2.01  It is clear that there is no single mechanism for resolving all types of banks—all resolution authorities look at their powers as a ‘toolkit’, with the determination of the appropriate tool for the job being made on a case-by-case basis.

2.02  This raises the question of whether the tools which currently exist cover all potential cases, or whether there remain lacunae in coverage. We know that certain institutions—for example relatively small US community, mid-size, or even regional banks—can be resolved, because the Federal Deposit Insurance Corporation (FDIC) has a long, successful track record of doing so. We know that the new single-point-of-entry (SPE) resolution model works for certain US global systemically important banking groups (US G-SIBs), and so on. However, the question that we have to ask is whether the suite of tools that we have covers the whole universe of banks and banking groups. In order to demonstrate the coverage of these methods it is necessary to consider a hierarchy of approaches to bank failure. This is broadly as follows:

1. 1.  restructuring of the business, involving asset and business sales, and recapitalization through raising new equity;

2. 2.  sale of all or a majority of the business by the purchase of assets and the assumption of liabilities (i.e. a purchase and assumption transaction) or otherwise;

3. 3.  write-down (after equity has been written down) or conversion to equity of subordinated or senior debt;¹

4. 4.  liquidation;

5. 5.  state aid.²

(p. 10) 2.03  The first two stages may involve the use of a number of resolution tools. These tools have the effect in practice of moving assets around the group to be resolved, and are in practice no more than preliminary steps to the process of resolution. Their primary usefulness is the creation of separate pools to which separate tools may be applied—thus, for example, a transfer of certain assets and liabilities to a bridge bank may enable that bridge bank to be sold as a going concern, thereby enabling bail-in of the remaining capital instruments. It is also true that these tools may be applied to have the effect of a resolution tool—thus, a transfer to a bridge bank leaving some creditors behind in the transferor entity may have the effect of bailing in those creditors—but for these purposes we concentrate on the economic effect rather than the legal name of these tools.

2.04  It is also important to be clear-eyed about the state—and option. There may be many reasons—not all of them economic—which might drive a state to rescue an institution. It should be pointed out that this is by no means a specific characteristic of banks—similar government intervention has been practised in many jurisdictions with firms ranging from car manufacturers to steelworks, and from shipbuilders to coal mines. The aim of designers of bank resolution architectures is to ensure that at all times government has an option not to intervene, not to prevent government from ever intervening.

2.05  Finally, in this context, real-world experience in the context of car manufacturers and steelworks has been that formal procedures, once rendered credible, are rarely used. The normal state of resolution for a business in the commercial world is a restructuring in which creditors voluntarily agree to a variation in their rights in order to optimize the prospects of the business continuing in existence and thus being able to repay a larger proportion of the amounts owed to them than would otherwise be the case. This is likely to be the position in the banking industry of the future. In order for a negotiated settlement to be successful, it is essential that the alternative to settlement be both unattractive and plausible. Creditors of banks in 2008 had no incentive to negotiate, and could reasonably await bail-out, because governments had no credible alternative course of action. This explains the apparent paradox that, provided resolution is plausible in theory, it will not be needed in practice, and to the extent that it is not plausible in theory, it will have to be demonstrated in practice.

2.06  In conceptual terms banks can be divided into the following classes.

2.07  In general the primary safeguard for depositors in institutions of this kind is the deposit guarantee scheme (DGS). Banks of this kind can be left to fail and their (p. 11) depositors can be left to their recourse to the DGS without systemic damage. It may frequently be the case that some commitment of DGS resources to recapitalization (sometimes referred to as DGS bail-in) may produce a lower cost to the DGS itself than allowing the institution to fail and paying claims, but either approach is effective for this purpose.

2.08  These institutions are in general too big for the relevant DGS to cover, and do not have long-term wholesale market creditors who can be bailed in without destabilizing the banking system. In such cases, resolution is managed through writing down the uninsured and unpreferred component of the deposit base. It is in respect of these institutions that the gone concern loss absorbing capacity (GLAC) tool is of most relevance, since it forces such institutions to maintain a minimum level of long-term non-deposit funding.

2.09  These institutions are those which it is envisaged will be resolved through what could be described as a ‘classical’ restructuring or bail-in of their wholesale funding. The problem is likely to be a determination of which wholesale obligations can be bailed in without disturbing or destroying the commercial basis of the institution.

