VoxEU Column Financial Markets International Finance

A proposal for countercyclical contingent capital instruments

Contingent capital requirements may reduce the problems of low-quality bank capital and excessively leveraged institutions, but they also risk being too complex. This column aims to strike an appropriate balance by presenting a proposal based on both macroeconomic and bank-level triggers for debt-to-equity conversion. It assesses how such a rule would have performed in identifying stressed banks in recent years.

Financial regulators and industry practitioners have recently considered contingent capital – any debt instrument that converts into equity when a predefined event occurs – as a possible tool for allowing banks to increase capital, even though some are sceptical (Goodhart 2010).

Contingent capital may pursue several goals, including improving the levels and quality of bank capital, providing extra financial resources for systemically important financial institutions, and developing credible countercyclical buffers. While there is relative consensus on the role of contingent capital in a gone-concern (insolvency absent public support) scenario (BCBS 2010b), its role in going-concern scenarios is more controversial. And, needless to say, the views of the industry and the supervisory community are not the same (Figure 1).

Figure 1. Banks vs supervisors

The two perspectives are indeed different, but not necessarily mutually exclusive. Clearly, the design of contingent capital instruments is key.

We propose using “countercyclical contingent capital” as a way to meet the countercyclical buffers introduced by the Basel Committee for Banking Supervision (2010a). In good times, banks would be allowed to issue contingent capital in order to build up the buffer; in bad times, contingent capital would be converted into common equity, thus providing banks with sufficient resources for bearing losses, possibly avoiding restrictions to credit supply (De Martino et al. 2010).

In our setting, while minimum capital requirements and regulatory capital remain the policy instruments for achieving micro-prudential stability, countercyclical buffers and (countercyclical) contingent capital are the instruments for pursuing the new macro-prudential objectives (see Libertucci and Quagliariello 2010).

While in principle our proposal seems to be well suited for countercyclical purposes, there are several practical issues to be addressed in order to balance the need for an effective tool with the willingness to keep capital instruments simple (Goodhart 2010). These are:

  • the choice and calibration of the triggers,
  • the conversion mechanism, and
  • the prudential treatment.

We deal with these in turn.

Choice of the triggers

We propose a double trigger in the conversion process from debt to equity. Conversion would be required when both the following take place:

  • The macro leg: The financial system is facing problems, as predefined by some quantitative rule.
  • The micro leg: The individual bank – while still in a going-concern status – shows weaknesses (for instance, in the form of a capital adequacy ratio below a predefined threshold, set above the minimum).

The interaction of the two triggers would give rise to a “quasi-default” status in which the bank is still able to meet regulatory requirements with its current capital, but additional capital injections are needed as aggregate risk is increasing.

As in the Squam Lake Working Group (2009) proposal, the macro leg identifies “cyclical” problems, whose severity (mild recession vs. crisis) depends on the calibration of the threshold. The micro leg, meanwhile, ensures that only weaker banks convert, providing right incentives to sound management.

For testing how different couples of micro and macro triggers would work, we focus on the top 15 banks in terms of total assets in eight countries (Canada, France, Germany, Italy, Spain, Japan, the UK, and the US) over the period 1994-2009. For the bank-specific trigger, we test accounting/prudential ratio (total capital ratio, Tier 1 ratio, leverage, and return on equity) as well as market-based variables (abnormal returns over different time horizons). For the macro trigger, we start with the returns of the domestic banking indices over different time horizons (3 months, 1 month, 2 weeks). We then move towards more genuinely macroeconomic indicators (e.g., the deviation of GDP from its trend, interest rates, etc.).

Probability of conversion

First, we look at the probability of conversion. We consider the double trigger infringed if both the bank-specific and the macro variables simultaneously drop below the threshold in any month of the year. We interpret the historical frequency of breaches as the ex ante probability of being converted. Our simulations suggest that, while there is obvious sympathy for prudential ratios from the regulatory community, they do not perform very well in all countries, particularly in the US, UK, and Canada where the prudential triggers are never pulled (i.e., the probability of conversion is zero). Common thresholds across jurisdictions with different definitions of supervisory capital do not work.1 We also find that market-based triggers show a better performance on average, but they may be more prone to death spirals and would not fit non-listed intermediaries.

Timing of conversions

We next examine when the triggers would have been hit in the past. Ideally, countercyclical contingent capital should convert when bad times are approaching. To address this point, we analyse when a given combination of triggers would have determined a conversion. Figure 2 shows the results, at country level, based on four different combinations of triggers:

  • A: Tier 1 ratio below 5% and GDP below its trend;
  • B: Tier 1 ratio below 5% and banking index return below -5%;
  • C: Two-week abnormal return below -6% and a 1.5 GDP gap;
  • D: Two-week abnormal and bank index returns below -6% and -5% respectively.

