Forecast and Backtesting of VAR Models in Crude Oil Market

Research & Reviews: Journal of Statistics and Mathematical Sciences

article

Yue-Xian Li¹  Jin-Guo Lian²  Hong-Kun Zhang² 

[1] Department of Mathematics and Statistics, Inner Mongolia Agricultural University;Department of Mathematics and Statistics, University of Massachusetts Amherst

关键词: Risk Metrics;    Value-at-risk;    GARCH-class models;    Forecasting;    Backtesting;   

来源: Research & Reviews

【 摘 要 】

The oil price has a very important effect on the world economy. In this paper, using data sets of Europe Brent and West Texas Intermediate (WTI) Cushing crude oil daily prices from Jan. 4, 2000 to Jan. 4, 2016, the VaR forecasting performance of GARCH-type models are analyzed and compared in a short horizon. Based on the Kupiecs POF-test and Christo ffersens interval forecast test, as well as a Back testing VaR Loss Function, the empirical results indicate that, for Europe Brent crude oil, EGARCH (1,1) has the best performance; while for WTI, APARCH (1,1) and GJR-GARCH (1,1) outperform other GARCH models. In fact, these results also give significant guidance on how to choose a better risk management model for the certain commodity of different companies even in the same time period.

【 授权许可】

Unknown