The End of Preventive Supervision?

The FDIC and the OCC recently issued a Joint Notice of Proposed Rulemaking (NPR) that would establish a definition of “unsafe or unsound practices” and change the standards for issuing matters requiring attention (MRAs) and matters requiring board attention (MRBAs). While advocates for these changes describe them as promoting a greater focus on core financial risks, they are much more likely to undermine attempts at preventive supervision.

Preventive Supervision

Effective banking regulation depends on preventive supervision. Spotting warning signs and excessive risk taking early allows supervisors to intervene when the bank is still salvageable. Although supervisors assign banks CAMELS ratings from 1 (best) to 5 (worst), a “3” rating often proved to be the most meaningful. Once a bank deteriorated to a “4” or especially a “5,” it was often too late. At that point you functioned less as a primary care physician and more like a coroner.

Preventive supervision may be the objective, but the practice can fall short of the aspiration.  Supervisors may spot reckless, sloppy, and imprudent activity but still be reluctant to crack down on the activity until the bank’s condition shows clear signs of deterioration. Supervisors can certainly expect push back. The CEO of one troubled bank once told a colleague of mine, “Leave me alone; I’m making money.” The bank later failed. Material loss reviews identify cases where supervisors didn’t act quickly or decisively enough. It’s harder to say how many banks would have failed but for the prompt actions of supervisors. Research by John Kandrac and Bernd Schlusche suggests, however, that lighter supervisor standards and fewer supervisors translate into more risk taking and more frequent and severe bank failures.

The NPR

The joint NPR seeks to make some fundamental changes to the nature of bank supervision. First, it would introduce a new threshold for what constitutes an unsafe or unsound practice. OCC had already established safety or soundness standards under Section 12 CFR 30. The standards are extensive but a little vague. As I had noted in an earlier post, Part 30 uses the word “appropriate” 73 times. Last year’s Supreme Court decision in Loper Bright made it easier for federal courts to second guess regulators. While Congress explicitly gave regulators discretion to establish safety and soundness standards, it might make sense for regulators to state more explicitly for what they have in mind.

The NPR is also vague. It claims to encourage a greater focus on “material financial risks” without defining the term beyond “those most likely to cause significant stress.” Well, that doesn’t help much. What is “material”? Who knows?

The proposal would also codify standards for MRAs. Current OCC guidance describes MRAs as communicating a bank’s deficient practices. They follow a “5 Cs” format, where supervisors communicate the concern, cause, consequences, required corrective actions, and the bank’s commitment to fixing the problem. Supervisory concerns must reflect actual weaknesses and not merely the falling short of best practices. The proposal would raise the bar for MRAs further to the same “material financial risk” standard as for unsafe or unsound practices. Ratings downgrades to “3” or below would likewise follow the same standard.

The table below summarizes current and proposed approaches:

Do these guys understand risk?

The proposal would limit unsafe or unsound practices to those that are “contrary to generally accepted standards of prudent operation AND if continued,” are “likely to— (A) Materially harm the financial condition of the institution; or (B) Present a material risk of loss to the Deposit Insurance Fund; or (ii) Materially harmed the financial condition of the institution.” The key term here is “likely.” The term “if continued” allows for identification of an unsafe or unsound act before it impacts an institution’s financial condition, but the agencies make clear that “the conduct must be sufficiently proximate to a material harm to an institution’s financial condition to meet the proposed definition.”

Establishing a standard of likely and material financial harm to even criticize a bank reflects a fundamental misunderstanding of the nature of risk. A risk can be excessive, unsafe, and unsound without a likely bad outcome. Consider driving while intoxicated (DWI). Drunk drivers are roughly 14 times more likely than sober drivers to get into an accident. That doesn’t make getting into a crash likely. In fact, only about 0.16% of drunk drivers will get into a crash on any given trip. Does that mean that we should stop DWI enforcement since a drunk driver is more likely than not to get home safely?

The Case of SVB

Closer to home, consider the failure of Silicon Valley Bank. By the end of 2021, SVB looked like an accident waiting to happen. The bank had roughly doubled in size in the space of one year with nearly all its growth going into long-term fixed rate MBS. More than 90% of its deposits were uninsured with further concentrations within the tech and venture capital sectors. Interest rate risk management, contingency funding plans, and liquidity stress testing were all deficient. But based on what was known at the time, was a material financial loss likely, and if so, when?

Even at the end of 2022, the bank met well capitalized thresholds with an above-average risk-based capital ratio. The bank also remained profitable with few nonperforming assets. SVB started to experience deposit outflows in 2022, with deposits down by $16 billion. However, SVB also was able to replace these deposits with FHLB advances and total assets remained steady.

Unrealized losses on securities had risen to alarmingly high levels during 2022, reaching more than 100% of Tier 1 capital. But by then it may have been too late. The optimal time for supervisory intervention would have been in 2021, when the bank was growing imprudently but there was still an opportunity to reduce the IRR exposure at a modest loss. Trying to hedge by mid-2022 would have only locked in the unrealized losses.

It’s not even clear that SVB would have met the “likely material loss” standard in late 2022. Most of its securities were in the held-to-maturity category where bank management and its accountants asserted SVB had the intent and ability to hold to maturity. The last major runup in interest rates was nearly 30 years earlier (1994) and proved short-lived, returning to previous levels in only about a year. Riding out the adverse rate cycle was not entirely implausible.

SVB’s deposit concentrations made the bank especially vulnerable to runs. But were such runs likely? The magnitude of SVB’s run was largely unprecedented. In 2008, WaMu lost 10% of its deposits over 16 days. SVB lost 25% of its deposits in one day with an additional outflow of 62% slated for the next day.

The supervision of SVB has received and deserves a lot of criticism. But at least the Fed had issued numerous MRAs, assigned a CAMELS rating of “3” during 2022, and rated bank management deficient. It’s not clear whether any of these would have happened under the proposed framework.

The FDIC’s Supervision of First Republic Bank might provide a better indicator of things to come. That bank’s failure cost the FDIC $15.6 billion. Yet until March 31, 2023, the bank had no outstanding MRBAs, with both management and liquidity rated “1.” The FDIC made some eleventh-hour ratings downgrades and added an MRBA to try to appear slightly less ridiculous.

The NPR indicates that the proposal “reflects the agencies’ judgment and experience that their supervisory resources are best focused on practices that are likely to materially harm an institution’s financial condition, such as risks that are more likely than other risks to lead to material financial losses, bank failures, and instability in the banking system.” While it’s hard to argue with risk-based supervision, that’s not what’s happening here. Rather, the regulators are borrowing the narrative from bank lobbyists (especially the Bank Policy Institute) that MRAs somehow distracted banks and their supervisors from addressing the real risks. I’ve pointed out the absurdity of this argument here, here, and here.

The real problem behind the 2023 failures (as well as those in 2008 and in the 1980s) came from supervisors’ failure to quickly and decisively address problems when they were still tractable. In other words, a lack of preventive supervision. The proposal would make the situation worse.

Implementing the proposal will make it much more difficult for examiners to address problems early. Imagine you’re a bank examiner and see a bunch of warning signs. The bank is growing quickly and aggressively, with significant asset and funding concentrations. Records and controls are weak and the second and third lines present little effective challenge. Management shows little inclination to fix the bank’s problems. Now imagine trying to convince a team of attorneys and political appointees that these problems are severe enough to make a financial loss both likely and material. You’ll probably need to wait until these weaknesses manifest themselves into losses large enough to meet some (as yet unspecified) materiality threshold. Then you’ll have to hope it’s not too late.