LONDON, THURSDAY 2 DECEMBER
Howard Davies
Chairman, Financial Services Authority, United Kingdom

Why bother?

We supervisors and central bankers have recognised pretty openly over the past couple of years that our favourite armchair, the 1988 Accord, is bursting at the seams - and nowhere more so than in London, where so many complex, innovative transactions are done by banks from so many countries.

The Accord was designed to work ‘on average’. But things don’t work on average any more. Revolutionary improvements in risk management techniques have allowed banks to focus on marginal capital, that is the extra capital needed to do the deal you’re thinking of doing. So the perverse marginal incentives (to lend to riskier corporates, for example) - of which regulators were perfectly aware even in 1988 - have become more of a problem. And since marginal incentives matter more, it becomes more important that hedging should be rewarded; or where that is not possible, at least not punished.

The 1988 Accord reversed the secular erosion of global banking capital, and capital ratios rose around the G10 for several years. So it did its job. But recently, it has become clear than a rising capital ratio does not necessarily mean improved capital adequacy.

Let me explain. Smart people try to design products that will be placed into one regulatory risk category while having different economic risk characteristics. Where the economic risk diverges from the regulatory risk, a bank can exploit this difference. This will be asymmetric: the bank can choose the economic risk if that is the lower, or the regulatory risk where that is the lower. The result is an erosion of capital standards. Given a fixed amount of risk, a sophisticated bank can achieve almost any ratio it wants, by clever use of credit derivatives, securitisation and so on.

So once these problems had become acute, we elected radical surgery.

Objectives

In revising the Accord, the Basel Committee defined four objectives. These are no secret; in fact they are set out in the June paper. The Accord should:

  • continue to promote safety and soundness in financial system;
  • continue to enhance competitive equality;
  • constitute a more comprehensive approach to addressing risks; and
  • focus on internationally active banks, but its principles should be applicable to others.

The first two of these – safety and soundness, and competitive equality – were also the objectives of the 1988 Accord. In a few minutes David Clementi will talk about the first, systemic risk, so I shall not dirty his bathwater. I know what he’s going to say, and I would have liked to say a lot of it myself. But he processed his words first.

I cannot entirely avoid the subject, however. When we spend our time reviewing the rules, we must be conscious of the global financial system and our place in it. The prudential supervision of financial institutions has a huge role to play in trying to prevent the build-up of concentrations of risk in the financial system. The pursuit of financial stability is the joint responsibility of financial supervisors, central banks, the IMF and so on, all of whom meet twice a year at the Financial Stability Forum. Avoiding or mitigating crises requires close co-operation between supervisors and others. Here, the Memorandum of Understanding agreed between the Bank of England, the FSA and the Treasury provides the framework within which we try to anticipate crises, and plan how to manage them.

Of the four objectives, the other two are new – and I want to focus on the last one: the notion that the new Accord should focus on internationally active banks, but its principles should be applicable to others.

The fourth objective: wide applicability

The 1988 Accord was designed to be applied only to internationally-active banks in the G10 countries that signed up. But in practice it has been applied in some form in over 100 countries. In many of those countries, it is applied to all banks. And in the EU it is applied, via Directives, to all banks (and, in the UK, building societies) and also to securities firms, fund managers and even some IFAs.

That it has been applied so widely is an enormous tribute to the Accord. And it is a good thing, creating as it does (in theory) a common benchmark for banking solvency around the world. The Accord is a model of parsimony, and strikes a balance between accuracy and simplicity that, almost perversely, has become over time less attractive to those banks for which it was designed, but no less attractive for others.

But this wonderful export performance brings with it challenges and responsibilities. The Accord has become the single de facto benchmark for banks’ solvency around the world. The challenge is to design a new Accord that could still be applied to banks large and small around the world, including the most sophisticated banks. That success raises at least two problems for those engaged in the current review.

The first is a franchise gap. Can the G10 supervisors really presume to design rules for all non-G10 countries?

I suggest that we cannot actually avoid it, for the new Accord will be implemented world-wide whether we like it or not. So the Basel Committee needs to take into account its unavoidable responsibility in this area. This reality is becoming increasingly clear to all of us on the Committee, and we have begun a number of outreach initiatives.

No international organisation can fix its membership for all time. Should we now widen the membership of the Basel Committee? We should be careful, as the Basel Committee’s relative leanness is a great advantage. The Committee works by consensus, which means each member effectively has a veto. But it gets things done. If we materially expand the Committee we risk paralysis, and we should not take this risk lightly.

