher-than-average real interest rates—are associated with the corrective real appreciation occurring mainly through currency appreciation rather than through higher inflation.

In general, Goldfajn and Gupta find that tight monetary policy does raise the probability of “success” that is, achieving the corrective real appreciation via currency appreciation. When the sample is restricted to cases where the banking sector is fragile, however, tight monetary policy seems to reduce the probability of success (although as the authors note, this latter result is based on very few eases and is not robust).

Goldfajn and Baig (1998), rather than defining and identifying crisis episodes from a broad sample of countries, focus on the very recent experience of five Asian countries, from mid-1997 through May 1998. Using daily data, they analyze the relationship between nominal interest rates and nominal exchange rates during the recent Asian crisis. A vector autoregression does not find a significant relationship—in either direction—for any of the five Asian countries. On the other hand, a panel regression using changes in interest rates and exchange rates yields a traditionally-signed coefficient over all the sample spans examined, although this is statistically significant only in some subperiods. Country-by-country regressions find a significant traditionally-signed coefficient in some periods for Indonesia, Korea, and the Philippines (the only significant coefficient with the opposite sign is found for Malaysia, and this is in one sub-period only). Goldfajn and Baig thus conclude that their study finds no evidence that higher interest rates lead to weaker exchange rates; if anything, there are periods where higher rates lead to stronger exchange rates.

Gould and Kamin (2000) examine the link between exchange rates and interest rates using weekly data. They attempt to control explicitly for the endogeneity of interest rates by entering variables proxying for the risk premia in the regression. The paper finds no effect of interest rates on exchange rates in either direction, which the authors suggest may be due to the difficulty of identifying such an impact in weekly data, especially in a short sample.

Tanner (2000) examines the relationship between domestic credit growth and exchange market pressure (i.e., the sum of exchange rate depreciation and reserve outflows scaled by base money). Using data for Brazil, Chile, Mexico, Indonesia, Korea, and Thailand, he finds that contractionary monetary policy tends to reduce exchange market pressure, consistent with the traditional view.

Basurto and Ghosh (2001) argue that the effect of an interest rate increase on the exchange rate should depend on whether the increase is perceived as temporary, and on how the decline in the interest rate is expected to be reversed (through an increase in the money supply or through lower inflation reversing the decline of real money balances). They also argue that, in terms of monetary aggregates, monetary policy was not especially tight in East Asia, and that high nominal interest rates, particularly in Indonesia, were a reflection of inflationary expectations rather than of tight monetary policy. Using an explicit monetary model of exchange rate determination to control for monetary policy, therefore, they examine whether higher real interest rates contributed to a widening of the risk premium (perhaps through the expectation of bankruptcies). With the exception of a brief episode in early 1998 in Korea, they find no such evidence.

For most countries, developments in real credit aggregates 82 tell a broadly similar story to real lending rates (Figure 5.6).83 In Indonesia and. to a lesser extent, in the Philippines, real credit rose sharply after the programs were approved and began to decline only in the second half of the first program year. In Brazil and Thailand, real credit increased initially as well, but these increases were considerably smaller, of shorter duration, and were followed by steady decline; in Korea and Mexico, real credit fell steadily throughout the first program year although the decline in Mexico was much steeper than in Korea. Real credit growth during the first program year was positive only in Argentina.

Figure 5.6.
Figure 5.6.

Broad Money and Banking System Credit in Real Terms1

Sources: Data provided by the national authorities; IMF, International Financial Statistics; and IMF staff estimates.1 Broad money and domestic credit at actual exchange rates, deflated by the consumer price index. Index, January 1994= 100 for Argentina, Mexico, and Turkey; index, January 1998= 100 for Brazil; index, December 1996= 100 for Indonesia, Korea, the Philippines, and Thailand. For Indonesia, since March 1999, the domestic credit data reflect transfers of nonperforming loans to Indonesia Bank Restructuring Agency (IBRA).

There appears to be a somewhat better correspondence between real GDP growth and real money and credit aggregates than with real interest rates. Appendix IV reports impulse response functions of real GDP growth (seasonally adjusted) to shocks in real money and real credit growth (seasonally adjusted), respectively, from a fourth-order vector auto regression. 84 The estimates suggest that the positive effects of an expansion in real money or real credit is generally quite short-lived, lasting a quarter or two, and the impact multiplier is no greater than 0.3–0.4, and generally lower.

