Regime-switching stochastic volatility: Evidence from the crude oil market

This paper incorporates regime-switching into the stochastic volatility (SV) framework in an attempt to explain the behavior of crude oil prices in order to forecast their volatility. More specifically, it models the volatility of oil return as a stochastic volatility process whose mean is subject to shifts in regime. The shift is governed by a two-state first-order Markov process. The Bayesian Markov Chain Monte Carlo method is used to estimate the models. The main findings are: first, there is clear evidence of regime-switching in the oil market. Ignoring it will lead to a false impression that the volatility is highly persistent and therefore highly predictable. Second. incorporating regime-switching into the SV framework significantly enhances the forecasting power of the SV model. Third. the regime-switching stochastic volatility model does a good job in capturing major events affecting the oil market. (C) 2009 Elsevier B.V. All rights reserved.