Identifying complex core–periphery structures in the interbank market

Abstract

This paper proposes a framework to identify the structure of a financial network and its evolution of over time, and presents an application to an interbank market with complete actual data. The framework is based on a methodology popular in the social network literature, namely the Stochastic Blockmodelling (SBM), which, we argue, is more general, transparent and richer in results than other proposed methodologies. In particular, we can identify the presence of multiple cores and peripheries, as well as different ways of interaction between them. We find that such a varied coreperiphery structure exists in almost all periods for different instruments analyzed. Also, in the case of term deposits, which account for two-thirds of interbank exposures, we find that far from being static, the structure underwent a transition in the period 2009-2015, with the core increasing its size. We also show that facts revealed by our approach cannot be observed in metrics commonly used to describe networks. Finally, we describe how the elements identified by our method can be used to single out sources and channels of transmission of systemic risk in a network of banks. Resumen Este trabajo propone un marco para identificar la estructura de una red financiera y su evolución en el tiempo, y presenta una aplicación a un mercado interbancario con datos efectivos y completos. El marco está basado en una metodología popular en la literatura de redes sociales, Stochastic Blockmodelling (SBM), la que, sostenemos, es más general, transparente y rica en resultados que otras metodologías propuestas en la literatura de redes financieras. En particular, nos permite identificar la presencia de múltiples centros y periferias, así como diferentes formas en que estos se relacionan. Encontramos que en casi todos los periodos existe una variada estructura centro-periferia para los distintos instrumentos analizados. Además, para depósitos a plazo, que representan dos tercios de las exposiciones interbancarias, encontramos que, lejos de permanecer estática, la estructura mutó en el período 2009-2015, con un centro que aumentó su tamaño entre otros cambios observados. También mostramos que los hechos derivados de nuestro análisis no pueden ser observados en métricas comúnmente usadas para describir redes financieras. Finalmente, describimos cómo los elementos identificados pueden ser usados para identificar fuentes y canales de transmisión de riesgo sistémico en una red de bancos.

Structural changes in the interbank market across the financial crisis from multiple core–periphery analysis

“…Bipartitivity emerges even at longer aggregation periods when using an extended stochastic block model that takes into account the different numbers of connections that different banks have. Carreño and Cifuentes (2017) also used stochastic block models to analyse the interbank market of term deposits and derivatives in Chile at the daily level. Their approach permits to identify multiple cores and peripheries-though they rarely observed a second core-and allows to characterise different forms of interaction between core and periphery.…”

GTST: A Python Package for Graph Two-Sample Testing

“…Graph two-sample hypothesis testing is a problem that frequently arises in various disciplines, for example in bio informatics [1], community detection [6], and risk management [3]. Graph two-sample hypothesis have mostly been performed by using graph statistics such has the degree centrality and shortest paths.…”