Applying these estimated elasticities out of the sample to the non-Asian crisis countries suggests that the observed declines in real GDP growth were broadly commensurate with the tightening monetary and credit conditions.85

In Asia, by contrast, the declines of real GDP growth in the first two quarters of 1998 generally exceeded the fall in real money or real credit growth rates (with the implied “elasticity” well above the estimated 0.3–0.4).86 This suggests that either tightening monetary conditions had an unusually large (but difficult to estimate) effect during the crisis in Asia or, more simply, that monetary developments themselves can account for only a small fraction of the observed output declines. Moreover, declines in real money or real credit generally led, or were contemporaneous with, declines in real GDP in Turkey and Latin America; in Asia, by contrast (especially in Indonesia, Thailand, and the Philippines), the sharp decline in real GDP growth led declines in real money or real credit (Figure 5.7).

Figure 5.7.
Figure 5.7.

Real GDP, Real Credit, and Real Money

(Quarterly growth rates, seasonally adjusted, in percent per quarter)

Source: IMF staff estimates.

To summarize, in a capital account crisis, monetary policy must focus on stemming capital outflows and preventing spiraling exchange rate depreciation and inflation. This was done relatively more successfully in Latin America than in East Asia, perhaps because the Latin American authorities were less reluctant to raise interest rates more aggressively. Moreover, given capital outflows, some combination of higher real interest rates (or credit rationing) and real exchange rate depreciation was unavoidable; there is little evidence to suggest that tighter monetary policy itself was responsible for the observed output declines in Asia.

Credit Markets and Quantity Rationing in the Asian Crisis Countries

There has been considerable debate on whether a credit crunch developed in Asia in the aftermath of the crises.1 Ghosh and Ghosh (2000) note that, while real lending rates were quite low, and even negative, in the Asian countries at the onset of the crises, credit conditions may nonetheless have been tight because of quantity rationing.

In order to capture such rationing, Ghosh and Ghosh apply an explicit disequilibrium model to estimate the supply of, and demand for, real credit during the crisis period. The demand for credit is assumed to depend on the real lending rate, and a vector of indicators intended to capture actual and expected economic activity (including real GDP growth, the moving-volatility of growth, the stock market index, and inflation). Credit supply is assumed to be a function of the real spread and real GDP growth (reflecting banks’ willingness to lend) and banks’ lending capacity. Lending capacity, in turn, depends on banks’ liquidity (total liabilities minus reserves minus cash-in-vault) and their capital capacity (the maximum amount of loans that can be supported with existing capital while respecting capital adequacy requirements).

The estimates suggest that tightening credit conditions at the onset of the crises resulted in quantity rationing to the tune of 10–20 percent (i.e., estimated real credit demand was some 10–20 percent higher than estimated real credit supply). Thereafter, however, the decline in economic activity and weakening aggregate demand resulted in a fall in credit demand as well. By the beginning of the second quarter of 1998, credit supply was no longer the binding constraint and there was no more quantity rationing (real interest rates had also risen, especially in Korea). (As emphasized in the paper, individual firms (especially small and medium-sized enterprises) may have continued to face credit rationing.)

Decomposing the factors behind the quantity rationing “credit crunch,” Ghosh and Ghosh find that about 10 percent of the decline in estimated credit supply is accounted for by the banks’ “willingness to lend” variables, and the rest by a fall in hanks’ lending capacity. The reduction in lending capacity, moreover, largely reflected a lack of adequate capital rather than insufficient liquidity (although, in Indonesia, there was a decline in liquidity for a period as well). As such, it is unlikely that further injections of central bank liquidity to the banking system would have done much to ameliorate the credit crunch.

The analysis suggests an important crisis dynamic which is consistent with the results reported above on aggregate supply and demand shocks. At the onset of the Asian crises, weakness in the banking system resulted in credit rationing, contributing to the supply shock. With the subsequent downturn in aggregate demand (and real interest rates now reflecting the withdrawal of real resources available to the economy), the demand for credit also fell, and credit supply was no longer the binding constraint. See the bar charts for Indonesia and Korea in the figure below.

uch05bx03fig01

Indonesia: Estimated Credit Demand and Supply

(In constant rupiah, 1995 = Rp 100)

uch05bx03fig02

Korea: Estimated Credit Demand and Supply

(In constant won, 1995 = 100)

1 The term “credit crunch” is often used broadly to mean a situation in which credit is “tight,”ather than necessarily quantity rationing. See Dollar and Hallward-Driemeier (1998), who use survey data on 1,200 Thai manufacturing firms in end-1997 and early 1998; Domac and Ferri (1998), who examine the relationship in Korea between increases in the spread between bank lending rates and treasury bond rates and industrial production; and Claessens, Djankov, and Ferri (1998), who assess the impact of the currency and interest rate shocks (between early 1997 and September 1998) on the liquidity and the solvency of a sample of East Asian firms.

