What a Network Measure Can Tell Us about Financial Interconnectedness and Output Volatility

This paper applies the PageRank algorithm, often used in network analysis, to capture multidimensional and high-degree, cross-border banking relations among countries. It provides a nuanced picture of financial interconnectedness that has not been available in the literature to date.Our measure, FIRank, shows the probability of connection to the network by any country or, equivalently, the share of all connections captured by each country, and provides relative rankings of countries according to their degree of interconnectedness. We show that the United Kingdom and the United States remain the ‘core’ in the global banking network over a thirty-three-year period, with most countries scattered in the ‘periphery’ despite considerable growth and change in the network.This finding contrasts with claims of an increasingly even distribution of connections reported in other studies using more limited network measures. Examining whether financial interconnectedness raises or lowers output volatility, we show that the relationship is nonlinear: initially, higher interconnectedness raises volatility, but beyond a critical level volatility is reduced. This is true in periods of smaller and idiosyncratic shocks but is even more pronounced in the GFC period of large shocks. The novelty of our approach lies in applying well-understood network measures to cross-border banking data to identify where countries rank in international financial interconnectedness with the global bank-lending network.Further, by explicitly analysing how the relative interconnectedness index is related to output volatility we provide new insights into the pros and cons of higher levels of international financial interconnection.(This abstract was borrowed from another version of this item.)