Andrew Gracie: Ending too big to fail - getting the job done

Speech by Mr Andrew Gracie, Executive Director of Resolution of the Bank of England, at Deloitte, London, 26 May 2016.

The views expressed in this speech are those of the speaker and not the view of the BIS.

Central bank speech  | 
09 June 2016
PDF version
 |  7 pages

When G20 Leaders were putting together the post-crisis financial reform agenda in 2009 in London and Pittsburgh, the imperative to find a solution to too big to fail (TBTF) was clear. Feelings were rightly raw given the scale of public support provided to banks and the damage to the global economy. But there was scepticism that resolution - managing the failure of systemically important financial institutions without taxpayer support and without crashing the financial system - would ever work. Naysayers thought that bail-outs would always be necessary for larger banks, especially those operating cross-border. The work that has been done since has met this challenge. With the publication of the final Total Loss-absorbing Capacity (TLAC) Standard1 in November last year a major milestone was passed. We have come a long way to making G-SIB resolution feasible and credible, and with implementation of the measures already agreed, and the tying of loose ends, the task will be complete. Today I want to review what we have done and what remains to be done to make G-SIBs resolvable.

But before turning to resolvability of firms let me say something first about the FSB Key Attributes (KAs) of Effective Resolution Regimes for Financial Institutions2 and the process of resolution planning, both of which are driving the building out of the new resolution paradigm.

The KAs, which were agreed by the FSB and endorsed by G20 Leaders at their summit in Cannes 2011, are the foundation on which all else has been built:

a. The KAs set out our shared objective as authorities - to be able to manage the failure of a G-SIB without adverse impact on system stability, and without recourse to taxpayer funds.

b. They list the minimum elements needed for a statutory resolution regime to be effective. As countries have implemented the KAs at national level, this has provided a convergent framework of resolution regimes internationally, smoothing the way for resolution to work consistently cross-border.

c. Finally, the KAs instituted a process of resolution planning for global systemically important banks (G-SIBs) in crisis management groups (CMGs).

The power of this should not be underestimated. The status quo pre-crisis was built on the belief moral hazard was better managed through constructive ambiguity. It is fair to say that this did not work out so well. The new paradigm is for authorities to be clear ex ante as to how resolution will work and to ensure it can. Clear with the market, so shareholders and creditors are under no illusion as to the risks they are taking. And clear as authorities with each other to confront the obstacles and difficult questions around firm failure upfront and find answers to them.

Resolution planning will be a constant from here for firms and authorities. CMGs held annually or semi-annually have been central to this effort. Incrementally, resolution plans for G-SIBs have come into focus in CMGs. The resulting plans have then been endorsed by senior policymakers from the authorities around the table in CMGs in the Resolvability Assessment Process (RAP). In RAPs for each firm, the plans have been reviewed, barriers to resolvability identified and agreement reached on how, and on what timeframe, they will be removed. This way commitment has been secured from all the authorities in CMGs to the plan ex ante. But perhaps more importantly, there is a network of people who have worked through these questions together and have a degree of mutual trust and reliance that will enable us to get to a co-operative outcome cross-border in a way that has often failed us as authorities in the past. We will need to continue to ensure that these contacts are maintained and deepened. Ultimately they are what will make resolution work for international banks.

This answers to the call that often comes from industry that we should nail down everything in Cooperation Agreements (CoAgs) or work towards some kind of global treaty. I do not think this is necessary to ensure resolution will work effectively. There is a level of detail and complexity in the resolution plans that does not lend itself to a quasi-legal document like a CoAG.

But more importantly emphasis on CoAgs misses how the process has evolved. The metaphor originally conceived for resolution plans was living wills. This was powerful, comprehensible by all. It captured the notion that the failure of large banks would not be left to chance in future but instead would be worked through in advance and part of what it would be to be a responsible, regulated entity. But it also gave the impression that each plan would be idiosyncratic, unique to the firm.

In fact, as time as gone on it has become clear that for G-SIBs many of the elements of the plans are similar and the identified barriers generic. Indeed discussions in CMGs for individual G-SIBs has translated into minimum requirements for resolvability at the level of the FSB. In the way that we are removing barriers, we are expressly setting out to provide a platform for cooperation to work, ensuring that home and host interests in this process are aligned.

