Abstract:
This paper provides an empirical network analysis of the Argentine interbank money market, commonly known as call market, based on data from the Central Bank of Argentina (BCRA). Its main topological features are described applying graph theory, focusing on the unsecured overnight loans settled from 2003 to 2017. The network, where banks are the nodes and the operations between them represent the links, exhibits low density, as is usual in financial networks, and a higher reciprocity than comparable random graphs. It displays a short average distance and its clustering coefficient remains above that of a random network of equal size. Both indicators show values in line with those reported for other interbank networks around the world. Furthermore, the network is prominently disassortative. Different node centrality measures are computed. It is found that a higher centrality enables a node to settle more convenient bilateral interest rates compared with the average market rate, identifying a statistical and economically significant effect by means of a regression analysis. The degree distributions fit better to a Lognormal distribution than to a Poisson or a Power Law. These results constitute a relevant input for systemic risk assessment and provide solid empirical foundations for future theoretical modelling and shock simulations.
