Over the three years since the crisis broke out in 1997, the five Asian countries—Indonesia, Korea, Thailand, Malaysia, and Philippines—managed impressive recoveries. The recoveries were faster than expected by anyone. The economies started to bottom out in the second half of 1998. The rebound of growth in 1999 was no less drastic than its free-fall. In Korea, for example, the growth rates showed a turnaround from −6.7 percent in 1998 to 10.7 percent in 1999.

Yung Chul Park and Jong-Wha Lee*

Over the three years since the crisis broke out in 1997, the five Asian countries—Indonesia, Korea, Thailand, Malaysia, and Philippines—managed impressive recoveries. The recoveries were faster than expected by anyone. The economies started to bottom out in the second half of 1998. The rebound of growth in 1999 was no less drastic than its free-fall. In Korea, for example, the growth rates showed a turnaround from −6.7 percent in 1998 to 10.7 percent in 1999.

The purpose of this paper is to make an assessment of this speedy adjustment from the crisis in East Asia. In particular, we analyze the macroeconomic adjustment process of the East Asian currency crisis in a broad international perspective. First, we assess the impacts of the crisis on GDP growth using a cross-country data set, which compiled all currency crisis episodes over the period from 1970 to 1995. From these cross-country data, we draw some stylized facts about the adjustment of key macroeconomic variables during the crisis. Then we investigate the critical factors that determine the adjustment process.

Our analysis of cross-country patterns shows that GDP growth rates drop with the eruption of a crisis but then recover quickly to the pre-crisis level in two or three years, showing a V-pattern of adjustment. Thereafter, the GDP growth rates tend to rise slightly above the pre-crisis levels, but then subside back to a more sustainable level. We also compare the adjustment patterns of GDP growth rates between two subgroups of the currency crisis episodes—one with conditional financial assistance from the IMF and the other without. We find that the adjustment process was much sharper in the group with the IMF program, compared to those without. That is, in the IMF program countries, GDP growth rates start to fall precipitously even before the eruption of a crisis but then recover to its pre-crisis level more quickly in two years.

The macroeconomic adjustment process in East Asia is in general consistent with these stylized patterns. However, the degree of initial contraction and following recovery has been far greater in East Asia than what the crosscountry evidence predicts. This paper tries to make an evaluation of what factors contributed to the sharper contraction and the quicker recovery in East Asia compared with the cross-country patterns.

As discussed in the third section, we believe that a large number of internal and external factors are responsible for the deeper crisis and the quicker recovery in East Asia. The origin and the nature of the shock, initial conditions, the development of external environments, and the stabilization and structural adjustment policies taken must have a significant consequence on the adjustment path as they did in the eruption of the crisis. From cross-country regressions based on the sample of previous crisis episodes, we find that exchange rate depreciations, expansionary macroeconomic policies, and favorable global environments are the critical determinants of the post-crisis recovery. Financial assistance from the IMF is found to have no independent impact on the recovery process.

We find that the quick recoveries in East Asia have been largely driven by the accommodating macroeconomic policies, favorable external environments, and more export-oriented structure. After Korea, Malaysia, and Thailand shifted to a relaxation of monetary and fiscal policies by the second half of 1998, their economies took off. The sharp real currency depreciations must have had a bigger impact on more open Asian economies. Favorable external developments helped the quick improvement in East Asian exports. In this sense, the East Asian process of adjustment is not much different from the stylized pattern from the previous currency crisis episodes over the period from 1970 to 1995. However, the stylized pattern of adjustment cannot explain why the crisis was more severe and the recovery has been much faster than what was expected from the previous experiences of crisis. This paper argues that the sharper adjustment pattern in East Asia is attributed to the severe liquidity crisis that was triggered by investors’ panic and then amplified by weak corporate and bank balance sheets.

The stylized pattern of real GDP growth from the cross-country episodes displays that the crisis-hit countries can recover their pre-crisis or non-crisis average growth rate in three years after the crisis. Hence, it raises a question of whether the East Asian economies will be able to return to the pre-crisis trend rate of growth.

Although the financial crisis of 1997 abruptly brought a halt to Asia’s period of robust growth, there was little in Asia’s fundamentals that inevitably led to the crisis. This paper discusses the long-term prospects for growth in East Asia. From the cross-country regressions, we find that there is no evidence for a direct impact of a currency crisis on long-run growth. This suggests that with a return to the core policies that resulted in rapid growth, the East Asian economies can again return to sustained growth.

The paper is organized as follows. The next section discusses the methodology for our cross-country analysis and presents central features in the macroeconomic adjustments of the crisis-hit countries. Then, using regression analysis based on the cross-country data, we assess the factors that can explain the behavior of GDP growth rates during the crisis. The third section reviews the recent recoveries in East Asia and compares them with the stylized patterns from the cross-country analysis. We analyze the driving forces of the faster recovery in East Asia. The fourth section discusses the issue of the sustainability of the current recovery. Concluding remarks are found in a final section.

Cross-Country Patterns of Adjustment to Currency Crisis


In order to assess the post-crisis adjustment of the crisis-hit countries, one needs first to define a currency crisis. Several alternative indicators and methods have been used in the literature to identify the year when a crisis erupted in each country. Frankel and Rose (1996) and Milesi-Ferreti and Razin (1998) used the nominal depreciation rate of the currency. Sachs, Tornell, and Velasco (1996), Radelet and Sachs (1998), and Kaminsky and Reinhart (1999) combined the depreciation rate with other additional indicators such as losses in foreign reserves, increases in the interest rate, and reversals in capital accounts to identify the crisis.

Each definition has limitations. A large-scale depreciation can occur orderly without a speculative attack. Identifying unsuccessful speculative attacks is a difficult task. Reliable data on reserves and interest rates in developing countries are often unavailable. Reserves or interest rates can change irrespective of an attack. Lee and Rhee (2000) suggest an alternative measure based on the initiation of an IMF stabilization program. But, countries often receive an IMF program after a crisis breaks out or without a currency crisis. Governments may sign an IMF agreement not necessarily because they need foreign exchange, but because they want austerity conditions to be imposed (Przeworski and Vreeland, 2000).

Since the purpose of this paper is not to improve the measure of a currency crisis, we use the conventional nominal depreciation rate of the currency as a benchmark measure. But, in contrast to Frankel and Rose (1996), we use quarterly data, instead of annual data, to define a currency crisis. That is, based on quarterly data, a country is judged to have had a currency crisis in the year when it has a nominal depreciation of its currency of at least 25 percent in any quarter of the year and the depreciation exceeds the previous quarter’s change in the exchange rate by a margin of at least a 10 percent. Thus, our definition captures the incidences of currency crises that were severe but short-lived perhaps due to successful interventions in the foreign exchange market. During the period from 1970 to 1997, the total number of currency crises was 260. We use a window of plus/minus two years to identify an independent crisis. That is, if there were a precedent crisis within two years before a crisis, we count it as a consecutive crisis, but not an independent one. This procedure yields a total of 192 currency crisis episodes.1

Then, we divide all crisis episodes into two groups based on whether the crisis-hit countries entered into an IMF program or not. We have compiled data on all types of IMF programs that include stand-by arrangements, extended fund facility (EFF) arrangements, structural adjustment facility (SAF), and enhanced structural adjustment facility (ESAF) over the period from 1970 to 1997.2 The program is identified by the year when the loans are approved. Thus, if a country received financial assistance from the IMF during or one year after the currency crisis, we consider it as a currency crisis with an IMF program. Note that the decision on participation in the IMF program following a currency crisis can be determined endogenously by various factors. A country may enter into agreements with the IMF when it faces a more severe foreign reserve crisis or a worse macroeconomic situation (Conway, 1994). But, relying on the IMF conditionality may be just a way to impose domestically unpopular austerity policies (Przeworski and Vreeland, 2000).

Table 1 shows a summary of data on currency crises based on our definition during the period from 1970 to 1997. There were 192 currency crisis episodes during this period. The number of crises was increasing over time, from 40 in the 1970s, to 69 in the 1980s and 83 in the 1990-97. The number of countries that experienced at least one crisis was 99.3 Thus, on average each country had 1.86 crises over the period. Out of the 192 crisis episodes, 71 of them participated in an IMF program.

Table 1.

Incidence of Currency Crises and IMF Program Participation

article image
Notes: A currency crisis is defined to occur in the year when a country has a nominal depreciation of currency of at least 25 percent in any quarter of a year and the depreciation rate exceeds the previous quarter’s change in the exchange rate by a margin of at least a 10 percent. If the country under a currency crisis received financial assistance from the IMF during the year or one year after the currency crisis, it is classified as a currency crisis with the IMF program participation. Our sample does not include the former Soviet Union countries and counts only independent crises by imposing plus/minus 2 years window.

Macroeconomic Adjustment During the Currency Crisis

On the basis of the currency crisis index, we investigate how the crisis-hit economies, on average, behaved during the five years prior to and following the crisis. We first look at the movement of real GDP growth rates and then investigate the sources of output changes by looking at the movements of GDP expenditure components and major macroeconomic policy variables in the typical crisis-hit country during the period before and after the crisis. We also construct a control group of “tranquil” observations. If a country had not been subject to any crisis within a window of plus/minus two years surrounding a specific year, it is counted as a non-crisis country in that specific year. The behavior of the macroeconomic variables between the two subgroups—one with conditional financial assistance from the IMF and the other without—is also compared.

We use the data for the period from 1970 to 1995. Thus we attempt to draw the stylized pattern of macroeconomic adjustment from the crisis episodes that had occurred prior to the Asian crisis. There are 176 independent currency crises during this period, and in 64 episodes the countries participated in an IMF program.

Real GDP Growth

Figure 1 shows the movements of average GDP growth rates during the five years prior to and following the crisis; that is, from t−5 to t+5 where t is the year of a currency crisis. For comparison, we include a straight line, which indicates the average GDP growth rate during the tranquil period that did not experience a currency crisis or enter into an IMF program within a window of plus/minus two years.

Figure 1.
Figure 1.

Changes in GDP Growth Rates During the Currency Crises

In general, we find that the growth rates, on average, exhibit a V-type pattern of adjustment over the period before and following the crisis. The growth rates during the pre-crisis period from three to five years prior to the crisis are slightly lower than the average during the tranquil period of 3.5 percent. The growth rate continues to decline over time, from 2.7 percent in t−4 to 1.1 percent in t−1, implying that economic conditions are aggravated prior to the eruption of a crisis.

