1. Introduction

When Thailand decided to float its baht, not many observers foresaw the real magnitude of the impending crises that later battered several Asian and Latin American countries; ironically most of these countries had been believed to be successful models of economic growth. The float of the Thai baht initially triggered depreciation of nearby countries’ currencies, including the Malaysian ringgit, the Indonesian rupiah, and the Philippine peso. At that time, depreciation of Southeastern Asian countries currencies was viewed, at best, as "correction" of overvalued currencies, and was believed to help to manage external balances. It was in late October 1997 that revealed the true nature of crises when Taiwan abandoned intervention on behalf of the New Taiwan dollar, which triggered speculative attacks on the Hong Kong dollar, the Korean won and some of Latin American currencies, especially the Brazilian real and the Argentine peso. These speculative attacks intensified in December 1997 when it became known that Korea’s reserves were almost depleted. By following January, the Indonesian rupiah lost 81 percent of its value relative to its July 1, 1997 level. Similarly, the Thai baht, the Malaysian ringgit, and the Philippine peso all lost their values by 56 percent, 46 percent, and 41 percent, respectively. During the short interval from October 1 to late December 1997, the Korean won lost its value by 55 percent. While these episodes of rapid depreciation of these currencies that were severely affected are well known, it needs to be pointed out that the volatility of these affected currencies also sharply increased to unprecedented levels during the crises and, more notably, the increase in volatility occurred almost simultaneously with the affected currencies.

This paper will investigate the linkage of the volatility of exchange rates of affected countries during the crisis. There can be many reasons why volatility of exchange rates is linked each other. One reason is that fundamentals of exchange rates, especially trade and investment, are related, and therefore any new information about fundamentals might affect the volatility of corresponding countries at the same time. Since foreign exchange markets are almost perfectly integrated on a 24-hour trading basis, a change in volatility of one currency due to the new information about fundamentals should be transmitted to the other currency’s volatility simultaneously. Most explanations that emphasize the importance of macroeconomic fundamentals view a contagious nature of currency crises in similar ways. However, the magnitudes of crises have often been observed to be too large to be explained by fundamentals only (Sachs, Tornell and Velasco 1996). The other reason for linkage between volatility of exchange rates is "market psychology." Even if there were no apparent common fundamentals between currencies, but speculations based on fads, noises or herd instinct might be transmitted. If, during the crises, foreign exchange traders are unable to identify the source of shocks, especially between global shocks and local shocks, trying to infer the source of shocks from observing exchange rates could lead to correlation between exchange rate volatility and transmission of idiosyncratic errors to the other exchange rates (King and Wadhwani 1990). Also, it has been pointed out that mere existence of currency crises in other countries increase the probability of attacks on domestic currency even after taking fundamentals into consideration (Eichengreen, Rose and Wyplosz 1996).

Hypothesis and Model

The model we will estimate is based on Engle, Ito and Lin (1990) and Kroner and Ng (1995). where is an exchange rate of country i at time t+1 and is an error term at t+1 and assumed to follow conditional hetersckedastistic process with being conditional variance of . Then we have

.

is i.i.d with

. (1)

Specifically, this paper will address the following questions. First, how volatility of each country’s currency is affected by other countries’ currencies? Because the implications on exchange rate volatility arising from common fundamentals as opposed to from market psychology, we can test these two views directly by examining the effect of past volatility of foreign countries’ on current volatility of the domestic country, i.e., . One problem that might arise is that even if volatility seems to transmit from one country to another, it might not mean that the volatility is due to the market psychology because two countries might have very similar fundamental problems. We will overcome this problem by the use of high frequency data. Second, we can compare the temporal behavior of volatility transmission for pre-crisis and post-crisis periods. If temporal behavior of volatility transmission (and ) is systemically different, then support for the fundamentalist’s view might be weakened. In addition to parameter stability test, we can also test for constancy of conditional correlation coefficients. Third, we can also examine impulse response of volatility during pre- and post- crises, which will show persistence of the effect of other currencies’ on the volatility of domestic currency. If fundamentalist view is correct, then the effect of previous domestic volatility should be greater. .

Data