A1215
Title: Measuring systemic risk with non-exchangeable dependence
Authors: Andreas Heinen - Universite de Cergy Pontoise (France) [presenting]
Sangwon Lee - CY Cergy Paris Universite (France)
Abstract: Non-exchangeable dependence breaks the symmetry between the response of an individual firm to market distress and the market's reaction to individual firm distress. A non-exchangeable bivariate copula is used to model the joint distribution of the daily returns of a set of major U.S. financial institutions and of the market index for the 2001-2016 period. Based on this model, systemic risk measures such as CoVaR, Exposure-CoVaR, and MES are computed. More specifically, both static and dynamic versions of a non-exchangeable Clayton copula are estimated, combined with parametric and non-parametric marginal GARCH models for the returns. Further, the systemic risk measures, both in- and out-of-sample, are backtested. It is found that measuring systemic risk using a non-exchangeable copula outperforms its exchangeable counterpart.
