Why banks are not too big to fail - evidence from the CDS market

Abstract

This paper argues that bank size is not a satisfactory measure of systemic risk because it neglects aspects such as interconnectedness, correlation, and the economic context. In order to differentiate the effect of bank size from that of systemic importance, we control for systemic risk using the CoVaR measure introduced by Adrian and Brunnermeier (2011). We show that a bank's contribution to systemic risk has a significant negative effect on banks’ credit default swap (CDS) spreads, supporting the too‐systemic‐to‐fail hypothesis. Once we control for systemic risk, bank size (relative to gross domestic product (GDP)) has either no or a positive effect on banks’ CDS spreads. The effect of bank size increases in the home country's debt ratio and turns positive already at moderate debt ratios. This result is consistent with the too‐big‐to‐save hypothesis. We show further that the effect of systemic risk rises sharply at the onset of the financial crisis in August 2007, but weakens after the failure of Lehman Brothers, reflecting changing bailout expectations. Taken together, our results suggest that banks are not too big to fail, but they may be too systemic to fail and too big to save. — Andreas Barth and Isabel Schnabel