2.10  In these cases the primary issue is to ensure that the allocation of losses does not result in runs that could destabilize the financial system. In such cases, it is important to ensure that long-term unsecured senior creditors are structurally, contractually, or legally subordinate to short-term unsecured senior creditors, and that senior creditors otherwise are treated equally, and in particular that long-term senior funding depositors are bailed in along with other long-term wholesale unsecured creditors.

2.11  It is one thing to say that we have a complete set of techniques in isolation, but another to say that they can be implemented in practice, and the argument is sometimes advanced that no matter how plausible bank resolution plans may seem in theory, political realities mean that they can never be implemented in practice. This argument is based on the assumption that internationally active banks can only (p. 12) be resolved with the active cooperation of multiple governments and resolution authorities around the world.

2.12  Although it is true that international resolution is likely to require multiple activities by multiple regulators, and it is certainly true that coordination between public sector entities would enhance the efficiency of resolution, it is not true that the lack of such cooperation would be a complete bar to resolution. At its simplest, where a single point of entry approach to resolution is employed, the only resolution activities which are required are those which take place at the group holding company level. Any subsidiary below that holding company which requires recapitalization will undergo a solvent restructuring involving the creation of new equity by the holding company. The aim of this structure is precisely to ensure that resolution is only required in respect of one entity in one country—all other group members should remain solvent and trading throughout the process. The only request which the resolution authority in the home country needs to make of resolution authorities in other countries should be that they do not take positive steps to interfere with the resolution process. It does not seem unreasonable for this level of cooperation to be expected.

2.13  It is true that a classical single point of entry approach requires a particular corporate structure, and not all systemically important banking groups have that structure. This has led to the further objection that since making all banking groups conform to the SPE corporate structure would be onerous or—in some cases—arguably impossible, the idea that global resolution for groups that do not already reflect this structure is fatally flawed.

2.14  This is incorrect, but in order to understand why it is necessary to begin with the understanding that the SPE approach is simply one way of achieving a policy outcome. The policy outcome which the SPE approach delivers is to pre-designate a minimum value of long-term unsecured creditors of the institution (whether by statute or otherwise³) as being those creditors whose claims will be written down (or off) in order to recapitalize the entity before any losses are imposed on any other creditors. In an SPE structure, this designation is effected by specifying creditors of the holding company as having this status. However, this is clearly not the only way of achieving that aim. Provided that the creditors concerned are easily and unambiguously identifiable, it is immaterial for this purpose whether they are creditors of any specific group company. What matters is that the terms of the instruments that they hold make clear, either explicitly, as with contingent convertible debt securities (CoCos), which convert into equity upon the occurrence of a specified trigger event such as a capital shortfall, or by necessary implication of the relevant applicable law, that (p. 13) they will be written down in order to recapitalize the bank before any excluded liabilities are written down, or alternatively, that they will be written down before any excluded liabilities are written down under the applicable statutory priority scheme. In principle it is easier for all concerned if the holding company is resolved and the bank subsidiary is untouched, but there is no strong reason why the designation approach should not produce the same result for the same value of designated debt.

2.15  Given this, it should be clear that size is pretty much irrelevant to resolvability—larger institutions need more ‘bail-in-able’ debt; smaller institutions need less or can rely on the relevant deposit insurance scheme, but that is about it. The possibility of resolution is independent of the size of the institution requiring resolution.

2.16  It may therefore be asked how concern about the potential for failure of systemically important institutions became sidetracked onto the issue of ‘size’? The answer is almost certainly that ‘size’ in this context has become confused with complexity. Larger institutions are generally observed to do more things in more places than smaller institutions, and one of the things which became clear during the crisis was that liquidation or a traditional insolvency approach—in which the institution to be resolved is separated into its different business lines and those lines are dealt with separately—becomes impracticable beyond a certain size of institution. Furthermore, the Lehman Brothers case showed that liquidation, without regard to ongoing business, is potentially destructive of value and gives rise to unfair inequalities of treatment of claimants when the business is broken up along legal entity and jurisdictional lines. This is not because separating out a large institution is impracticable per se—any institution can be dealt with in this way given sufficient time—but because in order to effect a resolution of a service provider, one of the key drivers is to avoid interruption of service. An extended closure of the institution concerned pending the completion of this analysis is not a practical approach to resolution.⁴ It is therefore the case that certain institutions can be said to be unresolveable using a traditional insolvency approach under pre-crisis law in most countries (i.e. not in accordance with the Financial Stability Board's Key Attributes for Effective Resolution Regimes). However, it is precisely in response to this discovery that alternative approaches have been developed which are not dependent on an insolvency-based approach, and can be applied at the group level. It is precisely this possibility of group level application which means that the technique can be applied to any firm, regardless of its size.

