金融資產報酬通常為厚尾且非常態為主，而過去多數文獻的模型以常態分配為假設，而Copula函數能夠依據個別資料之間的關聯性找出最適之分配，使得模型的運用上更加有彈性。 本文主要分別利用傳統避險模型、固定條件相關(CCC-GJR-GARCH)模型、動態條件相關(DCC-GJR-GARCH)模型以及以Copula-based GJR-GARCH模型，利用最小變異避險理論為避險績效衡量標準，依據樣本內及樣本外進行避險比率及避險績效的實證，找出最佳的模型，提供最適的避險比率之衡量與績效評估之比較。實證結果發現以Copula 為基礎的GJR-GARCH模型的避險績效較傳統OLS模型佳。 Financial asset returns are usually fat-tailed and non-Gaussian. In the past, most of the literatures have the normal distribution assumption. The Copula functions that based on the relationship between the individual assets have more flexible than the other models to find the optimal allocation. In this paper, we use traditional hedging model, fixed conditional correlation (CCC-GJR-GARCH) model, dynamic conditional correlation (DCC-GJR-GARCH) model and the Copula-based GJR-GARCH model for the estimation of the optimal hedge ratio and the hedging performance measure by the theory of minimum variance. Empirical results show that the Copula-based GJR-GARCH models perform more effectively in the in-sample test. In addition, all of the dynamic hedging models perform more effectively than OLS model in the out-sample test.
