Microsoft Word - 7. FINAL_Tarullo.docx BANK SUPERVISION AND ADMINISTRATIVE LAW Daniel K. Tarullo* This Article presents a systematic consideration of how administrative law doctrines apply to banking supervision, an unusual form of administrative practice that rests on an iterative relationship between banks and supervisors. First, it describes the rationales for, and process of, bank supervision. Second, this Article uses recent administrative law arguments lodged by banking interests against key supervisory practices as the springboard for an analysis of why our largely “trans- substantive” administrative law can be problematic in the context of specific mandates given by Congress to administrative agencies. It argues that courts considering how administrative law doctrine applies to agency practices must contemplate more fully the substantive law the underpins the mission and organization of the agency. When these statutory provisions are taken appropriately into account, arguments that supervisory practices are consistent with administrative law requirements are substantially strengthened. Third, this Article demonstrates how even a more tailored application of contemporary administrative law doctrines would miss a critical feature of banking supervision—that it is premised on an ongoing relationship between banks and supervisors. Judicial review of agency action usually focuses on discrete * Nomura Professor of International Financial Regulatory Practice, Harvard Law School. From 2009 to 2017, the author was a Member of the Board of Governors of the Federal Reserve System, and oversight Governor for Supervision & Regulation. Thanks to Scott Alvarez, Tim Clark, Cary Coglianese, Howell Jackson, Jack Jennings, Patricia McCoy, Lev Menand, Gillian Metzger, Todd Rakoff, Kevin Stiroh, and participants in the Yale Journal on Regulation 2019 Administrative Law Conference and a Harvard Law School faculty workshop for helpful comments on earlier versions of this Article. Of course, no views expressed in the paper should be attributed to any of them. 280 COLUMBIA BUSINESS LAW REVIEW [Vol. 2022 agency actions, thereby eliding this critical fact. As a result, administrative law doctrines such as the “practically binding” test for agency guidance are peculiarly inapposite. Lastly, this Article offers a tentative proposal for shifting the administrative law review of supervisory actions to focus on how banking agency processes manage the iterative nature of the supervisory relationship. I. Introduction ..................................................................... 281 II. The Nature of Bank Supervision ................................... 286 A. The Role of Bank Supervision .............................. 291 B. The Organization of Bank Supervision ................ 300 III. Administrative Law Doctrine and Supervisory Activities ................................................................................... 315 A. Procedural Requirements for a Binding Stress Test Regime .................................................................. 316 1. Background on Capital Regulation and Stress Testing ................................................. 318 2. Arguments for Notice-and-Comment on Supervisory Stress Model and Scenarios. ...... 324 3. Relevant Statutory Provisions on Capital Regulation. ...................................................... 332 B. Supervisory Ratings ............................................. 340 C. Supervisory Guidance and Communication ........ 347 1. Supervisory Guidance and MRAs. .................. 349 2. Interaction of Guidance, MRAs, and Ratings . 353 3. The Poor Fit Between Doctrine and the Supervisory Function ...................................... 363 IV. Is a Different Approach Feasible? ................................ 367 A. The Conceptual Mismatch Revisited ................... 368 B. Toward Greater Congruence of Supervision and Administrative Law .............................................. 376 1. Trans-Substantive Options for Reconciling Administrative Law and Supervision............. 377 2. An Alternative Approach ................................ 379 3. Internal Administrative Law and Judicial Review. ............................................................ 383 4. Some Qualifications. ....................................... 395 V. Conclusion ................................................................. 398 No. 1:279] BANK SUPERVISION AND ADMINISTRATIVE LAW 281 I. INTRODUCTION Bank supervision is an unusual form of administrative practice that sits uneasily within contemporary administrative law doctrines. The supervisory function involves “monitoring, inspecting, and examining financial institutions” so as “to ensure that an institution complies with [applicable] rules and regulations, and that it operates in a safe and sound manner.”1 Rule enforcement, including through various forms of on-site examination or inspection, is hardly unique to banking agencies.2 My focus here is principally on that second task of supervisors, referencing the statutory authority of banking agencies to prohibit “unsafe or unsound” banking practices,3 which can include just about anything a bank is doing that may materially affect its financial soundness. Supervision is an iterative process of communication between banks and supervisors.4 It routinely involves the identification of potentially unsafe and unsound practices in the manifold aspects of bank activity unaddressed by legislative rules, both for banks generally and for individual banks.5 Though supervisory communication is not legally binding, it is intended to affect bank practice. It usually, though not invariably, does.6 Predictably, this influence of specified supervisory expectations rests in part on the prospect of formal enforcement actions under the quite capacious statutory 1 BD. OF GOVERNORS OF THE FED. RSRV. SYS., THE FEDERAL RESERVE SYSTEM: PURPOSES AND FUNCTIONS 74 (10th ed. 2016). 2 See, e.g., 29 U.S.C. § 657 (2018) (granting the Secretary of Health and Human Services the power to “enter, inspect, and investigate places of employment”). 3 12 U.S.C. § 1818(b)(1) (2018). 4 See OFF. OF THE COMPTROLLER OF THE CURRENCY, COMPTROLLER’S HANDBOOK: BANK SUPERVISION PROCESS 7 (2019). 5 See id. at 46. 6 See id. at 40–45. Information about how these procedures affect bank behavior is based on the author’s experience at the Federal Reserve. 282 COLUMBIA BUSINESS LAW REVIEW [Vol. 2022 enforcement authority granted to the three Federal Reserve banking agencies—the Federal Reserve, the Office of the Comptroller of the Currency (OCC), and the Federal Deposit Insurance Corporation (FDIC). But it also grounded in two other elements of banking law. One is the statutory requirement for supervisory ratings of banks, which can affect their rights and obligations. The other is the wide range of approvals required for various bank activities. Many of these approvals are decided with reference to statutory factors such as the banking organization’s overall condition or managerial capabilities. Banking agency evaluation of such factors derives substantially from supervisory experience with the applying bank. These characteristics of bank supervision are by no means new. But as supervision became more rigorous in the period following the Global Financial Crisis of 2007-2009 (“Financial Crisis” or “Crisis”), commentators associated with, or sympathetic to, the banking industry have argued that there are important administrative law deficiencies in supervision. Some have argued that the notice-and-comment requirements of the Administrative Procedure Act (APA)7 should apply to key supervisory actions such as scenario design for supervisory stress tests of banks’ capital adequacy.8 Others have complained that the proliferation of informal supervisory directions is untethered from statutory authority and that there is excessive secrecy around the decisions taken by banking agencies on a range of issues.9 Longstanding advocacy by smaller banks for judicial or other independent 7 5 U.S.C. § 553(b) (2018). 8 See COMM. ON CAP. MKTS. REGUL., THE ADMINISTRATIVE PROCEDURE ACT AND FEDERAL RESERVE STRESS TESTS (2016), https://www.capmktsreg.org/wp-content/uploads/2019/05/The- Administrative-Procedure-Act-and-Federal-Reserve-Stress-Tests- Enhancing-Transparency.pdf [https://perma.cc/8XF8-T4WH] [hereinafter “CCMR Paper”]. 9 See Guidance, Supervisory Expectations, and the Rule of Law: How Do the Banking Agencies Regulate and Supervise Institutions?: Hearing Before the S. Comm. on Banking, Hous., & Urban Aff., 116th Cong. 42 (2019) [hereinafter Tahyar Testimony] (statement of Margaret E. Tahyar, Partner, Davis, Polk & Wardwel LLP). No. 1:279] BANK SUPERVISION AND ADMINISTRATIVE LAW 283 review of the safety and soundness ratings assigned by supervisors was renewed as supervisory bank ratings were downgraded during and after the crisis.10 To date there have not been many judicial cases involving supervisory actions. But a legal challenge to certain supervisory actions, which was said to be under consideration by some banks prior to the change in leadership at the banking agencies following President Donald J. Trump’s election, could follow a return to more rigorous policies by President Joseph R. Biden’s appointees. In this Article, I use the prospect of these challenges as an entry point for considering how administrative law does, and should, apply to the supervisory function. I draw three conclusions from this analysis. First, while there are good arguments that core supervisory actions pass muster under conventional readings of current administrative law, it is not entirely clear how some practices would fare in a court challenge. In particular, aggressive judicial expansion of the “practically binding” test for determining whether an agency policy must be channeled through notice-and-comment rulemaking could strike at the core of the supervisory relationship—the use of guidance and other forms of supervisory communication to affect bank behavior in ways that pull up short of legal compulsion. The uncertainty is due in part to the relative paucity of court cases dealing with supervision. But it also arises in part from the heavily trans-substantive slant of much contemporary administrative law. Doctrines developed in one regulatory 10 There is some irony in this upsurge of complaints from banks and their supporters about excessive or arbitrary supervision, insofar as a recurring academic narrative is that bank supervisors are regularly captured by the banks they supervise. See infra notes 193–196 and accompanying text. A good example may be found in many of Ed Kane’s papers. See, e.g., Edward J. Kane, Changing Incentives Facing Financial Services Regulators, 2 J. FIN. SERVS. RSCH. 265 (1989). For a review of criticisms and ideas for mitigating capture, see Lawrence G. Baxter, Essay, Capture in Financial Regulation: Can We Channel It Towards the Common Good?, 21 CORNELL J. L. & PUB. POL’Y 175 (2011). The financial crisis reinforced this narrative, which was featured prominently in congressional hearings and media reporting. 284 COLUMBIA BUSINESS LAW REVIEW [Vol. 2022 context may be applied in others without considering the sometimes substantial variation in statutory mandates and organizational realities. Second, when due account is taken of statutory banking law, administrative law analysis of most supervisory practices becomes very favorable. The broad targets of banking law, such as “unsafe or unsound” practices and systemic risk, reflect the heterogeneous nature of financial intermediation. Statutory provisions recognize, and require, the distinctive administrative function of supervision. In deciding how the general terms of the APA, contained in Title 5 of the United States Code, apply to supervisory practices, provisions of Title 12 cannot be ignored. That is, while the APA of course applies to all administrative actions not explicitly exempted, the judicial determinations of how its terms and exceptions apply should take account of the mandates established by Congress in substantive legislation such as the banking laws. Third, the traditional focus of administrative law review on a specific agency action could present its own problems if supervisory actions were more frequently challenged. Banking supervision above all entails an ongoing relationship between a bank and agency officials. The origins and effects of most supervisory actions and communications connect to other actions and communications. Thus, while applying administrative law doctrine with a recognition of the statutory foundations of bank supervision would be preferable to a homogenized mode of review, there could still be a gap between doctrinal emphasis and the realities of the supervisory relationship for both banks and the banking agencies. This conclusion suggests that an even more differentiated administrative law treatment of banking supervision could be warranted—one that focuses on the agencies’ process for managing the supervisory relationship, within which non-legally-binding communications are supposed to influence bank behavior. Part II begins with a description of the basics of contemporary banking supervision, including its statutory sources, and a more precise account of how it complements bank regulatory rules. This account requires some No. 1:279] BANK SUPERVISION AND ADMINISTRATIVE LAW 285 explanation of the nature of financial risk, and thus the task of prudential regulation. The opacity, heterogeneity, and fast- changing nature of much financial activity compound the difficulty of capturing in rules all the combinations of circumstances that determine whether a bank’s practices are “unsafe or unsound.” Good supervision can often counteract new or increasing risks more expeditiously and efficiently than generally applicable regulatory rules. Appropriately conceived and implemented supervisory activities thereby allow safety and soundness regulations to be calibrated less stringently and more efficiently. Part III presents a doctrinal analysis of potential challenges to three important forms of supervisory activities: supervisory stress testing in setting minimum capital requirements; assignment of supervisory ratings, which carry tangible consequences for banks; and the use of supervisory guidance and various informal means of communication, which are central to the supervisory relationship. In all three cases, the supervisory practices can be well-defended solely by reference to the trans-substantive application of doctrine that characterizes much contemporary administrative law. But in two of those three cases, there is at least some doubt as to how a court might rule.11 The discussions of stress testing and supervisory guidance go on to specify how the substantive provisions of banking law relevant to supervision should be factored into the procedural analysis of administrative law. Part IV considers the implications of one conclusion from Part III—that contemporary administrative law may require an effort to place the square peg of the overall supervisory relationship through the round hole of an administrative law built on review of specific actions. Sometimes it will fit through with no problem; sometimes it will not. Whatever the outcome in a specific case, there will often seem a lack of congruence between the doctrinal status of the practice and how it measures up to norms of accountability, fairness, and due consideration of policy options. This Part begins with a 11 Ironically, the exception is the ratings process, which actually presents the strongest normative case for more fully developed agency procedures. 286 COLUMBIA BUSINESS LAW REVIEW [Vol. 2022 brief historical review that helps explain the current disconnect between administrative law and banking supervision. It then presents a possible alternative—a framework for internal administrative law and practice that would evaluate challenges to specific supervisory actions within the broader context of the supervisory relationship. Such an approach would, to be sure, validate certain actions that some banks may want to challenge under current doctrine. But by requiring more regularity and transparency in the overall supervisory structure, it could also have advantages for banks, especially the smaller banks for which expensive court challenges to administrative action are rarely a realistic option. While my focus is squarely on the unusual characteristics and statutory foundations of bank supervision, I refer at several places in the discussion to the potential relevance of my analysis for administrative law in other substantive regulatory areas. Although the very particularity of the supervisory function in banking law that underlies much of my argument counsels caution in extrapolating to other substantive areas, Part V concludes by briefly drawing together and expanding upon some of these references. II. THE NATURE OF BANK SUPERVISION As a historical matter, bank examination and supervision predated bank regulation,12 at least as we today understand regulation to be grounded in a set of rules found in the United States Code and the Code of Federal Regulations.13 The 12 Terminology can be confusing here, since bank “regulation” is often used to refer both to the whole enterprise of prudential regulation— including legislation, notice-and-comment rules, and supervision—and more narrowly to the set of regulations promulgated by banking agencies and collected in the Code of Federal Regulations. In this Article, I will try to avoid confusion by using “regulation” to refer to the former and “regulatory rules” to refer to the latter. 13 The states and, following passage of the National Bank Act, Act of June 3, 1864, ch. 106, 13 Stat. 99, during the Civil War, the federal government dispatched examiners to be sure banks were meeting the conditions and requirements under which they had been chartered. For No. 1:279] BANK SUPERVISION AND ADMINISTRATIVE LAW 287 vestiges of this history can be seen in the organization and prominence of supervision within the federal banking agencies. It is also reflected in the statutes applicable to banking organizations. Contrary to the assertions of some representatives of banking interests,14 the supervisory role of the banking agencies is well-grounded in law.15 Statutory provisions explicitly assume, and expect, a supervisory function that is distinct from regulation and that reaches well discussions of the historical origins of bank supervision in the United States, see BRAY HAMMOND, BANKS AND POLITICS IN AMERICA FROM THE REVOLUTION TO THE CIVIL WAR (1957). The early evolution of federal supervision by the Comptroller is recounted in ROSS M. ROBERTSON, THE COMPTROLLER AND BANK SUPERVISION: A HISTORICAL APPRAISAL (1995). For a recent historical account showing the far-reaching role contemplated for supervision, see Lev Menand, Why Supervise Banks? The Foundations of the American Monetary Settlement, 74 VAND. L. REV. 951 (2021). 14 A lawyer whose firm regularly represents large banks and the president of an organization that describes itself as “conducting research and advocacy on behalf of America’s leading banks” testified before the Senate Banking Committee. Greg Baer, Welcome to the Bank Policy Institute, BANK POL’Y INST. (July 16, 2018), https://bpi.com/welcome/ [https://perma.cc/YME9-3H2Q]. They claimed, respectively, that “[t]he word supervision . . . appears nowhere in the legal framework governing the banking sector[]” or “the authorizing statutes for the examination process.” Tahyar Testimony, supra note 9; Guidance, Supervisory Expectations, and the Rule of Law: How Do the Banking Agencies Regulate and Supervise Institutions?: Hearing Before the S. Comm. on Banking, Hous., & Urban Aff., 116th Cong. 30 (2019) [hereinafter Baer Testimony] (statement of Greg Baer, President, Bank Pol’y Inst.). Both Mr. Baer and Ms. Tahyar are wrong. An obvious example is in the legislation creating the Federal Financial Institutions Examination Council, described infra, notes 78–83 and accompanying text, which included in its instructions to the new Council “recommendations to promote uniformity in the supervision of” depository institutions. Financial Institutions Regulatory and Interstate Rate Control Act of 1978, Pub. L. No. 95-630, § 1002, 92 Stat. 3641, 3694 (codified at 12 U.S.C. § 3301). More broadly, the words “supervisory” and “supervisors” appear literally dozens of times in the U.S. Code. See, e.g., infra notes 17–26 and accompanying text. 15 As explained in infra Part III, contrary to the arguments cited in the preceding footnote, supra note 14, much of current bank supervisory practice is consistent with current, trans-substantive understandings of administrative law. Once the statutory foundation of supervision is considered, the argument for its validity becomes even stronger. 288 COLUMBIA BUSINESS LAW REVIEW [Vol. 2022 beyond enforcement of regulatory rules.16 The Federal Reserve Act prohibits the Board of Governors from delegating to the regional Reserve Banks its functions “for the establishment of policies for the supervision and regulation of depository institution holding companies” and requires the Board of Governors to assess such fees on large bank holding companies as “are necessary or appropriate to carry out the supervisory and regulatory responsibilities of the Board.”17 Whenever the deposit insurance fund operated by the FDIC suffers a “material loss” because of the failure of a bank, the inspector general of the relevant federal banking agency must make a report “reviewing the agency’s supervision of the institution” to determine the causes of the loss and to make recommendations to the agency for preventing future such losses.18 In 1994, Congress required each federal banking 16 As with banking “regulation,” the term “supervise” and its variants are sometimes used more generically to refer to a banking agency or to specify firms covered by a statutory provision, as in “large interconnected bank holding companies supervised by the Board of Governors.” 12 U.S.C. § 5322(a)(2)(I) (2018). The statutory provisions cited in the text and accompanying footnotes that follow all are used in the more precise sense, drawing distinctions between supervision, on the one hand, and regulation, enforcement, or examination on the other. 17 Federal Reserve Act of 1913 § 11, 12 U.S.C. §§ 248(k), (s) (emphasis added). The Act also grants the Board authority to “examine” any depository institution or affiliate for purposes of deciding its eligibility for access to the discount window. Id. § 248(n). Thus, in a single provision of the Federal Reserve Act, Congress has shown that it uses “examine,” “supervise,” and “regulate” to mean different things. See also id. § 242 (“The Vice Chairman for Supervision shall develop policy recommendations for the Board regarding supervision and regulation of depository institution holding companies and other financial firms supervised by the Board, and shall oversee the supervision and regulation of such firms.” (emphasis added)). Similarly, a 2014 provision clarifying the application of certain accounting principles to bank holding companies with insurance subsidiaries specified that it did not “limit the authority of the Board under any other applicable provision of law to conduct any regulatory or supervisory activity of a depository institution holding company[.]” Insurance Capital Standards Clarification Act of 2014, Pub. L. No. 113-279, sec. 2, § 171(7)(c)(B), 128 Stat. 3017, 3018 (codified at 12 U.S.C. § 5371(c)(1)(B)) (emphasis added). 18 12 U.S.C. §1831o(k)(1)(A) (emphasis added). Following the savings and loan crisis of the late 1980s, this section was added by the Federal No. 1:279] BANK SUPERVISION AND ADMINISTRATIVE LAW 289 agency to establish an independent intra-agency process “to review material supervisory determinations made at insured depository institutions,”19 which are defined to include not only examination ratings, but also “the adequacy of loan loss reserve provisions” and “loan classifications on loans that are significant to an institution.”20 Other statutory provisions show that “supervision” is not subsumed under “examination” or “enforcement” authority.21 In the 1978 legislation creating the Federal Financial Institutions Examination Council, Congress supplemented the Council’s mandate to establish uniform examination standards for all insured depository institutions with the instruction to “make recommendations for uniformity in other supervisory matters.”22 The statute give the explicitly non- exclusive examples of “classifying loans subject to country risk, identifying financial institutions in need of special Deposit Insurance Corporation Improvement Act of 1991, Pub. L. 102-242, sec. 131, § 38(k)(1)(A), 105 Stat. 2236, 2263 (amending the Federal Deposit Insurance Act). While this provision does not define “supervision,” the reports of material loss reviews by inspectors general of the banking agencies make clear that they interpret supervision as encompassing more than regulation and formal enforcement actions. See, e.g., DEP’T OF THE TREASURY OFF. OF INSPECTOR GEN., OIG-16-052, SAFETY AND SOUNDNESS: ANALYSIS OF BANK FAILURES REVIEWED BY THE DEPARTMENT OF THE TREASURY OFFICE OF INSPECTOR GENERAL, REPORT 14–15 (2016) (summarizing supervisory shortcomings and recommendations from fifty- four material loss reviews, including absence of appropriate informal enforcement actions and MRAs). 19 Riegle Community Development and Regulatory Improvement Act of 1994, Pub. L. No. 103-325, 108 Stat. 2160, 2218 (codified at 12 U.S.C. § 4806(a)) (emphasis added). 20 12 U.S.C. § 4806(f)(1)(A). 21 In his Senate testimony, Mr. Baer of the Bank Policy Institute asserted that “[t]here is a large difference between examining a firm and supervising it,” with the implication that the latter is somehow unauthorized. Baer Testimony, supra note 14. It is true that there is a lot more to supervision than an annual examination. See id. As explained in the text, it is decidedly not the case that the distinct role of supervision is unrecognized by Congress. 22 Financial Institutions Regulatory and Interstate Rate Control Act of 1978, Pub. L. No. 95-630, § 1006(b), 92 Stat. 3641, 3695 (codified at 12 U.S.C. § 3305(b)) (emphasis added). 290 COLUMBIA BUSINESS LAW REVIEW [Vol. 2022 supervisory attention, and evaluating the soundness of large loans that are shared by two or more financial institutions.”23 In its most recent significant amendments to banking law, in early 2018,24 Congress again indicated its recognition of supervisory authority distinct from enforcement authority. While limiting the circumstances under which a higher capital requirement could be applied to certain commercial real estate lending, Congress added that “[n]othing in this section shall limit the supervisory, regulatory, or enforcement authority of an appropriate federal banking agency to further the safe and sound operation of an institution.”25 Finally, 23 Id. An especially telling example of congressional awareness and embrace of a distinct supervisory role is provided by a 1989 amendment to the statutory provision authorizing the role of the Comptroller of the Currency in overseeing banks placed in conservatorship. The original language in the 1933 law provided that the Comptroller should “cause to be made such examinations of the affairs of such bank as shall be necessary to inform him as to the financial condition of such bank.” Bank Conservation Act, Pub L. No. 73-1, § 204, 48 Stat. 1, 3 (1933) (providing relief for the national emergency in banking). The 1989 amendment changed the language to read that the Comptroller is “authorized to examine and supervise the bank in conservatorship as long as the bank continues to operate as a going concern.” Financial Institutions Reform, Recovery and Enforcement Act of 1989 (FIRREA), Pub. L. No. 101-73, sec. 803, § 204, 103 Stat. 183, 443, (codified at 12 U.S.C. § 204) (emphasis added). 24 Economic Growth, Regulatory Relief, and Consumer Protection Act, Pub. L. No. 115-174, 132 Stat. 1296 (2018) (codified as amended in scattered sections of 12 U.S.C. and 15 U.S.C.). 25 Economic Growth, Regulatory Relief, and Consumer Protection Act § 214, 12 U.S.C. § 1831bb(e) (emphasis added). There are other examples of statutory provisions distinguishing supervision from examination and enforcement. See, e.g., 12 U.S.C. § 3109(a) (banking agencies “may disclose information obtained in the course of exercising supervisory or examination authority to any foreign bank regulatory or supervisory authority . . .” (emphasis added)); id § 3903 (instructing banking agencies to make an evaluation of banks’ exposure to foreign countries “for use in banking institution examination and supervision.” (emphasis added)); id. § 3910(a)(2) (audit by Government Accountability Office “may include a review or evaluation of the international regulation, supervision, and examination activities” of a federal banking agency (emphasis added)); id. § 5362(b) (Federal Reserve may recommend that primary financial regulatory agency for a subsidiary of a bank holding company “initiate a supervisory No. 1:279] BANK SUPERVISION AND ADMINISTRATIVE LAW 291 some statutory provisions refer to supervisory “actions” or “determinations” in contexts clearly outside formal enforcement proceedings. As part of its legislative response to the savings and loan crisis that began in the late 1980s, Congress required federal banking agencies to maintain records of “all informal enforcement agreements and other supervisory actions” related to any subsequent administrative enforcement proceedings.26 A. The Role of Bank Supervision Although a full exposition of prudential bank regulation is beyond the scope of this Article, a brief explanation provides some context for understanding the role of supervision.27 Prudential bank regulation is traditionally motivated by moral hazard and negative externality problems. Moral hazard inheres in government-provided deposit insurance programs, access to discount window and other central bank liquidity facilities,28 and, less directly, the well-grounded belief that government will support banks when liquidity action or enforcement proceeding” and may take that action if unsatisfied with other agency’s response (emphasis added)). 