B0894
Title: Estimation of systemic risk in semi-parametric dynamic models based on the empirical distribution of residuals
Authors: Loic Cantin - CREST (France)
Christian Francq - University of Lille and CREST (France)
Jean-Michel Zakoian - CREST (France) [presenting]
Abstract: In semi-parametric conditional location-scale models, commonly used systemic risk measures (such as the CoVaR) involve conditional moments and conditional quantiles of the joint distribution of two innovation processes. To estimate the latter quantiles, we rely on the estimation of the empirical joint distribution of two residuals series. We establish stochastic equicontinuity of the empirical joint cdf, allowing us to prove the consistency and asymptotic normality of estimators of the innovations quantiles. As an application, asymptotic confidence intervals for systemic risk measures are derived. Numerical illustrations based on simulated and real data will also be presented.