The FSB's TLAC standard is a good example of how the dynamic of resolution planning in CMGs has provided the impetus for the policy changes. Put crudely, if the KAs set out what legal powers resolution authorities needed, they did not describe the resolution transaction - how the resolution powers should be applied to G-SIBs.

Work since in CMGs has confirmed the transaction is a bail-in. And, in a way, this is what the TLAC standard really represents - agreement among authorities that it is not feasible or credible to take a G-SIB apart at the operating company level during a resolution weekend. Rather the key operating companies need to be recapitalised by bail-in to support continuity in critical economic functions (CEFs), and to buy time for the subsequent reorganisation of the firm (to the degree that is required). The precise tool used, and the point in the group at which resolution powers are applied, may vary. But at the heart of all the resolution plans drawn up in the CMGs is a bail-in transaction.

If TLAC implies a bail-in transaction, what is required to make this transaction work? I want to cover three dimensions: the quantity, quality and distribution of instruments to be bailed in.

The principles that cover the Term Sheet provide the best guide to the policy intent. As regards quantity, the principles provide a robust minimum global standard so that G-SIBs can fail without placing the rest of the financial system or public funds at risk of loss. The principles require resolution authorities to make a prudent assumption about losses incurred on the way into resolution and the need for prudent valuation at that point. Resolution authorities must also ensure that a firm can continue to meet conditions for authorisation and to be sufficiently well capitalised to command market confidence post-resolution.

These principles are reflected in the final TLAC calibration: we worked on the basis that minimum capital requirements would be eroded on the way into resolution, and that G-SIBs would need to be recapitalised to at least a level to continue to meet minimum conditions for authorisation and maintain market confidence on the way out. The FSB agreed that TLAC corresponding to 18% of risk weighted assets (RWAs), or 6.75% of leverage, would provide the minimum necessary to meet these goals, but that more may be necessary depending on firm-specific considerations. Indeed, in transposing TLAC in domestic regulation, several jurisdictions have already indicated that they consider higher requirements to be appropriate for G-SIBs based in their jurisdictions. But the common floor provides a base level that should allow G-SIBs to operate cross-border with something of a level playing field.

If this is the agreed minimum end-point for TLAC, in the course of the last year, on the basis of a quantitative impact study (QIS), we also agreed a viable transition path for G-SIBs to implement TLAC. This is reflected in the two-stage conformance period under which G-SIBs need to comply with TLAC of 16% RWAs and 6% leverage by 2019 and at least 18% and 6.75% by 2022. It was felt that extending implementation to 2022 provided sufficient time for firms to reissue maturing debt as TLAC or complete any additional issuance in a way that minimises any impact on the cost of funding or the supply of credit to the real economy.

Turning to the quality of TLAC, the principles covering the Term Sheet set out: the need for bail-in to be legally enforceable on TLAC; that bail-in should not give rise to systemic risk; that TLAC instruments should be stable, long-term claims; and that holders should have clarity on the order in which they would incur losses.

Taken together, these principles underscore the importance of subordination of TLAC to operating liabilities. Subordination provides clarity to investors and avoids bail-in being complicated or inhibited by compensation claims that arise if creditors are treated less favourably than in insolvency. This need to ensure that the bail-in of TLAC instruments is credible as well as feasible was reflected too in the provisions in the final Term Sheet for a 5% limit on liabilities that could sit pari passu with TLAC and for exclusion of structured notes. Making TLAC consistent with bail-in also led to clarification in the Term Sheet that after 2022 all TLAC should be issued out of resolution entities: debt issued out of SPVs and, in groups subject to SPE strategies, Tier 2 subordinated debt issued to the market out of operating companies will no longer count as TLAC.

Where TLAC is issued in a group brings me to the final key dimension of TLAC - distribution. One of the big innovations of TLAC is to go beyond previous capital accords to address the distribution of loss absorbency within groups. This was always going to be necessary for resolution which operates at legal entity, not group level. And it is essential to engender trust and time consistency between authorities cross-border. As it is set out in the principles covering the Term Sheet, host authorities must have confidence that there is sufficient loss absorbing and recapitalisation capacity available to subsidiaries in their jurisdictions with legal certainty at the point of entry into resolution. Internal TLAC - prepositioned loss absorbency down-streamed from the resolution entity to material subgroups, equivalent to 75-90% of the subgroup's needs - is designed to achieve this, without pushing up overall TLAC requirements for the group as a whole. It also provides a mechanism in an SPE strategy through which by converting the internal TLAC via contractual terms, the key operating subsidiaries can be kept outside resolution and the losses can be transmitted to the top of the group for the resolution to take place in one legal entity, be that a holding company or a parent bank, in one jurisdiction. Again this addresses the concerns that some have expressed about making resolution work cross-border in different legal regimes.

