本研究透過Richard, Evarist 及 Jon(2007)所提出的simplified multivariate generalized autoregressive conditional heteroscedasticity(S-GARCH)模型進行期貨避險,利用其模型特性,將多變量的參數估計式透過單變量的方式簡化,解決一般多變量GARCH模型在多資產的擴展上容易增加過多的參數導致自由度的損失以及數學估算上的困難,藉此檢測S-GARCH模型除了估算容易外,相較於其他模型是否亦具備較良好的避險績效。 本篇研究使用三種不同產業類別的期貨現貨日資料進行雙變量期貨避險分析,研究資料取自於Datastream。依本篇實證結果指出,S-GARCH模型在指數期貨、外匯期貨及農產品期貨中,其模型不但收斂容易,並且避險績效在外匯期貨中有不錯的表現。另外,本篇研究採用資料滾動(rolling)的方式進行迴歸分析,探討市場波動度與模型績效表現之間的關係,實證結果顯示,當市場波動度越高時,將伴隨著較差的避險績效。
This dissertation investigates the futures hedging performance via the simplified multivariate generalized autoregressive conditional heteroscedasticity (S-GARCH) model proposed by Richard, Evarist and Jon. S-GARCH solves the problems of convergence and computational complexity encountered in multivariate GARCH models by decomposing the multivariate process into their univariate counterparts which greatly simplifies the estimation. This study investigates if S-GARCH has better hedging effectiveness compared to other multivariate GARCH models. The hedging performance of index, foreign exchange and agricultural futures are investigated. Empirical results show that S-GARCH is easy to converge and has good hedging performance for foreign exchange data. In additions, a regression analysis for the unhedged volatility and variance reduction show that when the volatility is high, there is a tendency for a worse hedging performance.