Structural Policies

Structural policies also could play a major part in the response to capital account crises both by restoring confidence and by addressing underlying weaknesses. In the absence of credible measures to address underlying structural weaknesses, monetary and fiscal policies alone are unlikely to restore confidence and renew capital inflows. In fact, in the case of structural policies, the confidence effect is arguably more important, at least in the short run. Whereas monetary and fiscal policies have an immediate or medium-term impact on the economy, structural measures, by their very nature, normally require time to take effect. In the short run, therefore, their impact is largely limited to restoring confidence. This confidence effect puts structural reforms very much at the center of capital account crisis programs.87

Nonetheless, there are important trade-offs in determining the structural agenda in capital account crisis programs. On the one hand, reforms that merely “paper over” underlying weaknesses, or that are not perceived as credible, are unlikely to do much to engender confidence, 88 Indeed, as underscored by the experience with bank closures in Indonesia, attempts at reform that are half-hearted, or lack credibility more generally, may well exacerbate the loss of confidence and deepen the crisis. Moreover, there may be domestic political economy considerations that make it easier for the authorities to tackle difficult structural reforms in the aftermath of a crisis than in calmer times.89 On the other hand, a desire for comprehensiveness may dilute and undermine the reform effort, strain the authorities’ implementation capacity, and unnecessarily antagonize interest groups, thereby eroding public support for the program. In practice, the dividing line is not always clear-cut, and programs have no doubt erred on both sides—not tackling underlying weaknesses sufficiently aggressively in some cases, while trying to take on too many reforms at once in others.

Figure 5.8 compares the formal structural conditionality in capital account crisis programs to conditionality in other programs supported by stand-by or extended arrangements (with programs in transition economies shown separately), 90 Two features of the data are immediately apparent. First, with the exception of Indonesia, there was significantly less structural conditionality in capital account crisis programs than in other IMF-supported programs (particularly IMF-supported programs in the transition economies). Second, by and large, conditionality in capital account crisis programs was focused on the specific structural weaknesses of the countries.

Figure 5.8.
Figure 5.8.

Structural Conditionality1

(Structural measures per year of program)

Source: IMF, MONA database; IMF-supported country program documents.1 Performance criteria, structural benchmarks, and conditions for completion of review; three sectors of conditionality (public enterprise restructuring, social safety net, and agriculture) did not contain any measures in the capital account crisis countries. The programs in Argentina and Mexico contained no structural conditionality.2 First year of the three-year Stand-By Arrangement (SBA), which was subsequently replaced by an Extended Fund Facility (EFF).

Thus, in Turkey (1994), the core of the structural measures centered on improving the state of the public finances and the health of the banking sector. In addition to “traditional” tax and spending measures found in most IMF-supported programs, budgetary discipline was to be improved by an overhaul of the social security system, limiting transfers to state enterprises, and an ambitious privatization program (of which, however, only privatization was subject to formal conditionality). A blanket guarantee on all domestic and foreign currency deposits helped avert a full-scale run on the banking system, allowing the authorities to focus on improving supervision and regulation.91

In Brazil (1998), likewise, structural reforms centered on improving long-term public finances, especially those of the provinces. At the national level, the budgetary process was to be guided by the Fiscal Responsibility Act (a structural benchmark under the program), bolstered by the introduction of a nationwide value-added tax and, on the expenditure side, reforms of the social security system and social spending programs.

Mexico (1995) and Argentina (1995) differed from the other programs reviewed here, in that financial sector weaknesses were not viewed as being the cause of the crises but rather its side effect. Particularly in Mexico, the crisis was mostly macroeconomic in nature—an overvalued exchange rate and an unsustainable current account deficit—and the program, which did not include any structural conditionality, reflected this.92 Although the banking sector developed significant difficulties—with nonperforming loans growing by 50 percent during the period December 1994–March 1995–the financial sector crisis followed the currency crisis. Costs of financial sector restructuring in capital account crisis countries are reviewed in Box 5.4.

Brazil, Turkey, and Mexico all offer examples where, arguably, the implementation or design of the structural reforms did not go sufficiently far in addressing the underlying weaknesses. In Mexico, for instance, inclusion of financial sector reform measures under the program might have speeded up the reform process. In Brazil, the initial announcement of the program helped calm markets and stem capital outflows. Subsequently, however, as Congress failed to pass key legislative changes (including increases in social security contributions and several tax increases), and there were increasing doubts about provincial finances, confidence was eroded, and by early 1999 Brazil faced a full-blown capital account crisis.

