3.01  The easiest mental model to use in understanding bank groups is probably the Marxist model. Marx regarded society as composed of the ‘base’—the forces and relations of production which constitute economic reality—and the ‘superstructure’—culture, institutions, and social norms. Base determines superstructure, and a failure to perceive the realities of the base—or a belief that superstructure has any ‘real’ existence—constitutes false consciousness. Bank groups can usefully be considered using this paradigm. The ‘base’ comprises the systems and processes which conduct the banks’ day-to-day business, whereas the ‘superstructure’ is the legal construct which sits on top of the base. In analysing the bank itself, a focus on legal structures is a form of false consciousness. In determining whether an entity can continue to function, what matters is not whether the legal entities are solvent on an accounting basis, but whether the underlying systems are continuing to operate. The failure of Lehman Brothers International (Europe) provides a dramatic demonstration of this proposition—when the systems stop working, the institution is finished, and the notional solvency or otherwise of the legal entities is a detail for historians rather than a material fact.

3.02  The reason that this is important in the international context is that in most situations where banks operate through different national subsidiaries, it is highly likely that their operational, payment, and functional activities will be conducted through bank-wide systems. Even in contexts where bank regulators have required national subsidiarization they have generally not gone so far as to require the maintenance of separate free-standing national operational systems—generally because such a requirement would add substantially to the service costs incurred by national customers. However, in the absence of such a requirement, the question of the possible survivability of the national subsidiary is a function of the continuing existence of the underlying systems. This has a number of consequences. One is that if the architecture of the bank is such that the system concerned is effectively operated by the troubled institution, then the failure of that institution will necessarily cause cessation of operations throughout the group. In order to address this issue without (p. 28) fragmenting operational systems in a way which would create massively increased costs, it is clearly necessary to create some degree of independence for the function concerned. However, any significant reconstruction of bank systems would impose costs which are very significant on banks, and such costs (payable as they are out of profits) would directly impact capital levels and further inhibit banks’ ability to produce demand deposits and other forms of money and create extensions of credit. In short, it is by no means clear that any G-SIB break-up would be feasible without requiring in effect a complete reconstruction of the bank itself.¹

3.03  Lawyers (perhaps justifiably) tend to perceive corporate groups on a legal entity by legal entity basis. However, this is not in general how groups are either managed or resourced. A simplified model of a conventional bank group might be as shown in Figure 3.1.

• View full-sized figure

Figure 3.1  Simplified model of a conventional bank group

3.04  The key point here is that each of these layers will be subdivided. Legal structure will be subdivided into individual legal entities, IT infrastructure will be subdivided into different systems, and management structure will be subdivided into business areas. These subdivisions are not necessarily congruent with the subdivisions at other layers.

3.05  Sometimes one or more of these will conjoin. In Figure 3.1, the column on the right is most likely to represent a newly acquired subsidiary, where management, systems and legal structure will all—at the point of acquisition—be discrete. However, over time the process of integration of such business into the parent group is likely to result in any of the three layers being merged—booking may be transferred to a new legal entity, management may be restructured within the wider group, and IT systems may be integrated. None of these processes are generally related to the others—banks do not generally prioritize legal structures when designing management processes or IT systems, or IT functionality when designing trade booking structures.(p. 29)

3.06  The effect of all this, however, is to expose as an illusion the idea that because business is conducted within a particular legal subsidiary, it is therefore segregated—or capable of being easily divided from—the other activities of the group. A subsidiary is, in legal realist terms, simply a few lines in a company registry—the question of whether a particular business can be separated and easily sold from a group is much more likely to be determined by its management and control structures than by the legal substructure of its contracts.

3.07  Thus far we have considered the bail-in of a bank. However, most large banks are members of groups, and it is frequently the case that in a bank group there is an unregulated bank holding company above the bank.

• View full-sized figure

Figure 3.2  Resolution at the bank level

3.08  In the case illustrated in Figure 3.2, if the resolution were to be conducted at the bank level the effect of the resolution would be to break the group structure (since the bank would cease to be a subsidiary of the holding company)—and would be likely to push the holding company into insolvency (since the ability to liquidate the shares would become deferred and its dividend flow from the regulated bank would cease). This problem clearly would not arise if the senior borrowing were primarily at the holding company level. However, practices vary amongst banks as to whether funding is (a) raised at the holding company level and downstreamed, (b) raised at the level of the bank itself, or (c) raised at both levels. Thus it is impossible to make any general assumption as to where in the group eligible debt will be raised.

3.09  This means that the structure of any resolution must be adapted to the specific case of the bank group concerned. This can be most easily understood by considering the case of the slightly more complex bank group illustrated in Figure 3.3.