Figure 2. Timing of conversions

At first sight, the combination between a market-based trigger and a macroeconomic trigger (C) seems to work more accurately, determining most of the conversion in severe economic recession episodes. Similarly, market-based triggers – even when roughly calibrated (D) – work fairly well, determining most of the conversions in market crisis times. By contrast, the dynamics of the double trigger based on prudential indicators (A and B) is less clear-cut.

Accuracy of conversions

The ability of different indicators to successfully identify troubled banks is also key. For this reason, we examine how successfully the different double triggers would have identified banks in distress in the period 2007-2009.

We build on Laeven and Valencia (2010) for the definition of troubled banks. In 2007-09, they identify five countries with distressed banking sectors (France, Germany, Spain, the UK, the US); in these countries, they find eight distressed banks. We use this list of banks to analyse the performance of the four double triggers presented above in terms of Type I and Type II errors.

Results show that prudential-based triggers (versions A and B) would have been pulled for four banks only: one of them is a distressed bank according to Laeven and Valencia. Thus a sizeable Type I error. Market-based triggers (C and D) would be triggered more often. Trigger C would have determined 45 conversions in 2008-09: four distressed banks are captured among these observations. Finally, trigger D would have determined a lower number of conversions (18 in total), correctly identifying four distressed banks. Type I error is much lower, but at the cost of very high Type II errors, although their size might be largely overestimated because of the very strict criteria used for selecting troublesome banks.

Case studies

In order to show how countercyclical contingent capital would perform at bank level in real life, we present two case studies. These examples mimic the functioning of our proposed contingent capital for two banks during the crisis: a troubled bank and a sound one.

Figures 3 and 4 describe the dynamic of both micro- and macro-trigger variables for a given bank. Moreover, for the threshold values used before in Figure 2, they highlight periods where the double trigger would have been pulled.

Figure 3.

Figure 4.

The case studies are as we would expect. First, the triggers based on Tier 1 ratio are pulled extremely rarely: in all four examples, they would have never been activated. This result applies both to the sound bank (as expected) and to the poorly performing bank. On the other hand, market-based indicators successfully identify the poorly performing bank’s distress situations. But this accuracy does not come without a cost, represented in this example by the false alarm for the sound UK bank.

The definition of conversion mechanisms

The conversion mechanism of contingent capital is another important determinant of its marketability. In fact, the conversion rate determines the number of common shares that the holders of convertible debt receive when the triggering event occurs. Accordingly, the conversion rate determines the share of losses that subscribers and existing shareholders bear – a choice intimately depending on policy objectives.

We support a mechanism based on a capped variable number of shares, which penalises the shareholdings more but avoids wallpapering. The risk of dilution provides the correct incentives for existing shareholders to monitor managers; at the same time, since the cap means that debt holders may also suffer some losses, market discipline is preserved.

Prudential treatment

We believe that banks should be allowed to use contingent capital for meeting the buffer requirement before conversion; however, contingent capital would not be eligible as regulatory capital and it would not be available for meeting minimum capital requirements. After conversion, the contingent capital would fade out, leaving the bank with an increased quantity of common equity to rely on.

As regards the features required for countercyclical contingent capital instruments, they have to be paid fully at issuance (i.e., no unfunded instruments allowed) to avoid counterparty credit risk and prevent contagion effects when the trigger is pulled. They should be either perpetual or at least long-dated and permanent in order to avoid the risk of rollover. Flexibility of payments does not seem to be relevant to the extent that such instruments are used to meet the buffer requirement.

Conclusions

In taking the decision on the possible use of contingent capital for prudential purposes, regulators will trade-off the potential benefits of this tool and the risk that a complex design leaves room for excessive financial innovation and arbitrage opportunities. To what extent countercyclical contingent capital design is consistent with the willingness of the supervisory community to keep capital instruments as simple as possible is open to debate.

The opinions expressed are those of the authors and do not necessarily reflect those of the Bank of Italy.

References

Basel Committee on Banking Supervision (2010a), Countercyclical capital buffer proposal, Consultative document, July.

Basel Committee on Banking Supervision (2010b), Proposal to ensure the loss absorbency of regulatory capital at the point of non-viability, Consultative document, August.

De Martino, G., Libertucci (2010), M., Marangoni, M. and Quagliariello, M., “Countercyclical Contingent Capital (CCC): Possible Use and Ideal Design”, Banca d'Italia Occasional Paper, n. 71, September.

Goodhart, Charles AE (2010), “Are CoCos from Cloud Cuckoo-Land?”, VoxEU.org, 10 June.

Laeven L. and Valencia F. (2010), “Resolution of Banking Crises: the Good, the Bad, and the Ugly”, IMF Working Paper, no. 146.

Libertucci M. and Quagliariello M. (2010), “Rules vs. discretion in macroprudential policies”, VoxEU.org.

Squam Lake Working Group on Financial Regulation, (2009), An Expedited Resolution Mechanism for Distressed Financial Firms: Regulatory Hybrid Securities.


1 Also significant public injections in some countries may have affected banks’ solvency positions and, thus, our results.

315 Reads