Amidst all the complexities of financial regulation, perhaps no policy parameter is more important than the minimum level of equity capital that banks must have in relation to the size of their overall exposures. In short, how leveraged can they be? The Global Crisis of 2008 caused so much havoc in large part because banks had grossly excessive leverage – of 40 or 50 times. With such a thin equity buffer, and with debt bad at absorbing losses, taxpayers had to bail out the banks.
Much stronger capital buffers are therefore fundamental to banking reform. But how much stronger? Perhaps surprisingly more than seven years on, this question has not been fully decided. But on 29 January the Bank of England published its framework for the systemic risk buffer (Bank of England 2016). This consultation paper deserves the attention of economists generally, not just specialists in bank regulation.
This column lacks space for a wider discussion of the economics of bank capital, which would go into the Modigliani-Miller theorem, debt overhang, ratchet effects, and so on (for a strong and challenging book on the subject see Admati and Hellwig 2013). But two general points can be made at the outset.
- First, systemic bank failures produce enormous negative externalities.
Making banks more resilient therefore has massive positive externalities – i.e. benefits to society not internalised by the banks themselves. Only when banks are very safe indeed does the wedge between private and social interests become small.
- Second, the main argument that banks deploy against higher equity requirements – that equity is a costly form of funding for them – gives a reason to favour higher equity requirements.
Equity is costly to the extent that banks are risky. It is in the public interest to contain risks from banks – especially those providing core services such as current accounts – which is best done by more capital, not less.
A good way to approach the current UK policy issue is Chart A of the Bank of England paper, reproduced as Figure 1 below, which illustrates the capital requirements proposed to be in place by the start of 2019. The notion of ‘capital’ here goes wider than shareholders’ common equity capital. In the jargon it is known as ‘Tier 1’ capital, which also includes limited more-or-less equity-like capital such as contingent capital instrument (‘co-cos’). But capital that is not pure equity can be at most 1.5% of the minimum required capital. Equity capital is not locked away in some vault. It is simply one of a number of elements of the bank funding mix, but a very important element for the public interest because only equity capital is assuredly loss-absorbing. The bulk of the funding will be deposits, bonds, and so on. Thus the figure depicts only a small fraction of bank funding – indeed much less than the 11% figure on the right.
Figure 1. 2019 Tier 1 capital requirements
That is because the percentages are relative to risk-weighted assets (RWAs). The idea is that capital requirements should reflect the fact that some assets (such as loans) are riskier than others. Risk-weighting failed hopelessly in the run-up to the Crisis but is being greatly improved. Moreover, a cap on overall leverage (i.e. with no risk-weighting) is being put in place internationally. This says that capital must exceed 3% of total assets. On this basis leverage can be no more than 33 times. Non-experts will think that this allows a potentially dangerous amount of leverage. Non-experts are right. Admati and Hellwig would require equity of at least 20% of total assets, so leverage of at most five times.
The blue slab in Figure 1 is the minimum capital that any bank must always maintain. The purple slab is a buffer to absorb losses while a stressed bank keeps going and repairs itself. The level of the blue and purple slabs comes from the international Basel III agreements. They add up to 8.5% of capital relative to RWAs, or 7% in terms of common equity capital. The brown layer is an add-on that the regulator might apply to some banks on an individual basis. The red has to do with macro-prudential regulation. Over the cycle a system-wide buffer can be expanded and contracted. For now it is zero.
The live policy issue concerns the yellow slab – the systemic importance buffers. There are two of these. First, some UK banks are considered to be of such global systemic importance as to be required to have a so-called G-SIB buffer based on their global exposures. Second and the subject of the recent Bank of England consultation paper, there is the systemic risk buffer that is to apply to ring-fenced banks that are domestically systemically important. The two buffers are not added together. Loosely speaking, whichever is the higher applies.
Ring-fencing and the ICB
Ring-fencing is a structural reform that was recommended by the UK’s Independent Commission on Banking (ICB 2011) [disclaimer: which I chaired] and is now in law and being implemented. It is a form of separation between retail banking and investment banking. Core retail banking activities (domestic retail deposit-taking, etc.) must be conducted in an independent ring-fenced bank with its own capital requirements, which may not carry out various investment banking activities, though they may take place in the rest of the banking group to which the RFB belongs. Ring-fencing aims to give retail banking a degree of insulation from global shocks, better ways of resolving banks that get into trouble, and it provides a way of having safer capital requirements for systemically important domestic banks while allowing international business to operate from the UK on the basis of international regulation.
The ICB recommended that any ring-fenced bank with RWAs greater than 3% of UK GDP should have a systemic risk buffer of 3% of RWAs in terms of equity on top of the Basel 7% (or on top of 8.5% of the wider definition of Tier 1 capital). Ring-fenced banks with RWAs between 1% and 3% of GDP would have to meet a lower supplemental requirement on a sliding scale. The ICB recommended that leverage constraints be tightened in parallel.
While 3% is a large increment in relative terms, in an ideal world I and others on the ICB would have gone higher. But in practice there were constraints arising from geographic arbitrage, functional arbitrage (i.e. inefficient migration of business to non-banks), and issues of transition in a weak macro-economy. A judgement had to be made, and ours was the 3% extra equity requirement, which would have applied to the bulk of British retail banking. This would be a substantial addition to the equity capital backing the system. Starting from such a low base, every percentage point matters a lot.