Costs of Financial Sector Restructuring

Capital outflows put heavy burdens on domestic banking systems, especially when associated with currency depreciations. The deposit base shrinks as agents run away from domestic currencies and the loan portfolio deteriorates as a result of the economic contraction induced by the falling currency. Recapitalization costs of crisis-affected banks depend on deterioration of the loan portfolio, the speed of currency stabilization, and capital endowment and loan loss provisioning in the banking system.

In Argentina, Turkey, and Brazil the recapitalization costs were modest, owing to relatively sound banking systems prior to the crisis and a Fast return of confidence. In Argentina, the banks lost some one-fifth of their deposits and one-third of banks were taken over, forced to merge, or went bankrupt. The share of non-performing loans increased by 15 percent, but thanks to large efficiency gains during 1991–94, proper loan loss provisioning, and functioning banking supervision, the overall health of the banking system was sustained with little explicit cost. The situation in Turkey was similar: the fast return of currency stability and business confidence limited the cost of bank recapitalization, which was done mostly through interest rate spreads and high yields on government paper. The extent of the damage to bank portfolios was not clear from the official data, mostly because of underreporting of nonperforming loans. In Brazil, the return of confidence was also rapid and private banks were relatively unaffected. The damage to state-owned financial institutions was much larger, however, and the outstanding bonds financing their restructuring amounted to 6 percent of GDP in mid-2000.

In contrast, in the remaining countries, the recapitalization costs were much larger, owing to weak banking systems, lack of prudential regulation, and protracted periods of currency instability. In Mexico, the share of nonperforming loans more than doubled and the total cost of bank recapitalization was revised to about 15 percent of GDP. The damage to bank portfolios was even higher in the four Asian countries: the peak-crisis estimates of the share of nonperforming loans were twice as high as the early-crisis estimates in Korea, the Philippines, and Thailand, and three to four times as high in Indonesia (see the table). The estimates of recapitalization cost—with the exception of the Philippines—dwarf the Mexican case and the domestic budgets are likely to bear the brunt of those costs.

Asian Crisis Countries: Nonperforming Loans and Recapitalization Cost

(As a percent of total loans)

article image
Source: Berg (1999).

As a percent of GDR Recapitalization costs in Indonesia have been revised upward repeatedly. They are now estimated to be at around 47 percent of GDP, assuming that future asset recoveries approach 10 percent of GDP.

In Turkey, even though the announcement of the structural program served to stem capital outflows—and helped restore inflows—implementation of the structural reform agenda soon petered out, perpetuating structural weaknesses (especially in the fiscal accounts) that eventually necessitated another IMF-supported program. Likewise, in Mexico, it is at least debatable whether the program missed an important opportunity to undertake a comprehensive reform of the financial sector, which subsequently suffered from protracted difficulties, 93

The Asian programs provide a sharp contrast in terms of the scope and depth of the structural measures (Box 5.5), and indeed have been criticized on grounds that the structural reform agenda was overly ambitious.94

The programs in Thailand and Korea were similar in that most of the formal structural conditionality pertained to financial sector reforms (Figure 5.6). These reforms—which included measures for bank restructuring and resolution, improvements in financial supervision and prudential regulation, the establishment of frameworks for debt workouts, and other closely related measures—were clearly integral to addressing the underlying weakness that had caused the crises. As such, even if the precise pace, sequencing, and modalities of the reforms may be debated, it would be difficult to argue that the reforms themselves were unnecessary or unfocused.95 Beyond these formal conditions, however, both programs included a large number of ancillary measures. While many of these were supportive of the financial and corporate sector reform efforts (or were related to enhancing social safety nets), others—though worthwhile reforms in themselves—were of peripheral importance to the immediate crisis, 96

The Indonesia 1997 stand-by is exceptional among the programs reviewed here in terms of the number and the scope and of structural measures subject to some form of conditionality (Figure 5.8). One reason for this broad brush approach was that the original program was primarily intended as a precaution against contagion (rather than dealing with an immediate crisis), and sought to reform the financial sector and deregulate the economy more generally. Events soon overlook the program, however, and the state of the banking system deteriorated dramatically as markets concluded that the authorities were not genuinely committed to undertaking reforms, with widespread runs even on solvent banks amid growing political uncertainty and confusion about the coverage of deposit guarantees.97 Massive liquidity support had to be provided by the central bank to preserve the payments system, derailing monetary policy. Subsequently, the focus of the structural reform agenda had to be sharpened, concentrating on rehabilitating the core banking system, strengthening the social safety net, and restructuring corporate debt in the context of improved governance both of the corporate sector and of the country more generally.

