Stability of Core-Periphery Interbank Networks

Abstract

This paper examines the stability of core-periphery interbank networks in a static simulation framework. We also compare the results of the full-scale simulation with a mean-field approach. The main results are as follows: (i) the model exhibits n-dependence: compared to the benchmark network with 25 banks, our system with 200 banks requires much lower capital ratios to induce any defaults at all, since the relative importance of one initially defaulting institution decreases with a growing number of banks; (ii) subtle changes in the parameters governing the network structure induce large swings in the number of defaults not only via the changing network topology, but also indirectly by changing the asset allocation between core and periphery banks; (iii) when the proportion of interbank assets in the system is large, a further increase in that proportion can stabilize the system; (iv) increasing the connectivity among periphery banks greatly increases system stability. (v) While the mean-field approach closely replicates the simulations results for the early rounds of default, it is unable to reproduce the simulation results for later rounds.