Combining Value-at-Risk and Expected Shortfall measures

Tail risk measures such as Value-at-Risk (VaR) and Expected Shortfall (ES) have become essential tools for meeting the stringent global standards of bank management and regulation. In this direction, several approaches for forecasting VaR and ES have been proposed in the literature. Nevertheless, regardless of the approach used, there are many sources of uncertainty that could significantly affect the accuracy of VaR and ES measures. To mitigate the influence of these sources of uncertainty, we propose a novel forecast combination strategy based on the Model Confidence Set. Our results reveal that the proposed combined predictors provide a good alternative for adequately forecasting tail risk measures.