So if those are the trees, what is the wood? Agreement of the TLAC standard signifies that authorities in FSB have put in place the last major piece of the financial reform architecture. They have agreed the amount of gone concern loss-absorbing capacity that is needed to balance the going concern requirements in Basel III. TLAC is also based on a shared approach to resolution through bail-in. In other words, the end goal is in sight.

So what happens next? For TLAC the priority is national implementation. With three years to 2019 and the need for G-SIBs to set out their capital plans for investors, there is no time to waste. US and Swiss authorities started the ball rolling. We in the UK followed at the end of last year with a Consultation Paper on the Bank's approach to setting a minimum requirement for own funds and eligible liabilities (MREL). We are considering the responses received and are aiming to come out with a final policy statement soon.3 I hope that greater certainty about requirements and firms' plans to meet them will reduce the risk premium that attaches to debt financing now and drives a wedge in pricing between instruments that are economically equivalent, particularly between the US and Europe.

Clarification of national approaches to implementation of TLAC will also provide the platform on which to build internal TLAC. There is technical complexity to work through here - how in keeping with the high level principles in the Term Sheet to design the instruments. For example how should internal TLAC be subordinated, should it always be issued directly to the resolution entity, what maturity should instruments have and how should they adjust as the risk profile of the material sub-group changes? And perhaps most importantly how should the joint contractual trigger that allows internal TLAC to be written down outside of resolution work in practice? The FSB is working on these issues and a consultation paper will be issued later in the year.

Implementation of internal TLAC may also prompt the questions as to how internal TLAC relates to existing capital requirements - is there trapped capital, located in a subsidiary or holding company but which cannot be moved to the subsidiary that needs it? If so, in a world of internal TLAC, can it be released? Where should buffers be located in groups? Host regulators will still want subsidiaries to hold adequate capital and potentially additional buffers if these entities are systemically important to the domestic economy. But they may for example become more comfortable about a higher share of capital being held at group level. CMGs and supervisory colleges will provide the setting to think these issues through.

Finally, now we have TLAC, we need to complete the work on how we will use it. That is, how in practice we will execute a bail-in. Part of this is ensuring that the right mechanics are in place. For example, that between the G-SIB, share registrar and the relevant Central Securities Depositories (CSDs) or International CSDs, as resolution authorities we have worked out the plumbing to find a way to distribute the shares already in issue (or new equity) to the bondholders that will take ownership of the firm via the bail-in. Most of this is a matter of building on existing technology and market conventions that are used in liability management exercises and sovereign or other debt restructurings. A difference perhaps is that as regulators we need to ensure shareholder control thresholds are observed and that we are comfortable with any resulting change in control.

Equally important is the question of how and when we will carry out valuations for the purpose of the bail-in and how we will comply with listing and disclosure requirements through the process. In the event of a G-SIB failure we are unlikely to be in a position to fix bail-in amounts at the resolution weekend. Rather, come Monday morning, we will announce, first the universe of debt that has been immobilised (and thus the solvency we can put back into the balance sheet of the bank in resolution); and second the independent valuation process that will ensue. Valuation and setting the terms of the bail-in might take several months. We may have suspended listings on the equity or debt we have immobilised - in some jurisdictions instruments and listings might be cancelled altogether - but it does not mean we can escape making some disclosure on the condition of the firm in the intervening period. We want, at FSB level, to think this through so as to have a path for bail-in execution, that works domestically and cross-border, meets securities law requirements and is transparent to shape and manage market expectations.

I have described how, on the Monday morning of the resolution, by pointing to the debt that we have available to bail in, we can communicate that the G-SIB has sufficient resources to be recapped and carry on. Indeed part of the Monday morning announcement will be confirmation by relevant regulators that, at operating company level at least, all relevant entities continue to meet their capital requirements. But while TLAC is necessary to achieve this, it is not in itself sufficient.