26 Comprehensive Thrift and Bank Fraud Prosecution and Taxpayer Recovery Act of 1990, Pub. L. No. 101-647, sec. 2547(a)(1), § 8(u)(7), 104 Stat. 4859, 4887 (codified at 12 U.S.C. § 1818(u)(6)). See also 12 U.S.C. § 4806(a) (requiring “an independent intra-agency appellate process . . . to review material supervisory determinations made at insure depository institutions” (emphasis added)). 27 The adjective “prudential” confines the scope of the regulatory actions referenced to those intended to protect the safety and soundness of banks or the financial system as a whole. There is extensive regulation of banks—in areas such as consumer protection and anti-money laundering measures—that is not included under this rubric. For an accessible introduction to both the prudential and non-prudential elements of financial regulation, see JOHN ARMOUR ET AL., PRINCIPLES OF FINANCIAL REGULATION 51–79 (2016). 28 For an explanation of the traditional moral hazard concerns, see MICHAEL S. BARR, HOWELL E. JACKSON & MARGARET E. TAHYAR, FINANCIAL REGULATION: LAW AND POLICY 266–68 (3rd ed. 2021). 292 COLUMBIA BUSINESS LAW REVIEW [Vol. 2022 strains or solvency concerns across banks threaten the operation of the financial system as a whole.29 Moral hazard concerns are especially pronounced with respect to the largest financial firms, the failure of any one of which could have far-reaching consequences for the entire financial system. As has been well-documented, large banks enjoyed an effective funding subsidy in the pre-Crisis period.30 Investors and counterparties incorporated a lower risk premium in lending to those firms, apparently reflecting their judgment that they would be made whole through government assistance to large banks during periods of stress.31 Although there is some evidence that post-Crisis regulatory reforms have reduced this subsidy,32 there is little doubt that the government continues to provide what is effectively a form of systemic risk insurance to banks and, indeed, the entire financial system of which banks are one important part.33 29 For an explanation of the dynamic behind this more indirect form of moral hazard, see Emmanuel Farhi & Jean Tirole, Collective Moral Hazard, Maturity Mismatch, and Systemic Bailouts, 102 AM. ECON. REV. 60, 61–62 (2012). 30 See Elijah Brewer III & Julapa Jagtiani, How Much Did Banks Pay To Become Too-Big-To-Fail and To Become Systemically Important?, 43 J. FIN. SERV. RSCH. 1, 30–32 (2013); Kenichi Ueda & Beatrice Weder di Mauro, Quantifying Structural Subsidy Values for Systemically Important Financial Institutions, 37 J. BANKING & FIN. 3830, 3840 (2013). 31 See Brewer & Jagtiani, supra note 30; Ueda & Weder di Mauro, supra note 30. 32 See NICOLA CETORELLI & JAMES TRAINA, FED. RSRV. BANK OF N.Y. STAFF REP. NO. 859, RESOLVING ‘TOO BIG TO FAIL’ 12 (2018), https://www.newyorkfed.org/medialibrary/media/research/staff_reports/sr8 59.pdf [https://perma.cc/68SA-TVDY]. 33 The COVID-19 induced crisis in early 2020 brought forth once again a multitude of liquidity facilities from the Federal Reserve to support an even broader range of financial assets. See Funding, Credit, Liquidity, and Loan Facilities, FED. RSRV. (Apr. 12, 2021), https://www.federalreserve.gov/funding-credit-liquidity-and-loan- facilities.htm [https://perma.cc/RRY4-PP8J] (describing various programs the Federal Reserve established, including the Municipal Liquidity Facility, a Main Street Lending Program, and a Commercial Paper Funding Facility). No. 1:279] BANK SUPERVISION AND ADMINISTRATIVE LAW 293 A complementary rationale for prudential regulation is that serious financial strains on large financial institutions can produce severe negative externalities for the financial system.34 That is, even if moral hazard could be substantially reduced with a credible prospect of bank resolution and imposition of losses on shareholders and creditors, there may still be large negative externalities.35 The actual failure of large banks could result both in a downward spiral in asset prices resulting from fire sales36 and in a major contraction of credit availability.37 More generally, the fact that banks affect the money supply through their creation of demand deposits gives the government a major stake in their stability. Distinctive characteristics of traditional banking and certain other forms of financial intermediation complicate the development of a regulatory framework to address the motivating concerns. Bank balance sheets can be quite opaque, in that the characteristics and quality of assets may be difficult to discern accurately, even when the basic classification of an asset is known.38 The relative speed with which a bank’s principal productive input—money—can be redeployed to other “products” (i.e., other kinds of lending or investment) distinguishes banks from most other commercial activities, in which the introduction of new products carrying a different risk/reward calculus is usually both slower-moving and more transparent. The risks associated with loans carrying the same terms and with very similar borrower profiles may vary materially among banks, depending on the 34 See ARMOUR ET AL., supra note 27, at 57–59. 35 Id. 36 FIN. STABILITY BD., EVALUATION OF THE EFFECTS OF TOO-BIG-TO-FAIL REFORMS 12 (2020), https://www.fsb.org/wp-content/uploads/P280620-1.pdf [https://perma.cc/VW73-42W4]. 37 Naoaki Minamihashi, Credit Crunch Caused by Bank Failures and Self-Selection Behavior in Lending Markets, 43 J. MONEY, CREDIT & BANKING 133, 134 (2011). 38 For an example of one such accounting trick banks may use, see Morgan Housel, The Blatant Opacity of Banks’ Balance Sheets, THE MOTLEY FOOL (Apr. 6, 2017, 1:14 AM), https://www.fool.com/investing/general/2009/08/19/the-blatant-opacity-of- banks-balance-sheets.aspx [https://perma.cc/Y343-HKYP]. 294 COLUMBIA BUSINESS LAW REVIEW [Vol. 2022 composition of the rest of their portfolios.39 These characteristics of banking together mean that external recognition of a bank’s deterioration by depositors, investors, and others may materially lag the reality. Finally, because banks rely significantly on uninsured deposits and other forms of short-term funding, this opacity also makes them susceptible to funding runs. Depositors and other counterparties are aware that they may have insufficient information to evaluate a bank’s solvency and may rationally withdraw funding at a hint of trouble.40 The resulting contraction of liquidity can cripple credit intermediaries and, consequently, the entire economy. Different types of regulatory measures are used to contain moral hazard and reduce the risk of significant negative externalities.41 Activities restrictions are intended either to keep insured depository institutions and their affiliates out of especially risky areas, to constrain banks’ use of insured deposits to compete with businesses lacking access to similarly subsidized resources, or both. For similar reasons, where insured depository institutions are allowed to affiliate with other financial firms, such as broker-dealers, structural regulation limits transactions between them. As activities and structural measures were relaxed in the three decades preceding the Global Financial Crisis of 2007-2009, prudential regulation has come to rely more on capital and liquidity regulation.42 Capital provides a buffer against losses arising from all assets and activities, regardless of whether bankers and 39 For example, other things being equal, an identical group of one thousand consumer auto loans will be less risky for a bank with little other exposure to asset classes sensitive to increases in the unemployment rate than for one already heavily concentrated in such assets. 40 MARCO CIPRIANI & GABRIELE LA SPADA, FED. RSRV. BANK OF N.Y. STAFF REP. NO. 956, SOPHISTICATED AND UNSOPHISTICATED RUNS 4 (2020) https://www.newyorkfed.org/medialibrary/media/research/staff_reports/sr9 56.pdf [https://perma.cc/W96F-3LW8] (discussing how sophisticated investors tend to be the ones who run pre-emptively). 41 See BARR et al., supra note 28, at 194–248. 42 See DANIEL K. TARULLO, BANKING ON BASEL: THE FUTURE OF INTERNATIONAL FINANCIAL REGULATION 33–35 (2008). No. 1:279] BANK SUPERVISION AND ADMINISTRATIVE LAW 295 regulators correctly anticipated the greatest sources of risk. Because of moral hazard considerations and the negative externalities associated with a large incapacitated bank, socially optimal capital levels will generally be higher than levels optimal for the bank’s shareholders. Among the clearest lessons from the Crisis was that prevailing capital requirements had been woefully inadequate, both in conception and in enforcement.43 Liquidity regulation, which is best understood as complementing capital regulation, attempts to moderate the temptation of banks to rely more and more on short-term funding, which is less expensive in good times but always more susceptible to runs.44 The same opacity of bank balance sheets and the same potential for fast-moving changes in funding, lending, and investing that motivate much of bank regulation suggest that exclusive reliance on regulatory rules will not be the most effective or efficient means to achieve prudential goals. Contemporary banking supervision serves five functions that augment and complement regulatory rules. First, regular examinations and ad hoc inquiries facilitate the enforcement of regulatory rules. The opaqueness of bank balance sheets means that external monitoring alone may not identify certain compliance problems, such as a decline in borrowers’ ability to repay. Second, supervisors can provide bank boards and senior management an informed, independent assessment of their own banks’ practices. They can also help disseminate best practices in risk management and related areas without compromising the proprietary information that may be associated with the bank(s) that pioneered these practices. 43 BASEL COMMITTEE ON BANKING SUPERVISION, BASEL III: A GLOBAL REGULATORY FRAMEWORK FOR MORE RESILIENT BANKS AND BANKING SYSTEMS 2–4 (2011). 44 For more on the risks in short-term runs, see ANTOINE MARTIN, DAVID SKEIE, & ERNST-LUDWIG VON THADDEN, FED. RSRV. BANK OF N.Y. STAFF REPORT NO. 444, REPO RUNS (2010), https://www.newyorkfed.org/medialibrary/media/research/staff_reports/sr4 44.pdf [https://perma.cc/4JQL-P9ZK]. 296 COLUMBIA BUSINESS LAW REVIEW [Vol. 2022 Third, supervisors stop banks from arbitraging regulatory rules. Regulatory arbitrage of categories established in quantitative capital, liquidity, and other rules is a continuing enterprise by banks, their consultants, and their lawyers. Supervisory oversight can detect when banks have changed the composition of their assets and liabilities so as to obtain more favorable regulatory treatment without any reduction in actual risk to the institution.45 Fourth, supervision can be an efficient way to fill in the lacunae in banking rules, which are inevitable when even quite detailed rules are applied to the varied and often complex activities of large banks. Risks in the banking system can be quite heterogeneous. The eight U.S. banks of “global systemic[] importan[ce]”46 have diverse business models and thus different risks to revenue and asset quality. Good supervision can identify and take relatively quick mitigating action in ways targeted to the particular risks posed. Regulatory rules could attempt to address the wide range of risks, including those not specifically anticipated. However, because many risks are hard to foresee and potential deviations from prudent practice are legion, the rules might have to be much more stringent and much blunter to capture all these contingencies. The result would be greater costs to lending and other financial activities than with supervision, and thus a less efficient means for achieving a comparable level of prudential protection. Alternatively, supervisors could respond to any perceived shortcoming in a bank’s risk management by requiring it to hold more capital. Again, though, this response might be costlier for prudent financial intermediation than improvements in risk management. 45 Can detect certainly does not necessarily mean will detect. Moreover, what is referred to by banks as regulatory “optimization” includes not just arbitrage of the sort described in the text, but also changes in the composition of bank assets and liabilities that reduce regulatory charges by lowering risk—precisely what the regulators intended. 46 FIN. STABILITY BD., 2021 LIST OF GLOBAL SYSTEMICALLY IMPORTANT BANKS (G-SIBS) (2021), https://www.fsb.org/wp- content/uploads/P231121.pdf [https://perma.cc/Y87D-8ZZC]. No. 1:279] BANK SUPERVISION AND ADMINISTRATIVE LAW 297 Similarly, the fifth supervisory function is to identify bank innovations or shifts in activities carrying risks that were not contemplated in the drafting of existing regulatory rules. Depending on the circumstances, the appropriate response will be supervisory guidance to one or a small number of banks, industry-wide supervisory guidance, an enforcement action, an early start on a new regulatory rule, or some combination of these measures. The fast-moving potential of financial activity will often counsel one or more supervisory measures during what may be an extended period of agency analysis of the emerging bank practice and consequent rule- making process. In theory, the optimal approach to prudential regulation should be a mix of generally applicable rules and of supervisory oversight that necessarily entails considerable discretion.47 The right mix of regulatory rules and supervisory oversight will vary over time and with the nature of bank risks. Similarly, the intensity of supervisory oversight— especially the relative number and scope of supervisory demands on banks—will vary across banks. While the first two supervisory functions—enforcement and independent assessment—are important, they will not usually raise issues at the intersection of supervision and administrative law. A simple example can illustrate how the last three supervisory functions—monitoring arbitrage of regulatory rules and addressing gaps and new situations not previously contemplated—complement regulatory (or statutory) rules and can implicate issues of discretion and administrative formality. One of the best-established 47 For a somewhat different, but congruent, assessment of the efficiency potential of supervision, see Kevin J. Stiroh, Exec. Vice President, Fed. Rsrv. Bank of N.Y., Remarks at Bank Regulation, Lending and Growth, The Bank Policy Institute and Columbia University’s School of International and Public Affairs, New York City: Policy Efficiency in Supervision (Mar. 1, 2019), https://www.newyorkfed.org/newsevents/speeches/2019/sti190301 [https://perma.cc/4AH9-KBD6]; see also BEVERLY HIRTLE & ANNA KOVNER, FED. RSRV. BANK OF N.Y. STAFF REPORT NO. 952, BANKING SUPERVISION: THE PERSPECTIVE FROM ECONOMICS, (2020), https://www.newyorkFederal Reserve.org/research/staff_reports/sr952 [https://perma.cc/8FR6-V3QZ]. 298 COLUMBIA BUSINESS LAW REVIEW [Vol. 2022 supervisory practices is scrutiny of the risk management process accompanying an extension of credit by a bank. Since the late 1980s, U.S. banks have been subject to risk-weighted minimum capital requirements, which are formalized as regulatory rules.48 As the term implies, these requirements assign a risk-weight to each asset or class of assets on a bank’s balance sheet. The minimum loss-absorbing capital that the bank must maintain is set at a percentage of the total of the bank’s risk-weighted assets.49 The assignment of risk weights and specification of a minimum percentage capital level in regulatory rules require the exercise of judgment by the banking agencies. Loss functions and minimum capital levels have been set based on historical loss experience adjusted for observed trends in the volatility of asset values, creditworthiness of identifiable classes of borrowers, and other factors.50 The requirements 48 Joseph G. Haubrich, A Brief History of Bank Capital Requirements in the United States, ECON. COMMENT. Feb. 28, 2020, at 1, 2, https://www.clevelandfed.org/newsroom-and-events/publications/economic- commentary/2020-economic-commentaries/ec-202005-evolution-bank- capital-requirements.aspx (on file with the Columbia Business Law Review). 49 To illustrate using a simplified example, single-family mortgages may be risk-weighted at 50%, unsecured loans to corporations at 100%, and Treasury bills at 0%. In such a case, the risk-weighted value of a $300,000 mortgage would be $150,000, a $300,000 corporate loan would be $300,000, and $300,000 of Treasury bills would be 0. If that percentage were, say, 7%, then the minimum level of bank capital (basically the difference between the value of all the bank’s assets and all its liabilities) required for this stylized balance sheet would be $31,500 (($150,000 + $300,000 + 0 = $450,000) x .07 = $31,500). 50 The empirical basis for the first minimum risk-weighted capital requirements, as established in 1988, is difficult to discern. Post-Crisis requirements, on the other hand, were developed after regulators calculated losses actually suffered by banks during the Crisis Period. See Beverly Hirtle, How Were the Basel 3 Minimum Capital Requirements Calibrated?, FED. RSRV. BANK OF N.Y.: LIBERTY ST. ECON. (Mar. 28, 2011), https://libertystreeteconomics.newyorkFederal Reserve.org/2011/03/calibrating-regulatory-minimum-capital- requirements.html [https://perma.cc/9PLQ-7PHY]; BASEL COMM. ON BANKING SUPERVISION, CALIBRATING REGULATORY MINIMUM CAPITAL No. 1:279] BANK SUPERVISION AND ADMINISTRATIVE LAW 299 are not calibrated to cover all conceivable losses associated with a bank’s assets and activities, present and future. If a bank is disproportionately making the riskiest kinds of loans within various asset categories, which by assumption will likely result in losses well above historical norms for that asset class, then the risk-weight assigned to that form of loan in the regulatory rule will not be adequate to cover the losses that may result if the asset class is subject to an unexpected shock.51 Here is where supervision comes in, as explained in the next Section. Although the expressed views of supervisors as to how a bank should alter its practices are not legally binding without some type of enforcement action, they are usually respected to at least some degree.52 While the reasons for this respect for non-binding direction are difficult to document, they seem to rest on a combination of three considerations that may be more or less weighty in particular instances. First is the possibility of significant adverse legal consequences if a formal enforcement action follows the failure of a banking organization to respond to supervisory concerns. Second is the potential for reputational damage with customers, investors, and politicians that can accompany a publicized dispute with supervisors. Third, because the bank-supervisor relationship involves repeated interaction, bank management will usually want to avoid getting on the wrong side of its supervisors, even when the supervisory direction pertains to minor matters. A bad relationship with supervisors can have a number of consequences unfavorable for the bank, such as more extended scrutiny of bank operations, lower supervisory ratings (which themselves carry consequences), or delays in approval of bank applications for acquisitions or various other activities for which agency REQUIREMENTS AND CAPITAL BUFFERS: A TOP-DOWN APPROACH (2010), https://www.bis.org/publ/bcbs180.pdf [https://perma.cc/XC9A-FS7J]. 51 See TARULLO, supra note 42, at 80 (describing capital arbitrage). 52 Knowledge of banks’ reaction to supervisory guidance is based on the author’s personal experience working at the Federal Reserve. 300 COLUMBIA BUSINESS LAW REVIEW [Vol. 2022 permission is required.53 Whether or not supervisory pronouncements are “practically binding” in the sense that some courts and scholars have used that term,54 they usually influence bank practices. Still, banks will often question, sometimes resist, and occasionally outright ignore supervisory guidance. It is worth noting that the dynamics reflected in the preceding explanations for bank compliance with supervisory expectations are in some ways preferable for banks, relative to a hypothetical world in which there was no bank supervision as we know it—only regulatory rules and enforcement.55 The supervisory process can be an efficient way to clarify the practices that supervisors will regard as satisfactory and thereby reduce regulatory uncertainty.56 It can also help banks head off problems that might develop into the objects of enforcement actions, with the assurance that no formal enforcement action will be forthcoming as long as a remediation plan is well implemented.57 B. The Organization of Bank Supervision The specifics of supervisory organization and practice vary somewhat across the three federal banking agencies, partly because the agencies have somewhat different responsibilities 53 As has been pointed out by those worried about supervisors being too lax, the fact of repeated interactions can cut in both directions. A dedicated supervisory team at a large bank has at least some incentive to maintain a smooth relationship with bank officials so as to facilitate its access to information from the bank. 54 See, e.g., Gen. Elec. Co. v. EPA, 290 F.3d 377, 383 (D.C. Cir. 2002); Cass Sunstein, “Practically Binding”: General Policy Statements and Notice- and-Comment Rulemaking, 68 ADMIN L. REV.491 (2016). 55 For a discussion of this point, see Guidance, Supervisory Expectations, and the Rule of Law: How Do the Banking Agencies Regulate and Supervise Institutions?: Hearing Before the S. Comm. on Banking, Hous., & Urban Aff., 116th Cong. 48–50 (2019) (prepared statement of Patricia A. McCoy, Professor of Law, B.C. L. Sch.), https://www.banking.senate.gov/imo/media/doc/McCoy%20Testimony%204 -30-19.pdf [https://perma.cc/S585-LFWZ]. 56 See id. at 49. 57 See id. No. 1:279] BANK SUPERVISION AND ADMINISTRATIVE LAW 301 and partly because similar functions rarely evolve identically in separate agencies. For example, while the FDIC has backup supervisory authority for all banks because of its status as the insurer of bank deposits and as the potential resolution authority for large bank holding companies, it is the primary federal regulator only for banks that are state-chartered and not members of the Federal Reserve System.58 The OCC supervises all nationally-chartered depository institutions, which run the gamut from community banks with less than a billion dollars in assets to the commercial bank subsidiaries of the four behemoths: J.P. Morgan Chase Bank, Bank of America, Wells Fargo, and Citibank, each of which has assets of between $1.6 and $3.3 trillion.59 The Federal Reserve is the primary federal regulator of both state-chartered banks that have joined the Federal Reserve System and of the holding companies of all insured depository institutions. As consolidated regulator, it has supervisory authority over the holding companies that own ninety percent of all insured depository institutions, including all the larger ones. Most small- and medium-sized banks with holding companies do not have sizeable non-bank affiliates such as broker-dealers.60 For these institutions, the Federal Reserve usually does little additional supervision at the consolidated level, relying instead upon the work of the OCC or FDIC (or, for state member banks, its own staff) in 58 As a result of the 2019 merger between BB&T and SunTrust, the resulting Truist bank is by far the largest under FDIC supervision, with assets in excess of $500 billion. See Federal Reserve Statistics Release, Large Commercial Banks, FED. RSRV. (Dec. 31, 2021), https://www.federalreserve.gov/releases/lbr/current/default.htm [https://perma.cc/WFC6-TJJU]. 59 See id. 60 Under the Bank Holding Company Act, an insured depository institution may generally affiliate with non-bank financial firms so long as they are separately incorporated within the holding company and certain limitations on transactions between them are observed. Bank Company Holding Act of 1956, 12 U.S.C. §§ 1842–1843 (2018). 302 COLUMBIA BUSINESS LAW REVIEW [Vol. 2022 supervising the insured depository institution that constitutes essentially all the activities of the consolidated firm.61 At the other end of the spectrum, most of the largest banks have extensive operations outside their insured depository institutions, including broker-dealers, asset managers, commodities brokers, wealth management units, and other financial firms. As explained below, the Federal Reserve has a much different and more elaborate supervisory structure for these firms.62 Notwithstanding the variations alluded to earlier, supervisory organization and practice across the three banking agencies are similar in most particulars that are relevant for this Article, except for the Federal Reserve’s oversight of the most significant holding companies. Banks are grouped into supervisory portfolios based on size and, to a somewhat lesser degree, complexity.63 For the largest banks, the agencies dedicate a team of supervisors exclusively to that bank. They thus have an essentially ongoing presence at the institution and engage in “continuous monitoring” of the firm.64 Specialists in various risk topics, such as credit, 61 Bd. of Governors of the Fed. Rsrv. Sys., SR 16-4, Relying on the Work of the Regulators of the Subsidiary Insured Depository Institution(s) of Bank Holding Companies and Savings and Loan Holding Companies with Total Consolidated Assets of Less than $50 Billion (Feb. 17, 2021), https://www.federalreserve.gov/supervisionreg/srletters/sr1604.htm [https://perma.cc/C999-QPVD]. 62 See supra notes 99–104 and accompanying text. 63 There are more formal divisions of portfolios at the OCC and Federal Reserve. The OCC has portfolios for large banks, midsize banks, federal branches and agencies of foreign banks, and community banks. The Federal Reserve has portfolios for large banks, foreign banks, regional banks, and community banks. The Federal Reserve also maintain a special portfolio for the largest, most systemically important institutions. See supra note 96– 104 and accompanying text. 64 “Continuous monitoring” is “intended to enable each firm-focused supervisory team to ‘develop and maintain an understanding of the organization, its risk profile, and associated policies and practices’ . . . as well as to identify gaps or issues that might lead the team to do more in- depth analysis.” THOMAS EISENBACH ET AL., FED. RSRV. BANK OF N.Y., SUPERVISING LARGE, COMPLEX FINANCIAL INSTITUTIONS: WHAT DO SUPERVISORS DO? (2015), No. 1:279] BANK SUPERVISION AND ADMINISTRATIVE LAW 303 market, operational, compliance, support the dedicated teams of supervisors. For mid-sized and small banks, there are no dedicated teams of supervisors. Instead, each bank is subject to an on-site examination every twelve or eighteen months.65 Banks of all sizes must file regular reports with their supervisors, which form part of the basis for off-site monitoring.66 The banking agencies also conduct targeted exams, focusing on a particular form of activity or risk, which may apply to anything from a single bank to a whole portfolio of banks at one (or sometimes all three) of the agencies.67 The three federal banking agencies publish detailed manuals for use by supervisors. These manuals contain an explanation of the laws and regulatory rules to which the banks are subject and often instruct supervisors on how to check for compliance with those requirements.68 The banking https://www.newyorkfed.org/medialibrary/media/research/epr/2017/epr_20 17_what-do-supervisors-do_eisenbach.pdf?la=en. [https://perma.cc/U94Y- 3KWB]. 65 Banks with less than $3 billion in assets that are well-capitalized and well-managed are eligible for the longer eighteen-month examination cycle. All other banks musts be examined every twelve months. 12 U.S.C. § 1820(d). 66 See Aaron Zabler, Safety and Soundness Offsite Examination Work: Safety and Soundness Update, FED. RSRV. BANK OF MINNEAPOLIS (Sept. 15, 2014), https://www.minneapolisfed.org/article/2014/safety-and-soundness- offsite-examination-work [https://perma.cc/FT4R-UH66]. 67 One regular interagency targeted exam is the Shared National Credits Program (SNC), whereby the three banking agencies coordinate in assessing credit risk trends and risk management practices associated with the largest and most complex credits shared by multiple regulated financial institutions. For a report of the outcome of a recent SNC exercise, see BD. OF GOVERNORS OF THE FED. RSRV. SYS, FED. DEPOSIT INSR. CORP. & OFF. OF THE COMPTROLLER OF THE CURRENCY, SHARED NATIONAL CREDITS PROGRAM: 3RD QUARTER 2016/1ST QUARTER 2017 EXAMINATIONS (2017), https://www.federalreserve.gov/newsevents/pressreleases/files/bcreg20170 802a1.pdf. [https://perma.cc/4P69-JYCV]. 68 See Supervision Manuals, BD. OF GOVERNORS OF THE FED. RSRV. SYS. (last updated Nov. 30, 2021), https://www.federalreserve.gov/publications/supmanual.htm [https://perma.cc/C54Q-7S9U]; Comptroller’s Handbook, OFF. OF THE COMPTROLLER OF THE CURRENCY, occ.treas.gov/publications-and- resources/publications/comptrollers-handbook/index-comptrollers- 304 COLUMBIA BUSINESS LAW REVIEW [Vol. 2022 agencies also periodically publish supervisory guidance,69 which is frequently directed to the banks and the field personnel supervising the banks. Guidance can take different forms. It may specify how supervisors are expected to monitor implementation or compliance by banks with applicable regulations.70 It may also draw attention to observed trends in a banking activity that raise safety and soundness concerns and indicate ways in which potential risks might be avoided.71 handbook.html [https://perma.cc/U9VU-K785] (last visited Mar. 7, 2022); Consumer Compliance Examination Manual, FED. DEPOSIT INS. CORP. (last updated Aug. 16, 2021), https://www.fdic.gov/resources/supervision-and- examinations/consumer-compliance-examination-manual/index.html [https://perma.cc/E92B-E766]. 