Board of Governors of the Federal Reserve System

It's a great pleasure to be with you today at the ABA Banking Law Committee's annual meeting. I left the practice of law—and immersion in the company of lawyers—closing in on 20 years ago now, but there have been many times during my long sojourn among businessmen and economists that I have reflected with fondness and some nostalgia on the famous adage of Harrison Tweed (the "Tweed" of Milbank, Tweed, a reformer of the bar, the "most democratic of aristocrats," and the last man to unironically wear a cape in the lobby of the Chase Manhattan Plaza) which most of you can no doubt recite by heart: "I have a high opinion of lawyers. With all their faults, they stack up well against those in every other occupation or profession. They are better to work with or play with or fight with or drink with than most other varieties of mankind." Speaking here feels a lot like coming home.

This afternoon, I would like to talk with you about the outwardly mundane but increasingly consequential topic of bank supervision. Twenty years ago, when I would have been among your number at this meeting, this would have been my cue to pull out my Blackberry and start checking my emails. The structure and content of regulation was both intellectually interesting and professionally meaningful; I considered bank supervision, by contrast, as both too workaday and too straightforward to merit the commitment of much legal horsepower or personal attention. I could perhaps have been excused by the callowness of youth, yet it was a common view at the time. Having now been immersed for the last two years both in the practice of supervision and in the complementary relationship between the regulatory and supervisory processes, I realize that this wasn't true then, and is certainly not true now. It is not a drafting accident that the Dodd-Frank Act gave my position at the Federal Reserve the title of Vice Chairman for Supervision. Notwithstanding the extensive reform of bank regulation after the crisis, which has had much consequence for the industry (most of it salutary) it is the process of examination and supervision that constitutes the bulk of our ongoing engagement with the industry and through which our policy objectives are given effect.

This division of labor is important for lawyers and policymakers to think about deeply because the processes of regulation and supervision are necessarily different in crucial respects. Regulation establishes a binding public framework implementing relevant statutory imperatives. Because a rule is designed to apply generally, rules must be based on general principles intended to achieve general aims, rather than reverse-engineered to generate specific effects for specific institutions. Given their general applicability, there must be a general process for all those with an interest—industry, academics, citizens, Congress—to have notice of, and opportunity to comment on all rules, ensuring that all potential effects and points of view are taken into account in the rule's crafting. And given their general function, rules must be clear and public: Those affected must know what to expect and what is expected.

Supervision, by contrast, implements the regulatory framework through close engagement with the particular facts about particular firms: their individual capital and liquidity positions, the diverse composition of their distinct portfolios of assets, their business strategies, the nature of their operations, the strengths and weaknesses of their management. Much of the granular information used by supervisors is, accordingly, proprietary and confidential, and many of their judgments and decisions are closely tailored to specific circumstances.

Given the strong public interest in the safe, sound, and efficient operation of the financial industry and the potential for hair-raising and widespread adverse social consequences of private misjudgment or misconduct in that industry, close and regular supervision of this sort can help us all sleep restfully. Yet, the confidential and tailored nature of supervision sits uncomfortably with the responsibilities of government in a democracy. In the United States, we have a long-standing, well-articulated framework for ensuring that regulations conform with the principles of generality, predictability, publicity, and consultation described above. Supervision—for good reason, in my view—is not subject to this formal framework. But it is currently not subject to any specific process constraint promoting publicity or universality. This leaves it open to the charge, and sometimes to the fact, of capriciousness, unaccountability, unequal application, and excessive burden.

Here, then, is a conundrum. We have a public interest in a confidential, tailored, rapid-acting and closely informed system of bank supervision. And we have a public interest in all governmental processes being fair, predictable, efficient, and accountable. How do we square this circle? In my time with you today, we will not do more than scratch the surface of this question. It is a complex and consequential issue that, for decades now, has received far too little attention from practitioners, academics, policymakers and the public. Evaluating this question will be a significant focus of mine going forward, and I hope that there will be much discussion in many fora from which we at the Fed, and at other regulators, can learn. So today, I simply want to open the exploration of some these conceptual issues, and then offer some specific suggestions—by no means comprehensive—on some obvious and immediate ways that supervision can become more transparent, efficient, and effective.

