Temporal networks in the analysis of financial contagion

This paper studies the dynamics of contagion across the banking, insurance and shadow banking sectors of 16 advanced economies in the period 2006-2018. We construct Granger causality-in-risk networks and introduce higher-order aggregate networks and temporal node centralities in an economic setting to capture non-Markovian network features. Our approach uncovers the dynamics of financial contagion as it is transmitted across segments of the financial system and jurisdictions. Temporal centralities identify countries in distress as the nodes through which contagion propagates. Moreover, the banking system emerge as the primary source and transmitter of stress while banks and shadow banks are highly interconnected. The insurance sector is found to contribute less to stress transmission in all periods, except during the global financial crisis. Our approach, as opposed to one that uses memoryless measures of network centrality, is able to identify more clearly the nodes that are critical for the transmission of financial contagion. JEL Classification: C02, C22, G01, G2