• View full-sized figure

Figure 3.3  Resolution of a more complex bank group

In this case, the group has creditors at multiple levels and within multiple legal entities.

3.10  The starting point for consideration of this situation should be the fact that it is desirable for both the group and for its regulator that creditors should be clear (p. 30) which parts of the group they are exposed to. Creditors of the holding company will clearly consider themselves exposed to the group as a whole, and creditors of the booking vehicle will consider themselves exposed—in credit terms at least—solely to that booking vehicle. However, for the US subsidiary bank, for example, the question of whether that bank would, in difficulty, be able or entitled to call upon the resources of the remainder of the group would have to be determined as part of the living will process.

3.11  What all of this comes down to is nothing more complex than that the resolution structure should reflect the existing structure of the group and the expectations of creditors as to their counterparty exposures. This is neither unreasonable nor overly challenging, and certainly does not represent an insuperable barrier to the establishment of resolution regimes, at least where the bank holding company and the entities issuing eligible debt are incorporated in the same country. The position is more challenging where the group has a more subsidiarized structure containing large entities issuing debt that are not incorporated in the same country as the bank holding company, as the local regulator will not have direct powers over the holding company to implement the resolution. Again, this is not a fundamental objection to the adoption of a resolution approach by regulators for those groups for whom resolution is possible and at present most large banking groups with a bank holding company structure have their bank holding company incorporated in the same jurisdiction as their principal banking entities. It also potentially aligns the interests of regulators in effective resolution planning with the interests of banks in operating integrated legal entities operating internationally through branches rather than subsidiaries.

(p. 31) 3.12  However, cross-border issues should not mean that it is impossible to implement a resolution in a case where the troubled bank or its bank holding company also has troubled foreign subsidiaries. In many cases, the parent will have substantial assets that can be used to recapitalize its subsidiaries and its subsidiaries will have significant intra-group borrowings from their parent bank or bank holding company owing to the downstreaming of funding raised at the parent level. The bail-in of debt at the level of the parent bank or bank holding company should create enough capital to give it the capacity to contribute assets to its subsidiaries or write off or convert into equity loans to its subsidiaries, enabling their recapitalization as part of the overall process, even if the local regulator does not have an effective resolution regime. This may have the result of insulating the creditors of those subsidiaries against losses, at the expense of the parent’s creditors, but this will not be unfair or create moral hazard as long as it is known in advance that debt incurred at the parent level is structurally subordinated to debt incurred at the subsidiary level. If it is, market forces will cause the return on debt incurred at the parent level to go up to reflect the greater risk of such debt and the return on debt at the subsidiary level to go down to reflect the lower risk of such debt. In effect, creditors at the subsidiary level will be forced to internalize the cost of their preferred position, thus driving out any moral hazard, and creditors at the parent level will be fully compensated for the additional risk they bear. In addition, if local regulators have a local resolution regime with corresponding powers, there should be ways in which the group’s lead regulator and the local regulators can coordinate the exercise of their powers to produce an appropriate result, without the need (outside the European Union) for complex international treaties which could take many years to negotiate.

3.13  A resolution of a bank subsidiary within an integrated group would ordinarily require external senior creditors of that bank subsidiary to be issued new shares in that subsidiary. However, such issue could result in the subsidiary ceasing to be part of the integrated group; an outcome which might not be practical either for the subsidiary bank or the group. Consequently, if the resolution authority wishes to preserve the integrity of the group, it may arrange for the written-down creditors of the subsidiary to be given new equity in the group parent. The write-down or -off of the eligible debt in the subsidiary would create capital reserves in the subsidiary. The issue of the shares in the holding company does not necessarily require any intermediate step of requiring the subsidiary to issue additional shares or debt to its parent company, but the mechanics would depend on local corporate law sensibilities. There is no reason why this should not be done by statute, and if the bank and the holding company are established in the same jurisdiction, a legislative solution in that jurisdiction should be capable of being crafted.

3.14  In the context of groups, it is also important to note that a consistent policy would be required as regards debts to (p. 32) the top-tier parent. The question of whether debts to the top-tier parent should be treated differently from any other debt in the context of a resolution is not straightforward. However, it is by no means clear that it is necessary to have a single answer to this question on a global basis—the optimum solution would seem to be that this issue should be addressed between individual banks and their lead regulators as part of the ‘living wills’ discussion. Again, this may be easier to accomplish in the ‘targeted’ approach where banks continue to be able to issue senior debt that is not eligible for bail-in alongside subordinated debt that is eligible for bail-in.

