First let me express my thanks, and some surprise, to the organizers for the invitation to make some pre-dinner remarks on the progress of banking reform since the crisis.1 Surprise because I doubt I’m the appetizer that some of you might have hoped for.
My title – banking reform nine years on – might have been snappier, in our decimal world, as ten years on. After all, the interbank markets started to seize up in August 2007. But of the mid-September anniversaries, Northern Rock in 2007 seemed a bit too British for such an international gathering as this. So Lehman Brothers in 2008, and the global mayhem that followed, is my reference point.
The collapse of Northern Rock, however, turned out to be a harbinger about what would happen on a global – or at least north Atlantic – scale a year later. But few picked up on it, or even on the collapse of Bear Stearns the following March. Thus came the near-collapse of market capitalism in the autumn nine years ago. Economics – and politics – has not been the same since.
The leverage question: The great divide
Twenty minutes allows discussion of only one point, so I shall focus on the leverage question pure and simple – what minimum fraction of bank funding should be common equity capital?
Risk weights are for another day. There won’t be time to talk about liquidity, and anyway the serious liquidity problems tend to be manifestations of underlying solvency problems. Neither will I talk about the structural reform, though that is what the UK’s Independent Commission on Banking is best known for, except to make two observations.
The first is that ring-fencing is being implemented in the UK, and on time. Credit to the Prudential Regulation Authority for its firmness of purpose on that front. The second is that, besides the Volcker rule, there is remarkably little serious structural reform anywhere else. The admirable Liikanen report of 2012 for the EU (European Commission 2012), and talk of a Glass-Steagall law for the 21st century in the US, seem to be going nowhere. The EU version of the Volcker rule is so narrowly defined as to make it almost a sideshow, a distraction from the deeper reform proposed by the Liikanen group. In short, we have structural reform in the UK and business as usual everywhere else, despite what happened nine years ago.
The leverage question is one of the most fundamental issues for a market economy. On its answer there is a great divide.
The general (albeit not universal) opinion expressed by those in the financial sector – regulators, not just banks – is that reform since 2008 has got us to about the right place. Banks generally can have leverage of 30 or so, and the largest institutions 25 or so. (The US limits leverage more than the global norm, but there are accounting differences.) The authorities now, we are told, “are focused on not increasing overall capital requirements across the banking sector. In short, there will be no Basel IV” (Carney 2016, emphasis added).
But the general (ditto) opinion among economists outside the financial sector is that banks should be required to have at least twice as much equity capital relative to their exposures as the prevailing regulatory settlement. To say twice is indeed to understate things. A truly formidable group of economists with great expertise in finance, Nobel laureates included, wrote to the Financial Times in 2010 to criticise Basel III (Admati et al 2010). For them, at least 15% of bank assets should be funded by equity – “the social benefits would be substantial … and the social costs would be minimal, if any”. On that view, Basel III allows four times too much leverage. For Mervyn King a 10% equity base – three times Basel III – would be “a good start”.
So one group or the other, if not both, would appear to be wrong by a large margin, on a policy question of deep importance. For the sake of the public interest one might hope that the official view is right. Alas, I struggle to see how it can be. Let me explain why.
Ten times more?
On a superficial view, however, we can relax. In his 3 July letter to the leaders of the G20, the chair of the Financial Stability Board states with great prominence that “the largest banks are required to have as much as ten times more of the highest quality capital than before the crisis” (emphasis subtracted). This has become something of a mantra of Governor Carney, used even on the day after the Brexit vote, and later echoed in Parliament by then Prime Minister Cameron.2
I am perplexed that so curious an observation is thought worthy of such repeated emphasis. Its import is that Basel II was hopelessly lax, and allowed leverage of 250 times. Ten times better than hopelessly lax is not a useful measure.
And when one looks at the actual leverage ratios of the major banks, the improvement relative to the period before the crisis is positive but modest. Bank of England figures show that the 2016 leverage ratios of the major UK banks were better by a factor of about 1.25 than in the period five years before the crisis (Bank of England 2016, Chart B2). Even relative to the position in the thick of the crisis, the improvement is by a factor of two or so, nothing remotely like ten.
It is fair to note that the governor’s “ten times” calculation is for CET1 – “capital of the highest quality” – whereas the leverage ratios just mentioned are for Tier 1 including elements such as CoCos, the merits of which as going-concern capital are very questionable given their low triggers and the dependence of those triggers on accounting measures that lag events.
