Systemic risk in networks

Abstract

Systemic risk, i.e., the risk that a local shock propagates throughout a given system due to contagion effects, is of great importance in many fields of our lives. In this summary article, we show how asymptotic methods for random graphs can be used to understand and quantify systemic risk in networks. We define a notion of resilient networks and present criteria that allow us to classify networks as resilient or non-resilient. We further examine the question how networks can be strengthened to ensure resilience. In particular, for financial systems we address the question of sufficient capital requirements. We present the results in random graph models of increasing complexity and relate them to classical results about the phase transition in the Erdös-Rényi model. We illustrate the results by a small simulation study.