June 6, 2002
The Road to Basel II: Good Intentions and Imposing Challenges
I want to thank Pam Martin for inviting me to join RMA for this conference on capital. These conferences are of enormous value in keeping the industry well informed on developments in the area of capital — and thus in assisting the industry to contribute to the process of policy development.
As many of you know, RMA's involvement in the process doesn't end when the last attendee has left for home. We in the regulatory world have come to count on RMA's thoughtful advice — and on its criticism, as well — in helping us to make sure that our approach to capital regulation makes sense, both for supervisors and the industry.
The process of revising the 1988 Basel capital Accord — Basel I, as it's now nostalgically known — began modestly and sensibly enough. Hailed as a breakthrough when it was formulated, the 1988 Accord aged quickly — and not altogether gracefully.
Within just a few years of its adoption, the Accord was the subject of numerous criticisms, many of which centered on its failure to differentiate adequately among assets of varying risks. Indeed, with each passing year it became increasingly clear that the Basel rules were even proving counterproductive in some respects. For example, by encouraging some institutions to move high quality assets off the balance sheet, the Basel rules were having the effect of reducing the average quality of bank loan portfolios.
By 1998 it became clear that a fundamental overhaul of the capital rules was called for, and in June 1999 the Committee sought comment on an initial draft of a proposal to replace the 1988 Accord with a more risk-sensitive framework.
Before work commenced in earnest on the proposed new Accord, the Basel Committee adopted five objectives to guide its efforts.
- First, any new capital rule should at least maintain the current overall level of capital in the banking system.
- Second, it should promote competitive equality and a level playing field for international banks.
- Third, it should take a comprehensive approach to addressing risks.
- Fourth, its approach to capital adequacy should be appropriately sensitive to the degree of risk inherent in a bank's positions and activities.
- And, finally, the Committee agreed that while a new capital rule should focus on internationally active banks, its underlying principles should be suitable for application to banks of varying levels of complexity and sophistication.
I don't think any of us anticipated how challenging it would be to translate these principles into a new Accord that would gain acceptance from the many constituencies. But it didn't take us long to find out.
The June 1999 consultative draft was criticized by the industry as being too simplistic and placing too much reliance on external credit rating agencies for establishing risk weights for commercial loans. In response to this criticism, the Committee took on the task of developing a much more comprehensive and risk sensitive capital rule.
It expanded the number of working groups, task forces, and subgroups to address some of the concerns that had been expressed during the initial comment periods. At one time or another, there have been no fewer than 20 of these sub-committees, conducting studies and writing papers on one aspect or another of the new Accord.
Our second consultative paper for a new Accord, however, which was released for comment in January 2001, generated even more controversy. Banks reported that, contrary to the Committee's intent of leaving overall capital levels about the same, the January 2001 proposals would generate significantly higher amounts of required capital.
Even the rating agencies, which might have been expected to applaud a system that would create an even greater demand for their output, worried publicly about conflicts of interest and the possibility that they would be perceived as unduly influenced by the regulators.
In fact, very little of the proposal escaped criticism; but if there was a common theme to much of what we heard, it was that, in seeking to avoid oversimplification, the Committee had swung too far in the opposite direction. Many critics worried that Basel II's three-pillar superstructure — capital, supervisory oversight, and market discipline — was overly complex.
Because of the great differences among countries in legal structures, banking practices, and regulatory approaches, there was considerable skepticism that a single set of detailed, prescriptive rules could ever be expected to apply to all large internationally active banks. An example of the challenges facing the Committee is the difficulty it has had simply agreeing on an acceptable definition of what constitutes "default."
Furthermore, critics argued that the Committee was slipping dangerously into a micro-management role by assuming responsibility for deciding what constituted the only acceptable methodology for risk management — the basis of the proposed Accord's approach to the calculation by banks of their regulatory capital.
The proposed capital charge for operational risk also attracted an enormous amount of attention. Unfortunately, it was almost all negative. Commenters disagreed with the definition of operational risk, with the risk proxies by which it would be measured, and with the amount of capital banks would be expected to set aside for that risk.
And, regarding Pillar 3 — the requirement that banks increase the amount of information they publicly disclose, in order to enhance market discipline — critics pointed out that it was difficult to standardize disclosures across banks and nations, and that the industry would be required to disclose far more information than the market would be able to use.
In response to these criticisms, the Basel Committee has sought to scale back parts of the proposed rule and is recalibrating the formulas by which capital charges will be determined, with the hope that it will avoid a system-wide increase in required capital.
