摘要
本文以美國西德州及英國布蘭特原油爲標的，針對1990年波斯灣戰爭使石油市場大幅波動的時期，以相對應的原油期貨進行避險。本文以持有現貨部位探討空頭避險策略作爲研究主題，並假設不考慮交易成本的前提下，修正誤差爲常態的假設，改以Politis (2004)提出之厚尾分配，應用GARCH模型、ARJI模型、GARCH-NoVaS模型與ARJI-HT模型，針對不同避險期間，進行樣本外避險績效的評估。實證結果如下：一、在本研究的研究期間內，假設誤差項爲厚尾分配，其避險績效皆較常態假設優良，表示厚尾分配的設定能有效捕捉到資產報酬率的特性，提高模型的配適能力，提升樣本外的避險績效。二、ARJI模型所計算之避險績效較GARCH模型優良，顯示加入跳躍的因素後，模型更能掌握在短時間內的不確定性，並精確捕捉原油價格波動性，使得避險績效較佳。三、各模型在預測期間之避險績效，大致上均較未避險時之報酬變異降低約70%~80%，因此投資者仍可以規避其價格波動的風險。 實證結果建議投資者在進行操作時，以西德州原油期貨進行避險者，宜採用ARJI模型估計；而以布蘭特原油期貨進行避險者，宜採GARCH-NoVaS模型作爲避險模型，可有效提升避險績效，降低投資風險。
並列摘要
Because of the economic recession always comes with continuous rise in price of oil. Consequently, hedging of oil price has become a crucial issue. Although the GARCH model can capture the time-varying volatility of price, and the ARJI model captures the jump dynamics of price, they are still not good enough to correct fat-tailed property of returns innovation. For this reason, this study employs the GARCH model, ARJI model and GARCH-NoVaS model that accommodate the heavy-tailed returns innovation proposed by Politis (2004) to further examine the hedge performance of WTI crude oil and Brent crude oil, respectively under alternative hedging periods during the Gulf War in 1990. The empirical results show that hedging during high volatility period can reduce variance about 70%~80%. The ARJI model generates superior hedge performance to GARCH model. Moreover, the assumption of GARCH residual in heavy-tail distribution is more appropriate than normal distribution, so that models which accommodate with heavy-tail returns innovation have better hedge performance than traditional return specification. Overall, this paper suggests using the ARJI model to enhance the hedge performance for investors in WTI crude oil markets, while using the GARCH-NoVaS model to abate investment risk for them in Brent crude oil markets.