Beyond the Power Law: Uncovering Stylized Facts in Interbank Networks

TL;DR

This paper analyzes the distribution of interbank network characteristics in Russia from 1998 to 2004, finding heavy-tailed distributions best described by stretched exponential and log-normal models, with network topology changing during crises.

Contribution

It provides a comprehensive analysis of interbank network distributions beyond power laws, identifying the best-fitting models for various network attributes and their robustness across different periods.

Findings

Heavy-tailed distributions dominate the data.

Stretched exponential and log-normal fit the full data range best.

Network topology significantly changes during crisis periods.

Abstract

We use daily data on bilateral interbank exposures and monthly bank balance sheets to study network characteristics of the Russian interbank market over Aug 1998 - Oct 2004. Specifically, we examine the distributions of (un)directed (un)weighted degree, nodal attributes (bank assets, capital and capital-to-assets ratio) and edge weights (loan size and counterparty exposure). We search for the theoretical distribution that fits the data best and report the "best" fit parameters. We observe that all studied distributions are heavy tailed. The fat tail typically contains 20% of the data and can be mostly described well by a truncated power law. Also the power law, stretched exponential and log-normal provide reasonably good fits to the tails of the data. In most cases, however, separating the bulk and tail parts of the data is hard, so we proceed to study the full range of the events. We find that the stretched exponential and the log-normal distributions fit the full range of the data best. These conclusions are robust to 1) whether we aggregate the data over a week, month, quarter or year; 2) whether we look at the "growth" versus "maturity" phases of interbank market development; and 3) with minor exceptions, whether we look at the "normal" versus "crisis" operation periods. In line with prior research, we find that the network topology changes greatly as the interbank market moves from a "normal" to a "crisis" operation period.