Abstract:
Systemic risk measures have gained popularity in the recent finance literature and they are widely applied for detecting systemic risk contributions of financial institutions. It has also been suggested that a regulator should base capital requirements upon the contribution to systemic risk of each single institution. We analyse market data based systemic risk measures such as Delta CoVaR, MES and SRISK. We find that the uncertainty about their estimates is high. Using bootstrap techniques, we develop a nonparametric hypothesis test to assess if these measures are statistically different between institutions. We find that it is hard to rank institutions using their CoVaR or MES, while SRISK gives good results. We conclude that confidence intervals, incorporating the uncertainty of the measure should be provided and that systemic risk measures should be applied with caution, especially when they are used for important purposes such as the design of a new regulatory framework, because they can lead to expensive bad decisions.