3.15  A further problem potentially arises within groups as regards the transmission of capital. In general, where a member of a group has surplus capital, if another member of the group is in need of capital, a number of mechanisms exist for transferring that capital within the group.

3.16  At its simplest, the transferring entity can use its excess equity to purchase new shares in the transferee entity. However, this is generally not permitted where capital is to be transmitted upwards, since subsidiaries generally cannot buy shares in their own parent. Alternative mechanisms exist—the entity which has excess capital can distribute its excess capital up the chain in the form of dividends until it reaches the parent holding company, at which point the parent holding company can downstream the excess capital to the subsidiary which is short of capital. This is, of course, subject to any restrictions which may be imposed by local corporate law.

3.17  An alternative is the indirect creation of capital by the forgiveness of intra-group debt. This is an effective mechanism (cancellation of debt results in an automatic increase in shareholders’ funds), but relies on there being forgivable debt in place, and on the directors of the forgiving company being confident that ‘giving away’ a company asset is within their powers and duties.²

3.18  Another alternative is the capital contribution—a straightforward gift of money or other property from one company to another—although there are sometimes accounting difficulties with having capital contributions recognized as capital.

3.19  In practice, there are a host of tax, accounting, and regulatory rules which can inhibit the use of any of these mechanisms. These rules are difficult enough in one jurisdiction, but rapidly become a major obstacle when transfers between a number of different jurisdictions are involved.(p. 33)

3.20  Bank groups are protean—not only are they very different one from another, but also they may change significantly as the business of the bank changes. As should have been clear from the foregoing, generalization about ‘bank groups’ are impossible because each large bank group is to some extent unique, and even simple generalizations about ‘the holding company’ or ‘the group’ are liable to counterexamples. However, it is to some extent possible to separate bank groups into broad types, and we suggest here a taxonomy which may enable some progress to be made in addressing resolution options.

3.21  For the purposes of the examples that follow, we have divided creditors into three broad types:

• Banking creditors, meaning retail and wholesale depositors and creditors arising out of the provision by the bank of payment and custody services;

• Investment business creditors, meaning swap counterparties, trading counterparties, exchanges, clearing systems, and other investment business counterparties (including repurchase agreement (repo) counterparties).

• Capital structure creditors, meaning long-term unsecured creditors of the bank, including bondholders and other long-term unsecured finance providers.

• View full-sized figure

Figure 3.4  The ‘bank holdco’ model

3.22  In Figure 3.4 we see a more or less ‘empty’ holding company holding a bank with a large balance sheet. Assets not held within the bank itself will generally be held by subsidiaries of the bank. Funding is likely to be raised primarily at the bank level, since any funding raised at the holding company level is structurally subordinated to funding raised at the bank level. However, regulators will require a certain type and amount of financial funding to be raised at the holdco level, (p. 34) since this will be the first in the firing line in the event of single-point-of-entry (SPE) resolution at the holdco level. In general derivatives, markets and trading business will all be done out of the bank rather than the holdco, since the bank will be the most creditworthy member of the group and will ensure that counterparties have the lowest risk exposure (and therefore the lowest costs of dealing with it).

3.23  In the context of this institution, two issues arise. One is that it is unlikely that short-term unsecured investment business creditors will be written down through the resolution mechanism, as long as the bank group has sufficient capital structure liabilities to absorb losses and fully recapitalize the bank group. This is because avoiding such write-downs is likely to maximize the residual value of the bank for the benefit of its capital structure financial creditors, reduce the risk of runs that could destabilize the banking system, and allow operations that are critical to the market to be continued—three important policy objectives for the resolution authority. Further, short-term unsecured bank depositors are also unlikely to be written down for similar reasons, as long as the group has sufficient capital structure liabilities to absorb losses.

3.24  Mechanically, resolving the holdco model is in some respects the easiest challenge. If creditors are at the level of the holdco, it is a relatively simple matter to extinguish their claims and issue them with new shares in the holdco. The holdco’s assets can then be downstreamed to the bank.

3.25  The position of creditors of the holding company is not, however, as simple as it seems. Orthodox corporate finance would dictate that by becoming creditors of the holdco, they have voluntarily accepted a position where they are structurally subordinated to creditors of the bank. It is therefore highly arguable that such creditors should be written down first before direct creditors of the bank are affected. However, most creditors of holdcos assume that status as a result of diligent enquiry as to the credit standing of the holdco concerned. If the structure of that holdco was that all of the funding which it raised would be downstreamed as equity, it would be clear that they were in fact no more than synthetic equity investors in the relevant subsidiary.