69 The banking agencies publish their guidance under different headings. The Federal Reserve has a special publication series, “Supervision and Regulation Letters” (referred to as “SR Letters”). See Supervision and Regulation Letters, BD. OF GOVERNORS OF THE FED. RSRV. SYS. (last updated Nov. 2, 2020), https://www.federalreserve.gov/supervisionreg/srletters/srletters.htm [https://perma.cc/AX4Z-QHBP]. The OCC and the FDIC release their supervisory guidance in publication series that contain other information of interest to banks (e.g., proposed or final regulations, notices of changes in organization), labelled “OCC Bulletins” and “Financial Institution Letters,” respectively. See Weekly Bulletin, OFF. OF THE COMPTROLLER OF THE CURRENCY, https://www.occ.gov/topics/charters-and-licensing/weekly- bulletin/index-weekly-bulletin.html [https://perma.cc/3CTS-AJ4S]; Financial Institution Letters, FED. DEPOSIT INS. CORP., https://www.fdic.gov/news/financial-institution-letters/index.html [https://perma.cc/GVX5-P6KS]. 70 For an example of guidance on how supervisors are expected to monitor implementation, see Bd. of Governors of the Fed. Rsrv. Sys., SR 14- 4, Examiner Loan Sampling Requirements for State Member Bank and Credit Extending Nonbank Subsidiaries of Bank Holding Companies in the Regional Banking Organization Supervisory Portfolio (revised Dec. 6, 2021), https://www.federalreserve.gov/supervisionreg/srletters/sr1404.htm [https://perma.cc/WY78-NCDJ]. 71 For an example of how guidance draws attention to observed trends, see Bd. of Governors of the Fed. Rsrv. Sys., SR 21-19, The Federal Reserve Reminds Firms of Safe and Sound Practices for Counterparty Credit Risk Management in Light of Archegos Capital Management Default (Dec. 10, 2021), https://www.federalreserve.gov/supervisionreg/srletters/SR2119.htm [https://perma.cc/4LNL-VJ2G]. No. 1:279] BANK SUPERVISION AND ADMINISTRATIVE LAW 305 Proposed guidance that is not time-sensitive and is intended to remain in place indefinitely as part of the supervisory framework is generally, thought not always, published in the Federal Register for public comment before being finalized. Supervisors have a variety of ways to penetrate opaque bank balance sheets and thereby identify problems that could pose a threat to the safety and soundness of a bank.72 Similarly, when a bank enters a new activity or offers a new product, or when an existing activity or asset class is growing quickly, a lack of experience or capacity at the bank can lead to losses that would not generally be expected from a bank that is well-established in the product or activity. Here the supervisory emphasis will be upon assessing the bank’s ability to proceed without undue risks to safety and soundness. When shortcomings in bank practices are identified, there are various supervisory responses, depending on the specifics of the problem and, in at least some cases, on a bank’s own preferences. Often the supervisors will send the bank a notice of a Matter Requiring Attention (MRA) or, less frequently, a Matter Requiring Immediate Attention (MRIA).73 As we will see later, the number and scope of MRAs issued in the post- Crisis period became controversial with banks and, in fact, with some senior supervisors.74 72 For example, they can test a selection of loan files to see if the underwriting process followed by a bank is sufficiently rigorous. They can examine the bank’s risk management policies and processes more generally. They can watch for early signs of abnormally high levels of non-performing loans. They can evaluate the loans or other assets to be sure those assets are being correctly categorized for purposes of risk-weighting. These matters are often quite specific to a bank (though banks are often found to share certain shortcomings in, for example, the underwriting process). 73 BD. OF GOVERNORS OF THE FED. RSRV. SYS., SUPERVISION AND REGULATION REPORT 16 (2018), https://www.federalreserve.gov/publications/files/201811-supervision-and- regulation-report.pdf [https://perma.cc/XB9Y-BJFS]. For my illustration in supra note 72 involving underwriting practices, these written notices would instruct the bank to correct whatever deficiencies have been identified in the underwriting or risk management processes. 74 See supra Section III.C. 306 COLUMBIA BUSINESS LAW REVIEW [Vol. 2022 Another form of supervisory communication is a Memorandum of Understanding (MOU) with the bank.75 An MOU usually reflects a supervisory concern that has persisted for some time and a judgment that outstanding MRAs have not been satisfactorily addressed. While an MRA draws attention to a problem and gives a general indication of what the bank needs to do to address that problem, a MOU will often contain commitments for quite specific remedial actions. Failure to implement these actions in a timely fashion can lead to agency initiation of proceedings to issue a cease-and- desist order which, for sufficiently serious problems, may also be issued without prior MRAs. Unlike the three kinds of supervisory communication already mentioned, a cease-and- desist order is public and legally binding. It is issued pursuant to express statutory authority76 and subject to the familiar procedures for written notice, hearing, and judicial review contained in the APA.77 All banking organizations receive supervisory ratings each year (or, for smaller banks qualifying for less frequent exams, every eighteen months). For insured depository institutions, banking agencies use the CAMELS system, developed by the Federal Financial Institutions Examination Council (FFIEC). The FFIEC, an interagency entity composed of principals from the agencies that regulate depository institutions,78 was created as part of the Financial Institutions Regulatory and Interest Rate Control Act of 1978.79 The legislation was 75 FED. DEPOSIT INS. CORP., INFORMAL ACTIONS, RMS MANUAL OF EXAMINATION POLICIES § 13.1, https://www.fdic.gov/regulations/safety/manual/section13-1.pdf [https://perma.cc/S8GK-3H6Y]. 76 12 U.S.C. § 1818(b) (2018). 77 Pub. L. No. 79-404, 60 Stat. 237 (1946) (codified at 5 U.S.C. § 551 et seq.). 78 At present, the FFIEC consists of the Comptroller of the Currency, the Chair of the FDIC, a Federal Reserve Governor, the Director of the Consumer Financial Protection Bureau, the Chair of the National Credit Union Administration, and the Chair of a State Liaison Committee created by the legislation. 12 U.S.C. § 3303(a). 79 Pub. L. No. 95-630, § 1004, 92 Stat. 3641, 3694 (codified at 12 U.S.C. § 3303). No. 1:279] BANK SUPERVISION AND ADMINISTRATIVE LAW 307 passed against the backdrop of an increase in bank failures, which proponents attributed in part to the absence of uniform standards and cooperation among the regulatory agencies, including a “[c]ompetition in laxity.”80 The FFIEC was instructed to “prescribe uniform principles and standards for the federal examination of financial institutions” by the federal financial regulatory agencies and “make recommendations to promote uniformity in the supervision of these financial institutions.”81 Shortly after its creation, the FFIEC adopted the CAMEL system, technically known as the Uniform Financial Institutions Rating System. The system was significantly, though not fundamentally, revised in 1996.82 The Federal Reserve and the FDIC have solicited 80 S.REP. 95-1273, at 6 (1978); see Bd. of Governors of the Fed. Rsrv. Sys., SR 19-3, Large Financial Institution (LFI) Rating System (Feb. 26, 2019), https://www.federalreserve.gov/supervisionreg/srletters/sr1903.htm [https://perma.cc/2N3D-3V8A]; Large Financial Institution Rating System; Regulations K and LL, 83 Fed. Reg. 58,724 (Nov. 28, 2018) (to be codified at 12 C.F.R. pts. 211, 238). Other parts of the legislation expanded the cease and desist authority of the agencies to reach officers and directors of banking organizations, as well as the banks themselves. For a discussion, see Alan F. Garrison, Note, The Financial Institutions Regulatory and Interest Rate Control Act of 1978, Federal Banking Agencies, and The Judiciary: The Struggle To Define the Limitation of Cease and Desist Order Authority, 44 WASH. & LEE L. REV. 1357 (1987). 81 Pub. L. No 95-630, § 1002, 92 Stat. 3641, 3694 (codified at 12 U.S.C. § 3301). 82 In addition to modifying a number of descriptions and definitions, the 1996 changes added the “S” in the form of the component of Sensitivity to Market Risk and increased the emphasis on risk management practices in each component, particularly Management. These changes were made through a notice-and-comment procedure. Uniform Financial Institutions Rating System, 61 Fed. Reg. 37,472 (notice July 18, 1996); Uniform Financial Institutions Rating System, 61 Fed. Reg. 67,021 (notice Dec. 19, 1996). While notice-and-comment procedures were followed, FFIEC did not characterize its action as a rulemaking. Its authorizing statute is ambiguous as to the extent of rule-making authority by FFIEC itself (as opposed to its constituent agencies). The banking agencies quickly implemented the new standards, which remain in place today. The three banking agencies quickly adopted the new standards to be effective January 1, 1997, but did so in different ways. The FDIC published the text of the revised FFIEC standards in the Federal Register, characterizing it as a “policy statement.” 308 COLUMBIA BUSINESS LAW REVIEW [Vol. 2022 public comment on the use and impact of ratings, which could be a prelude to another significant revision.83 Under the revised CAMELS system, which added an “S” to the old CAMEL system, examiners rate Capital, Asset Quality, Management, Earnings, Liquidity, and Sensitivity to Market Risk on a one-to-five scale. Each bank is given a composite rating on the same scale. The ratings process differs somewhat among the three banking agencies, and among portfolios within agencies. Roughly speaking, though, the relevant team—whether one with a continuing presence in a large bank or a group assembled for an annual examination— recommends ratings, which are then subject to some form of review by senior supervisory officials in the regional offices of the agencies (or, in the case of the Federal Reserve, the regional Reserve Banks) and Washington. The Federal Reserve has two rating systems for bank holding companies: one for most holding companies and one for “Large Financial Institutions” (LFI), which are generally those with more than $100 billion in assets. The first is the RFI/C(D) rating system, established in 2004, which uses the same numerical ranking as CAMELS: The components are Risk Management, Financial Condition, and Potential Impact of the non-bank affiliates on the depository institution(s) Uniform Financial Institutions Rating System, 62 Fed. Reg. 752 (notice Jan. 6, 1997). The Federal Reserve simply issued a brief press release noting its adoption of the new standards, which were attached to the release. Press Release, Bd. of Governors of the Fed. Rsrv. (Dec. 24, 1996), https://www.occ.gov/news-issuances/news-releases/1997/nr-occ-1997- 29.html [https://perma.cc/6WQ5-LQAQ]. The OCC does not appear to have issued any public notice at all, though in March 1997, it issued a Q & A clarifying some issues that had been raised by the new system, which it noted had been in effect since January 1. News Release 1997-29, Off. of the Comptroller of the Currency, Common CAMELS Questions Answered (Mar. 17, 1997), https://www.occ.gov/news-issuances/news-releases/1997/nr-occ- 1997-29.html. [https://perma.cc/4DSL-DCA3]. 83 Press Release, Bd. of Governors of the Federal Rsrv. Sys. & Fed. Deposit Ins. Corp., FDIC and Federal Reserve Request Information on Use and Impact of CAMELS Ratings (Oct. 18, 2019) https://www.federalreserve.gov/newsevents/pressreleases/bcreg20191018a. htm. [https://perma.cc/57S2-CMGT]. No. 1:279] BANK SUPERVISION AND ADMINISTRATIVE LAW 309 within the holding company.84 The composite rating also incorporates the rating assigned to the relevant depository institution(s) by primary Federal Reserve regulator(s). The LFI system, adopted in November 2018,85 is still being phased in. The three components—Capital Planning and Positions, Liquidity Risk Management and Positions, and Governance and Controls—are meant to better track the changes in supervision of the largest banking organizations since the financial crisis.86 Notable is the emphasis on capital and liquidity, which are each the subject of regular supervisory testing across these large firms. The Federal Reserve has decided to use four non-numeric ratings for the LFI system— Broadly Meets Expectations, Conditionally Meets Expectations, Deficient-1, and Deficient-2.87 At first glance, this shift from numeric to verbal ratings may look like the euphemistic substitution of verbal formulations for letter grades at some academic institutions. However, as is apparent from the Federal Reserve’s explanation of the change, the new formulations—and the nature of the supervisory conclusions conveyed with them— are intended to provide more clarity to banking organizations as to what, if any, remedial action is expected of them.88 This distinction is especially relevant for larger firms, which are 84 The Risk Management and Financial Condition each have four subcomponent ratings. The Risk Management they sub-components are Board and Senior Management Oversight; Policies, Procedures, and Limits; Risk Monitoring and Management Information Systems; and Internal Controls. The Financial Condition sub-components are Capital Adequacy, Asset Quality, Earnings, and Liquidity. See Bd. of Governors of the Fed. Rsrv. Sys., SR 19-4 (Attachment #2), RFI Rating System (Feb. 26, 2019), https://www.federalreserve.gov/supervisionreg/srletters/sr1904a2.pdf [https://perma.cc/Q3EL-9ZY9]. 85 Large Financial Institution Rating System; Regulations K and LL, 83 Fed. Reg. 58,724 (Nov. 21, 2018) (to be codified at 12 C.F.R. pts. 211, 238). 86 See id. at 58,724. 87 Id. 58,729. 88 See Bd. of Governors of the Fed. Rsrv. Sys., SR 19-3, Large Financial Institution (LFI) Rating System, supra note 80. 310 COLUMBIA BUSINESS LAW REVIEW [Vol. 2022 evaluated on a far broader and deeper set of expectations than smaller banks. In providing a formalized and, either literally or effectively, quantified supervisory view of the condition of banks, the ratings act as a kind of grading system. They are conveyed to management and boards of directors of the banks, along with an explanation of the reasons for the ratings. They are used internally to track the overall health of and identify trends in the industry. They can also help benchmark efforts to apply generally consistent evaluations of banking organizations. Consequences of two sorts ensue from low ratings. First, more supervisory attention is usually directed at lower-rated banks, often including MRAs or other communication of processes or conditions at the bank that need to be improved. Second, specific regulations89 and statutory provisions90 allow, or in some cases require, stricter supervisory treatment for banks with lower composite ratings. The Federal Reserve’s explanation of the new LFI system makes clear which ratings 89 For example, in order for a small bank to qualify for less frequent on- site examination, it must be rated a 1 or 2. 12 C.F.R. §§ 208.64(b)(3), 337.12(b)(3) (2021); see also id. § 5.51(b)(7)(i) (national banks with composite rating of 4 or 5 must notify OCC of changes in directors or senior executive officers, actions that may be disapproved by OCC). 90 Perhaps the most important of these is 12 U.S.C. § 1841(o)(9)(A)(i) (2018), which defines “well managed” as “a CAMEL composite of 1 or 2 (or equivalent rating under an equivalent rating system.”). All insured depository institutions of a bank holding company must be “well managed” in order for that BHC to be able to own certain kinds of non-bank financial firms, such as those conducting investment banking or private equity activities. Id. § 1843(l)(1). A banking organization that has previously qualified as a “financial holding company” and then ceases to be well- capitalized or well-managed is subject to a so-called 4(m) agreement, which requires correction of the deficiencies that led to the lower rating (or breach of another statutory requirement). Until the corrections are made, the Federal Reserve “may impose such limitations on the conduct or activities of that financial holding company or any affiliate of that company as the Board determines to be appropriate under the circumstances and consistent with the purposes of this [Act].” Id. § 1843(m). No. 1:279] BANK SUPERVISION AND ADMINISTRATIVE LAW 311 will lead to these statutory consequences.91 The FDIC also relies significantly on ratings in implementing the statutory requirement that deposit insurance be risk-based.92 Examination reports and supervisory ratings are not published, either for individual banks or in the aggregate.93 Indeed, banks may not release their own ratings without the permission of their supervisor.94 The ratings process has revealed a kind of supervisory procyclicality that paralleled bank lending behavior. In the years leading up to the Crisis, ratings were generally high.95 Following the onset of the Crisis and the Great Recession that followed, bank ratings were set substantially lower. While this shift is in part explicable by the obvious fact that the balance sheets of all banks had been adversely affected by the severe economic conditions, the management and risk management capacities of banks did not change dramatically between mid-2007 and 91 See Large Financial Institution Rating System, Regulations K and LL, 83 Fed. Reg at 58,730. 92 E.g., 12 C.F.R. § 327.9(a)(4). The statutory requirement for a risk- based premium system is in 12 U.S.C. § 1817(b)(1). 93 In its Quarterly Banking Profile, the FDIC identifies the number of “problem banks” and their aggregate assets. FED. DEPOSIT INS. CORP., QUARTERLY BANKING PROFILE: FOURTH QUARTER 2021, at 31 (2021), https://www.fdic.gov/analysis/quarterly-banking- profile/qbp/2021dec/qbp.pdf [https://perma.cc/39SV-3K6K]. Problem banks are those rated 4 or 5. Id. There have been a few studies undertaken and then published by the regulatory agencies themselves on trends in past ratings. See, e.g., William Bassett, Seung Jung Lee & Thomas Popeck Spiller, Estimating Changes in Supervisory Standards and Their Economic Effects, J. BANKING & FIN., Nov. 2015, at 21. 94 Off. of the Comptroller of the Currency, OCC 2019-15: Supervisory Ratings and Other Nonpublic OCC Information: Statement of Confidentiality (Mar. 25, 2019), https://www.occ.gov/news- issuances/bulletins/2019/bulletin-2019-15.html [https://perma.cc/U6YY- ELER]. 95 Knowledge of banks’ ratings is based on the author’s personal experience working at the Federal Reserve. Cf. FED. DEPOSIT INS. CORP., QUARTERLY BANKING PROFILE: FOURTH QUARTER 2009, at 17 tbl.II-B (2009), https://www.fdic.gov/analysis/quarterly-banking- profile/qbp/2009dec/qbp.pdf [https://perma.cc/7ZVA-JJ7P] (showing the number of “problem” deposit insurance funds rose from 76 in 2007 to 702 in 2009). 312 COLUMBIA BUSINESS LAW REVIEW [Vol. 2022 late 2008. Instead, supervisors may have realized that they had not adequately assessed the shortcomings in risk management and other characteristics of banks and so appropriately rated many banks lower. In at least some cases, supervisors may have erred on the side of lower ratings in response to widespread criticism of pre-Crisis supervision. In the years following the Crisis, the most frequent (though not the only) complaint from bankers had less to do with the initial downgrades than with what they perceived to be the frustrating vagueness in supervisory standards for banks to follow to restore their higher ratings. Because the ratings, examination reports, and some other forms of supervisory communications are not publicly available, there are relatively few studies of the effectiveness of supervision. Those that assess supervisory effectiveness directly tend to come from researchers at bank regulatory agencies with access to supervisory information not available to outsiders.96 A study from researchers at the Federal Reserve Bank of New York concluded that firms supervised in a relatively intense fashion as compared to roughly similar firms, have lower volatility of earnings and market returns, hold less risky loan portfolios, and engage in more conservative loan loss reserving practices.97 One of the more 96 In order to protect firm-specific supervisory information, these papers typically aggregate that information, so independent evaluation of these studies is not easy. In recent years academic researchers have used some surrogates for supervisory intensity in studying the impact of supervision. A review of the literature on supervision is included in HIRTLE & KOVNER, supra note 47. 97 BEVERLY HIRTLE, ANNA KOVNER & MATTHEW PLOSSER, FED. RSRV. BANK OF N.Y. STAFF REP. NO. 768, THE IMPACT OF SUPERVISION ON BANK PERFORMANCE 2, 47 (2016), https://www.newyorkfed.org/medialibrary/media/research/staff_reports/sr7 68.html [https://perma.cc/Y4SX-SR6T]. This study was updated to consider the effects of supervision over time, especially during banking industry downturns. The principal findings of the initial study were confirmed. See Uyanga Byambaa et al., How Does Supervision Affect Bank Performance During Downturns?, FED. RSRV. BANK OF N.Y.: LIBERTY ST. ECON. (Apr. 8, 2020), https://libertystreeteconomics.newyorkfed.org/2020/04/how- does-supervision-affect-bank-performance-during-downturns [https://perma.cc/3BGK-UVH4]. No. 1:279] BANK SUPERVISION AND ADMINISTRATIVE LAW 313 notable findings was that the lower risk profile of the more intensively supervised firms was not accompanied by lower returns or asset growth.98 While there are some variations in the mode of supervision of the different portfolios of firms by the banking agencies, the Federal Reserve takes a different and more intense approach in its supervision of the eight U.S. banking organizations that have been designated as being of “global systemic importance.”99 This new approach began with the creation of the Large Institution Supervision Coordinating Committee (LISCC) following the Financial Crisis.100 Supervision of the LISCC firms differs in several respects from that of even the largest regional banking organizations such as US Bank, Truist, and PNC. First, there is considerably more coordination of ongoing supervisory activities, which is achieved through various committees 98 HIRTLE, KOVNER & PLOSSER, supra note 97, at 1–3, 29; Byambaa et al., supra note 97. 99 The eight U.S. firms are Bank of America, Bank of New York Mellon, Citigroup, Goldman Sachs, JP Morgan Chase, Morgan Stanley, State Street, and Wells Fargo. Large Institution Supervision Coordinating Committee, BD. OF GOVERNORS OF THE FED. RSRV. SYS. (Dec. 18, 2020), https://www.federalreserve.gov/supervisionreg/large-institution- supervision.htm [https://perma.cc/PQ7Q-SBVN]. Identification of banks as being of global systemic importance (“G-SIB”) is done through the joint efforts of the Financial Stability Board and the Basel Committee on Banking Supervision, both international groups of central bankers, regulators, and other government officials. See 2018 List of Global Systemically Important Banks (G-SIBs), FIN. STABILITY BD. (Nov. 16, 2018), http://www.fsb.org/wp-content/uploads/P161118-1.pdf [https://perma.cc/XQ2G-NMBG]. The Federal Reserve is not bound by this international identification process, but participates fully in it and has agreed with the list of U.S. banks so identified. Originally the U.S. operations of four foreign banks were also included in the LISCC portfolio. Those banks were removed in late 2020 and early 2021. Bd. of Governors of the Fed. Rsrv. Sys., SR 20-30, Financial Institutions Subject to the LISCC Supervisory Program (Mar. 31, 2021), https://www.federalreserve.gov/supervisionreg/srletters/sr2030.htm [https://perma.cc/ST2K-RCUC]. 100 The organization of the LISCC is described in Large Institution Supervision Coordinating Committee, BD. OF GOVERNORS OF THE FED. RSRV. SYS., supra note 99. 314 COLUMBIA BUSINESS LAW REVIEW [Vol. 2022 chaired by staff from the Board or Reserve Banks, but with substantial participation by supervisors from all Reserve Banks having responsibility for the LISCC firms. Supervisors undertake both recurring and ad hoc “horizontal reviews” of LISCC firms in areas of concern, meaning that the eleven firms are examined simultaneously using a single set of metrics or standards. An important motivation for creation of the LISCC was the Board’s conclusion that the pre-Crisis framework for supervision of these institutions had proved wanting.101 The dedicated supervisory teams had substantial autonomy from the Board staff, with little effective oversight. This situation had predictably led to substantial variation in the relative rigor and consistency of supervision at the largest banks. The absence of communication among dedicated teams and the failure to aggregate both quantitative and qualitative information about risks at those banks meant that the Federal Reserve had a very incomplete picture of the range of risks, both correlated and uncorrelated, that existed within these most important actors in the U.S. financial system. The creation of the LISCC removed a considerable amount of autonomy, especially from some of the previously quite independent teams based in the Federal Reserve Bank of New York. But, through the LISCC process, it also gave those teams and their superiors at the Reserve Banks a voice in the development of supervisory policies applicable to all firms in the portfolio. A second distinction between the LISCC and other supervisory portfolios is its leadership. The LISCC is composed of the most senior supervisory staff at the Board and the relevant Reserve Banks, along with senior staff from other Divisions of the Federal Reserve Board.102 This 101 See Jon Hilsenrath, Washington Strips New York Fed’s Power, WALL ST. J. (Mar. 4, 2015), https://www.wsj.com/articles/washington-strips-new- york-feds-power-1425526210 (on file with the Columbia Business Law Review). 102 Large Institution Supervision Coordinating Committee, BD. OF GOVERNORS OF THE FED. RSRV. SYS., supra note 99. No. 1:279] BANK SUPERVISION AND ADMINISTRATIVE LAW 315 “interdisciplinary” character of the LISCC103 is intended to introduce into the supervisory process the economic, financial, and institutional perspectives of other parts of the Federal Reserve System, so as to inform the formulation of supervisory policies for this group of a dozen highly systemic firms. Third, LISCC supervision is informed by the work of a Monitoring and Analysis Program that draws on a broad range of information sources beyond the supervisory process “to facilitate the identification and exploration of topics and risks related to LISCC firms that are new, changing, misunderstood, or underappreciated, to ensure that supervision of LISCC firms adapts to changes in the financial industry and broader economy.”104 The information produced by this process is used to complement supervisory observation of bank practices. Identifications of potential correlated vulnerabilities may lead to supervisory examination of the assets or activities in question. III. ADMINISTRATIVE LAW DOCTRINE AND SUPERVISORY ACTIVITIES This Part considers whether three important forms of supervisory activity are inconsistent with the APA and other applicable statutory procedural requirements: the use of stress testing in setting minimum capital requirements for large banks; the assignment of supervisory ratings to banks and bank holding companies; and supervisory guidance and other forms of supervisory communications. Although the possible vulnerability of these activities to administrative law challenges differs, three points emerge from the analyses taken as a whole. 103 Wall Street Reform: Oversight of Financial Stability and Consumer and Investor Protections: Hearing Before the Comm. on Banking, Hous. & Urb. Affs. on Examining the Agencies’ Overall Implementation of the Dodd- Frank Wall Street Reform and Consumer Protection Act, 113th Cong. 44 (2013) (statement of Daniel K. Tarullo, Governor, Bd. of Governors of the Fed. Rsrv. Sysm) (The LISCC “fosters interdisciplinary coordination”). 104 Id. 316 COLUMBIA BUSINESS LAW REVIEW [Vol. 2022 First, the difficulty in conclusively determining the status of some of these activities under current administrative law doctrines arises in part because those doctrines have been developed or applied without due regard for the distinctive nature of the quite dissimilar mandates Congress has given administrative agencies. Some assumptions underlying the doctrines simply do not fit the realities of bank supervision. Second, when the substantive framework established by Congress for banking regulation is considered alongside administrative law doctrines, the argument that most supervisory activities are procedurally consistent with applicable law becomes very strong. Indeed, it is notable that banking interests arguing on legal grounds against various supervisory actions have generally ignored relevant banking law provisions. Third, even if the analysis is appropriately tailored in this way, there may still be an uneasy fit between some generally applicable doctrine and the iterative nature of the bank supervisory relationship. In some cases, the result could be unnecessary and inefficient constraints on supervisory activity. In one case, the result is arguably too much procedural leeway for banking agencies. This third point leads to the discussion in Part IV of how administrative law might take a different approach that both reflects the characteristics of bank supervision and respects the norms and requirements in the APA. A. Procedural Requirements for a Binding Stress Test Regime Complaints from bank proxies about the Federal Reserve’s stress testing regime are in several respects distinct from most other concerns about supervision. First, they are the most extensively developed. A kind of informal brief making the administrative law arguments has been published by a group that can fairly be characterized as sympathetic to the industry.105 Second, because of the centrality of capital 105 In September 2016, the Wall Street Journal reported that a number of large banks were considering filing suit based on these arguments. Emily No. 1:279] BANK SUPERVISION AND ADMINISTRATIVE LAW 317 regulation and stress testing to post-Crisis prudential regulation, these complaints are the most direct challenge to the efficacy and efficiency of a bank regulatory system that combines rules and supervision. Third, the objects of the industry’s procedural complaints—the supervisory model used in the stress tests to calculate losses and the hypothesized adverse scenarios—are undoubtedly part of an administrative process intended to be binding on the banks. In this respect they differ from complaints about supervisory communications that the agency characterizes as legally non- binding but that banks believe are practically binding. These complaints about stress testing are evaluated in two steps. First, the charge of inconsistency with notice-and- comment requirements is considered and rebutted solely with reference to the doctrines relied on by banking interests. This step illustrates how administrative law requirements should be interpreted and applied with considerable attention to the substance of the regulatory mandate being carried out by the agency. The second step introduces two statutory provisions on capital regulation ignored in the complaints about stress Glazer & Ryan Tracy, Bank Groups Weigh Legal Challenge to Fed Stress Tests, WALL ST. J. (Sept. 1, 2016), https://www.wsj.com/articles/bank- groups-weigh-legal-challenge-to-fed-stress-tests-1472751110 (on file with the Columbia Business Law Review). Later that month, a paper making the administrative law case against elements of the stress testing program was released by the Committee on Capital Markets Regulation, a group consisting mostly (though not entirely) of individuals with strong ties to the banking industry, including current bank employees, trade association employees, industry consultants, and former bankers. CCMR Paper, supra note 8. A description of the Committee and a list of its current members are posted on its website, Members, CMT. ON CAP. MKTS. REGULATION, https://www.capmktsreg.org/members/ [https://perma.cc/FVL9-FQ2B] (last visited Mar. 7, 2022). No case was ever filed. The industry may have concluded, rightly as it turns out, that President Trump’s appointees to the Federal Reserve would take substantial steps to accommodate the banks’ requests. See Press Release, Bd. of Governors of the Fed. Rsrv. Sys., Federal Reserve Board Finalizes Set of Changes that Will Increase the Transparency of its Stress Testing Program for Nation’s Largest and Most Complex Banks (Feb. 5, 2019), https://www.federalreserve.gov/newsevents/pressreleases/bcreg20190205a. htm [https://perma.cc/9WRC-VL29]. 