I. Introduction According to conventional wisdom, the rigidity of pre-1997 pegged exchange rate regimes in East Asia was one of the main causes of the currency crisis. (1) The basis for assuming that East Asia's pre-crisis exchange rates were rigid is the declarations of central banks about exchange rate policies. In practice, however, the declared exchange rate regime itself may not indicate rigidities--a tightly controlled managed float regime may be more rigid than a pegged regime with a wide band, ceteris paribus. As Ghosh et al. (1995) and Reinhart (2000) point out, declared exchange rate regimes may differ from the actual characteristics of exchange rates. Therefore, it is worthwhile to pursue a statistical examination of East Asia's pre-crisis exchange rates in order to identify their distinctive characteristics. We also investigate the issue of whether systematic relationships existed among East Asia's bilateral exchange rates vis-a-vis the U.S. dollar in the pre-crisis period. For example, was there a systematic relationship between the Malaysian ringgit-U.S, dollar exchange rates and the Thai baht-U.S, dollar exchange rate? The existence of such systematic relationships in the pre-crisis period would lend further support to the contagious nature of the Asian currency crisis. At the same time, the pre-crisis existence of relationships raises the interesting question of how the peg regimes were able to last so long in light of their vulnerability to external shocks. Evidence shows that intra-day, daily, or weekly exchange rates are approximated by martingales, and the returns exhibit conditional volatility, fat-tail distribution, skewness, and excess kurtosis. Autoregressive conditional heteroscedasticity (ARCH) models have been used successfully to describe foreign exchange rate data. More recently, Engle and Gau (1997) examine the conditional volatility of the exchange rates under the target zone regime of the European Monetary System (EMS). They find that exchange rates within the EMS show the same characteristics as free-floating exchange rates except for strong negative autocorrelations. Conditional volatility in exchange rates has been attributed to the irregular arrival of new information in the market. Engle, Ito, and Lin (1990) interpret volatility persistence in foreign exchange markets to either the time it takes market traders to act on new information or the autocorrelation of news across countries arising from potential policy coordination. They employ Ito and Roley's (1987) decomposition of the intra-day yen-U.S, dollar exchange rates based on the operating times of different markets to show volatility spillover across markets. On the other hand, Baillie and Bollerslev (1990) examine systematic relationships in the conditional returns or variances of the hourly bilateral exchange rates of the United Kingdom, Germany, Switzerland, and Japan vis-a-vis the U.S. dollar. They use the robust inference procedures of Wooldridge (1990) and Bollerslev and Wooldridge (1992) in order to handle the skewness and excessive kurtosis in the data. The results fail to show systematic relationships among the four major exchange rates. Our paper investigates the behaviour of the precrisis daily bilateral exchange rates of four Newly Industrializing Countries (NICs) of Hong Kong, South Korea, Singapore, and Taiwan, and four members of the Association of Southeast Asian Nations (ASEAN-4), Indonesia, Malaysia, Philippines, and Thailand, vis-a-vis the U.S. dollar. We perform preliminary analyses and robust misspecification tests to explore the distinctive characteristics of East Asian exchange rates. We also look for empirical evidence of systematic relationships among the region's precrisis bilateral exchange rates vis-a-vis the U.S. dollar. Although our paper is essentially a study of the behaviour of exchange rate rather than exchange rate policy or regime, we do attempt to draw some policy implications from our results. …
Read full abstract