Topological structure of the interbank exposure network and systemic risk

Framing systemic risk as a problem of financial contagion, I studied how system robustness is affected by its topological structure. My simulation results show that, firstly, the more heterogeneous networks – such as scale-free and coreperiphery networks – are on average more robust against default contagion than the random network. However, the performances of the former are sensitive to specific configuration of the network and location of the initial shock. In particular, when shocks are targeted at the most connected banks, the scale-free and core-periphery networks become fragile and will collapse once capitalization level falls below certain threshold. Secondly, the extent of contagion surges drastically for random and scale-free networks once there are market-wide problems, such as correlated shocks or asset price dynamics. This suggests that financial contagion, no matter how small it was under normal circumstance, can evolve exponentially once market condition deteriorates. Lastly, while the core-periphery network appears to be superior than the other two networks in most scenarios, an effective regulatory policy requires that the set of core banks should be empirically identified.