2.17  This is not, of course, to underplay the importance of complexity. As noted above, a prerequisite for resolution is that resolution authorities should be able to get their (p. 14) arms easily and quickly around the problem to be solved, and in particular to be able to establish reasonably accurately the size of the problem which needs to be solved. This is likely to require continuing work on the part of banks and resolution authorities (acting individually and through crisis management groups), and has not yet been delivered. However, there seems little doubt that it can be and will be, as mandated by the Financial Stability Board (FSB) and G20.

2.18  There is no mechanical rule which can be applied to say with certainty which firms should be resolved, and which may be safely left to fail in a traditional insolvency proceeding. The question of whether any particular institution need to be resolved is determined by the condition of the economy—and the market—at the relevant time. The exemplar for this is the situation in the United Kingdom in 2008, when the UK government was forced—at the height of the crisis—to guarantee the obligations of London Scottish Bank, which at that time had around £250m of deposits. The rationale for this decision was the fear that at that particular point the collapse of any UK bank, no matter how small, might set off a devastating run on the UK banking system as a whole. This startling demonstration of the fact that at the height of a crisis the classes of ‘too-big-to-fail’ and ‘too-small-to-notice’ might overlap was a reminder that any deposit-taking institution might, in certain circumstances, require resolution.

2.19  One of the most important issues with resolution is its impact on the entitlements of participants in the resolved entity. The best starting point for this analysis is to remember that the difference between creditors and equity investors is that equity investors are entitled to everything that is left after the creditors have been paid. Thus, a measure which reduces the claims of creditors by £1 therefore increases the net assets to which equity holders are entitled by £1. Where the value of equity pre-resolution has been reduced to zero, it is a simple matter to say that if creditor’s claims are written down, creditors should receive all the new equity created by the write-down of their claims. However, real life is rarely this tidy.

2.20  It is vitally important in considering bank failure to remember that ‘failure’ in the context of a bank has an entirely different meaning from the term ‘failure’ in the context of a normal firm. A normal firm may continue trading for as long as it has £1 of positive asset value and a reasonable expectation that that value will grow rather than shrink. A bank, by contrast, must close its doors—or more accurately, will be closed by its regulator—as soon as its capital falls below the mandated value set by its regulator or it is likely to run out of a sufficient amount of cash or other liquid assets to satisfy its obligations as they come due. Thus, a failed bank is almost necessarily still solvent, at least in the sense that an insolvency practitioner would understand the term. This, in turn, means that the failed bank is likely to have a significantly positive net asset value, and in turn that its equity may still have a significantly positive value.

(p. 15) 2.21  This raises the fascinating question of the value of such equity. Equity in a bank which is no longer considered viable by its regulator exists in a kind of Schrodinger state, as can be seen in the rescue of Bear Stearns by J.P. Morgan. Although shares of Bear had a book value of US$84 per share immediately prior to its failure, J.P. Morgan initially agreed to pay just US$2 per share, on the basis that had it not been prepared to guarantee the obligations of Bear Stearns, the almost inevitable outcome would have been liquidation, in which it was assumed that the resulting outcome for shareholders would have been complete wipeout.

2.22  There is an interesting discussion of this point in Starr International Company, Inc v United States.⁵ At the point at which the Federal Reserve advanced its rescue loan to AIG, it was common ground that had the loan not been advanced, the firm would have become insolvent, but because of the making of the loan, it did not do so. However, part of the terms of the loan were that then existing equity holders were diluted by the issuance of new equity to the US Treasury equal to 79.9 per cent of the equity of AIG on a fully diluted basis. The US Federal Court of Claims recently held—inter alia—that the issuance of new equity was unauthorized and amounted to an illegal exaction by the US Treasury, but that the US government was not required to compensate the equity holders for any lost value resulting from the unauthorized dilution, because their equity would, but for the Federal Reserve intervention, have been valueless.

