Research Insight - Backtesting Expected Shortfall - December 2014

In this white paper, we join the debate over Expected Shortfall versus VaR by introducing three model-independent, nonparametric back-test methodologies for Expected Shortfall, which we find more powerful than today's standard Basel VaR test. Our three Expected Shortfall back-test's generally require the storage of more information, but we find no conceptual limitations nor computational difficulties. In fact, one of the proposed back tests does not require any additional storage of data from a normal VaR back-test. In addition, we affirm that elicitability has to do with model selection and not model testing, making it almost irrelevant for choosing a regulatory risk standard.