Backtesting Marginal Expected Shortfall and Related Systemic Risk Measures

This paper proposes an original approach for backtesting systemic risk measures. makes it possible to assess the measure forecasts used identify financial institutions that contribute most overall in system. Our procedure is based on simple tests similar those generally backtest standard market measures such as value-at-risk or expected shortfall. We introduce a concept of violation associated with marginal shortfall (MES), and we define unconditional coverage independence these violations. can generalize any MES-based (SES), (SRISK), delta conditional ([Formula: see text]CoVaR). study their asymptotic properties presence estimation investigate finite sample performance via Monte Carlo simulations. An empirical application panel U.S. conducted validity MES, SRISK, [Formula: text]CoVaR issued from bivariate GARCH model dynamic correlation structure. results show this provides valid MES SRISK when considering medium-term horizon. Finally, propose early warning system indicator future crises deduced backtests. quantifies how much measurement error by forecast at given point time which serve detection global reversals. was accepted Kay Giesecke, finance.