2.23  The truth of this proposition can in turn be tested by considering the outcome of the Lehman Brothers administration in the United Kingdom. Upon the failure of Lehman Brothers, it was widely expected that creditors of the UK firm would lose out badly, and claims on the entity traded as low as 30 per cent of their face value. However, the final announcement by the administrators confirms that the eventual payout to creditors was 110 per cent of the face value of their claims (the reason that the payment was more than 100 per cent was that claims in UK insolvency attract interest at what is currently a penal rate (8 per cent simple interest per annum)), and this in turn ensures that there is very unlikely to be any surplus for shareholders). However, the key point here is that creditors were paid in full with a surplus, and the expectation that equity claimants could be disregarded was demonstrated to be—at least—oversimplistic.

2.24  Finally, we note that it must be conceded that not all banks can be resolved. The issue here is that in order to resolve a bank using normal resolution techniques, it is necessary to have a reasonably clear picture of the current state of the bank and a clear pathway towards solvency. Put simply, you can’t know how to fill a potential hole if you can form no reasonable estimate at all of the size of the hole.⁶ Recent (p. 16) experience has taught us that in some cases the quality of the assets of a failed institution is so bad and its capitalization and liquidity position so poor that it is practically impossible to create a worst-case scenario. Without such a scenario, credible resolution is impossible. However, such institutions pose a relatively simple problem. A firm which is not capable of being recapitalized and reorganized through resolution has truly failed, in that the continuation of its business, no matter how desirable, is no longer possible. It must fail and be liquidated in the ordinary fashion—the question is simply one of who pays for that failure. Provided that the answer to this question is that the loss falls on the capital holders, the relevant DGS and uninsured creditors, there is no reason to conclude that impossibility of resolution has any very significant consequences.

2.25  Much of the discussion about bank resolution is predicated on the basis of the simplifying assumption that a bank is a single entity. In economic terms this may be correct, but in legal terms it is clearly not. Most banks, and all systemically important banks, operate as groups of legal entities. In legal terms, groups do not exist—it is only the companies which comprise the group which can enter into contracts, incur liabilities, or fail. This is not, however, the way that economists (or people generally) see the world. Businesses are generally thought of as single undertakings—‘Ford’ or ‘BP’ are unitary concepts. Thus for a lawyer it makes perfect sense to talk of a group being partially insolvent, in that some of its components are insolvent whilst others are not. For non-lawyers, however, the concept is almost meaningless—it is like speaking of a human being as being partly dead.

2.26  However, in the same way that it is possible in emergencies to preserve the life of a living organism by removing dead parts, it is possible in emergencies to preserve the franchise value of parts of bank groups by allowing other parts to become insolvent. To press the analogy slightly further, the question of whether this is possible or not rather depends on the functions of the parts being amputated. There are some parts of a group whose removal can be accomplished without damaging the business of the group as a whole, but there are others whose removal entails the immediate and automatic extinction of the entire organism. It is by no means always crystal clear which is which.

2.27  There is therefore no automatic answer to the question ‘What are we trying to resolve—the group or the bank?’—the only meaningful answer is ‘It depends’. Consequently, it is necessary to think about bank resolution tools not only in the context of individual legal entities, but also in the context of how those tools could be applied to bank subsidiaries within a group, to parent companies of banks, and potentially to non-bank subsidiaries of banks. This is a difficult piece of analysis. To complicate matters further, bank groups are by no means uniform, and different (p. 17) bank managements have different strategies as to how the economic activity of the bank should be reflected in the legal structure of the group.

2.28  Finally as regards groups, one of the key differences in the use of these tools is that the distinction between unitary and ‘archipelago’ banks—that is, between banks whose business is conducted as a single entity versus those banks whose structure is deliberately divided into multiple separable banks—is not relevant to the practicability or otherwise of resolution.⁷ The resolution of an archipelago bank will in general be done as a series of separate resolutions of the troubled subsidiaries concerned (i.e. multiple points of entry), with at least some banking subsidiaries escaping resolution altogether. However, once this allocation has been made, the resolution of the subsidiaries concerned will proceed along the same lines as the resolution of a unitary bank group. The difference here is simply where within the group the resolution action is taken. The topic of resolution at the group level is considered in greater detail in Chapter 3 of this work.

