國際原油價格波動幅度之研究與預測

Abstract

本篇研究以三大指標性原油價格為研究對象，分別為西德州原油、北海布蘭特原油及杜拜安曼原油，資料來源為經濟部能源局網站，期間選取自2000/1/3至2007/2/7，研究發現西德州原油價格、北海布蘭特原油價格以及杜拜安曼原油價格，彼此間之變化極具關聯性，且雖然該等原油價格呈現逐年攀升的趨勢，但觀察其波動率時間序列圖形，卻可發現其波動幅度相反的呈現逐年縮小的現象。 而利用自我相關條件異質變異數模型，進行三大指標性原油價格波動性模型的估計中，本研究以標準的GARCH模型和納入風險溢酬考量的GARCH-M模型及考慮槓桿效果的TGARCH模型作為估計波動性工具，進行相互比較，實證結果指出利用樣本內資料所建立的波動性模型中，以TGARCH模型收斂效果最好。而在三大指標性原油價格的波動性模型中風險溢酬效果皆不顯著。在布蘭特原油價格波動性模型及西德州原油價格波動性模型中槓桿效果有顯著影響。 再則針對樣本外資料進行波動性的預測能力評估，由於樣本外期間較短，使其無法產生與樣本內資料相同的一致性結論，例如布蘭特原油價格的波動性預測以GARCH模型的預測能力最佳，但杜拜安曼原油價格的波動性預測，則是以GARCH-M模型的預測能力最佳。

This thesis studies the volatilities of three major crude oil prices: Western Texas, Dubai & Amman, and Brent. The data are daily oil prices from January 3, 2000 to February 7, 2007. Changes of these three oil prices significantly relate to each other, with similar price moving curves. These three major oil prices all rose higher and had been tripled from 2004 to 2006. However, their volatilities became smaller. By using three kinds of autogressive conditional heteroscedasticity models which are standard GARCH, GARCH-M, and TGARCH models to estimate volatilities of in-sample three major oil prices, it is found that TGARCH model including leverage effect fits best the actual volatilities. However, to forecast volatilities of three major oil prices, the TGARCH model is no more the best because of the short out-of-sample period. Besides, the risk premium effects in three major oil prices were all insignificant. The leverage effects were significant in Western Texas and Brent oil prices.