Reflections on the Capital Accord, risk management and accounting
Speech by Andrew Crockett, General Manager of the Bank for International Settlements and Chairman of the Financial Stability Forum, at the Bankers Club Annual Banquet held in London, 4 February 2002.
President, Your Excellencies, My Lords, Mr Governor, Aldermen, Sheriff, Ladies and Gentlemen, this evening I would like to take the opportunity to share with you some personal reflections at the intersection of risk management, accounting and the proposed New Basel Capital Accord. This will involve recalling briefly how we got here, highlighting what I consider as the key merit of the proposed Accord and reading the tea leaves on what issues are likely to attract attention in the future.
In order not to abuse your patience so late in the day and after such a splendid banquet, I would like to make only four points.
First, while the New Capital Accord has been criticised for being too complex, much of its complexity is the inevitable result of three highly desirable features, viz increased risk sensitivity, wide applicability, and the shift of responsibility for risk measurement for capital purposes towards banks themselves.
Second, arguably the greatest of the many merits of the New Capital Accord is that of hardwiring the credit culture in banking, by which I mean helping to develop and consolidate better awareness, pricing and management of credit risk. This is, inevitably, an evolutionary process.
Third, while banks have made remarkable improvements in credit risk assessment methodologies in recent years, one aspect that deserves greater attention is the measurement and pricing of credit risk over the business cycle.
Finally, looming large on the horizon is the issue of appropriate accounting in banking. One key question here - on which I shall focus today - is how to provision for credit losses and value loans more generally. What is needed is an open and constructive dialogue among all the parties involved despite their different perspectives, viz accountants, supervisors and market participants.
Let me elaborate briefly on each of these points.
I. The Capital Accord: too complex?
One cannot have one's cake and eat it too. It is not possible to have a Capital Accord that reflects risk accurately, that is suitable for banks of widely differing degrees of sophistication, that shifts the responsibility for measuring risk away from regulators to banks, and which, at the same time, is nice and simple. We might dream of one, but reality quickly intrudes. Consider the three desirable features in turn.
First, increased risk sensitivity. All of us have strongly wished for greater risk sensitivity. The lack of differentiation of risk in the original Capital Accord was heavily criticised by banks and observers. The Accord was no doubt a major milestone, underpinning a safer and sounder banking industry. But over time the well known distortions generated by inadequate differentiation in credit quality had become increasingly apparent.
If life is not simple, nor is credit risk measurement. Credit risk is multidimensional. Measuring it correctly requires breaking it down into individual components, viz probabilities of default, loss given default, exposure at default, correlation of defaults across exposures. And the techniques currently used vary considerably across types of borrower and lines of business (eg corporate and retail, securitised vs on-balance sheet portfolios, etc). Given current risk measurement technology and even abstracting from differences across institutions and countries, simplicity is beyond reach.
Next, wide applicability. Partly a victim of its own success, the Capital Accord has become applicable to institutions of widely different degrees of sophistication and from countries with widely different legal traditions and business cultures. But, as banks themselves have repeatedly pointed out, one size does not fit all. The answer has been to introduce a menu approach, with options capable of catering for the needs of a heterogeneous banking population. But more options mean more complexity. And beyond sheer multiplicity, the relationship amongst them becomes important. An obvious example is the need to calibrate the risk weights to provide incentives to shift from the cruder to the more sophisticated approaches. Easier said than done.
Finally, the responsibility for measuring risk. The New Accord is intended to induce some shift in the responsibility for measuring risk away from regulators towards banks. Philosophically, the shift is analogous to the one that has already occurred with market risk, where banks can now use their own models subject to validation by supervisors. Of course, limitations in current credit risk methodologies have set limits to this shift. But the direction is unmistakable, as most obvious in the advanced IRB approach. We all agree that this shift is welcome; indeed, you have clamoured for it for a long time. But such a shift inevitably calls for a strengthening of other checks and balances on banks' risk measurement and management. Hence the explicit recognition and strengthening of Pillar 2 (supervisory review) and Pillar 3 (market discipline, through heightened disclosure). More and taller pillars inevitably mean greater complexity.
The bottom line is simple: there is no free lunch. Increased complexity is unavoidable. To be sure, the Committee is working hard to limit the complexity that may have resulted from the way in which the Accord was presented or the speed at which the whole exercise has been undertaken. But, ultimately, the potential for simplification hinges on the development, streamlining and enhanced reliability of banks' own risk measurement and management systems. In the longer term, therefore, progress is in your hands.
