SATURDAY 25 SEPTEMBER
Howard Davies
Chairman, Financial Services Authority

The role of risk management in dealing with volatility is a very broad subject, which could take one in a number of different directions. But the organisers have helpfully provided a list of four examination questions for us to address. And, since I went to the kind of school which prided itself on examination technique, and where boys had drummed into them the requirement always to answer questions on the paper, I will dutifully do so now.

Question 1. “Does widespread use of statistical risk management tools, such as value-at-risk, contribute to increased systemic volatility?”

Answer: not necessarily. In principle, VaR and PFE, supported by good information, should reduce volatility by reducing the likelihood of concentrations and highly leveraged positions, but if misused they can give a dangerous false sense of security.

Question 2.“Can instruments of liquidity provision at the national level, such as contingent credit lines, provide effective protection from currency runs?”

Answer: perhaps, though the case is at best not proven so far.

Question 3.“To what extent can new financial instruments offer opportunities for risk-pulling (an interesting idea, which I thought about for some time until I realised it was a misprint for risk-pooling) and risk-reduction?”

Answer: in principle, a lot. But in practice they also create opportunities for risk enhancement and risk concentration.

Question 4.“what limits, if any, should be placed on risk-taking by institutions that currently are not regulated?”

Answer: I will tell you next April, when I have completed my report on HLIs for the Financial Stability Forum.

Ladies and gentlemen, thank you for your attention,. Have an enjoyable weekend.

Sadly, I know that this is an examination in which one sentence answers will not suffice. They do not earn the non-existent fee one receives for speaking at these seminars. So I will expand a little on one or two of these points, and particularly on the implications of enhanced volatility for regulators. I will try to be as practical as possible. Not because we are not interested in the theory, but because I think that is likely to be the most distinctive contribution I can make.

But it may be worth saying, first, that financial regulators should not fall into the trap of regarding volatility in financial markets as unambiguously a Bad Thing.

There is no doubt that the dramatic increase we saw in the volatility of financial markets in developing economies in 1997 and 1998 was painful, in both financial and in some cases social terms. Though in relation to the financial systems of those countries it might be regarded as a form of stress testing which has produced results which will in the long run be beneficial to those systems and their supervisors.

And prompt adjustment in prices in financial markets, rather than long drawn out agonies, may be socially efficient in that it encourages early adoption of necessary changes in behaviour.

Economists have long recognised that different régimes generate different amounts and types of volatility. What matters most to regulators, and to individual firms, is how far that volatility, or it’s damaging consequences, can be and are avoided by hedging strategies. And – the core topic for today – how far risk management techniques can both model the impact of volatility in advance, and provide contemporaneous and retrospective methods of mitigating its potentially damaging consequences.

It is fair to say, at the outset, that – overall – financial institutions in the developed world survived the turmoil of ’97 and ’98 remarkably well. So their risk management systems cannot have been as bad as all that. But we have to recall that Western banks were well capitalised, after remarkably prosperous years. And many models failed to predict losses at all accurately, through inadequate correlations and unwarranted assumptions about the relationship, or lack of relationship, between the value of positions and collateral.

So what lessons have we as supervisors learnt about the way in which the regulatory techniques should adapt and evolve to cope with new instruments, new types of institution and new risks?

We attach increasing importance to four features of our regulatory régime which go some way towards developing an appropriate response.

First, supervisors need to be more focused, and systematic, in their assessment of the risk characteristics of an individual firm. So, over the last three years, beginning in the Bank of England and continuing at the FSA, we have instituted a new risk assessment process which delivers a calibrated risk profile of each bank in our care. We assess business risks and control risks first separately, then alongside each other, and tell each bank how we rate them against their peers. A bank might, overall, be a low risk institution but with some high risk areas, which will then be the focus of our regulatory effort. And we tell banks straightforwardly whether we think that their control environment is as rigorous as that operated by comparable firms in comparable markets. We are perhaps helped in doing this by the large number of institutions we supervise in London. That does give us a large database of major institutions on which to base our assessments.

Of course these assessments cannot remain snapshots and, for the firms involved in the most volatile markets, we need a continuous dialogue with their management to keep our evaluations up to date. That is quite resource intensive, but we – and I believe the firms themselves – regard it as worthwhile. We also pay continuous attention to the economic and market environments in which the firms operate.

Second, we increasingly believe that it is important to look at institutional risks in the round, and to look at the impact of different markets across different types of institution. We have an increasing number of bank assurers in the United Kingdom and of course banking and securities business has traditionally been transacted in the same group in our régime.

Looked at from the market or counterparty perspective, many of our insurance companies are active in derivatives markets, and when we did our trawl of LTCM counterparties, we found some in the insurance sector as well as among banks and securities firms. Perhaps we are conditioned by our structure to take this view, but we increasingly believe that integrated cross-sectoral regulation is crucial to reach a full understanding of the interaction of risks between institutions, between business sectors and between markets.

Third, there is clearly a need for regulators to enhance the skill base of their front‑line staff. That is not easy to achieve, given the staff turnover we face in markets like London’s, where well qualified regulatory staff are much in demand. It requires some careful thought to be given to the organisation of regulatory expertise. It may simply not be realistic to think that one can educate and train every banking supervisor to be able to assess a value-at-risk model, or to look hard at capital adequacy. So we have instituted cross-institutional teams, in some areas, to look at areas where a particularly high level of expertise is required. So we have a traded risk department which covers a wide range of institutions. And we set up a targeted team to look at exposures to HLIs, and how they were managed, in the aftermath of the LTCM crisis. We are now developing some further ideas on theme based supervision which could, over time, amount to quite a sizeable change in the way we deploy our resources.

