Jon Cunliffe: A 21st century approach to dealing with failed banks
Speech by Sir Jon Cunliffe, Deputy Governor for Financial Stability of the Bank of England, at the1st Single Resolution Board Annual Conference, Brussels, 29 April 2016.
It is hard now to think of a time when there was ever much sympathy for the owners of banks. The years following the financial crisis have been marked by a scarcely diminishing public anger about the bailing out of banks by the taxpayer; the 'heads I win, tails you lose' privatisation of bank profits and socialisation of bank losses.
That has not always been the case. When the City of Glasgow Bank failed in 1878 there was a great outpouring of sympathy for the hardship faced by the bank's shareholders. A relief fund to help them was established, raising nearly £400,000 from members of the public - around £35 million in today's money - including through a public recitation of the works of Shakespeare. The government did not step in.1
The world, of course, was very different then. The shareholders of the City of Glasgow Bank, like the shareholders of many other British banks at the time, stood to lose much more than the value of their shares in the bank. They had unlimited liability for the losses of the bank. And the losses were large - about 50% of the bank's assets. The holder of a single share would have been liable for the equivalent of four times the salary of a teacher of the time.
It is not surprising therefore that within about a decade unlimited liability had all but disappeared from the British banking scene, to be replaced with regimes that limited the liability of shareholders in one way or another.2
I would not propose that we go back to a system of unlimited liability for bank shareholders. But the City of Glasgow Bank episode is interesting because it shows that the public policy challenges of dealing with failing banks are not new.
When a bank fails, the money has gone. Someone has to pay for the losses, however. The question is 'who pays?' This question has probably be-devilled public policy since banks were invented.
'Who pays?' - the past
It goes back much further than late Victorian Britain. The Code of Hammurabi - written by the King of Babylon in around 1,800BC - included a provision that if the harvest failed due to adverse weather then the 'debt-tablet' would be washed in water and interest on the loan would be cancelled for that year.
The public authorities of 14th century Barcelona hit upon a novel policy approach to the liability of bank owners - not only would bankers who failed be publicly vilified by the town crier, they would be forced to live on bread and water until they repaid their creditors.3
How to deal with failed banks is of course only part of a much larger public policy issue - how to deal with the failure of commercial enterprises generally. Arguably, the development of the limited liability corporate body - in which the owners' liability is limited to their investment - and corporate insolvency, is one of the most important foundations of modern capitalism.
But banks are different for two reasons. First, their underlying business of maturity transformation means they depend on confidence. Loss of confidence in one bank can quickly become a run on all banks. And given their leverage and interconnectedness, failure of one bank can lead to failures of others.
And second, as banks have developed they have increasingly become the main providers of money in modern economies - 96% of money in the UK is in the form of claims issued by banks. Failure can mean that large numbers of people lose most of their access to money - if not their money itself - and to the financial system while the bank goes through insolvency.
As banks developed and grew larger in the twentieth century, the question of 'who pays when the bank fails?' continued to challenge policy makers.
Once the liability of a bank's shareholders was increasingly limited the emphasis shifted to the bank's creditors, including its depositors. However the dangers of exposing short-term creditors, like ordinary depositors, to losses were demonstrated in the early twentieth century in the US. Bank failures leading to depositor losses led to numerous bank runs making the whole system unstable and liable to runs at the first sign of bad news. The episode demonstrated the costs of bringing into doubt the 'moneyness' of banks' deposit liabilities.
As a result, policymakers realised that depositors were not best placed to absorb losses and federal deposit insurance was established in the US with the creation of the FDIC in 1933. The answer then to the question of 'who pays?' was first the shareholders, then unprotected depositors and finally the surviving banks that as a whole funded the insurance - with bridging loans from the taxpayer if necessary.
Interestingly, industry-funded deposit insurance was not adopted in the UK until 50 years later; the Bank of England preferred to rely on the 'Governor's eyebrows' for both supervision and for dealing with failing banks - using suasion to get the industry as a whole to provide support, such as the 'Lifeboat' which was created in 1973.4
With the liberalisation and globalisation of banking at the end of the last century, banks became larger, more complex and more interconnected. It became increasingly unclear whether the longer-term creditors could be made to pay and whether large banks could be allowed to fail.
The case of Continental Illinois demonstrated that deposit insurance was not enough. The danger of loss of confidence and contagion made the authorities reluctant to bail in longer-term uninsured creditors. It was in giving evidence on the subsequent bailout - which cost over $1bn - that the then Comptroller of the Currency admitted that the largest eleven banks in the US were 'too big to fail'.