318 COLUMBIA BUSINESS LAW REVIEW [Vol. 2022 testing. There is a strong case to be made that these statutes entirely preclude the administrative law arguments against stress testing practices. At the very least, they reinforce the claims made in the first step of the analysis for applying administrative law doctrines consistently with the substantive statutory scheme created by Congress. 1. Background on Capital Regulation and Stress Testing To place the legal issues in substantive regulatory context, a bit of background on capital regulation and stress testing is needed. Capital requirements remain central in the post- Crisis prudential regulatory regime. Banks with less than $100 billion in assets are subject only to the simpler point-in- time capital requirements.106 Minimum capital requirements for banks with assets of between $100 billion and $250 billion are set every other year based on the results of a stress test.107 Banks with assets of over $250 billion have their capital requirements set based on the results of an annual stress test.108 In 2020 and 2021 stress testing was used as the basis 106 As originally passed, the Dodd-Frank Wall Street Reform and Consumer Protection Act (Dodd-Frank Act) required stress testing for all banks with $50 billion or more in assets. Pub. L. 111-203, § 165(a)(2)(B), 124 Stat. 1376, 1424 (2010). In 2018, the Economic Growth, Regulatory Relief, and Consumer Protection Act raised the asset threshold for stress testing to $100 billion and required that banks with between $100 and $250 billion in assets be stress tested only “on a periodic basis.” Economic Growth, Regulatory Relief, and Consumer Protection Act, Pub. L. No. 115-174, § 401(e), 132 Stat. 1296, 1359 (2018) (codified at 12 U.S.C. § 5365 note). The Federal Reserve currently treats “on a periodic basis” as every other year. Press Release, Bd. Governors Fed. Rsrv. Sys., Federal Reserve Board Releases Scenarios for 2019 Comprehensive Capital Analysis and Review (CCAR) and Dodd-Frank Act Stress Test Exercises (Feb. 5, 2019), https://www.federalreserve.gov/newsevents/pressreleases/bcreg20190205b. htm [https://perma.cc/M2SA-AZRS]. 107 12 U.S.C. S 5365 note (2018). 108 Id. No. 1:279] BANK SUPERVISION AND ADMINISTRATIVE LAW 319 for calculating capital requirements for, respectively, thirty- three and twenty-three large banks.109 As noted in Part II, capital requirements have traditionally been generated from regulatory estimates of average past losses associated with various asset classes. These average loss parameters do not factor in how losses across asset classes may be correlated in some states of the world, nor how losses across a range of institutions may weaken the financial system as a whole. Traditional capital requirements, which do not vary with changes in economic conditions, all suffer to a greater or lesser degree from a form of pro-cyclicality.110 Stress testing is a dynamic, forward-looking approach to setting minimum capital levels that can potentially overcome some of the traditional limitations of static capital requirements.111 Using an economic model, stress tests measure the impact on bank losses and revenues of a hypothesized severely adverse scenario.112 Stress testing thus projects the amount of capital that would be needed to absorb bank losses in an unlikely, but plausible, tail event.113 Stress testing can reveal the degree of loss correlations among asset classes. By testing all large bank balance sheets simultaneously, stress testing gives a more accurate picture 109 BD. OF GOVERNORS OF THE FED. RSRV. SYS, DODD-FRANK ACT STRESS TEST 2020: SUPERVISORY STRESS TEST RESULTS 1 (2020), https://www.federalreserve.gov/publications/files/2020-dfast-results- 20200625.pdf [https://perma.cc/XG67-6YA8]; BD. OF GOVERNORS OF THE FED. RSRV. SYS, DODD-FRANK ACT STRESS TEST 2021: SUPERVISORY STRESS TEST RESULTS 1 (2021), https://www.federalreserve.gov/publications/files/2021- dfast-results-20210624.pdf [https://perma.cc/7SJV-8CGV]. 110 To take one example, the higher earnings and lower default rates experienced during the peak of an economic cycle allow more lending (because earnings increase capital and rising asset prices increase collateral values) and can give a misleading picture of how the bank will fare once the economy turns down and default rates on all those additional loans increase. For a more complete explanation of the procyclical effects of many capital requirements, see Rafael Repullo & Javier Suarez, The Procyclical Effects of Bank Capital Regulation, REV. FIN. STUD. 452 (2013). 111 Policy Statement on the Scenario Design Framework for Stress Testing, 12 C.F.R. pt. 252 app. A, § 1(e) (2021). 112 Id. § 2(b). 113 Id. § 1(f). 320 COLUMBIA BUSINESS LAW REVIEW [Vol. 2022 of the risks to the financial system. Depending on the sophistication of the economic model, some of the unhelpful pro-cyclical features of point-in-time capital requirements can be mitigated.114 Following the Financial Crisis, the Federal Reserve adopted a notice-and-comment rule that requires large banks annually to undergo a stress test, which is used to help set minimum capital requirements for those banks.115 As explained in the preamble to that rule, the regulation was motivated in part by the substantial depletion of capital in some large banks on the eve of the Financial Crisis because of large dividend distributions.116 Thus, the rule was intended to ensure that systemically important banks maintain sufficient 114 There is an ongoing debate around the efficacy of, and optimal approach to, stress testing. For an introduction to the issues, see Richard J. Herring & Til Shuermann, Objectives and Challenges of Stress Testing, in HANDBOOK OF FINANCIAL STRESS TESTING (J. Doyne Famer et al., eds., 2022). This Article focuses only on the administrative law issues surrounding the Federal Reserve’s use of stress testing and does not consider these important policy issues. 115 Capital Plans, 76 Fed. Reg. 74,631, 74,632 (Dec. 1, 2011) (to be codified at 12 C.F.R. pt. 225). The rule has subsequently been amended, as reflected in the discussion in the text. Regulations Q, Y, and YY: Regulatory Capital, Capital Plan, and Stress Test Rules, 85 Fed. Reg. 15,576, 15,576 (Mar. 18, 2020) (codified at 12 C.F.R. § 225.8 (2021)). The changes, while consequential as a matter of capital regulation policy, do not significantly affect the administrative law issues considered here. 116 The preamble reads: During the years leading up to the recent financial crisis, many bank holding companies made significant distributions of capital, in the form of stock repurchases and dividends, without due consideration of the effects that a prolonged economic downturn could have on their capital adequacy and ability to continue to operate and remain credit intermediaries during times of economic and financial stress. The final rule is intended to address such practices, building upon the Federal Reserve’s existing supervisory expectation that large bank holding companies have robust systems and processes that incorporate forward-looking projections of revenue and losses to monitor and maintain their internal capital adequacy. Capital Plans, 76 Fed. Reg. at 74,632. No. 1:279] BANK SUPERVISION AND ADMINISTRATIVE LAW 321 capital on an ongoing basis so that they would continue to meet minimum capital requirements and remain viable intermediaries even were the hypothesized stress and ensuing losses to occur. The Federal Reserve regulation calls for the calculation of a “stress capital buffer” for each bank, which reflects the maximum net loss it could be expected to suffer during the hypothesized severely adverse scenario.117 In order to avoid restrictions on its capital distributions through the period covered by that stress test,118 each bank is required to maintain its risk-weighted capital levels at or above the sum of the stress capital buffer, minimum regulatory requirements, and any other applicable buffers.119 In its annual process, the Federal Reserve uses its own supervisory stress testing model to project bank losses and revenues under a hypothetical severely adverse scenario. The model inputs the details of each bank’s balance sheet and business lines to determine the amount of equity capital that a bank would need to remain adequately capitalized even after the hypothesized scenario has occurred. Shortly after completion of the stress test, the Federal Reserve provides each bank with a notice and explanation of its stress capital buffer.120 A bank may request reconsideration of the buffer,121 in accordance with 117 12 C.F.R. § 225.8 (2021). 118 Id. § 225.8(h)(2)(ii) (detailing restrictions). 119 The eight systemically important U.S. banks included in the LISCC portfolio are each assigned capital surcharges, which are treated as a buffer requirement on top of minimum capital requirements. 12 U.S.C. § 225.8 (2018). The Federal Reserve also has authority to impose a “countercyclical capital buffer” on large banks, though to date it has not done so. All buffer requirements are added together and then added to the applicable minimum common equity requirement of 4.5% on a risk-weighted basis. 12 C.F.R. § 225.8 § 217.11(c). So, for example, a systemically important bank with a surcharge of 2% and a stress capital buffer of 3.5% of risk-weighted assets would have to begin restricting capital distributions if its common equity capital level fell below 10% (4.5 + 2 + 3.5). 120 Id. § 225.8(h)(1). 121 Id. § 225.8(h)(2). 322 COLUMBIA BUSINESS LAW REVIEW [Vol. 2022 procedures laid out in Board regulations.122 Where reconsideration is requested, the tentative stress capital buffer is reviewed by a group of Federal Reserve experts independent of those who originally conducted the stress test.123 Except in unusual circumstances, each bank is to be notified by August 31 of its final stress capital buffer,124 which then becomes effective on October 1.125 While the Federal Reserve did provide some information on its supervisory economic model, its initial position was that releasing detailed information on the model code would enable banks to adjust the risk profile of assets so that they are disproportionately concentrated right below thresholds in the model that would trigger a higher loss function.126 It has always disclosed the results of its stress tests on a bank- specific basis, including loss rates for broad categories of assets and anticipated revenues during the stress period.127 Banks and outside analysts can thus approximate the supervisory model’s loss functions for particular asset classes. But without the model code (or something close to it), it is difficult to determine precisely which attributes of a specific asset led to its inclusion in one loss function rather than 122 Id. § 225.8(j). 123 See Press Release, Bd. Governors Fed. Rsrv. Sys., Federal Reserve Board Announces Large Bank Capital Requirements, Which Will be Effective October 1 (Aug. 10, 2020), https://www.federalreserve.gov/newsevents/pressreleases/bcreg20200810a. htm [https://perma.cc/Q2DQ-JXAT]. 124 12 C.F.R. § 225.8(h)(4)(i) (2021). 125 Id. § 225.8(h)(4)(ii). 126 Daniel K. Tarullo, Governor, Fed. Rsrv., Speech at the Yale University School of Management Leaders Forum: Next Steps in the Evolution of Stress Testing (Sept. 26, 2016), https://www.federalreserve.gov/newsevents/speech/tarullo20160926a.htm [https://perma.cc/NA44-FNLC] (“Full disclosure would permit firms to game the system—that is, to optimize portfolio characteristics based on the parameters of the model and take risks in areas not well-captured by the stress test just to minimize the estimated stress losses.”). 127 The most recent results are presented in BD. OF GOVERNORS OF THE FED. RSRV. SYS., DODD-FRANK ACT STRESS TEST 2020: SUPERVISORY STRESS TEST RESULTS, supra note 109, at 27–42. No. 1:279] BANK SUPERVISION AND ADMINISTRATIVE LAW 323 another.128 Following the appointment by President Trump of a new Chair and Vice Chair for Supervision, the Federal Reserve released substantially more information on the models.129 The Federal Reserve also adopted a rule explaining how it approaches scenario development.130 It releases the details of its stress scenarios, though only shortly before the test is run on the bank balance sheets as they stood at some point in the preceding quarter.131 This timing does not permit banks to adjust their balance sheets to respond to the scenario. While there is considerable continuity in the nature of a severe recession, the accompanying idiosyncratic features of the scenario and a hypothesized shock to traded asset prices may be changed from year to year to reflect both the range of potentially serious risks and the relative likelihood of a shock actually occurring.132 Though the regulation does not require 128 For banks themselves, each year of stress tests likely increases their information about the model, because they are aware of the specifics of their assets and can observe how loss rates change. Because the Federal Reserve provides notice of significant changes in modelling (without, at least to this point, providing the changed code itself), the banks can control for the scenario and determine whether differing loss rates for their assets are attributable to model change or to differences in the assets. Outside analysts do not have the benefit of this information unless, for their own reasons, bank choose to disclose it. 129 See Enhanced Disclosure of the Models Used in the Federal Reserve’s Supervisory Stress Test, 84 Fed. Reg. 6,784 (notice Feb. 28, 2019). 130 Policy Statement on the Scenario Design Framework for Stress Testing, 78 Fed. Reg. 59,533, 59,538 (proposed Dec. 15, 2017) (codified at 12 C.F.R. pt. 252). The Board of Governors subsequently revised the scenario in ways not relevant to the issues discussed in this Article. See Amendments to Policy Statement on the Scenario Design Framework for Stress Testing, 84 Fed. Reg. 6651 (Feb. 28, 2019) (codified at 12 C.F.R. pt. 252). 131 For the banking book (traditional lending) and for the entire balance sheet of most of the participating banks, the balance sheet on the last day of the preceding quarter is used. For the handful of firms with extensive capital market activities, the Federal Reserve varies the “as of” date of the trading book balance sheet varies from year to year, in a modest effort to reduce gaming by banks. Policy Statement on the Supervisory Design Framework for Stress Testing, 12 C.F.R. pt. 252 app. A § 3(a). 132 For an example of the scenario used for the most recent round of stress testing, see FED. RSRV., 2021 STRESS TEST SCENARIOS (2021) 324 COLUMBIA BUSINESS LAW REVIEW [Vol. 2022 it, the practice has been that three members of the Board of Governors, including the Chair meet annually with staff to decide upon the idiosyncratic features for that year’s test.133 Thus, the scenarios reflect an up-to-date judgment by the Board of Governors on the most significant economic and financial risks to be reflected in the stress tests. 2. Arguments for Notice-and-Comment on Supervisory Stress Model and Scenarios. The principal administrative law challenge to stress testing is that the models and scenarios are APA “rules” because they are future-oriented and de facto set capital requirements for the largest banks, and thus are subject to the notice and comment requirement.134 In essence, the argument is that each year’s stress test requires a new notice- and-comment procedure covering the proposed scenario and any changes made in the supervisory model. A different argument might be that the supervisory model code must be disclosed as part of providing adequate notice to affected parties of the impact of the stress test and capital planning rules. There are two rejoinders to the core argument. One is that each bank’s stress capital buffer is the result of an informal adjudication. Under the rule set forth by the Supreme Court in Bennett v. Spear, the stress test itself meets neither of the two conditions for final agency action—that it be the “consummation” of the agency’s decision-making process and that it result in legal obligations.135 If it is not a “final agency action,” then it cannot be subject to the notice-and-comment requirements of a legislative rule, which by definition creates a binding legal right or obligation. Each year, the Board https://www.federalreserve.gov/newsevents/pressreleases/files/bcreg20210 212a1.pdf [https://perma.cc/PC74-D7FR]. 133 Information about how the Board of Governors for the Federal Reserve decides on Stress Test factors is based on the author’s experience at the Federal Reserve. 134 CCMR Paper, supra note 8, at 11–12. 135 See Bennett v. Spear, 520 U.S. 154, 177–78 (1997). No. 1:279] BANK SUPERVISION AND ADMINISTRATIVE LAW 325 determines if each bank has built up enough capital at the time of the test to withstand possible future losses. This determination follows the detailed analysis conducted in the stress test, following both an initial notice to banks and an opportunity for each bank to contest that determination. Only then does the stress buffer meet the Bennett test of being the “consummation” of the decision-making that creates an obligation for the bank.136 Indeed, the current regulation governing the stress capital buffer makes clear that there is no final agency action until these later steps have been completed.137 The procedures laid out in the regulation for notice and explanation of the determination, as well as the opportunity for reconsideration, are more than adequate to meet the minimal requirements for informal adjudications under the APA.138 The proffered rebuttal to this argument is that, because the models and scenarios are developed in advance of the stress test itself, they are prospective in effect and must be part of the rule-making. “Adjudications” must be backward- regarding, carried out against the backdrop of pre-existing statutory standards or regulatory rules.139 But this rebuttal begs the question of why the Federal Reserve’s original stress testing capital rule did not itself provide the relevant standard to determine whether the bank has “demonstrated an ability to maintain capital above . . . minimum regulatory [standards] . . . under expected and stressful conditions through the planning horizon.”140 In each stress testing cycle, 136 See id. 137 12 C.F.R. § 225.8(h)(4)(i) (2021). 138 For APA requirements for informal adjudication, see BEN HARRINGTON & DANIEL J. SHEFFNER, CONG. RSCH. SERV., R46930, INFORMAL ADMINISTRATIVE ADJUDICATION: AN OVERVIEW (2021). 139 See CCMR Paper, supra note 8, at 12–13. 140 Capital Plans, 76 Fed. Reg. 74,631, 74,632 (Dec. 1, 2011) (to be codified at 12 C.F.R. pt. 225). The rule has subsequently been amended by Regulations Q, Y, and YY: Regulatory Capital, Capital Plan, and Stress Test Rules, 85 Fed. Reg. 15,576, 15,576 (Mar. 18, 2020) (codified at 12 C.F.R. § 225.8 (2021)). See supra note 115. As discussed in the text, similar standards are found in other Federal Reserve regulations and in Dodd- Frank itself. 326 COLUMBIA BUSINESS LAW REVIEW [Vol. 2022 the Federal Reserve applied this standard on a bank-by-bank basis, using the stress test, including the supervisory model and scenarios, as a tool in that evaluation. That is, its decision to object could have been understood as a judgment on whether the capital levels and assets as previously generated by the bank and its capital distribution intentions as previously stated were consistent with this standard. This response to a potential argument by banks has not fundamentally changed with the Federal Reserve’s adoption of the stress capital buffer approach. Since the operative decision by the Board is no longer an objection to a capital plan, but the establishment of the stress capital buffer determined by the annual stress test implementing the statutory requirement, the relevant standard may now be found in the legislative and regulatory language governing stress testing. Congress indicated that stress tests were to be used to determine “whether such companies have the capital, on a total consolidated basis, necessary to absorb losses as a result of adverse economic conditions.”141 This language is echoed as the stated purpose of the Federal Reserve’s regulation governing stress testing.142 Even if we assume, for the sake of argument, that the supervisory stress model and the stress scenarios annually adopted by the Board of Governors are rules that would otherwise be subject to notice-and-comment requirements, there is a strong case for applying the statutory good cause exception.143 While that exception has generally, and rightly, been narrowly applied over the years,144 the nature of capital regulation and stress testing support its application here. 141 12 U.S.C. § 5365(i)(1)(A) (2018). 142 12 C.F.R. § 252.41(b). 143 The APA specifies that its procedural requirements do not apply “when the agency for good cause finds (and incorporates the finding and a brief statement of reasons therefor in the rules issued) that notice and public procedure thereon are impracticable, unnecessary, or contrary to the public interest.” 5 U.S.C. § 553(b)(B) (2018). This is referred to as the “good cause exception.” See Little Sisters of the Poor Saints Peter and Paul Home v. Pennsylvania, 140 S.Ct. 2367, 2374 (2020). 144 See., e.g., Mid-Tex Elec. Co-op, Inc. v. F.E.R.C., 822 F.2d 1123, 1132 (D.C. Cir. 1987) (noting that the exception should be “narrowly construed No. 1:279] BANK SUPERVISION AND ADMINISTRATIVE LAW 327 There are not many precedents close to point, and “the ‘good cause’ inquiry is inevitably fact- or context- dependent.”145 Courts considering the good cause exception have generally focused on whether notice-and-comment would undermine the substantive aim of the contemplated agency action, as opposed to a simple delay or additional inconvenience for the agency. The D.C. Circuit characterized the public interest prong of the rule as intended to prevent situations in which “announcement of a proposed rule would enable the sort of financial manipulation the rule sought to prevent.”146 The Ninth Circuit observed that the APA “is not a device by which an agency may be forced to adopt a less effective regulatory program in order to more effectively comply with notice and comment procedures” and that “[t]he existence of the good cause exception is proof that Congress intended to let agencies depart from normal APA procedures where compliance would jeopardize their assigned missions.”147 While set in a different factual context, a group of cases involving price controls imposed under laws enacted in the and reluctantly countenanced”). In fact, many rules are issued without notice-and-comment, in reliance on either the APA good cause exception or the enabling statutory authority. However, many of these appear to involve interim rules followed by notice-and-comment, statutory deadlines for rulemaking, or instances where the statute essentially determined the content of a rule. See U.S. GOV’T ACCOUNTABILITY OFF., GAO-13-21, FEDERAL RULEMAKING: AGENCIES COULD TAKE ADDITIONAL STEPS TO RESPOND TO PUBLIC COMMENTS 17–18 (2012). 145 Mid-Tex Elec. Co-op, 822 F.2d at 1132. 146 Utility Solid Waste Activities Grp v. E.P.A., 236 F.3d 749, 755 (D.C. Cir. 2001). The court, referring back to the canonical ATTORNEY GENERAL’S MANUAL ON THE ADMINISTRATIVE PROCEDURE ACT (1947), was simply describing its understanding of the public interest prong of the good cause exception, which it did not find implicated in the case before it. The Attorney General’s Manual itself did not elaborate beyond saying that the public interest prong of the exception could be applicable where “the issuance of financial controls [would take place] under such circumstances that advance notice of such rules would tend to defeat their purpose.” ATTORNEY GENERAL’S MANUAL ON THE ADMINISTRATIVE PROCEDURE ACT 31 (1947). 147 Riverbend Farms, Inc. v. Madigan, 958 F.2d 1479, 1484 (9th Cir. 1992). 328 COLUMBIA BUSINESS LAW REVIEW [Vol. 2022 1970s supports the application of the exception for stress testing.148 These cases found that the public interest prong of the exception applied, because advance notice of specific price controls would incentivize affected producers to increase prices before the controls took effect, thereby undermining the price stability aims of the regulations. Here, making the model code and scenarios available in advance to banks would “precipitate activity by affected parties that would harm the public welfare”149 and, in the process, undermine the regulatory purpose of the stress test by allowing banks to game the system. They could take risks in areas not well- captured by the parameters of the supervisory stress model just to minimize the estimated stress losses. With effective advance notice of key scenario features, they could adjust their balance sheets based on their knowledge of which assets would fare relatively well under the scenario/model combination. Stress testing would then be less effective, quite possibly substantially so.150 An additional concern is that the 148 See, e.g., see Mobil Oil Corp. v. Dep’t of Energy, 728 F.2d 1477, 1491– 92 (Temp. Emer. Ct. App. 1983); Tasty Baking Co. v. Cost of Living Council, 529 F.2d 1005, 1015 (Temp. Emer. Ct. App. 1975); Nader v. Sawhill, 514 F.2d 1064, 1068 (Temp. Emer. Ct. App. 1975); Reeves v. Simon, 507 F.2d 455, 458–59 (Temp. Emer. Ct. App. 1974); DeRieux v. Five Smiths, Inc., 499 F.2d 1321, 1332 (Temp. Emer. Ct. App. 1974). Most far-reaching of the statutory authorities was the Economic Stabilization Act of 1970, Pub. L. No. 91-379, 84 Stat. 799, which President Nixon used to impose controls between late 1971 and early 1973. Edward Cowan, Mandatory Wage-Price Controls Ended Except In Food, Health, Building Fields; Nixon Calls For Voluntary Compliance, N.Y. Times (Jan. 12, 1973). The authority to control prices lapsed in 1974. Edward Cowan, Wage-Price Controls Quietly Expire, N.Y. TIMES (May 1, 1974), https://www.nytimes.com/1974/05/01/archives/wageprice-controls-quietly- expire-special-to-the-new-york-times.html [https://perma.cc/8FR2-KYZK]; see Economic Stabilization Act Amendments of 1973, Pub. L. 93-28, § 8, 87 Stat. 27, 29 (extending the Economic Stabilization Act until May 1, 1974). 149 Mobil Oil Corp., 728 F.2d at 1492. 150 The CCMR Report supra note 8, addressed the good cause exception, arguing that it is inapplicable to either scenario design or model code. As to the former, CCMR identifies the “impracticality” prong as the relevant one. But, as explained in the text, it is really the public interest prong that is at issue, since the value of stress testing is undermined if the banks have the scenarios in advance. As to the latter, CCMR argues that No. 1:279] BANK SUPERVISION AND ADMINISTRATIVE LAW 329 delay entailed in a notice-and-comment proceeding would risk making the stress scenario stale.151 gaming of the model is not an adequate reason for withholding the code because the Federal Reserve could “object to a bank’s capital plan if [it] suspected a bank of gaming the tests.” Id. at 19. The CCMR suggests that the Federal Reserve could develop what would necessarily be an elaborate monitoring system, including analysis of the data it receives throughout the year, in order to determine if gaming is present. This effort to require a massive (and likely less than effective) monitoring exercise by the Federal Reserve to uncover what will almost surely occur (and, presumably, be hidden by banks to the degree possible) almost proves the public interest need for maintaining the secrecy of the code. The CCMR suggests elsewhere in the report that it is unclear if the stress test would be gamed and that the burden should be on the Federal Reserve to prove that gaming has taken place. Id. at 21–22. Considering that capital “optimization” is a constant effort by banks, and a rich source of fees to consultants, there is more than a bit of disingenuousness in this recommendation. Finally, CCMR notes that the good cause exception has, in the limited circumstances in which it has applied, involved only temporary deviations from notice-and-comment requirements. Id. at 18. The relevance of this point is unclear, since, by their very nature, knowledge of price controls is obviously not a threat to the public interest once those controls have been published. The potential for disclosure of the model code to undermine the utility of stress testing is ongoing. 151 Some indication of what might be in store for the Federal Reserve were it to put the scenarios out for notice and comment was provided in a blog post by an employee of a bank trade association after the scenarios were published in 2018. Jeremy Newell, The Federal Reserve’s 2018 CCAR Scenarios: A Look at Process, BANK POLICY INST.: BLOG (Mar. 2, 2018) https://bpi.com/the-Federal Reserves-2018-ccar-scenarios-a-look-at-process [https://perma.cc/C9JE-CL5A]. The post suggested that the agency would need to provide a convincing explanation for why it chose “a 65% drop in the stock market, and not 50% (or 70%),” why it chose a 30% decline in housing prices, why the yield curve was assumed to steepen as it did in the scenario, why unemployment rose so suddenly, and “[w]hat about the rest of the 28 variables.” Id. As explained in the text, the demand that each variable in a stress scenario be “right” reflects a misunderstanding of stress testing. However, the more salient point is that this blog post suggested the intention of the trade association representing most of the largest banks in the country to litigate each element of the Federal Reserve’s scenario. Even if, and perhaps especially if, the Federal Reserve retained the basic elements of its scenario after notice and comment, the banks would have had a period of months to make strategic adjustments to their balance sheets. 330 COLUMBIA BUSINESS LAW REVIEW [Vol. 2022 Stress testing is not like using a model to develop a regulation that, for example, limits emissions of polluting substances. In such a case, industry adherence to the standard developed with the help of a model would itself achieve the regulatory purpose. But a stress test is exactly that—a test. Neither regulators nor bankers can count on correctly anticipating what the next source of severe stress will be. Recall the origins of the 2007-2009 Financial Crisis. Had regulators used only one stress scenario from year to year, and that scenario had not correctly anticipated the decline in housing prices and shock to mortgage-backed securities, the banks might still have been badly undercapitalized when the Crisis hit. The force of this point is further strengthened by the onset of the COVID-19 crisis in early 2020. No prior scenario developed by the Federal Reserve had contemplated the dramatic halt to economic activity associated with stay-at-home and other social distancing measures initiated to stop the spread of the virus. That is why variation in scenario elements, such as shocks to different asset classes and changes in yield curve shapes and interest rates, is critical to the value of stress testing.152 While analogies to stress testing have limitations, stress testing does in key respects resemble a test administered in a college course. If students are given the test in advance, most, if not all, of them will do very well. But since the exam can have covered only a small proportion of the material covered in the course, the professor will have no idea how much they know about the subject matter of the whole course. Similarly, if banks have the model and the scenario in advance, most, if not all, of them will produce balance sheets suggesting they will be comfortably capitalized should the scenario occur. But these results could paint a very misleading picture of their actual resiliency, because the single severely adverse scenario 152 In theory, the Federal Reserve might run dozens, perhaps hundreds, of scenarios to try to capture a much greater proportion of unanticipated risks. As an administrative and organizational matter, the idea seems impractical at present, though improved information technology systems may someday make it feasible. No. 1:279] BANK SUPERVISION AND ADMINISTRATIVE LAW 331 is—like a college exam—selective: It has tested only for one possible set of risks.153 A different line of argument could be that the model code should have been disclosed as part of the notice-and-comment rulemaking for the stress testing rule. This argument rests on cases holding that disclosure of various forms of technical data relied on by an agency in developing a rule is required as part of the notice mandated by APA § 553(b).154 There is a long- running debate as to the consistency of this line of cases, initiated by a 1973 D.C. Circuit opinion,155 with the Supreme Court’s subsequent admonition in Vermont Yankee156 against courts creating procedural requirements beyond those 153 A counter-analogy that has been used from time to time is that the stress test should be like posting a speed limit. If you want banks to go no higher than fifty-five mph, then you should put up a sign saying that clearly by releasing the model and possibly the scenarios in advance. See Ryan Tracey, Quarles: Fed To Propose More Transparent Stress Tests, WALL ST. J. (Dec. 1, 2017), https://www.wsj.com/articles/quarles-fed-to-propose-more- transparent-bank-stress-tests-1512176369 (on file with the Columba Business Law Review). I hope the text has shown why this analogy is inapposite. While the outcome one wants in the case of speed limits is for vehicles to travel below the posted speed, the outcome that regulators want is not that banks hold precisely enough capital to be able to buffer losses should the correlations in the model and the hypothesized scenario actually turn out to be the case. Instead, overall resiliency is the aim, with the stress test a useful, but necessarily incomplete way of calibrating capital requirements. 154 See, e.g., Cal. Wilderness Coal v. U.S. Dep’t of Energy, 631 F.3d 1072, 1089 (9th Cir. 2011) (modeling data “critical” to the ability of states to consult meaningfully on a congestion study that formed the basis for designation of an electric transmission corridor); Am. Radio Relay League v. FCC, 524 F.3d 227, 236–40 (D.C. Cir. 2008) (holding that agency must release technical studies and data on which it relied in promulgating rule to regulate the radio spectrum); Portland Cement Ass’n v. Ruckelshaus, 486 F. 2d 375, 393 (D.C. Cir. 1973) (requiring agency to release results of testing of emissions at existing plants underlying regulation to adopt emissions controls under the Clean Air Act because “[i]t is not consonant with the purpose of a rule-making procedure to promulgate rules on basis of . . . data that, [to a] critical degree, is known only to the agency.”). 155 See Portland Cement Ass’n, 486 F.2d at 393. 156 Vermont Yankee Nuclear Power Corp. v. Nat. Res. Def. Council, Inc., 435 U.S. 519 (1978). 332 COLUMBIA BUSINESS LAW REVIEW [Vol. 2022 explicitly stated in the APA.157 One need not take a definitive side in that debate, however, to conclude that the Federal Reserve should have the discretion to withhold the model code from disclosure. The reasonable position that courts should at times elaborate APA requirements in the context of specific administrative practices is strongest if it tailors those elaborated (or, de facto, additional) requirements to the substantive regulatory activity in which an agency is engaged. The requirements may thus vary from context to context. As mentioned earlier, the nature and use of the supervisory stress model are quite different from the models used by agencies such as the EPA and FCC in the line of cases requiring disclosure. In those circumstances, the model was itself generating, or contributing to the generation of, the desired regulatory outcome, for example, a limit on the emission of particulate matter. Knowledge of the model on the part of the regulated firms would not change that outcome. Stress testing, to reiterate, uses modeling to test a bank’s resiliency under different hypothesized conditions. Banks’ knowledge of the model will change their behavior and thereby compromise the regulatory goal of adequate capitalization of systemically important banks. 3. Relevant Statutory Provisions on Capital 157 In American Radio Relay League, then-Judge Kavanagh’s observed: Portland Cement stands on a shaky legal foundation (even though it may make sense as a policy matter in some cases). Put bluntly, the Portland Cement doctrine cannot be squared with the text of § 553 of the APA. And Portland Cement’s lack of roots in the statutory text creates a serious jurisprudential problem because the Supreme Court later rejected this kind of freeform interpretation of the APA [i]n its landmark Vermont Yankee decision. Am. Radio Relay League, 524 F.3d at 246 (Kavanagh, J. concurring in part and dissenting in part). In Little Sisters of the Poor Saints Peter & Paul Home v. Pennsylvania, 140 S.Ct. 2367, 2385–86 (2020), the Court reiterated its Vermont Yankee doctrine. No. 1:279] BANK SUPERVISION AND ADMINISTRATIVE LAW 333 Regulation. Up to this point, my rebuttal to arguments that administrative law requires the supervisory model and stress scenarios to be notice-and-comment rules has rested in part on the nature of capital regulation and stress testing. That is, I have tried to distinguish precedents developed in other substantive areas based on the differences between capital regulation and those other areas. My argument for the application of the good cause exception similarly relies on the peculiarities of stress testing. Needless to say, I believe this attention to differences among substantive regulatory areas and regulatory agencies is the right approach to judicial review of administrative action. Still, there will always be some room for disagreement on how general administrative law requirements should be applied to any substantive regulatory program, including stress testing. Once relevant provisions of statutory banking law are considered, however, any question on the breadth of the Federal Reserve’s discretion in this area should fall away.158 There are two key statutory provisions. The first is the requirement in section 165 of the Dodd-Frank Act of enhanced prudential standards for large bank holding companies.159 158 It is telling that the CCMR Report barely mentions banking law. See CCMR Paper, supra note 8. 159 Dodd-Frank Act, 12 U.S.C. § 5365 (2018). With the amendments enacted in 2018, that provision currently requires a range of enhanced prudential standards for banks with assets equal to or greater than $250 billion, allows the Federal Reserve to impose such standards for banks with assets equal to or greater than $100 billion, and requires supervisory stress tests for all banks with assets equal to or greater than $100 billion, though only on a “periodic basis” for banks with assets between $100 billion and $250 billion. Economic Growth, Regulatory Relief, and Consumer Protection Act, Pub. L. 115-174, § 401(e), 132 Stat. 1296, 1359 (2018) (codified at 12 U.S.C. § 5365 note). In addition to the enhanced prudential standards required by subsections (a) and (b), section 165 contains various other prudential requirements for large banks. Some, like stress testing, 12 U.S.C. § 5365(i) (2018), and counterparty credit exposures, id. § 5365(e)(3), are mentioned in the text. Others, such as the requirement for resolution planning, are not directly relevant to the questions addressed here. Id. § 5365(d). 334 COLUMBIA BUSINESS LAW REVIEW [Vol. 2022 These standards, which are to be “more stringent”160 than those applicable to smaller banks, must “increase in stringency” based on the risk posed by banks to financial stability, safety and soundness.161 The Board of Governors is instructed, in developing these standards, to “differentiate among companies on an individual basis or by category, taking into consideration their capital structure, riskiness, complexity, financial activities . . . , size, and any other risk- related factors that the Board . . . deems appropriate.”162 The Board is required to establish these more stringent requirements in certain areas, including risk-based capital requirements and leverage limits,163 and is given discretion to establish such requirements in a number of other areas.164 A later subsection of this same provision creates a requirement for annual stress testing of large banks.165 Though that section does not specify that the results of the stress tests must be used in setting the more stringent minimum capital levels, the Federal Reserve has combined the two mandates in its approach to capital regulation. Aside from a requirement to consult other financial regulators,166 the statute does not specify any procedures for the Federal Reserve to follow in implementing enhanced prudential standards. But at least two features of the statute should be relevant to courts in deciding what the APA requires. First, the capital measures required by the statute are called “standards,” whereas other subsections of the same statute either require or give discretion to the Federal Reserve to enact “regulations.”167 Indeed, one such subsection requires that the Board “by regulation, shall prescribe standards” to 160 12 U.S.C. § 5365(a)(1)(A). 161 Id. § 5365(a)(1)(B). 162 Id. § 5365(a)(2)(A) (emphasis added). 163 Id. § 5365(b)(1)(A)(i). 164 Id. § 5365(b)(1)(B). 165 Id. § 5365(i). 166 12 U.S.C. § 5365(b)(4) (2018). 167 For example, the Board “may issue regulations that require []bank[s] . . . to maintain a minimum amount of contingent capital that is convertible to equity in times of financial stress.” Id. § 5365(e). No. 1:279] BANK SUPERVISION AND ADMINISTRATIVE LAW 335 limit the exposure of banks to any single counterparty.168 The fact that the requirement for capital “standards” makes no similar reference to the Board acting “by regulation” at least opens the possibility that Congress contemplated more procedural flexibility in implementing those standards. This inference is strengthened by the second relevant provision, which authorizes the Board to vary standards on an individual bank basis if it so desires. That second relevant statute is section 908 of the International Lending Supervision Act of 1983 (ILSA).169 ILSA was enacted as part of a package of congressional responses to the Latin American debt crisis of the early 1980s, which had revealed inadequate levels of capital at some large, internationally active banks.170 Section 908 includes both a strong charge and broad authority for the banking agencies in setting capital requirements: Each appropriate Federal banking agency shall cause banking institutions to achieve and maintain adequate capital by establishing minimum levels of capital for such banking institutions and by using such other methods as the appropriate Federal banking agency deems appropriate.171 The statute makes clear the discretion the banking agencies have in carrying out this mandate: Each appropriate Federal banking agency shall have the authority to establish such minimum level of capital for a banking institution as the appropriate Federal banking agency, in its discretion, deems to be 168 Id. § 5365(e)(1). 169 International Lending Supervision Act § 908, 12 U.S.C. § 3907. 170 See Ellen W. Smith, New Control on Global Debt: The International Lending Supervision Act of 1983, 17 CORNELL INT’L L.J. 425 (1984) (reviewing the background and context of ILSA). 171 12 U.S.C. § 3907(a)(1) (emphasis added). The House report on the bill stated that this provision was enacted to provide “a stronger, unambiguous statutory directive to the regulators to strengthen banks’ capital positions.” H.R. REP. NO. 98-175, at 45 (1983). 336 COLUMBIA BUSINESS LAW REVIEW [Vol. 2022 necessary or appropriate in light of the particular circumstances of the banking institution.172 This language was at least in part a response to a Fifth Circuit decision earlier in 1983 that had set aside an OCC cease-and-desist order for a bank to increase its capital as not supported by substantial evidence.173 In a later decision, a chastened Fifth Circuit confirmed that the plain language of section 908, read in conjunction with the APA’s exclusion from judicial review of an agency action “committed to agency discretion by law,”174 precluded judicial review of an FDIC directive to a specific bank to raise its capital ratios.175 The banking agencies have broad discretion in the means they choose to implement the decisions they make on bank capital requirements. In addition to the “use other methods” language quoted above, section 908 provides that an agency may enforce its decision through a cease-and-desist order176 172 12 U.S.C.§ 3907(a)(2) (emphasis added). 173 First Nat’l Bank of Bellaire v. Comptroller of the Currency, 697 F.2d 674 (5th Cir. 1983). Again, the legislative history of ILSA elaborates clearly the motivation for the provision, stating that Bellaire “clouded the authority of the bank regulatory agencies to exercise their independent discretion in establishing and requiring the maintenance of appropriate levels of capital.” S. REP. NO. 98-122, at 16 (1983). The Committee further explained: The Committee believes that establishing adequate levels of capital is properly left to the expertise and discretion of the agencies. Therefore, in order to clarify the authority of the banking agencies to establish adequate levels of capital requirements, to require the maintenance of those levels, and to prevent the courts from disturbing such capital, the Committee has provided a specific grant of authority to the banking agencies to establish levels of capital[.] Id. 174 5 U.S.C. § 701(a)(2) (2018). 175 Fed. Deposit Ins. Corp. v. Bank of Coushatta, 930 F.2d 1122, 1129 (5th Cir. 1991). 176 12 U.S.C. § 3907(b)(1) (2018). This subsection actually states that failure of a bank to maintain the capital level required by a banking agency may be deemed by the agency an “unsafe or unsound practice” within the meaning of 12 U.S.C. § 1818. 12 U.S.C. § 3907(b)(1). Under section 1818, banking agencies issue cease-and-desist orders when they have determined No. 1:279] BANK SUPERVISION AND ADMINISTRATIVE LAW 337 or a “directive” to a bank.177 A banking agency may require a bank “to submit and adhere to a plan acceptable to the . . . agency describing the means and timing by which the banking institution shall achieve its required capital level.”178 Finally, the statute authorizes the agency to “consider such banking institution’s progress in adhering” to a required capital directive whenever the bank seeks approval “for any proposal which would divert earnings, diminish capital, or otherwise impede such banking institution’s progress in achieving its minimum capital level.”179 The agency “may deny such approval where it determines that such proposal would adversely affect the ability of the banking institution to comply with such plan.”180 Although these two statutes were enacted in response to two different banking crises separated by a quarter century, they actually complement one another quite well and are directly linked. In addition to the requirement for more stringent prudential requirements included in section 165, Dodd-Frank added a sentence to section 908 calling on the banking agencies to make the capital standards required under this section or other provisions of Federal law for insured depository institutions countercyclical so that the amount of capital required to be maintained by an insured depository institution increases in times of economic expansion and decreases in times of economic contraction, consistent with the safety and soundness of the insured depository institution.181 that such a practice exists. Id. § 1818(d). These orders are judicially enforceable. Id. § 1818(i). 177 12 U.S.C. § 3907(b)(2)(A). 178 Id. § 3907(b)(2)(B)(i). 179 Id. § 3907(b)(3)(A). 180 Id. § 3907(b)(3)(B). 181 Dodd-Frank Act, Pub. L. No. 111–203, § 616(c), 124 Stat. 1376, 1615 (2010) (codified at 12 U.S.C. § 3907(a)(1)) (adding this provision in an amendment to the International Lending Supervision Act of 1983, Pub. L. 98-181, § 908(a)(1), 97 Stat. 1153, 1280 (codified as amended at 12 U.S.C. § 3907(a)(1)). 338 COLUMBIA BUSINESS LAW REVIEW [Vol. 2022 This amendment invites, or perhaps directs, the banking agencies to use their existing section 908 authority to address capital regulatory shortcomings newly revealed by the Financial Crisis. As noted earlier, one of the many potential advantages of using stress testing to help set capital requirements for large banks is that it can ameliorate the pro- cyclical bias of standard point-in-time capital requirements. Section 908 maps comfortably onto the approaches successively adopted by the Federal Reserve to use stress tests in setting capital requirements. Minimum capital requirements for the large banks included in the program are established by the Board of Governors after an evaluation of the impact of a severe stress scenario on each bank’s losses and revenue, based on its specific assets and business models. Following section 908(a), the Board sets the capital requirement for each bank that, “in its discretion, [it] deems to be necessary or appropriate in light of the particular circumstances of the banking institution.”182 Should a bank nonetheless, for example, declare and distribute dividends that would bring it below the minimum level determined by the Federal Reserve, the Board could issue either a capital directive183 or a cease-and-desist order. 182 12 U.S.C. § 3907(a)(1) (2018). 183 There could be a question as to whether banks could argue that the Federal Reserve is precluded from using capital directives to enforce CCAR results under the principle enunciated in U.S. ex rel. Accardi v. Shaughnessy that agencies must comply with their own regulations. 347 U.S. 260, 267 (1954). The Federal Reserve’s current procedural regulation for issuing capital directives, 12 C.F.R. §§ 263.80–263.85 (2021), includes some steps not currently followed in the original CCAR program and the current approach of stress capital buffers. See Edward Stein, Stress Test- Based Capital Directives (Apr. 26, 2018) (unpublished manuscript) (on file with the Columbia Business Law Review). Indeed, it may be prudent for the Federal Reserve to amend this regulation to more explicitly integrate it with the elaborate capital planning requirements it has subsequently put in place. However, even without such an amendment, there would be nothing to stop the Federal Reserve from following its current regulation should a bank deviate from a capital plan to which the agency has not objected. That is, it can reasonably argue that the bank-specific capital requirements are determined using the stress test process under the broad mandate of section 908(a) and that enforcement in the event a bank ignores those requirements No. 1:279] BANK SUPERVISION AND ADMINISTRATIVE LAW 339 The judicial review normally attending cease-and-desist orders has not been available for capital requirements following enactment of ILSA. The Tenth Circuit squarely rejected a bank challenge to a cease-and-desist order for increased capital on the ground that the “in its discretion” language applied only to capital directives.184 The court reasoned that section 908 “forecloses” review of agency “imposition of capital requirements because it commits the setting of capital levels to [agency] discretion without giving us any standard to determine the correctness of the [agency] decision.”185 The stress test and associated capital requirements can thus be understood as an exercise of the essentially unreviewable discretion granted the Federal Reserve to set capital requirements for each banking organization it supervises, based on its assessment of the risks faced by that bank. Based on section 908 alone, against the backdrop of Dodd-Frank section 165, the Federal Reserve has authority simply to announce to each of the covered banks what its capital requirement will be. Seen through this lens, the elaborate apparatus for developing, validating, and applying the stress test to the large banks that meet the criterion for systemic importance established by Congress is a considerably more structured and self-disciplining mechanism than the law requires. These practices make for better policy. However, they, much less the enervating steps of disclosing the model code or delaying scenario design for notice-and-comment, are not legally mandated. is a separate matter under section 908(b). See International Lending Supervision Act of 1983 § 908, 12 U.S.C. § 3907. The procedures contemplated in the regulation could be somewhat cumbersome for the Federal Reserve, but they could still be followed. 184 Frontier State Bank Oklahoma City v. Fed. Deposit Ins. Corp., 702 F.3d 588, 594 (10th Cir. 2012). 185 Id. at 597. The court further noted that “[w]hile this choice leaves banks in the position of enduring any vicissitude attending the exercise of the regulator’s discretion, Congress is permitted to prioritize the safety of the banking system over banks’ interest in avoiding subjective or even harsh agency decisions.” Id. 340 COLUMBIA BUSINESS LAW REVIEW [Vol. 2022 B. Supervisory Ratings As described in Part II, the assignment of ratings to banks every twelve or eighteen months is central to the supervisory function, carrying both supervisory and regulatory consequences. This is especially true for all but the handful of institutions to which the agencies have assigned dedicated supervisory teams. For the supervisors of those thousands of banks, the assignment of CAMELS ratings follows directly from the examination process. For the banking organizations with dedicated supervisory teams, the assignment of ratings is the occasion for reviewing all the concerns and observations of the preceding year. Important as the ratings process may be, there are few administrative law arguments available to banks aggrieved by the ratings they receive. Statutory procedural requirements are minimal, and judicial review of the substance of supervisory ratings is unlikely to provide meaningful relief. In creating the FFIEC, Congress required that the “uniform principles and standards” for examination of depository institutions “shall be applied” by the banking agencies.186 It has also implicitly required the Federal Reserve to assign ratings to bank holding companies by conditioning “financial” holding company status on the holding companies, as well as their insured depository subsidiaries, being well-capitalized and well-managed.187 Within the categories of administrative action contemplated in the APA, the assignment of ratings must be a form of informal adjudication, since it is clearly not rule-making and equally clearly not required to be “on the record after [an] opportunity for agency hearing,”188 and thus not formal 186 12 U.S.C. § 3305(a). 187 12 U.S.C. § 1843(l)(1)(C). This requirement was added by section 606(a) of the Dodd-Frank Act, Pub. L. 111-203, 124 Stat. 1376, 1607, which amended the Bank Holding Company Act of 1956, Pub. L. 84-511, 84 Stat. 511. Under the 1999 Gramm-Leach-Bliley Act, the requirements for well- managed and well-capitalized applied only to the insured depository institutions of a holding company wishing to become a financial holding company. Pub. L. No. 106-102, § 103(a), 113 Stat. 1338, 1346. 188 5 U.S.C. § 554(a) (2018). No. 1:279] BANK SUPERVISION AND ADMINISTRATIVE LAW 341 adjudication. As such, it is procedurally confined only by the minimal standards provided for all agency actions under the APA and any additional statutory requirements. The only such requirement of any significance is that there be “an independent intra-agency appellate process . . . to review material supervisory determinations made at insured depository institutions,”189 including examination ratings.190 The only meaningful requirement for this process other than its fairly loose prerequisite of “independence”191 is each banking agency’s obligation to ensure that an appeal can be “heard and decided expeditiously” and that safeguards exist to protect appealing banks from retaliation by examiners.192 Not surprisingly considering the consequences associated with ratings, complaints from banks about the ratings system have been commonplace. In the years following the Financial Crisis, the most frequent concerns pertained to perceived inconsistencies in ratings across banks in similar circumstances, overly long and demanding processes for restoring a higher rating after it has been downgraded, and the potential for downgrades because of deficiencies in areas not previously identified by supervisors as a specific subject of focus or because supervisory expectations have increased since previous examinations.193 The first is something of a perennial; the last two are probably at least in part a product of the post-Crisis adjustment of supervisors to what was regarded as insufficiently rigorous pre-Crisis supervision. Other reported concerns include the subjectivity of the Management rating,194 that enforcement actions may more or 189 12 U.S.C. § 4806(a). 190 Id. § 4806(f)(1)(A)(i). 191 The statute defines “independent appellate process” as one in which the review is done by “an agency official who does not directly or indirectly report to the agency official who made the material supervisory determination under review.” Id. § 4806(f)(2). 192 Id. § 4806(b). 193 Information about concerns is based on reports from supervisors to the author and direct conversations with bankers during that period. 194 See, e.g., Baer Testimony, supra note 14, at 11; Jeff Bater, Banks Welcome FDIC Chief’s Call for Review of Ratings System, BLOOMBERG L. (July 11, 2018, 2:33PM), https://news.bloomberglaw.com/banking- 342 COLUMBIA BUSINESS LAW REVIEW [Vol. 2022 less automatically result in a downgrading of Management,195 and that the statutory appeals process is largely ineffective.196 There is merit in some of these complaints, though not as much as some banks and their representatives would assert. However, current administrative law likely holds little potential for change that would be of much help to banks, especially smaller banks. Hurdles lie both in the availability of judicial review and its practical effect even if available.197 law/banks-welcome-fdic-chiefs-call-for-review-of-ratings-system [https://perma.cc/V6SE-E5LE]. As reporting reveals, bank complaints about the “subjectivity” of the Management rating actually vary a fair amount, sometimes in ways at odds with each other. See Baer Testimony, supra note 14 For example, while one former executive at a large bank quoted in the article suggested that supervisors should rely on “objective” data, an executive of a trade association representing community banks was quoted as saying that “[e]xaminers should be looking at how the bank’s going to perform in the future as opposed to how it’s been doing.” Id. The latter approach, while it might incorporate objective information, surely requires considerable judgment by examiners. It is worth noting that in its 1996 revision of the ratings system, the FFIEC responded to comments about the importance attached to risk management: The revised rating system reflects an increased emphasis on risk management processes. The Federal supervisory agencies currently consider the quality of risk management practices when applying the UFIRS, particularly in the management component. Changes in the financial services industry, however, have broadened the range of financial products offered by institutions and accelerated the pace of transactions. These trends reinforce the importance of institutions having sound risk management systems. Uniform Financial Institutions Rating System, 61 Fed. Reg. 67,021, 67,023 (notice Dec. 19, 1996). 195 Tahyar Testimony, supra note 9, at 43. 196 See Julie Anderson Hill, When Bank Examiners Get It Wrong: Financial Institution Appeals of Material Supervisory Determinations, 92 WASH. U. L. REV. 1101, 1160 (2015) (arguing the appeals process lacks consistency among regulators, results in few appeals, and lacks transparency). 197 Two bank trade associations have suggested, though not quite said, that it is illegal for a banking agency to downgrade the rating of a bank based on anything other than a violation of law—meaning either contravention of a statute or regulatory rule, or a practice serious enough to No. 1:279] BANK SUPERVISION AND ADMINISTRATIVE LAW 343 A recent Seventh Circuit case has raised hopes among some banking interests that courts may be willing to review at least the non-capital elements of supervisory ratings. In Builders Bank v. FDIC,198 the court reversed the district court’s dismissal of the case for lack of jurisdiction and rejected the FDIC’s alternative argument for dismissal that, under the doctrine enunciated in Heckler v. Chaney, there was “no meaningful standard against which to judge the agency’s exercise of discretion.”199 Writing for the court, Judge Easterbrook took note of Frontier State Bank, Oklahoma City v. FDIC,200 the Tenth Circuit decision holding that agency decisions on bank capital requirements were committed to agency discretion.201 After reaching that conclusion, the court constitute an “unsafe or unsound” practice within the meaning of 12 U.S.C. § 1818(b). Letter from Gregory A. Baer, President, Bank Pol’y Inst. & Rob Nichols, President, Am. Bankers Assoc. to Mick Mulvaney, Acting Dir., Bureau of Consumer Fin. Prot. (Nov. 13, 2018), https://www.aba.com/- /media/documents/letters-to-congress-and-regulators/bpi-aba-joint-pfr-on- supervisory-guidance-fdic.pdf?rev=10ba04e070ee4a36914d0e282cf9a669 [https://perma.cc/Y86H-77L3]; Letter from Gregory A. Baer, President, Bank Pol’y Inst. & Rob Nichols, President, Am. Bankers Assoc. to Robert A. Feldman, Exec. Sec’y, Fed. Deposit Ins. Corp. (Nov. 5, 2018), https://www.aba.com/-/media/documents/letters-to-congress-and- regulators/bpi-aba-joint-pfr-on-supervisory-guidance- fdic.pdf?rev=10ba04e070ee4a36914d0e282cf9a669 [https://perma.cc/Y86H- 77L3]. This position may actually be a lobbying request for something within the discretion of the banking agencies. If it is indeed intended as a statement of what current law requires, however, it is extremely difficult to square with the creation and regular reaffirmation by Congress of the uniform ratings system. Nowhere do 12 U.S.C. § 3305 and other statutory provisions referencing supervisory ratings state or imply that ratings are dependent on the standards for cease-and-desist orders in § 1818(b). On the contrary, if that were the case, ratings would be largely superfluous. That is, if Congress intended the position favored by the bank trade associations, it could simply have made the limitations on financial holding companies provided for in § 1843(m) flow from an adjudicated violation of § 1818, rather than as a result of a bank receiving a rating below a 2. 198 846 F.3d 272, 274 (7th Cir. 2017). 199 Heckler v. Chaney, 471 U.S. 821, 830 (1985). 200 702 F.3d 588, 594 (10th Cir. 2012). 201 Builders Bank, 846 F.3d at 276 (citing Frontier State Bank, 702 F.3d at 597–604). 344 COLUMBIA BUSINESS LAW REVIEW [Vol. 2022 had considered on the merits whether certain portions of a cease-and-desist order unrelated to capital requirements should be set aside as arbitrary and capricious.202 Judge Easterbrook reasoned that the same distinction between capital and non-capital issues could open the door for review of ratings elements unrelated to capital adequacy.203 Builders Bank was hailed as a potentially “significant victory for the banking industry.”204 The holding, as carefully phrased by Judge Easterbrook, is certainly defensible. But it seems unlikely that it will meaningfully increase banks’ ability to challenge supervisory ratings successfully, for four reasons. First, the Seventh Circuit concluded little beyond the absence of a clear bar to judicial review of non-capital CAMELS elements. The court left open the possibilities that one or more other elements could be committed to agency discretion and that the bank’s nominal challenge to other CAMELS ratings was in fact an indirect challenge to the FDIC’s conclusion on capital adequacy.205 Second, Judge Easterbrook questioned whether the assignment of a rating was a “final administrative action” within the meaning of 5 U.S.C. § 704.206 He suggested, without holding, that only a rating serving as the basis for a clear legal consequence would be reviewable.207 Thus, banks 202 Frontier State Bank, 702 F.3d at 597–604. 203 Builders Bank, 846 F.3d at 276. Judge Easterbrook also suggested that obvious mistakes in a banking agency’s decision on capital adequacy, such as an arithmetic error, might be subject to judicial correction. Id. 204 Lalita Clozel, Bank Wins Right To Sue Regulator Over Camels Rating, AM. BANKER (Mar. 20, 2017), https://www.americanbanker.com/news/bank-wins-right-to-sue-regulator- over-camels-rating [https://perma.cc/PMG9-BDPV] (quoting banking lawyer). 205 See Builders Bank, 846 F.3d at 276. None of these open issues was resolved upon remand, because the case was disposed of on other grounds by the district court, whose decision was then affirmed by the Seventh Circuit, also in an opinion by Judge Easterbrook. See Builders NAB LLC v. Fed. Deposit Ins. Corp., 922 F.3d 775 (7th Cir. 2019). 206 Builders Bank, 846 F.2d at 275. 207 Id. No. 1:279] BANK SUPERVISION AND ADMINISTRATIVE LAW 345 might need to await an enforcement action for unsafe or unsound practices, the conclusion of a 4(m) agreement, or the recalculation of its deposit insurance premium to have a court examine the merits of the rating. Third, while the Frontier Bank court assessed the portions of the FDIC’s order dealing with liquidity, management, and asset growth issues, it upheld the FDIC on each.208 Given how bank-specific and technical such issues can be, the court appropriately deferred to the agency’s conclusions once it had assured itself that there was sufficient evidence to establish the reasonableness of the findings. In the ratings context, where the bank’s entire operations are under examination, the amount of evidence available to support supervisory conclusions on any one of the elements will almost certainly be greater than that available to support a finding of a specific “unsafe or unsound” practice forming the basis for a cease- and-desist order. If judicial reviews of supervisory ratings loom, the general counsels of the banking agencies will be sure to advise their supervisors to include in examination reports anything that might be helpful on appeal. Fourth and finally, even if a bank can overcome these hurdles and convince a court to find part of a CAMELS rating to have been arbitrary and capricious, the usefulness of that outcome to the bank will probably be limited.209 For smaller banks the expense of a federal district court case will itself be a strong disincentive. Moreover, the rating and its attendant supervisory and regulatory consequences may well remain in place during the pendency of the litigation. Also, on remand there will be an opportunity for a banking agency to supplement its conclusions with new evidence, perhaps drawn from bank activity subsequent to the filing of the initial case. 208 Frontier State Bank, 702 F.3d at 597–604. 209 Since the APA does not allow for money damages, it does not appear that any additional deposit insurance premiums paid during the period in which a lower rating was in place could be recouped. It is unclear whether there is any other basis for attempts to recover the difference between insurance premiums resulting from a lower rating. See Builders NAB LLC, 922 F.3d at 777. 346 COLUMBIA BUSINESS LAW REVIEW [Vol. 2022 In short, even a somewhat generous application of standard administrative law doctrines would be of only limited help to banks in dealing with supervisory ratings they considered unwarranted. The thousands of small banks will find little beyond a symbolic victory in Builders Bank. To be sure, a large bank that did not find litigation costs prohibitive might occasionally find it cost effective to sue. Yet, even the large banks would likely gain little from a review of the “management” component rating about which their surrogates have complained. Management care and competence are not easily reduced to the kinds of objective indicators that can form an important starting point for assessing capital, liquidity, earnings, and perhaps asset quality. Indeed, were the agencies to attempt to objectify the criteria for the rating, the result would almost surely be a set of over-specified requirements that would justifiably strike many bank executives as inappropriate for rating the management of any individual bank.210 The modest legal requirements for the ratings process stand in contrast to its importance in the supervisory relationship. As both the culmination of a supervisory evaluation and a starting point for future supervisory oversight, it is central to this relationship, especially for the thousands of smaller banks for which there is no continuous monitoring (and thus engagement) by a dedicated on-site team. Even with the potential for some judicial review of supervisory ratings, it seems unlikely that current administrative doctrine law will deliver the relief that banks—especially smaller banks—will find useful as a practical matter. My overall argument in this Article is that the nature of bank supervision, including but not limited to, its statutory foundations, argues for shaping administrative law doctrine 210 Furthermore, in the be-careful-what-you-wish-for department, if courts were unexpectedly to begin second-guessing supervisory judgments as to banks’ risk management procedures and performance, the agencies would be incentivized to use their unreviewable discretion to issue directives for higher capital when a bank’s risk management deficiencies worried them. No. 1:279] BANK SUPERVISION AND ADMINISTRATIVE LAW 347 so as clearly to allow certain supervisory communications and judgments that undeniably can affect bank conduct. But the same consideration of taking due account of the supervisory relationship argues for asking more of the banking agencies in assigning supervisory ratings than current administrative law doctrine requires of them.211 That is, a principled departure from trans-substantivity in administrative law should not be unidirectional, but should instead take us where the nature and Congressional mandate of bank supervision lead. As Part IV will argue, adapting APA requirements to the realities of the bank supervisory relationship would entail more effective review of ratings decisions than arguably exists at present. As a group, banks would surely do better with a more responsive internal agency review process that placed the weight of the agency leadership behind ratings reviews than with highly legalized, judicially-centered review. The existence of such a process could in turn provide courts with a more effective means of providing appropriate checks on agency discretion in the ratings process. C. Supervisory Guidance and Communication This Section examines the consistency of supervisory guidance with generally applicable administrative law doctrines. In many respects, the use of guidance and various forms of communication associated with it best exemplifies the bank supervisory function. While the stronger argument is that current practices are consistent with administrative law, there is more uncertainty on this topic than on the two supervisory actions considered in prior sections. The uncertainty is in part a consequence of the famously muddled state of doctrine on what types of administrative guidance are 211 The banking agencies have taken steps to enhance the attractiveness of their internal appeals processes for ratings and other significant supervisory actions. Most recently, the FDIC has proposed creation of an independent Office of Supervisory Appeals, whose members would be drawn from outside the agency. Guidelines for Appeals of Material Supervisory Determinations, 85 Fed. Reg. 54,377, 54,377 (notice Sept. 1, 2020). This change may be of some benefit to banks, though concerns about damaging supervisory relationships may well remain. 348 COLUMBIA BUSINESS LAW REVIEW [Vol. 2022 contained within the APA exception to notice-and-comment rulemaking for “general statements of policy.”212 In addition, because some forms of supervisory communication, such as MRAs, have few close parallels outside banking regulation, there is limited direct precedent on which to draw in projecting how they would be treated by courts under current law. Finally, as I will explain following the doctrinal analysis, this characteristic form of supervision reveals most clearly the sometimes-awkward fit between elements of administrative law and the longstanding statutory mandate for a supervisory function. The issues are best illustrated with a concrete example. I have hypothesized a set of supervisory actions in order to focus on the interaction of banking agency guidance with notices of MRAs and bank ratings decisions, two forms of administrative actions peculiar to banking supervision. Before proceeding, I should note that in the last several years, the banking agencies have moved closer in practice to the more legalized approach to guidance advocated by various banking interests. That direction was reflected to some degree in a 2018 interagency statement.213 It was reflected much more explicitly in various speeches of the Federal Reserve Governor who was the Vice Chair for Supervision from October 2017 through October 2021.214 At the request of those 212 5 U.S.C. § 553(b) (2018). Judicial comments on the confused state of the doctrine go back decades. See, e.g., Noel v. Chapman, 508 F.2d 1023, 1029–30 (2d Cir. 1975) (commenting that the distinction between a rule for which notice-and-comment is required and a general statement of policy is “enshrouded in considerable smog”). Quite recent judicial comments along the same lines confirm that the confusion has not been dispelled in the intervening period. See, e.g., Nat’l Mining Ass’n v. McCarthy, 758 F.3d 243, 251 (D.C. Cir. 2014) (inquiry into how to classify an agency action “turns out to be quite difficult and confused”). 213 Bd. of Governors of the Fed. Rsrv. Sys., SR 18-5: Interagency Statement Clarifying the Role of Supervisory Guidance (Sept. 12, 2018), https://www.fdic.gov/news/press-releases/2018/pr18059a.pdf [https://perma.cc/J6HJ-NHHF]. 214 See, e.g., Randal K. Quarles, Vice Chair for Supervision, Remarks at the American Bar Association Banking Law Committee: Spontaneity and Order: Transparency, Accountability, and Fairness in Bank Supervision (Jan. 17, 2020), No. 1:279] BANK SUPERVISION AND ADMINISTRATIVE LAW 349 same banking interests, the banking agencies together proposed a rule to make a modified version of 2018 statement internally binding.215 Early in 2021 the three agencies separately adopted final versions of essentially the same rule.216 Although this Section uses this rule as a starting point for my discussion, it is concerned less with the policy choices of a specific set of banking agency leaders than with the status of supervisory guidance in administrative law. 1. Supervisory Guidance and MRAs. The view of supervisory guidance held by the banking agencies under the leadership of President Trump’s appointees is summarized in the interagency “guidance on guidance”: The Board issues various types of supervisory guidance, including interagency statements, advisories, bulletins, policy statements, questions and answers, and frequently asked questions, to their respective supervised institutions. A law or regulation has the force and effect of law. Unlike a law or regulation, supervisory guidance does not have the force and effect of law, and the agencies do not take enforcement actions based on supervisory guidance. Rather, supervisory guidance outlines the Board’s supervisory expectations or priorities and articulates the Board’s general views regarding appropriate practices for a given subject area. Supervisory guidance often provides examples of practices that the agencies generally consider consistent with safety- and-soundness standards or other applicable laws and https://www.federalreserve.gov/newsevents/speech/quarles20200117a.htm [https://perma.cc/ZR5P-MPX7]. 215 Role of Supervisory Guidance, 85 Fed. Reg. 70,512, 70,515 (proposed Nov. 5, 2020) (to be codified at 12 C.F.R. pts 4, 262, 302, 791, 1074). 216 Role of Supervisory Guidance, 86 Fed. Reg. 9260 (Feb. 12, 2021) (to be codified at 12 C.F.R. pt. 4) (OCC final rule); Role of Supervisory Guidance, 86 Fed. Reg. 120,79 (Mar. 2, 2021 (to be codified at 12 C.F.R. pt. 302) (FDIC final rule); Role of Supervisory Guidance, 86 Fed. Reg. 18,173 (Apr. 8 2021) (to be codified at 12 C.F.R. pt. 262) (Board of Governors final rule). 350 COLUMBIA BUSINESS LAW REVIEW [Vol. 2022 regulations, including those designed to protect consumers. Supervised institutions at times request supervisory guidance, and such guidance is important to provide insight to industry, as well as supervisory staff, in a transparent way that helps to ensure consistency in the supervisory approach.217 This statement is notable for several reasons. First, it reiterates that supervisory guidance is not legally binding, as a regulatory rule or agency adjudication would be. Just about everyone agrees with that proposition. Second, in saying that the agencies “do not take enforcement actions based on supervisory guidance,” the concept of guidance implicit in the statement departs somewhat from the idea of guidance often articulated by courts and other agencies. As “general statements of policy,” guidance is frequently characterized in those other contexts as expressing an agency’s enforcement priorities or intentions. The legal issue often, and often confusingly, then turns on whether the guidance has left any room for enforcement discretion. Here, though, the banking agencies have created their own ambiguity. While they echo the position of some banking interests that attempt to detach guidance from enforcement altogether,218 it is not clear how fully they are embracing that position. The ambiguity is underscored by a third point about the rule, which is really a question: If supervisory guidance is not a basis for enforcement, what is the supposed significance of the supervisory “expectations” and “approach” referred to in the statement? As this Section will show, answering that question helps explain why current administrative law doctrine on guidance seems so inapposite to the supervisory function. Before turning to that question, a bit more factual background on guidance and MRAs is needed. 217 12 C.F.R. pt. 262, app. A (footnote omitted). This language, which also appears in the OCC and FDIC rules, is unchanged from the 2018 interagency statement on interagency guidance. 218 See Baer Testimony, supra note 14, at 29 (“[T]he law [is] becom[ing] clearer that guidance is nonbinding and cannot serve as the basis for an MRA.”). No. 1:279] BANK SUPERVISION AND ADMINISTRATIVE LAW 351 Guidance is often put out for notice and comment, especially when there is no special urgency associated with the subject. It is sometimes issued jointly by two or more banking agencies. Even where the banking agencies agree on identical guidance, they issue it in somewhat different ways.219 Periodically, previously issued guidance documents will be withdrawn because they responded to a transitory circumstance, have become outdated, or have been superseded by subsequent guidance or regulations.220 The amount of detail varies considerably among guidance documents. Guidance generally does not identify a specific practice or safeguard as best, much less required. However, some guidance contains extended lists of considerations or qualitative objectives relevant to some aspect of bank operations. Some guidance identifies elements of risk management that supervisors regard as important. So, for example, while the interagency guidance on home equity 219 The OCC addresses “OCC Bulletins” to both banks and examination personnel. See, e.g., Off. of the Comptroller of the Currency, OCC 2022-1: Civil Money Penalties: Notice Adjusting Maximum Civil Money Penalties for 2022 (Jan. 19, 2022), https://www.occ.gov/news- issuances/bulletins/2022/bulletin-2022-1.html [https://perma.cc/5CXK- X677] (addressing bulletin to “Chief Executive Officers of All National Banks, Federal Savings Associations, and Federal Branches and Agencies; Department and Division Heads; All Examining Personnel; and Other Interested Parties”). The FDIC publishes its “Financial Institution Letters” without explicit addressees, but clearly directed at banks. In addition to the heads of supervision at each of the regional Reserve banks, the Federal Reserve often also addresses its written guidance to banks. See, e.g., Bd. Of Governors of the Fed. Rsrv. Sys., SR 22-2: Status of Covered Savings Associations and Holding Companies of Covered Savings Associations Under Statutes and Regulations Administered by the Federal Reserve (Jan. 13, 2022), https://www.federalreserve.gov/supervisionreg/srletters/SR2202.htm [https://perma.cc/4ADK-FGWM] (“To the office in charge of supervision and appropriate supervisory and examination staff at each federal reserve bank and banking organizations supervised by the federal reserve.”). 220 In 2016, for example, the Federal Reserve withdrew seventy-eight guidance documents. Bd. of Governors of the Fed. Rsrv. Sys., SR 16-9: Inactive Supervisory Guidance (Apr. 21, 2016), https://www.federalreserve.gov/supervisionreg/srletters/sr1609a1.pdf [https://perma.cc/43TN-7YEK]. 352 COLUMBIA BUSINESS LAW REVIEW [Vol. 2022 loans that was issued in 2014 did not prescribe a specific process that banks must follow as the draw-down period for the loan nears its end, it did include the following: Management should begin reaching out to borrowers well before their scheduled end-of-draw dates to establish contact, engage in periodic follow-up with borrowers, and respond effectively to issues. Lenders often find that successful outreach efforts start at least six to nine months or more before end-of-draw dates, with simple, direct messaging. Many successful programs have required several attempts to contact borrowers to achieve the most effective timing and messaging.221 As noted briefly in Part II, MRAs are a form of administrative action peculiar to a specifically supervisory function. They are not enforcement actions. They are typically communicated as part of a report to a bank following a supervisory examination. The Federal Reserve describes them as follows: MRAs call for action to address weaknesses in processes or controls that could lead to deterioration in a banking organization’s soundness; may result in harm to consumers; or that have caused, or could lead to, noncompliance with laws and regulations. When weaknesses are acute or protracted, Federal Reserve examiners may recommend that management take action more quickly by issuing a “matter requiring immediate attention,” or MRIA.222 While an MRA is not itself an indication that an institution is currently in an unsafe or unsound condition or otherwise troubled, the Federal Reserve notes that a large number of 221 Press Release, Off. of the Comptroller of the Currency, Bd. of Governors of the Fed. Rsrv. Sys., Fed. Deposit Ins. Corp., Nat’l Credit Union Admin. & Conf. of State Bank Supervisors, Interagency Guidance on Home Equity Lines of Credit Nearing Their End-of-Draw Periods (July 1, 2014), https://www.fdic.gov/news/press-releases/2014/pr14052a.html [https://perma.cc/7A8P-UUNB]. 222 BD. OF GOVERNORS OF THE FED. RSRV. SYS., SUPERVISION AND REGULATION REPORT, supra note 73, at 16. No. 1:279] BANK SUPERVISION AND ADMINISTRATIVE LAW 353 outstanding MRAs may result in a downgrade to a banking organization’s supervisory rating.223 MRAs are usually resolved without escalation to a formal enforcement action, but the failure to correct deficiencies cited in prior MRAs may be a factor in the agency’s decision to bring a formal enforcement action, as well as in its evaluation of the overall condition of the institution.224 2. Interaction of Guidance, MRAs, and Ratings One probable reason for the confusion in judicial doctrine is that “guidance” takes so many forms, and has such different effects, across agencies.225 In evaluating claims that a particular agency pronouncement should have followed a notice-and-comment rulemaking process, courts need to grapple with these variations in practice. To illustrate the unique set of issues raised by the interaction of guidance, MRAs, and ratings decisions in bank supervision, consider the following hypothetical situation. This hypothetical is based on actual published guidance covering model risk management.226 It incorporates the issuance of an MRA that refers to the guidance and an eventual ratings downgrade that, in turn, refers back to the MRA. A twenty-one-page document explains the role of quantitative models in banks’ risk management, as well of some of their limitations. Roughly half the document addresses the issue of model validation and controls around model use. Although it does not recommend, much less 223 Id. at 16. 224 Id. at 21. 225 For a survey of the panoply of forms of agency guidance, and the ways agency use it, see Nicholas R. Parrillo, Federal Agency Guidance and the Power to Bind: An Empirical Study of Agencies and Industries, 36 YALE J. ON REGUL. 165 (2019). 226 Bd. of Governors of the Fed. Rsrv. Sys., SR 11-7 Attachment: Supervisory Guidance on Model Risk Management (Apr. 4, 2011), https://www.federalreserve.gov/supervisionreg/srletters/sr1107.htm [https://perma.cc/49ZH-DLNQ] [hereinafter “Model Risk Guidance”]. This guidance was issued jointly with the Office of the Comptroller of the Currency. 354 COLUMBIA BUSINESS LAW REVIEW [Vol. 2022 mandate, specific practices, it does set forth and explain three core elements of an effective validation framework. That explanation refers to a series of processes and safeguards that banks should have in place to manage pertinent risks properly. The text includes hortatory, fairly general statements, such as “model aspects should be subjected to critical analysis by both evaluating the quality and extent of developmental evidence and conducting additional analysis and testing as necessary.”227 Suppose now that a state member bank used a consulting firm to develop and update some of the models it uses for assessing the credit risk associated with its retail and commercial lending. In their regular monitoring and examination of the bank’s operations, Federal Reserve supervisors discovered that the bank’s models had not been updated to take account of relevant changes in delinquency and default experience. This failure has likely caused the model outputs to understate significantly the bank’s potential exposure in the event of even a mild recession. Upon further investigation, the supervisors found that bank employees conducted little oversight of the vendor’s maintenance of the model. While the bank routinely provided the vendor with information on the bank’s credit experience with the exposures being modeled, employees did not regularly check to see if that information had been appropriately incorporated into the revised model code. The supervisors issued an MRA criticizing the failure of the bank to validate the model in accordance with the expectations laid out in the guidance. In doing so, they cited 227 Id. at 11. The guidance continues in a similar vein: Comparison to alternative theories and approaches should be included. Key assumptions and the choice of variables should be assessed, with analysis of their impact on model outputs and particular focus on any potential limitations. The relevance of the data used to build the model should be evaluated to ensure that it is reasonably representative of the bank’s portfolio or market conditions, depending on the type of model. Id. No. 1:279] BANK SUPERVISION AND ADMINISTRATIVE LAW 355 to numerous parts of the guidance, including that “[b]anks are expected to validate their own use of vendor products.”228 Then, in their annual ratings assignment exercise a few months later, supervisors downgraded the management and overall ratings of the bank. As is customary in Federal Reserve practice, the supervisors sent a letter explaining each of the component ratings. In explaining the management downgrade, the letter characterized the bank’s model validation process as wholly inadequate and its oversight of its vendor as not in keeping with sound risk management principles. While there was no mention of the guidance document, several shortcomings identified in the letter paralleled considerations included in that document. The letter went on to say that problems with the bank’s model raised significant questions about the bank’s capacity to manage its lending in a safe and sound fashion. It further indicated that supervisors had observed little progress in remediation of these shortcomings following the issuance of the MRA. How would this set of actions fare under current administrative law if reviewed by a court following a challenge by the bank? As noted earlier, it is difficult to say with complete confidence. To qualify for the § 553(b) exception as a “general statement of policy,” guidance may not be legally binding.229 As with all banking agency guidance, the supervisory guidance on model risk management does not by its terms purport to be a legislative rule changing legal rights or obligations of the bank.230 For many years, that fact would 228 Id. at 16. 