Assessing the interconnectedness of financial institutions using the extreme value theory

Abstract

We explain a way to determine the systemic importance of financial institutions. We classify ten European banks (2004-2007) based on their interconnections. These are determined by computing co-crash probabilities. We define these CCPs as probabilities that a pair institution undergoes an extreme variation of CDS spreads at the same time. These extreme variations are supposed to follow a Generalized Pareto Distribution. These distributions are time-varying and we determine the parameters of the Generalized Pareto Distribution using the Generalized Pareto Regression. These parameters depend on explanatory variables which are recognized as determinants of CDS spreads of financial institutions. This helps to consider heterogeneity with respect to time.