The market got the message. For the largest, most complex banks, the answer to the question 'who pays?' was likely to be the taxpayer. Despite central banks' attempts to engender uncertainty over whether banks would be bailed out, despite the cases of BCCI and Barings in the 1990s in which two medium-sized banks were allowed to fail, expectations that banks would be bailed out became entrenched.
The impact has been identified and quantified; the bailout of Continental Illinois in 1984 artificially reduced the funding costs of all other banks that were considered 'too big to fail' by more than a quarter of a percentage point (32bps).5 And while fluctuating over time, these 'implicit subsidies' remained sizable throughout most of the 1990s and 2000s.
And indeed, expectations of bailouts were generally right. Very few banks were allowed to go into insolvency in the global financial crisis. Lehman was very much the exception that proved the rule. The impact of Lehman's failure revealed that things had not moved on since Continental Illinois; the official machinery in most jurisdictions was completely unable to handle the failure of a large - or in the UK case even a medium-sized - banks in a way that did not pose risks to the financial system.
So in most cases banks were bailed out. The risks of contagion, loss of confidence in the system and disruption to the provision of banking to the economy were simply too great. But the costs had grown since the Continental Illinois episode of the 1980s. In the UK, for example, the banking sector had grown from around 100% of GDP to more than 500% of GDP in 2008. And the UK government had to inject 13 times more money into RBS than had been injected into Continental Illinois.6
Is too big to fail inevitable?
The post-crisis period has seen a vast amount of work to come up with a better way to deal with a failed bank and a better answer than 'the taxpayer' to the question of 'who pays?'
Some have argued that it cannot be done. The crisis, they argue, showed that large, complex banks do indeed have to keep operating. They are too interconnected and provide too large a share of essential services to the economy to be allowed to abruptly stop operating.
This line of argument leads to the conclusion that if we want to end too big to fail, we need to end being big itself. We need to go back to a world of small, less complex banks. With the help of deposit insurance, these could be wound up quickly and simply without great disruption to the economy and society as a whole.
A world of smaller, simple banks may appear tempting. But something important would I think be lost. As the Independent Commission on Banking argued in their Report, splitting up banks would reduce diversification benefits.7 While the cost of failure of small banks may be lower, these banks may hence be more likely to fail. And a large number could fail simultaneously if they take on correlated risks, as was the case in the US Savings and Loans crisis in the 1980s. Something else might be lost: banking relies on information. Large, multi-national banks can use their existing customer knowledge to offer a customer services across the globe and across different business lines. This allows information to flow freely across borders and products and supports international trade and investment.
I think it is possible to have a world in which large banks can fail in an orderly way - without the contagion and damage to the economy that in the past has forced the taxpayer to come in and absorb losses. But there is no single, silver bullet to achieve that; rather it requires the application of a comprehensive set of policy and powers. It requires a major change in the regulatory landscape and the transition will be difficult.
Progress on resolution
The good news is that there has been very substantial progress towards this new world. In the UK certainly we are now in a much stronger position in terms of dealing with bank failures than we were in 2008. Internationally, we have agreement on key minimum standards for dealing with large globally systemic banks.
But the hard work of implementing the new powers, policies and standards is just beginning. Banks have to make the transition to the new world. This will require sustained effort on the part of the regulatory and of the political authorities.
It will not be costless. And as the costs of ensuring orderly failure of large banks in bad times will fall on those banks in normal times, there will be pressure to skimp on what is necessary. I want next to review briefly the progress we have made on the five main planks of this programme of work.
(i) Improved resilience
The first plank is to keep banks out of trouble in the first place. This means ensuring that banks have sufficient resilience as going concerns to absorb losses without getting into trouble or having to cut back lending to the economy in times of stress. For the larger, systemic banks, it means having even higher resilience, not so much because of the extra risks they run but because of the greater damage they can do to the economy if they go wrong.
I do not want to go into great detail on this today, but some headline numbers are worth highlighting. UK banks have roughly doubled the proportion of assets they fund with equity since the end of 2007, and they have cut the share of total funding accounted for by short-term wholesale funding from nearly 30% in 2007 to 10%. They have tripled their liquid asset holdings. Trading assets have been reduced by a third and inter-bank exposures have shrunk by two thirds.
In addition, the major UK banks have demonstrated they are resilient to severe but plausible hypothetical stress tests designed by the Financial Policy Committee and PRA Board, and to recent real stresses in global financial markets - which saw UK bank debt spreads increase by much less than in previous episodes of stress.
(ii) Resolution tools and powers
The second plank is making sure the authorities have the necessary powers and machinery to manage the failure of a bank.
Astounding though it may seem, the UK went into the crisis with no single institution formally responsible for dealing with failing banks. It was assumed the Governor of the Bank of England's eyebrows would deal with such problems as they had done it in the past. But the Bank had always relied on informal methods - it had no statutory powers in this area.
Nor was there any special insolvency regime for banks that gave priority to the needs of depositors and ensured they did not lose access to their accounts for an extended period.
As a result, when Northern Rock got into trouble and a buyer could not be found, the Government had no alternative but to nationalise it - shareholders were wiped out and subsequently the Government was forced to inject £1.4 billion of capital into the bank. A year later, when faced with the much bigger challenges of RBS and HBOS, the Government likewise had to step in.8
The situation was similar in other countries. While the US had a resolution regime, this only covered the insured deposit-taking entities within a banking group and proved unable to deal with the failure of large, interconnected groups. Similarly, SEC-regulated investment banks were outside of the scope of resolution regimes. While some European countries had bank-specific insolvency regimes, they too did not have regimes that satisfy modern standards for bank resolution that would have guaranteed the continuity of critical economic functions.
The shortcomings in terms of public institutions and powers have now been addressed in the key jurisdictions. We have, of course, yet to see how the new resolution powers operate in practice. But we now have a crucial, far reaching capability that we lacked entirely eight years ago.
We need to be clear at this point what we mean by 'resolution' powers and authorities. 'Resolution' does not mean 'resurrection' of a fallen bank. Rather, it means stabilising the bank so that, over time as far as is necessary, it can be either split up, restructured or parts of it put into insolvency in an orderly way.
These new responsibilities and powers, crucially, do not only cover what happens when a bank can no longer carry on operating; they also cover how a bank organises and finances itself when it is operating normally.
A key lesson of the crisis is that the structure and financing of a bank when it is operating normally can make it impossible to deal with its failure in an orderly way. If banks are not structured and financed in a way that supports orderly resolution before they run into trouble, the authorities' options can be very constrained when the bank fails.
(iii) Bank structures
This brings me to the third plank of ending too big to fail. Banks before the crisis were not organised to be resolvable.
The third plank is to ensure that banks are structured in a way that supports resolution and that barriers to resolvability are removed. This can be a daunting challenge - the average number of separate legal entities at G-SIBs is around 1,000 and in some cases is much higher.9
This may seem a rather gruesome business. It is not usual for a firm to spend a huge amount of time planning for its demise. It is certainly an onerous business for all concerned. In the UK, the Bank has to prepare resolution plans for around 200 banks and building societies, based on information provided by the firms. But it is a necessary business. Non-financial firms can fail and go into insolvency without widespread damage to the economy and financial system. Banks, once they get above a certain size, cannot.
(iv) Loss absorbency
The fourth plank is to ensure in advance that banks are financed in a way that supports resolution. This is perhaps the most difficult part of the puzzle as it goes to the heart of that ever difficult question of 'who pays?'
Clearly, the shareholders stand first in the line. And by strengthening bank capital we have ensured that the shareholders can absorb a much greater amount of loss.10 But, unless we go back to the concept of unlimited liability we must assume that if a bank is unable to continue operating, the shareholders' capital has effectively been exhausted. And clearly, the insured depositors - whose losses are covered by the rest of the industry - stand last in line. In between, there is an army of bank creditors, ranging from unprotected depositors, corporates, interbank liabilities, repos, and derivatives, to trade creditors, pension liabilities, contingent liabilities, tax liabilities, and, of course, the unsecured wholesale funders of the bank who have bought longer-term bank debt.
The lesson of the crisis was that it can be very difficult to bail in some of this group for a number of reasons. It is not simply the risk of contagion. Some of the liabilities are linked to the continuing operation of the bank, and bailing them in would mean bringing the bank to an abrupt stop. Some are politically difficult or would cause greater economic damage than others. And this army of creditors tends, in most jurisdictions, to be a single class that has to be treated equally in insolvency and it is impossible to separate them out.
The lesson of the crisis is clear. If the bank that fails on Friday evening has to be stabilised so that it can be back in operation on Monday morning, it must have sufficient creditors whose claims can credibly and quickly be converted to equity over the weekend. If it does not, the authorities will have only the options of insolvency - with all the disruption that entails - or a bailout by the taxpayer. In that case, history will be likely to repeat itself.
A prompt and sufficient recapitalisation that restores solvency should enable a bank to regain authorisation requirements. It should also enable a bank, if needed, to access central bank liquidity on the same terms as other banks. Provision of central bank liquidity to solvent banks is not a bailout. It is part of what central banks do when necessary. In essence what is needed is to settle the question of 'who pays' unambiguously and in advance. We need to pre-position a specific type of creditor who can absorb loss if the bank fails.