To close the franchise gap, I think we need genuine and deep consultation between the Basel Committee and supervisors around the world. The Committee has a liaison group with other central banks and supervisors, which is playing and needs to play a full part in the review. At the same time, all those involved in framing proposals in Basel need to consider applicability outside the G10 not as an afterthought, but as a core part of the analysis.

The second challenge posed by our export success is this: can we design rules that can be applied across the world? is the new Accord potentially applicable in countries with less well-equipped supervisors, with shallower or non-existent capital markets, with lower disclosure standards? And we must have One Accord. Having one broadly applicable agreement puts great pressure on governments and supervisors around the world to adopt consistent minimum standards of prudential supervision. We all benefit from this pressure, particularly here in London. I know that Bill McDonough, who chairs the Committee, agrees with me. So does Andrew Crockett, who chairs the Financial Stability Forum.

The question is in the regulatory ether at the moment. It has much in common with an EU question: can we design laws that are appropriate for large national clearing banks and small regional co-operative banks? That question is also implicit in the Committee’s fourth objective. In fact, it is worthwhile asking why the Committee focuses on the internationally-active. In practice, the Accord is the standard for domestic banks in the G10 too. Risk in the international financial system is not posed only by the activities of the internationally-active. The collapse of a domestic banking system will have significant, negative international effects. Look at Japan.

That’s the easier EU question. The harder one is: can we design laws that are appropriate for a big universal bank, a bulge bracket securities house, a tiny mutual mortgage lender, a fund manager and a two-man IFA? Senator George Mitchell has some spare time now; but even he might struggle with this one. These are strikingly diverse businesses. For example, for most investment firms, credit risk is not the most significant risk, but no business is immune from operational risk. The ‘other risks’ capital requirements will therefore be particularly important for our investment firms.

The Basel question also has a lot in common with a more parochial question: can we design a common rule book for all the institutions authorised and supervised by the FSA, as we are trying to do?

In all cases, it is not likely to get any easier to design one set of rules for all. But I believe we still can. Let me suggest some principles that apply to the Basel question.

i) Four principles for Lycra regulation

In order to have rules that are appropriate to different banks, we need a menu approach. For example, we might have one approach that is a quite accurate measure of the risks, but is expensive to implement. The other is a simpler standard approach. I am thinking of the Market Risk Amendment (CAD and CAD2, in Eurospeak).

The market risk approach is a good model. Indeed, in the field of credit risk it may be that we need more than two options. We need a standard approach – using rating agencies, according to our proposal. We need to use banks’ internal ratings. At the FSA, we guess that, at least in the EU, the range of institutions that it likely to want to use internal ratings could be so wide that we shall either impose excessive standards on some or ‘dumb down’ others. This is a smaller version of the same problem, and the answer is the same – more than one internal ratings approach. And ultimately, although not in this round of changes, we shall I am sure see credit risk models being used. With ‘other risks’ too, we shall almost certainly need a standard approach for the majority of institutions while leaving open the possibility – indeed, the hope – of proper models being used in time by the more sophisticated.

Secondly, where even simple rules impose a burden that is not justified by the prudential benefit, we need waivers. Again, the CAD allows institutions that undertake zero or small amounts of trading not to bother with the CAD framework, since the risks are not material. The new rules for interest rate risk in the banking book are likely to give a mirror-image waiver to firms that really only trade.

Where different options are available for the same risk, should each approach deliver comparable levels of capital requirements? In the context of market risk, we said no. This was not an attempt to favour big banks. The VaR approach is a more accurate measure of market risk than the standard approach. The standard approach needs to be calibrated with caution in order to ensure that a bank will achieve a minimum degree of soundness whatever its portfolio. On average, this will result in a higher capital charge. Is this unfair? No! To see this as an unequal outcome would be to miss the point. It is a matter of equal opportunity rather than equal outcome. In market risk, all banks are given the opportunity to invest in order to achieve the minimum standards required for the VaR approach.

The same considerations should apply to credit risk and operational risk. It may not be easy to calibrate, and the incentive to adopt the accurate approach need be too strong, but accurate techniques for measuring credit risk should result in slightly lower capital requirements than broad-brush ones.

My second principle is this: the Basel Accord does not work without an infrastructure to support it. The Asian and other crises have shown that successful banking supervision cannot work in isolation – it requires solid foundations. Clear property rights, minimum standards of disclosure, freedom from political interference and corruption - for banks and their supervisors, and minimum standards of supervision remain preconditions for successful implementation of the 1988 Accord. So any country wanting to implement the Accord must get the Basel Core Principles done first. If they do not do so, their banks will not be able to qualify for a preferential risk weighting.

