Journal of Network Theory in Finance

Risk.net

Harmonic distances, centralities and systemic stability in heterogeneous interbank networks

Gábor Fukker

  • Rigorous approach to network-based centrality measures with new extensions
  • Comparing the relationship between centralities and losses in the event of an institutional failure
  • Finds that the usefulness of centrality measures might be limited but a linear combination of centralities could lead to better results in forecasting losses
     

This paper investigates the effects of contagion in interbank-lending networks, with a special focus on the theoretical grounding of centrality measures. I introduce a new systemic importance measure based on the harmonic distance of Acemoglu, Ozdaglar and Tahbaz-Salehi from 2015. Then, motivated by their theoretical results for concentration centrality in a later work of the same year, I compare this measure with well-known centrality measures that have already been applied in the systemic risk literature, and that do not take into account the structure of a contagion mechanism. I derive an explicit formula of size-adjusted harmonic distances and extend it by using liquid assets for a heterogeneous banking system. The simulation results on extended Barabasi–Albert-type networks and complete networks do not confirm that these new measures would perform better than “off-the-shelf” centralities, but a linear combination of these measures forecasts losses better than any of the individual measures. The time variations of centrality measures are presented in an interbank network. I also test for the scale-free property of the Hungarian interbank-lending network and suggest that network measures be used as systemic risk indicators in systemic stress indexes. Finally, this work points out that, in some cases, the application of centrality measures might be misleading, and specific clearing algorithms provide more reliable results.

To continue reading...

You need to sign in to use this feature. If you don’t have a Risk.net account, please register for a trial.

Sign in
You are currently on corporate access.

To use this feature you will need an individual account. If you have one already please sign in.

Sign in.

Alternatively you can request an indvidual account here: