本研究應用Hamilton and Susmel (1994) 提出的馬可夫轉換自我迴歸異質條件變異數 (SWARCH) 模型來解釋台灣股票市場日報酬率的浮動。傳統ARCH模型的條件變異數通常存在高持續性，但假使有任何結構性變動，此時高持續性可能出現虛假情形，例如政策改變或重大事件發生。本研究要估計變異數持續性及狀態轉換的影響。實證結果得到 SWARCH模型比傳統ARCH模型能提供較好的資料以減少變異數持續性，並且在狀態轉換下此模型可掌握當時重大事件所造成的股市波動。 This paper examines the volatility of daily stock market returns in Taiwan using a Markov Switching Autoregressive Conditional Heteroscedastic (SWARCH) model developed by Hamilton and Susmel (1994). Conventional ARCH models usually exist the highly persistence to conditional volatility but this persistence may be spurious if there is any structural change in the conditional volatility, such as policy changes or news events. We investigate the evolution of volatility persistence and the effect of regime shifts. Empirical results show that the SWARCH model provides a better description of the data, which implies a much lower degree of volatility persistence than conventional ARCH models. Furthermore, the volatility regimes identified by our model appear to correlate well with major events.