Throughout this difficult period, Committee Chairman Bill McDonough of the New York Fed has been a source of pragmatic good judgment. He's been an effective spokesman for the Committee and its work; he's been open and honest in addressing the limitations of what we've produced; and he's been totally committed to producing an Accord that works, however much time it takes to get there.
When the original deadlines for comment, publication, and implementation of the new Accord were no longer realistic, Bill was quick to recognize the need to extend them. As a result, we're now looking at the possibility that the new Accord won't be finalized until 2003, with an implementation date of the end of 2006 or perhaps even later.
To be sure, all of the energy expended by the Committee and innumerable staff members from member countries, working with the banking industry, has produced important positive results. The Committee has improved the approach to operational risk, lowering the proposed charge from 20 percent of total regulatory capital to 12 percent, and it may go lower still.
And it's laid the groundwork for a more sophisticated approach to op risk — the so-called Advanced Measurement Approach, or AMA, which would look more to a bank's internal assessments, as with credit risk. We've been concerned that the attractiveness of this approach would be reduced if a "floor" were imposed, and we will continue to argue against any floor on the op risk charge. The Committee has also cut back significantly on the volume of disclosure that would be required under Pillar 3, and, in the area of retail credit, staff has issued a working paper on alternative measures to gauge unexpected losses.
But the question is whether this will be enough. All the criticism and delay have cast a cloud over the Basel process, causing some to speculate whether the Committee will be able to produce a generally acceptable product. We have already been through two elaborate consultative packages; now there will be a third.
But as Chairman McDonough has made clear, the Committee won't issue the next consultative package (CP 3) until it has thoroughly road-tested the Committee's revisions to the January 2001 proposal with the banking industry. This fall, therefore, the Committee will be undertaking a Quantitative Impact Study of the revised Accord to assess its overall impact on banks and the banking system.
Like its predecessors, the next QIS will ask banks to apply the Basel proposals hypothetically in their institutions, as a work-in-progress, and to calculate for the Committee the capital charges that would result. The Committee has announced that QIS 3 will begin in October of this year, with the results due back to the Committee by early next year. The Committee will carefully consider the results, with the hope of issuing a revised draft Accord during the first half of 2003.
Because it's so important that QIS 3 provide a complete and accurate picture of the expected results of a new Accord, just filling some of the still-missing pieces of the new framework by October will be a significant challenge. In this process, the Committee remains especially focused on three issues:
- Balancing the need for a risk-sensitive Accord with the need to be sufficiently clear and flexible that banks can use it effectively;
- Ensuring that the Accord leads to an appropriate treatment of credit to small- and mediumsized enterprises, which are important for economic growth and job creation; and
- Finalizing calibration of the minimum capital requirements to bring about a level of capital that, on average, is approximately equal to the requirements of the present Basel Accord and that includes an incentive for banks to develop risk-sensitive internal ratings-based systems.
While much of this work is very impressive, to some degree it skirts the fundamental issues that continue to lead some observers to question whether we will ever achieve a workable Accord that is true to the five principles I mentioned at the outset of my remarks. Let me now take up some of those issues.
The first is complexity.
Even the proposed Accord's staunchest defenders concede that the proposal is exceedingly complex. But they argue that the new rules need to be complicated to address the complexity of the institutions that apply them. And they are quick to point out that hundreds of pages of the Accord's rules will apply exclusively to the biggest, most sophisticated banks; for non-complex institutions much of that detail will be irrelevant. Nonetheless, I continue to be concerned about the proposal's complexity. The January 2001 proposal stood at a daunting 500 pages, chock full of highly technical language and arcane mathematical formulations. And even that proposal was incomplete. Given the importance of getting banks' capital requirements right, bankers, examiners, legislators, and policy makers need to understand the structure and content of the new rules, and I believe that the sheer size of the capital proposal has been an obstacle to clear understanding and meaningful commentary. When the Committee issues what is anticipated to be its final capital proposal next year, it should strive to set forth a reasonably concise set of black-letter rules that lay out the structure of the new Accord in readily understandable form, with such elaborating detail as is absolutely necessary confined to annexes. We shouldn't attempt to draft language to cover every conceivable contingency that might arise. I'll comment later about the implications of overcomplexity for competitive equality. We in the U.S. must also keep in mind that before the new Accord can become effective for our banks, we'll have to go through a formal notice-and-comment rulemaking process. I am not being entirely facetious when I observe that complying with our "plain English" rule for administrative rulemaking may be a formidable challenge if the proposal is not significantly simplified.
