Quantifying Systemic Risk with Factor Copulas

We propose a tail dependence based network approach to study systemic risk in a network of systemically important financial institutions (SIFIs). We utilize a flexible factor copula-based method which allows us to measure the level of extreme risk in a portfolio when dependence is driven by one or several factors. We identify the most ``connected'' SIFIs based on an eigenvector centrality approach applied to copula-implied dependence structures as ``central'' SIFIs. We then demonstrate that the level of systemic risk implied by such SIFIs chosen as conditioning factors in the factor copula setup exceeds that which is implied by non-central SIFIs in terms of portfolio Value-at-Risk and the portfolio return under stress. This study contributes to the quantification and ranking of the systemic importance of SIFIs which is important for setting adequate capital requirements in particular and stability of financial markets in general