Inevitably, the question arises whether the reform strategy, especially in the Asian countries, was appropriate. One view holds that it is wrong to start restructuring financial institutions and strengthening the regulatory framework in the midst of a crisis. Specifically, it is argued that the hasty closure of financial institutions weakens confidence, while tightening regulatory standards worsens the credit crunch. On this view, governments should have tried to buttress confidence through unconditional liquidity support, and should have started dealing with the underlying problems only once a degree of calm in the financial markets had been achieved. In fact, the introduction or enforcement of capital adequacy requirements (CARs) was phased in; in Korea and Thailand, 8 percent CAR standards were already in place, but loan classification and loan loss provisioning standards were relatively lax. Classification rules were tightened in two steps in Korea; while in Thailand, the implementation of stricter loan classification began only in mid-1998 (to be completed over the next year and half). In Indonesia, loan classification and provisioning was tightened in October 1998; at the same time, CARs were temporarily lowered. But there is still a question of whether a slower pace of reform would have been better.

Although it is difficult to prove the counterfactual, putting off financial restructuring until quieter limes was probably not a realistic option. In the absence of immediate actions to assess and address the weakness in the balance sheets of financial institutions, unconditional liquidity support or comprehensive guarantees would have created enormous moral hazard, and likely would have resulted in further deterioration in balance sheets. In the case of Mexico, it is at least plausible that an earlier resolution of the banking system problems would have lowered the ultimate cost of financial sector restructuring. Moreover, as the experience in Indonesia illustrates, massive liquidity support would have been difficult to sterilize, resulting in a loss of monetary control. Finally, given that at least the broad dimensions of the problems in the financial system were well known, the failure to act at an early stage would have raised doubts about the governments’ resolve to deal with the underlying problems—and this would likely have weakened confidence at home and abroad.

Structural Measures in IMF-Supported Programs in the Asian Crisis Countries

The Asian capital account crisis programs were notable for their much greater structural content than those in Latin America and Turkey.

Thailand

In Thailand, owing to weak supervision and prudential regulation, together with a sharp decline in asset prices, the share of nonperforming loans more than doubled between end-1996 and March 1997. The national authorities soon recognized that many, if not most. Thai finance companies were insolvent or at least severely undercapitalized. The reform strategy consisted of closing bankrupt institutions, issuing a blanket deposit guarantee to calm markets, and eventual restructuring and rehabilitating of Thai financial institutions with the establishment of the Financial Sector Restructuring Agency. Financial sector reform was the centerpiece of the IMF-supported program, with several structural performance criteria, including: preventing deposit interest rate competition by insolvent banks; provision for taking over non-viable institutions; introduction of new loan classification rules; and, subsequently, on closing insolvent banks, and fully recapitalizing undercapitalized institutions in the course of 1998.

By the Second Review of the IMF program, in early 1998, it was clear that this ambitious timetable was not achievable. In particular, financial institutions were having difficulty raising sufficient capital and the legal system was incapable of speedy treatment of bankrupt cases. The program was therefore broadened to include strengthening of the legal and judicial system, and two new performance criteria—on introduction of new loan classification and provisioning guidelines, and on the signature of Memoranda of Understandings with all financial institutions regarding their recapitalization plans.

Subsequent reviews resulted in further broadening of the structural reform agenda, particularly in corporate sector restructuring and privatization procedures, bank consolidation, and a variety of legal issues. The critical mass of measures was implemented by end-1998, and by early-1999, bolstered by the pick-up in economic activity, major banks had begun the process of their recapitalization. Moreover, the simplification of legal procedures allowed for fast progress in resolving corporate sector insolvencies.

Korea

In Korea, banks’ excessive exposure to the weakening chaebols (industrial conglomerates), their reliance on short-term foreign borrowing, and poor prudential regulation and supervision all made the economy vulnerable to a shift in sentiment following the Thai crisis.

The IMF-supported program, approved in late 1997, focused on financial sector restructuring, corporate governance, and capital account liberalization—as well as labor market reforms and trade liberalization. As in Thailand, it soon became clear that the original program underestimated the initial crisis and its implementation schedule was too ambitious. By the time of the First Quarterly Review, an agreement with foreign creditor banks had been reached, allowing the program to focus on addressing underlying structural weaknesses. Foremost was the need for bank restructuring and recapitalization, and the improvement of prudential supervision and regulation. The capital account was liberalized, mainly to remove distortions that had promoted short-term, unhedged foreign borrowing and to allow greater foreign participation in Korean financial markets.