A second issue is the Accord's approach to regulatory capital for the largest and most sophisticated institutions. As you know, the new rules would reward banks that have developed the most advanced internal risk ratings systems by allowing them to use those systems in the calculation of their capital requirements — the so-called "internal ratings-based approach," or IRB. Banks with less developed capabilities would have a somewhat less complex methodology, while banks with basic risk management systems would utilize risk weights and capital charges set by the Committee under a standardized approach. Capital charges would be calibrated to provide incentives for banks to make the investment required to put in place more advanced risk management systems, and thus to move from the standardized approach ultimately to the advanced IRB approach.
The concept is sound enough. But as the Committee has worked with banks over the past 18 months, it's become clear that risk management systems are still evolving in the private sector. We need to be cautious that Basel II does not stultify private-sector innovation by forcing banks to invest prematurely in a single government-dictated approach that may not reflect the best practices that might otherwise evolve.
That was a message that came across loud and clear from the IRB pilot conducted by an interagency team of U.S. examiners this past year. The pilot targeted a group of banks assumed to be most likely to become eligible for the advanced IRB approach, with the objective of gathering information about how closely the approach to risk management being followed by banks matched the framework proposed under the draft Accord.
Although the interagency team hasn't yet completed its review, its findings to date should give us pause. What we've found is that while the science of risk management has been improving at U.S. institutions — progress attributable in part to the spur provided by the Basel process — the current state of the art doesn't yet meet the standard proposed by the Basel Committee. For example, Basel II appears to set out higher expectations for the completeness and independence of the internal ratings process than the processes at our largest banks currently reflect.
Moreover, while many banks already employ such concepts as probability of default and loss given default, we'll also need to make sure that those concepts are being applied in a consistent and reasonably accurate manner. The systems banks use to monitor, analyze, and evaluate the effectiveness of their own risk ratings systems, while generally appropriate for the internal management purposes for which they were designed, are not yet adequate for supervisors to rely on for determining regulatory capital requirements. Additional control mechanisms may be needed to assure bank management — and us — that there is integrity to the capital calculations that emerge from their methodologies.
Another issue is the availability of historical data related to losses. Most banks seem to know that there is a data issue. However, it is not clear that banks know the exact nature and dimension of the issue. Many banks are worried about this requirement, fearing that we are asking them to build data systems that will be too costly. They're asking what the bare minimum is in terms of what they need to do to comply.
This suggests that banks may not understand that they will probably have to spend much more on data collection than they are currently contemplating. At a minimum, banks will need to have data systems that allow them to "backtest" their ratings assignments.
This sobering reality has significant implications for the Basel process. It suggests that the new Accord may end up imposing on banks the Committee's own conception of what an effective risk management program should look like. Indeed, I've speculated that we could even get to the point where banks may simply throw up their hands and ask the regulators to prescribe a uniform risk-rating matrix for all banks.
The danger of our taking on such as role is significant. In fact, one important predicate of the IRB approach is that banks will use the same risk rating systems they use in the everyday management of their business, not some system designed just to satisfy Basel II. Our goal should be to make certain that the new capital framework reflects the best contemporary practice. We must be very careful to avoid micro-managing the institutions that we supervise.
But by imposing our own views on the industry, even with the best intentions, the Committee could be discouraging innovation in the marketplace — innovation that could well produce results superior to an approach designed by regulators. The last thing we want to do is to force the industry to invest in systems that will soon be incompatible with future market-driven developments. The new Accord needs to be flexible.
The Level Playing Field is another issue.
The principle of competitive equality and a level playing field for international banks is an admirable one, and an appropriate goal of the Committee's efforts. Yet one must question whether the exceedingly complex and highly prescriptive approach to capital reflected in the Committee's work to date will really foster competitive equality.
I recognize that there's an element of irony in this, because it was precisely with the view of promoting greater international uniformity that a prescriptive capital rule was adopted in the first place. The assumption was that differences in national supervision — differences in supervisory philosophy and in the availability of supervisory resources — would lead to divergent implementation if supervisors were left to their own discretion.
This same philosophy is prevalent in Basel II, where it's assumed that highly prescriptive and detailed rules, along the line of the Committee's proposals, will not only promote uniformity, but will make it more difficult for supervisors to promote the competitive interests of their own banks through a very permissive and discretionary application of general principles.