II. The Capital Accord: hardwiring the credit culture
Hence the second point I would like to make today. Increased complexity should not make us lose sight of what, to my mind, is the fundamental and more long-lasting merit of the New Accord, viz helping to hardwire the credit culture in banking. In other words, the Accord is making a major contribution to fostering and consolidating better credit risk management in banking. I have little doubt that, a few years from now, we will look back on this contribution as a salient achievement.
The state of credit risk management in the banking industry is rather paradoxical. Credit risk is simultaneously the new and old frontier. New, because, until recent years, so little had been done at the conceptual and practical level to address it. The most evident symptom is the extraordinary dearth of data which makes it difficult to obtain reliable estimates: most banks have systematically been throwing this data away, not realising that it could represent their ultimate comparative advantage. Old, because, since the origins of the industry, credit risk has been by far the most common source of banking distress and failure.
The reason for this paradox is not hard to find. It is the cosy environment in which banks operated in much of the postwar period under the protective umbrella of tight regulation of the financial system, which limited competition and ensured quasi-rents. Following financial liberalisation globally, this environment is a fading memory. Nowadays, more than ever, risk management holds the key to survival and is a source of competitive advantage, the basis for sustained, as opposed to fleeting or illusory, profitability.
Admittedly, the progress in this area, especially by the most risk-conscious and innovative institutions, has been remarkable in recent years. I recall that, when I first raised this issue in the annual lecture to the British Bankers Association in 1995, I did not expect that advances would proceed so speedily. The key contribution of the proposed New Accord is to seek to foster and, in particular, spread best practice. It is to extend to as many institutions as possible the progress made by the leading ones. Hence the inbuilt incentives to move to the more advanced variants of the Accord, in the form of somewhat lower minimum capital requirements. I think that the educational role of the Accord is already evident. It has substantially raised the level and depth of the debate concerning credit risk management.
Even so, the challenges for supervisors and bank management should not be underestimated. For supervisors, the required upgrading of skills is substantial. For bank management, especially its most senior echelons, it is vital that risk measurement and management is not uncritically entrusted to models and black boxes. The role of judgment, honed by long-standing experience in the business, remains fundamental.
III. Risk management: credit risk over the business cycle
The role of judgment and experience is especially important in the context of the third point I would like to make: despite advances in recent years, an area where progress has not been adequate and which should receive greater attention is how risk varies over the business cycle. Let me elaborate.
Many of the banking crises seen in industrial and emerging market countries since at least the late 1980s arguably reflect instances in which the mutually reinforcing nexus between financial and real developments has gone too far. In our liberalised financial environment, access to external finance has become more plentiful. And it is more intimately driven by perceptions of, and appetite for, risk. These move strongly with economic activity, reinforcing each other. Hence the highly procyclical nature of credit, asset prices and market indicators of risk, such as credit spreads. As a result, financial imbalances, and associated distortions in the real economy, tend to grow in cyclical expansions. Their subsequent unwinding has often overwhelmed the lines of defence that were in place.
To me, this suggests that we are better at judging the relative or cross-sectional risk of different instruments, institutions and counterparties than at assessing how risk, especially system-wide risk, evolves over time. Indicators of risk tend to decline during upswings and to be lowest at or close to the peak of the financial cycle, ie just at the point where, with hindsight, we can see that risk was greatest. And yet, there is a sense in which risk increases during upswings, as financial imbalances build up, and materialises in recessions.
These difficulties in measuring the time dimension of risk have potentially increased in significance with the New Accord, which ties minimum capital requirements to banks' own risk assessments. Hence the need to strengthen efforts to improve our understanding of the problem. What is needed is to make sure that defences are built up in good times, when it is easier and cheaper to do so, to be available in bad times. This makes sense from the perspective of individual institutions and, even more so, from that of the system as a whole. Generalised retrenchment once risk materialises can potentially exacerbate a downturn and hence the deterioration in credit quality for all.
What, then, could be done? This is an exceedingly difficult problem, since solving it runs against the grain of human nature and competitive market forces. But let me identify some possible lines of action. First, we need longer horizons for risk assessment than seem prevalent in the industry. The longer the horizon, the harder it is to have confidence in the continuation of good times. We have to escape from the tyranny of the annual cycles of accounts and taxes. Longer horizons are important for provisioning - as I will discuss later - and for capital purposes. In particular, it is unrealistic to think that capital can be raised so easily under stress, especially if stress is generalised. Second, we need to strengthen prudent instincts at the expense of competitive instincts. Supervisors should encourage the use of conservative assumptions for capital calculations and collateral valuations, with due attention to business cycle effects. Finally, and more specifically, banks should explicitly take into account the greater cyclical variability in minimum requirements in the New Accord. This means holding higher capital cushions that can be partly drawn down in downswings. Ideally, cushions that rise during upswings would be desirable.