2.30  Further, any intervention in a solvent business is a violation of the freedom to conduct a business as guaranteed by Article 16 of the Charter of Fundamental Rights.⁸ However, the BRRD states (recital 17) that the limitation of that fundamental right is ‘necessary in order to strengthen the business of institutions and (p. 18) avoid institutions growing excessively or taking excessive risks without being able to tackle setbacks and losses and restore their capital base’, and ‘it requires preventative action to the extent that it is necessary to address the deficiencies and therefore it complies with Article 52 of the Charter of Fundamental Rights’.

2.31  Although these concerns are expressed in the context of EU law, the fundamental issues are serious constitutional and human rights law issues in almost all countries. The international position is exhaustively surveyed in an International Bar Association report⁹ which reviewed the position in the G20 countries, and concluded that:

All jurisdictions have confirmed that any constitutional/human rights protection which do accrue to investors in shares or bonds, or contractual counterparties, can be limited under the laws of their jurisdiction if the limitation is imposed in the public interest. In many cases, there are requirements that limitations will be proportionate to the public interest at stake, imposed by statute (rather than by regulation or through the exercise of an administrative discretion) and/or that some form of compensation is attributed.

2.32  Finally, one of the key elements of the resolution planning process is the idea that resolution authorities should be able to require changes to the structure and organization of institutions and their groups. This can conflict with Article 16 of the Charter of Fundamental Rights, the right to conduct business. As a result, authorities’ discretion in this direction should be restricted to what is necessary to ensure resolvability.

2.33  Ironically, the United States may be one of the countries in which the conflict between insolvency laws and individual economic rights is almost never debated today. This lack of any serious debate is almost certainly a result of two historical factors. First, the US Constitution expressly authorizes the US Congress to enact bankruptcy laws,¹⁰ which has always been seen as the authority to create reasonable exceptions to any contractual or other property rights in the event of insolvency. Second, the United States engaged in a vigorous debate over the conflict between economic rights and economic regulation during the Great Depression, and the proponents of a broad power to engage in economic regulation won the day, with the Lochner line of cases, which had granted constitutional protection to economic rights, being overruled and widely discredited.¹¹ While the US Constitution clearly imposes some limits on the bankruptcy power of Congress to interfere with settled (p. 19) contractual and property rights, it is not clear what those limits are and they have not been very protective since the Lochner line of cases were discredited.

2.34  The most serious debate over the scope of the federal bankruptcy power to interfere with economic rights is reflected in litigation over the original terms of the AIG rescue and the unilateral change in the original terms of the rescue of Fannie Mae and Freddie Mac. The plaintiffs in AIG challenged the power of the US Treasury to acquire equity in AIG as part of the government’s compensation for the Federal Reserve’s rescue of that firm. While the plaintiffs nominally won in the Federal Court of Claims, the court refused to award them any damages on the ground that AIG would have been forced into bankruptcy and the entire value of the plaintiffs’ shares would have been wiped out without the rescue. The plaintiffs in the Fannie Mae and Freddie Mac litigation have challenged the power of the US Treasury and the Financial Housing Finance Authority (FHFA) to unilaterally amend the terms of the original deal between the US Treasury and Fannie Mae and Freddie Mac, to implement the so-called net worth sweep, which gave the US government the exclusive right to all of the future net worth of Fannie Mae and Freddie Mac.

