Applying new volatility estimators to the 2020 oil price crash—Was it predictable?

Given the increasing importance of negative skewness and structural breaks in oil returns, this paper delves into significant daily fluctuations in WTI crude oil prices during the 2020 oil price crash by developing new AR–EGARCH models. Our new investigation of the 2020 oil price crash with highly effective models uncovers several key findings. Firstly, incorporating skewed distribution errors, both skew- t and skew-GED errors, effectively captures crude oil price jumps during the crash. Secondly, integrating structural breaks is also effective in capturing the substantial crude oil price fluctuations during the crash. Thirdly, for WTI crude oil, our new models, the fully integrated AR–EGARCH models, which consider not only skewed distribution errors of skew- t or skew-GED errors but also structural breaks, provide volatility estimates that have predictive power for the volatilities of our not fully integrated models. In contrast, the volatility estimates from our not fully integrated models do not predict the volatilities of the fully integrated models. Additionally, our fully integrated AR–EGARCH models demonstrate explanatory power for crash risks in oil prices, while other comparable existing models do not have such power. This is one of the most novel pieces of evidence from our work. Finally, four, five, and six weeks prior to the 2020 oil price crash, our fully integrated AR–EGARCH models consistently provide stronger predictions for the oil price crash compared to other existing models. This demonstrates how our new models can be applied in real-world practice and also highlights the novelty of our work.