本研究主要探討條件變幅自我迴歸(Conditional AutoRegessive Range CARR)模型與一般化自我迴歸條件異質變異數(Generalized AutoRegressive Conditional Heteroskedasticity GARCH)模型,在比較富邦金控股市報酬波動的預測績效何種模型較為準確,樣本期間為2011年1月3日到2015年3月13日。藉由單根檢定、ARCH模型檢定,來檢測兩模型對股市報酬的預測績效有無顯著差異。 實證結果發現CARR模型與GARCH模型的預測績效與波動預測能力比較上,CARR模型較GARCH模型更能預測績效與掌握波動性。此一推論與Chou(2005)的結論具有一致性,在捕捉波動性變動的過程上,CARR模型的表現皆優於GARCH模型。
The current study aimed to compare the degree of accuracy that CARR and GARCH models had on volatility forecasting performance of stock returns of Fubon Financial. The sample period covered from January 3rd , 2011 to March 13th, 2015. Unit root test and ARCH model test were employed to examine if significant differences existed in the forecasting performance of stock returns between the two models. The empirical results revealed that in terms of forecasting performance and volatility forecasting, CARR performed significantly better than GARCH, echoing Chou’s (2005) statement that in the process of capturing volatility, CARR was superior to GARCH.