2.35  The situation that is most likely to give rise to a serious claim for compensation under the Takings Clause of the US Constitution is one in which a bank is placed into an FDIC receivership before it is balance-sheet insolvent and the FDIC substantially undervalues the bank based on fire-sale prices, distributes that deflated value to the most senior claimants, and wipes out the equity and other junior claims. If the equity or other junior claimants can show that the FDIC substantially undervalued the bank or its assets and distributed a windfall to a third-party purchaser or the senior claimants, the junior claimants should have a strong argument that the FDIC effectively ‘took’ a portion of their property rights without giving the junior claimants just compensation as required by the Takings Clause. The solution to this problem would be for the FDIC to preserve the relative priority of the junior claims, just in case the bank or its assets turn out to be worth more than the FDIC’s (p. 20) valuation based on fire-sale prices, by giving the junior claimants warrants or other participation interests allowing them to participate in any unexpected upside.¹²

2.36  It is sometimes argued that bank resolution is unnecessary, since the objective which it seeks to secure—continuation of services to bank users—is already secured through depositor protection measures such as deposit insurance and depositor preference. In some respects this is clearly wrong, since depositor preference in many countries (in particular the EU) does not extend to anything other than retail and small corporate depositors measures, and deposit protection fund coverage is limited in value. However, even for retail depositors, falling back on a deposit guarantee is a poor alternative to continuation of service. This was dramatically illustrated in the United Kingdom in the aftermath of the failure of Northern Rock, as retail depositors discovered that the fact that they would get their deposit back eventually was not going to help them to pay the rent or buy groceries that week (or, at that time, for many weeks to come).

2.37  This means that bank resolution and depositor protection measures must coexist—they are not alternatives. It is therefore necessary to consider the interaction of the two regimes.

2.38  In general, depositors are protected up to specified limits by deposit guarantee schemes (DGSs), which have the effect of passing any loss that would otherwise be suffered by such depositors through to the other banks in the relevant banking system. The fact that a DGS distributes losses within a system makes clear why such schemes are irrelevant in a systemic crisis—claims on a DGS are valueless if all of the contributories are unable to contribute.¹³ However, in the event of institutional (p. 21) failure within a basically sound system, the DGS is the primary protection for depositors within that system.

2.39  The fact that calls on a DGS are allocated to other banks in the relevant system also provides (in theory at least) an economic incentive for banks to monitor each other’s risktaking, and gives a strong interest in the efficient resolution of the failed bank in a way which minimizes losses to insured depositors, and therefore to them.

2.40  It is therefore important to all systemic participants that the involvement of the DGS in a bank resolution is structured so that the resources of the DGS are deployed in a way which minimizes losses to its contributors. This means that the DGS should not be restricted only to bearing losses after failure, but should be able to deploy its resources—cooperatively with other parties—to reduce the magnitude of the losses actually suffered. This is sometimes referred to as ‘bailing in the DGS’, but this concept is unhelpful, since it implies voluntary assumption of losses by the DGS. In a classical DGS rescue, the DGS expends a small amount of money voluntarily in order to avoid the risk of suffering a greater loss at a later stage.

2.41  A further element to be addressed is the fact that in general immediate funding for a DGS is provided by the relevant government. All DGSs are either explicitly or implicitly guaranteed by their governments, on the basis that no government would in practice allow the DGS in its territory to collapse (although, as was seen in the Iceland case, a government may well be prepared to cut loose overseas depositors in its banks in order to rescue its domestic system¹⁴). Thus where a DGS is called upon for any commitment, the immediate source of the credit exposure behind that commitment is the relevant government, even if the ultimate cost is charged back to the DGS members over a somewhat longer timescale. This point has been used to argue that involvement of the DGS in bank resolution is in fact nothing more than a disguised form of government bail-out, since in the immediate response to the crisis it is the government’s credit which is committed. It is argued that since this approach is in principle no different from a direct commitment of taxpayer’s funds accompanied by the imposition of a tax on the banks over a subsequent period (as has been the case in, for example, the United Kingdom), the two are identical.

2.42  It should be noted as a preliminary point that this argument is equally applicable to pre-funded and post-funded schemes. Although in pre-funded schemes it is possible that very small institutions may be resolved using only the funding already in existence, pre-funded schemes do not in general hold sufficient funds (p. 22) to meet even a significant fraction of the deposits held by a G-SIB or D-SIB.¹⁵ A pre-funded scheme is simply a scheme where the scheme operator has sufficient funds in its pocket to cover some level of expected loss—for any level above that, its recourse to its sponsoring government is the same as that for an unfunded scheme.

