Influence Estimation in Interbank Lending Networks: a Machine Learning Approach

Abstract

Motivated by the financial crisis of 2007-2008, the study of systemic risk in the interbank lending market has grown rapidly in recent years. The identification of influential nodes in the interbank network lies at the heart of this field, shaping policy decisions of regulatory entities such as the Federal Reserve and the European Central Bank. This thesis implements two methods to estimate the contagious influence of banks in the Italian interbank lending market. The first method derives a unified measure of centrality by identifying optimal combinations of established centrality measures in network science. The second method is a novel approach derived from the field of viral marketing, learning relationships between banks from the spread of observable behaviors across the interbank market.