Modelling the volatility of TOCOM energy futures: A regime switching realised volatility approach

Abstract

This paper combines the Heterogeneous Autoregressive Realised Volatility (HAR-RV) model and the Markov Regime Switching (MRS) approach to estimate and forecast volatility of energy futures contracts traded at the Tokyo Commodity Exchange (TOCOM). The proposed MRS-HAR-RV model allows the dynamics of the realised volatility to change as market conditions change. The dataset consists of intraday prices for gasoline, kerosene and crude oil futures. Estimation results suggest that MRS-HAR-RV model can capture dynamics of price volatility of energy futures better than alternative models. However, out-of-sample forecast evaluation results show that MRS-HAR-RV can only produce better forecasts for more liquid contracts. Moreover, MRS-HAR-RV model seems to less over-predict and more under-predict the volatility compared to HAR-RV, HAR-RV-CJ, GARCH, and MRS-GARCH models. • Realised volatility of TOCOM energy futures is regime dependent. • MRS-HAR-RV model can better capture the dynamics of realised volatility. • MRS-HAR-RV is more likely to under-predict realised volatility. • MRS technique helps reduce the over-prediction in realised volatility.