#1: The big Implications of sPOE and MPOE Resolution Structures
The development of resolution frameworks has revolutionized (or should revolutionize) many aspects of bank regulation. If the core purpose of regulation is to protect the financial system, then understanding the likelihood and impact of failure is central to the exercise. Resolution frameworks divide into two basic approaches to bank structure; this choice will have important implications for host regulation.
Multiple-Points-of-Entry Banks (MPOE): One approach to managing a group is to think of subsidiaries as independent “portfolio companies” with a common shareholder. MPOE groups define certain subsidiaries as resolution units and reserve the right to walk away from them in crisis, to protect the rest of the group. Each resolution unit is designed as a self-contained, individually resolvable entity, in line with local requirements. In a sense, these groups have accepted misallocation risk in exchange for the right to walk away from a local entity. This can be a practical solution for certain types of banks – especially those that are built as a federation of locally oriented retail banks.
Single-Point-of-Entry Banks (SPOE): The other major category is SPOE groups, which treat subsidiaries as part of a naturally interconnected enterprise – a single organization that acts as a “source of strength” for its entities around the world. For these groups, resolution happens at the top entity only, and resources are pushed downstream from parent to subsidiary as needed to support critical activities throughout the world. This is the dominant structure for big banks today, applying to perhaps 90% of the world’s 30 most systemically important banks.
The choice of MPOE structure simplifies the regulatory analysis for the host. The group has emphasized that the subsidiary is on its own, so these subsidiaries should naturally be regulated as stand-alone banks. They should be subject to capital and regulatory requirements in line with stand-alone local banks of similar scale and activity. In this way, hosts achieve “certainty” – at least to the same extent they achieve it for similarly situated domestic banks. In contrast, SPOE groups are built on the premise of group support and source of strength. That can provide different and important advantages to a host regulator; we discuss how to ensure this support in Box 2.
#2: Achieving “Certainty” for Hosts under SPOE
As we saw in Box 1, the distinction between MPOE and SPOE structures has major implications. Regulation for an MPOE subsidiary can be built on a simple stand-alone entity framework. However, SPOE groups are built on the premise of group support and source of strength. But how can hosts be sufficiently certain that support will be available, sufficient, and genuine? A few elements are critical:
Resolution Credibility: If the group does not have a credible, well-resourced resolution plan, then hosts are exposed to the same pitfalls they faced in 2008. Without such a plan, each country will have to sort out its own problems via local bankruptcy or local resolution. Even if the bank group is bailed out, national authorities are likely to fight over who should foot the bill, as in the acrimonious case of Fortis in the Benelux region. A strong resolution plan, well-funded with external TLAC, is essential.
Walkaway Risk: Even if there is a strong resolution plan at the top that can recapitalize all operating subsidiaries, hosts are exposed to the risk that the group chooses not to support a weak subsidiary. A value maximizing group could improve its economic position by discarding a subsidiary with very negative equity. Walkaway risk is unlikely for a going-concern group because of important franchise, market, and reputational pressures, and has been very rare in the modern era. However, host regulators worry that group support could falter under the legal conditions of bank resolution, which may require resolution authorities to consider walkaway options if the group fails.18 This risk can be addressed in two ways: via prepositioning of some capital to tilt the NCWOL calculations in favor of group support or by legal tools, such as guarantees or support agreements.
Transparency on Central Reserve: Our model shows that “friendly” hosts can be put into a risky position if they do not respond to capital trapping by others. They rely on the availability of the central reserve, which could be impaired by the actions of more aggressive hosts. Indeed, perhaps the worst position is to be a lone “no-fencer” in a ring-fencing world.19 Transparent and fair allocation is critical to support cooperation. This is similar to the “negative pledge” covenants that bond investors use to protect themselves: make sure that lenders rank equally and that the company doesn’t pledge assets to secure other loans ahead of the existing loan group.
Endgame Considerations: The stress on hosts will be highest when the group is nearing failure. At this point, further support from the top will be unreliable at best, so hosts are likely to want sufficient tangible prepositioned resources most at this point. These resources will be important to a host for two reasons: for assurance against walkaway in resolution; and for locally resourced protection against poor resolution execution by the home.
Credible host arrangements will need to address all four elements to achieve a sufficient degree of certainty. It will be impossible to achieve absolute certainty – we are not attempting the impossible task of reducing bank risk to zero. But the overall package should achieve at least the same level of safety as a local stand-alone bank of similar size under domestic capital rules.
Our model bank analysis shows that strictly local (i.e., ring-fenced) support is highly inefficient for SPOE groups. Hosts can achieve significantly better risk outcomes with lower amounts of local capital if the regime is built with strong, resilient group support distributed under transparent rules. As we consider the prudential implications of the new resolution architecture for SPOE, arrangements like our “straw man” proposal can produce significantly better outcomes for both groups and hosts.
#3: Major Capital Categories and Their Allocation
Capital Minimum: Basel standards require a minimum equity of 4.5% of risk-weighted assets. We accept this as a minimum for both the group and for each major subsidiary, and assume that any breach would trigger a failure event.
Going-Concern Buffers: Most common equity at large banking groups is now driven by the numerous buffer requirements (capital conservation buffer, G-SIB buffer, etc.) and other rules (e.g., CCAR stress tests). These increase the effective requirements for capital far beyond the official Basel minimum. The average G-SIB Tier 1 capital ratio is now over 14%, suggesting that regulatory and bank management buffers are close to 10%. This provides a resource that can be drawn down to absorb losses in the case of a future downturn and is the primary capital resource we modeled in our analysis, up to the capital minimum threshold.
Resolution Funding (GLAC27): Large amounts of subordinated debt have been issued to provide a prepositioned resource for bank resolution. These GLAC resources would be converted – or “bailed-in” – to create fresh equity and support critical operations and key subsidiaries. The outcome is similar to Chapter 11 reorganization, but with more preplanning (and much faster). In the U.S., GLAC debt resources have been scaled to roughly $1 trillion (on top of more equity). This is ample to restore group capital, even under tough assumptions and 2008-scale conditions. Several other key jurisdictions have achieved similar levels of resourcing.
These resource pools for buffers and GLAC are massive. But there has been little discussion on how best to deploy them across a group to address the twin challenges of resilience in going concern and credibility in resolution. We propose to utilize capital buffers – resources designed for drawdown in adverse conditions – as the natural tool for resilience and recovery. For SPOE firms, these buffers should be housed in a central reserve to avoid misallocation risk and avoid unnecessary failures. In contrast, capital minimums and GLAC are designed to ensure acceptable endgame results. They are naturally less mobile and therefore prime candidates for prepositioning.
In the 10 years since the financial crisis hit, extensive reforms have rebuilt the foundations of banking. The effort has been centered on the Financial Stability Board’s (FSB) four major reform pillars, including stronger capital and ending the concept of “too big to fail.” But could a little-known “fifth column” – the rise of widespread ring-fencing – threaten to undermine these achievements?
Ring-fencing is shorthand for local control around certain assets or activities, typically requiring dedicated pools of local capital and liquidity.1 The rise of widespread ring-fencing is perhaps a natural reaction to the recent crisis, but it can lead to unexpected and adverse outcomes. It harms competition, economic resilience, and growth.2 In particular, it can actually damage bank safety, confounding the original intent of the exercise.
In a recent working paper, we developed an analytic framework to assess the effects of ring-fencing for a simple model bank.3 The results showed that widespread ring-fencing can increase the risk of bank failure, in some cases dramatically. While precise quantification naturally depends on specific assumptions, the analysis suggests that the rise of ring-fencing is a major issue and could be at the center of the next crisis.
In this article, we summarize that analysis and outline a potential solution: a plan for transparent “mutual disarmament” to restore bank resilience while addressing legitimate host concerns. The development of post-crisis resolution architecture forces a new look at home-host relations given the stark differences for subsidiary support between the multiple-points-of-entry (MPOE) and single-point-of-entry (SPOE) frameworks (see Box 1). It also forces us to consider how best to satisfy the twin objectives of resilient recovery and credible resolution. However, the new resolution framework also gives us a large pool of subordinated total loss-absorbing capacity (TLAC4) – resources that could be harnessed to solve the problems raised by ring-fencing.
Our framework tries to balance the need for certainty for host jurisdictions (subsidiaries) with the need for flexibility for the home jurisdiction (parent). A rough outline of the framework includes:
Moderate capital prepositioning to protect hosts. The parent should have sufficient net investment in key subsidiaries to enforce cooperation and deter “walkaway risk.” Such “skin in the game” is also important to avoid “no creditor worse off than in liquidation” (NCWOL) challenges in the event of resolution.
Top-up rules to maintain host safety (perhaps via binding contracts to support needy entities, like the support agreements used in U.S. living wills.)
Prepositioning should mostly be in subordinated, gone-concern TLAC, plus a minimum equity requirement (but no local equity buffers).
The remaining group capital (equity buffers) should be kept in a central/mobile reserve, in order to ensure resilience. (Equity buffers make up perhaps two-thirds of the total equity layer today.)
Before the crisis, regulators were fairly comfortable hosting international banks, at least from well-run jurisdictions. Competent home-country regulators would supervise the group and address any problems that emerged. Those assumptions held up well for many years and supported international growth, diversification, and competition. But this implicit home-host trust was shattered by the crash-landing failures of major banks such as Lehman Brothers and Fortis in 2008, which imposed surprise losses on host countries. Banks were now said to be “international in life, but national in death.”5
In response, many jurisdictions moved to protect national interests and adopt stringent controls around local subsidiaries, including capital restrictions, liquidity restrictions, and operational support requirements.
They decided to ring-fence early, far in advance of potential stress conditions.6 Some moves were public, such as the U.S. intermediate holding company rule and the European Union’s response (the proposed intermediate parent undertaking), but others happened under the radar. Indeed, ex ante ring-fencing has become widespread even within the EU banking union,7 and Brexit will likely create further barriers. These proposals have emerged as national initiatives, without a substantial international debate or a global framework. As these policies proliferate, they threaten to undermine bank resilience (see Figure 1 for a schematic view of this shift).
Our analysis from here will focus specifically on geographic ring-fencing and the effect of capital Balkanization.8 The purpose of locking in local resources is to create extra safety for your part of the bank. But does ring-fencing actually make banks safer?
A sole ring-fencer can gain an advantage. There is a large risk reduction for a first mover if other jurisdictions do not match that decision. The first ring-fencer benefits from both a) local capital and b) the ability to tap a large central reserve (see case #2a, Figure 2). However, trapping capital for one entity reduces resources for others, and their risks begin to increase. In the real world, other jurisdictions will respond – typically with countervailing ring-fencing policies to improve their own position. This shrinks the pooled central reserve further, harming the resilience of the group. Eventually, the central reserve can dry up and all jurisdictions become worse off.9 The risk of failure can increase by 5x or even 15x compared with an “integrated bank,” where internal resources are mobile.10
This threat to solvency is caused by “misallocation risk” – the risk that a bank has enough capital overall but cannot get those resources to the right subsidiary in time to avoid a local failure.11 (See Figure 3.) The risk increase is only 4.8x (case 3 in Figure 2) if the subsidiaries can fail without harming their affiliates but grows to 15.1x (case 4) if the failure of one subsidiary ultimately leads to a cascade of failure throughout the group (interaffiliate contagion). Misallocation with contagion is a nightmare scenario, but it’s unfortunately plausible unless we take action to understand and address it.
A Home-Host Prisoner’s Dilemma
If retaliation is pervasive, the outcome for everybody – even the first ring-fencing country – is worse off than when this process started. Local entities become more likely to fail, potentially dramatically so. This is a “prisoner’s dilemma” – an economic paradox in which each participant seeks to achieve a local benefit but ends up worse off when others also pursue their own incentives.12 If local incentives are sufficiently strong, a “tragedy of the commons” can seem inevitable.
The rules of the actual prisoner’s dilemma also stipulate that the participants cannot communicate to achieve a better outcome. However, bank regulators can cooperate and build ex ante structures and communication mechanisms to share in the global gains from a more enlightened framework. What should they do?
In a recent speech, Federal Reserve Governor Randal Quarles took a broad view of the ring-fencing discussion, and discussed liquidity and capital ring-fencing, late versus early ring-fencing, and the importance of transparency in the pursuit of a balanced and stable outcome.13 In particular, he highlighted the need to understand the different objectives of home and host regulators. To plot our escape from the prisoner’s dilemma, let’s first examine these perspectives:
- The Home-Country View: A regulator that supervises a bank with headquarters in its jurisdiction (i.e., a home regulator) will focus primarily on the health of the consolidated group. First of all, the group is a direct responsibility. In addition, the other problems of local collapse or home-host conflict simply fall away if the consolidated group is healthy. The home regulator will naturally focus on group solvency and work to avoid misallocation risk and resource trapping.14 In Quarles’ language, group flexibility will be the prime objective for the home.
- The Host-Country View: Host regulators are charged with protecting local financial stability and national interests. A strategy of naïve trust is not credible in the wake of 2008. As Quarles put it: hosts need certainty. They want enough control and resources to ensure that activities undertaken in their jurisdiction meet local standards of safety and that they will be treated fairly. The resulting policy implications are pretty simple under MPOE but become more interesting under a SPOE, “source of strength” regime (see Box 2 for a deeper dive into host concerns).
Ring-fencing lies at the heart of home-host relationships. Subsidiarization – with full local capital and liquidity – can provide hosts with a high degree of certainty.
But the analysis above shows that such widespread local capital trapping will make banks riskier. It backfires as other hosts retaliate and leads to a deeply counterproductive result.
How can we address these legitimate host concerns without the drawbacks of ring-fencing? Can we, in the words of Quarles, find a good “balance of flexibility for the parent bank and certainty for local stakeholders” (emphasis added)?
Branches: One solution is branching, an ancient strategy that ties local and foreign interests into a single legal entity. This avoids the problem of local-only failure altogether.15 A fully branched structure can solve the host problems mentioned above and support a more efficient and resilient business model. One positive element of both the recent U.S. and EU proposals has been the decision to exclude branches from their ring-fencing initiatives.
While branching is a simple and time-tested solution, there are limitations. For example, many countries require a subsidiary structure for certain activities (e.g., insured retail deposits that are backed by the national government).16 Branching is an effective and constructive solution where it is allowed, but it is unlikely to be the entire solution in the near term. We therefore have to grapple with the issue of subsidiaries.
Subsidiaries: We used our model to explore various capital structures for banking groups. For a given amount of total capital, our goal was to minimize risk at both the group and the subsidiary level. We tested various approaches, including ones similar to the current FSB approach.17 These structures generally scored poorly on flexibility and resilience. We did not achieve satisfying results until we expanded our approach to consider going and gone concern capital resources separately and took a broader view of the different needs of recovery and resolution (see Box 3 for a high-level discussion of these components).
We believe that the best balance of resilience and certainty can be achieved through a mix of funded and unfunded capital commitments, together with a transparent approach to the timing and allocation of surplus resources. Some elements of this proposal are consistent with current regulatory thinking, but others depart significantly from the status quo. Our proposal rests on a mix of the following:
- Moderate Prepositioning using Gone-Concern Loss Absorbing Capital (GLAC): A modest amount of ring-fencing provides the first element. The purpose, however, is not to replicate a miniature stand-alone bank (per traditional thinking under the Basel Concordat) but rather to deter walkaway risk and create a defensible endgame capital solution for a host.
Composition: Capital and resolution arrangements are most critical as the group comes under serious stress. Our approach is to focus on what’s needed near the point of failure and work back from the endgame. We want a consistent strategy that avoids restructuring the group under stress, so we propose to adopt a sufficient endgame capital position ex ante. In the endgame, hosts should not care if the local resources are equity or GLAC as long as the total is big enough; both are subordinated to support critical local functions, and GLAC can be turned into equity under local control if necessary. We choose GLAC as the primary prepositioning resource because it is designed specifically to support recapitalization in resolution. Hosts will demand such protection near the point of failure, which means GLAC is inherently less mobile and should be the natural prepositioning tool. Importantly, that choice frees up going-concern equity for the mobile “group reserve,” which strengthens recovery capacity and resilience.
Sizing: Recent FSB rules proposed that major subsidiaries carry internal TLAC equal to 75% to 90% of the external TLAC requirements for a stand-alone bank, but we are unaware of any quantitative work behind this proposed range. In our analysis of host needs (see Box 2), we need sufficient resources to address both “walkaway risk” and “endgame considerations.” Because we also propose a top-up rule (see below), walkaway risk is short-term in nature (to the next top-up date), and can be deterred by a relatively modest amount of capital.20 Endgame considerations are the dominant constraint. In a resolution (or near-resolution) situation, hosts will likely want to hold at least a minimum “normal” going-concern capital structure when they exit resolution, to support critical functions after GLAC has been bailed in.21 This could be set at 8% of RWA (per standard Basel calculations) or perhaps slightly higher to provide a margin of safety.
We propose host capital prepositioning at 10.5%, partly for conservatism and also to utilize some Basel benchmarks. The 10.5% is comprised of 4.5% in equity (the Basel minimum) together with 6% in GLAC (modeled on the FSB benchmark of 33% of 18% TLAC). This gives us a substantial quantum of prepositioned subordinated resources to protect hosts, and also shifts the NCWOL calculation to force parent support (i.e., home cooperation) in resolution. Our bank solvency analysis already considered those capital components to be naturally immobile and unusable for recovery, so they had no effect on failure risk in our model.22 This frees up all equity buffer capital for a central reserve to be allocated to relieve stress wherever it arises. In this way, equity buffers can fulfill their core resilience function with maximum efficiency.
- Top-Up requirements: What happens if the value of prepositioned resources drops as a crisis intensifies? We propose a capital maintenance rule whereby the group tops up the subsidiary for significant losses, to restore the economic position of the subsidiary and maintain a positive net investment position for the parent. This also provides an early and frequent check of the parent’s commitment to central SPOE-style support.23 It tests this commitment under real-world conditions, without having to wait until conditions become dire.
The large U.S. banks have adopted a contractual approach to subsidiary support as part of their Title I resolution planning requirements. These “secured support agreements” (SSAs)24 are designed to ensure that the group supports key operating subsidiaries with sufficient capital and liquidity to remain viable throughout resolution and beyond. SSAs must be legally robust under the conditions of resolution. The SSA approach has a number of useful elements already, and perhaps could be extended to clarify the overall plan for intragroup support and underpin international cooperation.
A Straw-Man Proposal: In summary, we look to integrate these prepositioning and top-up mechanisms as follows:
- Minimum going-concern equity of 4.5% plus gone-concern GLAC of 6% of risk-weighted assets in prepositioned resources; plus
- Regular top-ups in the case of any deficit. These top-ups are drawn from buffer CET1 capital, which would be kept at the home, in a central reserve.
This package provides a total of 10.5% of RWAs in prepositioned, subordinated resources (plus replenishment rights). This provides a robust response to walkaway risk.25 If converted in resolution, it provides a strong in extremis capital cushion, one that is consistent with a solid capital ratio today. However, the decision to not preposition any going-concern equity beyond the required minimum is also important. The bank therefore retains flexibility to use these resources wherever the need arises, which eliminates the addressable misallocation risk.26 This structure allows us to achieve a low failure risk, identical to the “integrated bank” because we use the other components of TLAC to satisfy host-specific safety concerns. We believe it provides a strong answer to Quarles’ dilemma of how best to maximize flexibility for the home and certainty for the host.
We believe our straw-man proposal provides a credible path out of the prisoner’s dilemma of endemic ring-fencing. It provides a better, more transparent outcome for both homes and hosts, using both ex ante structure and transparency to control the pressures of ring-fencing.
The current regulatory approach has mostly transposed rules for external capital without considering the broader implications of how to handle equity buffers and GLAC allocation.28
By utilizing the resources of the new bank capital framework and resolution, we can develop a much more resilient structure, one that takes advantage of the “high road” suggested last year by Bank of England Governor Mark Carney:
A decade of hard-fought financial reform creates enormous opportunities. It’s all too easy to give into protectionism, but the road less taken is often the most rewarding. All the conditions are in place for following the “high road” of mutual recognition and cooperation both with Europe and across the G20.29
Both home and host regulators benefit from a high-road approach, built around a well-designed, verifiable program of “mutual disarmament.” Our straw-man approach looks to improve group resilience by replacing hard ring-fencing with a broader framework, while also addressing the legitimate concerns of host countries. Similar issues arise in the liquidity arena, which some believe are even more serious than capital ring fencing; we believe they should likewise be addressed with a similar mix of moderate prepositioning and contractual top-up.
This would provide a robust solution for cross-border global banks, even under 2008 levels of stress. It also provides critical transparency on how the process should work, from a starting point of full capital down through to the endgame position. This transparency is important to set out ex ante, to meet Quarles’ maxim: “Trust everybody – but brand your cattle.”
It’s crucial to update the rules for cross-border financial engagement to account for the new architecture of resolution for the pressures of today. It will become even more important as the world economy diversifies away from the North Atlantic. In the next 20 years, the seven biggest emerging countries – the “E7” – will grow to double the GDP size of the major industrialized nations that comprise the G7 today.30 This is a dramatic change from the old world, where G7 economies held a consistently dominant economic and financial position. While our straw man may not be the final or perfect answer, we believe it offers substantial improvements over the status quo. We hope it helps start a debate on how to build a durable “high road,” and re-establish cross-border trust on a firm base of enlightened, clear, and well-funded self-interest.
The full methodology paper can be seen here: https://papers.ssrn.com/sol3/papers.cfm?abstract_id=3085649
The views expressed are the author’s own. n
1 Ring-fencing typically involves both a structural requirement and resourcing rules that trap capital or liquidity (often via separate rules). Ring-fencing can also emerge from local preference requirements, such as U.S. rules that require assets booked in local branches to first satisfy local liabilities before they can be used elsewhere. Ring-fencing can also occur on a product basis, such as U.K. ring-fencing for retail activities.
2 See Mario Draghi, European Central Bank president, for an extensive discussion on how ring-fencing has impaired monetary transmission across the EU and reduced the ability of EU banks to cushion economic shocks – in contrast to the U.S., which has benefited from “banking union” (i.e. widespread interstate banking) since the 1980s and 1990s). https://www.ecb.europa.eu/press/key/date/2018/html/ecb.sp180511.en.html
3 See D. Wilson Ervin Working Paper: “The Risky Business of Ring-Fencing.” The approach uses a Monte Carlo process to produce numerous economic scenarios for a model banking group with four equally sized subsidiaries. Failure is deemed to occur when capital buffers are exhausted and the subsidiary has reached a minimum level of capital. The author would like to thank Bogdan Ianev for extensive quantitative support and insight on this project. https://papers.ssrn.com/sol3/papers.cfm?abstract_id=3085649
4 TLAC is a liability side resource (like equity). Some people think of allocating this resource to support loss-absorbing capacity across key subsidiaries. Others prefer to think of the benefit of TLAC in terms of the corresponding surplus assets it adds to the parent entity of the group, which can be invested in its subsidiaries. Both frameworks are useful to explain certain elements of the system; they deliver similar results because double-entry accounting matches assets and liabilities.
See also Davis Polk FinReg Blog (posted August 15, 2018) for an interactive spreadsheet tool that highlights these relationships. https://www.davispolk.com/files/internal_ringfencing_interactive_illustration_-_locked.xlsx
5 Variously attributed to Mervyn King, Charles Goodhart, and Thomas Huertas.
6 Some people highlight the risk of late ring-fencing (under stress conditions), and argue that ex ante ring-fencing is better. We agree that it’s important to avoid surprises, and that a clear-headed analysis of endgame pressures is essential. However, the primary effect of early ring-fencing is simply to bring the problem forward and leads to the counterproductive outcomes we highlight in Figure 2. In this article, all references to ring-fencing refer to the early or ex ante type, unless otherwise specified.
7 See comments by Andrea Enria (chair of European Banking Authority) in EUROFI magazine (April 2018), and Danièle Nouy (chair of the SSM) at the EU Parliament Economic and Monetary Affairs Committee (testimony of June 19, 2018).
8 Capital ring-fencing is an active topic in the current FSB debates because of its implication for internal TLAC allocation. However, some would argue that liquidity ring-fencing is actually an even more pressing concern.
9 The exhaustion of the central reserve can be accelerated by local gold plating – several countries have adopted local internal TLAC policies that are effectively higher than similarly situated local banks or international standards. Another pressure point is the “sum-of-the-parts” issue: Most subsidiaries also have some intragroup exposures, which means the sum of the assets (or RWA) across all subsidiaries will be larger than the net consolidated balance sheet. If internal capital or TLAC is based off the local balance sheets, then capital trapping will have an even more powerful effect when compared with consolidated figures.
10 See our Working Paper, op. cit., for methodology details. Our paper is a first attempt to assess the impact of ring-fencing on bank safety and propose a policy response; refinements to either would be welcome. Many assumptions are still crude (e.g., no fat tails, a highly simplified bank structure, etc.). However, the policy imperative does not depend on exact math. If refined models reduced the impact from 5x to 3x, the outcome would still deserve a major policy response.
11 We assume an SPOE strategy, which applies to the vast majority of G-SIBs (see Box 1 for details).
12 See Davis Polk FinReg blog: “FSB Finalizes Guiding Principles on Internal TLAC” posted 10 July, 2017, for an early discussion of this issue in the context of internal TLAC. https://www.finregreform.com/single-post/2017/07/10/fsb-finalizes-guiding-principles-on-internal-tlac/
13 See Randal Quarles, “Trust Everyone – But Brand Your Cattle” (https://www.federalreserve.gov/newsevents/speech/quarles20180516a.htm). Implicit in his speech (and our analysis) is that the main goal of a national regulator is to reduce the risk and cost failure on his or her watch (i.e., for the specific entities they are charged with supervising). For a more mercantilist assumption (minimization of national outflows), see Bolton and Oehmke, “Bank Resolution and the Structure of Global Banks,” forthcoming at the Review of Financial Studies.
14 Even home countries face group vs. local conflicts. Several jurisdictions impose forms of ring-fencing for local entities, such as U.S. national depositor preference, or U.K. retail ring-fencing. These rules increase group risk somewhat in pursuit of other policy objectives. To simplify our task, however, we assume that home regulators focus only on the consolidated entity.
15 We base our discussion on an idealized branch that participates fully and equally in the enterprise of the bank. In practice, many jurisdictions deviate from the ideal, in some cases materially. Jurisdictions that impose subsidiary-type rules on branches (e.g., dotation capital) are likely to achieve little except additional group risk.
16 See Bank of England, SS10/14, September 2014, for a policy discussion of branches vs. subsidiaries. They support subsidiarization for critical domestic sectors (e.g., retail deposits – especially where insured by a national government agency – or large domestic lenders). Otherwise, branches are seen as superior, as long as the bank is from a well-regulated jurisdiction and carries strong capital and has a credible resolution strategy.
17 The current FSB framework generally requires subsidiaries to carry going-concern capital amounts at full Basel levels (100%) and internal TLAC requirements at 75% to 90% of G-SIB stand-alone external TLAC requirements.
18 Most resolution regimes incorporate a “no creditor worse off than liquidation” (NCWOL) test. This requires resolution authorities to provide creditors with at least the liquidation value in resolution, and provides an important creditor protection.
19 If aggressive jurisdictions trap most of the group’s capital, then it can become difficult to re-allocate resources for others. This puts the “friendly host” into the unenviable position of a weak-capital entity inside a weakened group.
20 Our model assumes valuations are done quarterly and estimates walkaway risk using simple option probabilities (i.e., it ignores other franchise elements that could also induce support). It uses the same underlying assumptions as our core model, using an equity-only framework to avoid negative solvency; the results suggest risk is minimized at around 4% to 6% of RWA.
21 This is primarily to support going concern operations at the subsidiary in a whole-bank SPOE resolution, but this also provides emergency resources in the unforeseen event that the host is forced to resolve the subsidiary on its own.
22 Because GLAC can only be used in resolution, and minimum equity was the trigger for failure, we did not allow them to be used up or reallocated as a firm suffered losses. They were effectively excluded from our drawdown process, so using them to provide certainty for the host was effectively “free” and did not have any adverse effect on modeled failure rates.
23 Hosts can enforce compliance via legal tools if the top-up is contractual, or via other regulatory sanctions, such as: dividend or activity restrictions, public shaming, or (in extremis) local resolution.
24 See (e.g.) JPMorgan, 2017 Public Resolution Plan, pp. 35-44.
25 This provides roughly 50% of the FSB external TLAC standard; if the official FSB calibration range for internal TLAC remains at 75% to 90%; the top-up obligation could perhaps be counted to satisfy the remaining increment above prepositioning.
26 In Basel terminology: minimum CET1 equity and GLAC are prepositioned at the subsidiary; all of the remaining elements (mostly buffers that are designed for drawdown in crisis) are held centrally at the group level.
27 GLAC (gone-concern loss absorption capacity) is subordinated capital (typically long-term debt) that can be converted to equity in resolution to recapitalize the bank and operating subsidiaries. GLAC is a subset of TLAC (total loss-absorbing capacity), which consists of both going-concern capital (e.g., equity) and GLAC. See FSB: http://www.fsb.org/wp-content/uploads/r_141015.pdf
28 For example, the U.S. IHC imposes capital rules that are broadly similar to U.S. BHC requirements, such as RWA, leverage, and TLAC ratios and CCAR stress tests. Net IHC constraints typically lie in a range that is similar to (or higher) than BHC capital. Exceptions have begun to emerge in areas that are most obviously linked to resolution considerations, like internal TLAC.
29 Mark Carney, Governor, “The High Road to a Responsible, Open Financial System,” Bank of England, April 7, 2017. https://www.bankofengland.co.uk/speech/2017/the-high-road-to-a-responsible-open-financial-system
30 PricewaterhouseCoopers, “The World in 2050,” February 2017.
Wilson Ervin is a Vice Chairman at Credit Suisse in the group executive office. He works on a variety of strategic projects, especially policy reforms related to bank capital and ending “too-big-to-fail.” Ervin also chairs the Credit Suisse Americas Foundation and the Impact Investment Advisory Council. Prior to his current role, Ervin was the Chief Risk Officer of Credit Suisse, a member of the Executive Board, and chair of the Capital Allocation and Risk Management Committee. From 1990 to 1998, he worked at Credit Suisse Financial Products, where he headed new product development. Before 1990, he held various roles in capital markets (both fixed income and equity), including Australia investment banking, and the Mergers & Acquisitions group at Credit Suisse. Ervin received his A.B., summa cum laude, in economics from Princeton University.