The Basel Committee is fully aware of the issue and is taking steps to address it. The Financial Services Authority and the Bank of England have been at the forefront of these efforts. Modifications to the risk weight curve and heightened reliance on stress testing, notably for macroeconomic factors, are obvious possibilities. Language in the proposals aimed at encouraging longer horizons for risk assessment is yet another. Even so, as in other areas, the largest scope for progress is in your hands. This is because the Basel Committee's objective is to establish standards in accordance with industry's own practices; the Committee's choice of a one-year horizon for the quantification of risk is the most obvious example.
IV. Accounting: provisioning and loan valuation
Looking further ahead, the broader and not entirely unrelated question of appropriate accounting for the banking industry looms large on the horizon. While the debate has begun, it is bound to intensify in the future. Admittedly, it is hardly possible to do justice to it in the little time left at my disposal. Given its importance, however, it would be wrong for me to neglect it altogether. Let me make just a few general observations.
First, mounting pressure to address the issue is coming from different quarters. In addition to various incremental modifications to current practices, accountants have, most notably, been developing controversial proposals in favour of fair value accounting (FVA). These have been strongly opposed by much of the commercial banking industry. For their part, banking supervisors have for some time been working towards a greater understanding and even promotion of a degree of further harmonisation in loan valuation and provisioning practices across countries. And, above all, global financial markets have become increasingly intolerant of the gulf in practices across jurisdictions, as it severely hampers and obscures the comparison of the attractiveness of investments.
Second, the stakes are high. Some economists might argue that accounting is just a veil, of little or no significance for economic decisions. I am convinced of the contrary. Accounting practices are a key filter through which we observe reality, measure value and profit. As such, they have, directly and indirectly, a profound influence on those decisions, including as anchors for regulatory and tax arrangements. They affect the efficiency and stability of our economies. And this influence acquires special significance in the banking industry, the nerve centre of our economic systems. Think, for instance, of the tragic impact of the delayed recognition of credit losses in a number of banking crises - crises whose costs in terms of GDP have often run in double digits.
Third, the debate will no doubt highlight the tension between the different perspectives of the various interlocutors. Accountants, for instance, tend to stress "true and accurate valuations" as well as their verifiability. Banking supervisors lay comparatively more emphasis on prudence and may be more willing to accept less immediately verifiable valuations provided that they promote banking soundness. And bankers are primarily concerned with the consistency between the various proposals and the way they internally run their business.
The emerging debate on provisioning and loan valuation neatly illustrates this point. Consider it from the perspective of the procyclicality issue raised before. It is well known that current provisioning practices impart a strong procyclicality in bank profits. This is because provisions are essentially backward-looking: credit losses have to be highly probable before a (specific) provision can be taken. This also means that provisioning falls in a boom, so that measured capital may be overstated in upswings. Banking supervisors are generally more favourably disposed towards various forms of upfront provisioning that could limit procyclicality. Accountants, by contrast, together with tax authorities, are opposed to such practices, seeing them as a way of artificially smoothing, and possibly even manipulating, profits. And FVA proposals, whatever their many merits, could, from this perspective, actually exacerbate procyclicality, given the behaviour of market prices over the cycle.
This illustration, while specific, is symptomatic of a broader inevitable tension between legitimate alternative perspectives. This tension will have to be addressed and overcome. What is needed is an open and constructive dialogue between the various parties involved. I have little doubt that, just as in the case of credit risk, this dialogue will greatly raise the level of the debate and help us deepen our understanding of the issues. At present, even a common language has not yet fully emerged, clouding the discussion. Possible solutions could range from compromises over accounting standards to prudential adjustments on the part of supervisors based on the standards arrived at by the accounting profession. Here, too, the banking community can make an important contribution to reaching a solution.
To conclude, if there is a common thread in my remarks today it is the pivotal role that the banking community, that is, all of you, can play in contributing to address the issues I have identified. This is the logical result of the profound philosophical shift in recent decades towards endowing financial firms with greater responsibility for their actions, a shift reflecting the broader acceptance of the merits of free markets. In this new environment, more than you might think, the key to progress is in your own hands. Ultimately, it depends on your enlightened self-interest.
Thank you.