Fourth, we are firmly convinced that it is vital to look at risk management in the context of its position within the firm as a whole, and to look at the management structures and behavioural patterns which surround it, as well as at the details of the VaR models themselves. It is vital to ensure that VaR results are embodied into a proper control structure and that senior management are committed to sound risk management and to sponsorship of a good risk aware culture. It is also important to ensure that there are appropriate accountability structures in place. We have recently proposed to the market a new régime of approved persons which aims to capture more accurately than before the individuals who are responsible for position taking and risk in different parts of the business, and to pin specific regulatory responsibilities on them.

Senior management should be able to articulate the risk appetite of the firm. And they should be able to ensure that all aspects of the way the firm is managed, including crucially its remuneration policies, are consistent with that risk appetite. We have often found firms where senior management claim that they have a particular approach to risk in different markets but where the incentive structure put in place for their traders, and heads of trading, push them to take on higher risks in pursuit of highly leveraged short term personal returns.

I said at the start that I did not propose to look at the details of risk models. I know, from having read an extremely interesting paper he has written on risk management lessons from LTCM, that Philippe Jorion will have more to say on that subject. Our own view is that the key for supervisors is to ensure that stress testing is performed to examine key vulnerabilities to which the organisation is exposed. And furthermore that risk assessment is, in practice, firmly related to business decisions. Is there evidence that when risk models show that the institution is exposed, key business decisions such as cutting positions, hedging or mitigating risks, diversification, or simply running down the book, are in practice taken? We have found too many occasions where the output of VaR models is regarded as an interesting academic exercise which is not hard wired into business strategy choices.

These are some general lessons about the way in which supervisors should respond. And I would add two or three observations about more recent events, from our own experience.

We think it particularly important to ensure that new ideas on good risk management practices are properly embedded in the firms we supervise. Immediately after the LTCM episode there were many in the City who said that nothing clever was needed in response, just the application of sound banking principles: know your customer, and all that. For a time, there was no doubt that banks tightened up on their exposures to highly leveraged institutions. But just as the learning curve in these markets is steep, so is the forgetting curve, and we have already identified examples of firms beginning to loosen up and to accept less disclosure from their counterparties – which was, of course, where we came in. So we have been following up, along with other banking supervisors, the recommendations in Jan Brockmeijer’s excellent report for the Basel Committee. The job of managing risk is, of course, one for management. And supervisors have no comparative advantage in developing new risk management tools. But we can, I think, be a useful stimulus to promote good practice.

We have visited a range of banks and securities firms to establish the extent of their compliance with the sound practices in that paper and the areas where further work is required. That has shown a mixed picture, but some progress overall. And we think it important to go beyond that and to assess the extent to which the more sophisticated risk management ideas in the Corrigan/Thieke Counterparty Risk Management Policy Group report are being followed. This may require the introduction of regulatory incentives to persuade managements to take improvements seriously.

We have also, again as part of a Basel Committee exercise, been examining the performance of VaR models in conditions of the highest volatility. As might be expected, there were many instances of so-called “back testing exceptions” – instances where trading losses exceeded the one day VaR estimate. Some banks fell in to the “yellow zone”, in other words they experienced between five and ten exceptions. But there were no banks classified in the red light zone (defined as ten or more exceptions in a 250 day period). The size of exceptions included some that fell between two and three times the one day VaR. But none exceeded three times the one day VaR, which gives some support for the three times multiplier imposed by regulators in the face of much industry criticism three years or so ago.

This evidence needs to be put alongside claims that the models “broke down” over the period. No evidence has emerged to us to suggest that VaR models which are operated in banks that meet both the qualitative and quantitative criteria failed to provide significant capital this time, though it is fair to say that evidence remains anecdotal, and, as the best health warnings say, the past is not necessarily a guide to the future. So we are not in any way diminishing our own focus on the assessment of models, rather the reverse.

Finally, I am conscious that I ducked an answer to the last question, on whether limits should be placed on risk taking by institutions not currently regulated.

Pedantically, I could say that it is not possible to impose limits on unregulated institutions. But I guess the question is, should the regulatory net be drawn more widely?

That is indeed the question which the group I chair on HLIs, reporting to the Financial Stability Forum, have been asked to address. And the G7 finance ministers have made it clear that in doing so we should look at both indirect regulatory approaches, operating through controls on regulated firms, and direct approaches, whether by the imposition of disclosure requirements or, indeed, a full scale régime of registration and prudential supervision.

At this stage, it would be fair to say that there is widespread consensus that some improvements in our ability to understand the potential systemic risks created by highly leveraged unregulated institutions are needed. There is strong support for measures to enhance the transparency of markets and counterparties. The President’s Working Group here in the US proposed some changes which could have quite far reaching impact on disclosure and transparency of these institutions. And others have gone further to propose full-scale regulatory régimes. It is our task over the next six months to assess the pros and cons of these arguments and present a considered report to the Financial Stability Forum at its next meeting in April. I can only tell you now that I am, personally, as interested, and indeed intrigued as you no doubt are to know just how this will turn out.

More Speeches: