Modelling the Realized Volatility of Crude Oil Futures Market : A Regime Switching-Har with Measurement Errors Approach

Measurement errors and structural breaks are important factors affecting the forecasting accurancy of volatility in the oil futures market. To capture these key points in one uniform forecasting model, we propose the MRS-HARQ-type model by simultaneously introducing the modification of measurement errors and Markov regime-switching in HAR models. The results show that using a regime switching-HAR with measurement errors approach can significantly improve predictive accuracy and accurately portray the structural breaks of crude oil market and the MRS-HARQ-RS model with realized semi-variance has the best forecasting performance. The proposed model provides a useful alternative to the usually employed HAR models