Capital rules are pointless, for example, without adequate provisioning. There are many countries where banks are allowed to get away for years with inflating their reported capital, and there are some countries where banks have even been encouraged to do so.

The third principle: take account of country circumstances. This is not an excuse for a race to the bottom, seeking competitive advantage wherever you can find it, but a recognition that we do not live in a superstate. The entire cultural, legal and financial substrate of a country is crucially relevant to supervision. Stark differences in taxation, accounting systems and insolvency laws can render it rather pointless to tinker with details of capital adequacy requirements.

The fourth principle is closely linked to the third: if one of the pillars is weak, lean on the others more. Take market discipline. Information has less impact in countries where the financial markets are not deep and liquid. In such countries, a greater intensity of supervision and higher capital requirements are generally needed to make up for the lack of market discipline.

If market discipline is weak, it is all the more important to open up markets to foreign firms. I believe there is a need for developing countries to make much stronger commitments to remove barriers to entry at the forthcoming GATS talks at the World Trade Organisation.

Equally, where supervisors are weak, under-resourced and so on, they need to rely more on other pillars to protect others.

So, in many countries, 8% is simply not enough.

As a concluding remark, let me say that emerging market supervisors may see the Basel proposals as difficult, resource-intensive, maybe needing new legislation. And they will not be alone. But if we get the new proposals right, they should also see the proposals as offering a huge opportunity for supervisors from these countries to put in place systems able to cope with the challenges of the next decade.

The three pillars of the temple of Baal

The Basel paper was published as a single document for a good reason: it should be considered as a whole. The three pillars must be looked at together.

Why three pillars? Capital requirements alone do not deliver, and never have delivered, depositor protection or a stable financial system – because they are not fully incentive-compatible - and so all the Basel supervisors do more. Much of the rest of the day will focus on the first pillar, so I shall focus on the other two. They are no less important.

Supervisory review

The second pillar of Basel – supervisory review – is pretty similar to what we at the FSA do now for banks. There is much more interest in this pillar in other countries, particularly in the EU, and that is because the implications are greater, almost revolutionary in places.

What prudential supervision is about is helping protect other people from the failure of the institution by trying to ensure the institution is adequately run. An adequately-run institution needs to know why it’s in business. It needs to have a strategy and some idea of where its revenues will come from. It needs to know what kind of risks it faces and, preferably, to try to measure them. It needs to know what kinds of risks it wants to face and take measures to eliminate the rest. And it needs to have some way of telling how much capital it needs to deliver an acceptable risk-adjusted return to shareholders.

One option included in the supervisory review pillar is individual minimum capital ratios, which you will know as trigger ratios if you’re at a bank supervised by FSA. While setting individual capital requirements is not easy, we think it is better than not doing so at all. It demands supervisory discretion and consistency, skills and controls. In the UK, we banking supervisors have laboured to improve our risk assessment processes in the last few years and we continue to do so. We do not think we have achieved perfection, but we are convinced that it is right to make these judgements.

An equal capital requirement across all types of bank/firm is not a level playing field, because some institutions pose more of a risk to depositors or to the system than others, and not all of this risk is captured in the minimum capital framework. Some institutions ‘pollute’ the financial system more than others; and in order to preserve confidence in the financial system while using scarce capital efficiently, the greatest ‘polluters’ should have the most capital. Compared to this ideal benchmark, flat rate requirements are a waste of capital. Even worse, they are a subsidy from firms that pose less risk to others to those firms that pose more risk – which is, surely, a distortion of competition.

The Treasury have recently confirmed that we shall have a general duty to take competition into account in the course of fulfilling our statutory responsibilities, but that competition will not be one of our statutory objectives. That is entirely right: competition is a means, not an end. In general, competition is the best way to make people better off. One of the best ways to protect consumers’ interests is by encouraging an efficient financial system. In order to do this we must be open to innovation, and in order to be open we must allow competition. Where we choose to restrain competition, we shall have to have a good reason for doing so, by reference to our statutory responsibilities. But a lot of financial regulation is built on the fact that in many financial activities there is not enough shared information for a market to function efficiently – so competition doesn’t always achieve the best that can be done. In prudential regulation I regard our trigger ratio regime not as a distortion of competition, but as a heavy roller with which to iron out the bigger bumps in the playing field.

To return to the second pillar, setting individual ratios is a paradigm shift for many supervisors, with serious resource and cultural implications. There are also legal problems in many countries. But (at least in Europe), the principle that different firms deserve different capital requirements appears to be widely accepted, and that principle is clearly explained in Chapter IV of the Commission’s paper.

