本研究利用樣本期間1995年至2008年台灣股票市場上市公司資料，探討如何利用風險來解釋與預測超額報酬，透過在模型中加入獨特性變異數做為影響投資機會變數之變異數的代理變數，來修正單獨利用股票市場變異數衡量風險時所高估的部分，同時亦比較不同計算方式之獨特性變異數對預測結果的影響。 研究結果顯示，僅考慮股票市場變異數時，不能預測台股超額報酬，若加入獨特性變異數後，則兩者能共同預測台股超額報酬。不同計算方式之獨特性變異數於樣本內預測時都表現良好，樣本外預測時，保留越長的樣本外期間，方能凸顯加權平均獨特性變異數優於平均獨特性變異數。 In this paper, we use the idiosyncratic variance and stock market variance jointly forecast Taiwan stock market excess returns. We construct value-weighted and equal-weighted average idiosyncratic variance, and then compare their impact on the results. Our empirical results show that the idiosyncratic variance and stock market variance can jointly forecast Taiwan stock market excess returns, but stock market variance alone does not. The performances from different ways to calculate the idiosyncratic variance are the same in the in-sample forecasts. However, if the out-of-sample periods to be retained are longer , value-weighted average idiosyncratic variance will be better than equal-weighted idiosyncratic variance in the out-of-sample forecasts.