The Bank of England’s proposed policy
The Bank of England, in its recent policy proposal, has made a rather different judgement. A ring-fenced bank with assets less than £175 billion (about 10% of GDP) would have no systemic risk buffer requirement at all. (Northern Rock in 2007 is a case in point.) The 3% increment would apply only to ring-fenced banks with assets above £755 billion, 40% of GDP. There are no such banks. The proposal would add just 0.5% of RWAs to total equity capital in UK banking, perhaps 0.2% of total (i.e. unweighted) assets. Leverage constraints would, quite rightly, move in proportion to elevated capital requirements.
By contrast, the ICB’s recommendation on extra equity capital would apply the full 3% uplift to ring-fenced banks with assets above £160 billion or so.1 Thus the Bank of England’s policy is significantly softer.
One reason why the Bank’s proposal does not add a greater amount of capital to the system is its modesty relative to capital requirements applied internationally to some UK banks to protect global stability. That capital may be down-streamed to meet the domestic requirement.2 The ICB proposal by contrast would build substantial extra capital, relative to global requirements, to boost the resilience of the UK banking system.
The Bank’s approach to the systemic risk buffer accords with the policy position was set out in December in Bank of England (2015). On that analysis, the UK banking system “only has a little more capital to build, in aggregate, by 2019”. And the Bank’s “assessment of the appropriate level of capital is substantially lower than earlier estimates of the appropriate level of equity for the banking system, including those that were produced by the Basel Committee on Banking Supervision”. This conclusion is related to a paper by Bank staff (Brooke et al. 2015) which suggests that the minimum Tier 1 capital requirement for UK banks should be 10-14% in typical risk environments.
Even on its own terms it is not clear that the Brooke et al. paper justifies as mild an approach to equity capital requirements as the Bank of England proposes (as compared, say, to what the ICB recommended). For one thing, the sum of the blue, yellow and purple slabs in Figure 1 is below the stated 11% figure unless all banks have a 2.5% systemic risk buffer, which on the Bank of England proposals they do not. Significant bank-specific (brown slab) capital is needed to get up to 11%.
The staff paper would be easier to assess if it engaged with the prominent work of Admati and Hellwig (2013), which goes unmentioned. When two analyses of the same problem arrive at very different answers, it is helpful to know what variations of assumptions and/or method explain the difference in results. Admati (2016) does however engage with the analysis of Brooke et al. (2015), and gives reasons why she finds it “fundamentally flawed”.
The Bank of England gives four reasons for its downward revision of bank capital policy. They concern complementary and fully anticipated elements of the package of banking reforms, and therefore seem a curious basis for a downward revision of the core element of stronger capital buffers by 2019. Two of them – better supervision and use of the countercyclical capital buffer at some future point – were factored into the ICB’s analysis. The other two – more effective resolution arrangements (with bail-in debt) and structural separation in the form of ring-fencing – were ICB recommendations. Bail-in debt, a welcome development, is no substitute for effective loss-absorbency in the first place. Indeed a regulatory distinction is made between the “going-concern” loss-absorbency of equity capital and the “gone-concern” loss-absorbency of e.g. bail-in debt. Ring-fencing, far from being a reason to go easy on capital requirements, facilitates stronger capital buffers.
In sum, there are reasons to query the justification for the Bank of England’s apparent policy softening on bank equity requirements. Perhaps the Bank is perfectly correct, but its policy on a matter of this importance should be questioned. No-one can confidently know the answer, but that in itself is a reason for caution – i.e. not risking equity requirements that are too low. The downside risks are much larger than the costs of having an equity/debt mix in bank funding that is a bit higher than optimal.
Meanwhile the markets have something to say. In the early weeks of 2016 bank stocks and related instruments have dropped sharply. This is a timely market reminder that the resilience of the banking system cannot be taken for granted. Policy in this area really matters, and calls for vigorous economic debate.
Admati, A. (2016), “The missed opportunity and challenge of capital regulation”, National Institute Economic Review.
Admati, A. and M. Hellwig (2013), The Bankers’ New Clothes: What’s Wrong with Banking and What to Do about It, Princeton, NJ: Princeton University Press.
Bank of England (2015), Supplement to the December 2015 Financial Stability Report: The Framework of Capital Requirements for UK banks.
Bank of England (2016), The Financial Policy Committee’s Framework for the Systemic Risk Buffer.
Brooke, M., O. Bush, R. Edwards, J. Ellis, B. Francis, R. Harimohan, K. Neiss and C. Siegert (2015), “Measuring the macroeconomic costs and benefits of higher UK bank capital requirements”, Financial Stability Paper No 35, Bank of England.
Independent Commission on Banking (2011), Final Report: Recommendations.
 The £160 billion figure is derived as follows. A ring-fenced bank with RWAs = 3% of UK GDP has total (i.e. unweighted) assets = 8.5% of GDP if the average risk weight is calibrated in line with the leverage cap, i.e. is 3/8.5 or 35%. And 8.5% of £1.9 trillion GDP is £161.5 billion. A higher average risk weight would produce a lower sum.
 Down-streaming mustn’t go too far. If down-streamed capital is securely within the ring-fence, as it must be for ring-fencing to work, beyond a point down-streaming could erode capital guarding against risks to global stability from other operations of the banking group.