Effects of Forecasters Disagreement on the Crude Oil Volatility : A Garch-Midas Approach

In this paper, we investigate whether the forecasted crude oil prices from the Survey of Professional Forecasters conducted by the European Central Bank contain information for the Brent crude oil return volatility predictions. With a variety of GARCH-Mixed Data Sampling, i.e., GARCH-MIDAS specifications, our in-sample estimation results suggest that the oil market is more volatile when disagreement among the forecasters increases, and the variance model including both the realized variance and forecasters disagreement fits the data the best. Our out-of-sample forecasting results indicate that including forecasters disagreement into the GARCH-MIDAS significantly improves oil return volatility prediction. We demonstrate in the real world application that considering crude oil forecasters disagreement when forecasting volatility has important implications for the VaR risk management