2.43  There are a number of important distinctions which can be made between DGS bail-in and government bail-out. However, the most important is simply that the costs incurred by a DGS should ultimately fall back on the other members of the financial system concerned.¹⁶ There is an element of intertemporal allocation here, in that the aim of the government intervention is generally to transfer the liability from the system during the crisis, when it is weak and less able to absorb the losses, to the same system post-crisis, when it will have recovered and will be more able to absorb the losses. However, assuming that the system remains in being, there is never any issue of the exposure incurred by government falling anywhere other than on the system participants concerned in proportion to their liability to contribute to the scheme—a liability which bears no other resemblance to a tax. Put simply, in order to argue that a DGS bail-in constitutes taxpayer assistance, it is necessary to argue that DGS contribution levies are taxes—a proposition which would, at the very least, sound strange to any competent government statistician or administrative court.

2.44  Cross-border issues with DGSs are notoriously difficult. The current trend is towards requiring at least retail (and therefore insured) deposit-taking to be conducted within local subsidiaries in each country—an approach which has the merit of ensuring that each local banking subsidiary is part of the local DGS. However, where deposits are taken through a cross-border branch network the problem swiftly becomes intractable—the Icesave experience in the United Kingdom¹⁷ dramatically illustrated the fact that a foreign DGS is ultimately backstopped by the sovereign of the jurisdiction in which the bank concerned is established, and if that sovereign is unable or unwilling to finance the compensation of foreign depositors, this creates a gap which will—in practice—need to be filled by the banks or government of the host jurisdiction. The structuring of DGSs as regards cross-border deposit-taking remains an unresolved issue.

2.45  The final point to make in this context is that DGSs are tools created to absorb losses from a single bank failure where that failure occurs within a solvent banking system. In a systemic crisis, a DGS may well turn from being part of the solution to being part of the problem, and in particular may become a channel for the transmission of systemic risk. The point is clearly made in the BCBS/IADI Core Principles for Effective Deposit Insurance Systems:

A deposit insurance system is not intended to deal, by itself, with systemically significant bank failures or a ‘systemic crisis’. In such cases all financial system safety-net participants must work together effectively. In addition, the costs of dealing with systemic failures should not be borne solely by the deposit insurance system, but dealt with through other means such as by the state. (p. 3)

2.46  This point is also picked up in the Iceland decision. In defending discrimination by the Icelandic authorities against non-Icelandic depositors under the Icelandic depositor protection arrangements, the court said that ‘EEA [European Economic Area] states enjoy a wide margin of discretion in making fundamental choices of economic policy in the specific event of a systemic crisis’. Put simply, not only are DGSs not designed to be used in the event of systemic crises, but it may be unwise to count on their being available in the event of such crises. This was, of course, equally visible during the Cyprus crisis—once all of the banks in a particular country are in trouble, recourse to the DGS of that country ceases to be a relevant consideration.

2.47  Increased focus on the role of the DGS in resolution, and on the possible function of a DGS as a transmitter of financial vulnerability within the banking system, has resulted in increased interest in depositor preference.

2.48  Depositor preference means a legal priority in insolvency for depositor creditors of a failed bank over claims of all other senior, unsecured creditors (subject perhaps to limited wage and trade claims). Depositor preference has a long history in the United States and other countries, and is generally based on the idea that depositors are the people whom bank regulation seeks to protect, and that therefore in any competition between depositors and others for the assets of the bank, depositors should have preference. However, a number of countries have actively rejected the notion, on the basis that the resulting interference with the rights of other creditors stunts the ability of the institution concerned to do business.

2.49  Depositor preference is also sometimes territorial. Thus, for example, in the United States, deposits payable in the United States (domestic deposits) are preferred, such that after bank capital the burden of losses incurred by a failed bank falls on the (p. 24) other creditors of the bank, including the holders of deposits payable solely outside the United States (foreign branch deposits),¹⁸ before it falls on the holders of domestic deposits. The effect of this arrangement in practice has been that the assets of most failed institutions have been insufficient to satisfy the claims of both the holders of domestic deposits and the holders of any claims junior to domestic deposits. For example, based on our review of ten of the largest institutions to fail in the United States in 2008 and 2009, domestic deposits generally accounted for at least 97 per cent of the failed institution’s liabilities,¹⁹ and the value of each institution’s assets averaged just 65 per cent of total liabilities at the time of failure.²⁰ Under these conditions, the assets will be insufficient to satisfy the claims on all domestic deposit liabilities; as a result, nothing will be left for general unsecured claims or any other claims junior to domestic deposits.