What else, then, is required to stabilise a firm in resolution, to make it resolvable? I want to set out the other work we have done, and are continuing to do, to stabilise a G-SIB going through a resolution process. The first is ensuring continuity in financial contracts in keeping with the KAs. Across FSB members, we have addressed this issue domestically in our respective jurisdictions with statutory powers to stay close-out. But that has left a risk of cross-border close-out under foreign law contracts. That could be solved by statute if hosts had powers recognise or support home country resolution actions. Indeed this is included in the KAs. But in some important cases, including the US, resolution regimes have been implemented without statutory powers of recognition. Absent a reliable statutory approach, changes in all relevant contracts were needed instead.

This underlines the importance of the achievement we have made in getting the ISDA Protocol4, and credit is due to industry and ISDA for making that happen. Most of the G-SIBs have already adhered to the expanded Protocol covering not only OTC derivatives but also securities financing transactions. By the end of this year, the remaining advanced economy G-SIBs will also have adhered. But more importantly, with the jurisdictional modular protocol in place and regulation requiring adherence published for the largest jurisdictions the way is clear for the rest of the market, in particular buyside firms, to adhere. The Protocol effort was built around the principle that entry into resolution in itself should not give counterparties the right to close out, at least not while a bank in resolution continues to perform. This encapsulates the new paradigm of resolution. We need to ensure that other market contracts and conventions are set up in a similar way. I want to mention one where we have just finished work in FSB and one where we are just starting.

The former is operational continuity. Last autumn, FSB published for consultation draft guidance5 for how shared services IT and operational processes should be set up in G-SIBs for them to be resolvable. Again this builds on resolution planning work in CMGs and identification via the RAP process of operational continuity as a generic barrier to resolution. The draft guidance not only addresses the setup of contracts and service level agreements (SLAs) with third parties and internal service companies to provide for continuity at point of entry to resolution but also restructuring or wind-down of services post bail-in.

Relatedly we have started work in FSB on continuity of access to financial market infrastructures (FMIs6). This is not only at the level of the FMI itself but also where G-SIBs connect to FMIs indirectly through agency banks or correspondent banks. The same principle should apply here - if the bank in resolution remains authorised and continues to perform on its obligations then it should be able to secure continuity of access to services provided by FMIs. But unlike the market standard agreements addressed in the Protocol, scheme rules and commercial agreements provided by agency banks vary and preserve a good deal of discretion for the FMI or G-SIB acting as a quasi FMI. We are exploring how far that discretion can, or needs to be, constrained. Ultimately, there is a potential collective action problem we need to address. It should be in the interests of FMIs and agency banks to allow the resolution to succeed. After all, in an SPE strategy the legal entity they are dealing with is likely to stay outside resolution. But the temptation must be there, if you can in resolution, to grab collateral or to increase margin calls on an extraordinary basis. This is the behaviour we saw at various points in the crisis. We need to plan to avoid a repeat of that.

This brings me to the other consultation paper7 issued recently by FSB, related to funding in resolution. If it is clear on Monday morning that a bank in resolution can meet its obligations falling due, then the incentive for counterparties to remove liabilities or seize collateral should be reduced. The funding in resolution principles represent a step toward achieving this. They put the emphasis on self-insurance, or firms having adequate liquid resources and the internal infrastructure to identify needs and mobilise collateral, to be able to see them through the resolution process. But they also describe a potential role for credible public backstops to instil confidence and manage the perception that the firm cannot self-insure. A backstop for funding in resolution is likely to be separate from ordinary central bank facilities. But that should not necessarily mean that the bank in resolution is cut off at the central bank. To some extent the same principle that we want to apply to other market participants in dealing with a bank in resolution should apply also to central banks - that is, if the bank remains authorised and continues to perform, it should have continuity. As long as a bank in resolution continues to meet the eligibility criteria applied to other banks, it should not, as a general rule, be excluded from central bank operations. The adequacy of a G-SIB's liquid resources through resolution is part of the resolution planning discussion in CMGs. In a cross-border setting, coordination around resources and availability of backstops will depend on the resolution strategy - in an MPE strategy responsibilities will sit locally with the authorities of each resolution entity; in an SPE strategy the home authority will need to play a stronger coordinating role. But this discussion will stand on the firm foundation provided by TLAC. Whether for market counterparties, central banks or backstop providers, lending decisions are much easier to contemplate where solvency is assured in the operating entities that need funding.

