本研究利用Bollerslev (1990) 所提出之固定相關 ( Constant Conditional Correlation,CCC ) 及 Engle (2002) 提出動態條件相關係數 ( Dynamic Conditional Correlation,DCC ) 多變量GARCH模型估算人民幣和港幣、人民幣和美金、人民幣和歐元、人民幣和新加坡幣、港幣和歐元、港幣和新加坡幣、美金和新加坡幣、歐元和新加坡幣兌新台幣之匯率,所組成的八組外匯投資組合風險值。比較CCC-GARCH和DCC-GARCH兩模型於GARCH(1,1)基礎下,不同信賴水準之預測風險值的能力,並以回溯測試應用Kupiec 檢定及RMSE 資金運用效率,兩評估風險預測績效指標做分析衡量。實證結果顯示DCC-GARCH(1,1)-N模型相較能處理金融資產厚尾和波動叢聚現象,表風險控管績效能力較優,因此適合選作外匯投資組合風險值預測績效模型。
In this study, use the Constant Conditional Correlation (CCC) proposed by Bollerslev (1990) and Dynamic Conditional Correlation (DCC) proposed by Engle (2002) Multivarite GARCH Model, to estimate on exchange rate CNY and HKD / NTD, CNY and USD / NTD, CNY and EUR / NTD, CNY and SGD / NTD, HKD and EUR / NTD, HKD and SGD / NTD,USD and SGD / NTD, EUR and SGD / NTD, composed of Foreign Exchange Rate Portfolio for Value at Risk (VaR). By comparing the CCC-GARCH and DCC-GARCH two models based on the GARCH(1,1), different confidence levels of predict ability to Value at Risk, based on the Kupiec in back-testing and RMSE for capital efficiency, two risk prediction performance indicators to assess for measure analysis. The result shows that DCC-GARCH(1,1)-N model compare to deal with financial assets for feature fat-tail and volatility clustering, is the better ability to risk performance control, therefore suitable selected as forecasting performance foreign exchange rate portfolio for Value at Risk model.