Following the 2007-08 Global Financial Crisis, regulators around the world sought to design a new framework that would result in “bail-ins” instead of “bailouts” of large financial institutions. One important tool in that framework was a debt-equity hybrid instrument: contingent convertible bonds, or CoCos. Regulators encouraged banks to issue CoCos as liabilities on their balance sheets to increase their regulatory capital buffers. If banks fell into financial trouble, CoCos could be “triggered” and either convert to equity or be written down to zero. The goal was to improve the issuing banks’ stability in times of stress. In response, international banks have issued over $1 trillion dollars’ worth of CoCos over the past 15 years.

This regulatory framework was put to a serious test in March 2023. In the matter of a few days, shockwaves emitting from the failure of Silicon Valley Bank in the United States were felt around the world.[1] The shock spread to Switzerland and, in particular, to Credit Suisse. Even though Credit Suisse held billions of CoCos on its balance sheet prior to the panic, the Swiss government had to engineer an emergency bailout to merge it with UBS. CoCos were unable to save Credit Suisse from failure.

In a recent paper, we attempt to address two important sets of questions, using the Credit Suisse collapse as a quasi-natural experiment. First, were banks with more CoCos perceived by the market as safer, on average, than banks with fewer (or no) CoCos? In other words, we aim to provide a systematic analysis of the market reaction to banks that held CoCos during March 2023. In doing so, we hope to shed light on the efficacy of CoCos during a time of true financial panic. Second, given the extreme stress placed on banks during the spring of 2023, how has the CoCos market evolved afterward? Have banks changed the contractual terms of newly issued CoCos to, for example, make themselves safer or to make CoCos more attractive to investors? Has the composition of issuing banks changed?

We explore the first set of questions by examining abnormal returns of common stocks and changes in credit-default swap spreads. The results suggest that holding additional CoCos was not associated with better outcomes during the March 2023 panic. Higher levels of CoCos were associated with worse equity performance and a greater increase in probability of default. If the market had more confidence in CoCos, one would expect to see more positive reactions. Yet our results are more consistent with the market realizing that CoCos were not providing the stability that was originally anticipated.

We also check for the possibility of selection bias – that is, whether inherently weaker banks issued more CoCos to begin with. We find that issuance prior to 2023 was not significantly associated with weaker bank characteristics. Indeed, earlier research has shown that weaker banks – those with more volatile assets, with greater leverage – were the ones that issued fewer CoCos.[2] This is consistent with earlier research that also found that better-capitalized banks and less-risky banks were more likely to issue CoCos.[3] Consistent with these earlier findings, our results show that the negative market reaction is less likely to be driven by the market realizing that the issuers are weaker.

With respect to our second set of questions, we use data (including hand-coded data) on contractual terms. While we observe some changes to non-price terms, we do not observe any systematic patterns. Moreover, an individual bank’s experience during March 2023 does not seem to predict the bank’s adjustment of subsequent contract terms. Interestingly, however, we find some changes in issuer characteristics. New issuers are less liquid than their predecessors, and they are mostly concentrated in continental Europe. In comparison, banks in the United Kingdom have pulled back from the CoCos market, and Chinese banks have largely stopped issuing CoCos.

Our paper contributes to the broader literature on the CoCos regulatory experiment. Since the experiment’s inception, numerous finance and legal scholars have argued that CoCos could be valuable in recapitalizing banks during times of stress since their conversion or write down would reduce banks’ liabilities.[4] Others described how CoCos could encourage banks to improve risk management and raise capital promptly.[5] CoCos were billed as a mechanism that could impose greater market discipline on banks[6] and possibly avoid future government bailouts.[7] Subsequent empirical analysis found that CoCos might have positive effects for banks and reduce systemic risk.[8] Some evidence suggests that CoCos can generate risk-reduction benefits and lower costs of debt for their issuers, as shown by declines in issuers’ CDS spreads.[9]

In contrast, our study lends support to the view that the benefits from issuing CoCos may be less than expected. This finding is in line with earlier research, such as evidence that suggests that the commonly used market-based triggers can lead to self-fulfilling bank runs.[10] Other studies have found that CoCos might even raise the probability of a bank run, impose a negative externality on other banks, and discourage equity holders from providing additional capital in times of stress.[11] Together, these results support the view that CoCos alone are unlikely to eliminate the need for public bailouts.[12]

Our research is part of a growing literature investigating the Credit Suisse collapse in March 2023. To our knowledge, our paper is the first to (1) analyze how all CoCo issuers reacted during the events of March 2023 and (2) document how issuers and buyers of CoCos did or did not adapt in the aftermath. We find that banks which had more CoCos on their balance sheets did not have better market outcomes or better stability outcomes in March 2023. In fact, the banks with more CoCos appear to have fared worse in response to the collapse of Credit Suisse. With respect to the change in the CoCos market, newly issued CoCos have generally kept the same contractual terms in the aftermath of the panic. What’s different are issuer characteristics. New issuers do not appear to be as liquid, on average, and they are more geographically concentrated in Europe, which is a cautionary sign for the next time CoCos are triggered to boost an ailing bank.

ENDNOTES

[1] Choi and Zhang (2025).

[2] Goncharenko et al. (2021).

[3] Avdjiev et al. (2020) and Abdallah and Rodrıguez Fernandez (2022).

[4] Flannery (2005, 2015) and Culp (2009).

[5] Bolton and Samama (2012) and Calomiris and Herring (2013).

[6] Flannery (2005) and Crawford (2017).

[7] Coffee (2011).

[8] Kund and Petras (2023) and Gupta et al. (2021).

[9] Avdjiev et al. (2020).

[10] Sundaresan and Wang (2015).

[11] Chan and van Wijnbergen (2014).

[12] Avgouleas and Goodhart (2015).

This post comes to us from Albert H. Choi, Jacob E. Gerszten, and Jeffery Y. Zhang at the University of Michigan Law School. It is based on their recent article, “Limits of Contingent Convertible Bonds: Evidence from the Credit Suisse Collapse,” available here.