The Importance of Transparency
Let me begin by delving a little more deeply into the distinction between regulation and supervision and the process applicable to both. In delegating to agencies such as the Fed the significant power to write regulations, Congress has codified a regulatory process that emphasizes transparency. This process was born in the 1930s, in the tumult of government expansion that was the New Deal, when Congress began a decade-long debate over how to manage the new regulatory state. The result, the Administrative Procedure Act (APA), was, I should note, developed with the active involvement of the American Bar Association. The APA continues to serve as the basis for the public disclosure and participation required for agency rule-writing and for the judicial review affected parties are guaranteed to challenge rules.

This transparency is intended to prevent arbitrary, capricious, and thus ineffective regulation by inviting broad public participation and mandating a deliberate public debate over the content of proposed rules. One obvious purpose of this transparency is to provide clarity and predictability: it helps make clear how agencies are considering exercising their discretion. The significant process protections in laws such as the APA are also meant to ensure fairness. The wisdom behind this approach is that fairness both helps bring forth more considered and effective regulations and builds respect for and adherence to the law, which is essential for enforcement. Transparency is central to our ability to assert that our rules are fair.

Not everything that government does, however, can be accomplished in exactly the same way that regulations are written. One of these things is bank supervision.

Bank Supervision
Banks are subjected to supervision, in addition to regulation, as an additional form of government oversight because of their complexity, opacity, vulnerability to runs, and indispensable role in the economy, enabling payments, transmitting monetary policy, and providing credit. The government provides a safety net to banks in the form of deposit insurance, and in return, banks are subject to government oversight that mimics some of the monitoring that the private sector would provide, absent the government safety net. The bank regulatory framework sets the core architectural requirements for the banking system, but it isn't enough to set the rules and walk away like Voltaire's god. The potential consequences of disruption in the financial system are so far-reaching, and the erosion of market discipline resulting from the government safety net sufficiently material, that it is neither safe nor reasonable to rely entirely on after-the-fact enforcement to ensure regulatory compliance. Supervisors are in a good position to monitor individual firms' idiosyncratic risks. And in addition to what they do at individual banks, supervisors monitor for risk that may be building among clusters of banks or across the banking system. These "horizontal" exams across multiple banks help highlight new or emerging risks and help examiners understand how banks are managing these risks.

Through their engagement with banks, supervisors promote good risk management and thus help banks preemptively avert excessive risk taking that would be costly and inefficient to correct after the fact. Where banks fall materially out of compliance with a regulatory framework or act in a manner that poses a threat to their safety and soundness, supervisors can act rapidly to address the failures that led to the lack of compliance or threat to safety and soundness.

This is a crucial point: Supervision is most effective when expectations are clear and supervision promotes an approach to risk management that deters bad behavior and decisions by banks. Clearly communicating those expectations is essential to effective supervision, and in a larger sense, clear two-way communication is the essence of effective supervision. Supervisors rely on banks to be frank and forthcoming, and supervisors in turn can help secure that frankness by explaining what their expectations are and why their expectations are reasonable, not arbitrary or capricious. Greater transparency in supervision about the content of our expectations and about how we form our expectations and judgments can make supervision more effective by building trust and respect for the fairness and rationality of supervision.

I don't believe the Federal Reserve has communicated as clearly as it could with the banks we supervise. More transparency and more clarity about what we want to achieve as supervisors and how we approach our work will improve supervision, and I have several specific proposals.

Broadly speaking, these actions fall into three categories: (1) large bank supervision, (2) transparency improvements, and (3) overall supervisory process improvements.

Large Bank Supervision
Last fall, we completed a cornerstone of the recent banking legislation to tailor our rules for regional banks. This was entirely consistent with a principle at the heart of our existing work: Firms that pose greater risks should meet higher standards and receive more scrutiny. Our previous rules relied heavily on a firm's total assets as a proxy for these risks and for the costs the financial system would incur if a firm failed. This simple asset proxy was clear and critical, rough and ready, but neither risk sensitive nor complete. Our new rules employ a broader set of indicators, like short-term wholesale funding and off-balance-sheet exposures, to assess the need for greater supervisory scrutiny.

