Andrew Gracie: Making resolution work in Europe and beyond - the case for gone concern loss absorbing capacity

Speech given by Mr Andrew Gracie, Executive Director of Resolution, Bank of England, at the Bruegel breakfast panel event, Brussels, 17 July 2014.
Central bank speech  | 
25 July 2014
PDF version
 |  7 pages


It is nearly three years since the FSB published its Key Attributes of Effective Resolution Regimes (KAs) and, as we look forward to the G20 summit in Brisbane in November, it is worth taking stock of the progress that has been made to end Too Big to Fail.

In Europe we now have the Bank Recovery and Resolution Directive (BRRD)1. Broadly speaking this maps to the FSB Key Attributes: the BRRD establishes a comprehensive set of legal powers and articulates the process through which recovery and resolution planning will take place to ensure that firms develop and maintain effective recovery plans, and that the necessary actions are taken to ensure that they are resolvable. And the objective is in line with the Key Attributes: both view resolution in terms of ensuring the continuity of critical functions without recourse to public funds.

So it will be a real step forward for us here in Europe when the BRRD is implemented. The 1 January 2015 deadline for doing so is fast approaching and all Member States are working hard to revise their national laws and bank resolution frameworks by then.

Given this progress, I propose to talk about what a resolution transaction looks like in practice before turning to some of the preconditions for resolution to be feasible and credible.

I will focus in particular on making liability structures compatible with resolution. The FSB was asked by G20 leaders to prepare, by the time of the Brisbane summit, proposals on the adequacy of global systemically important banks' (G-SIBs) loss absorbing capacity when they fail. Such loss absorbing capacity is sometimes referred to as "gone concern loss absorbing capacity", or GLAC. The equivalent provision in BRRD is the Minimum Requirement for Own Funds and Eligible Liabilities, or MREL. In both contexts the concept is essentially the same: both GLAC and MREL aim to ensure that banks have in place an adequate stock of liabilities that can be used to cover losses and meet recapitalisation needs in a resolution.

I will concentrate my remarks on large international banking groups, or Global Systemically Important Banks (G-SIBs) in the parlance of the FSB.

The BRRD applies to all credit institutions and certain investment firms. Member States will therefore need to think through how MREL should apply to other firms.

But, crucially, EU G-SIBs are often active in third countries and non-EU firms have extensive EU operations. This explains why the FSB's proposals on GLAC are a crucial part of ending the problem of Too Big to Fail and in effect internationalise the concept of MREL beyond the EU.

Resolution transaction

The resolution of a G-SIB is likely to involve bail-in. Why do I say this, when we have other resolution tools available? I say this because of the importance of protecting critical functions. It is simply not credible to think that a G-SIB could be dismantled over a resolution weekend without a destabilising disruption to critical functions. Finding a private sector purchaser capable of taking on the business - in whole or in parts - is likely to be even more difficult.

Bail-in enables us to stabilise a firm. It buys the time that will allow for an orderly reorganisation or wind down of all or part of a failing firm, in order to address the underlying cause of the failure. There is more than one way to effect a bail-in: either through the use of write-down and conversion powers directly, or via a bridge institution. In a bridge, liabilities are exposed to loss by leaving them behind in an insolvency process with the activity that is to continue transferred to a new legal entity. In contrast, where the bail-in tool is applied directly, liabilities are written down or converted into equity within the existing legal entity.

Applying the bail-in tool directly has a number of advantages over the bridge approach, at least in the context of a resolution at the level of an operating company. In particular, because no new entity is required, it will not be necessary over a resolution weekend to authorise a new entity, or to change the contract documentation to which the firm is a party, and memberships of Financial Market Infrastructures can remain intact.

But, whichever approach is pursued, the essential elements of the transaction are the same. These are set out in the KAs and are included in some detail in the BRRD:

  • Entry into resolution: The trigger for entry into resolution is twofold2: first, a firm must be failing, or likely to fail to meet its conditions for authorisation; second, it must be reasonably likely that no action is available to remedy this situation within a reasonable timeframe.

This is intentionally forward looking; it is designed to allow recovery actions to be taken to remedy a breach, or likely breach, of threshold conditions, when the existing owners and management of the firm can take actions that might be successful in rectifying the breach. But, when recovery is not reasonably likely, resolution authorities have the option to put the firm into resolution and deploy their toolkit. And the BRRD goes beyond this - it includes the obligation that, when these two triggers are met, capital instruments are written down or converted to the extent necessary to meet resolution objectives, putting in statute the Basel Committee's requirement that capital must be fully loss-absorbing when firms meet the point of non-viability.

