- 期刊
- OpenAccess
原油價格預測及甲苯和二甲苯價格均衡關係之實證研究
Crude Oil Price Forecasting and the Equilibrium Between Toluene and Xylene: New Evidence
摘要
預測油價對國家經濟命脈有直接影響，聯產品市場隱含交易訊息，選取西德州中級原油（West Texas Intermediate, WTI）和甲苯（toluene）現貨價，自2007年至2018年4月共2,832筆日資料，以WTI向量誤差修正模型（WTI vector error correction model, WTI VECM），並與類神經網路（artificial neural networks, ANN）比較。研究貢獻為使用聯產品價差與原油價差長期均衡後，再用來預測油價，兩方法樣本外的預測曲線（2018年5月）呈現完美重合，均方根誤差（root mean squared error, RMSE）均為1.11。正確的油價預測可用來套利，亦可降低風險。
並列摘要
Oil price forecasting directly affects the economic artery of the country. Markets of joint products contain trading information. To forecast oil price movements, I build a West Texas Intermediate vector error correction model (WTI VECM) and find a long-term equilibrium between WTI spot price spreads and joint product spot price spreads. Then, this study adopts the rolling window technique and detects the time-variation property. To evaluate forecasting performance, the artificial neural network (ANN) is used as a benchmark. The sample data is selected from WTI and Free On Board (FOB) Korea spot prices daily of toluene and xylene between 2007 and April, 2018. Interestingly, I found two overlapping curves of out-of-sample results (RMSE = 1.11) from May, 2018. An accurate forecasting can bring arbitrage opportunities and reduce transaction risks.