Network effects and systemic risk in the banking sector

This paper provides a review of recent research on the structure of interbank relations and theoretical models developed to assess the contagious potential of shocks (default of single units) via the interbank network. The empirical literature has established a set of stylized facts that includes a fat-tailed distribution of the number of banks, disassortativity of credit links and a pronounced persistence of existing links over time. These topological features correspond to the existence of money center banks, the importance of relationship banking and the self-organization of the interbank market into a core-periphery structure. Models designed to replicate these topological features exhibit on average more contagious potential than baseline models for the generation of random networks (such as the Erdös-Renyi or preferential attachment mechanisms) that do not share the stylized facts. Combining different channels of contagion such as interbank exposures, portfolio overlaps and common exposure to non-financial borrowers, one typically finds that different contagion channels interact in a distinctly nonlinear way.