  • Early termination: For resolution to work effectively, resolution authorities need to ensure that counterparties do not use entry into resolution of a firm as grounds for closing out derivatives and other contracts. Disorderly close out could generate exactly the adverse impact of failure on the broader system that resolution is intended to avoid. For that reason, powers to apply a stay on contract terminations and cross defaults are part of the Key Attributes. This power is included in the BRRD, along with to apply a stay power in recognition so that resolution actions taken by non-EU resolution authorities can be enforced within the EU. There are also provisions in the BRRD that will give resolution authorities the power to impose requirements that where EU firms are operating outside the EU, in the absence of a recognition power in the host regime, their financial contracts include terms that ensure the actions of the home resolution authority are enforceable.3

  • Bail-in: So to the bail-in where again the KAs and the BRRD are aligned4. The most important thing to note here is that bail-in must respect the creditor hierarchy. Under BRRD, a small number of liabilities are excluded from bail-in - for example those with an original maturity of less than seven days, and in exceptional circumstances some additional liabilities may be excluded - for example in order to maintain continuity in critical functions. But the principle is clear that liabilities will be written down and converted according to the creditor hierarchy and, within a given class, all affected creditors will be treated equally. Moreover, there is an explicit safeguard that creditors shall be left no worse off following the application of the bail-in tool than they would have been in insolvency. This enables resolution authorities to act with sufficient speed and flexibility, whilst providing shareholders and creditors with an appropriate degree of certainty as to how they will be treated.
  • Valuation: Determining the amount of write-down and conversion of debt required to recapitalise a failed bank and calculating the counterfactual insolvency valuation for the no creditor worse off test are at the heart of the bail-in. A series of valuations are therefore required in the course of a resolution transaction, as the BRRD sets out in some detail5. First, a "going concern" valuation of the firm's assets and liabilities will be used to the gauge the size of any bail-in that is required. A core purpose of this valuation is to quantify expected future losses and so the recapitalisation need at the point of resolution. Second, an equity valuation is needed to inform the terms of the exchange and the allocation of the net asset value generated via the bail-in to the liability holders that have been written down or converted. And, finally, a no creditor worse off valuation measures the counterfactual valuation of the firm in an insolvency to ensure that shareholders or creditors are left no worse off than they would have been in an insolvency. The EBA's technical standards in this area will be a key determinant of our approach to valuation.
  • Exchange: A desirable feature of resolution by bail-in, is to be able to transfer to those bond holders subject to bail, any compensation they are due as soon as possible. Compensation would take the form of the firm's shares or other securities and the resolution would be structured so as to deliver as quickly as possible to the relevant bond holders an interim instrument which is tradeable but also facilitates the change of ownership that follows a bail-in to meet the relevant regulatory requirements. To this end we have been working on Certificates of Entitlement (CEs) which may provide a possible solution. CEs, issued by the resolved bank, represent the contingent right to receive the firm's shares as compensation. Over the resolution weekend, trading and settlement of the firm's bonds will be blocked and CEs will be credited to the bondholders' accounts in the clearing systems (through their custodians). The bondholders, now CE holders, will have a beneficial interest in the ordinary shares of the firm. Once the terms of the bail-in are finalised, conversion ratios for the CEs into shares will be established and CE holders will be able to exercise their rights under the CEs resulting in legal title and voting rights being transferred to those bailed in bondholders entitled to compensation. In order to receive legal title and voting rights to the shares, the CE holders are required to provide evidence of their underlying beneficial ownership of the bonds, confirm that they are not acting as a controller and provide instructions for settlement. In this way beneficial ownership of the bank in resolution will stay in the private sector throughout the process.
  • Reorganisation: Finally, where a firm has been stabilised through a bail-in, this must be accompanied by a comprehensive restructuring plan. Bail-in is - explicitly - a prelude to reorganisation and the stability that it creates buys the time to achieve this6. How much time is required will depend on the options available and how easily and quickly they may be implemented but it is essential to address directly the causes of failure and to restore market confidence so that the firm can access funding and operate normally. Part of the resolution planning process with firms is discussion of the options available for reorganising critical functions in a resolution, whether they are to continue within the firm, to be transferred elsewhere in the market, or to be wound down. Authorities may require firms to make changes ex ante in the way they are organised or operate in order to be adequately resolvable. Structural reforms have been introduced in some jurisdictions to guarantee that options to separate critical functions in resolution exist. The creation of ring-fenced banks in the UK is one such example.

