Forecasting the volatility of crude oil futures: New evidence from jump-induced volatility

Abstract

This paper proposes an augmented heterogenous autoregressive (HAR) model with time-varying jumps to forecast the realized volatility (RV) of crude oil futures. Jump-induced volatility of crude oil futures is obtained from a GARCH-jump process, then used to augment the HAR model. The results based on both the in-sample and out-of-sample analyses suggest that jumps offer added information for forecasting the RV of crude oil futures, surpassing the incremental information contained in the crude oil implied volatility index (OVX). Various robustness tests confirm these findings. Our findings have key implications for energy market investors, risk managers, and policymakers.