229 5 U.S.C. § 553(b) (2018). 230 See Pac. Gas & Elec. Co. v. Fed. Power Comm., 506 F.2d 33, 38–39 (D.C. Cir. 1974). In Pacific Gas & Elec. Co., the court noted: The critical distinction between a substantive rule and a general statement of policy is the different practical effect that these two types of pronouncements have in subsequent administrative proceedings. A properly adopted substantive rule establishes a standard of conduct which has the force of law. In subsequent administrative proceedings involving a substantive rule, the issues are whether the adjudicated 356 COLUMBIA BUSINESS LAW REVIEW [Vol. 2022 likely have been the end of the matter for analytic purposes. In the last couple of decades, however, many courts have applied some version of a “practically” binding test for determining whether the guidance needed to go through APA notice-and-comment procedures for rulemaking.231 I use the term “some version” because courts do not consistently articulate a particular statement of the test. So, for purposes of this illustrative hypothetical, I will use two formulations that, while consistent with one another, have somewhat different emphases. In General Electric Company v. EPA, an oft-cited 2002 case, the D.C. Circuit stated that “an agency pronouncement will be considered binding as a practical matter if it either appears on its face to be binding or is applied by the agency in a way that indicates it is binding.”232 More recently the Fifth Circuit stated the test somewhat differently, even as it quoted General Electric in elaborating its approach: “We evaluate two facts conform to the rule and whether the rule should be waived or applied in that particular instance. The underlying policy embodied in the rule is not generally subject to challenge before the agency. A general statement of policy, on the other hand, does not establish a ‘binding norm.’ It is not finally determinative of the issues or rights to which it is addressed. The agency cannot apply or rely upon a general statement of policy as law because a general statement of policy only announces what the agency seeks to establish as policy. A policy statement announces the agency’s tentative intentions for the future. When the agency applies the policy in a particular situation, it must be prepared to support the policy just as if the policy statement had never been issued. An agency cannot escape its responsibility to present evidence and reasoning supporting its substantive rules by announcing binding precedent in the form of a general statement of policy. Id. (internal footnotes and quotations marks omitted). 231 For a summary and critique of this line of cases, see Cass R. Sunstein, “Practically Binding”: General Policy Statements and Notice-and- Comment Rulemaking, 68 ADMIN. L. REV. 491 (2016). 232 Gen. Elec. Co. v. EPA, 290 F.3d 377, 382 (D.C. Cir. 2002) (citations omitted). No. 1:279] BANK SUPERVISION AND ADMINISTRATIVE LAW 357 criteria to distinguish policy statements from substantive rules: whether the rule (1) impose[s] any rights and obligations and (2) genuinely leaves the agency and its decision-makers free to exercise discretion.233 Although the Fifth Circuit quoted the General Electric formulation in explaining its second criterion, it followed its own statement of the test in focusing more on the degree to which the agency statement was “practically binding” on agency personnel, as opposed to the private actors being regulated. Both kinds of “practically binding” analysis may be found in the caselaw. It is not always clear whether the different emphases reflect different understandings of the test or simply the greater relevance of one or the other to the facts in each case. Here, the bank would argue that the “guidance” establishes rule-like requirements for model management and validation. It could point to the comprehensiveness of the guidance and to its assertion that “all banks should confirm that their practices conform to the principles in this guidance for model development, implementation, and use, as well as model validation.”234 The detail in the rest of the guidance document is not, the bank would argue, a mere statement of enforcement priorities or an indication of a regulatory approach the agencies intend to develop in the future. In the words of the D.C. Circuit decision finding an EPA “guidance” document to have required notice-and-comment rulemaking, “[i]t commands, it requires, it orders, it dictates.”235 Banks are instructed to establish and maintain practices to meet the expectations of the guidance, and supervisors are instructed to be sure they do. The Federal Reserve could respond that its guidance document was meant primarily to educate both supervisors and banks as to the problems associated with risk modeling. 233 Texas v. United States, 809 F.3d 134, 171 (5th Cir. 2015), aff’d per curiam mem. by equally divided court, 579 U.S. 547 (2016) (internal quotation marks omitted) (quoting Profl’s & Patients for Customized Care v. Shalala 56 F.3d 592, 595 (5th Cir. 1995)). 234 Model Risk Guidance, supra note 226, at 21. 235 Appalachian Power Co. v. EPA, 208 F.3d 1015, 1023 (D.C. Cir. 2000). 358 COLUMBIA BUSINESS LAW REVIEW [Vol. 2022 While the document speaks in places about what banks “should” do,236 those are hortatory statements principally illustrating the need for banks to take account of risks associated with modeling. That language contrasts with legally binding regulations, which generally use the verbs “shall” or “must,” rather than “should.”237 Furthermore, the guidance does not prescribe specific protocols or practices for banks. While the document describes the considerations and experience that will inform supervisory assessment of any specific bank, it does not provide a rule-like basis for determining whether the bank’s approach is consistent with its safe and sound operation. Indeed, the Federal Reserve could counter the above-quoted language that the bank might argue is rule-like by noting the language that immediately follows: “Details of model risk management practices may vary from bank to bank, as practical application of this guidance should be commensurate with a bank’s risk exposures, its business activities, and the extent and complexity of its model use.”238 Thus, the Federal Reserve could argue, the guidance is only the starting point for the supervisory analysis that follows. Whatever formulation of the test they choose, most courts end up assessing agency action other than the “guidance” statement that is directly at issue, because it is often through these other actions that the practically binding effect of a legally non-binding statement becomes apparent. Here the 236 E.g., Model Risk Guidance, supra note 226, at 16 (“Banks should require the vendor to provide developmental evidence explaining the product components, design, and intended use, to determine whether the model is appropriate for the bank’s products, exposures, and risks.” (emphasis added)). 237 Courts may place considerable weight on an agency’s choice between clearly imperative, or more qualified, language. Compare Appalachian Power Co., 208 F.3d at 1023 (“[T]he entire Guidance, from beginning to end—except the last paragraph—reads like a ukase.”) with Ctr. for Auto Safety v. Nat’l Highway Traffic Safety Admin., 452 F.3d 798, 809 (D.C. Cir. 2006) (pointing out agency’s use of terms such as “in general” and “would not normally” in finding agency communication not required to be issued through notice-and-comment procedures). 238 Model Risk Guidance, supra note 226, at 21. No. 1:279] BANK SUPERVISION AND ADMINISTRATIVE LAW 359 focus would be on the subsequent issuance of an MRA and eventual downgrade of the bank’s management rating. Banking interests have argued that “violation” of guidance should not be the basis for MRAs or ratings downgrades, precisely because the guidance is supposed to be legally non- binding.239 Because a ratings downgrade can result in binding restraints on acquisitions or other activities by banking organizations, a bank might argue that the invocation of the model risk management guidance in my hypothetical amounts to its having been “applied by the agency in a way that indicates it is binding,” in the terms of the General Electric test. Even if a ratings downgrade does not result, the bank might contend that the guidance is binding because it felt compelled to act in response to the MRA to avoid other supervisory consequences. The Federal Reserve, in turn, could argue first that an MRA is not even a final administrative action within the meaning of the APA and thus not subject to judicial review.240 An agency action is final only if it is “both ‘the consummation of the agency’s decision making process’ and a decision by which ‘rights or obligations have been determined’ or from which ‘legal consequences will flow.’”241 In and of itself, an MRA does not appear to satisfy the second part of the test. Ratings decisions, even those involving only the non-capital elements of bank ratings, may not be immediately reviewable either, though once concrete consequences for a bank follow from a ratings decision, the argument for administrative finality becomes stronger.242 239 See Baer Testimony, supra note 14, at 35. 240 See 5 U.S.C. § 704 (2018) (“A preliminary, procedural, or intermediate agency action or ruling not directly reviewable is subject to review on the review of the final agency action.”). 241 National Mining Ass’n v. McCarthy, 758 F.3d 243, 250 (D.C. Cir. 2014) (quoting Bennett v. Spear, 520 U.S. 154, 177–78 (1997)) (emphasis in original). Additional discussion of the final agency action rule is at supra notes 134–138. 242 Judge Easterbrook suggested as much in Builders Bank, 846 F.3d at 275. But since the FDIC had acquiesced in immediate review of its rating, the court was not required to decide the issue head-on. Id. 360 COLUMBIA BUSINESS LAW REVIEW [Vol. 2022 Second, the Federal Reserve could argue that, in the context of banking law, the issuance of supervisory determinations such as MRAs is specifically contemplated by Congress. As discussed in Part II, numerous statutes contemplate supervisory action as distinct from “regulatory” action in the sense of notice-and-comment rules, or “enforcement” action in the sense of a proceeding resulting in a formal, legally binding order. The agency could argue that, in providing for an independent, intra-agency channel for appealing “material supervisory determinations,” Congress has recognized that supervisory actions can have an impact on bank practices short of legally enforceable orders or even ratings decisions.243 More broadly, the statutory foundations for a supervisory function should be relevant in judicial evaluations of procedural requirements for the exercise of that function. Finally, the Federal Reserve can argue that the supervisory examination that preceded issuance of the MRA itself entailed an evaluation of the actual practices of the bank against the standard of potentially unsafe or unsound banking practices, as set forth in § 1818. A decision on a ratings downgrade would similarly involve an overall assessment of risk management by the bank. So, unlike cases in which a permit is alleged to have been denied for failure to follow a specific piece of agency guidance,244 or where an agency inspection was more likely to occur if a firm did not take on obligations beyond those required by statute and regulation,245 here the banking agency might insist that its MRA reflected a kind of de novo conclusion concerning the potential for the bank’s model risk management to pose a potential risk to its safety and soundness. Thus, while the 243 12 U.S.C. § 4806(a) (2018). To be clear, § 4806 does not include MRAs in its definition of “material supervisory determinations.” Id. § 4806(f)(1). The point of this argument by the Federal Reserve would be that Congress contemplates supervisory actions that have a practical effect on banks while falling short of being legally binding. 244 See Pac. Gas & Elec. Co., 506 F.2d at 36. 245 See Chamber of Com. v. U.S. Dep’t of Lab., 174 F.3d 206, 208 (D.C. Cir. 1999). No. 1:279] BANK SUPERVISION AND ADMINISTRATIVE LAW 361 guidance certainly reflected an agency policy, the policy was applied to the “particular situation,”246 and the conclusion that the bank’s practices required remediation was in fact supportable without reference to the guidance. The bank was not “practically bound” to adopt specific practices, and the supervisors were not practically bound to find the bank’s practices problematic by the fact of the guidance alone. This last point suggests that, as with the General Electric formulation, the Fifth Circuit’s test of whether the supervisors are “genuinely” free to exercise their discretion may accommodate arguments on both sides. The bank can argue that the guidance essentially required the supervisors to take action if the bank itself had not, for example, put in place procedures to evaluate assumptions and “alternative theories.”247 The agency, again, could argue that the supervisors evaluate model risk management against the potential for safety or soundness problems, and that they retain discretion to decide in each case whether a bank’s practices meet that threshold. The guidance, in other words, is just that. By now the temptation, if not necessity, for a court to delve much more deeply into the facts must be obvious. Will a practice of conducting lengthy discussions among supervisors concerning the adequacy of a specific bank’s practices before MRAs are issued or ratings changed indicate that the guidance itself is not really binding, even if it is mentioned? Must a court inquire not only into the decision-making process for the bank challenging the guidance, but also the same process for every bank to which the guidance applies, in order to learn how much real discretion exists? After all, if some banks have avoided MRAs and ratings downgrades even where their systems do not track the guidance closely, the agency’s arguments that the guidance is not practically binding in the context of the challenging bank will be strengthened. On the other hand, if the failure of banks to 246 Pac. Gas & Electric Co., 506 F.2d at 39. 247 See Model Risk Guidance, supra note 226, at 11 (“Comparison to alternative theories and approaches should be included.”). 362 COLUMBIA BUSINESS LAW REVIEW [Vol. 2022 implement a kind of validation practice mentioned in the guidance consistently results in MRAs being issued, does at least that piece of guidance begin to look more “binding”?248 And if one part of the guidance is quite specific, even if most of the same guidance document is not, does that make the guidance look more binding?249 While I think the agency would have the better of the argument in the hypothetical case presented here, it would be 248 Quite apart from the potential breadth and complexity of the inquiry imagined by the questions in the text, there could be practical concerns about maintaining the confidentiality of supervisory information pertaining to all the banks not involved in the litigation challenge by one bank to agency guidance. 249 For example, the Interagency Guidance on Leveraged Lending, 78 Fed. Reg. 17,766 (notice Mar. 22, 2013) largely took the same approach as the Model Risk Guidance, supra note 226—that is, a lengthy explanation of risks, along with various fairly generally stated considerations and concerns for managing risks. But it also included the following: “Generally, a leverage level after planned asset sales (that is, the amount of debt that must be serviced from operating cash flow) in excess of 6x Total Debt/EBITDA raises concerns for most industries[.]” Interagency Guidance on Leveraged Lending, 78 Fed. Reg. at 17,773. While the qualified “generally” suggested that not every loan greater than six times EBITDA (earnings before interest, taxes, depreciation, and amortization) posed a safety and soundness risk, and in fact numerous banks continued to make such loans following the issuance of the guidance, banks might have argued that the specificity of the formula had effectively removed line supervisors’ discretion, at least to the degree of requiring an examination of any such loans. That is, the bright line established by the guidance itself made it more “binding.” In fact, supervisory discretion to decline to act in the fact of such loans remained. But it is certainly the case that the objective formulation provided a starting point for supervisors in the effort to moderate the growth of what appeared to be inadequately underwritten leveraged loans. Still, the starting point was just that. The subsequent supervisory inquiry engaged bank officials, who could (and did) try to show why the loans in question were not excessively risky because of factors not captured in the EBITDA metric. In fact, the complaint of banks and their supporters about more precise thresholds such as the EBITDA formula is somewhat ironic since banks themselves often ask for such specifics to help them better manage their own supervisory relationships. A cynical, but perhaps accurate, view is that banks are in favor of bright line guidance thresholds providing safe harbors but not bright line thresholds resulting in greater supervisory scrutiny. No. 1:279] BANK SUPERVISION AND ADMINISTRATIVE LAW 363 rash to predict with certainty how a court would decide. The muddled state of doctrine on the policy statement exception to notice-and-comment requirements may be due in part to its being often bound up with the closely related doctrine of administrative finality, and in part to differing predispositions among circuit court judges. But it is probably also a byproduct of the way in which cases posing this issue are presented to courts, an explanation to which I now turn as a prelude to offering an alternative approach to administrative law’s application to banking supervision. 3. The Poor Fit Between Doctrine and the Supervisory Function The hypothetical fashioned for the preceding discussion was designed to present a somewhat difficult case under current doctrine. To illustrate the mismatch between current administrative law doctrine and banking supervision, consider a variant on the hypothetical that presents a much sharper case. Suppose now that, in response to the growing use of artificial intelligence (AI) in credit models intended to largely displace humans from involvement in certain classes of credit risk decisions, the banking agencies issue guidance to supplement the existing model risk management guidance. The new guidance indicates that the use of AI in place of loan officers is acceptable for certain classes of smaller denomination loans to individuals and businesses, but that the relative novelty of an underwriting approach with minimal or no human involvement requires special attention to model validation and outcomes analysis. Accordingly, the guidance states that, for the time being, a comparison of actual experience with loan performance and model predictions should be conducted at least monthly to be sure that unanticipated risks will not build. The guidance also sets forth in some detail the kinds of model development, validation, and governance considerations addressed in the existing Model Risk Management guidance, reformulated to address the differences in an AI underwriting process from conventional practices. Like existing guidance, the 364 COLUMBIA BUSINESS LAW REVIEW [Vol. 2022 supplemental document does not include specifics such as the acceptable quantitative deviation of actual loan performance from model predictions. Suppose now that supervisors automatically issue MRAs to any bank that does not conduct the predictions/outcomes analysis at least once a month. They nearly automatically issue MRAs to any bank that has not demonstrably incorporated the processes for development, validation, and governance. While, again, it is clear that the specifics of those processes may vary, it is equally clear that supervisors will not be receptive to arguments that a form of validation identified in the guidance is unnecessary for a particular bank’s model. Finally, suppose that a failure to remediate AI model shortcomings identified in MRAs has been highly correlated with downgrades in management ratings or with a continuation of already downgraded ratings. The foregoing hypothetical is quite unrealistic in light of past supervisory practice, but its sharpness will be useful for heuristic purposes. The factors mentioned in cases such as General Electric make it more likely than in the previous hypothetical that a court would conclude the guidance should have been issued as a legislative rule (though the doctrinal confusion is such that here too the outcome is uncertain). It obviously removes many, if not most, of the complicating questions around “practically binding” assessments. Almost on its face, the fact pattern implies that the discretion of line supervisors to engage in a kind of de novo assessment of a firm’s AI model risk management has been substantially constrained. And there are near automatic negative consequences for any banking organization that “violates” the guidance, in the sense that the bank has not adopted risk management practices that take account of the concerns and considerations reflected in the supervisory document. A court ruling that the AI guidance did not qualify for the “general statement of policy” exception would carry with it the unstated premise that the agency’s regulatory aim could be achieved through notice-and-comment rulemaking, but with more procedural formality. In bank supervision, however, this premise may be unfounded. This possibility can be explored No. 1:279] BANK SUPERVISION AND ADMINISTRATIVE LAW 365 by asking why the agencies might have chosen guidance in the first place. The most obvious answer is the one most criticized by those who would significantly restrict the scope of the general statement of policy exception—the delay associated with a notice-and-comment process.250 However, in opting for guidance over rulemaking the banking agencies may not have been motivated solely, or even principally, by time considerations. Indeed, the banking agencies frequently do put proposed guidance out for comment, though the process for guidance is typically shorter than that for a legislative rule.251 But, like the Model Risk Management guidance, most supervisory guidance documents are not especially rule-like. Instead, they provide an explanation, sometimes extensive, of relevant factors and risks associated with a bank process or product. Even when 250 The agencies might well fear that banks could unknowingly assume considerable risk from reliance on AI credit models in the year or more it would take to issue a rule. To the degree that these circumstances are truly exigent, an agency could invoke the § 553(b) exception where notice-and- comment procedures would be “impracticable.” 5 U.S.C. § 553(b) (2018) (“[T]his subsection does not apply . . . when the agency for good cause finds . . . that notice and public procedure thereon are impracticable[.]”). Courts have at times been somewhat grudging in applying this exception and, where they do, they often expect that the agency will conduct a notice-and- comment rulemaking process after it has put an interim rule into immediate effect. See JOHN F. MANNING & MATTHEW C. STEPHENSON, LEGISLATION AND REGULATION: CASE AND MATERIALS 931–32 (4th ed. 2021). Following Little Sisters of the Poor Saints Peter & Paul Home v. Pennsylvania, 140 S.Ct. 2367 (2020), agencies may have greater scope for using interim final rules. However, while the majority’s permissive approach was stated in broad terms with no hint of special circumstances in that case, id. at 2385–86, it remains to be seen whether the Court will take a similarly deferential approach to agencies in future cases in which it has less sympathy for the agency’s action. 251 This difference is at least partly because the agencies will not need to adopt the litigation defense mode associated with notice-and-comment rulemaking, such as responding to virtually every unique comment and including long, often detailed explanations and justifications of the rule in the Federal Register notice. Both these practices are defensive measures taken in response to judicial decisions invalidating agency rules as either procedurally deficient or arbitrary and capricious. 366 COLUMBIA BUSINESS LAW REVIEW [Vol. 2022 guidance documents are relatively lengthy, they generally do not include anything like the specificity one would find in legislative rules such as capital and liquidity requirements, counterparty credit exposure limits, and permissible activities for affiliates of bank holding companies. In the present hypothetical, the agencies would be unable to draft rules for the use of AI models in credit decisions that even approached this level of specificity. Experience with these models is limited, the nature and use of the models is evolving rapidly, and the appropriate risk management practices likely vary materially from one bank to another. If one forced the agency to enact this kind of guidance as a rule, it would necessarily be quite general in places and would still require the very kind of supervisory assessment and bank-specific application contemplated in the guidance before there could be a determination that it had been “violated.”252 As suggested earlier, the alternative to a vague regulatory rule could be a more precise but very blunt and restrictive one. Alternatively, a materially higher capital requirement would try to assure bank resiliency in the face of uncertain risks associated with AI models. An outright prohibition on certain forms of models or certain products associated with them would try to prevent risk management errors associated with new methodologies. Each of these alternatives would almost surely be more socially costly than an iterative process in which supervisors and bank risk managers applied generally stated guidance based on the expertise and experience of each firm. In short, supervisory guidance documents are usually not intended to serve the same purpose as legislative rules. They are not intended to be treated as equivalent to a violation of a regulatory rule or a finding of an “unsafe or unsound” practice. 252 Contrast the supervisory position postulated in the text with the well-grounded belief of some commentators that requiring legislative rules in place of guidance could incentivize agencies to issue vague rules that they would then apply broadly in subsequent enforcement actions. See, e.g., John F. Manning, Nonlegislative Rules, 72 GEO. WASH. L. REV. 893, 896 (2004). With much supervisory guidance, I suggest, the banking agency would not choose a vague rule; it would have little choice other than to adopt one. No. 1:279] BANK SUPERVISION AND ADMINISTRATIVE LAW 367 Instead, they are a critical part of the supervisory function.253 They focus the attention and efforts of line supervisors in fulfilling their statutory and institutional responsibilities for evaluating the conditions and management of banking organizations, especially with respect to activities that are heterogeneous across firms, rapidly changing, or both. They are intended to help supervisors identify vulnerabilities and risks before those problems ripen into serious problems for a bank. Once identified, those risks are addressed through appropriate supervisory channels, ranging from informal communications by dedicated supervisory teams to MRAs to ratings changes and associated remedial programs. In many instances, this approach allows for considerable involvement by the bank itself in framing measures to mitigate the risks identified. This interaction reflects the fact of an ongoing supervisory relationship, as contemplated by Congress and built out by the agencies. These processes have consequences, to be sure. They are supposed to. But those consequences are defined and managed within the supervisory function itself. IV. IS A DIFFERENT APPROACH FEASIBLE? The doctrinal analyses in Part III argued that the statutory foundations of supervision, along with the substantive features of bank regulatory statutes, should be considered in determining the consistency of supervisory activities with administrative law. When they are, those activities should generally fare well against legal challenges. Yet Part III also revealed a conceptual mismatch between significant strands of administrative law and the supervisory function. The mismatch arises because supervision is grounded in an ongoing relationship in which many elements 253 This is not to say that a desire to avoid the procedural and substantive requirements associated with legislative rules has never played a role in agency decisions. I suspect that it has from time to time. The point, rather, is that there is a well-grounded role for guidance documents in supporting the supervisory function, itself demonstrably distinct from regulatory rule enforcement. 