So what does all this tell us? First and foremost it tells us that failing firms are not being resurrected in a bail-in. The reality of a G-SIB is that its activities cannot be shut down immediately and it will need to be adequately recapitalised to sustain its critical functions and stabilise the situation while a reorganisation or orderly wind-down takes place. In order to achieve this, firms must have a liability structure that allows them to absorb losses and to restore solvency to the point that a G-SIB can be reauthorised, and command market confidence and access to normal central bank facilities following the resolution.

Loss absorbing capacity in resolution

That is what GLAC and MREL are meant to help achieve. If G-SIBs do not have adequate loss-absorbing capacity when they fail, then the resolution authority may not be able to stabilise them without recourse to public funds. This will mean the pernicious problem of "Too Big To Fail" will remain unsolved, and the destructive link between banks and their sovereigns will not be severed.

But what exactly does "adequate" loss absorbency in resolution mean? That will depend on the size and systemic importance of a firm; a very small bank with a few hundred depositors may not need any loss-absorbing capacity as its critical functions may be limited to the insured deposits it provides, which are covered by the deposit guarantee scheme (DGS). Any cessation in service is unlikely to cause financial instability - although, even in the case of the smallest banks, it is important that insured depositors are able quickly to recover their funds via the DGS. But, as I have already noted, given their size and complexity, it is important for G-SIBs that there is adequate GLAC to allow the bank to be stabilised without disrupting its critical functions. This has implications for the size, quality, location and disclosure of G-SIBs' GLAC in G-SIBs.

Taking size first: in order that critical functions are not disrupted, prudential authorities around the world must be able to permit the firm in resolution to continue to provide its critical functions (e.g. deposit taking). This implies that, at a very minimum, there must be sufficient GLAC to restore minimum capital requirements. Whatever backward-looking accounting may have indicated, experience suggests there is usually little or no equity value left in a bank once it has failed. The prudent, fair and realistic valuation that resolution authorities must conduct in resolution (as required in the BRRD) will likely crystallise previously unrealised losses or expected losses on the asset side of the bank's balance sheet. This suggests that GLAC should at least be equivalent to minimum Tier 1 capital requirements. It is likely to need a capital position at least as strong as other banks in the market. And for a bank that has undergone resolution to command market confidence, it is likely to need at least as much capital as other banks in the market. It would be sensible therefore for G-SIBs to maintain sufficient GLAC to be able at least to restore fully regulatory capital buffers above capital minima, in order to maintain market confidence that they would be able to absorb any further losses during the restructuring phase without re-entering resolution. One final point on quantum - the ability of stressed systemically important banks to raise their capital position by shedding assets in short order is not sufficiently reliable to warrant a lower GLAC requirement for G-SIBs.

So, we can see how much loss-absorbency we might need, but how much of what, exactly? In principle, it should be possible to expose every unsecured, uninsured liability of a bank to loss in resolution, just as would happen in insolvency. But, from experience, we know that some uninsured, unsecured liabilities are not as reliably loss absorbing in resolution as others and, in practice, it has been difficult to break into the senior class. In part this is due to an absence of legal powers to facilitate this - and the BRRD addresses that problem. But in part it is because of the potential adverse consequences of taking resolution action within a class which includes liabilities that it is difficult to bail-in and where, although we have the power to exercise discretion in exceptional circumstances, it is undesirable and legally risky to rely on this discretion. Just as we require G-SIBs to maintain high-quality, long-dated capital instruments that can reliably absorb losses in a going-concern in order to reduce the probability that they fail, so too we should require G-SIBs to maintain high-quality, long-dated liabilities that can reliably absorb losses in a gone-concern, in order to reduce the probability that they fail in a disorderly way.

So, which liabilities are sufficiently reliably loss absorbing in resolution to count towards a firm's GLAC? Well, any liability that meets the following conditions:

  • First, it must be sufficiently likely that the liability will actually be there when the firm enters resolution. The BRRD requires that, in order to count toward MREL, liabilities must have a residual maturity of at least 12 months.
  • Second, the liability must be within the scope of the relevant authority's resolution tools. That means that instruments issued under foreign law must contain explicit contractual provisions to bring them within scope of the authority's powers, unless there is explicit and watertight statutory provision for this.
  • Third, the instrument must be easy to value in resolution, and operationally straight-forward to expose to loss (and this is easier for debt than it is for deposits) without disrupting the critical functions of the G-SIB. Otherwise it may not be possible to stabilise the firm quickly whilst achieving the resolution objectives.
  • Fourth, exposing the liability to loss must not be likely to give rise to valid legal challenge or "No Creditor Worse Off" compensation claims (NCWO) for breach of the NCWO creditor safeguard.