Work in FSB to date has concentrated on what is necessary to stabilise G-SIBs at the point of entry to resolution, whether changes to firms to make them resolvable - like TLAC and operational continuity requirements - or to market contracts and conventions - like the ISDA Protocol or access to FMIs. But bail-in is not the end of the task. It is a prelude to whatever reorganisation is necessary to restore a firm to viability or to wind it down. Importantly though bail-in buys the time for that process to be orderly.

Post bail-in restructuring runs as a thread through some of the other initiatives I have mentioned. It is part of the operational continuity principles. And, as part of the access to FMI work we are keen to investigate how easy it would be in resolution to dismantle the quasi-FMI functions - payment services, clearing and custody - provided by G-SIBs to clients, financial and non-financial. However easy or difficult it is to transfer or wind-down services provided by a G-SIB, clients may have contingency providers themselves or may want to rely on the portability of their positions to move. If they are to move we need to find a way for that to happen in an orderly way.

Post bail-in restructuring is also where ring-fencing of UK banks fits in. Parliament has enacted legislation to ensure there will be no doubt about the separation of retail activity in the major UK banking groups at the point of failure. The PRA now has the task of implementing that legislation to complete the work of making ring-fenced banks separable ex ante. Similar developments in the restructuring space are underway in other jurisdictions. The splitting off of domestic retail banks in Swiss G-SIBs and the requirement for large foreign banks in the United States to form Intermediary Holding Companies (IHCs) are examples. Another aspect being explored in resolution plans is the scope for orderly, solvent wind-down of trading books in investment banks in recovery or resolution. This goes to the general principle that it is important for us as resolution authorities to understand the options we have to restructure critical functions if we need to in resolution. I can imagine that in time via planning in CMGs we will see more convergence in approaches in this area across jurisdictions.

In conclusion, much has been done to achieve what is necessary to make G-SIB resolution work. TLAC is symptomatic of the distance travelled and represents a true game-changer. Resolution planning will continue in CMGs to bring refinements in some of the other areas I have described - operational continuity, access to FMIs, funding in resolution and post bail-in restructuring. Over the coming months we will continue implementing these to make resolution work in practice for large, cross-border firms.


Bank of England (2015) 'The Bank of England's approach to setting a minimum requirement for own funds and eligible liabilities (MREL)', Consultation on a proposed Statement of Policy, available at

Financial Stability Board (2014) 'Key Attributes of Effective Resolution Regimes for Financial Institutions', available at

Financial Stability Board (2015) 'Guidance on Arrangements to Support Operational Continuity in Resolution', Consultative Document, available at

Financial Stability Board (2015) 'Guiding principles on the temporary funding needed to support the orderly resolution of a global systemically important bank ("G-SIB")', Consultative Document, available at

Financial Stability Board (2015) 'Principles on Loss-absorbing and Recapitalisation Capacity of G-SIBs in Resolution: Total Loss-absorbing Capacity (TLAC) Term Sheet', available at

International Swaps and Derivatives Association (2015) 'ISDA 2015 Universal Resolution Stay Protocol', available at

1  Financial Stability Board (2015) 'Principles on Loss-absorbing and Recapitalisation Capacity of G-SIBs in Resolution: Total Loss-absorbing Capacity (TLAC) Term Sheet'.

2  Financial Stability Board (2014) 'Key Attributes of Effective Resolution Regimes for Financial Institutions'.

3  The UK's consultation on the minimum requirement for eligible funds and liabilities, through which the UK will implement the TLAC standard for relevant firms, was published on 11 December 2015. See Bank of England (2015) 'The Bank of England's approach to setting a minimum requirement for own funds and eligible liabilities (MREL)'.

4  International Swaps and Derivatives Association (2015) 'ISDA 2015 Universal Resolution Stay Protocol'.

5  Financial Stability Board (2015) 'Guidance on Arrangements to Support Operational Continuity in Resolution'.

6  In the US these are known as Financial Market Utilities (FMUs).

7  Financial Stability Board (2015) 'Guiding principles on the temporary funding needed to support the orderly resolution of a global systemically important bank ("G-SIB")'.