Structural components of the program were further modified at the Second and Third Reviews, although their main thrust remained similar (the IMF paid special attention to the reform of commercial banks, while the World Bank focused on corporate sector restructuring). In light of the greater-than-expected economic downturn, the social safety net also received greater prominence. By early 1999, with the recovery well under way; short-terms structural measures—such as the first phase of hank and corporate sector restructuring—had been undertaken. A number of longer-term measures (including bank restructuring that required substantial financial injections, and institution-building) still remained outstanding, however.

Indonesia

In Indonesia, major structural weaknesses centered on excessive regulation in domestic trade, banks were saddled with nonperforming loans, and large, unhedged borrowing in foreign currency. As in other Asian countries, the growth of nonperforming loans was caused by rapid expansion of the financial sector, poor supervision and regulation, unrestricted “connected-lending,” and excessive property sector exposure. However, the extent of problem loans was difficult to assess because of non-transparent ownership and cross-holdings of equity and loans.

The original program, approved in early November 1997, and intended mainly as a precaution against contagion, sought to reform the financial sector and deregulate the economy more generally. Financial sector goals included transparent bank rehabilitation through the budget, revision of prudential regulations, and elimination of restrictions on bank lending. Deregulation of the economy was to be achieved by eliminating import and marketing monopolies and the expansion of activities open to foreign participation. To this end, a plethora of performance criteria and benchmarks were established, mostly dealing with state owned banks, tariff reduction, and increases in utility prices.

By the time of the First Review in April 1998, the state of the banking system had deteriorated dramatically, as markets concluded that the authorities were not genuinely committed to undertaking reforms, and confusion about the coverage of deposit insurance resulted in widespread rims even on solvent banks. A three-tiered approach to the banking crisis was adopted: blanket deposit insurance for two years; establishment of the Indonesian Bank Restructuring Agency; and the proposal of a framework for corporate restructuring. These steps, the thrust of which was basically unchanged from the original program, were detailed in a series of prior actions, performance criteria, and structural benchmarks. In addition, to help offset some of the price effects of the proposed structural reforms, the social safety net was strengthened.

Political uncertainty and program delays eroded confidence further: the exchange rate plummeted in mid-1998, partly because the announcement effects of the external debt restructuring under the Frankfurt agreement had short-lived benefits, and the crisis was reaching calamitous proportions. Although the longer-term goals of the program remained much the same, the immediate focus shifted to rehabilitating the core banking system and providing a social safety net.

The initial focus of the Extended Arrangement, approved in August 1998, was bank restructuring, social safety net issues, and corporate governance. By the time of the Second Review in October 1998, some progress in bank restructuring had been achieved, although estimates of the costs had risen to about 30 percent of GDP, while other reforms continued to lag.

By mid-1999, with a nascent recovery under way, program discussion centered almost exclusively on structural issues. Progress on recapitalization of private banks was encouraging, but losses of the state owned banks—accounting for more than three-quarters of all deposits—continued to mount. Measures in the critical areas of loan classification, asset recovery, and corporate restructuring were still overdue. In the ensuing months, progress was slow.

Philippines

Unlike the other Asian countries, the Philippines already had an IMF arrangement in place at the time of the crisis. While the Extended Arrangement, originally approved in 1994 as a precautionary program, included a number of structural measures, key vulnerabilities in the financial sector—excessive exposure to the property sector, weak prudential regulation, and reliance on short-term foreign currency borrowing—remained unresolved. Fortunately, in contrast to some other countries, the asset bubble was in a relatively early stage. The second, six-month extension of the EFF outlined two main measures for strengthening the banks: exposure to the real estate sector would be limited to 20 percent of banks’ portfolios, and the maximum value of real estate collateral was lowered from 70 percent to 60 percent. Foreign exchange exposure risk was limited by requiring that 30 percent of the value of foreign liabilities be kept in short-term liquid assets. The subsequent stand-by arrangement sought to strengthen financial sector supervision and prudential regulation.

In turn, financial sector restructuring and dealing with problem loans often required corresponding changes in the corporate sector and, in some instance, reforms of the labor market (Korea) and the legal system (Thailand). Nevertheless, in retrospect, it is certainly questionable that all of the measures included in the broad structural reform agendas were critical to the program. Striking the right balance is a difficult task, particularly in capital account crises where the rather nebulous concept of “restoring confidence” is so critical to the program’s success. The issue is being examined, however, in the context of the IMF’s review of structural conditionality. The main lesson seems to be that, like fiscal and monetary policies, structural reforms should be directed at specific weaknesses of the economy—and not simply at “restoring confidence” in general.

