Leverage the network: a stress test framework based on DebtRank

Stefano Battiston
University of Zurich

We develop a novel framework to monitor systemic risk in financial systems, that computes in a stylized but testable way, the impact of shocks of various nature on the financial system. We disentangle the contributions of first-round effects due to direct exposures to external assets, second-round effects due to interbank exposures, as well as for third round effects due to fire sales of the shocked assets. A contagion module computes the impact of a shock according to various possible mechanisms, including the traditional default cascade and the more recent DebtRank indicator of systemic impact. A second, network generation module, allows to estimate the precision over the computation of such impact and more in general to cope with partial or missing information on banks mutual exposures. It does so by generating Monte-Carlo samples of networks that are topologically plausible and also coherent with the total lending/borrowing available from balance-sheet data. As a demonstration, we carry out a stress--testing exercise on a dataset of listed European banks and we find that second-round and third-round effects dominate first-round effects.

Back to Workshop I: Systemic Risk and Financial Networks