The study on the relationship between American and Canadian stock index markets

[[abstract]]本研究選取美國道瓊工業指數、加拿大標準普爾/多倫多證交所 (TSX)綜合指數和摩根史坦利公司所編製的世界指數為探討，採用1990年1月3日至2006年12月27日之週指數資料。以美國三個月期的國庫券利率為無風險利率的替代變數。利用ADF單根檢定法、共整合檢定、ARCH效果、ICSS演算法、單變量GARCH-CAPM模型、向量自我迴歸模型與雙變量ICSS-BEKK模型，分析美國股市與加拿大股市是否具有投資風險分散的效果，並且探討美國道瓊工業指數超額報酬率與加拿大TSX綜合指數超額報酬率之間一階動差的相互關係與二階動差的關聯性。更進一步地，以實證說明美國道瓊工業指數超額報酬率與加拿大標準普爾/多倫多證交所綜合指數超額報酬率之間是否存在顯著的波動外溢效果。經由實證分析本研究將結論歸納如下三點： 一、在單變量GARCH-CAPM模型中，我們發現模型內若有考量波動性的轉變，能有效的降低美國道瓊工業指數超額報酬率與加拿大TSX綜合指數超額報酬率波動持續性的問題。 二、在VAR模型一階動差的分析中，美國道瓊工業指數超額報酬率與加拿大TSX綜合指數超額報酬率之間具有均數相依的關係，為雙向的因果關係。 三、在雙變量GARCH模型 (ICSS-BEKK)的實證分析中，發現美國道瓊工業指數超額報酬率與加拿大TSX綜合指數超額報酬率之間確實存在顯著的波動外溢效果。[[abstract]]In this paper, we choose American Dow Jones Industrials Index, Canadian S&P/TSX Composite Index and MSCI World Index. Using weekly Wednesday-close stock index value from 3 January, 1990 through 27 December, 2006 to examine the relationship between American and Canadian stock index markets. It is well known that volatility persistence is overestimated if regime shifts are ignored in the standard GARCH model. Therefore, we use an iterated cumulative sums of squares (ICSS) algorithm to detect time periods of sudden changes in the variance of excess return in American and Canadian stock index markets. According to the empirical findings, we conclude the results as following: (1)In univariate GARCH-CAPM model, our findings indicate that accounting for volatility shifts can usefully reduce the persistence in volatility. (2)In VAR model, we find that there exists a bi-directional causal relationship between American and Canadian stock index markets. (3)When we incorporate the information obtained on sudden changes in volatility in a bivariate GARCH model, we find the evidence of significant volatility spillover effect between American Dow Jones Industrials index and Canadian S&P/TSX Composite Index. These findings have important theoretical and practical implications for investors and portfolio managers to make international investment decisions more correctly