Cited By

  • View in gallery

    Quarterly Fiscal Impulses and Real GDP Growth

    (Seasonally adjusted, in percent)

  • View in gallery

    Changes in Real Interest Rates and Real Exchange Rates

  • View in gallery

    Private Capital Flows and Ex Post Dollar Rates of Return1

  • View in gallery

    Private Capital Flows and Ex Ante Dollar Rates of Return1

  • View in gallery

    Nominal and Real Overnight and Lending Rates

  • View in gallery

    Broad Money and Banking System Credit in Real Terms1

  • View in gallery

    Real GDP, Real Credit, and Real Money

    (Quarterly growth rates, seasonally adjusted, in percent per quarter)

  • View in gallery

    Indonesia: Estimated Credit Demand and Supply

    (In constant rupiah, 1995 = Rp 100)

  • View in gallery

    Korea: Estimated Credit Demand and Supply

    (In constant won, 1995 = 100)

  • View in gallery

    Structural Conditionality1

    (Structural measures per year of program)

  • Agénor, Pierre-Richard, C. John McDermott, and E. Murat Uçer, 1997, “Fiscal Imbalances, Capital Inflows, and the Real Exchange Rate: The Case of Turkey,European Economic Review, Vol. 41 (April), pp. 81925.

    • Search Google Scholar
    • Export Citation
  • Aghion, Philippe, Philippe Bacchetta, and Abhijit Banerjee, 2000, “A Simple Model of Monetary Policy and Currency Crises,European Economic Review, Vol. 44 (May), pp. 72838.

    • Search Google Scholar
    • Export Citation
  • Alesina, Alberto, and Allan Drazen, 1991, “Why Are Stabilizations Delayed?American Economic Review, Vol. 81 (December), pp. 117088.

    • Search Google Scholar
    • Export Citation
  • Basurto, Gabriela, and Atish R. Ghosh, 2000, “The Interest Rate-Exchange Rate Nexus in the Asian Crisis Countries,Staff Papers, Special Issue—IMF Annual Research Conference, International Monetary Fund, Vol. 47, pp. 99120.

    • Search Google Scholar
    • Export Citation
  • Berg, Andrew, 1999, “The Asia Crisis: Causes, Policy Responses, and Outcomes,IMF Working Paper 99/138 (Washington: International Monetary Fund).

    • Search Google Scholar
    • Export Citation
  • Berg, Andrew, Catherine Pattillo, 1999, “Are Currency Crises Predictable?: A Test,Staff Papers, International Monetary Fund, Vol. 46 (June), pp. 10738.

    • Search Google Scholar
    • Export Citation
  • Blanchard, Olivier J., and Danny Quah, 1989, “The Dynamic Effects of Aggregate Demand and Supply Disturbances,American Economic Review, Vol. 79 (September), pp. 65573.

    • Search Google Scholar
    • Export Citation
  • Boorman, Jack, Timothy Lane, Marianne Schulze-Ghattas, Ales Bulir, Atish R. Ghosh, Javier Hamann, Alex Mourmouras, and Steven Phillips, 2000, “Managing Financial Crisis: The Experience in East Asia,Carnegie–Rochester Conference Series on Public Policy, No. 53 (Amsterdam: North-Holland).

    • Search Google Scholar
    • Export Citation
  • Burnside, Craig, Martin S. Eichenbaum, and Sergio Rebelo, 1999, “Prospective Deficits and the Asian Currency Crisis,Policy Research Working Paper No. 2174 (Washington: World Bank).

    • Search Google Scholar
    • Export Citation
  • Calvo, Guillermo A., 1996, “Varieties of Capital–Market Crises” (unpublished; College Park: University of Maryland). Available via the Internet: http://www.bsos.umd.edu/econ/ciecrp.htm

    • Search Google Scholar
    • Export Citation
  • Calvo, Guillermo A., 1999, “Fixed versus Flexible Exchange Rates: Preliminaries of a Turn-of-Millennium Rematch” (unpublished; College Park: University of Maryland). Available via the Internet: http://www.bsos.umd.edu/econ/ciecrp.htm

    • Search Google Scholar
    • Export Citation
  • Calvo, Guillermo A.,, and Enrique G. Mendoza, 1996, “Mexico’s Balance–of–Payments Crisis: A Chronicle of a Death Foretold,Journal of International Economics, Vol. 41 (November), pp. 23564.

    • Search Google Scholar
    • Export Citation
  • Chalk, Nigel A., and Richard Hemming, 2000, “Assessing Fiscal Sustainability in Theory and Practice,IMF Working Paper 00/81 (Washington: International Monetary Fund).