That said, the composition of our supervisory portfolios has not yet been aligned with our recent tailoring rules. For example, the Large Institution Supervision Coordinating Committee (LISCC) portfolio includes all Category I firms, which have the greatest risk profile, along with certain Category II and Category III firms, which are less systemic. Other Category II and Category III firms, on the other hand, are supervised under our large and foreign banking organizations (LFBO) portfolio.

Since the crisis, we have been giving significant thought to the composition of our supervisory portfolios, and, in particular to whether and how we should address the significant decrease in size and risk profile of the foreign firms in the LISCC portfolio over the past decade. Because of these changes, which I will describe in more detail momentarily, I believe there is a compelling justification to make changes today to the composition of the foreign banks in the LISCC portfolio.

Separately and in keeping with the goal of transparency, I think it is important that all the Fed's supervisory portfolios have a clear and transparent definition. Today nearly all of our supervisory portfolios have such a crisp and clear formulaic definition specified in the public domain, but the LISCC portfolio does not. My goal is to develop, prospectively, a clear and transparent standard for identifying LISCC firms. My preferred approach for achieving this objective would be to align the LISCC portfolio with our recent tailoring categorizations. I believe we should draw the LISCC line to coincide with Category I. The justification for this line-drawing is that Category I firms pose the most systemic risk and require the most supervisory attention. In this state of the world, Category II and III firms would remain subject to heightened supervisory standards that are commensurate with their risk profile.

Allow me to draw out what this approach could mean for the foreign banks that currently are in the LISCC portfolio. Since 2010, these four banks have significantly shrunk their U.S. footprint, and their U.S. operations are much less risky than they used to be. Since 2008, the size of the LISCC FBOs' combined U.S. assets has shrunk by about 50 percent, and they have reduced the assets at their broker-dealers from a peak of $1.9 trillion in 2008 to $340 billion today, a reduction of over 80%. In addition, the estimated systemic impact of the LISCC FBOs today is much smaller than the U.S. GSIBs. The average method 1 GSIB score of the combined U.S. operations of the LISCC FBOs is less than a quarter of the average GSIB score of the six non-processing U.S. GSIBs.

Thus, if any foreign banks move out of the LISCC portfolio based on this de-risking, they would move into the LFBO portfolio, where they would be supervised alongside other foreign and domestic firms with similar risk profiles. Notably, this change in supervisory portfolio would have no effect on the regulatory capital or liquidity requirements that currently apply to the four LISCC FBOs. Similarly, the change would not result in a loss of insight into the activities of these firms.

In the same spirit, I think we should consider publishing the internal procedural materials that the Fed uses to supervise the LISCC firms, sometimes referred to as the Program Manual. The Manual contains a description of the main supervisory processes for identifying risks and our approach for addressing them. Publishing the Manual would help the public and the banks better understand why we take the actions that we take as supervisors and would demystify some of our processes. If we took these two simple steps—defining LISCC firms and publishing the Program Manual that governs our supervisory approach—it would go a long way in helping to make our supervisory practices more understandable and accessible without undermining supervisory effectiveness.

Let me now turn to the ratings framework that applies to all large holding companies. A firm's supervisory rating, which is confidential, is important because it affects things such as the firm's ability to engage in mergers and acquisitions and to enter new lines of business. Just over a year ago, the Board began implementing a new ratings framework for large holding companies called the large financial institutions (LFI) ratings framework. The LFI ratings framework focuses on three components of a firm's operations: capital, liquidity, and governance and controls. We inaugurated the LFI ratings framework for LISCC firms in January 2019 and for other large holding companies at the beginning of this year.

