The triple (T3) dimension of systemic risk: Identifying systemically important banks

This paper proposes a new systemic risk measure based on a multi-way analysis. The systemic risk is composed of two different components: the time and the cross-dimension. The first refers to the accumulation of banking risk and its interaction with the business cycle, while the second concern, the high-level concentration of the specific risk on relevant financial institutions. Thanks to the three-way analysis, we can, on the one hand, estimate a new measure of systemic risk based on these both components, and, on the other hand, we can transparently identify the SIBs (systemically important banks). To this end, we have empirically evaluated and compared the marginal expected shortfall (MES), the SRISK measure and the conditional value at risk (CoVaR) based on a representative sample of Eurozone institutions listed on the stock exchange for the period from June 2005 to May 2018. Our results show how these systemic risk measures produce different systemic risk classifications for the same bank. The findings, therefore, highlight the fragility and structural dependence of these measures, which may not be used for the estimation of a stable rank. Applying three-way factorial analysis, we show how our measure gives a more stable score. Moreover, our index is the first one to be composed of both the cross-section and the temporal components, essential elements for a proper assessment of systemic risk. © 2020 John Wiley & Sons, Ltd.