I quantify high-dimensional conditional value-at-risk (CoVaR) measures, including multi-CoVaR (MCoVaR) and vulnerability-CoVaR (VCoVaR), to assess systemic risk arising from distress across multiple financial institutions. MCoVaR captures system-wide risk under the assumption that all institutions are distressed, whereas VCoVaR reflects scenarios in which at least one institution breaches its value-at-risk threshold. Using a two-step maximum log-likelihood approach based on multivariate dynamic conditional correlation (DCC)-generalized autoregressive conditional heteroskedasticity (GARCH) models, I first estimate univariate GARCH parameters via quasi-likelihood methods and then estimate multivariate DCC parameters. Analyzing 246 U.S. financial institutions (73 banks, 52 brokers, 34 real estate firms, and 87 insurers) at both institutional and sectoral levels, I identify heterogeneous contributions to systemic risk across groups. Statistical tests confirm that each group plays a significant role in systemic vulnerability, although their relative importance depends on the specific systemic risk measure employed. Overall, the results provide new insights into the drivers of systemic risk associated with multivariate institutional distress.
扫码关注我们
求助内容:
应助结果提醒方式:
