How useful is market information for the identification of G-SIBs?

The Basel Committee on Banking Supervision (BCBS) framework used to identify global systemically important banks (G-SIBs) is based on banks' balance sheet information, leaving information derived from market data untapped. Among the most widely used market-based systemic risk measures, Adrian and Brunnermeier's (2016) Delta-Conditional Value at Risk (ΔCoVaR) best captures the system-wide loss-given-default (sLGD) and conditional impact concepts underlying the BCBS GSIB methodology. In this paper we investigate, using a global sample of the largest banks, whether a score based on ΔCoVaR could be useful for ranking G-SIBs or for calibrating an alternative G-SIB indicator weighting scheme. In our first analysis we find that the ΔCoVaR score is positively correlated with all five of the systemic importance categories of the BCBS framework. However, considerable information/noise with regard to the ΔCoVaR score remains unexplained. Before more is known about this residual, a score based on ΔCoVaR is difficult to interpret and is inappropriate for identifying G-SIBs in a policy context. Besides, we find that a ranking based on ΔCoVaR is subject to substantial variability over time and across empirical specifications. In our second analysis we use ΔCoVaR to place the current static weighting scheme for G-SIB indicators on an empirical footing. To do this we regress ΔCoVaR on factors derived from the G-SIB indicators. This approach allows us to focus on the part of ΔCoVaR which can be explained by balance sheet information which alleviates the identified issues of interpretability and variability. The derived weights are highest for the cross-jurisdictional activity (43%) and size (27%) categories. We conclude that ΔCoVaR is not suitable for use as an alternative G-SIB score but could be useful for policymakers to pursue an empirically grounded weighting scheme for the existing G-SIB indicators.