한국, 미국, 일본과 아시아 금융위기 국가의 주식시장간 조건부상관관계 추정과 정보확산속도에 관한 연구

As the financial markets of emerging nations have become internationalized and integrated nowadays, much prior research shows that co-movements between the international stock markets have sharply increased, especially during financial crises such as Black Monday. This suggests that the analysis of the time-varying conditional correlation is important. The timevarying conditional correlation depends on both the degree of connection between nations and composition of optimal portfolios for asset allocation and risk management. In order to show how the co-movements between financial markets have been during the financial crises, we have investigated dynamic conditional correlation for stock returns of sample nations during the Asian crisis period and IT bubble crisis. We also examine the contagion effect and the relationship between the information diffusion speed and the contagion effect. The analysis period for the paper is from January 1990 to December 2007. The sample data include Korea, three Southeast Asian nations (Thailand, Indonesia, and Malaysia) which have experienced the Asian financial crisis, the U.S., and Japanese stock markets, which can be considered advanced stock markets. We estimate dynamic conditional correlation between sample nations. We also use an estimator, which is calculated through the Fisher transformation, in order to analyze the contagion effect. Also, we test information diffusion speed by the autoregressive structure of timevarying volatility of the GARCH model. Finally, we test the robustness of the contagion effect using weekly stock returns as well as daily stock returns, and analyze whether or not the increase in volatility during the crisis period affects the contagion effect beyond the information diffusion speed. Empirical results can be summarized as follows. First, the estimation of the dynamic conditional correlations between sample nations satisfies the necessary conditions of a model estimator. Furthermore, the conditional correlation is time-varying as time passes. Second, the contagion effect among Korea and Southeast Asia (Thailand, Indonesia, and Malaysia) is significant but not between these nations and the U.S. This result implies that the Asian nations are strongly connected with the U.S. economy during the Asian financial crisis. Third, we have investigated how the information diffusion speed has impacted on the contagion effect. The information diffusion speed of the three Southeast Asian nations is faster than Korea and Japan during the Asia crisis period. This indicates that the information diffusion speed depends on the relative size of the economy. To check whether the results are robust across the frequency of the data set, we use the weekly stock returns beyond the daily stock returns. The estimation results of the weekly data are slightly different from the previous results of the daily data. This may suggest that when non-synchronicity of trading and the institutional differences of stock markets are significant, the use of unsuitable data sets for analysis may not guarantee reliability for risk management and international portfolio management in which the dynamic conditional correlation plays an important role.