Speech by Andrew Bailey, Managing Director, Prudential Business Unit at the Bank of America Merrill Lynch Conference, London
Thank you for inviting me to speak today. I want to talk about the challenges bank supervisors face in determining the appropriate approach to assessing the capital requirements for banks. Scarcely a new subject I know, but nonetheless one that is at the top of the agenda of regulators, both macro and microprudential. My starting point today is not whether banks are, or are not, adequately capitalised – I will come to that – but the framework by which supervisors assess capital needs, a subject that is attracting a great deal of attention.
I will put my cards on the table at the start. I am not a believer that supervisors should rely on any single data point when assessing a given firm’s stability. We need to assess measures of capital, liquidity and leverage. The art of supervision is to look at a situation from several angles and seek thereby to identify weaknesses. Good supervision is about judgement. The psychologist and behavioural economist Daniel Kahneman has described good judgement as, ‘the product of a solid grasp of intuition coupled with empirical validation where possible’.
There was too much box-ticking in the past, but I do not intend there to be in the future. I was dismayed to hear recently a commentator saying that, ‘supervisors will never be more than box-tickers’. Wrong. If you don’t agree, meet my colleagues. I was also interested to read recently a claim that today supervisors are unable to stand up to the banking lobby. Wrong. We accept that in a judgement-based approach to supervision, there will be disagreement. We don’t seek disagreement for the love of it, but it is part of what we do.
In terms of how supervision is done, my preference is to start with a risk-based approach to assessing capital needs and then back that up with other risk-based approaches and a simple non-risk based approach, like a leverage ratio. The general approach is also the one that we are developing for insurers by designing early warning indicators to act as a check on the modelled risk-based approach.
Why do I start with a risk-based approach? Because there is a logic to the approach of the supervisor mirroring how the firm manages risk for itself. I would be horrified if a firm said to us, ‘“you know, we don’t bother to assess the risk we are taking’. Some history is relevant here. Modern risk-based capital requirements began with the Basel I Accord in the late 1980s. I was in Banking Supervision at the Bank of England then as we implemented Basel I.
It had simple, crude, risk weights, but it was better than anything that had come before. In the Basel I era, I remember that conversations with banks often began with them saying: ‘we are happy to discuss risk-based capital requirements as you the supervisors measure them, but that is not how we manage risk’. This statement would not have been bad had the Basel I approach acted as a restraint on how firms took risk, whether or not it was how they measured risk. Unfortunately, that was at best only partially true.
Basel II is much criticised, and there is much to criticise, but one thing we should not forget is that it had barely been implemented when the financial crisis began, indeed Basel II was not implemented in the US at all. The excessive leverage and risk-taking had grown up in the Basel I world.
Interestingly, in the US, unlike in Europe, banks were effectively constrained from balance sheet expansion by a regulatory leverage ratio. But their response to this constraint appears to have been to seek a higher return on equity through holding riskier assets drawn from, for instance, sub-prime and leveraged loans.
When we look at the causes and types of bank failures in the crisis, the point is often made in this country that the problems appear to have been located in mortgage and commercial property lending as much as investment banking.
That is true. Basel I was flawed in that it failed adequately to cover investment banking activities that grew after the late 1980s, arguably because of the regulatory arbitrage. Basel II failed to fill these gaps adequately. But Basel I, through its simple risk weights, also encouraged the property-lending boom. One part of the story goes something like this: demutalised building societies came under pressure to generate higher returns for shareholders and thus devised business models that pushed them towards greater leverage and larger risks for a given amount of lending – in other words, higher LTV lending. And they also exploited another flaw in Basel I, the ability to securitise loans without properly assessing the impact on the viability of the business models. Thus, Northern Rock grew its high LTV, low-margin lending and developed a securitisation operation that was ultimately too big for the bank to sustain.
Now to be clear, this is not a defence of Basel II. Far from it. Basel II introduced a model-based approach alongside the so-called standardised approach. It created complexity, and it encouraged the use of models generally, including for assets that are not, in my view, susceptible to robust modelling. Basel II was also implemented in what looks like too much haste – it may not have seemed that way at the time I accept. But bad models were allowed through the gate. Commercial property was a bad case in point. One has to ask whether commercial property loans are too lumpy to be modelled. But, modelled they were.
And, we are now undoing that with the so-called slotting approach, which involves putting loans into risk buckets according to key characteristics.
Supervisors have taken, broadly, four actions to correct the failings of Basel I and II. First, to raise sharply the capital requirements for trading book and investment banking activities. In doing so, we are changing the terms of trade for investment banking of this sort, and we are, of course, seeing the consequences. No apologies there I’m afraid.
Second, we are, as I have just said, taking some assets out of the models regime because they should not have been there in the first place.
Third, where assets are in the models regime, we are moving to create floors to prevent low modelled risk weights driving overall capital requirements down to imprudent levels. There is no doubt that an industry has grown up to arbitrage models and we will not in future let that lead to irresponsible practices. Let me take the case of mortgage risk weights, which has attracted recent attention. Prime mortgages are the sorts of assets that should be susceptible to modelling – there are a lot of them and they have been around for a long time. We still require the models calibration to take account of the early 1990s recession. Specifically, we don’t want models to use the relatively benign loan losses of recent years to reduce capital held. Why?
The recent recession has been accompanied by very low interest rates, which have kept loan losses down, but this may or may not be the picture in the future. It would be unwise to bet on it happening, and particularly when we know that loan forbearance is quite widespread.
