This paper examines statistically and systematically the five causes of the Asian currency crisis exposed by the IMF, using a probit model. The paper shows that the two causes of the IMF are persuasive. The Asian currency crisis tends to occur when the ratio of foreign reserve to total debt is low and the progress of financial deregulation without regularity is great. The paper also shows that the general causes found by previous researches never fit the case of this crisis, but that the trade linkage of each country to the first victim country helps explain the causes of this crisis.
Some Asian currencies in the wake of the flotation of the Thai baht in 2 July 1997 collapsed, and the currency crisis spread to other countries in the Asian region. To restore these affected countries, the International Monetary Fund (IMF), World Bank (WB), Asian Development Bank (ADB) and the G-7 countries provided the US$ 117.7 billions of assistance together with the program of economic reform. The IMF (IMF, 1998a, pp. 23–25; IMF, 1998b, pp. 82–106; WB, 1998a, pp. 1–72; ADB, 1998a, pp. 19–37) reviewed the causes of this crisis as follows: (i) current account deficits as a trigger, (ii) large external debt (overdependence on short-term capital flows) and a great progress of financial liberalization without regularity as expressing a fragile financial sector and (iii) maintenance of relatively fixed exchange rates and of rather high interest rates as expressing policy weakness.1 However, the IMF did not test statistically these causes of the crisis.
The purpose of the paper is to examine statistically and systematically the causes of the Asian currency crisis exposed by the IMF, using a probit model.
The recent literatures on currency crises tend to capture the cause of crises and the prediction for the outburst of crises in terms of probability, by using the probit model.2 This trend of research is in the practical and political requests. Frankel and Rose (1996) and Eichengreen et al., 1995 and Eichengreen et al., 1996 used data from 1971 through 1993 to characterize common causes of many currency crises. Crises tend to occur with high probability, depending on a low output growth, a low level of foreign direct investment, a high inflation rate and a high growth of domestic credit.3Glick and Rose (1998) examined the Asian currency crisis in 1997 together with the Russian currency crisis in September 1998 and the Brazil currency crisis in January 1999, though they did not examine the Asian crisis in isolation.4 Therefore, there is naturally no papers related to the statistical test of the causes of the Asian currency crisis exposed by the IMF. In Section 2, we define the Asian currency crises, explain the statistical methodology and describe the variables for the causes (focusing on the causes of (i) and (ii) exposed by the IMF) and the dataset. In Section 3, we test and evaluate the causes of the crisis, compared with the previous researches and alternative causes of the crisis. In Section 4, we draw some conclusions.
We have shown statistically and systematically by a probit model that the low ratio of foreign reserve and the great progress of financial liberalization without regularity out of the five causes of the Asian currency crisis exposed by the IMF are persuasive. These are measures of a fragile financial sector. However, the other causes, current account deficits and large external debt (overdependence on short-term capital flows), are not denied. The general causes pointed out by Frankel and Rose (1996), Eichengreen et al. (1996) never explain well the Asian currency crisis. In this sense, the IMF model gives a best explanation to this crisis.
The concept of the trade linkage to the first victim country (Thailand) found by Glick and Rose (1998) helps explain the causes of the crisis. When we added their original idea of TRADE1 to the IMF model, this revised IMF Model 1 was not more effective in explaining the Asian currency crisis. However, if the slightly revised our idea of TRADE2 is added, the revised IMF Model 2 obtains a better performance in a point that the two types of R2 are greatly improved and the effects of the variables on the probability of crisis are significantly larger (more obvious). The robustness of all results is supported by OLS regression of exchange rate depreciation on these variables.
The implication of the results of the revised IMF Model 2 is as follows. Two hypotheses for the causes of the Asian currency crisis exist. According to one view (Radelet and Sach, 1998a, Radelet and Sach, 1998b, Ito, 1998 and Ito, 1999, the extent and depth of the crisis should not be attributed to a deterioration in fundamentals (i.e. a fragile financial sector), but rather to panic on the part of domestic and international investors (i.e. panic seems to pass contagiously from Thailand to other countries which have a strong trade linkage to Thailand). The other view (ADB, 1998a, Corsetti et al., 1998, IMF, 1998a and IMF, 1998b is based on deterioration in fundamentals caused by the structural and policy distortions. The revised IMF Model 2 supports simultaneously the two hypotheses and provides the compromise to this controversy.