    • Search Google Scholar
    • Export Citation
  • Chand, Sheetal K., 1992, “Fiscal Impulses and Their Fiscal Impact,IMF Working Paper 92/38 (Washington: International Monetary Fund).

    • Search Google Scholar
    • Export Citation
  • Chang, Roberto, and Andres Velasco, 1998, “Financial Fragility and the Exchange Rate Regime,NBER Working Paper No. 6469 (Cambridge, Massachusetts: National Bureau of Economic Research).

    • Search Google Scholar
    • Export Citation
  • Chu, Ke-Young, and Sanjeev Gupta, 1998, eds., Social Safety Nets: Issues and Recent Experiences (Washington: International Monetary Fund).

    • Search Google Scholar
    • Export Citation
  • Claessens, Stijn, Simeon Djankov, and Giovanni Ferri, 1998, “Corporate Distress in East Asia: Assessing the Impact of Exchange Rate and Interest Rate Shocks” (unpublished; Washington: World Bank).

    • Search Google Scholar
    • Export Citation
  • Claessens, Stijn, Simeon Djankov, and Tatiana Nenova, 2000, “Corporate Risk Around the World,Policy Research Working Paper No. 2271 (Washington: World Bank).

    • Search Google Scholar
    • Export Citation
  • Cline, William R., 2000, “The Role of the Private Sector in Resolving Financial Crises” (unpublished; Washington: Institute for International Finance).

    • Search Google Scholar
    • Export Citation
  • Corsetti, Giancarlo, Paulo Pesenti, and Nouriel Roubini, 1999, “What Caused the Asian Currency and Financial Crisis? Part II: The Policy Debate,Japan and the World Economy, Vol. 11, No. 3, pp. 30573.

    • Search Google Scholar
    • Export Citation
  • Dell’Ariccia, Giovanni, Isabel Gödde, and Jeromin Zettelmeyer, 2000, “Moral Hazard and International Crisis Lending: A Test,paper presented at the first IMF Research Conference, Washington, November.

    • Search Google Scholar
    • Export Citation
  • Dollar, David, and Mary Hallward-Driemeier, 2000, “Crisis, Adjustment, and Reform in Thailand’s Industrial Firms,World Bank Research Observer (International), Vol. 15 (February),No. 1, pp. 122.

    • Search Google Scholar
    • Export Citation
  • Domaç, Ilker, and Giovanni Ferri, 1998, “The Real Impact of Financial Shocks: Evidence from Korea,Policy Research Working Paper No. 2010 (Washington: World Bank).

    • Search Google Scholar
    • Export Citation
  • Dornbusch, Rudiger, and Alejandro Werner, 1994, “Mexico: Stabilization, Reform, and No Growth,Brookings Papers on Economic Activity: 1, Brookings Institution, pp. 253315.

    • Search Google Scholar
    • Export Citation
  • Drazen, Allan, and Vittorio Grilli, 1993, “Benefits of Crises for Economic Reforms,American Economic Review, Vol. 83 (June), pp. 598607.

    • Search Google Scholar
    • Export Citation
  • Edwards, Sebastian, 2000, “Contagion,The World Economy, Vol. 23 (July), pp. 873900.

  • Eichengreen, Barry, 1999, “Is Greater Private Sector Burden Sharing Impossible?Finance & Development, Vol. 36 (September), pp. 1619.

    • Search Google Scholar
    • Export Citation
  • Fane, George, and Ross H. McLeod, 1999, “Lessons for Monetary and Banking Policies from the 1997–98 Economic Crises in Indonesia and Thailand,Journal of Asian Economics, Vol. 10 (Fall), No. 3, pp. 39513.

    • Search Google Scholar
    • Export Citation
  • Feldstein, Martin, 1998, “Refocusing the IMF,Foreign Affairs, Vol. 77 (March–April), pp. 2033.

  • Feldstein, Martin, 1999, “A Self–Help Guide for Emerging Markets,Foreign Affairs, Vol. 78 (March–April), pp. 93109.

  • Flood, Robert P., and Nancy P. Marion, 1998, “Perspectives on Currency Crisis Literature,IMF Working Paper 98/130 (Washington: International Monetary Fund).

    • Search Google Scholar
    • Export Citation
  • Frankel, Jeffrey A., 1999, “The Balance Between Adjustment and Financing,paper presented at the International Monetary Fund seminar,Key Issues in Reform of the International Monetary System,Washington, May.

    • Search Google Scholar
    • Export Citation
  • Furman, Jason, and Joseph E. Stiglitz, 1998, “Economic Crises: Evidence and Insights from East Asia,Brookings Papers on Economic Activity: 2, Brookings Institution, pp. 135.

    • Search Google Scholar