本文主要利用馬可夫狀態轉換模型與ICSS-GARCH 模型分析美國、英國與日本股價報酬率之緩長記憶與狀態轉換的現象，並探討狀態轉換對波動持續性的影響。再根據馬可夫狀態轉換模型與ICSS 演算法所檢測出的狀態轉換點，以一般相關係數、條件相關係數與動態相關係數探討在狀態轉換下，國際股票市場的關聯性與蔓延效果。本文發現美國、英國、日本指數報酬率之條件相關係數與動態相關係數皆顯著為正，表示國際股市已經存在共同移動關係，而發生結構轉換讓相關係數顯著增加或是減少，表示國際股市普遍出現蔓延效果，建議投資人應考慮鄰近區域國家與主要股市之市場表現及金融外生衝擊為決策變數，以分散風險與達成最佳資產配置。

This research uses the long memory model, the Markov regime switching GARCH (MS-GARCH) model and iterated cumulative sums of squares (ICSS) GARCH model to investigate volatility persistence and structural changes in the American, United Kingdom, and Japan stock market. Additionally, we use time-varying correlation coefficients to estimate the constant correlation coefficients (CCC) and dynamic conditional correlation (DCC) by multivariate GARCH model. The main purpose identifies the contagion effect of international stock markets, and testes whether the mean of the estimated CCC and DCC coefficients in the period of turmoil after the crisis differs from that in the stable period before the crisis.Empirical findings show that the return has long memory and regime switching in volatility cause spurious long memory in the international stock markets. The model which considers regime switching has less persistence. There is more persistence at high volatility states than low volatility states by MS-GARCH which has better predict performance.Additionally, contagion effects existe in the international stock markets. Evidence suggests that investors need to consider financial impacts in international stock markets when making their investment decisions.