2.50  Once depositor preference has been established, the structure of a bank is generally likely to change such that any creditor who is not a depositor may seek to become a secured creditor. The fewer the unsecured creditors, the larger the risk to those creditors, and therefore the smaller the pool of unsecured creditors is likely to be. This means that the introduction of depositor preference may result in a greater likelihood that depositors (or the DGS) will have to be bailed-in in the absence of any other bail-in-able senior creditors.

2.51  Ironically, the domestic deposit liabilities of the US G-SIBs (except for Morgan Stanley Bank) accounted for a substantially smaller percentage of their total (p. 25) liabilities compared to smaller more domestic banks.²¹ All of the G-SIBs (except Morgan Stanley Bank) had significant amounts of liabilities that are subordinate to domestic deposit liabilities, including foreign branch deposits, federal funds liabilities, trading liabilities, other borrowed money, subordinated debt, and other liabilities.²² As a result, the value of the assets of such a bank in receivership is far more likely to be sufficient to satisfy all the claims on domestic deposits, thus resulting in no loss to the US deposit insurance fund and allowing for some distribution to junior claimants.²³

2.52  Territorial depositor preference has sometimes been alleged to be disguised protectionism. This was the allegation which was rejected in the Iceland case,²⁴ but in that case the rejection was based on the fact that the systemic crisis which the Icelandic government faced was so severe that it could not have met its obligations extraterritorially. In the United States, by contrast, territoriality is a matter of course, since all claims on deposits payable in the United States have preference over claims on foreign branch deposits that are payable solely outside the United States (although not foreign branch deposits that are dually payable at the foreign branches and in the United States).²⁵ There is more or less universal agreement that such structures are destructive and should be avoided. However, this is broadly where the consensus ends.

2.53  In general, depositor preference comes in two forms: preference up to the insured limit, and unlimited preference.²⁶ Depositor preference up to the insured limit is, properly viewed, not depositor preference but DGS preference—the depositor will be compensated out of the DGS, and is therefore indifferent to the preference (p. 26) position, whereas the contributories to the DGS will get the benefit of the preference through their subrogated claim against the failed institution.

2.54  Unlimited depositor preference is generally confined to certain classes of depositors (e.g. it is unusual for deposits placed by financial institutions to receive this preference). However, protected types of depositors include individuals, small- to medium-sized enterprises (SMEs), pension funds, and charities.

2.55  The United States, in particular, has an unlimited depositor preference regime which covers (broadly) all deposits payable in the United States (but not foreign branch deposits payable solely outside the United States). This has the immense advantage of avoiding the necessity for ‘line-drawing’ for protected classes. The FDIC believes that this unlimited depositor preference has facilitated US bank resolutions, since it means that entire deposit books can be easily and rapidly dealt with provided that the total assets of the institution in resolution exceed the value of the deposit book. However, an object-lesson in the dangers of this approach was given in the collapse of the Cypriot banks, which were almost exclusively deposit-funded, in that once the value of the deposit book falls below the asset value (plus the resources of the DGS), the allocation of pain amongst depositors involves an emergency triage of depositors into classes who are to be more or less affected by the loss, and such emergency triage is both politically painful and, almost by definition, unjust in certain cases.

2.56  The introduction of depositor preference significantly changes the dynamics of bank resolution. This is because by changing the position on liquidation, depositor preference automatically changes the extent to which the creditor concerned can be bailed in without breaching the ‘no creditor worse off than in liquidation’ (NCWOL) principle. Thus, where a bank has eighty of assets, eighty of deposit liabilities, and twenty of other liabilities, the insolvency position will be that the depositors will make full recovery and the other creditors will receive nothing. This situation may have advantages—for example, the FDIC experience is that this is helpful for the conclusion of purchase and assumption (P&A) transactions, since if the assets exceed the value of preferred deposits by even one cent, it then becomes possible to transfer all of the preferred deposits to a replacement institution (or a bridge bank) without contravening the ordinary rule of priority. The converse of this position is that unsecured creditors other than depositor creditors commonly receive no recovery at all in FDIC resolutions.