A comparison of Value-at-Risk and Expected Shortfall and their impact on regulation
It has been argued that the risk measurement techniques the latest Basel accords suggest lack the mathematical modeling and conservatism needed to address: (a) cases of extreme loss or market stress; and (b) the complexity of new financial instruments, like credit default swaps. This study evaluates this claim by comparing Value-at-Risk, the risk measure suggested in the accords, to Expected Shortfall, a coherent alternative, using Dow Jones CDX North American Investment Grade and American International Group (AIG) CDX data before, during and after the financial crisis. The methods of historical simulation, bootstrapped historical simulation, filtered historical simulation and parametric approaches based on the Normal, Student's T and Gumbel distributions are tested. It is found that the same approaches which are used for non-crisis regimes are not appropriate for crisis regimes. Furthermore, some commonly suggested approaches for calculating VaR are found to be inappropriate for this data and certain time regimes.
Omar, Yumna, "A comparison of Value-at-Risk and Expected Shortfall and their impact on regulation" (2013). ETD Collection for Fordham University. AAI3564941.