Forecasting volatility in the petroleum futures markets: A re-examination and extension

• Published in: Energy economics 2020-02, Vol.86, p.104626-12, Article 104626
• Main Authors: Hasanov, Akram Shavkatovich, Shaiban, Mohammed Sharaf, Al-Freedi, Ajab
• Format: Article
• Language: English
• Subjects: Confidence; Distribution functions; Economic analysis; Economic forecasting; Economic models; Energy economics; Forecasting; Futures; Futures market; Mathematical models; Model testing; MSGARCH; Petroleum; Petroleum futures; Prices; Robustness; Rolling window; Statistical analysis; Statistical models; Structural breaks; Switches; Volatility; Volatility forecasting; Windows
• ISSN: 0140-9883; 1873-6181
• DOI: 10.1016/j.eneco.2019.104626

This paper examines the volatility models and their forecasting abilities for four types of petroleum futures contracts traded on the New York Mercantile Exchange. The aim of this paper is twofold. Firstly, it replicates and carries out the robustness checks using the rigorous model confidence set test on the out-of-sample volatility forecast analysis undertaken by Sadorsky (Energy Economics, 2006; 28, 467–488) through the same statistical models but with the extended data on daily prices of petroleum futures. Our test results largely confirm the findings obtained in the replicated paper. Secondly, our paper also explores the relevance of some statistical complexities (e.g., model optimality, regime switches, and alternative distribution functions) in volatility forecasting through a large number of moving windows. Our results, in general, show that accounting for the model optimality, structural breaks, and using the asymmetric heavy-tailed distribution functions in the estimations lead to significant forecasting accuracy gains. •This paper replicates and carries out the robustness checks on the volatility forecast analysis undertaken by Sadorsky (2006).•Our test results largely confirm the findings obtained in the replicated paper.•This study also explores the relevance of some statistical complexities in volatility forecasting.•The results show that accounting for the model optimality, structural breaks, and asymmetric distributions lead to considerable forecasting accuracy gains.