近年來國際原油價格劇烈的波動,常導致投資人承受巨額損失,因而使原油期貨成為避險的金融商品之一。由於原油價格常因稀少事件,而產生價格不連續現象,本研究利用Chan(2003)的雙變量CBP-GARCH模型,估計最小變異數避險組合條件風險值,最後藉Kupiec(1995)的概似比檢定法與Christoffersen(1998)的條件涵蓋檢定法進行回顧測試,以評估風險值模型的準確性。 研究發現,雙變量CBP-GARCH模型的最小變異數避險組合條件風險值模型通過回顧測試,而未避險模型與雙變量DCC-GARCH模型均未通過回顧測試。有鑑於此,雙變量CBP-GARCH模型的最小變異數避險組合條件風險值模型準確性較高,此乃因雙變量CBP-GARCH模型能捕捉跳躍動態過程與跳躍相關之故。因此若僅考慮資產價格間的動態波動性過程,會造成低估風險現象而容易使投資人承擔超過預期的損失,此結果可作為投資人避險的參考。 International crude oil prices of volatility severely make investors bear huge loss in recent years. Thus, crude oil futures become one of financial instruments of hedge. The crude oil prices bring out discontinuous phenomena, because of the rare events. In this study, it estimates the conditional value-at-risk of the minimum variance hedging portfolio by using the bivariate CBP-GARCH model proposed by Chan(2003). Moreover, this study is to evaluate the accuracy of the bivariate CBP-GARCH model by using backtesting method based on likelihood ratio test proposed by Kupiec(1995) and conditional coverage test proposed by Christoffersen(1998).The empirical results are as follows. The conditional value-at-risk model of the minimum variance hedging portfolio by using the bivariate CBP-GARCH model passes the backtesting; however, the conditional value-at-risk model of the minimum variance hedging portfolio by using non-hedge model and the bivariate DCC-GARCH model do not pass. The conditional value-at-risk model of the minimum variance hedging portfolio by using the bivariate CBP-GARCH model has high accuracy; it is because that it can capture dynamic jump process and jump correlation. Therefore, if we just consider the dynamic volatility process, it could underestimate risk and let investors bear the loss than expected. This result can be used as a hedge reference for investors.