Herman van Dijk (Erasmus University Rotterdam)

joint with Knut Are Aastveit and Jamie Cross

A novel and numerically efficient quantification approach is proposed to forecast uncertainty of the real price of oil using a combination of probabilistic individual model forecasts. The combination method extends earlier approaches that have been applied to oil price forecasting, by allowing for sequential updating of time-varying combination weights, estimation of time-varying forecast biases and facets of mis-calibration of individual forecast densities and time-varying inter-dependencies among models. To illustrate the usefulness of the proposed approach, an extensive set of empirical results is shown about time-varying forecast uncertainty and risk for the real price of oil over the period 1974-2018. We show that our combination approach systematically outperforms commonly used benchmark models and combinations approaches, both in terms of point and density forecasts. The dynamic patterns of the estimated individual model weights are highly time-varying, reflecting a large time variation in the relative performance of the various individual models. Our combination approach has built-in diagnostic information measures about forecast inaccuracy and/or model set incompleteness, which provide clear signals of model incompleteness during three crisis periods. Finally, to highlight that our approach can also be useful for policy analysis, we present a basic analysis of profit-loss and hedging against price risk.