3.26  This is not, however, the position of any bank holding company in real life. Bank regulators have been concerned for many decades about the idea of ‘double leverage’—that is, the situation where a parent raises debt capital and downstreams it to its subsidiary as equity capital, thereby enabling the subsidiary to raise more debt on the strength of its enhanced equity capital position. The regulatory tool which is usually applied to prohibit this is consolidated supervision, which requires that a bank group should have sufficient equity at the consolidated level to meet the risks on its total asset portfolio. A bank group which engaged in significant double leveraging would in general simply cease to meet its consolidated capital ratio and be closed down by its regulators. In addition to this, rating agencies generally (p. 35) publish for any rated bank group a ‘double leverage ratio’ which makes clear to investors the extent to which debt has actually been downstreamed as equity. The upshot of all of this is that a person who becomes a creditor of a holdco in fact does so on the basis that (a) it is highly likely that money attributable to debt financing will be downstreamed as debt and not as equity, and (b) that although the person has no direct controls or covenants to ensure that outcome, she can in general rely on regulators to ensure that it is done, and on rating agencies to tell her if it is not done.

3.27  What follows from all of this is that to say that a creditor of a bank holdco has voluntarily accepted the status of a subordinated creditor of the holdco’s subsidiaries is in some way wide of the mark. Unless corporate finance law in the jurisdiction of the subsidiary deems intragroup creditors to be subordinated to external creditors,³ or intragroup creditors contractually agree to be subordinated to external creditors, intragroup creditors are entitled to no worse treatment than the treatment which would be accorded to any other creditor of the subsidiary concerned. This position is illustrated in Figure 3.5.

• View full-sized figure

Figure 3.5  Structure of creditors

3.28  The senior creditors of the holdco would expect the holdco as internal creditor of the bank to be in exactly the same position as the external creditors of the bank. The bank has, in aggregate, 200 creditors; if it fails, those creditors are entitled to be treated pari passu, and the recoveries of the holdco as internal creditor will be no different from those of the external creditors of the bank—if the external bank creditors recover 50 cents on the dollar, the holdco can expect to recover the same on its internal credit to the bank.(p. 36)

3.29  The difference, of course, is that in order to recapitalize the bank subsidiary, it will be necessary for the holdco to downstream assets to the bank, including by forgiving any internal credit to the bank, with the external holdco creditors being written down to the extent the remaining assets of the parent (including its ownership in the recapitalized bank) are insufficient to satisfy all of the claims of the external holdco creditors. There are fundamentally two options here. One is to write down all creditors of the bank equally. This could result in the bank passing outside the holdco group, since the aggregate amount of the claims of the external senior creditors of the bank could be greater than the amount of the holdco’s claims on its internal credit to the bank. The other is to discriminate between different creditors—internal and external, bank and group—according to some other plan. The key issue here is that if creditors come to believe that the latter might be the approach taken, it will be extremely difficult for them to value their investments unless the authority concerned is prepared to make a clear statement as to what that plan may be, and to stick to it in practice.

• View full-sized figure

Figure 3.6  The ‘big bank’ model

3.30  In Figure 3.6 we see the traditional European bank structure, in which a large bank is itself the top company of the group. Assets not held within the bank itself will generally be held by subsidiaries of the bank. Funding is raised primarily at the bank level, although secured creditors may be creditors of ringfenced subsidiaries. In general, the ‘big bank’ is likely to do its derivatives, markets, and trading business out of the main legal entity, since this will be the most creditworthy member of the group and will ensure that counterparties have the lowest risk exposure (and therefore the lowest costs of dealing with it).

• View full-sized figure

Figure 3.7  The ‘bank/non-bank holdco’ model

3.31  In Figure 3.7 we see a holding company which owns a bank and a non-bank investment firm. These activities are likely to be ringfenced by local legislation into a ‘bank chain’ and a ‘non-bank chain’, with restrictions on transactions between the two sides of the group below the level of the holding company. In this case, it is more likely that significant capital structure liabilities may have been incurred at (p. 37) the parent company level, since lenders at that level will have access to a larger asset pool than lenders to the bank. It is possible that significant capital structure liabilities will have been incurred at the bank level. Indeed, it is possible that all three components—the bank, the investment firm, and the holding company—may have incurred capital structure liabilities.

3.32  We need to begin with a hypothesis as to where in the group the loss has been incurred. For the purposes of this chapter we will assume that the loss has been incurred in the banking part of the group.

3.33  At the level of the bank itself, the issues here are no different from the ‘big bank’ model. Considering the position of the investment firm immediately raises the ‘dead in parts’ problem. It should be remembered that in this context it is highly likely that the bank and the investment firm will share the same branding, the same advertising campaign, and the same IT, processing, and payment systems. As a result, it may well be the case that the survival of the investment firm will be entirely dependent on the survival of the bank. Clearly, if the resolution can be conducted entirely at the group level, that is likely to be the optimal solution.

• View full-sized figure

Figure 3.8  The ‘global multi-bank’ model

3.34  In Figure 3.8 is a more or less empty holding company that owns a number of banks—generally incorporated in different jurisdictions and subject to some degree of restrictions on their interconnection. In this case, it is likely that at least some capital structure liabilities have been incurred at the holding company level, although it is likely that some (but perhaps not all) of the subsidiary banks will also have incurred some external capital structure liabilities.

3.35  Resolving the global multi-bank is more interesting than in the previous cases. The architecture of the global multi-bank is generally in response to pressures from national regulators who require national businesses to be undertaken by separately capitalized local subsidiaries. Since we have hypothesized that the holding company is ‘empty’ (i.e. has no economic activity of its own), it must follow that the loss causing the crisis must have been experienced in one or other of the bank subsidiaries. The effect of a write-down of the holdco’s capital (p. 38) structure liabilities is to create new equity at the holdco. The holdco can create new equity at the bank which has suffered the loss by forgiving intra-group debts owed by the bank to the holding company in respect of funding previously received. However, if the holding company has insufficient assets that can be downstreamed (including an insufficient amount of intragroup receivables to forgive) to the troubled subsidiary bank in order to recapitalize the bank, there could in extremis arise the possibility of bailing in external capital structure financial creditors of the troubled bank in order to recapitalize the troubled bank. It might even be possible to bail in any external capital structure financial creditors of the sister banks and distribute any of their excess assets to the bank that needs to be recapitalized.

3.36  The permutations in this regard are complex and difficult. Considering the group above, if Bank A gets into trouble and its own capital structure liabilities are insufficient to recapitalize it, should the capital structure creditors of Bank B be called on? If they are, how does the capital get transferred from Bank B to Bank A? What if Bank C (which has no eligible debt) gets into difficulties—should the capital structure creditors of Banks A or B be written down and their excess assets be contributed to Bank C to recapitalize it? To complicate matters further, if the resolution of Bank A results in majority control of Bank A being transferred to the capital structure creditors of Bank A, those creditors may take advantage of their status as the controlling shareholders of Bank A to restrain the new capital thus created from being transferred elsewhere within the group.

3.37  It is clear that these are no more than illustrations of broad classes of group structures, and it should also be clear that in each case the theoretical deployment of (p. 39) exposures would be dependent primarily on the type and volume of funding raised at each legal entity within the group.

3.38  Much of the discussion about bank regulation and resolution has been conducted in the context of a discussion about branches and subsidiaries—in particular, the extent to which resolution authorities may decide to protect their national economies by requiring overseas banks operating in them to operate through a domestic subsidiary. The argument, in broad terms, is that branch status reduces the level of control available to national resolution authorities in respect of the national businesses. The basis of this approach is articulated in the International Monetary Fund (IMF) paper on ‘Subsidiaries or Branches—Does One Size Fit All?’:⁴

In the absence of effective international cooperation in [bank resolution], resolution of institutions, in the event of a failure within a banking group, may be less costly and less destabilizing if these entities are organized as subsidiaries.

3.39  This suggests that in general it should be possible for a state to argue that it has a positive public interest in requiring banks to operate through subsidiaries in its jurisdiction because ‘healthy subsidiaries that operate independently of the parent may, in principle, be better able to survive the failure of the parent or other affiliates’.⁵ It is important to emphasize that the IMF position is that such concerns are only valid whilst there is no substantive progress internationally on satisfactory cross-border resolution regimes.

3.40  It is also important to note that it is generally an error to think about ‘subsidiary’ status as being a monolithic set of concepts common to all countries. The rules applicable to foreign branches vary wildly in different jurisdictions, and range from countries such as India and Brazil, where branches face local capital and liquidity charges identical to those applied to subsidiaries and are required to maintain boards with local representation, to the countries of the European Union, amongst whom branching is an absolute right with no interference (in theory) permitted to the host regulator. Even between these extremes the position varies considerably—for example, many would expect to find the rules of the United Kingdom and the United States to be reasonably similar as regards foreign branches, but as Table 3.1 illustrates, the differences remain substantial.

3.41  Thus it is important to remember that the regulation of branches is already protean, with no globally established norms for dealing with them.(p. 40)

3.42  There is a commonly held belief amongst some commentators (and politicians) that the position of creditors and authorities, as regards a bank business in a country, is improved if that business becomes a separately incorporated subsidiary in that country. As noted above, this is simply wrong. It is incorrect to say that branch or subsidiary architectures are necessarily superior or inferior to each other—these are simply legal concepts which trigger particular legislative provisions, and the provisions triggered have compensating advantages and disadvantages in both cases.

3.43  The easiest way to address this is to begin with the taxonomy of banks put forward by the Federal Reserve in the discussion section of its notice of proposed rulemaking for enhanced prudential standards applicable to certain large foreign banking organizations (‘FBO proposal’).⁶ This divides federal branches into aggregators and net recipients of intercompany funding. An aggregator branch is established to accumulate deposits (or other forms of finance) in its host jurisdiction and to remit those deposits to the bank’s offices and affiliates outside the host jurisdiction. One of the reasons given by the Federal Reserve for the FBO proposal was that US branches of foreign banks have become aggregators, and that their effect is to use US deposits to fund the accumulation of assets by its non-US affiliates outside the United States. The concern with aggregators is that in the event of a financial crisis, the assets financed by host country deposits are outside the control of the host country resolution authority concerned, and if the home authority is not prepared to remit the assets (or their value) to the host country resolution authority, the host country authority may be left with an insufficient amount of assets to satisfy all the claims against the branch. Requiring an aggregator branch to become a subsidiary may solve this problem if it has the effect of ensuring that the subsidiary has enough assets to pay its liabilities. However, if the problem to be addressed is simply that assets owned within the jurisdiction are located outside the jurisdiction, it seems completely irrelevant where the legal entity which is the asset owner is located. Put another way, it should be possible to require any asset owned by a branch to be returned in the same way and to the same extent that it would be possible to require an asset owned by a subsidiary to be returned.

(p. 42) 3.44  Net recipient branches are the mirror image of aggregator branches. A net recipient branch exists to make loans and/or originate assets in the host country, deploying funding raised outside it. A net recipient branch is always ‘solvent’, to the extent that local assets generally exceed local liabilities. However, a net recipient branch is likely to be highly illiquid, since a central model of funding is almost invariably accompanied by centralized liquidity management. It is possible for a net recipient branch to exist in subsidiary form provided that the business is very small relative to that of the parent bank—if it is not, then exposure limits within the subsidiary bank will cap its recourse to group funding. In general, regulators should prefer net recipient branches to be branches, since otherwise limits on intra-group exposures will disrupt the business model.

3.45  Regulators’ concerns about net recipient branches are generally focused on the withdrawal of credit issue. If a bank with a net recipient branch comes under stress, it is a fairly safe assumption that the bank will cease to roll over credits in the net recipient branch, repatriating assets to the parent to assuage the stress. Hence the argument that requiring a net recipient branch to subsidiarize will prevent this happening.

3.46  A further issue relates to costs of funds. Assume two countries, A and B. A is solvent, has a strong economy, and therefore relatively lower costs of funds for domestic institutions. B has a weak budgetary position and a declining economy, and therefore higher costs of funds. It makes perfect sense for banks established in A to establish net recipient branches in B, and this development should be beneficial for both A and B—A because it derives an increased return on investment, and B because this will reduce the cost of funds to industry in B. Authorities in B may therefore have reasons to incentivize the establishment of net recipient branches. Conversely, however, the authorities in A may object to their banks behaving in this way, since they will put their assets at increased risk of default or appropriation. This is an example of a situation where the home authority may prefer overseas operations to be done through a subsidiary whilst the host authority would strongly prefer these activities to be done through a branch. However, if we reverse the hypothesis, and assume that banks in B seek to deal with the position by establishing aggregator branches in A, the authorities in A may be unwilling to assent to this unless the business in A is done through a subsidiary.

3.47  The strength and weakness of national supervisors is also an issue. A subsidiary enables a host authority to exercise some greater degree of supervision over the business practices of the entity concerned. However, a more accurate way of looking at this is that a host country regulator who permits an overseas bank to open a branch in its jurisdiction places considerably more faith in the home regulator of that bank than one that requires a subsidiary. Thus, regulators who perceive themselves as ‘strong’ may be unwilling to permit branches of banks which they perceive as being established in home jurisdictions with ‘weak’ regulators.

(p. 43) 3.48  Finally, there is the sovereign issue. Although in theory sovereigns do not stand behind their banks, it is unquestionably true that the failure of a sovereign may result in the failure of the banks in that country. Thus a supervisor in a host country needs to assess both the risks inherent in the home country government and the extent to which contagion risk may damage that business. If the home country sovereign is financially weak, the host country authorities are likely to require subsidiarization to limit contagion. However, where the home country sovereign is financially strong, the host authority may prefer a branch structure.

3.49  It is also important to consider this issue from the perspective of a creditor of the business concerned. From a creditor’s perspective, in most cases their wish would be to deal with a branch. A creditor of a branch may pursue assets belonging to the bank concerned anywhere in the world they can be found, whereas a creditor of a subsidiary is limited to those assets which can be proved to be owned by that subsidiary. Possibly more importantly, where the local business is in difficulties, company law in many jurisdictions places severe limits on the extent to which a parent company can advance funds to a subsidiary which is insolvent and not likely to resume business. Thus if local business is subsidiarized, the parent company concerned may be forced by corporate benefit and other similar rules to refuse to provide additional capital to the subsidiary, leaving creditors of the subsidiary with no recourse to other parts of the group.

3.50  Thus far we have simplified this discussion by looking at the two-party situation. However, for SIFIs, the three-party situation is almost equally important. This situation arises where a bank incorporated in Country A deals through its branch in Country B with a customer in Country C. Since money is the easiest thing in the world to move across borders given modern technology, this is a pattern which occurs with increasing frequency. In general, arrangements of this kind are unproblematic when the bank is operating through a branch structure—the customer is a customer of the bank. It is generally immaterial both to the customer and to the bank whether the customer deals with an office of the bank in Country A or Country B—the customer’s legal rights and remedies are generally the same. With a subsidiaries structure, by contrast, the customer must elect on day one which legal entity to deal with,⁷ and the result of that election will determine the customer’s rights and remedies.

3.51  Finally, a commonly observed phenomenon in the resolution literature has been that non-lawyers tend to place too much importance on legal concepts. In law, and for regulatory reporting purposes, there are significant differences between a foreign branch and a foreign subsidiary. However, from a legal realist perspective, (p. 44) both are simply businesses headquartered outside the parent bank’s home country conducting business in a host jurisdiction. The host regulator for any branch has always had considerable flexibility as regards some aspects of the regulation of the branch—for example, host regulators control branch liquidity, the appointment of senior staff in the branch, compliance by the branch with local customer protection rules, and a number of other issues, in many cases in the same way in which they would control the activities of a separately incorporated local subsidiary. One way of putting this is that for a large banking branch business, incorporating that branch as a subsidiary would not have a dramatic impact on the day to day activities of the head of compliance of that business.

3.52  It has always been the case that supervisors of bank branches have a degree of control over the liquidity of the branches which they supervise—the general principle of bank regulation is that capital is for home state supervisors but liquidity is for host state supervisors. The logic of this may be open to question—in particular, this arrangement is designed for a world in which branches act exclusively within the boundaries of their host country, and becomes illogical if the branch does a significant amount of cross-border business itself—but it is nonetheless the general rule.

3.53  Requiring fixed liquidity pools on a branch-by-branch basis, however, would rapidly become onerous and reduce efficiency. More importantly, this is a security policy which comes with a heavy price, both in terms of cost to the bank and in terms of restricting the business of the branch. Consequently, regulators were historically prepared to permit arrangements under which the bank, under appropriate supervision, managed its liquidity position centrally.

3.54  It is now clear that the problem with these arrangements was that, as established, they gave little or no thought to the question of how the central pool would be allocated in the case of crisis. The spectre before every host regulator is the outcome of the Lehman collapse, where a ‘sweep’ arrangement which collected group liquidity into one particular entity had operated on the evening of the day prior to the failure, leaving the entities from which the cash had been swept hopelessly illiquid (and potentially unresolvable) the following business day. It is, however, an error to conclude that the problem was the sweep. The problem was not the sweep, but the fact that no thought had been given on any side to the question of how the funds swept should have been disposed in the event of a crisis. Accepting that this is a problem which needs to be solved, there are simpler, less costly, and more effective tools available to address it than subsidiarization.

3.55  The sensible approach to the problem would be an agreement among resolution authorities as to the disposition of the liquidity pool concerned. If we assume for this purpose that the liquidity pool is managed at the group level under the (p. 45) supervision of the home state supervisor, all that would be necessary to create such an arrangement would be for the members of the group’s crisis management group to agree on the relative levels of liquidity which would be allocated to the various subsidiaries and branches, and agree that in the event of the commencement of group resolution, those amounts would be made available on demand to the relevant group businesses. In effect, the host supervisor would determine a liquidity requirement for the hosted business and communicate that requirement to the home supervisor. The home supervisor would then assume the obligation to ensure that the liquidity pool of the group as a whole was at least equal to the sum of the local requirements.

