CoCo bonds and the risk appetite of banks: sweet or sour relationship

Mahmoud Fatouh and Ioana Neamțu

Since 2009, contingent convertible (CoCo) bonds have become a popular instrument European banks use to partially meet their capital requirements. CoCo bonds have a loss-absorption mechanism (LAM). When LAM is triggered, the bonds convert to equity capital or have their principal written down, providing more loss-absorbing capacity while a bank is still a going concern. The existing literature argues these bonds could increase risk-taking if shareholders gain at the expense of CoCo holders when the trigger is hit. In our two papers, we assess this argument theoretically and empirically. We show that the risk-taking implications of CoCo bonds rely on the direction and the size of the wealth transfer between shareholders and CoCo holders when LAM is triggered.


The failure of subordinated debt instruments in recapitalising failing, but solvent financial institutions in the 2007–08 global financial crisis prompted interest in CoCo bonds (first suggested by Flannery (2005)).

As the figure below shows, in addition to basic properties of any bond (rate, maturity etc), CoCo bonds have two design features:

  1. The loss-absorption mechanism, under which the bonds would convert to equity or have their principal written-down.
  2. The triggering event (trigger), which can be discretionary or mechanical (book-based or market-based).

Banks subject to Capital Requirements Regulation (CRR) or legislation based on the CRR may use CoCo bonds to meet up to 25% of minimum Tier 1 capital requirements. Jurisdictions subject to CRR include the EU and Norway and the UK has its own version of the CRR. CoCo bonds are referred to in prudential regulation as additional Tier 1 (AT1) capital. To count as bank regulatory capital, CoCo bonds need to have book-based triggers (based on a Common Equity Tier 1 (CET1) capital ratio). The use of book-based triggers has sometimes been criticised by academics. This is because accounting variables may not always reflect economic changes and capture risks in a timely manner. Additionally, CoCo bond issuers may have the ability to window-dress accounting ratios to avoid conversion. Most of the literature advocates for market-based triggers (eg Bolton and Samama (2014), Berg and Kaserer (2015)). Authors argue that such triggers can help avoid problems associated with the use of accounting values (slowness in reflecting economic developments, and issuers’ incentives to window-dress to avoid conversion). However, market-based triggers also face criticisms, including: i) potential manipulation by speculators to force conversion, in what is known as death spirals (Pennacchi et al (2014)); and ii) possible multiplicity or absence of equilibrium prices (Sundaresan and Wang (2015)), which could make the share price an unreliable conversion trigger.

The risk-taking incentives and CoCo bonds

The issuance of debt instruments may incentivise the issuer to take higher levels of risk. This is because when risk increases, a wealth transfer from debt holders to existing shareholders occurs; existing shareholders enjoy the gain if the risk pays-off, whereas everyone, including debt holders, lose if things go wrong (due to limited liability). In addition, CoCo holders face another possible wealth transfer. When the trigger is hit, CoCo holders have their bonds exchanged for equity shares or face a (temporary or permanent) write-down in the principal of their bonds. Depending on the design and covenants of the CoCo contract, this could cause wealth to transfer from CoCo holders to the current shareholders or vice versa.

The figure below illustrates how the triggering mechanism of a CoCo bond works. A stress which prompts CoCo conversion can have different implications depending on whether the CoCo bonds are principal write-down or conversion to equity. In two out of three cases (highlighted in green) the equity holders gain, and in one they lose (highlighted in red).

In our theoretical paper, we use a simple set up similar to that of Holmström and Tirole (1998) to examine the impact of CoCo bond issuance on the risk-taking incentives of the issuer, and investigate the key factors influencing this impact. Our analysis shows that whether CoCo bond issuance leads to higher risk-taking depends on the direction of the wealth transfer when the trigger is hit. That is, risk-taking increases if triggering the loss-absorption mechanism of a CoCo bond is associated with a wealth transfer from CoCo holders to shareholders. This is because existing shareholders receive the gains if the risk pays-off, but share the losses with CoCo holders, even before all equity capital is wiped-out. Conversely, a wealth transfer in the opposite direction (from shareholders to CoCo holders) reduces risk-taking incentives. Based on that, CoCo bonds could be a useful tool to reduce the riskiness of the bank activities. External investors can use them to control risk-taking incentives of the issuing bank. This, however, requires careful design of the CoCo contracts, especially in terms of conversion rate and the implied reputational costs.

