Post-Mortem Examination of the International Financial Network

Abstract

As the recent crisis has forcefully suggested, understanding financial-market interconnectedness is of a paramount importance to explain systemic risk, stability and economic dynamics. In this paper, we address these issues along two related perspectives. First, we explore the statistical properties of the International Financial Network (IFN), defined as the weighted-directed multigraph where nodes are world countries and links represent debtor-creditor relationships in equities and short/long-run debt. We investigate whether the 2008 financial crisis has resulted in a significant change in the topological properties of the IFN. Our findings suggest that the crisis caused not only a reduction in the amount of securities traded, but also induced changes in the topology of the network and in the time evolution of its statistical properties. This has happened, however, without changing the disassortative, core-periphery structure of the IFN architecture. Second, we perform an econometric study to examine the ability of network-based measures to explain cross-country differences in crisis intensity. We investigate whether the conclusion of previous studies showing that international connectedness is not a relevant predictor of crisis intensity may be reversed, once one explicitly accounts for the position of each country within the IFN. We show that higher interconnectedness reduces the severity of the crisis, as it allows adverse shocks to dissipate quicker. However, the systemic risk hypothesis cannot be completely dismissed and being central in the network, if the node is not a member of a rich club, puts the country in an adverse and risky position in times of crises. Finally, we find strong evidence of nonlinear effects, once the high degree of heterogeneity that characterizes the IFN is taken into account.

Keywords: financial networks, crisis, early warning systems

JEL Classification: E65, F30, G01