3.56  It may be objected that local insolvency laws may make it difficult for authorities to make such transfers. However, this is to misunderstand the nature of liquidity management. While liquidity is highly relevant during business as usual, it may be even more relevant at the recovery and resolution stages of managing bank failure. The absence of sufficient liquidity can result in insolvency by forcing a bank to sell assets at fire-sale prices and thus lock-in capital-depleting losses.

3.57  The easiest way to approach this is to consider the role of capital in a subsidiary. Begin with a regulator who wants a branch in its jurisdiction to convert to a subsidiary. The most immediate impact is that it will harm the recourse of the creditors of the business, who now have claims only on the local assets owned by the subsidiary and not on the global assets owned by the bank itself. The trade-off is that the subsidiary will have an excess of assets over liabilities (since it is required to maintain capital), and that excess will in theory be available to the host state regulatory authority to discharge local creditors. However, it should be noted that this does not change the position of, or benefit, local creditors, as an equivalent surplus will necessarily be maintained at the global bank level. What the subsidiarization requirement does is to restrict the creditor’s claims to a narrower pool of assets under the control of the host state resolution authority, as opposed to a larger pool under the control of the home state authority.

3.58  It is therefore the case that when we speak of ‘capital’ in a subsidiary, what we mean is that the subsidiary has more assets than liabilities. It is perfectly possible to do an equivalent analysis for branches, and in other areas of analysis (notably tax law), the concept of ‘branch capital’ is unproblematic. However, a simple ‘assets less liabilities’ calculation tends to break down on the fact that although assets of a bank may be attributable to a particular branch, in general liabilities are not, and although it is possible to establish which assets can reasonably be said to be ‘assets of the branch’, in the absence of complex branch ringfencing provisions in funding arrangements (which would themselves have significant impact on the costs of those arrangements), it is not possible to say that a particular liability is a liability of one branch (p. 46) but not another, unless it is payable solely at a particular branch or branches outside a specified jurisdiction.

3.59  However, what is important in this context is to look at the problem which the host authority is trying to solve in the first place. The primary focus of the host authority is precisely to protect ‘local’ creditors—that is, local owners of local liabilities. Viewed from this perspective, it is quite easy to identify which creditors are ‘branch’ creditors—they are those creditors in the jurisdiction concerned who can reasonably look to the resolution and regulatory authorities in that jurisdiction to protect their interests. The paradigm case here would be local retail depositors, but local corporates who bank with the branch and local business who use the branch for payment services would both also fall within that class. The basis of this selection is the local regulator’s obligation to protect customers in its jurisdiction, and following this line of argument gives us a plausible approach to identifying local creditors—that is, local creditors for this purpose are those creditors with whom the bank would not be permitted to deal without the authorization which the branch or the subsidiary has been given by the local regulator.

3.60  This would give us a way of addressing local capital requirements, in that it suggests that what the local regulator should require is a surplus of local assets over local liabilities. If the bank has no local liabilities, then the local regulator should be unconcerned; if the bank has more local liabilities than local assets, the local regulator will consider itself exposed and take steps to limit that exposure. A requirement that local assets exceed the claims of local creditors would seem to be effective in this regard.

3.61  A further issue—and one which creates a great deal of difficulty—is that an intelligent local resolution authority might argue that it is no use having local assets if those local assets are capable of being seized by overseas creditors so as to leave local creditors unprotected. This would argue for local ringfencing, which is an almost complete block to resolution. However, this again is to elide two different problems. In a resolution all that is needed is a presumption that local resolution authorities may apply local assets in satisfaction of the claims of local creditors. If this is sufficient for the resolution (using a no-creditor-worse-off-than-in-liquidation yardstick), then the problem is solved. However, in a liquidation, the issue goes away, since equality of treatment of creditors will be the appropriate outcome. The key point here is that in resolution assets are not distributed—the structure of a resolution is to put the entity being resolved in a position that it can continue in business without a cash and asset outflow. Thus a resolution can be conducted on the basis that local assets can be used to support the local business without imperilling the order of distribution of assets on an insolvency.

3.62  The overall effect of these two principles would be to require banks operating in host jurisdictions to increase the amount of their assets in these jurisdictions, to restrict (p. 47) to some extent the use of capital and liquidity resources around the enterprise, to require a greater degree of focus on location of assets and liabilities, and to increase significantly the amount of internal analysis and regulatory consent required for developing new lines of business, opening new business lines, or entering new markets. All of these have the effect of increasing fragility within the institution, since any restriction on the institution’s ability to deploy its assets to meet a crisis will increase the chances of the crisis overwhelming the institution.