Do CoCo bonds increase risk-taking empirically?

In our empirical paper, we look at the implications of the implied wealth transfer from CoCo holders to existing shareholders for risk-taking incentives, and the potential interactions with macroeconomic uncertainty and competition in the banking system. There is a potential sample selection bias, which may deter the validity of our conclusions. That is, if banks elect to issue CoCo bonds intending to take more risk, the increase in risk is due to ex-post riskier behaviour, rather than CoCo bond issuance. We refer to this as regulatory arbitrage, and use several econometric techniques to test for it.

We then examine whether CoCo bonds have increased or decreased risk-taking in practice by analysing the impact of CoCo bonds issuance in the United Kingdom between 2013 and 2018. Specifically, we estimate whether CoCo bond issuance is associated with higher levels of risk and whether this association is stronger for CoCo bonds that are more non-dilutive. In our setup, non-dilutive CoCo bonds are those with LAMs that imply wealth transfers from CoCo holders to shareholders. Meanwhile, LAMs of dilutive CoCo bonds imply wealth transfers in the opposite direction, as the figure below demonstrates.

We calculate the expected wealth transfer from CoCo holders to shareholder per equity share in the following way:

Expected wealth transfer per equity share

= Prob(trigger event occuring) x (value equity sharebefore_trigger
– value equity shareafter trigger)

The probability of the trigger event occurring is the probability that bank’s CET1 to risk-weighted asset ratio falls to 7%. Based on this method, we find that all 46 CoCo bonds issued in the UK are non-dilutive, implying a potential wealth transfer from CoCo bond holders to shareholders when the trigger is hit.

Empirical analysis and results

We find no clear evidence of regulatory arbitrage by CoCo bond issuers. All but one test for selection bias reach the same conclusion: despite potential ex-post incentives, banks with a higher risk appetite are not more likely to issue CoCo bonds.

Our analysis shows that CoCo bond issuance has a statistically significant positive correlation with asset risk (asset beta) of the issuing banks. Not surprisingly, the strength of this correlation increases with the size of the expected wealth transfer from CoCo holders to the current shareholders. So the more shareholders gain from conversion or write-down of CoCo bonds, the higher the risk the bank takes.

We perform a similar analysis on market risk (equity beta), default risk (credit default swaps spreads), and accounting-based risk (z-score). CoCo bonds are associated with higher levels of market risk. Nonetheless, the estimated size of the wealth transfer from CoCo bond holders to equity holders does not increase risk based on this measure. Default risk is expected to have an inverse relationship with CoCo bonds: higher capital means a lower probability of default. Our results confirm this intuition, and yet show that if gains from CoCo conversion are expected, then the bank is perceived to be riskier (higher CDS spreads). Finally, we do not find any correlation between the accounting based risk measure and CoCo bond issuance. This may be due to a slower adjustment of accounting measures or too few observations.


Our analyses suggest that CoCo bond issuance is associated with higher levels of asset risk, market risk and insolvency risk. The level of these risks appear to be higher when the conversion (write-down) of CoCo bonds is expected to transfer wealth from CoCo holders to existing shareholders. As such, we argue that the proper design of CoCo bond contracts should control for this wealth transfer/dilution. This may help limit the potential risk-taking implications associated with the issuance of these bonds, contributing to the resilience of the banking system. An alternative way to limit these effects could be to compensate managers with manager-specific CoCo bonds. These bonds could transfer wealth from managers to shareholders, better aligning managers’ incentives. For example, they could be write-down CoCo bonds with a relatively high trigger. Our findings suggest that the issuance of such bonds could enhance the resilience of individual banks and the whole banking system.

Mahmoud Fatouh works in the Bank’s Prudential Framework Division and Ioana Neamțu works in the Bank’s Banking Capital Policy Division.

If you want to get in touch, please email us at or leave a comment below.

Comments will only appear once approved by a moderator, and are only published where a full name is supplied. Bank Underground is a blog for Bank of England staff to share views that challenge – or support – prevailing policy orthodoxies. The views expressed here are those of the authors, and are not necessarily those of the Bank of England, or its policy committees.

One thought on “CoCo bonds and the risk appetite of banks: sweet or sour relationship

Comments are closed.