本篇文章比較兩種非線性之共整合模型(包括考慮狀態變數改變的修正性共整合模型及不對稱門檻共整合模型)對於股票指數與股票指數期貨之間的長期關係衡量,同時也比較此二種模型是否比原始簡單共整合模型更具有較好的衡量能力。並利用三種不同共整合模型來衡量2003年至2010年之間台灣MSCI指數與恆生股價指數及此二指數之期貨指數的長期關係。本篇文章主要的實證結論有兩項。第一、實證中說明股價指數與期貨指數之間具有共整合之效果,亦即兩者之間具有長期的關係,且整體而言,非線性共整合模型對於衡量股價指數與期貨指數之間的關係效果較好(與線性共整合模型相比)。第二、在兩種非線性共整合模型中,於衡量股價指數與期貨指數之間的長期關係而言,考慮區域轉換共整合模型之衡量績效比門檻共整合模型更好。同時,區域轉換共整合模型也偵測到2007到2008年次順位債券風暴對於此而金融市場所產生的衝擊與影響。
In this paper, two non-linear cointegration models, including a modified cointegration method considering regime shifts and a threshold cointegration model, are compared with each other. Also, the results of the two non-linear modelsare compared with the original simple cointegration model to check if they are better. Thus, three models are applied to estimate the long-term relationship between two equity index returns and their corresponding futures index returns. Two data sets including MSCI Taiwan equity index and Hang-Seng equity index, from the first of 2003 to the end of 2010, are employed in this study. In the empirical test, we obtain two major results. The first, the evidence shows that the cointegration exists between the index returns and corresponding index futures returns. Generally, the estimation performance of non-linear cointegration model is between than the one of linear cointegration. The second, the cointegration model with regime shifts generally performs better than threshold cointegration model, although the later one really has some advantage in examining the asymmetric effect of residual sequence. Furthermore, the regime switching model captures the impact of subordinated-debtcrisis in 2007 to 2008.