Disclosure

Turning to the third pillar, much of the rationale for regulation rests on problems caused by what economists, with their command of economics but not of English, call asymmetries of information: that is, financial institutions know much more about their operations than anyone else. Other things being equal, the more firms disclose, the less the asymmetry, and the less we regulators need to butt in. But this doesn’t justify disclosure by everyone of everything: in regulation as in couture, complete transparency is not practical.

Information is not free to produce and it’s not free to process. Small investors do not have strong incentives to conduct full analysis of published information, or even to acquire the skills to do so, which is why we have depositor protection and investor compensation schemes. Someone needs to analyse the often highly complex information published by financial institutions in order to transmute dull disclosure into lustrous discipline. This analysis is costly – very costly. So the disciplinary benefit has to be worth the effort.

But we are in favour of at least as much information disclosure as we have now, and we support efforts to increase disclosure. Indeed we are involved in international disclosure initiatives on more fronts than I can count.

One important way of reducing systemic risk is to build into the system better incentives for institutions to take into account their own risks individually. One way in which the new proposals try to do this is by introducing disclosure into Pillar 1. In order for a sovereign exposure to qualify for a risk weighting below 100%, the sovereign must subscribe to the Special Data Dissemination Standards (disclosure standards). This reinforcing link between the first and third pillars looks promising.

Not all proposals made so far will end up being implemented. There is a proposal receiving attention, particularly in the US, and promoted more than once by the Economist, that first, banks should be required to issue subordinated debt, on the grounds that subordinated debt holders have interests more similar to depositors than to shareholders. The second part of the proposal is that supervisors should precommit to some regulatory intervention in the event that the spread on a bank’s debt exceeds a certain amount.

At this stage, we are not convinced. There are a lot of ‘if’s. For one thing, you have to have liquid and efficient markets. You have to be sure that spreads reflect the market’s perception of the idiosyncratic risks and not much else. In the UK and elsewhere last year, spreads increased very significantly, and I do not believe that the entire jump in spreads could be justified by changing assessment of the risks to the UK banking sector. There were other factors at play, a general loss of liquidity for example. If we as supervisors pre-commit to a specified action when a margin exceeds some percentage, then can avoid regulatory forbearance, we may also end up taking unhelpful actions. And secondly, it may work at best for large institutions but not for everybody. Those who promote the scheme need to explain how it can work for small banks and firms. So while we at the FSA are not convinced by this idea, we would welcome more work on the idea and in the area of disclosure more generally.

ii) The European Commission paper

The approach in the European Consultation Document is broadly consistent with that in the Basel Committee's June Consultation Paper, and we thoroughly welcome that. The EU members states also endorse the three-pillar approach.

In some areas, though, the Commission has made clear progress over the Basel work, filling in some of the gaps and, I hope, giving you more to chew on. Most notably: other risks, interest rate risk in the banking book, credit risk mitigation – which takes up an amazing twenty pages of the Brussels paper - and supervisory review, as I mentioned earlier.

We would like the implementation of the new Basel Accord and of the new Directive – let’s call it CAD3 – to happen at the same time. You may recall that the Basel Market Risk Amendment was due to be implemented at the beginning of last year, while CAD2 was only agreed over the summer. This caused us a big problem. In the UK we delayed the implementation of the Basel package and implemented both together at the end of September last year. We don’t want to have to do something similar this time. Seven other EU member states will face the same issues, of course.

I’m afraid it’s by no means certain that we can achieve simultaneity. Although eight EU nations, plus the Commission, are represented at the Basel Committee and so guarantee a certain commonality, both processes are uncertain in their timing. Speakers later in the day will have more to say about the work that has been going on in Basel since June and how and when it will be published. But Basel could move swiftly. I’m not convinced that it will - so much remains to be done, and we must consult on the new work, not simply impose it – but it could. If it does go very smoothly, then I think we shall have to agree on a delay before implementation. The 1988 Basel Accord was phased in only by the end of 1992, after all. This time round, the delay would be both to allow institutions and supervisors prepare for the new rules, and to allow the EU process to run its course. The EU is too big for the Basel Committee to ignore. And this time around, to avoid dislocation, I wonder whether we should not simply set a deadline but a single agreed implementation date. This would not be driven by fears of competitive inequality – after all, we want the new rules to deliver roughly the same amount of aggregate capital – but because significantly different treatments of similar exposures could cause migrations of business, a concentration of risks in different regions, and enormous international regulatory arbitrage.

The Basel Accord can be amended without a legalistic process. Amending EU Directives requires a very complicated and often lengthy co-decision process involving Council, Parliament and Commission. This paper represents Stage 0 of that process. But since much of the time will be taken up with political discussions, we need to be doing the thinking as soon as possible.

More Speeches: