Remarks by Eugene A. Ludwig Comptroller of the Currency before the American Enterprise Institute

I am delighted to join this distinguished panel as we launch another phase of the public debate over ways to keep our financial system strong. It is not surprising that our panel should convene under the auspices of the American Enterprise Institute, which, under the leadership of Chris DeMuth, has sponsored some of the most significant work I know of on the most urgent policy questions of our day.

When financial regulators get together these days, umbrella supervision is one of the things that they are likely to talk about. One cannot avoid being struck by the apparent disconnect between a fragmented financial regulatory structure and the increasing consolidation of the financial services industry we regulate. The financial system is increasingly dominated by very large, highly diversified, global firms that do not stop at traditional sectoral boundaries, but engage in every type of financial business, from commercial lending to investment banking to insurance underwriting. These firms manage their risks globally, rather than looking at risks in individual entities within the company. Yet these companies are typically supervised by multiple functional regulators, each responsible for only a piece of the firm.

It is this juxtaposition of geographically diverse companies offering multiple new products being supervised by an array of functional regulators that leads many to call for some level of coordination or group oversight. In response to these calls, some have articulated the idea of "umbrella supervision."

The metaphor of the umbrella has obvious appeal. It evokes safety and security. But we need to move beyond imagery to specifics. What exactly do we mean by umbrella supervision? What role do we envisage for the umbrella supervisor? What entities would it oversee? And how would umbrella supervision differ from other proposals that have been put forward for improving the supervision of diversified financial firms?

I will organize my remarks by examining three basic approaches to the supervision of global financial entities — approaches that entail widely divergent costs and benefits. Each of these arguably could be what is meant by umbrella supervision. The first two of these approaches could be viewed as a species of umbrella supervision through some may see it otherwise. To my mind, however, they satisfy to a considerable degree what is most appealing about the umbrella concept.

The first approach is "supervision from a group-wide perspective." Its premise is that regulators cannot adequately supervise global, diversified financial firms unless they see all of the firm's risk exposures across all its legal entities, and unless they are in a position to assess the risk management systems of the firm as a whole.

If the current system of bank supervision did not already embody this group-wide perspective, I would certainly agree that the system needed to be changed. But group-wide supervision is, of course, one of the foundations of bank supervision in the United States, as it is in all of the G-10 countries. This goes back at least to 1979, when the Basle Committee on Banking Supervision adopted the following principle:

"It should be a basic principle of banking supervision that the authorities responsible for carrying it out cannot be fully satisfied about the soundness of individual banks unless they are in a position to examine the totality of each bank's business worldwide...All parent supervisory authorities should, within the context of their own systems and present circumstances, be required to give effect to the agreed principle that the capital adequacy and the risk exposure of all their banks be examined and assessed on the basis of the totality of their international activities."

In practice, consolidated supervision of U.S. banks is typically carried out at the level of the dominant bank within the firm. There is an obvious reason for this: the dominant bank typically accounts for the vast majority of the firm's total assets. In the case of large national banking companies, non-bank assets represent only about 10 percent of the consolidated total assets. Consequently, the primary bank supervisor — whoever that may be — possesses the most extensive knowledge of the bulk of the firm's activities.

But the primary supervisor does not limit its risk analysis to the bank, but instead often looks beyond the bank to the activities of non-bank subsidiaries and affiliates. It also assesses all the risks borne by the bank, including those risks that originate in non-bank subsidiaries and affiliates; and it assesses the adequacy of the firm's overall risk management systems, even when those systems extend beyond legal entity lines. I agree that we may need more of this kind of group-wide perspective to deal with increasingly complex conglomerates. But providing this broad perspective is well within the bank supervisor's reach, where the bank itself is the focal point for the company's risk management systems. Therefore, the bank supervisor can obtain necessary information from the bank and its affiliates, and also can verify transactions flowing between the bank and its affiliates. For some companies, this assessment will be complemented by information obtained from other functional supervisors.

The upshot is this: the primary supervisor seems well situated to undertake whatever may be needed in the way of supervision from a group-wide perspective. One must question what would be added by creating another layer of regulation in the form of an additional type of umbrella supervisor. For, if the primary bank supervisor is doing its job, it is already taking into account the risks that non-bank affiliates may pose to the bank.

The second "umbrella" approach to improving the supervision of diversified financial firms is the one that, until recently, had received the most international attention. It focuses on information-sharing, using as its vehicle what the Joint Forum on Financial Conglomerates (of which I am a member) referred to as a "convenor." The role of the convenor would be to facilitate communication and information-sharing among the various functional and national supervisors of a diversified financial firm. The convenor would gather information from functional regulators about the entities that they supervise, and disseminate that information to the other functional regulators. It would also have the authority to convene meetings among functional supervisors to deal with emergencies as well as routine situations.

This approach has the considerable virtue of allowing existing functional regulators — each expert in its own area of supervision — to continue to do their jobs with a minimum of interference or disruption. Depending upon how much information the functional supervisors are asked to provide, it would involve a minimum of additional burden. And it could well help functional regulators assess risk, by providing them with additional information and a point of contact for use when troubles arise.

I am certainly in favor of developing better mechanisms for information-sharing between supervisors. But we need to be clear on the limitations of this approach, and be aware of the pitfalls that we need to avoid.

First, periodically gathering, collating, and distributing information, particularly for a multinational firm with numerous legal entities supervised by numerous supervisors in different countries, is likely to impose a great deal more cost than might initially be apparent. I remember my own surprise at learning what it would cost respondents if we added just a few additional items to the "Call Report," the federal government's uniform report for banks.

Different entities gather information in different ways, often to match their own risk management, public disclosure, and accounting needs. Asking these entities for information often involves requiring them to undertake new information collections themselves, and this undertaking can be quite costly. Even organizing existing information in a new way can involve costs that are hardly trivial. We should be particularly cautious about imposing these costs when the benefits of collecting and disclosing the information are not clear.

And we have to wonder just how much value a convenor would add, particularly since information exchanges between supervisors can and do take place already on a bilateral basis. Indeed, it is not clear that we would make it easier for supervisors of the different parts of a banking group to share information and to communicate with each other by requiring that all these communications pass through a third party. These arrangements tend to reduce information sharing to a routine: a fixed set of information that supervisors exchange at set intervals. Routinized information sharing could well be both more costly and less useful than improving bilateral communication channels so that a supervisor facing a particular problem can more easily ask his counterpart for the precise piece of information that he needs.

The value of a convenor is even more questionable if we are talking only about public information, which supervisors ought to be able to obtain without the help of a convenor. If, on the other hand, we are talking about confidential information that will be gathered and disseminated to numerous regulators worldwide, this raises a confidentiality issue of no small proportions. A leak occasioned by a wide distribution of such information may actually increase risk, particularly in times of stress.

Let me be clear: I believe that we do need to improve information-sharing among supervisors. But we must be careful to avoid setting up unwieldy formal arrangements that cost more to operate than they can deliver in benefits, or that serve to impede rather than to facilitate supervisory cooperation. My conclusion is that we might do better to concentrate our efforts on improving coordination, rather than on designating coordinators. To my mind, the first steps we take should be to strengthen bilateral communication channels, probably starting with information-sharing arrangements in emergencies, where the need is greatest and the ongoing cost is least.

In seeking the benefits of coordination, it is important that we proceed with some care. There is genuine risk that precipitous action to achieve the worthwhile goal of improved coordination would actually decrease safety and soundness. For regulators, as for doctors, the first rule should be: "First, do no harm."

So far, I have talked about two "umbrella" or top-down approaches to improving supervision of large, diversified financial firms: first, relying on the primary supervisor to employ a group-wide perspective in its supervision and, second, mechanisms for information-sharing.

A third approach, which some equate with umbrella supervision, is the creation of an overseer — an authority distinct from and above the functional supervisors, including the primary supervisor. This "full scale" umbrella supervisor would have full responsibility for the entire firm.

Unlike the first two approaches, which, if properly implemented, could improve the supervision of global financial firms, full scale umbrella supervision of banking firms seems to me to have more costs than benefits.

My greatest objection to this formulation of umbrella supervision is that it adds an additional layer of regulation to what is already a heavily supervised sector. And what exactly does this additional layer contribute to the effectiveness of bank supervision? Not a group-wide perspective: as I mentioned earlier, the primary supervisors are ideally situated to practice supervision from a group-wide perspective. There seems little to gain from transferring responsibility for consolidated supervision to an umbrella supervisor who is farther removed from the bulk of the banking firm's activities.

Nor will an additional layer of supervision necessarily improve information sharing. And the benefits of information sharing can be gained by improving cooperation between functional supervisors, without creating an additional layer of regulation. It is unclear what else a full scale umbrella supervisor might add to the supervision of global financial firms.

Moreover, beyond the fact that the benefits are questionable, it is likely that such umbrella supervision will carry a high price. Redundant layers of regulation increase both the direct budgetary cost of regulation and the burden that regulation imposes on the regulated entities. Multiple layers also tend to slow supervisory decision-making, and —what is worse — to blur accountability for supervisory decisions. Functional supervisors may be tempted to wait for word from their umbrella supervisor before acting — or not act at all — since the umbrella supervisor is likely to share the ultimate responsibility for any problems.

Full scale umbrella supervision would also tend to be seen as extending the federal safety net to non-banks — for example, deposit insurance guarantees, access to the discount window, daylight Fedwire overdrafts, and even the concept of too-big-to-fail — with all of the accompanying moral hazard problems.

As a strong believer in the free market and free market solutions, I have to wonder whether creating a new regulator — a full scale umbrella supervisor — is the best way to improve coordination. Would the benefits of this kind of umbrella supervision outweigh its costs and risks, including the potential for a perceived extension of the federal safety net to an entire financial conglomerate? There is a growing consensus that we should be moving in the opposite direction: limiting the scope of the safety net and relying to a greater degree on transparency, disclosure, and market discipline to help ensure the strength of the financial system.

I have just outlined three types of oversight for financial conglomerates. One or all of these arguably could be called umbrella supervision. All three approaches involve both benefits and burdens. It is critically important that we study the issue carefully to ensure that the trade-off makes sense.

In this regard, let me suggest three important principles that should guide our efforts to enhance coordination in the supervision of financial conglomerates:

• First, as I mentioned earlier, we should build on the expertise of functional supervisors through bilateral information-sharing arrangements; the primary emphasis should be on coordination, not coordinators.
• Second, we should strive to maximize efficiency and to minimize duplication and burden. Duplication and burden impose severe costs not just on the financial system as a whole, but on individual consumers in terms of pricing and availability of products and services.
• Third, we should recognize that, whatever model of coordination we choose, some additional burden will result. If we decide that this burden is justified, we should make sure that we apply it as equitably as possible to all similarly situated entities so that we do not distort the playing field in a way that may itself create safety and soundness concerns.

I would like to thank AEI and Chris — and my colleagues on our panel — for helping us move the discussion forward.