بازیابی بحران ارز: برخی از حقایق تجربی

Using a comprehensive data set of over 100 developing countries, this paper examines how economies recover from a currency crisis. Some of the regularities found are listed below. First, it takes less than 3 years for the GDP growth rate to fully recover its tranquil-period average. The level of GDP, however, remains permanently below its initial trend, suggesting that the shocks underlying a crisis are persistent. Second, the credit crunch problem lasts much longer than the reduction in the GDP growth rate. Third, the pre-crisis credit expansion and reserve inadequacy, which have been widely recognized as the main causes of a crisis, also are closely related to the extent of the post-crisis recession. Fourth, crisis episodes that are caused mainly by illiquidity, rather than by insolvency, tend to exhibit a sharp recovery.

The goal of this paper is to examine how economies evolve after the eruption of a currency crisis. Since the Tequila and Asian crises, there has been a resurgence of interest in currency crises. However, most of the studies on currency crises examine the causes of a crisis, and the issue of what happens after a currency crisis still remains relatively unexamined. How long does it take for the economy to fully recover from a crisis, and which economies recover faster than others? This paper aims to answer these questions by providing some stylized facts about macroeconomic adjustment of crisis-hit economies. Some of the previous studies on currency crises documented how various macroeconomic indicators evolve before and after a crisis (Eichengreen et al., 1995, Aziz et al., 2000 and Frankel and Rose, 1996). These studies focused on the causes of a crisis, however, and only briefly examined the aftermath of the turbulence. Among the existing studies that looked more directly at the post-crisis adjustment, Bordo et al. (2000) examined the average duration and the severity of a currency crisis. However, this study only investigated the variation in the recovery process over time (between the pre-1973 and the post-1973 periods) rather than the variation between countries. In addition, while Gupta et al. (2001) and Lee and Park (2001) examined the cross-country differences in the pace of recovery, their results are likely to be biased because of the endogeneity problem. Gupta et al. (2001) used the difference in the GDP growth rate between three pre-crisis years and two post-crisis years as a measure of the pace of recovery (the dependent variable) and various macroeconomic indicators in the three pre-crisis years as the explanatory variables. Thus, what they estimated out may be simply the correlation between the explanatory variables and the pre-crisis GDP growth.1Lee and Park (2001) estimated an equation that relates the GDP growth rate in post-crisis years to contemporaneous macroeconomic indicators. In this paper, using a comprehensive data set of roughly 100 developing countries, we re-examine the macroeconomic adjustments in the aftermath of a currency crisis. In particular, we choose our variables so that the possibility of the aforementioned endogeneity problem is reduced. Additionally, unlike most previous studies, we pay special attention to possible outliers in the data. We believe that careful treatment of outliers is particularly important in analyzing a data set from many developing countries. The main findings of this paper are listed below. First, although GDP growth falls initially after a crisis, it fully recovers its tranquil-period average in less than 3 years. This suggests that the effects of a currency crisis on growth are not as persistent as is often believed. The level of GDP, however, tends to be permanently lowered by a crisis since the growth rate in the post-crisis period does not exceed its tranquil-period average. A “technical rebounding” or return to the initial growth path is not a common occurrence in our data. Second, the credit crunch problem is much more persistent than the reduction in the GDP growth rate. The real credit growth, for instance, does not show any clear improvement over the post-crisis years that we consider. Third, the extent of the investment/lending boom and reserve inadequacy in the pre-crisis period is important for the pace of post-crisis growth. Existing studies have reported that the extent of domestic credit growth and the relative magnitude of the short-term debt and liquid assets are two of the most robust indicators of a future currency crisis. We find that these factors are important in determining the pace of post-crisis growth, as well. Among the two factors, the pre-crisis investment/lending boom appears to be a more robust determinant. Fourth, currency crises that lead to a so-called V-shaped recovery (an initial drop followed by a sharp pickup in the growth rate) tend to be caused mainly by illiquidity, whereas a U-shaped recovery (an initial drop followed by sluggish recovery) is typically preceded by an investment boom. This paper is organized as follows. Section 2 presents the definitions of the crisis and the data set used in this paper. Section 3 reports stylized facts about the macroeconomic adjustments during the 3 years after the eruption of a crisis, and Section 4 examines why the pace of post-crisis growth differs across economies. The V-shaped recoveries and the U-shaped recoveries are examined in Section 5. Section 6 concludes the paper.

Most existing studies on currency crises investigate what causes a crisis and pay little attention to what happens afterwards. In this paper, we examined the macroeconomic adjustments of countries in the aftermath of a currency crisis. In addition, while previous studies emphasized the turbulence in the financial sector, we focused on the real sector of the economy represented by the GDP growth rate. Our findings suggest that a currency crisis has long-lasting effects on an economy. Even if the growth rate regains its potential value after 2 or 3 years of recession, the level of GDP remains permanently below the initial growth path that would have prevailed in the absence of the crisis. Both real domestic credit and foreign borrowings remain below the previous trend for many years, resulting in a prolonged credit crunch. We also found that, across crisis episodes, the severity of recessions can be systematically explained by various initial conditions. In particular, pre-crisis credit expansion and reserve inadequacy, which previous studies named as the key causes of a crisis, appear to affect the pace of post-crisis growth as well. This suggests that economies with a greater ex ante probability of a crisis are also more likely to experience a deeper recession. In addition, we examined the relative importance of illiquidity and insolvency in causing a crisis and their linkage with the V-shaped recovery. Using credit expansion and reserve inadequacy as the proxy variables, we found that liquidity crises are likely to exhibit a V-shaped recovery. Many issues are left for future research. For instance, the endogeneity of policy variables has not been properly handled in our analysis. Using time-lagged values of the policy variables will not solve the endogeneity problem as long as expectations on future performance of the economy are embodied in the current values. Future research is needed to resolve this problem. Additionally, it will be important to investigate the mechanism through which a pre-crisis investment boom deters the post-crisis recovery.