The growth rate increases slightly in the crisis year, which confirms that most currency crises have indeed been expansionary. As in Gupta et al. (2000), we also find that about 70 percent of the currency crises in our sample lead to an output increase in the crisis year. The average GDP growth rate of the crisis-hit countries remains at about 1.9 percent over the crisis year and one year after. But GDP growth recovers to its non-crisis level quickly in three years after the crisis, reaching 4.0 percent in t+3, that is about 0.5 percentage point higher than the average of the non-crisis economies. Thus, the growth rate tends to exceed its pre-crisis or tranquil period average, indicating that after a crisis the country’s level of GDP returns to the level of its pre-crisis growth path. Eventually, the growth rate tapers off and returns to the level of the tranquil period in four and five years after the crisis. This V-type pattern and the speed of recovery are broadly consistent with the findings in Hong and Tornell (1999) and Gupta et al. (2000).

Figure 1 compares the behavior of the GDP growth rates between the two subgroups- one with conditional financial assistance from the IMF and the other without. We find that the adjustment process shows a much sharper V-type in the program countries than in the non-program countries. The program countries start with lower growth rates of around 1.2 percent in t−4 and continue to slow down. They reach the trough, where the growth rate is -1.2 percent, in one year prior to the initiation of the currency crisis.

This magnitude of decline in growth rates is much larger than that of the non-program countries. At the trough, the growth rate of the crisis-hit program countries is about 4.7 percentage points lower than that of the non-crisis economies. Thereafter, rebounding from the deeper trough, the program countries show a quicker recovery. The GDP growth rate begins to recover from the crisis year and reaches its pre-crisis level quickly within two years after the onset of a crisis. The non-program countries also begin to recover a year after the crisis and then their growth rates stabilize at the non-crisis level from t+3.

The fact that the program countries have much lower growth rates than the non-program countries confirms that only a very serious macroeconomic situation forces a country to enter into agreements with the IMF. Nevertheless, it is intriguing that the crisis-hit countries show a quicker recovery from a deeper recession with the participation in the IMF program.

GDP Expenditure Components

Figure 2 shows the movements of the components of GDP expenditure during the five years prior to and following the crisis. Panel (a) shows that the share of private consumption expenditure in GDP remains stable over the period. In other words, consumption moves closely with GDP. The adjustment pattern is similar in both program and non-program countries. For the overall period, the consumption to GDP ratio in the crisis-hit economies exceeds the noncrisis tranquil period average, indicating that private consumption is high in the crisis-hit countries and even after a crisis these countries’ level of private saving does not increase to the level of the non-crisis countries.

Figure 2.
Figure 2.
Figure 2.

Changes in GDP Expenditure Components During the Currency Crises

Panel (b) of Figure 2 shows that in contrast to consumption, the investment (private plus public investment) rate shows more fluctuations. The level is no higher in the crisis-hit economies than in the non-crisis countries. For four to five years prior to the crisis, the investment ratio remains below the average level of tranquil observations of 22.8 percent. Thus, a stylized fact is that the crisis-hit countries have had ‘over-consumption’ but not necessarily ‘overinvestment,’compared to the level of the non-crisis countries. In the crisishit countries, the investment rate tends to decline during the pre-crisis period, reaching 19.8 percent in the crisis year. After the crisis, the investment rate increases gradually but does not return to the level of the pre-crisis or tranquil period, remaining at 20.9 percent for five years following the crisis. A popular claim regarding the role of the IMF conditionality is that the austerity program has an adverse effect on investment. Panel (b) of Figure 2 seems to support this claim. The IMF program countries have experienced a more severe investment contraction than the other group in the pre-crisis period, as the investment ratio declines continuously from 21.4 percent in t−5 to 18.9 percent in t. In the post-crisis period of the crisis-hit countries in which an IMF program is introduced, the investment rate does not recover to the pre-crisis level, remaining at 19.7 percent in t+4 and 20.1 percent in t+5. In contrast, the investment rate returns to the pre-crisis level in the non-program crisis-hit countries in five years after the crisis.

In the crisis-hit countries, domestic expenditure or demand is either slowly recovering or remains permanently below the pre-crisis level. In contrast, export demand shows a quick recovery during the post-crisis period. Panel (c) shows that in the crisis-hit countries, real export growth rates jump from less than 1 percent in t−1, to 3.0 percent in the crisis year and to 5.9 percent in t+1, and then remain at over 5 percent over the post-crisis period. For both program and non-program countries, export growth during the post-crisis period is faster than that of the pre-crisis or tranquil period, and thus leads a strong recovery. Consequently, as shown in Panel (d), after the currency crisis the export share increases permanently above the pre-crisis level. But, note that on average the export share in all crisis-hit countries is still lower than the non-crisis average.

During the early post-crisis period the quick recovery of export growth is accompanied by a contraction of import demand. The pattern of import reduction is more conspicuous in the program countries where import growth rates are negative in the pre-crisis period as well as the crisis year. Panels (e) and (f) of Figure 2 show that although the growth rate of imports recovers to the pre-crisis and non-crisis average in two years following the crisis, its share in GDP remains below the non-crisis average of 35.5 percent. The growth of exports and imports shows that the current account to GDP ratio improves quickly after the crisis. Thus, net exports tend to lead the recovery in the crisishit countries.

Macroeconomic Policy Indicators

Public consumption is an indicator of fiscal policy. Panels (a) and (b) of Figure 3 show that public consumption growth rates tend to slow slightly in the crisis year, and then recover to the pre-crisis as well as non-crisis average. But in the first year following the crisis, there is contrasting behavior between the program countries and non-program countries. While the public consumption growth rate is over 5.0 percent for the non-program countries, it is −0.8 percent for the program countries in the year of t+1. This confirms that an agreement with the IMF introduces a contractionary fiscal policy in the program country. Reflecting this sharp contraction in public consumption expenditure, the ratio of public consumption to GDP declines quickly in t+1 with the IMF program. The ratio remains at the level lower than the pre-crisis or non-crisis average in both program and non-program countries even five years after a crisis.

Figure 3.
Figure 3.
Figure 3.

Macroeconomic Policy Indicators During the Currency Crises

Like fiscal policy, monetary policy of the program countries contrasts sharply with that of the non-program countries. Panel (c) of Figure 3 shows that the real money supply growth rate remains positive throughout the years following the crisis and increases over time to return to the pre-crisis level in five years after the crisis in the non-program countries. In contrast, in the sample of the crisis-hit countries with IMF program participation, money supply growth is negative. Thereafter it returns to the pre-crisis average growth rate. The sharp reduction in money supply in the program countries implies that, as with fiscal policy, participation in an IMF program brings in tight monetary policy in the crisis-hit economy.

It is claimed that a currency crisis often develops into a banking crisis. As international lending declines suddenly, a weak banking sector is unable to play a proper intermediation role. Banks reduce the supply of credit to the private sector. Panel (d) shows that credit supply growth indeed slows down in the crisis-hit countries. For four to five years prior to the crisis, the real credit growth rate is 7.4 percent. Thereafter credit growth rates decline over time, reaching −1.6 percent in the crisis year. Even five years after the crisis, credit growth does not return to the level of the pre-crisis or tranquil period. The slow down of real credit growth is more pronounced in the IMF program countries. The supply of real credit declines by more than 8 percent in the year following the crisis and thereafter continues to slow down throughout the postcrisis period.

The robust growth of net exports during the post-crisis period is likely to be related to the real exchange rate depreciation associated with (or caused by) the currency crisis. Figure 4 shows that a currency crisis causes a sharp real depreciation of the exchange rate by about 15 percent in the crisis year. The real exchange rate also depreciates by 5.3 percent in the year following the crisis. Thereafter, it appreciates about 2 percent per year. Hence, the real exchange remains depreciated after the crisis. The pattern of adjustment in the real exchange rate is similar in both the program and non-program countries.

Figure 4.
Figure 4.

Change in Real Exchange Rate During the Currency Crises

Determinants of the Post-Crisis Recovery

We believe there are a large number of factors that determine the stylized pattern of adjustment in real output growth in the crisis-hit countries. Broadly speaking, there are four major factors that influence the adjustment pattern: (i) the origin and nature of the shock, (ii) initial conditions, (iii) domestic policies, and (iv) the external environments.

Origin and Nature of the Shock

The origin and nature of the shock that has provoked a crisis can influence the evolution of the crisis. Many currency crises can be attributed to macroeconomic mismanagement—large budget deficits and consequent monetary expansion in a fixed exchange rate regime—as in the Latin America debt crisis in the early 1980s. In this case, real depreciation of the currency and contraction of domestic absorption help to restore internal and external balance, leading to improvement in the economy.

Investors’ panic can intensify the effects of speculative attacks on currency. In particular, when the capital account is liberalized, a bad expectation by foreign investors can easily lead to a sudden reversal of foreign lending, thereby causing a significant contraction of the domestic economy. The adverse impact will be magnified if domestic corporations and financial institutions are heavily leveraged by large, unhedged, and short-term foreign currency debts. When a sharp and unexpected depreciation wreaks havoc with highly leveraged corporate and bank balance sheets, a sudden reversal of capital flows exacerbates the downturn in investment and the economy (Krugman, 1999; Aghion, Bacchetta, and Banerjee, 2000). But, once the investors’ panic calms down and foreign capital resumes to flow in, the economy can rebound quickly back to its long-term trend.

Initial Conditions

Differences in initial conditions could result in different patterns of adjustment. For example, structural variables such as per capita output and openness could be important in determining the pattern of post-crisis recovery.

The level of initial per capita GDP can influence the growth rate in the postcrisis period. In growth theory, a country with a lower initial per capita GDP is in a more favorable position for future growth. The fundamental idea is that the gap in existing capital and technology between the current and steadystate levels offers a chance for rapid “catching up”, via high rates of capital accumulation as well as the diffusion of technology from more technically advanced economies. In addition, when a currency crisis leads an economy to a lower level of per capita income relative to that of its own trend, the subsequent growth rate by which the economy rebounds to its potential would be higher.

Openness can also influence the speed and extent of the post-crisis recovery. When the economy is more export oriented, a quicker improvement in the current account follows a currency devaluation. Lee and Rhee (2000) argue that the quick recovery of the Korean economy may have been possible because of its openness and export orientation. An export oriented economy benefits more from devaluation after the crisis, and a subsequent improvement in the current account could in turn help restore foreign investors’ confidence and hence stability in the foreign exchange market.

Several studies also point out that the behavior of macroeconomic variables prior to the crisis can influence the degree of real output contraction. For example, a rapid expansion of bank credit or lending boom during the pre-crisis periods is critical to the post-crisis recovery (Sachs, Tornell, and Velasco, 1996; and Hong and Tornell, 1999). Gupta et al. (2000) find that the higher the size of short-term external debt and the amount of private capital flows are in the years prior to the crisis, the more severe is the contraction of output during the crisis-period.

Policy Factors

Macroeconomic and structural reform policies implemented by the government for crisis management can play a key role in the post-crisis adjustment of real output. Fiscal policy has a direct impact on domestic demand. Monetary policy plays a critical role in determining domestic consumption and investment.

In addition to the macroeconomic stabilization policies, structural reform programs can have significant effects on the adjustment path. It is often argued that structural reforms introduced by the IMF play a catalytic role in resuming foreign trade and private capital inflows, and thus contribute to the fast recovery of a crisis-hit economy as the commitment to the reform program improves foreign investors’ confidence in the economy. The critics of IMF programs, however, argue that the implementation of financial restructuring in conjunction with contractionary macroeconomic policies can make a credit crunch more severe than otherwise after the crisis.

For external demand, a larger depreciation of the exchange rate is expected to increase export earnings while cutting down import demand to improve the current account.

External Environment

A global economic environment is also critical to the post-crisis adjustment of crisis-hit countries. Business fluctuations of the world economy can influence post-crisis growth as they have a substantial impact on the terms of trade and export earnings of the crisis-hit country.

To the extent that the relevant data are available, we carry out an empirical assessment of the factors determining the pattern of post-crisis recovery. The explanatory variables that we consider to explain the speed and the extent of post-crisis recovery include per capita real GDP in the crisis year, world economic growth, which is an average of per capita GDP growth rates of a crisis-hit country’s trading partners weighted by its trade share, an interactive term of the real exchange rate depreciation with openness (trade-GDP ratio), real public consumption growth, and real money supply growth. We also include the investment rate.4

The regression also includes a dummy variable for the IMF program countries to see if participation in an IMF program had any impact on the recovery process. Upon entering an agreement with the IMF, a member government subscribes to the IMF conditionality which typically entails fiscal austerity, tight monetary policy, and currency devaluation. Since we include macroeconomic policies variables separately in the regression, the dummy variable may be able to capture the effect of the IMF program participation in postcrisis recovery.

We also control the differences in country-specific factors that may influence the potential growth path, by including the average growth rate for three to five years prior to the crisis. However, we do not include pre-crisis macroeconomic policy variables in the regressions, for the impact of these variables on the post-crisis recovery are extensively discussed in Hong and Tornell (1999) and Gupta et al. (2000). Also, we cannot incorporate any variables that measure structural vulnerabilities of the corporate and financial sectors due to the lack of broad cross-country data.

The dependent variable in the regression is the average growth rate of real GDP during the post-crisis period over k years.5


where GDPi, t+j is real GDP for country i in the j years after the crisis year (t) and N is the number of crisis episodes in our sample. Then, yi, t+k represents the real GDP growth rate, averaged over the post-crisis period of k years. Because we are mostly interested in short-term recovery, we choose k from 1 to 5. In the previous literature, k was often chosen arbitrarily, and thus crosssection data in which each country had only one observation was used for empirical investigation. Our framework differs significantly in that we use panel data. Thus, we utilize information in both cross-section and time dimensions. Our regression specification is as follows.


where x denotes the vector of the explanatory variables. Note that some independent variables such as real GDP in the crisis year, pre-crisis average growth rate, and an IMF program dummy are identical across all five equations. Fiscal policy variable is included as an average over the period from the crisis year t to the post-crisis year t+k, while monetary growth and real exchange depreciation variables are included as an average over the period from the crisis year t to the post-crisis t+k−1.

We estimate this system of the five equations by a seemingly unrelated regression (SUR) technique that corrects for heteroskedasticity in each equation and correlation of the errors across the equations.

Table 2 displays our estimates of the basic regression for post crisis recovery at various horizons, based on a total of 101 previous crisis episodes during the period from 1970 to 1995.

Table 2.

Determinants of the Pace of Recovery from the Currency Crises: A sample of 101 crisis episodes between 1970 and 1995

article image
Notes: Standard errors reported in parentheses. Levels of statistical significance indicated by asterisks: *95 percent; ** 90 percent. The system has 5 equations, where the dependent variables are the average real GDP growth rates over k years from the crisis year, t. The system is estimated by the seemingly unrelated regression (SUR) technique, which allows for different error variances in each equation and for correlation of these errors across equations. Each equation has a different constant term, which is not reported. An increase in real exchange rate indicates a real appreciation.

We find a strong and statistically significant negative relation between the initial real per capita GDP and the post-crisis growth rate at all horizons, implying that countries with lower per capita income tend to have larger increases in GDP growth over the period after the crisis. The impact of initial GDP on the post-crisis recovery is much larger in the year following the crisis, but then becomes smaller in the later years of the post-crisis period. The estimated coefficients imply that a 10 percentage point drop in per capita GDP in the crisis year is associated with a 0.2 percentage point (2.04*ln(0.9)) increase in GDP growth in the first year after a crisis erupted, but with a 0.1 percentage point on average over five years after the crisis.

The world growth variable also has a significantly positive coefficient in most of the regressions. The estimated coefficients imply that a one percentage point increase in world per capita GDP growth is associated with about 0.5 percentage point increase in GDP growth of the crisis-hit country in the postcrisis period.

The results also confirm the strong association between investment and GDP growth over the period of adjustment in the crisis-hit economies. The coefficients show that an increase of 10 percentage points in the ratio of investment to GDP is typically associated with an increase in the growth rate of about 1.3 percentage points per year.

Among the macroeconomic policy variables, the fiscal variable (measured by public consumption growth) turns out to be most significant for the recovery in all post-crisis periods except for the year of t+1. The estimated coefficients imply that an increase of the public consumption growth rate by 10 percentage points leads to an increase in the GDP growth rate by 0.5˜0.9 percentage points.

In contrast to the positive and significant contribution of fiscal policy, monetary policy turns out to be less important for post-crisis recovery. The average growth rates of real money supply are insignificant in all equations. One might argue that the weak effect of monetary policy on real output even in the short run is not credible. However, in our view, the real impact of monetary policy is ambiguous in the crisis-hit economies. Contractionary monetary policy as part of IMF programs can contribute to post-crisis growth as it helps stabilize prices and improve the current account.6

The results show that the interactive term between trade share and exchange rate depreciation variables have a significant impact on the post-crisis GDP growth only in a few years following a crisis. The estimated coefficients show that for the country with the average openness ratio of 0.6, a real exchange depreciation of 10 percent raises real GDP growth rate by about 0.4 percentage point per year over the four years after the crisis.

We also examined whether the agreements with the IMF had any impact on the post-crisis recovery. The estimated coefficients turn out to be statistically insignificant. Hence, there is no evidence that IMF programs had any significant impact on the recovery process after a currency crisis, when other factors were controlled.7

Macroeconomic policies may have an additional impact on growth by influencing the level of investment. Table 3 shows the results of regressions for the investment rate. We find that both public investment and real money supply growth play a quite significant role in promoting investment from the beginning of the post-crisis period, while exchange rate depreciation is insignificant. The estimated coefficient for public investment suggests that an increase of 1 percentage point in the ratio of public investment to GDP contributes to an increase in the total investment rate by between 1.3 and 1.5 percentage points. Hence, public investment increases total investment more than one for one, implying that public investment does not ‘crowd out’ equal amount of private investment from domestic sources by competing in product markets or financial markets. Thus, public investment, perhaps by improving the condition of social infrastructure, stimulates private investment and thus contributes to the post-crisis recovery by augmenting capital accumulation. An increase in real money supply growth by about 10 percentage points leads to an increase in the investment-GDP ratio by about 0.9~1.3 percentage points per year over the two years following the crisis.

Table 3.

Regressions for Investment Rate in the Post-Crisis Period

article image
Notes: Each equation is estimated by the least squares method. Robust standard errors reported in parentheses. Levels of statistical significance indicated by asterisks; *95 percent; ** 90 percent. Constant term is included, but not reported.

Assessments of the Recovery Process in East Asia

Macroeconomic Adjustments in East Asia

The economic turmoil that broke out in Thailand in July 1997 swept through East Asia and its devastating impacts were much more severe than anyone had expected. The countries that fell victim to the crisis suffered a sharp reduction in real income. In 1998, the growth rate plunged from the pre-crisis average of 7.0 percent to −13.2 percent in Indonesia, −10.4 percent in Thailand, −7.5 percent in Malaysia, −6.7 percent in Korea, and −0.6 percent in the Philippines. However, since 1999 the five crisis-hit Asian countries have managed impressive recoveries, which have been faster than the similar episodes of recovery in other parts of the world before. The rebounding of the growth rate in 1999 was no less drastic than its free-fall. Korea stood out as the best performer in that year by growing at 10.7 percent. For the other countries, the growth rate ranged from 5.4 percent in Malaysia to 0.2 percent in Indonesia.

With the passage of time, the recovery process has gained momentum. The growth outturn in 2000 is estimated to be higher than that of 1999 in four of the affected economies—Indonesia, Thailand, Malaysia, and the Philippines. In Korea, the growth rate slowed from 10.7 percent to 8.3 percent.

Figure 5 shows the GDP growth rates of the five affected economies. The adjustment process in East Asia that can be inferred from changes in the growth rates seems to be in general consistent with the stylized V-pattern we observe from the previous crisis episodes. However, the East Asian experience is in marked contrast to the stylized pattern of adjustment in GDP growth in that the degree of initial contraction and subsequent recovery has been far greater than what can be predicted from the previous cross-country evidence.

Figure 5.
Figure 5.

Adjustment of Real GDP Growth Rate in East Asia

The initial GDP contraction in 1998 was largely caused by the collapse of investment: the level of domestic capital formation plummeted in all five countries in 1998. The contraction amounted to more than 30 percent in Indonesia, Malaysia and Thailand, 17 percent in Korea, and 9 percent in the Philippines (Table 4).

Table 4.

Macroeconomic Adjustment in East Asia

article image
article image
article image
Source: Asian Development Bank on-line country data (http://www.adb.org/Statistics/country.asp).Note: The share of expenditure components in GDP is constructed based on data in constant prices.

Compared to investment demand, private consumption fell to a lesser degree. The consumption-GDP ratio remained mostly stable in the crisis period, which is consistent with the cross-country stylized pattern. On the contrary, the investment-GDP ratio dropped sharply. In Korea, for example, it fell from 33.4 percent in 1997 to 22.0 percent in 1998. Investment demand started to recover somewhat in 1999 in Korea and Malaysia, but it has continued to decline in the other countries.

While domestic demand was sluggish, a large increase in net export paved way for the initial recovery of the Asian economies. Import demand declined in all of the crisis-hit countries in 1998 by a substantial amount, ranging from 22 percent in Korea and Thailand to 5.3 percent in Indonesia, while exports continued to grow or remained unchanged in all countries except the Philippines.

It is therefore clear that net exports led the recovery in East Asia. Figures 6 and 7 data demonstrate the pattern of adjustment in more detail. A close examination of the quarterly rates of GDP growth shows that both Korea and Thailand reached the trough as early as the second quarter of 1998, and Indonesia, Malaysia, and the Philippines two quarters later. Overall, the recession in East Asia bottomed out in the second half of 1998, less than a year after the crisis had broken out. As shown in Figure 7, the subsequent recovery in 1999 was led mostly by a surge in net exports. Over the post-crisis period the private consumption to GDP ratio has remained stable in all countries except Indonesia. In Indonesia, private consumption expenditure rose in 1998. In Korea and Malaysia, the investment rate started to increase from the latter half of 1998, whereas in the other countries the investment rate declined.

Figure 6.
Figure 6.

Quarterly Changes of Real GDP Growth in East Asia

(In percent, year over year)

Figure 7.
Figure 7.
Figure 7.

Quarterly Movements of GDP Components in East Asia

An increase in public investment appears to have contributed to the resurgence of total investment expenditure in Korea and Malaysia. Table 4 shows that in both countries the fraction of government capital expenditure in total investment jumped from 11 percent in 1997 to over 21 percent in 1998.

The large depreciation of currency has backed up the quick surge of net exports since 1998. Table 4 and Figure 8 show that the level of real effective exchange rates in the five crisis-hit East Asian countries depreciated by 22 percent on average, ranging from 12 percent in Thailand to 50 percent in Indonesia in 1998.

Figure 8.
Figure 8.

Real Effective Exchange Rate in East Asia

Factors behind the Speedy Adjustment in East Asia

A large number of internal and external factors are likely to have contributed to the pattern of macroeconomic adjustment to the crisis in East Asia. On the basis of the cross-country evidence and available information on the pattern of macroeconomic adjustment in East Asia, we attempt to identify some of the factors that have engineered the post-crisis recovery.

Macroeconomic Factors

According to the empirical examination of the stylized pattern of adjustments from the previous 160 currency crisis episodes over the period from 1970 to 1995, which show a V-type adjustment of real GDP growth, a large real depreciation, expansionary monetary and fiscal policy, and an improvement in the global economic environment have been responsible for the upturn of the crisis-hit countries. In this sense, the East Asian process of adjustment is not much different from the stylized pattern. The same factors contributed to the quick post-crisis recovery of the East Asian economies.

An important structural factor driving the speedy adjustment in East Asia may have been the region’s higher level of openness. With a relatively large trade sector and export-orientation, these economies benefited from a large depreciation of the real exchange rate. The level of openness in terms of the share of exports and imports in GDP ranges from 200 percent in Malaysia to 60 percent in Indonesia. Thus, compared to other crisis-hit economies before, the depreciation is likely to have had a bigger impact on the more open East Asian economies. Note that the size of real exchange depreciation in the East Asian countries was comparable to the average depreciation rate in the previous crisis episodes.

One special feature of the East Asia crisis is that compared to the crosscountry evidence, the impact of depreciation on real output showed up as early as one year after the crisis. The large real exchange depreciation therefore restored external balance without much delay in East Asia. The flexibility in the labor market may have facilitated this swift adjustment, since the shift of resources from the non-tradeables to the tradeables sector elicited by the massive real exchange rate depreciation requires flexible factor markets.

The quick improvement in East Asian exports has been supported by favorable external developments. The global economy was strong in 1999. The U.S. economy has been able to absorb a large amount of exports of the East Asian economies. U.S. per capita GDP growth rates were 3.3-3.4 percent in 1998 and 1999, and jumped to 4.4 percent in 2000, which by far exceeded the average growth rate of 2.0 percent over the period from 1970 to 1995. As we saw from the cross-country regressions, global economic growth has a strong impact on the post-crisis recovery, particularly in the early years following the crisis. The deterioration in the terms of trade that precipitated the crisis thus reversed in 1999. In particular, the increase in the prices of semiconductors helped to boost Korean, Malaysian, and Thai exports.

Concerning macroeconomic policy management, the swift change in policy stance toward expansion has supported a quick recovery of the crisis-hit economies. In Korea relaxation of monetary and fiscal policy began around in April of 1998. A comparison of the turning points in the adjustment process measured by growth rates of the quarterly GDP with the timing of policy changes, broadly confirms that easing of monetary and fiscal policy has quickened the pace of recovery in both Thailand and Malaysia (Figure 9). Thailand shifted to a modest relaxation of macroeconomic policy in June 1998, and its economy took off in the fourth quarter of the same year after zero growth in the preceding quarter. In particular, public consumption expenditure increased significantly in the latter half of 1998. It was not until the third quarter—the end of August—of 1998 that a relaxation of monetary and fiscal policy was announced in Malaysia, and its economy moved out of the trough a quarter later. In Indonesia, on the contrary, because of the continuing weakness of the Rupiah, monetary policy remained contractionary until the second quarter of 1999. But public consumption increased sharply in the third quarter of 1999. This expansion boosted output growth in 1999. In Philippines, monetary policy was slightly contractionary over the post-crisis period, while public consumption expenditure has been growing since the first quarter of 1999.

Figure 9.
Figure 9.

Policy Indicators in East Asia

The positive role of expansionary macroeconomic polices in the post-crisis recovery raises the question of whether the initial tightening of monetary and fiscal policy was too harsh, and maintained for too long, and as a consequence deepened the crisis. In order to deal with the crisis itself—stopping bank runs, protecting the payment system, and stemming capital outflows, the IMF prescribed tight monetary policy together with fiscal austerity, which initially led to a sharp increase in interest rates. The contractionary monetary and fiscal policy has been criticized by many, including Radelet and Sachs (1998) and Feldstein (1999), as having been unnecessary because these countries were suffering from a liquidity problem. They imply that the traditional IMF prescriptions may have done more harm than good as they drove many highly leveraged but viable firms out of business, thereby deepening the downturn of the economy. The contribution of the initial IMF austerity programs remains controversial. On the other hand, it is quite clear that the swift change of macroeconomic policy stance toward an expansionary one helped these economies recover quickly. Fiscal policy had become contractionary immediately after the crisis, but was reversed quickly to be expansionary. The change in monetary policy stance then followed. Once the depreciation of the currency was arrested and stability returned to the foreign exchange market, the authorities of the crisis countries were able to adjust gradually the interest rates downward and expand the money supply.

Panic and Balance Sheet Effect

The contraction of real income in the East Asian countries that suffered the crisis was much larger and the subsequent recovery of these countries has been much faster than what can be predicted from the previous crises. There must be additional factors that have contributed to the deeper contraction and the quicker recovery in East Asia. We consider that the East Asian crisis has an aspect of a severe liquidity crisis caused by investors’ panic. This nature of the crisis must have an important role in the macroeconomic adjustment during the crisis.

There is general agreement that a fixed peg to a currency basket dominated by the U.S. dollar when the current account was piling up deficits was one aspect of policy mismanagement that triggered the crisis in Thailand. In a recent paper, Williamson (2000) shows that had it been implementing a BBC (basket, band, and crawl) rule, Thailand might have staved off its crisis, because the country was suffering from a balance-of-payment crisis. However, the Thai crisis was contagious, as shown by Park and Song (2001a, 2001b), and even a good exchange rate management using the BBC rule could not have saved other crisis victims like Indonesia and Korea from the contagion.

Although macroeconomic policies and economic fundamentals of Korea and Indonesia were regarded sound and credible, many foreign investors simply moved out of East Asian financial markets when they realized that most of the East Asian countries would suffer from similar macroeconomic and structural problems that were driving Thailand to the brink of debt default. With the withdrawal of foreign lenders and investors from the region, other East Asian countries experienced a sharp liquidity crisis and balance sheet problems associated with a large currency depreciation, causing a region-wide crisis explicable by second and third generation crisis models. That is, the contagion of the Thai crisis set in motion a crisis characterized by self-fulfilling prophecy and balance sheet deterioration in other East Asian countries, which did not have serious balance-of-payment problems. Once hit by contagion, the BBC system was simply unable to stave off the crisis because the band could not be maintained.

Why did foreign portfolio investors panic so much and exhibit herd behavior? They initially moved into East Asia with large sums of money to be invested in all types of local securities and real assets by the mistaken notion that rapid growth in the region would be sustained or that their investments would be protected by government guarantees. Most of the foreign investors paid little attention to the structural problems of the financial and corporate sections that began to haunt East Asia before moving in. When these problems came to light in the midst of currency depreciation and interest rate increases, they were startled. The ensuing fear of losing their investments then drove them to a state of panic, and every investor was scrambling to the exit.

Thus, one critical factor that could explain both the initial sharper contraction and faster recovery is related to changes in the expectations of foreign investors and both domestic households and firms on economic prospects of the crisis countries. When foreign investors began to lose confidence in East Asian economies, capital flows abruptly reversed. As shown in Table 5, in 1997 private net capital flows reversed by $115 billion (from a $120 billion inflow in 1996 to a $5 billion outflow). It is no wonder that this large-scale shift in financial flows provoked deep contractions, huge depreciations and financial embarrassment. And the argument goes that once investors’ panic calms down and foreign capital inflows resume, the economy rebounds to its long-term trend.

Table 5.

Capital flows to the Five Asian Economies

(Billion U.S. dollars)

article image
e-estimateNote: The five countries include South Korea, Indonesia, Thailand, Malaysia, and the Philippines.Source: IIF, January 2001.

Immediately after the crisis, there was rampant speculation that the crisis countries might not be able to avoid foreign debt default and hence might have to declare a debt moratorium. The international financial community, including the international financial institutions, also did not hesitate to lay the blame on the East Asian countries for the crisis. With the emerging consensus that the crisis countries had profound problems that were more serious than had been realized before, the prospect for recovery in East Asia turned from bad to worse. Many were skeptical whether these countries had institutional capacity and political will to carry out the necessary structural reforms. Even if they had, the skeptics pointed out that these crisis countries would take many years to put their houses in order. Under these circumstances, it is quite possible that the households and firms as well as foreign investors came to believe that the crisis was a permanent shock which would lead to a new lower equilibrium in terms of output and employment than when the crisis was seen as a temporary shock. And this perception of permanency may have induced domestic consumers and investors to cut down their spending a lot more than otherwise during the first six months of the crisis. However, the extensive criticism of East Asia diminished and gradually gave way to a more optimistic outlook for the crisis economies, and the realization that the crisis might be a temporary phenomenon started sinking in the minds of consumers and investors, thereby encouraging their spending.

In restoring the confidence of foreign investors, large support packages by the IMF made some contribution. The funding helped to reduce the short-term liquidity constraints of the economies and provide resources to stem the exchange rate depreciation. There were other turning points. Korea, for example, reached an agreement with its creditor in February of 1998 to lengthen the maturities of the short-term foreign currency loans (Radelet and Sachs, 1998).8 After the agreement was reached, at least some of foreign credit facilities including trade credit were restored. With this restoration of credit, the fear of a debt default abated considerably.

A large decrease in aggregate investment demand during the crisis period suggests that corporate distress was one of the main factors responsible for the sharper contraction in output in East Asia. Structural weaknesses in corporate and bank balance sheets were often pointed out as the main channel through which the effect of foreign disturbances was magnified in the East Asian crisis (Krugman, 1999; and Stone, 2000).

The reversal of capital inflows combined with a sudden downward shift in expectations could lead to a sharp depreciation of the exchange rate. The large unexpected depreciation was much more disastrous in East Asia because most firms were highly leveraged. When the bulk of corporate debts are denominated in U.S. dollars while revenues and assets are in local currency, the depreciation deteriorates the balance sheets of firms and inflicts large losses.Table 6 shows that foreign exchange losses of the Korean firms amounted to more than 17 trillion won in 1997, which was about 3.8 percent of GDP.9 These losses together with the increase in foreign debt financing costs result in a decline in the present value of the equity of the corporate sector. Gray (1999) stimates that a 50 percent depreciation would have reduced the equity value of Korean corporations by 9 percent and that of Indonesian corporations by 21 percent. The lower equity value leads to lower investment.

Table 6.

Foreign Exchange Losses of the Korean Corporate Sector

(Billion won, percent)

article image
Source: Authors’ estimates based on the Bank of Korea, Financial Statement Analysis.

Balance sheets of financial institutions were also very vulnerable to the currency depreciation. Since in East Asia banks had a large amount of foreign liabilities in their balance sheets, they suffered losses emanating from the currency mismatch.10 In June 1997 the ratio of foreign liabilities to foreign assets of the banking sector ranged from 1.3 in Korea to 6.8 in Thailand (Table 7). Maturity mismatches also created another vulnerability. Korean data shows that short-term foreign liabilities were more than twice as large as short-term foreign assets (Table 8).

Table 7.

The Ratio of Foreign Liabilities to Foreign Assets of the Banking Sector1

article image

Gross foreign liabilities and assets of deposit money banks.

Source: ADB based on data from IMF, International Financial Statistics.
Table 8.

Foreign Assets and Liabilities Outstanding at Financial Institutions in Korea

(Billion U.S. dollars)

article image
Source: Bank of Korea.

Percentage of total assets.

Percentage of total liabilities.

Once banks and non-bank financial institutions suffer a sharp decline in profits and hence a substantial erosion of their capital base, they are downgraded by the rating agencies and often denied access to international financial markets. As experienced by many money-losing financial institutions in East Asia, foreign banks and other institutional investors simply cut the lines of credit they had offered through the inter-bank loan market and refused the rollover of short-term loans when their client institutions were in trouble. This refusal created serious liquidity as well as balance sheet loss problems at the East Asian financial institutions. Faced with the liquidity problem, many banks and non-bank financial institutions had to reduce their supply of loans in both local and foreign currencies drastically even to their viable loan customers.

The mounting losses caused by the deterioration of bank balance sheets was bound to increase the country risk premium of the crisis-hit countries. A rise in the country risk premium in turn pushes up the cost of capital and lowers the present value of the equity of the corporate sector. Gray (1999) estimates an 8 percent temporary rise in the country risk premium for a year leads to a drop of 7 percent in the present value of corporate equity in Korea and 2 percent in Indonesia.11

An increase in the interest rate and currency depreciation together with other shocks can reduce the equity value of the corporate sector below a threshold that triggers widespread default. The risk of default was higher in East Asia where firms were highly leveraged with a large amount of short-term liabilities. The firms with a larger share of short-term debt faced more difficulties in financing and were unable to service their debts: bankruptcies soared, thereby magnifying the crisis.

In the recovery process, macroeconomic stability plays a crucial role for the normal operation of viable firms. Stabilization of the exchange rate and interest rate improves the equity value of the corporate sector and thus promotes investment. Improved confidence leads to an increase in spending. The restructuring of the corporate sector is necessary in order to reduce the vulnerability of the corporate sector and thus prevent the future crisis. However, in the short run, a quick recovery can not be engineered unless there is resurgence of domestic demand.

Structural Reform and Recovery12

At the beginning of the crisis, there was widespread belief that the crisis countries’ commitment to structural reforms would be critical to the recovery in East Asia. The reforms were expected to help East Asia emerge from the crisis with more stable, transparent, and efficient financial and corporate sectors. This expectation of reform espousing a market-oriented system would then improve long-term growth prospects and, at the same time, restore market confidence, thereby inducing the return of foreign lenders and investors to the region.

Three years into the reform process, the crisis countries have accomplished a great deal in improving the soundness and profitability of financial institutions and alleviating corporate distress. The World Bank (2000, p.7) argues that “assertive structural adjustment helped restore credit flows and boosted consumer and investor confidence.” Yet, it is not clear whether and to what extent financial and corporate restructuring has contributed to the ongoing recovery. Most of the serious structural problems that were identified as the major causes of the crisis in Indonesia, Korea, Malaysia, and Thailand could not have been resolved over a span of two years. In fact, banks are still holding in their balance sheets a large volume of non-performing loans and remain undercapitalized in all four countries. Many corporations in the region are still unable to service their debts. As for institutional reform, new banking and accounting standards, disclosure requirements, and rules for corporate governance have been introduced, but they are not rigorously enforced. It will take many years for the new system to take root.

Since the crisis countries are not even half-way there in restructuring their financial institutions and corporations, it would be presumptuous to argue that the reform efforts have established a foundation for sustainable growth in East Asia. Nor, would it be correct to assert that the gain in efficiency through the restructuring, which is difficult to measure at this stage, has been one of the principal factors driving the recovery. The improvement in efficiency is likely to be realized and translated into high growth over a longer period of time, certainly longer than two years.

The available pieces of evidence also do not support the contention that market-oriented reforms have contributed to restoring market confidence in the East Asian crisis countries; it certainly did not appear to have during the first two years of the crisis. International credit rating agencies report that reforms in the banking sector in the crisis countries have not gone far enough to ensure that these economies would be able to forestall another financial crisis. Only toward the end of 1999, Moody’s and S&P upgraded the sovereign credit ratings of Korea and Malaysia to the lowest investment grade from speculation grade. By that time, the recovery was in full swing in East Asia. Journalistic accounts have abounded with similar concerns and continued to raise doubts regarding the effectiveness of the reform in the crisis countries. Under these circumstances, most foreign investors would find it risky to return to the crisis countries, but they have. Many of the foreign investors appear to have been lured back by the rapid recovery and substantial improvements in external liquidity resulting from large surpluses on the current account.

Reflecting recovery rather than ratings improvement, capital inflows in East Asia have been rising. Since policy changes and structural reforms are subject to many uncertainties and require a long time to take effect, international banks and global institutional lenders do not seem to have either the patience or ability to monitor and assess the effects of structural reforms. This is particularly true when they are preoccupied with the short-term performance of their portfolios.

Differences in Post-Crisis Performance among the Asian Countries

The five Asian countries most affected by the Asian financial crisis experienced a speedy recovery that was faster than anyone had expected. But the extent of the recovery from the crisis differed among the five countries. By the end of 1999 only Korea had surpassed its pre-crisis peak level of GDP. Malaysia and the Philippines did so later in 2000, while Thailand and Indonesia still need another year or so to recover to their pre-crisis output level.

Table 4 indicates that the difference in the post-crisis recovery in 1999 reflects mainly the difference in the performance of investment and export growth among the Asian countries. While the annual growth rate of export in 1999 amounted to 16.4 percent in Korea, 13.8 percent in Malaysia, and 8.9 percent in Thailand, it was −32.1 percent in Indonesia. After investment ratios had dropped sharply in the five Asian countries in 1998 due to the crisis, they showed a slow recovery in 1999 in both Korea and Malaysia. By contrast, in the other three Asian countries investment ratios contracted further in 1999.

The investment contractions reflect the significant distress in both the corporate and financial sectors. The financial crisis caused deterioration of firms’ balance sheets. Then, the deterioration of the balance sheets of firms caused a massive accumulation of non-performing loans at banks and other non-bank financial institutions. The accumulation of bad loans cut into profits and consequently decreased the equity value of the financial institutions. Decapitalized financial institutions as a result of the mounting losses were forced to curtail their lending to both viable and non-viable firms, thereby exacerbating the downturn of investment.

In a bank oriented financial system that characterizes the financial structure of the crisis-hit countries in East Asia, the repercussion of bank failures is much more pervasive and felt throughout the economy. Because of their dominance, therefore, banks are likely to bring down many viable firms than otherwise when they are not able to function as intermediary.

Data show that the investment and output contractions in the Asian countries are closely associated with the sluggish bank lending. Although the money supply began to expand in 1999 for the five Asian economies, the supply of bank credit in real terms continued to slow in three of them—Indonesia, Thailand, and the Philippines (Panels (b) and (c) of Figure 9). In fact, more than three years after the crisis, real credit supply remains below the pre-crisis level in those three countries. The investment ratio recovered most quickly in Korea where real credit increased at the highest rate over the post-crisis period.

Prospects for Long-term Growth in East Asia

As the recovery continues in East Asia, there is a growing hope that these economies will be able to return to the pre-crisis level of robust growth. In this section, we make an assessment of the long-term growth prospects for East Asia.

Impacts of a Currency Crisis on Long-Term Growth

We investigate the impact of a currency crisis on long run growth based on a cross-country regression framework. We control for all important growth determinants and then examine whether a currency crisis has had any independent impact on GDP growth in the long run.

A wide variety of external environment and policy variables will affect growth prospects by changing the long-run potential income and the rate of productivity growth. Based on the results from previous empirical research, we consider the following variables as the important determinants of long-run per capita income growth: (1) initial income, (2) human resources, (3) the investment rate, (4) exogenous terms of trade changes, and (5) institutions and policy variables including government consumption, rule of law, and openness.13 To measure the stock of human capital, we use the average years of secondary and higher schooling for population aged 15 and over, available from Barro and Lee (2001). The rule of law index is a measure of the quality of institutions, which is based on the evaluation by international consulting firms that give advice to international investors. The openness measure is based on Sachs and Warner (1995). This index is calculated as the fraction of years during the period that the country was considered to be open to trade and thus sufficiently integrated with the global economy. The evaluation of the country’s openness is made on the basis of four dimensions of trade policy: average tariff rates, quotas and licensing, export taxes, and the black market exchange rate premium.

Table 9 presents the results of regressions for per capita real GDP growth rate using the explanatory variables just described. The data are a panel set of cross-country data over the two decades, 1975-85 and 1985-95. The system of two equations is estimated by a seemingly-unrelated-regression (SUR) technique, which allows for the correlation of the errors across the equations.

Table 9.

Long-Run Impact of Currency Crisis on Per Capita Growth Rate

article image
Notes: The system has two equations, where the dependent variables are the growth rate of real per capita GDP for each of the two: 1975-85 and 1985-95. The estimations use the SUR (seemingly-unrelated) estimation technique, which allows the error term to be correlated across the two periods and to have a different variance in each period. Each equation is allowed to have a different constant term (not reported). Standard errors are shown in parentheses. The R2 values and the number of observations apply to each period separately.

The regressions show that most of the controlling variables are significant determinants of long-term growth. For instance, the coefficient on the log value of initial GDP is highly significant. Thus it provides strong evidence for conditional convergence: that is, a poor country with a lower initial income level grows faster, when the variables influencing the steady-state level of income are controlled. Specifically, the coefficient in column 1 of Table 9 imply that a country at half of income level of another country grows by 1.4 percentage points (=2.0*ln(2)) faster than the richer country.

We add to the regression a variable that measures the occurrence of currency crises. The variable is constructed with the number of currency crises that each country experienced during the past decade. We have used the number of crises over the period of 1970-75 for the first equation and over the period of 1975-85 for the second equation. Thus we test if an experience of a currency crisis can have an impact on growth in the next decade. The estimated coefficient turns out to be statistically insignificant, implying that there is no direct impact of currency crises on growth in the long run. In column (2) of the regression, we add a variable that represents the number of currency crises with the IMF program participation. we also found no significance for this variable.

Although there is no direct impact of a currency crisis on long-run growth, it would be possible that a currency crisis or IMF program can have an indirect impact on long-run growth by influencing the controlling variables. For instance, if the investment rate becomes permanently lower because of the postcrisis stabilization program in the crisis-hit countries, it would have a negative impact on growth in the long run. On the contrary, if IMF structural reforms improve the quality of institutions, then a currency crisis with the IMF program participation can have a positive impact on growth.

Sustainability of East Asian Growth

The quick turnaround of the Asian economy from the 1997 crisis has brightened the region’s economic prospects. Despite the impressive record of the recovery, however, not everyone is sanguine about East Asia’s future prospects. The World Bank and IMF, for example, are not optimistic about the prospects of these countries sustaining the ongoing recovery, largely because weaknesses of financial institutions and balance sheet problems of corporations still remain unresolved in the region.

The macroeconomic performance of the crisis countries in 2001 will provide important clues to the question of whether these countries will be able to return to the pre-crisis trend rate of growth. Up to 2000, the pattern of recovery in East Asia has been quite similar to that of Mexico after its crisis in 1994.

Although the financial crisis of 1997 abruptly brought a halt to Asia’s period of robust growth, there was little in Asia’s fundamentals that inevitably led to the crisis. The key to the Asian crisis was too much short-term capital flowing into weak and under-supervised financial systems. This suggests that with better financial management and a return to the core policies that resulted in rapid growth, the East Asian economies can again return to sustained growth (Radelet, Sachs, and Lee, 2001). The major factors that have brought the relatively high growth in East Asia were high rates of saving, good human resources, trade openness, and maintenance of good institutions. In terms of these fundamentals, East Asia still has strong potential for a sustained growth.

But, in the long-term, the growth rate will be lower than the previous precrisis average of 7 percent. The convergence factor, which was found to be quite strong in the cross-country growth regression in the last section, implies that the faster growth in the last decades will force the East Asian economies to grow at a slower pace in the next decade. That is, the East Asian countries now have a much smaller gap in reproducible (physical and human) capital and technical efficiency from their long-run potential levels than they had in the last decades. Hence, the East Asian economies will face a smaller chance for rapid “catching up” through high rates of capital accumulation as well as the diffusion of technology from more technically advanced economies in the next decade, and inevitably become adjusted to a lower growth path.

The coefficients in the cross-country growth regressions imply that the convergence factors alone makes the Asian economies grow by about 1.5 percentage points slower over the next decade, compared to the last decades in which they had started with less than a half of their current income. Hence, unless the economies could achieve substantial improvements in other fundamental factors, such as quality of institutions, real GDP would grow at about 5 percent per year.

Concluding Remarks

The contraction of real income in the East Asian countries that suffered the crisis that erupted in 1997 was much larger and the subsequent recovery of these countries has been much faster than what can be predicted from the previous episodes of crisis elsewhere. The purpose of this paper has been to identify some of the factors that may explain the severity of, and rapid recovery from, the crisis. According to our empirical examination of macroeconomic developments following the crisis in East Asia, including a V-type adjustment of real GDP growth, a large real depreciation, expansionary monetary and fiscal policy, and an improvement in the global economic environment have been responsible for the upturn of the crisis-hit countries. In this sense, the East Asian process of adjustment is not much different from the stylized pattern observed from the previous 176 currency crisis episodes over the period from 1970 to 1995. However, the stylized pattern of adjustment cannot explain why the crisis was so severe and why the recovery has been so much faster than what was expected from the previous experiences of crisis. This study argues that the East Asian financial upheaval was in a large measure a liquidity crisis caused by investors’ panic. Once the liquidity constraint was eased as it was during the first half of 1998, domestic demand has since surged again and the crisis countries have been able to move toward the pre-crisis path of growth.

Comments on Papers 10 and 11

Charles Adams

These two papers raise important questions about the 1997-98 crisis and its possible long-run effects on the affected economies. In my comments, I want to focus on two interrelated issues: (i) what the evidence presented in the papers can tell us about the possible long-run impact of the 1997-98 crisis on Korea’s growth rate; and (ii) why the output declines during the Asian crisis were much larger than in other currency crises. In addition, my comments will touch briefly on the evidence presented by Park and Lee on the short-run behavior of crisis economies with and without Fund programs.

Long-term growth prospects.

When the Asian crisis first erupted, there were many observers who wondered whether it signaled the end of the (so-called) Asian miracle. Either because of the damaging effects or, most likely, because the crisis was seen as reflecting deep-seated problems plaguing affected economies, skeptics were quick to start hammering the nails in Asia’s coffin. Subsequently, as the crisis economies began to bounce back quite quickly in 1998-99, questions were raised about the sustainability of the recoveries (a “dead-tiger” bounce?). There was, in short, a fair degree of concern in some quarters about the outlook for Asia in the aftermath of the crisis. Against this background, Barro’s paper provides a useful overview of the factors underlying the very favorable long-term growth performance of the Asian economies and both papers shed light on what the experiences from other currency crises can tell us about the long-run effects of the Asian crisis.

The approach adopted to assess whether the crisis will have long-run effects on growth is to look at a large number of currency-crisis episodes (defined in terms of the size and abruptness of the change in the exchange rate), and whether they have tended to be accompanied by persistent effects on real GDP. Periods of currency crisis are captured through dummy variables. The main (and, in my view, somewhat surprising) finding is that currency crises, on average, have not lead to long-run changes in growth and investment, once allowance is made for trend changes related to convergence and other effects.1 Applying this result to Korea would thus make one relatively optimistic about long-term economic prospects, at least in so far as the crisis would not be seen as causing damage.2 Notwithstanding this relatively sanguine conclusion however, it should be noted that the Korean stock market does not yet appear to be pricing in a sustained bounce back and investment rates are still to recover to pre-crisis levels, notwithstanding the pick up of the economy from the sharp slowdown in 1997-98.

What is one to make of the relatively benign conclusion on the long-run effects of currency crises on growth? There would seem, in my view, to be at least four reasons for caution.

  • First, there is the question as to how to interpret the results given that growth and the occurrence of currency crises should ideally be determined endogenously in a fully specified model. In short, there is a question about the conditions under which the effects of a crisis can meaningfully be assessed through adding “exogenous” dummy variables to growth equations. One possibility is that the crisis dummies are capturing the impact of unobserved variables that influence both the occurrence of crises and (potentially) long-run GDP growth. But, if this is the case, these variables should ideally be identified and included as independent variables in the equations explaining long-run growth. Alternatively, the specification may be capturing potential feedback between crises and growth. For example, crises may encourage political or economic change (for good or bad) that, in turn, influences the determinants of growth. Alternatively, the growth process itself may influence the probability of crises such as might be the case, for example, if a crisis was caused by over investment over a sustained period.

  • Second, the methodology used is obviously based on the premise that currency crises are, in some sense, alike and that findings can be generalized. Clearly, however, currency crises are not all alike and average results can conceal wide differences in outcomes. For example, in some cases, currency crises have been accompanied by banking crises or external debt problems and, in others, by political change or upheaval. My prior would be that the degree of persistence would likely depend on whether crises have been accompanied by problems in these other areas. In particular, currency and external debt crises seem to have had quite persistent effects on growth, as was the case, for example, in Latin America during the 1980s. And currency crises involving banking problems seem also to have large and quite persistent effects, especially when banking problems are not promptly addressed. Beyond this, there is also the possibility that currency crises may have different effects across different types of economies according to the currency denomination of external debt. In particular, the short-run effects of such crises may be positive in mature economies that borrow in their own currencies but negative in emerging-market countries that only borrow abroad in foreign currencies (see Krugman, 1999). For all of the above reasons, the zero long-run average effect might well conceal a wide range of different experiences across countries that it would be useful to understand before applying the “average” effect to other crises.

