Financial Bridges and Network Communities

Abstract

Network data are widely used to study financial connectedness and monitor financial stability. This naturally calls for flexible models that account for network variability and significant changes in the structure due to major financial shocks. In this paper, we propose a new Weighted Stochastic Block Model with change points that allows for detecting modularity and structural breaks in temporal networks. To make the inference tractable in high dimensions, we propose to combine Monte Carlo Markov Chain and Variational Bayes methods. By examining the European financial system, we uncover a dynamic community structure and highlight the pivotal role of community bridges in shock propagation during the Global and European sovereign financial crises. Keywords: Stochastic Block Models, Financial Networks, Network Communities, Financial Crises, Systemic Risk

Keywords: Systemic Risk, Financial Institutions, Network Communities, Financial Crises

JEL Classification: G12, G29, C51