Chiara Pederzoli (Università Bicocca Milano)
 

The recent financial crisis highlighted the importance of interconnections in the financial system and the need to measure the impact of contagion. Following the crisis a rich literature has been growing on the very same problem of defining systemic risk and the issues connected to its measurement. Despite these efforts, there is still no consensus either on the very same definition of systemic risk, or on a single risk measure. While different definitions can be found in the literature stressing different aspects, generally speaking systemic risk involves the whole financial system instead of the single institution and it spreads over the real economy. However, the existence of multiple definitions of systemic risk has naturally implied the development of a wide range of measures for gauging it. Regulators’ stress tests on banks further stimulated an academic debate over systemic risk measures and their predictive content. Focusing on marked based measures, Acharya et al. (2017) provide a theoretical background to use Marginal Expected Shortfall (MES) for predicting the stress test results, and verify it on the 2009 Supervisory Capital Assessment Program of the US banking system. The aim of this paper is to further test the goodness of MES as a predictive measure, by analysing it in relation to the results of the European stress tests exercise conducted by the European Banking Authority. As for the 2014 stress test exercise, our results underscore the importance of choosing the appropriate index to capture the systemic distress event. In fact MES based on a global market index does not show association with the stress test, in contrast to F-MES, which is based on a financial market index, and has a significant information and predictive power. By moving to analyse the most recent 2016 EBA stress test, we find slightly different results