Into the Trading Book: Estimating Expected Shortfall

Abstract: In light of the revised 2019 proposals constituting the Fundamental Review of the Trading Book, which amend the third Basel Accord, expected shortfall is set to replace value at risk as the risk measure dictating banks' capital reserving requirements for exposure to market risk. This paper examines how best to accurately estimate expected shortfall from a regulatory perspective by carrying out an array of non-parametric as well as parametric methods over the recent years of financial instability. While previous research has predominantly made use of stock indexes to proxy bank’s trading books, we not only employ the S&P500 Index, but also real profit-and-loss data of three large European banks. Through backtesting we identify a GARCH-Generalized Pareto distribution model (rooted in the peaks-over-threshold model) as yielding the most satisfactory ES forecasts for both the index and bank data sets, with the age-weighted historical simulation method also showcasing an all-around strong performance.