Backtesting Marginal Expected Shortfall and Related Systemic Risk Measures

These Matlab codes compute the unconditional coverage (UC) and independence (IND) backtesting tests for the systemic risk measures of the article “Backtesting Marginal Expected Shortfall and Related Systemic Risk Measures”. We estimate the MES, SRISK, SES and Delta CoVaR for a panel of 95 large US financial institutions over the period from January 3, 2000, to December 30, 2016. A GARCH-DCC representation is assumed for firms’ returns. Then, we apply the UC and IND backtests. The backtests are computed at a daily horizon (short-term forecasts) for MES, SRISK, SES and Delta CoVaR, and also at a monthly horizon (medium-term forecasts) for LRMES and SRISK. The scripts allow computing the test statistics, the p-values of the backtests, the systemic risk measure predictions and their adjusted counterparts that we use as early warning system indicator of financial crisis. These Matlab codes allow reproducing all the figures and tables of the empirical application of this article.

The article is available at SSRN: https://ssrn.com/abstract=3456052 or http://dx.doi.org/10.2139/ssrn.3456052