[[alternative]]Co-Movements of Taiwan and International Stock Markets

[[abstract]]本研究的目的是在探討臺灣、美國、日本、香港及深圳股市間股票報酬之共移性現象。應用Engle and Kroner (1995)所提出的一般正定多變量GARCH模型為主要的實證模型，並將模型中條件誤差項之分配假設為雙變量t分配，藉由對條件共變異數與條件相關係數之探討來驗證臺灣、美國、日本、香港及深圳股票市場間股票報酬共移性的時間變異特性，並獲致下列的結論：(1)除了日本與深圳股市這一組外，臺灣、美國、日本、香港及深圳股市間不論長期或短期彼此間都有相關性。(2)本研究不但印證各國股市間相關性非固定不變的現象；而且發現十個組合之條件相關係數存在正相關的機率會大於負相關的機率。(3)最後在俄羅斯金融危機（Russian financial crisis）導致美國股市在1998年8月31日星期一崩盤的事件分析中，顯示若投資者參考每日條件相關係數改變的過程來調整最適資產負債組合，則可擴大國際投資組合風險分散的潛在利益。[[abstract]]This paper examines the co-movements of stock returns between each pair of stock markets within five international markets by using generalized positive definite multivariate GARCH models. The errors are assumed to follow a multivariate Student-t distribution. We find that all pairs of markets, except Hong Kong-Japan, display significant permanent and transitory covariance. We also find that while conditional correlations between the returns are generally small, the correlations change considerably over time. An event analysis suggests that using diversification strategies for these conditional correlations is potentially beneficial.[[notice]]補正完