Forecasting Crude Oil prices Volatility and Value at Risk: Single and Switching Regime GARCH Models

Forecasting crude oil price volatility is an important issues in risk management. The historical course of oil price volatility indicates the existence of a cluster pattern. Therefore, GARCH models are used to model and more accurately predict oil price fluctuations. The purpose of this study is to identify the best GARCH model with the best performance in different time horizons. To achieve the target we forecast the daily and weekly price volatility of West Texas Intermediate (WTI) crude oil from January 1990 to October 2020 using GARCH single regime (GARCH (1,1), GJR-GARCH, EGARCH, HYGARCH and FIGARCH), and switching regime (MRS-GARCH and MMGARCH)  models.  We then evaluate the accuracy of forecasts resulting from different models with the help of traditional loss functions and at-risk value.  The in-sample results indicate the high accuracy of the MRS-GARCH model in terms of weekly data, but the out-of-sample results show the superiority of single-mode GARCH models. There is thus no uniformly superior procedure for forecasting oil price volatility in different time horizons.  The  evaluation of the forecasting performance of VaR functions shows that switching regime models do not significantly improve the accuracy of forecasts of crude oil price fluctuations.