  • Third, although they incorporate many of the standard determinants of growth, the estimated equations include few of the variables that have figured prominently in recent discussions of the Korean (and other Asian) crisis episodes. Most notably, there is no direct allowance for the structure and efficiency of financial intermediation, corporate governance, transparency etc. These are factors that have been emphasized, in particular, by Chopra et al. at this conference as potentially important for understanding the crisis in countries such as Korea. Since no direct allowance is made for these variables, the equations cannot easily be used to assess the implications of “reforms” in these areas for the long-term growth outlook and hence how the crisis—by spurring reformcould have long-lasting effects on growth. Moreover, if, as some at this conference have argued, these “new” structural variables are more important for growth than has been assumed in the past, their exclusion from the estimated equations raises the possibility that the equations have been misspecified.

  • Finally, another reason for caution is that the Korean and other Asian crises clearly differ from the average experience in terms of the severity of the output declines and the failure of investment to thus far recover significantly. Part of the reason for this may be that the Asian crisis was both a banking as well as currency crises. But, until the reasons for the differences are better understood, care should be used in extrapolating from other crises. In addition, we need to understand better whether the failure of investment to rebound in Asia since the crisis signals potential problems down the road for growth or is simply the reflection of past excesses. An important question, in this connection, is whether it will be possible for rapid growth to resume in Asia with less investment than before the crisis. This might be the case, for example, if structural reforms are successful in improving the efficiency of financial intermediation and would imply that the weakness of investment need not be cause for concern.

Severity of output declines.

As noted, the output declines during the Asian crisis were much larger than during the average currency crisis. Park and Lee, in particular, argue that the large output losses reflected the interaction between investor panic, liquidity problems, and the structure of balance sheets, specifically, currency/maturity mismatches and high leverage. In essence, the story is one in which economies were shifted from a “good” to a “bad” equilibrium by (self-fulfilling?) investor panic, leading to very large output losses. The degree of investor panic in Asia, including the very large reversal of capital inflows of over $100 billion in 1997-98, is well documented by Park and Lee, and the argument made that the “fragile” structure of balance sheets rendered Asian economies particularly vulnerable to the associated liquidity problems. The argument appears consistent with other work that has sought to identify the vulnerabilities that may have turned a potentially “minor” crisis into a “major” event. As is well known, however, it is not easy to allocate “responsibility” across liquidity problems caused by investor panic and more deep-seated weaknesses and insolvency problems. Even though many would agree on the importance of balance-sheet structure, there continue to be differences of view on its role in explaining the severity of the Asian crisis. More substantively, however, a key question is whether balance sheet effects would make the Asian crisis fundamentally different from other crises and lead to long-run effects different from those in the “average” crisis. My impression is that Park-Lee see the Asian crisis as very different from many earlier crises but, if this is the case, it is not clear whether the earlier crisis experiences will necessarily provide a good guide to assessing the effects of the Asian crisis.

Finally, let m e turn to the question of the behavior of crisis countries with and without Fund programs. The Park and Lee paper suggests that there have tended to be some potentially large differences in the short-run behavior of currency-crisis economies with and without Fund programs. Most notably, countries under Fund programs on average have tended to experience larger output contractions during crises, but seem also to enjoy sharper bounce backs. Although they recognize that care is needed in interpreting the results—including because countries with programs may have more severe problems than those without—Park and Lee suggest that the differences reflect a tendency for more contractionary policies to be adopted under Fund programs. Having made these observations, however, it is striking that Park and Lee do not make much of the fact that these differences between program and non-program cases do not apply in the Asian crisis countries. Looking at the data provided on the five Asian crisis economies (Table 4 in their paper), the range of outcomes for the 1997/98 output collapse is spanned by two countries with Fund programs: Indonesia (where the decline is largest) through the Philippines (where the decline is smallest). Interestingly, however, the output declines in Thailand and Korea (with Fund programs) do not appear very different to that in the one crisis country without a Fund program (Malaysia). This raises questions as to the reasons for the similarities and how they can be reconciled with the results for other crises. Possible questions include: Were the outcomes similar because Malaysia actually followed similar policies to those advocated by the Fund, at least until capital controls were imposed in late 1998? Alternatively, were the output declines similar, notwithstanding different approaches to crisis resolution, because of common factors across the Asian crisis economies? Did all these economies face the same panic-induced liquidity problems the authors regard as important? Although Park and Lee present charts on the behavior of different demand components and policy indicators in the crisis economies, the data do not point to a clear answer as to whether similar policies were followed and why short-term outcomes for output were similar.

Let me conclude by complementing the authors for these two interesting papers on the crisis. Even though we are not yet in a position to reach firm conclusions about the effects of the crisis, the papers have done an excellent job enhancing our understanding of the possible long-run implications for growth as well as short-run crisis dynamics.

Yong Jin Kim

I am honored to comment on papers by such distinguished scholars as Professors Barro, Park, and Lee. I will first briefly summarize their results and then comment on their papers.

Both papers analyze panel data for a large number of countries using the standard cross-country growth methodology. Professor Barro concludes from developments in the five Asian-crisis countries’ stock markets and the investment share in GDP, which was relatively high before the crisis and low after the crisis, that there may be long-lasting impacts from the crises. Barro finds that a somewhat different conclusion emerges from a broader study of currency crises, which suggests that the five Asian-crisis countries could experience a rapid V-type recovery. Thus, his expectations about the long-run impact of the crisis on the Korean economy are mixed.

Professors Park and Lee also show that countries that went through currency crises, including the five Asian-crisis countries, tend to have a rapid V-type recovery. They identify the recovery mechanism as follows: Currency crises depreciate exchange rates, thus increasing exports, and expansionary monetary and fiscal policies increase domestic absorption, contributing to the rapid recovery. Trade partners’ economic growth, changes in the terms of trade, and export-oriented policies are additional factors that can contribute to a rapid recovery. Park and Lee find that the recovery of investment tends to be rather slow. Developments in the five Asian crisis countries were similar to those in other countries that experienced currency crises, but with a deeper economic contraction right after the crises and a faster recovery. Based on panel regressions for many countries, Park and Lee also find that the experiences of crises, with or without an IMF program, do not affect economic growth rates over the following decade. The authors’ results are quite exact and clear.

I have three comments, which are mostly complementary to both papers. The first is that a fast recovery is one thing, but making an economy less vulnerable to crises is another. The second comment concerns how to measure the cost of a crisis. From a theoretical perspective, if the cost of a crisis is as negligible as the cost of a business cycle as argued by Lucas (1987), why are we bothered with crises. My last comment attempts to reemphasize the importance of the proper design of institutions, as Friedman, Johnson, and Mitton argue in their paper for this conference.

As we all know, there is a heated debate among economists and policy makers in Korea. Some argue that the post-crisis restructuring process has not corrected the flaws in the organization of firms and financial institutions that caused the crisis. Others argue that, evidenced by the strong macroeconomic data, the Korean economy is on the right track. The disagreement is not about the current state of the economy, but about whether the economy is healthy enough to not have another crisis in the future. With data available for only two years after the crisis, it will be very difficult to discriminate between these two arguments.

Just before the crisis strong Korean macroeconomic performance suggested a sound economy that was resistant to crises. The crisis was unexpected. This suggests that we should be very cautious about interpreting the rapid recovery to imply that the Korean economy is less vulnerable to crises in the future. We can construct a simple model in which the probability of having a crisis depends on firms’ debt-equity ratio, and given this probability, the crisis is caused by investors’ herd behavior, as in Kim and Lee (2001). In this setup, fiscal and monetary policies can speed up the economic recovery from the crisis, but do not decrease the probability of having a crisis in the future, unless those policies decrease the debt-equity ratio. Expansionary macroeconomic policies can even increase the probability of crisis by subsidizing insolvent firms with higher debt-to-equity ratios. As this model illustrates, a fast recovery and making the economy resistant to crises may be quite different problems.

Other studies imply that increasing transparency is essential to reduce the probability of having a crisis. Frankel and Rose (1996), for example, show that higher levels of foreign direct investment lower the probability of a crisis, as in Hong Kong and Singapore. Similarly, Wei and Wu (2001) show that corruption or poor public governance is associated with a higher loan-to-FDI ratio, and this is related to a higher incidence of a currency crisis.

In this context, it would be interesting to do nested tests linking macroeconomic performance with microeconomic data representing the quality of financial institutions, firms’ governance structure, and so on. Papers such as King and Levine (1993) show that the quality of financial institutions has a significant impact on economic growth.

My second comment is that if there is a trade off between the two goals of a fast recovery and making the economy more resistant to crises, how should limited government resources be allocated between them to maximize welfare. If making an economy completely crisis-free is very expensive, then there exist an optimal allowance of crises. To design an optimal allowance of crises to maximize agents’ welfare, calculations of the cost of crisis and the probability of having a crisis over a set of different economic systems will be unavoidable.

My third comment is related to how to restructure the organizations of firms and financial institutions to minimize agency costs. We have learned from the crisis that the agency problems of chaebol and financial institutions have had large economic costs. These phenomena are well phrased by “too big to fail,” “bet the bank,” and so on. Economic theory says that resources are allocated through two channels, organizations and markets, as Coase (1937) noted. In this context, the strategy of introducing more competition into the markets will not, by itself, result in a more efficient allocation of resources. As the cases of Hanbo, Daewoo, and Hyundai chaebol exemplify, more severe competition without properly designed governance, incentives, and monitoring systems will induce the owners of firms to invest more resources into cheating activities, than into productive activities. Some argue that if chaebol owners behave this way, they will fail in the future due to the mechanism of competitive markets. This may be true, but for the owners it is optimal to cheat. Without the proper incentive mechanism, cheating will continue and the welfare costs will be continuously borne by other agents. This is very obvious theoretically and empirically. In this context, making markets more competitive should go hand in hand with making organizations more transparent. The history of capitalism shows that better designs of organizations, mostly to deter agency problems, have continuously and painstakingly evolved. Can developing countries skip over this process simply by encouraging competitive markets and decreasing new regulations?

Let me summarize the third point. We should design the governance system such that it will provide the proper incentives through explicit rules of law, not relying on owners’ moral spirits, nor politicians’ discretion. Becker (1968) shows that criminals are rational agents maximizing their utilities, using all kinds of means, including illegal ones, and measuring costs and benefits rationally. Thus, his therapy is simple. Raise the cost for wrongdoings, by increasing the penalty and/or the probability of being caught.


  • Aghion, Philippe, Philippe Bacchetta, and Abhijit Banerjee, 2000, “Currency Crises and Monetary Policy in an Economy with Credit Constraints,Working Paper, Economics Department, Harvard University, forthcoming in European Economic Review, ISOM Volume.

    • Search Google Scholar
    • Export Citation
  • Asian Development Bank, 2000, Asian Recovery Report 2000, March, Asian Development Bank (Manila, Philippines).

