Tail Risks Everywhere and Crude Oil Returns: New Insights From Predictive Quantile Approaches

Abstract

This paper investigates the heterogeneous impact and predictive power of high-dimensional tail risks from global markets on crude oil returns across different market conditions. Quantile approaches are adopted allowing for flexible predictive distributions of oil returns that can depart from normality. The results demonstrate that external market tail risks significantly influence oil returns besides their own tail risks. Notably, an increase in tail risks leads to lower (higher) oil returns in bearish (bullish) markets. Using feature reduction-based quantile approaches, especially the LASSO-based quantile autoregression model, can effectively leverage high-dimensional tail risks for predicting the conditional distribution of oil returns. Furthermore, probability distortion provides a novel perspective to explain the heterogeneous impact and predictive power of tail risks. These findings help investors and regulators assess the potential risks of oil-related assets and formulate corresponding risk management strategies by accurately predicting the probability distribution of oil returns.