But global rules, no matter how carefully weighed and measured, are not a satisfactory substitute for judgment — especially in a field like financial risk management, where the state of the art is constantly in flux. No set of rules can be made to apply equally well to all conceivable situations — indeed, by definition, all contingencies cannot be foreseen or provided for. That's why the true test of any surgeon's skill is not in the way he deals with the textbook case, but rather in the way he responds to the unexpected, when there's no textbook to follow. The same thing is true of examiners. Judgment — and the ability to make individualized decisions — are crucial. We must not tie examiners' hands with inflexible rules.
Moreover, the somewhat simplistic notion that competitive equality across countries can be achieved, and permissive local favoritism avoided, by highly prescriptive rules fails to take into account the vast differences among countries in the supervisory apparatus through which the rules would be applied.
In the United States, we have a highly developed — some say intrusive — system of bank supervision. For example, the OCC has full-time teams of resident examiners on site at our largest banks — as many as 40 or 50 examiners at the very largest. In addition, most U.S. institutions are also subject to holding company supervision by the Federal Reserve, and in some cases by the FDIC and state supervisors. In other countries, by contrast, supervision may rely less on bank examiners as we know them, and more on outside auditors to perform certain oversight functions.
Given such disparities in the methods of supervision, I submit that U.S. banks are more likely to be subjected to more vigorous enforcement, and less likely to be the beneficiaries of permissive exceptions, than banks in countries whose supervisory practices fall at the other end of the spectrum — although I must say that when some of our non-U.S. colleagues on the Committee express fears about discretionary supervision, they sometimes seem to be pointing the finger at us — quite unjustifiably, I believe.
We already have one early sign that fears about a highly discretionary application of the new rules should be aimed offshore. In our IRB implementation pilot we have found that even our largest and best-managed banks have some distance to cover to meet the proposed Basel standards. Yet some of our foreign counterparts on the Basel Committee are quite confidently telling us that their banks are ready to go. I have some difficulty concluding that these judgments are based on as exacting an inquiry as we have been making. Needless to say, this doesn't bode well for the level playing field.
Last but not least, there's the Basel provision for operational risk.
In my view, operational risk should be defined as the risk that's directly related to the quality of a bank's internal controls. Thus, two banks engaged in an identical line of business may present vastly different quantities of operational risk when the internal control systems of one are significantly better than those of the other.
A one-size-fits-all approach to operational risk — such as a formulaic capital charge based on some percentage of gross revenues or a percentage of the charge for credit risk — while simple to apply, would disadvantage the best managed banks and provide undeserved advantage to the worst managed. Worst of all, it would provide no incentive to improve internal control systems.
For this reason I've repeatedly argued that operational risk is a subject peculiarly appropriate for assessment under the Pillar 2 approach — an approach that relies on supervisory analysis rather than numeric formulas. But there are many on the Committee who are very cautious about such a use of Pillar 2, believing it would be used by supervisors to provide competitive advantages to their banks and thus to undermine the "level playing field." Those holding this view find strong comfort — erroneously, I believe — in highly detailed prescriptive rules.
In the interests of breaking the logjam, U.S. representatives to the Basel discussions on op risk have developed an alternative methodology that we intend to propose for inclusion in the next consultative paper. That methodology contains two key elements. First, we believe that an operational risk capital charge under the Advanced Measurement Approach should be based on a bank's own internal risk management system, subject to eligibility criteria and reasonableness checks that would be evaluated through the supervisory process.
Second, we believe that there should be no separate capital floor for op risk. Instead, our proposal calls for a temporary overall floor for both credit and operational risk based on the total amount of regulatory capital required under the current Accord — a floor amount that we believe should be phased out after two years.
As you're well aware, the op risk issue has been a lightning rod for criticism of the Basel process generally. We believe that our approach offers hope of getting that whole process back on track. If it does, I continue to believe that the benefits will be substantial.
It's been a long and arduous road to revising the Basel Accord, and we're obviously still a long way from home. But I don't think that any of us regrets having taken the journey.
Regardless of the outcome, the steps we've already taken have advanced the causes of international cooperation, supervisory competence, effective risk management, and safety and soundness in the global financial system. Whether we ultimately get a new Accord or not, I think you'll agree that these are considerable accomplishments.
Let me once again tip my hat to RMA for its fine contributions to those ends.