His emphasis here on CET1 capital is welcome, but in those terms the leverage allowed under Basel III is greater still than appears at first sight. Thus a 3% minimum leverage ratio in terms of Tier 1 capital shrinks to 2.25% in terms of CET1 if, as is the case, non-CET1 can account for up to 25% of T1. Put reciprocally, 33 times leverage becomes 44 times CET1. And this is meant to be the resilient post-crisis world.
Moreover, these ratios are calculated on the basis of the book value of capital, albeit with some adjustments. Despite the terms in which people who know better sometimes talk about it, equity capital is not an asset that is easily counted. It is not an asset at all. It is a liability to the issuing bank – the difference between the estimated value of its loan assets and other exposures on the one hand, and its contractual obligations to depositors and bondholders on the other.
In short, it is a residual, the difference between two typically big numbers, of which the asset side is hard to measure given the nature of banking, and dependent on accounting rules. Even today, credit losses that are expected but not yet incurred count as ‘capital’ for European banks. When IFRS9 changes that, impairment losses will increase and ‘capital’ will decrease, by a moderate amount in steady state, but potentially sharply in a downturn. Accounting ‘capital’ also includes anticipated mark-to-market (or model) gains on the trading and available-for-sale books. But when markets are stressed, there is a significant risk that this ‘capital’ too will not be fully available to absorb losses.
Bank asset valuation is part of what the stock market does every trading day. A bank’s market capitalisation reflects a view of the value of current exposures, less obligations to depositors and bondholders, plus the franchise value of future profits in excess of the cost of capital. This last element is positive, unless bank managers are expected to make value-reducing decisions (in which case more capital is definitely a good idea). The market value of equity value also reflects the value of the put option arising from limited liability – i.e. that beyond a certain point losses are the bondholders’ or government’s problem.
All in all, one would normally expect a market-to-book ratio well north of one. Only with perfect competition and zero prospect of equity being insufficient to cover losses would it be close to one if book values of current exposures were correct.
So, when a bank’s price-to-book ratio is persistently below one, as is the case for some well-known large banks in Europe, there is a real question about whether book values, and hence the basis for regulation, is right. Of course, the market might well have things wrong, and markets had a bad record pre-crisis. But regulatory values did even worse, being very slow to recognise losses, much slower than markets.
A good case in point is HBOS. At the end of 2008, HBOS had loans and advances valued at £522 billion.3 It had recorded a profitable 2007 but took a charge of £1.3 billion for impairment losses for the first half of 2008, a quarter of 1% of asset value. Yet by the end of 2011 its impairment losses had accumulated to £52.6 billion, largely irrecoverable – more than 10% of asset value. HBOS had been meeting its regulatory capital requirements with room to spare, though during 2007 its Tier capital ratio slipped below the average for other banks, which, believe it or not, was then increasing. Bank share prices, on the other hand, declined steeply through the second half of 2007, and of course a lot further in 2008.
The general point is not to rely too heavily on regulatory values being correct – allow a decent margin of error – especially when the market-to-book ratios are below one. In those circumstances it would be a good idea to run and to publish stress tests on the basis of market, as well as accounting, measures of capital. The Bank of England, however, prefers not to adopt this suggestion, and argues that low market-to-book values can be explained by expected future cash flows being weak for reasons other than poor asset quality.
Well maybe, but that is not altogether comforting. While a bank with poor cash flows may be able to sell assets if needs be in a normal environment, doing so in conditions of systemic stress is another matter. As we learned nine years ago, there is then the fire-sale problem that assets, no matter how good they had been thought to be, can only be sold at distressed prices. In such conditions, it is not at all fanciful to imagine banks’ assets declining in value enough to wipe out 25-fold leveraged banks.
How to resolve the great divide? One approach is to build a model and do cost-benefit analysis. What, though, are the costs of a greater fraction of equity capital being required? Remember that for the authors of the Financial Times letter of 2010, the costs to society are minimal if any, while the benefits are clear. Increasing the equity/debt ratio is undoubtedly costly to banks, because of tax/subsidy reasons and the positive externality to bondholders and taxpayers of more equity. But these are not costs to society; if anything they are benefits. Thus, to objections that the Modigliani-Miller theorem does not apply to banks, the short answer is: no indeed, and that’s all the more reason for higher equity requirements.
