The beauty contest between systemic and systematic risk measures: Assessing the empirical performance

Abstract

Abstract To assess the empirical performance of systemic and systematic risk measures and to face some legitimate concerns in literature regarding the connections between those indicators, we investigate how the state (distressed or not) of a financial company at a given date is related to the corresponding risk indicators. Based on a combination of univariate and multivariate Cox regressions, our approach is applied to 2006–2010 data of 171 listed US financial companies (grouped in two subsamples, S&P and Non-S&P), on which we estimate different versions of nine popular systematic and systemic risk measures, along with leverage as control variable. Results reveal a strong prevalence of systemic measures and leverage over systematic ones, especially when the time distance between the indicator and the event tends to increase. For the S&P companies, a combination of SRisk and leverage provides a considerable improvement over the best stand-alone indicators, while Expected Shortfall emerges as the only systematic measure providing some information in addition to SRisk and leverage for Non-S&P companies.