Last month, the Wall Street Journal published an article indicating several big banks were considering a legal challenge to the Federal Reserve’s annual stress testing exercise based on a violation of the Administrative Procedure Act (“APA”). The article was short on details, and to those familiar with the APA – the statutory scheme governing all administrative agencies – it likely created more questions than it answered. Many observers, including me, interpreted the article as a rather naked attempt to get the Fed to reveal more details about the exercise, and not as a credible threat to sue.
Big banks have been complaining for several years about the opacity of the annual stress testing exercise known as the Comprehensive Capital Analysis and Review (CCAR). Given the amount of money firms spend to pass the exercise and the stakes involved – the ability to buy back shares and pay dividends – any attempt to get the Fed to share more details is understandable. However, it is rare for firms to sue their regulators (especially prudential regulators) and for good reason: The odds of success are low, and there is the potential to sour the relationship with the very folks who have the discretion to bring and settle enforcement actions against you.
More details on the potential legal strategy were revealed several weeks ago when the Committee on Capital Markets Regulation (“the Committee”), a nonprofit composed partly of academics and partly of financial industry executives, published a white paper outlining the proposed challenge to the Fed’s stress tests under the APA. The Committee’s main argument is: (1) The Fed can prevent banks from distributing dividends or buying back stock from their shareholders if they fail the stress tests; (2) this means that the tests “act as a de facto binding capital constraint on banks”; (3) because this constraint is binding, the CCAR stress scenarios should have been promulgated through notice-and-comment rulemaking under the APA; (4) since they were not, the tests may be invalid.
The paper recommends that the Fed disclose two key elements – details on the assumptions underlying the stress scenarios, and the models used by the Fed to project firm revenue and losses – for public comment under the procedures outlined in Section 553 of the APA.
Each firm’s CCAR results are based on its performance in the severely adverse scenario, which uses assumptions around data points such as GDP growth and the unemployment rate to paint a pretty grim picture of the world. Most firms think the scenario is too bleak, and more importantly – unrealistic. And while the white paper makes some reasonable arguments around why the Fed should release more details on how they develop these assumptions, it is unlikely additional details will have a meaningful impact on the outcome of the tests.
What the banks are really after are the Fed’s internal models, the black box so to speak. These models take firm supplied balance sheet information, apply the scenario assumptions, and spit out projections for revenue, losses, and regulatory capital ratios. Firms project this information using their own models as well, but discrepancies between their results and the Fed’s have left the firms scratching their heads and clamoring for more information.
It seems unlikely that the banks will get their hands on these models for several reasons. First, I doubt any bank will actually file a lawsuit given the potential regulatory blowback and poor optics of mega banks suing their regulator – especially in light of the recent Wells Fargo scandal. Deutsche Bank and the ICBA appear to recognize this reality as evidenced by the very first footnote to the paper which indicates their dissent.
If a lawsuit were to proceed, who would bring the claim? It is hard to see how an industry organization, like the Committee on Capital Markets Regulation, would have standing to sue. With public opinion of Wall Street near all-time lows, it’s hard to imagine any of the big banks having the chutzpah to file suit. Any bank filing such a challenge might attract more negative attention and speculation than it would like.
Moreover, even if a plaintiff were to step forward, the APA provides several exceptions to the notice-and-comment process. One of them is known as the “good cause” exception. Under this exception, agencies may promulgate binding rules without notice-and-comment, and without publishing the final rules, if the agency demonstrates that those procedures are impracticable, unnecessary, or contrary to the public interest. The white paper acknowledges the Fed’s argument that disclosure of the models could be against public interest because the banks may “game” the test by adjusting their balance sheets leading up to CCAR in order to perform well under the test, and then altering the balance sheet immediately afterwards. However, the Committee discounts this concern by noting that if the Fed has reason to believe a bank “gamed” the models, then they could reject the firm’s capital plan on qualitative grounds.
The white paper gives short shrift to two case law examples that support the government’s ability to utilize the “good cause” exception, and fails to mention other situations where these cases and their progeny provided the basis for waivers of the APA’s notice-and-comment requirement. These original cases relate to the government’s attempts to reduce oil prices during the 1970’s oil crisis, when the government’s imposition of price controls was not subject to notice-and-comment due to the valid concern that had the rule been disclosed, market participants would react in such a way that prices would go up further. In essence, the government was concerned about market participants “gaming” the rule before it was finalized, and the courts have continued to allow waiver of notice-and-comment where “the very announcement of a proposed rule itself can be expected to precipitate activity by affected parties that would harm the public welfare.”
The Fed has also expressed concern that disclosing additional model details could lead to model convergence. Model convergence is the notion that once the Fed releases the details of its models, all the banks will adjust their internal models to align with the Fed’s, thereby failing to account for idiosyncratic risks or tail risks not captured in the Fed’s models. This is a very real concern, as evidenced by the fact that leading up to the financial crisis, the majority of bank and regulatory models failed to account for the potential of a nationwide housing price decline to infect the health of the broader financial system. If a lawsuit were filed, the Fed would certainly cite possible “harm [to] public welfare” resulting from model convergence as grounds for applying the “good cause” exception.
The banking industry’s concerns about the CCAR process have not gone unnoticed by its regulators. Last week, Fed Governor Daniel Tarullo announced a series of reforms to the annual exercise. The most significant change is the requirement that firms hold a “stress capital buffer,” which Tarullo admits will increase the capital requirements for the eight largest banks while reducing the requirements for the remaining banks. But most of the other changes are designed to enhance CCAR’s transparency. These include:
- Assuming that balance sheets and risk-weighted assets remain constant over the severely adverse scenario horizon, as opposed to the previous assumption that balance sheets would increase over the horizon;
- Reducing the severity of the change in the unemployment rate;
- Tying the change in house prices to disposable personal income; and
- Providing more details on the different components of projected net revenues.
Tarullo also indicated that the Fed is considering additional steps to promote transparency. These include disclosing descriptions of changes to the models well in advance; phasing in any material changes over several years; and most significantly, “publishing data that represent typical bank portfolios of loans and securities with the losses [the Fed] would project for those portfolios under various stress scenarios.”
It remains to be seen if these changes will be enough to pacify the big banks. I suspect not, considering that in his speech Tarullo stated the Fed has no intention “to publish the full computer code in the supervisory model that is used to project revenues and losses.” If the banks want full access, they will have to follow through on their threat to sue.
 DeRieux v. Five Smiths, Inc., 499 F.2d 1321 (Temp. Emerg. Ct. App. 1975); Nader v. Sawhill, 514 F.2d 1064 (Temp. Emerg. Ct. App. 1975).
 See, e.g., Mobil Oil Corp. v. Dep’t of Energy, 728 F.2d 1477 (Temp. Emer. Ct. App. 1983); United Ref. Co. v. Dep’t of Energy, 585 F. Supp. 626 (W.D. Pa. 1984).
 Mobil Oil Corp., 728 F.2d at 1492.