As we gain more experience with LFI, we will be paying close attention to how the new rating system is working and whether it is achieving its intended purpose. There are two features of the ratings system that I will be particularly interested to monitor, and which may well require adjustment. These are the embedding of qualitative "risk management" standards in the capital and liquidity components of the ratings (as opposed to standardized quantitative measures of capital and liquidity adequacy) and the ascetic principle by which a firm's "well managed" status is determined by its lowest component rating, no matter how good the bank is at everything else.

Regarding our stress tests under the Comprehensive Capital Analysis and Review (CCAR), I continue to look for ways to make the tests more transparent without making them game-able and without diluting their potency as a supervisory tool. I will mention three of these transparency-enhancing ideas. First, I expect that we will continue to provide more transparency on the models used in CCAR. We started providing improved transparency on models last year, and as I have previously said, we will remain on that path until we have released substantial details on all of our key models. We also continue to consider ways to increase the transparency around the scenarios we use in CCAR, including, for example, by modifying our scenario design policy statement to provide greater transparency on the design of the global market shock component of the stress tests.

Second, I expect that as part of the stress capital buffer, we will give banks significantly more time to review their stress test results and understand their capital requirements before we demand their final capital plan. Firms are currently permitted to revise and resubmit their capital plans after receiving their stress test results. But it is done on a short timeframe, and allowing additional time would produce better results without in any way reducing the stringency of the stress tests. Fundamentally, I think banks will be better able to do intelligent capital planning if we provide them with their complete set of regulatory capital requirements before we require submission of a capital plan.

Third and finally, we continue to look for ways to reduce the volatility of stress-test requirements from year to year. We are considering a number of options, such as averaging outcomes over multiple years or averaging the results of the current year's stress test with the results of one or more previous years. Again, the goal here is not to make the tests less strenuous but to give banks a greater opportunity to plan for them and to meet our expectations ex ante rather than through an ex post remedial process.

Transparency Improvements
The next three actions I'm proposing also relate to improved transparency, and they would improve our processes for supervising all banks. The first would be to create a word-searchable database on the Board's website with the historical interpretations by the Board and its staff of all significant rules. Regulatory interpretations by Board staff have grown piecemeal over the decades and haven't consistently been treated as the valuable resource they are. The Board's website has select interpretations of many laws but does not provide a comprehensive, user-friendly collection of regulatory interpretations, FAQs, and commentary. This project will require some effort of course, as well as vigilance to keep the interpretations up to date, but I believe that the end result will be well worth it.

The second of these transparency actions would be putting significant supervisory guidance out for public comment. The Board already invites comments on its regulations, as required under the APA, and regularly invites comment on some supervisory guidance and statements of policy. This practice of seeking comment on guidance leads to better, more informed supervision and better engagement by banks. I would like the Board to seek comment on more supervisory guidance going forward.

Third and finally, as another improvement related to guidance, I support submitting significant supervisory guidance to Congress for purposes of the Congressional Review Act. Currently, the Fed does this for rules but not guidance. I support doing so for significant guidance because significant guidance, though nonbinding, can still have a material impact on bank behavior. I believe this step would enhance the Fed's accountability and help build support for supervisory guidance.

Overall Supervisory Process Improvements
The last category of proposals includes five areas of improvement that all relate to what we call the "supervisory process"—how we go about conducting our responsibilities. Like my other suggestions, these are all rooted in common sense with a view toward maintaining firm and fair supervision.

The first is to increase the ability of supervised firms to share Federal Reserve confidential supervisory information (CSI) with employees, affiliates, service providers, and other government agencies to promote greater compliance with laws and facilitate the response to enforcement actions. We have received feedback that our rules can prevent banks from sharing CSI with a wide variety of relevant parties who need to know this information in order to help the bank remediate identified supervisory issues. We issued a proposal last year to address this shortcoming in our CSI rules, and I expect the Board will be able to issue a final CSI rule later this year.

The second process improvement is having the Board adopt a rule on how we use guidance in the supervisory process. I would expect the rule to state that the Board will follow and respect the limits of administrative law in carrying out its supervisory responsibilities. In particular, consistent with the September 2018 interagency statement on guidance, we would affirm the sensible principles that guidance is not binding and "non-compliance" with guidance may not form the basis for an enforcement action (such as a cease-and-desist order) or supervisory criticism (such as a Matter Requiring Attention (MRA)). This rule would be binding on the Board and on all staff of the Federal Reserve System, including bank examiners.

The third and fourth process improvements relate to supervisory communication. The third improvement is to restore the "supervisory observation" category for lesser safety and soundness issues. This approach would provide supervisors with a tool—supervisory recommendations—for continuing to raise concerns about less pressing supervisory matters while focusing a bank's attention on the most urgent matters, those that would receive MRAs. We removed this category of supervisory commentary in 2013 to better focus bank management on deficiencies found during the supervision process. (By way of comparison, both the FDIC and OCC retained this tool.) On reflection, I think there is value in supervisory observations. They allow an examiner to give notice about a supervisory concern even if that concern has not risen to the level of an MRA.

The fourth process improvement would be limiting future MRAs to violations of law, violations of regulation, and material safety and soundness issues. MRAs are supervisory communications that identify areas where banks are out of compliance with applicable legal standards or otherwise are engaged in practices that create substantial safety and soundness risks. MRAs identify the source of the compliance failure, deficiency, or safety and soundness weakness and generally include an expected timeframe for remediation. MRAs are not legally binding and are not enforcement actions.

Nevertheless, MRAs carry weight because they can affect a bank's supervisory rating. In limiting MRAs to legal violations and significant supervisory concerns, we would take care to clearly define the breadth of what constitutes a "material safety and soundness issue." This distinction is important as a matter of fairness. Banks should be able to understand the line between MRAs significant enough to affect the bank's supervisory rating and less significant matters that don't affect a bank's supervisory rating but raise concerns that should be considered by banks. Greater fairness contributes to greater supervisory effectiveness. Together, the third and fourth process improvements would be calibrated to improve communications so that banks can focus on remediating key weaknesses while maintaining awareness of emerging ones. Ultimately, a bank that promptly corrects its material safety and soundness weaknesses will be better able to serve its customers and intermediate credit through a range of scenarios, including under stress.

The final process improvement is to make routine our existing practice of having an independent review of important supervisory communications and guidance documents. We want to make sure that our supervisory communications, including MRAs, focus on violations of law and material safety and soundness issues and that these communications don't mistakenly give the impression that supervisory guidance is binding. We already closely scrutinize MRAs issued to the LISCC firms and in horizontal reviews of other large domestic and foreign banks. This extra scrutiny is a sensible practice that should be regularized and expanded across our supervisory portfolios.

With respect to prospectively assessing future guidance, the key goals here would include reassessing the scope of key guidance documents, removing inappropriate bright lines from guidance, and removing any mandatory language from guidance. I will discuss each of these goals in turn.

As I mentioned, the Board adopted a final tailoring rule last year that adjusted the regulatory standards applicable to banks, based on their risk profile. I think it would be useful for us to review our guidance in light of this tailoring exercise, such as guidance on stress testing and capital planning, and to update the scope of guidance where appropriate.

Regarding bright lines, bright lines tend to carry the implication that the standard they are delineating is binding. For this reason, rules often include bright lines so that it is clear how to stay in compliance with the rule. Putting bright lines in guidance, even when the bright line is phrased as a "should" rather than as a requirement, blurs the line between guidance and rules, and for this reason, it is a practice we should avoid.

For the same reason, it is inappropriate to put mandatory language in guidance. This practice can create the same distortions as the use of bright lines.

Conclusion
Obviously, the incremental changes to our supervisory processes described above do not completely answer the question with which I began my remarks today: How can we square the public interest in agile supervision with the public interest in transparency and accountability? This should be an ongoing question of high priority, both at the Fed and more broadly among those who care about our system of financial regulation. Equally obviously, however, these suggestions would strengthen our practice of supervision and increase the vigor and credibility of our supervisors.

The changes to supervision since the crisis have made the financial system stronger and more resilient than it was before. The incremental changes I have outlined, to increase transparency, accountability, and fairness, would make supervision more efficient and effective, and our financial system stronger and more stable.