The fourth thing that supervisors have done is to apply judgement in overlaying the Pillar 1 modelled capital requirements with Pillar 2 capital buffers. Since 2008, required Pillar 1 capital in the major UK banks has increased from £151bn to £186bn. Pillar 2 capital buffers set by the FSA, in all, have increased during the same period from just under £20bn to £150bn. Put simply, in the regime up to 2008, there was no judgemental overlay of capital buffers, now there is such a buffer. This is a product of good judgemental supervision.
What is the role of a leverage ratio? For me, it is not the frontline tool, but rather the very useful back-up check. Why is it not in the frontline? Because it would not on its own prevent the two main causes of the crisis. First, savvy investment bankers would take assets off balance sheet into places where a leverage ratio may not be the best measure. And second, on its own it does not stop the arbitrage whereby the riskiness of each unit of assets rises as a means to circumvent the discipline of the leverage ratio. But, it is a very useful tool, because on their own, risk-based measures can be arbitraged too by claiming that the measured risk has fallen when the real risk has not.
The leverage ratio helps to put a floor into the system which stops that type of activity beyond a certain point.
Does this approach of using several tools leave the whole system of regulation too complicated? It certainly puts an emphasis on having good people in regulation, but I come back to the quote of the late 1980s, namely that good regulation has to understand how firms take risks. That does not, however, mean that a supervisor should allow a free-for-all and then (usually vainly) hope to understand it. We are doing a number of important things to counter complexity. First, the proposal on ringfencing from the Independent Commission has a strong element of separating more straightforward activities from the more complex. It should not be a matter of rules and good behaviour inside the ringfence and the Wild West outside, because the outside could damage the financial system, but we will be able to use different approaches on each. Second, the Prudential Regulation Authority will be very focused in its judgements on the things that matter for our objectives. For me, this is where simplicity is key – keep it very focused, but to do that we must have good people who can see the wood for the trees. That is what we are developing for the PRA. And third, as part of this, we are very focused on understanding business models, how firms make money, the risks they take. This, too, is a feature common to our supervision of banks and insurers.
To give you an example, the high return on equity and low cost of equity business model is dead. It should never had been alive, because it could only exist through a misunderstanding of risks. Moreover, banks can alter their cost of equity through the choices they make on risk.
Sometimes I hear the cost of equity explained as if it is exogenous, but it is not.
Let me turn now to the current situation. I am not going to tell you the answer to whether the UK banks need extra capital, if so, how much of what type, and how rapidly should it be put in. The answer will of course differ for each bank. Let me though set out some of the considerations that shape the answers, and in doing so remember that we have to take into consideration the sizeable increase in capital in the system since 2008. And let me also emphasise that the FPC does need to reach a conclusion on these issues soon, and in doing so reduce one contribution to uncertainty over the future.
The first consideration to note is that we face a situation where there is uncertainty around our judgements on the quality of a number of asset classes and their valuation. This is a reflection of the unusual and difficult economic conjuncture that we have been going through, combined with the challenge of determining the scale and nature of, for instance, loan forbearance.
This is particularly evident in areas like commercial property lending where there is a hump of loan refinancing to come and a quite sharp tiering of performance among assets.
Second, the euro area, and the possible impact of disorderly break-up, should that happen (and in saying this, I offer no view on whether it will happen), could have a sizeable impact on many banks. But here the FPC faces another very difficult judgement, namely what scale of so-called tail risk –the losses from low probability but high impact events – do we wish to see covered in the capital held now by banks? In other words, how much insurance should come from capital providers? The broad answer is a lot, but that does not tell us how much, and in the limit there are utterly catastrophic risks for which it is not sensible to hold capital as the answer.
Third, when we look at the low market value of some banks relative to their book value, how much of this should we attribute to lending margins being considerably lower than they were expected to be? Net interest margins are squeezed in the current environment. This is important because, to the extent this argument holds true, the action taken to deal with it, if any, could be different.
Another relevant consideration is to what extent it reflects different market valuations of banks’ tangible assets or of their intangible assets, such as goodwill, that are to some extent already and will to a greater extent under Basel III be accorded no value for regulatory capital purposes.
Fourth, to what extent are low market values caused by uncertainty among investors about the future regulatory and operating environment for banks? This could be caused by a number of factors, including ringfencing and capital policy as supervisors adjust risk weights. Fifth, by how much further should we adjust risk weights to reflect past inadequacies.
And, finally, how rapidly should we make any adjustments, and how should such adjustments fit with both the Basel III implementation timetable and the changes that firms themselves are making?
In conclusion, and pulling all of this together, there is a lot of talk about the need for simplicity. I agree. The proliferation of rules is unhelpful and particularly when it extends into making rules on how to supervise, a feature that we see in the EU processes. To return to Kahneman’s point, solid intuition is a part of supervision. But simplicity is not about one-club golf, and it is not about abandoning risk-based regulation. A simple timeline would suggest that Basel I did more to cause the crisis than Basel II.
But the latter would surely have made it even worse had it be in place for longer. Looking forwards, we do need strong judgemental supervision, and we do need the powers to separate trading activities into separate legal entities as the ICB has proposed. As Paul Volcker recently pointed out, customer banking involves a fiduciary duty, whereas trading with counterparties does not. My only caveat here, and it is a very important one, is that the fiduciary duty of customer banking was sadly lost in the wave of mis-selling.
Judgemental supervision has already been applied to correct the excesses of pre-crisis activities. Investment banks have been forced to lower leverage and raise capital, and we are seeing the consequences. The monoline mortgage bank model – in the commercial not mutual-sector has been restructured heavily already. No demutualised building society survives as an independent entity today. The high return on equity with low cost of equity business model is dead. Of course, there is more to be done, and I have set out the framework within which I think about the capital needs of the banking system. Above all, simplicity is about clear focus and a firm resolve.
Thank you
No comments:
Post a Comment