多數實證研究所使用的波動估計模型，皆隱含假設在樣本觀察期間內，波動（或波動的參數）為固定不變，此隱含假設了市場結構是沒有發生改變，也就是波動路徑的形式是一致的，本研究認為此隱含假設其實並不盡合理。因此，本文將傳統計量方法中，對於結構性改變之檢定運用至此情況，以檢測樣本期間內是否存在著結構性改變。若確實存有結構性改變，則找出結構性改變之時間點，據此將研究期間加以分割，求得並非永遠是固定不變之待估參數，進而求出更適合模型的波動，以增加模型的績效。本研究以Black-Scholes與Hul-White兩評價模型分別搭配歷史波動與GARCH模型，對台灣股價指數選擇權進行實證研究，所得之實證結果如下：(1) 在樣本期間內，對歷史波動模型，本文找出了3個結構性改變時間點，運用在Black-Scholes模型下，發現除了遠月賣權之少數情形外（例如深價內與價內），平均而言，“考慮結構性改變後”之訂價誤差明顯小於“不考慮結構性改變”。(2) 對GARCH模型，則找出了1個結構性改變點將全樣本區分為2個區段，運用至Hull-White評價模型下，也有同樣之結論，即“考慮結構性改變”能明顯改善模型的績效，並降低訂價誤差。。自從Black, F., and M. Scholes，在1973年發表了著名的B-S模型後，後續許多學者為了減少訂價誤差、改善模型，無不費盡心思；但在繁複的數學代價下，所獲得的可能只是誤差的些微改善。在此本文提供了另一種思維來增進績效，我們不改變先天模型的設定，而是從後天實務操作上著手做修改。 In traditional approaches the parameters of volatility model are usually assumed to be constant during the empirical study period, which implies that market structure has not been changed over the time. In practice, we can see the structure change easily happen in financial (options) market. This phenomena intrigue our interesting. In conventional approaches we usually remedy the models to improve the performances, therefore, there are many complexity models has been developed. But in practice their improvements are often restricted. This article focuses on the structure change point and provides an alternative thinking to enhance the performance. From the empirical study, we can see even the simple Black-Scholes model we can improve its performance radically. This research applies Black-Scholes and Hull-White models on TAIEX Options. In which the historical volatility model and GARCH model are used for estimation. In Black-Scholes model (applied with historical volatility), we find three structure change points during 1 July 2004 to 31 July 2005. The performance of the model is significantly better than the model that has not been considered with structure change, especially for the deep-out-the-money pricing of the call options. The reason could be that the buyers are more sensibility than the others. In the GARCH model we find one change point, similar to the historical volatility model, these models has been considered with structure change are significantly better than the others.