They might slightly increase banks’ intermediation spreads, but that is not a social cost if it corresponds to removal of an intermediation subsidy, and the reduction in crisis probability is positive for other forms of lending as well as for society generally. Enormously greater capital requirements could ultimately squeeze deposits, and the liquidity services they provide. But we are nowhere near that point. Bank equity is positive, not negative, for the liquidity of deposits. And as Hyun Shin (2016) has emphasised, “having soundly capitalised banks turns out to be vital for the transmission of monetary policy”.
There would be costs to society from an unduly hasty increase in capital requirements, which is why it made sense for Basel III to have an extended implementation period. But capital can be built up steadily over time. Retaining earnings rather than paying out dividends is an obvious way. For a healthy bank this does not strike me as undue hardship for investors. In effect, their dividends are simply being reinvested, like accumulation units in a tracker fund. An investor needing cash can sell a few units. Shareholders in unhealthy banks would much prefer the dividend pay-out, but such banks need equity to be retained, not paid out.
But let us take the Bank of England’s own model, which does have substantial costs of higher capital requirements, as a reference point (Brooke et al. 2015). The Bank has appealed to that analysis to justify its explicit lowering of its estimate of optimal bank capital. I believe the results of the analysis rely on assumptions about new resolution regimes and countercyclical policy that are, respectively, imprudent and untenable.
The governor says that “credible resolution strategies will reduce both the likelihood and probable impact of systemic bank failures, leaving the system less reliant ongoing concern capital to do the heavy lifting”.4 Gone-concern capital doing heavy lifting is a chilling thought. And if bail-in debt is so good at taking loss, why not do the equivalent with common equity instead? That it would be worse for banks from a tax angle is not a public policy reason to favour debt over equity.
On countercyclical policy, the Bank of England’s results “are explicitly focussed on the costs and benefits of higher capital on normal risk conditions (broadly equivalent to the ‘mid-point’ of the credit cycle)”, not its full range. Given that capital buffers are a safeguard for abnormal conditions, this makes no sense. Flood defences are not constructed for normal weather conditions. As the Bank of England paper says, crisis probability is “likely to be considerably greater at the peak of the credit cycle”, and higher capital ratios would then be desirable. This, in masterly understatement, is said to justify the use of the countercyclical capital tool. But there are no grounds to suppose that countercyclical policy can anticipate all elevations of risk. The premise of normal risk conditions is simply the wrong basis for the analysis.
Correcting for this, and taking a less rosy view of the efficacy of resolution arrangements, the model delivers a very different result – optimal capital levels almost twice the Bank of England estimate. This leaves Basel III stranded. It also makes it all the more puzzling that the Bank of England made such limited use of the systemic risk buffer for UK retail banks when setting that important policy parameter last year.5
I hope these remarks will not be seen as critical of the progress towards financial stability that bank capital regulation has made over the past nine years. Basel III is a huge improvement on Basel II. It should however be seen as a staging post, not an end-point, and built upon in the years ahead.
Admati, A et al. (2010), “Healthy banking system is the goal, not profitable banks”, Financial Times, 9 November.
Bank of England (2016), Financial Stability Report, November.
Brooke, M, 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 35, Bank of England.
Carney, M (2016), “Redeeming an unforgiving world”, speech at the IIF G20 conference, Shanghai, 25-26 February.
Dowd, K (2017), “Are bank capital levels really ten times higher than before the crisis?”, Adam Smith Institute, July.
European Commission (2012), “High-level Expert Group on reforming the structure of the EU banking sector”, Final Report.
Prudential Regulation Authority (PRA) and Financial Conduct Authority (FCA) (2015), The Failure of HBOS plc.
Shin, H (2016), “Bank capital and monetary policy transmission”, Panel remarks at “The ECB and its Watchers XVII” conference, Frankfurt, 7 April.
Vickers, J (2016) “The systemic risk buffer for UK banks: A response to the Bank of England’s consultation paper”, Journal of Financial Regulation 2(2): 264-282.
 This column is the author’s keynote address at the conference of the Financial Stability Institute, Bank for International Settlements, Basel, 18 September 2017.
 For further discussion, see Dowd (2017).
 These and other figures are taken from PRA and FCA (2015), in particular Table 1.2.
 Letter to the Treasury Select Committee, 5 April 2016.
 My critique of that policy is Vickers (2016).