  • Barro, Robert, 1997, Determinants of Economic Growth: A Cross-country Empirical Study, Cambridge MA, MIT Press.

  • Barro, Robert, and Jong-Wha Lee, 2001, “International Data on Educational Attainment: Updates and Implications,Oxford Economic Papers 53(3), pp. 241-263.

    • Search Google Scholar
    • Export Citation
  • Conway, Patrick, 1994, “IMF Lending Programs: Participation and Impact,Journal of Development Economics, 45, pp. 365-391.

  • Feldstein, Martin, 1998, “Refocusing the IMF,Foreign Affairs, Vol. 77, pp. 20-33.

  • Frankel, Jeffrey A., and Andrew K. Rose, 1996, “Currency Crashes in Emerging Markets: An Empirical Treatment,Journal of International Economics, 41, pp. 351-366.

    • Search Google Scholar
    • Export Citation
  • Goldfajn, Ilan, and Poonam Gupta, 1999, “Does Monetary Policy Stabilize the Exchange Rate Following a Currency Crisis,IMF Working Paper 99/42.

    • Search Google Scholar
    • Export Citation
  • Gray, Dale, 1999, “Assessment of Corporate Sector Value and Vulnerability: Links to Exchange Rate and Financial Crises,” World Bank Technical Paper No. 455.

    • Search Google Scholar
    • Export Citation
  • Gupta, Poonam, Deepak Mishra, and Ratna Sahay, 2000, “Output Response During Currency Crises,” mimeograph, IMF and World Bank, forthcoming in IMF Staff Papers.

    • Search Google Scholar
    • Export Citation
  • Hahm, Joon-Ho, and Frederic Mishkin, 2000, “Causes of the Korean Financial Crisis: Lessons for Policy,NBER Working Paper 7483.

  • Haque, Nadeem Ul, and Mohsin S. Khan, 1998, “Do IMF-Supported Programs Work? A Survey of the Cross-Country Empirical Evidence,IMF Working Paper WP/98/169.

    • Search Google Scholar
    • Export Citation
  • Hong, Kiseok, and Aaron Tornell, 1999, “Post-Crisis Development of Asia,” Korea Development Institute, mimeo.

  • Kaminsky, Graciela, and Carmen M. Reinhart, 1999, “The Twin Crises: The Causes of Banking and Balance-of-Payments Problems,American Economic Review, 89(3), pp. 473-500.

    • Search Google Scholar
    • Export Citation
  • Krugman, Paul, 1999, “Balance Sheets, the Transfer Problem and Financial Crises,” in International Finance and Financial Crises: Essays in Honor of Robert P. Flood Jr., edited by Peter Isard, Assaf Razin, and Andrew Rose, International Monetary Fund and Kluwer Academic Publishers (Washington, DC and Norwell, MA, USA).

    • Search Google Scholar
    • Export Citation
  • Lee, Jong-Wha, and Changyong Rhee, 2000, “Macroeconomic Impacts of the Korean Financial Crisis: Comparison with the Cross-country Patterns,Economics Department, Korea University Working Paper, forthcoming in World Economy.

    • Search Google Scholar
    • Export Citation
  • Lee, Jong-Wha, Kiseok Hong, and Changyong Rhee, 2001, “The Macroeconomic Adjustment during the Currency Crises,Kyung Je Hak Yon Ku, 49(2), pp. 227-253, in Korean.

    • Search Google Scholar
    • Export Citation
  • Milesi-Ferreti, Gian Maria, and Assaf Razin, 1998, “Current Account Reversal and Currency Crises: Empirical Regularities,IMF Working Paper 98/89.

    • Search Google Scholar
    • Export Citation
  • Park, Yung Chul, 2001a, “East Asian Dilemma: Restructuring Out or Growing Out?,” Essay in International Economics, No.223, Princeton University (Princeton, NJ, USA).

    • Search Google Scholar
    • Export Citation
  • Park, Yung Chul, 2001b, “A Post Crisis Paradigm of Development for East Asia: Governance, Markets, and Institutions,” Korea University, Economics Department, mimeo.

    • Search Google Scholar
    • Export Citation
  • Park, Yung Chul, and Chi-Young Song, 2001a, “institutional Investors, Trade Linkage, Macroeconomic Similarities, and the Contagious Thai Crisis,The Journal of Japanese and International Economies, 15, pp. 199-224.

    • Search Google Scholar
    • Export Citation
  • Park, Yung Chul, and Chi-Young Song, 2001b, “Financial Contagion in the East Asian Crisis—With Special Reference to the Republic of Korea,” in International Financial Contagion, edited by Stijn Classens and Kristin Forbes, Kluwer Academic Publishers (Norwell, MA, USA), pp. 241-265.

    • Search Google Scholar
    • Export Citation
  • Przeworski, Adam, and James R. Vreeland, 2000, “The Effect of IMF Programs on Economic Growth,Journal of Development Economics, 62, pp. 385-421.

    • Search Google Scholar
    • Export Citation
  • Radelet, Steven, and Jeffrey Sachs, 1998, “The East Asian Financial Crisis: Diagnosis, Remedies, Prospects,Brookings Papers on Economic Activity I, pp. 1-74.

    • Search Google Scholar
    • Export Citation
  • Radelet, Steven, Jeffrey Sachs, and Jong-Wha Lee, 2001, “Determinants and Prospects of Economic Growth in Asia,International Economic Journal, 15(3), pp. 1-30

    • Search Google Scholar
    • Export Citation
  • Sachs, Jeffrey, and Andrew Warner, 1995, “Economic Reform and the Process of Global Integration,Brookings Papers on Economic Activity I, pp. 1-118.

    • Search Google Scholar
    • Export Citation
  • Sachs, Jeffrey, Aron Tornell, and Andrei Velasco, 1996, “The Collapse of the Mexican Peso: What Have we Learned?Economic Policy, 22, pp. 13-56.

    • Search Google Scholar
    • Export Citation
  • Stone, Mark, 2000, “The Corporate Sector Dynamics of Systemic Financial Crises,IMF Working Paper WP/00/114.

  • Williamson, John, 2000, Exchange-Rate Regimes for East Asia: Reviving the Intermediate Option, Policy Analysis in International Economics No. 60, Institute for International Economics.

    • Search Google Scholar
    • Export Citation
  • World Bank, 2000, East Asia: Recovery and Beyond, May.

  • Coase, Ronald, 1937, “The Nature of the Firm,Economica, pp. 386405.

  • Becker, Gary S., 1968, “Crime and Punishment: An Economic Approach,Journal of Political Economy, Vol. 76, pp. 169217.

  • Frankel, Jeffrey, and A. Rose, 1996, “Currency Crashes in Emerging Markets: An Empirical Treatment,Journal of International Economics Vol. 41, pp. 35166.

    • Search Google Scholar
    • Export Citation
  • Kim, Yong Jin, and Jong-Wha Lee, 2001, “A Model of Self-fulfilling Crises,” mimeo.

  • King, R., and R. Levine, 1993, “Finance and Growth: Schumpeter Might Be Right,Quarterly Journal of Economics, Vol. CVIII, pp. 681737.

    • Search Google Scholar
    • Export Citation
  • Krugman, Paul, 1999, “Balance Sheets, the Transfer Problem, and Financial Crises,” in International Finance and Financial Crises: Essays in Honor of Robert P. Flood, Jr., ed. by P. Isard, A. Razin, and A. Rose (International Monetary Fund and Kluwer Publishers).

    • Search Google Scholar
    • Export Citation
  • Lucas, Robert E. Jr., 1987, “Models of Business Cycles,” Basil Blackwell.

  • Wei, Shang-Jin, and Yi Wu, 2001, “Negative Alchemy? Corruption, Composition of Capital Flows, and Currency Crises,NBER Working Paper No. 8187.

    • Search Google Scholar
    • Export Citation

The authors would like to thank Robert Barro, Richard Portes, and participants at the NBER Conference on Management of Currency Crises, Monterey, California, USA, March 2001 for their helpful comments on an earlier draft. Si-Yeon Lee and Do-Won Kwak provided able research assistance.

Used with permission. Forthcoming from Managing Currency Crises in Emerging Markets, Michael Dooley and Jeffrey Frankel (eds.), copyright 2002 by the National Bureau of Economic Research. All rights reserved.


Lee, Hong, and Rhee (2001) describe the data in more details. we are grateful to Kiseok Hong and Changyong Rhee for sharing their cross-country data set.


The data come from Lee and Rhee (2000), which compiled the information from the IMF, Annual Report for each year.


The sample does not include the former Soviet Union countries that experienced currency crises and subsequently received the financial assistance from the IMF in the early 1990s.


The investment rate can be considered as an endogenous variable. The regression results do not change qualitatively when we exclude investment rate in the regressions. Note that investment includes public investment in addition to private investment. The regressions for investment rate are presented in Table 4.


We have also estimated another specification by using the reversal of GDP growth rate between the crisis-hit (that is, t−1 and t) and the post-crisis period, instead of post-crisis GDP growth, for the dependent variable in the regressions. we find the results do not change much.


Goldfajn and Gupta (1999) find that the use of tight monetary policy is accompanied with a sharper recovery of output during the currency crises.


A problem can occur in this regression when the participation in the IMF program is endogenously determined. To avoid this simultaneity problem, we need to use an instrumental-variable technique. We do not implement this approach yet due to lack of an ideal instrument.


They did not do so voluntarily, but at the urging of the G-7 governments and the IMF, and only when they were convinced that they would be repaid with handsome returns.


According to Hahm and Mishkin (2000), the foreign liabilities accounted for about 16% in total corporate debt in 1997 in Korea.


In 1997 the foreign liabilities accounted for about 55% of banks’ total liabilities in Korea, 27% in Thailand, and 15% in Indonesia (ADB, 2000).


The high domestic interest rate, which aims at stemming rapid depreciation, has the same devastating effect on the value of corporate sector equity and thus investment.


See Park (2001a, 2001b) for more details.


Our specification closely follows Barro (1997) in selecting the explanatory variables.


Given the way equations are formulated, the Barro tests also implicitly consider whether there are long-run effects on the level of real GDP. On average, there are no effects on either the long-run level or growth rate of real GDP.


Somewhat surprisingly, Barro’s paper implies that the trend growth rate of the Korean economy had already declined sufficiently sharply by the latter half of the 1990s that actual growth over this period (including the effects of the crisis) was above trend. As best I can tell, this reflects the effects on trend growth of the sharp decline in Korea’s terms of trade in 1996/97 as well as convergence effects.