本文以台灣股票指數的資料為研究對象，結合Engle and Lee (1993)的要素模型與Chan and Maheu (2002)的ARJI模型組合成ARJI-Trend模型，並加入Bai and Perron(2003)的結構轉變分析，來探討金融資產報酬的特性。 實證結果證明：1.條件變異數與跳躍強度是隨著時間的改變而變動，並非固定不變。2.條件變異數中的恆常要素與短暫要素確實存在；且短暫要素會比恆常要素收斂得快，條件變異數會收斂為一個常數，符合趨勢本身為定態的條件。 藉由本文的實證結果，以期對ㄧ般法人機構或是投資人在進行對台灣股市投資時，尤其對股票衍生性商品之評價、避險、與交易等，能有較多的資訊作出正確的投資決策。 This article regards the stock index in Taiwan as the research object. we use ARJI-Trend model combining Engle and Lee (1993)Component model with Chan and Maheu (2002)ARJI model and a structural break analysis by Bai and Perron(2003), to study the identities of assets return volatility. The evidence significant indicates that : 1. We find significant time variation in the conditional variance and the conditional jump intensity. 2. We find that both permanent and transitory components of the conditional variance are really exist in the whole sample period; Transitory component will decay faster than the trend, the condition variance will eventually converge a constant since the trend itself is stationary. With the research result of this article, we expect to provide some organizations and individuals with more information to make better investment decisions in Taiwan equity market, especially in derivatives pricing, hedging, and trading ,etc.