There is a broad consensus amongst regulators and resolution authorities that capital instruments and long-term, vanilla unsecured bonded debt meets all of the above conditions, subject to a couple of caveats. One of the main ones being that if senior debt ranks alongside liabilities that are less reliably loss absorbing, such as sight or other short-term deposits and derivatives, then such debt cannot meet condition four above without bailing in instruments that fail condition three. This is because if equally-ranking liabilities are not exposed to loss, and the resolution authority departs from the creditor hierarchy, then the senior bond holders may have valid NCWO compensation claims. This could put public funds at risk. Equally as importantly, the protection afforded to senior bondholders by the NCWO compensation claim could mean that market discipline is diluted. This issue is related to a topic I will return to; adequate disclosure of GLAC, and the ability of creditors to be able to assess accurately their position in the creditor hierarchy and the corresponding risk their exposures have.

But first, I would like to point out that requiring eligible instruments to rank below excluded liabilities (such as derivatives), does not mean that G-SIBs will have to meet GLAC with traditional Tier 2 debt. There are various mechanisms available through which "senior" bonded debt could remain senior to traditional subordinated debt but junior to the general creditor class. For example, this could be achieved via contractual provisions; as resolution authorities are empowered to require for MREL in the BRRD. Another option is for G-SIBs to issue GLAC-eligible instruments from a holding company and downstream the proceeds (often referred to as "structural subordination"). Another option that has been suggested is either to prefer excluded liabilities or subordinate eligible liabilities in statute - although agreeing changes to national insolvency laws could prove challenging. Again, whatever the approach, adequate disclosure at the legal entity level will be crucial to allow creditors to identify easily their position in the creditor hierarchy. With such adequate disclosure, there is no a priori reason why the cost of meeting such a requirement would differ between the approaches.

A final, crucial, aspect in rationalising G-SIBs' liability structures to make them resolvable is to ensure that the loss-absorbing capacity is issued from the right place in the group. G-SIBs have hundreds, sometimes thousands, of legal entities. Applying resolution tools to each one is not possible. Therefore the resolution strategy for each G-SIB will involve resolving either the whole group, or parts of the group, by applying resolution tools to a single parent, or a small number of intermediate parent entities - or, in FSB parlance, resolution entities which serve as the point or points of entry to resolution. There must be mechanisms in place to ensure that key operating subsidiaries of each resolution entity have reliable access to the GLAC issued by the resolution entity. One way of achieving this is for the resolution entity to downstream the resources raised by external issuance of GLAC-eligible instruments to material subsidiaries as long-term debt that ranks junior to the subsidiaries' non-capital instruments. This pre-positioned intragroup GLAC could contain provisions that allow the losses to flow up to the resolution entity by converting the intragroup debt to equity without placing the subsidiary into resolution, if home and host authorities agree this is necessary. Since, in a cross-border resolution of a G-SIB, it is in the interests of both home and host resolution authorities that key subsidiaries that are not themselves resolution entities do not formally enter resolution, and instead resolution tools are applied only to resolution entities, this will incentivise cooperation between home and host and reduce incentives to ringfence assets locally in an uncooperative manner.

So, if there is sufficient GLAC of adequate quality that is distributed appropriately within G-SIB groups, we will have made a further, major step in making G-SIBs resolvable and ending TBTF. Again, let me emphasise the importance of disclosure at the legal entity level. It is important that investors in GLAC, and creditors of G-SIBs generally, can identify their place in the creditor hierarchy and therefore better estimate their expected losses should resolution become necessary. This will reduce moral hazard, improve certainty of outcomes in the event of failure, and so incentivise investors to ensure the firms they are investing in are run in a prudent manner.


I want to emphasise just how much progress has been made in last three years: resolution regimes and authorities are now in place where there were none previously in many countries. Resolution planning for systemic banks is taking place in Crisis Management Groups and will soon be taking place in EU resolution colleges too. We look forward to the guidance and technical standards from the European Banking Authority on how resolutions will be transacted. I think, in time, the EU BRRD will come to be seen as one of the most significant pieces of banking legislation to emerge from the financial crisis.

But we must finish the job. For effective resolution to be feasible (i.e. technically practicable) and credible (i.e. widely understood and believable), firms need to be resolvable. MREL is central to this. And for the world's largest banks, for almost all of whom the EU is either home or host, it is crucial that the FSB develops an internationally-agreed standard that establishes a common framework, like MREL, outside Europe and ensures that G-SIBs have sufficient appropriately loss absorbing capacity should they fail in order to be able to implement an orderly, effective resolution of any firm and therefore end "Too Big to Fail".

2  BRRD article 32.

3  BRRD article 55.

4  e.g. BRRD article 43.

5  BRRD articles 36 and 74 and associated technical standards.

6  BRRD article 43(3).