Clearly, getting a creditor to accept this in advance has a cost; there will be a premium for carrying such an explicit risk. That cost is, however, simply the counterpart of the hidden subsidy being given to banks by the implicit taxpayer guarantee that applies if they cannot be resolved in an orderly way.
We now have international agreement on the minimum amount and nature of this type of loss absorbing instrument - which goes under the unlovely name of 'TLAC' - that the largest, globally systemic banks should issue.11
And in the EU, under the Banking Recovery and Resolution Directive (BRRD), resolution authorities will need to set in advance, for each firm, the overall amount and nature of the resources that can absorb loss both when the firm is a going concern and when it can no longer operate - this goes under the equally unlovely name of MREL.
(v) International co-operation
The final plank is international cooperation. The hardest challenge in ending too big to fail is dealing with the failure of systemically important banks that operate in a number of jurisdictions. The crisis proved beyond doubt that we did not have the international machinery to handle this.
We have invested considerable effort into addressing these shortcomings by increasing coordination between authorities and ensuring that national authorities also have the right incentives to cooperate in a crisis. For example:
● We have developed firm-specific cooperation agreements (CoAgs) to increase cross-border coordination and information-sharing between home and host authorities.
● We have set up Crisis Management Groups (CMGs) to agree resolution plans for globally systemically important banks. The UK is involved in CMGs for four G-SIBs as home authority and 14 as a host authority. Through CMGs we are making concrete changes in the set-up of firms that will help ensure that when faced with a crisis, authorities will stick to their ex-ante agreements. So-called 'internal TLAC' is a notable example - it enables losses to be passed up to the relevant part of the group.
● In the European Union, resolution colleges have been established in accordance with the BRRD for cross-border firms that have operations in more than one Member State, and the UK expects to be involved in around 20 as home or host authority.
Impact of reforms to date
This multi-pronged programme of measures to address too big to fail is already having an effect. Ratings agencies have started to reduce their 'government support' uplifts for big banks. Spreads between senior and structurally subordinated debt of UK G-SIBs also suggests that resolution regimes are gaining credibility.
The realisation is setting in that bank creditors are at risk if the bank gets into trouble. There certainly appears to have been a jump in awareness at the beginning of this year. The average spread for senior debt issued by European G-SIBs increased by over 25% within little more than a week.
In part, this is a result of the entry into force of the BRRD, and in the UK the Bank's consultation paper on the setting of MREL.12 But it has almost certainly been reinforced by other events that have brought the issue into sharper relief.
As a result, perhaps not surprisingly, there have been calls to amend the regime before it has even been implemented, on the grounds that it will raise costs for banks and make them less able to lend to the real economy and that it is politically impossible to bail in bank creditors.
There are clearly some very important and difficult transitional issues here. And no major shift in preventative regulation is ever universally welcome. But, it is, however, possible to view this greater awareness of the new regime in a more positive light. I have mentioned already evidence that expectations of taxpayer bailouts have diminished.
Perhaps, even more important, one of the key benefits of ensuring that certain creditors know they can be bailed in is that those creditors have a much stronger incentive to monitor and, if necessary, constrain the risks banks are taking.13
Transition to new resolution regime
Moving to an effective resolution regime, including for the largest cross-border banks, will require a major transition. Many larger banks will need to make changes to their structure and financing. There will inevitably be higher costs as the implicit public subsidy is removed. There is no free lunch.
This transition needs to be carefully managed. So in conclusion, I want to set out the principles that the Bank of England is following to guide the implementation of the new regime.
First and perhaps most important is proportionality. The resolution requirements for a bank should be what is necessary to deliver the resolution strategy for that bank - no more and no less.
Large banks cannot be allowed abruptly to cease operating. They have to be stabilised and resolved. Resolution is not resurrection. It takes time. So the bank has to have enough debt that can be bailed in to restore its capital so it can continue to operate as an authorised firm while it is being resolved.
Smaller banks can be more easily separated into the parts that provide critical economic services and that can be sold and those that do not and can go into insolvency. Only the critical parts would be recapitalised in resolution. They need less bail-in debt.
The very smallest institutions do not need to be kept in operation. Their abrupt cessation of activity should not generate contagion or damage the economy. Insolvency rather than resolution is the proportionate strategy here. They do not need bail in debt.
The second key principle is gradualism. The Bank is proposing to allow the full four-year transition period for banks to meet their MREL requirements. This will allow the market for MREL-eligible debt to develop and allow banks to smooth out their debt issuance, minimising issuance and interest costs.
The third is clarity. Banks and their creditors need to have clarity - clarity in relation to the requirements of the new regime and clarity as to their liability if things go wrong. So we need to avoid uncertainty and get on and implement the EU requirements that came into force at the beginning of this year.
Bank creditors need to know where they stand if things go wrong. This is not just a case of making sure that those that buy bank debt that can be bailed in realise the risks of the instruments they have purchased. It also means making sure that creditors understand the resolution strategy of the institution to which they have lent money. Unprotected depositors in institutions that have no bail-in debt should be clear that in the event of trouble, they stand next in line after shareholders when it comes to absorbing losses.
As I've tried to show, the question of how to deal with failing banks and who pays is not new. But the answers have changed over time. We no longer think we should rely on a diet of bread and water, unlimited liability, Shakespearian recitals, ordinary depositors or the taxpayer to bear the burden of failed banks.
Our success can never be absolute. We can't expect to insulate fully all institutions from all external shocks, however large.14 We have, however, in my view, put in place the legal, financial and practical arrangements to resolve banks of all sizes.
But it will only be true if we implement the new regime on resolution fully. Both regulators and industry have an incentive to do so. If, next time there is a crisis, authorities face the same unenviable choice as in the crisis, if taxpayer bailouts cannot be avoided, then the case for breaking up banks and making them much, much simpler will be very hard to resist.
We have made very significant progress on a resolution regime for the 21st century. The powers, international standards and institutions are all in place. The authorities and the banking system have strong and common incentives to complete the implementation. It is in all our interests that we do so.
Afonso, G, Santos, J A, and Traina, J (2014) 'Do 'Too-Big-To-Fail' Banks Take on More Risk?', Federal Reserve Bank of New York Economic Policy Review 20(2).
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
Brooke M, Bush O, Edwards R, Ellis J, Francis B, Harimohan, R, Neiss, K and Siegert, C (2015) 'Measuring the Macroeconomic Costs and Benefits of Higher UK Bank Capital Requirements', Financial Stability Paper 35.
Button, R, Knott, S, Macmanus, C and Willison, M (2015) 'Desperate adventurers and men of straw: the failure of City of Glasgow Bank and its enduring impact on the UK banking system', Bank of England Quarterly Bulletin Q1.
Carney, M (2014) 'The future of financial reform', Monetary Authority of Singapore Lecture.
Financial Stability Board (2015) 'FSB issues final Total Loss-Absorbing Capacity standard for global systemically important banks', Press release, available at http://www.fsb.org/2015/11/tlac-press-release/.
Financial Stability Board (2015) 'Assessing the economic costs and benefits of TLAC implementation'.
Herring, R J and Carmassi, J (2014) 'Complexity and systemic risk', The Oxford Handbook of Banking, 77.
Independent Commission on Banking (2011) 'Final Report Recommendations'.
Marques, L B, Correa R and Sapriza H (2013) 'International evidence on government support and risk taking in the banking sector', International Monetary Fund Working Paper no 13/94.
Morgan, D P and Stiroh, K J (2005) 'Too big to fail after all these years', FRB NY Staff Report no 220.
Usher, A P (1943) The Early History of Deposit Banking in Mediterranean Europe, Harvard University Press.
1 See Button et al (2015).
2 There was a reluctance from banks and from depositors to limit shareholder liability. Unlimited liability, it was thought, reassured depositors and noteholders that their claims were safe. So hybrid solutions were established in which shareholders' liability was capped at some multiple of the capital they had invested, or in which they were subject to a call for extra capital if the bank got into trouble.
3 See Usher (1943).
4 The 'Lifeboat' was a Control Committee of the Bank of England and the English and Scottish clearing banks, consisting of senior representatives of each under the chairmanship of the Deputy Governor of the Bank, which first met on 28th December 1973.
5 See Morgan and Stiroh (2005).
6 The FDIC had injected a total of $2.5bn as subordinated debt and equity into Continental Illinois (around $5bn in 2009 money). This compares to at least $66.7bn (contingent) capital injections into RBS (using 2009 exchange rates).
7 See Independent Commission on Banking (2011).
8 Existing shareholders were not fully diluted in either RBS or HBOS. In the case of RBS, for example, they retained 16% of RBS's total share capital (30% of voting shares).
9 See Herring and Carmassi (2014).
10 See Brooke at el (2015).
11 See Financial Stability Board (2015).
12 $See Bank of England (2015).
13 The Financial Stability Board has calculated that a withdrawal of government guarantees could reduce banks' probability of failure by a third. This draws on evidence in IMF and Fed NY working papers that evaluate the relationship between the 'government support assumptions' that rating agencies express and measures of bank risk-taking.
14 See Carney (2014).