本研究目的是想運用Box-Cox與Log在誤差修正模型,不同的轉換方式,來探討臺灣加權股價指數期貨避險比例有何異同。本研究根據臺灣期貨交易所與臺灣證券交易所,從2007年6月1日至2009年12月31日的資料,共648個交易日。首先,我們先利用ADF檢定方式探討現貨與期貨的價格是否有單一根(UNIT ROOT)的特性。第二步,利用Box-Cox與Log兩種轉換方式,在誤差修正模型上建立長期與短期均衡的關係。第三步,我們比較兩者不同轉換之避險比例,利用平均絕對相對誤差(MARE),平均絕對誤差(MAE),均方誤差(MSE)三種衡量方法來比較,何種轉換方式較好。結果顯示Box-Cox它的避險結果會比Log來的好。
This paper is devoted to examine the forecasting performance of the future hedge ratio by using Box-Cox power transformation in Frror Correction Model (ECM) forecasts, in comparison with log-transformation. We examined the daily date from Taiwan Security & Fulures Exchange covering the period from June 1, 2007 to Dec 31, 2009. First, we test whether both Spot and Futures prices are unit root by using the ADF test. Second, we modeled the iong-term relationship and the shot-term relationship with Box-Cox and Log transformations. Third, we compared the performance of two-hedge rations by using mean absolute relative error (MARE), mean absolute error (MAE), and root mean square error (RMSE). Our results showed that the Box-Cox power transformation provides outperforming forecasts in Error Correction Morel (EMC) and is recommendable to forecasting practice.