368 COLUMBIA BUSINESS LAW REVIEW [Vol. 2022 of both bank operation and agency practice are relevant, whereas much administrative law doctrine focuses on discrete agency actions affecting a regulated entity. As we have seen, the square peg of a supervisory practice may often be slim enough to fit through a round doctrinal hole. But perhaps not always. And some supervisory practices such as ratings decisions may readily fit through even where they implicate the norm of consistent, non-arbitrary administration that informs much of the APA. There may be a similar mismatch between at least some administrative law doctrines and a broader range of regulatory areas. As Nick Parrillo concludes from his invaluable study of the use of guidance across different administrative agencies, key structural features of the administrative apparatus established by Congress motivate agencies to use, and regulated businesses to follow, guidance.254 These incentives exist regardless of whether agency officials subjectively intend to use guidance to legally bind. Parrillo documents and explains “the tendency of heavily regulated businesses to invest in positive relationships to their regulator[.]”255 Because of its historical and statutory roots, banking supervision presents perhaps the exemplary case of the doctrinal incongruity caused by a misunderstanding of how regulatory frameworks are structured. In this Part, I consider how this incongruity might be reduced in a way that pays due regard to both the APA and the statutory framework for banking regulation. A. The Conceptual Mismatch Revisited Some portion of the conceptual mismatch between administrative law and banking supervision is doubtless due to the long history of the latter, which at the federal level dates to the 1863 passage of the National Bank Act, and at the state level decades before that. The cultures and accumulated learning of banking agencies did not necessarily fit neatly with the fairly broad provisions of the APA, which in a sense 254 Parrillo, supra note 225, at 270. 255 Id. at 177. No. 1:279] BANK SUPERVISION AND ADMINISTRATIVE LAW 369 memorialized the resolution of a series of legal and policy disputes on the permissible roles of administrative agencies. Yet one can make the case that in 1946 there was a workable alignment of that codified set of requirements for agency procedures and judicial review, on the one hand, and the role of bank supervision on the other. Early supervision originated in the right to examine banks to determine if bank behavior warranted certain legal consequences—whether a bank’s federal charter should be revoked,256 whether specific bank officials should be removed from the bank,257 or, later, whether federal deposit insurance should be terminated.258 These consequences would ensue after an adjudicatory proceeding, either judicial or administrative. Thus, the broad “visitorial” powers of banking agencies to gain access to bank premises for purposes of examining its books and records helped them to make informed decisions on initiating one of these actions. As the Supreme Court noted in 1963, the availability of these sanctions meant that “recommendations by the agencies concerning banking practices tend to be followed by bankers without the necessity of formal compliance proceedings.”259 Today we would probably characterize these “recommendations” as supervisory guidance, though they were generally more bank-specific and unlikely to take the form of published, detailed documents. Because the case involved antitrust issues, the visitorial and enforcement powers of banking agencies were addressed only in passing. But the Court was clearly unconcerned with the supervisory dynamic. On the contrary, it praised the “efficacy” of this system.260 256 Act of June 3, 1864, ch. 106, § 53, 13 Stat. 99, 116. This Act was subsequently designated the National Bank Act. Act of June 20, 1874, ch. 343, § 1, 18 Stat. 123, 123 (later codified at 12 U.S.C. § 38). 257 Banking Act of 1933, Pub. L. No. 73-66, § 30, 48 Stat. 162, 193. 258 Banking Act of 1935, Pub. L. No. 74-305, § 101, 49 Stat. 684, 690– 91 (amending § 12B of the Federal Reserve Act of 1913, ch. 6, 28 Stat. 251). 259 United States v. Philadelphia Nat. Bank, 374 U.S. 321, 330 (1963). 260 Id. 370 COLUMBIA BUSINESS LAW REVIEW [Vol. 2022 Given the binary and potentially draconian nature of at least two of the three legal consequences just noted, both banks and banking agencies had some incentive to informally resolve supervisory concerns and thereby avoid forcing the agency to either seek the heavy sanction or allow banks to continue operating in ways that caused concern. One imagines, though, that most banks were usually confident that the banking agency would not terminate its deposit insurance in response to anything less than grave concerns about the bank’s condition. That supposition likely explains why the banking agencies continued to press for intermediate sanctions. They got their wish in 1966, when Congress added to the statutory provision for terminating FDIC insurance— the now familiar § 1818(b) authority to act against a bank261— and, after a later 1978 amendment, a director, officer, or employee engaging in any “unsafe or unsound practice.”262 Here, too, the recourse formally provided to supervisors was adjudicatory.263 Section 1818 has since been amended to provide the agencies with additional enforcement powers. The hearings conducted under almost all of those provisions are to be formal adjudications.264 Pre-existing authorities such as the OCC’s non-judicial remedies for a national bank’s violation of its 261 Federal Institutions Supervisory Act of 1966, Pub. L. No. 89-695, § 101, 80 Stat. 1028, 1046–47 (codified as amended at 12 U.S.C § 1818(b)(1)) (amending § 8 of the Federal Deposit Insurance Act of 1950, Pub. L. 81-797, 64 Stat. 873, 879). 262 Financial Institutions Regulatory and Interest Rate Control Act of 1978, Pub. L. No. 95-630, § 107(a)(1), 92 Stat. 3641, 3649 (codified as amended at 12 U.S.C § 1818(b)(1)) (amending § 8 of the Federal Deposit Insurance Act of 1950, Pub. L. 81-797, 64 Stat. 873, 879). 263 In a series of amendments over the years, Congress has added additional enforcement options for the bank regulatory agencies. Most important are the authority to remove or prohibit unsafe or illegal practices by a range of “institution-affiliated parties,” 12 U.S.C. § 1818(e) (2018), and the imposition of civil money penalties against banks and institution- affiliated parties, 12 U.S.C. § 1818(i). The procedural and substantive requirements for these various enforcement provisions vary both with the type of proceeding and with the severity of the penalty sought by the agency. 264 Id. § 1818(h)(1). The standard for judicial review of any enforcement actions is likewise that provided in the APA. Id. § 1818(h)(2). No. 1:279] BANK SUPERVISION AND ADMINISTRATIVE LAW 371 charter obligations similarly incorporated APA hearing and judicial review standards.265 With respect to these formal statutory penalties, then, the relationship between banking law and generally applicable administrative law continues to be both clear and sensible: The APA procedures govern, with whatever glosses the courts may place on the statutory language. Through examinations, and visitorial authority more generally, supervisors can make timely decisions on enforcement. Because statutory standards such as those for termination of deposit insurance and “unsafe or unsound” practices leave considerable room for agency interpretation and discretion, banks may be incentivized to follow supervisory “recommendations” to forestall agency recourse to formal proceedings directed at the bank practices at issue. Of course, this is far from the whole story. However closely the preceding, stylized account may approach the reality of the supervisory function sixty years ago, two legal trends since then have complicated matters considerably. First has been the steadily increasing role of regulatory rules in prudential regulation. As is regularly noted in administrative law discussions, the ascendancy of notice-and-comment rules in agency regulation was not contemplated at the time the APA was debated and enacted. This circumstance is doubtless responsible in part for the confusion attending so many issues in the interpretation of § 553, including the practically binding doctrine. The proliferation of prudentially-motivated rules has also affected the supervisory function by substantially expanding its role of filling in lacunae in regulatory rules. The opaqueness, heterogeneity, and rapidly changing character of modern bank activities pose a considerable challenge in trying to draft regulations. Questions about the consistency of current bank practices with even detailed notice-and-comment rules are routine. The rapid adjustment of banking products or practices, sometimes as a direct result of a regulation, often quickly raises a whole new set of unanticipated issues. Bank agency officials—usually through line supervisors—regularly provide formal or informal 265 See id. §§ 93(b)(6), (d)(1)(C). 372 COLUMBIA BUSINESS LAW REVIEW [Vol. 2022 guidance to banks. This practice both gives banks more certainty on agency views and frequently channels bank behavior. In banking, though, even as the expanding scope of regulatory rules reconfigured the longstanding supervisory function, the existence of that function has in turn been relevant for the formulation of many regulatory rules. As the example of minimum capital requirements in Part II explained, regulatory rules are premised on supervisory exams and—for the largest banks—continuous monitoring. Supervision helps ensure that underlying bank risk management practices are sound and that there is an effective way to address the inevitable unanticipated risks and opportunities for regulatory arbitrage that arise from the characteristics of modern financial activity. This premise is generally implicit, rather than explicitly stated. But that is not surprising, insofar as supervision was well-established before the proliferation of regulatory rules began. If one were to tell the banking agency officials who write and revise regulatory rules that there would no longer be any supervisory exams or continuous monitoring, they would surely rethink the content of those rules. The result would likely be a set of rules intended to capture more bank practices, including practices not specifically anticipated by the regulators. It is quite plausible that such regulatory rules would in the aggregate impose greater costs on current financial intermediation, while yielding lower benefits of reduced risks of bank insolvency and financial instability, than could have been achieved with supervision as a complementary regulatory tool. The second trend has been a series of amendments to banking law that have both established a more explicitly supervisory function for the banking agencies and multiplied the occasions on which supervisory judgment will invariably affect legally required approvals for various bank activities. The set of provisions establishing the supervisory function was explained in Part II. Statutory requirements for regulatory approvals of proposed bank actions or activities have been added by Congress both when it seeks to tighten No. 1:279] BANK SUPERVISION AND ADMINISTRATIVE LAW 373 regulation266 and when it enacts deregulatory measures that reduced structural or activities constraints on banking organizations.267 Requirements for regulatory approval are easy to explain in the former case. They may be less intuitive where Congress is deregulating, but regulatory approval of specific actions has sometimes been the price for expanded powers under the banking laws. These statutes give the banking agencies broad discretion, in part because they mandate consideration of many 266 See, e.g., the Bank Holding Company Act of 1956, Pub. L. No. 84- 511, 70 Stat. 133, which was directed at bank circumvention of pervious banking laws, such as branching restrictions and of the separation of banking from commercial activities. 267 For example, the Riegle-Neal Interstate Banking Act of 1994 removed most statutory prohibitions on interstate bank acquisitions, but required prior approval of any such acquisitions by the Federal Reserve Board. Riegle-Neal Interstate Banking Act of 1994, Pub. L. No. 103-328, §101(a), 108 Stat. 2238, 2339 (codified at 12 U.S.C. §1842(d)(1)) (amending the Bank Holding Company Act of 1956, Pub. L. 84-511, § (3)(d), 70 Stat. 133, 135). Similarly, the Gramm-Leach-Bliley Act, Pub. L. No. 106-102, 113 Stat. 1338 (1999), was an essentially deregulatory piece of legislation, best known for having repealed the Glass-Steagall prohibition on affiliations between insured depository institutions and investment banks. But, even in constraining the authority of the Federal Reserve to oversee the non-bank affiliates within bank holding companies, the Act reflected the assumption of a well-established supervisory function. Indeed, Subtitle B of Title I of the Act is entitled “Streamlining Supervision of Bank Holding Companies.” 113 Stat. at 1362. One provision in this subtitle, § 113, added a new § 10A to the Bank Holding Company Act. § 113, 113 Stat. at 1369. That section, entitled “Limitation on Rulemaking, Prudential, Supervisory, and Enforcement Authority of the Board” provided that the Board may not prescribe regulations, issue or seek entry of orders, impose restraints, restrictions, guidelines, requirements, safeguards, or standards, or otherwise take any action under or pursuant to any provision of this Act or section 8 of the Federal Deposit Insurance Act against or with respect to a functionally regulated subsidiary of a bank holding company unless that action was necessary to protect the depository institution. Id. This ill-advised provision was repealed by the Dodd-Frank Act, Pub. L. No. 111-203, § 604(c)(2), 124 Stat. 1376, 1601 (2010), which provided for much broader Federal Reserve oversight of the non-banking affiliates in bank holding companies. 374 COLUMBIA BUSINESS LAW REVIEW [Vol. 2022 factors.268 The evaluation of some of these factors will be informed by supervisory familiarity, such as the “risk presented by such depository institution to the Deposit Insurance Fund“269 and the potential for “unsound banking practices” if approval were granted.270 Some statutes call explicitly for recognizably supervisory judgments of managerial competence. For example, in deciding whether to approve a bank holding company’s acquisition of a bank, the Board of Governors is instructed to consider seven categories of factors,271 including “[i]n every case . . . the financial and managerial resources and future prospects of the company or companies and the banks concerned,”272 which is further explained as including “consideration of the competence, experience, and integrity of the officers, directors, and principal shareholders of the company or bank.“273 In deciding whether to approve a new branch of a bank, the FDIC is instructed to consider seven factors, including “[t]he general character and fitness of the management of the depository institution.”274 The import of these approval processes is twofold. First, as already suggested, approval of applications will depend in no small part on judgments as to the condition and capabilities of banking organizations already subject to supervision. It is hard to see how they could not be. The staff analysis presented to agency principals of, say, a proposed acquisition will 268 See, e.g., 12 U.S.C. § 1828(d)(1) (2018) (approval required for opening a domestic branch with reference to a list of factors to consider in 12 U.S.C. § 1816); id. § 1842 (prior approval of Board of Governors necessary for banking holding company acquisition of a bank or other holding company); id. § 1843(j) (authority of Board of Governors to disapprove bank holding company engaging in non-banking activities). 269 Id. § 1816(5). 270 Id. § 1843(j)(2)(A). 271 Id. § 1842(c). 272 Id. § 1842(c)(2). 273 12 U.S.C. § 1842(c)(5). 274 Id. § 1816(4). See id. § 1828(d)(1) (referring to the list of factors in 12 U.S.C. § 1816 when considering required approval for opening a domestic branch). No. 1:279] BANK SUPERVISION AND ADMINISTRATIVE LAW 375 necessarily reflect the supervisors’ assessment of these factors based on their prior experience with the bank. Second, with banking organizations aware that these factors will be assessed when they seek to expand or otherwise alter their operations, they will have an incentive to satisfy significant supervisory concerns prior to filing applications for regulatory approval.275 Of course, disapproval of any of these applications is surely a final agency action for which judicial review would be available. But the prior supervisory assessments called forth by the applicable factors will have substantially affected an agency’s evaluation of the application. The preceding recap helps explain the incongruity between contemporary administrative law doctrine and the supervisory function as it exists in the statutory framework for banking regulation. The APA was drafted upon the apparent but erroneous presumption that adjudication would remain the principal interface between agencies and regulated entities. As rulemaking expanded, judicially developed doctrines around rulemaking rested on the apparent but unrealistic presumption that a more or less fastidious promulgation of, and compliance with, regulatory rules was the best way to achieve the public interest in regulation, while guarding against arbitrary or overreaching administrative action. As with the topic of guidance, the disconnect between these and other doctrines on the one hand, and the realities of administrative practice on the other, is by no means limited to bank supervision. Others have examined this problem from a broader administrative law perspective.276 Again, though, the problem seems particularly 275 For an explanation of this incentive across a broader range of administrative areas, see Parrillo, supra note 225, at 192–94. 276 For two recent assessments that adopt very different perspectives but share the premise that judicial review and associated administrative law doctrines misperceive how agencies are structured and function, see Jennifer Nou, Subdelegating Powers, 117 COLUM. L. REV. 473 (2017); William H. Simon, The Organizational Premises of Administrative Law, 78 LAW & CONTEMP. PROBS. 61 (2015). A less recent treatment that identifies the growth of guidance and other informal means of administrative regulation as related to areas of regulation that entail repeated interactions 376 COLUMBIA BUSINESS LAW REVIEW [Vol. 2022 acute in the area of bank supervision, with its long history and statutory foundations. B. Toward Greater Congruence of Supervision and Administrative Law So long as there are few legal challenges to agency actions, there will be little impetus for achieving greater congruence between generally applicable administrative law doctrine and the supervisory function. However, as President Biden replaces the leadership of the banking agencies appointed by President Trump, both the regulatory and supervisory agendas are likely to shift from a direction of moderate deregulation to one of more rigorous oversight. This change may produce the bank-initiated litigation that was rumored to be under consideration prior to the 2016 presidential election. In that event, courts may need to confront head-on the administrative law issues discussed in Part III. The failure of doctrine to take account of the iterative relationship between banking agency officials and banks that characterizes the supervisory function may then lead to problematic decisions that confuse, rather than clarify, the permissible roles of supervision. While there are two readily available doctrinal paths that would produce more clarity,277 neither would reconcile the supervisory function with the norms of consistency, fairness, and accountability that motivate the provisions of the APA. The route to doing so may instead require a departure from the trans-substantive tendencies of contemporary administrative law, in favor of one that integrates the specifics of substantive agency mandates with procedural between agencies and economic actors can be found in Todd D. Rakoff, The Choice Between Formal and Informal Modes of Administrative Regulation, 52 ADMIN. L. REV. 159 (2000). 277 The first doctrinal path is laid out in Baer Testimony, supra note 14, at 35. The second has been explained and advocated for by Professors Gersen, Manning, and Elliott. See Jacob E. Gersen, Legislative Rules Revisited, 74 U. CHI. L. REV. 1705, 1712–13 (2007); John F. Manning, Nonlegislative Rules, 72 GEO. WASH. L. REV. 893, 896 (2004); E. Donald Elliott, Re-Inventing Rulemaking, 41 DUKE L. J. 1490, 1490 (1992). No. 1:279] BANK SUPERVISION AND ADMINISTRATIVE LAW 377 forms and requirements. After explaining why the more conventional approaches share a problematic conceptual premise with current doctrine, I will outline such an alternative. 1. Trans-Substantive Options for Reconciling Administrative Law and Supervision One alternative doctrinal path would be to extend the reasoning of court of appeals cases invalidating agency guidance that is “practically binding” on regulators and regulated entities to prohibit nearly any supervisory action other than those directly tied to notice-and-comment rulemaking or formal enforcement actions. In the rendition favored by some banking interests, this route would come close to eliminating the supervisory function by prohibiting any other communication that a bank might interpret as forcing a change in its practices. Guidance would effectively be relegated to the status of advice to banks on good practices. Written guidance and, presumably, less formal supervisory communications could not be considered as supervisors made decisions on MRAs, ratings, and other actions. MRAs would need to be based on a “violation[] of law.”278 This alternative would certainly go a long way to resolve any doctrinal ambiguity around supervisory actions such as stress testing, the use of guidance, and the relationship between guidance and MRAs. But it would do so by effectively reading supervision out of the law, without reconciling administrative law doctrines with the statutory basis for supervision.279 278 Baer Testimony, supra note 14, at 35. While their proposed rulemaking on the role of guidance moves several steps down this alternative road, the agencies have for now stopped short of this last demand by the petitioning banking interests that MRAs be reduced to courtesy notices to banks that they may be violating the law. See Role of Supervisory Guidance, 85 Fed. Reg. 70,512, 70,515–16 (proposed Nov. 5, 2020) (to be codified at 12 C.F.R. pts. 302, 791, 1074). 279 That is precisely what some banking interests have tried to do, despite the statutory foundations of supervision. See Baer Testimony, supra note 14. 378 COLUMBIA BUSINESS LAW REVIEW [Vol. 2022 A second, equally clean doctrinal path would focus solely on whether a supervisory communication is directly legally binding. This route would rest heavily on requirements of administrative finality before judicial review is available. It would also follow the argument favored by many academics280 for rejecting “practically binding” standards in light of the Supreme Court’s prohibition in Vermont Yankee281 and Mortgage Bankers282 of requirements for notice-and-comment rulemaking that go beyond the terse requirements of § 553 of the APA.283 This, of course, is not the direction in which courts have been heading. If judicial receptivity were to materialize, it would be sensible policy in some other areas of regulation, especially those in which subsequent, more formal agency actions are necessary for tangible consequences to ensue. In the banking context, it would be immeasurably better than the approach favored by banking interests, because it recognizes—rather than represses—the degree to which non- legally binding agency communication is critical to a sensibly functioning regulatory system.284 Still, like current doctrine and the alternative preferred by banking interests, it does not reflect the intended and actual nature of the bank supervisory relationship. It would focus predominantly, if not exclusively, on the single, legally binding agency action at the center of the litigation. Without reference to that relationship, it might still find fault with a later, consequential agency action that had been informed by prior supervisory assessments of a bank’s risk management practices. On the other hand, it could permit arbitrary, and even unauthorized, actions by supervisors that 280 See supra note 277. 281 Vermont Yankee Nuclear Power Corp. v. Nat. Res. Def. Council, 435 U.S. 519, 523 (1978). 282 Perez v. Mortgage Bankers Ass’n, 575 U.S. 92, 100 (2015). 283 This argument has not been offered by its academic proponents specifically in the banking law context, but as an argument to clarify the confusing line of cases addressing general statements of policy across the range of administrative activities. See Gersen, supra note 277, at 1712–13. 284 For a comprehensive and convincing demonstration of this broader point, see Parrillo, note 225. No. 1:279] BANK SUPERVISION AND ADMINISTRATIVE LAW 379 do affect bank practices, so long as these actions cannot be linked to definite legal effects in a final agency action. 2. An Alternative Approach Though they would likely reach opposite conclusions in many cases, neither trans-substantive option would reconcile the standards of the APA with those of banking law. An effective reconciliation would need to shape administrative law requirements around the specifics of substantive banking law and the organization of the agencies that Congress has created to administer it.285 While the norms underlying administrative law should be applied in all regulatory areas, including banking, courts could elaborate on the relatively flexible statutory provisions of the APA in a manner that allows for effective and efficient realization of the substantive statutory mandate given to banking agencies. 285 Professors Levy and Glicksman have identified numerous instances of “agency specific precedents,” in which courts appear not to have followed a trans-substantive approach to administrative law, but have instead developed doctrine to ratify an agency specific procedure. Richard E. Levy & Robert L. Glicksman, Agency Specific Precedents, 89 TEX. L. REV. 499, 500 (2011). While the authors are troubled by some such departures from a more universal application of administrative law, they are careful to distinguish instances where substantive enabling statutes have direct bearing on agency procedures: More fundamentally, the organic statute establishes the agency’s substantive mandate, authorizes the agency to take particular kinds of action to fulfill that mandate, and specifies the legal standards for taking such action. In any given administrative law case, the organic statute colors the administrative law issue—it determines what is at stake, dictates the type of action the agency may take to further its statutory mandate, provides the substantive test for determining the propriety of the agency’s action, and governs the kinds of evidence or information the agency (or party) will use to justify (or attack) the agency’s decision. These distinctive components of agencies’ organic statutes limit the universality of administrative law. Id. at 509. 380 COLUMBIA BUSINESS LAW REVIEW [Vol. 2022 To give some concreteness to this idea, I present here the outline of an approach by which courts would evaluate the legality of supervisory actions in the context of the internal processes maintained by a banking agency to implement and oversee its supervisory activities. That is, the legality of a specific supervisory measure would depend in large part on whether it had been taken within an administrative system that effectively managed supervisory relationships against the backdrop of statutory authorities and limits, with appropriate procedures and safeguards against arbitrary supervisory action. The approach I suggest would, in essence, make the quality of “internal” administrative law a central consideration in a court’s external review of supervisory actions. This linkage between the two spheres might seem a bit at odds with arguments that courts should refrain from too much scrutiny of internal administrative law, lest they disincentivize agencies from regularizing their functioning so as to promote internal effectiveness and accountability.286 , Gillian Metzger and Kevin Stack—prominent proponents of giving agencies space for internal administrative law—also recognize that there is “no clear line demarcating the two.” 287 Indeed, they favor encouraging internal administrative law through exercise of judicial deference, though they are skeptical that courts will often be so inclined.288 As explained below, I share some of that skepticism but hold out some hope that courts may be willing to grant some deference to individual, external supervisory actions that come out of a satisfactory internal process. It is important to note that the incongruity that I have identified exists between the supervisory function and some 286 See generally Gillian E. Metzger and Kevin M. Stack, Internal Administrative Law, 115 MICH. L. REV. 1239 (2017). Metzger and Stack catalogue a number of ways in which this can happen, such as by “treating agency attempts to bind internal agency officials as grounds for characterizing an agency rule as a legislative rule requiring notice and comment.” Id. at 1295. 287 Id. at 1251. 288 Id. at 1295–96. No. 1:279] BANK SUPERVISION AND ADMINISTRATIVE LAW 381 contemporary doctrines developed by the judiciary. Most provisions in the APA itself are sufficiently open-textured that its procedural requirements and protections against arbitrary and capricious action could be applied in a way that takes appropriate account of the supervisory relationship. Still, administrative caselaw does provide at least some support for this approach to judicial oversight of bank supervision, though not in contexts that could fairly be cited as direct precedent. I will mention two cases, one addressing bank supervision specifically and the other a much broader comment on an agency’s prerogative in fashioning its procedures. Together, these cases adumbrate the possibilities for a more tailored approach to judicial review of supervisory actions. First is the characterization of the supervisory relationship offered by Judge Douglas Ginsburg, writing for the D.C. Circuit in a case rejecting an effort by bank shareholders to compel the OCC to produce privileged bank exam information in connection with a derivative suit: Bank safety and soundness supervision is an iterative process of comment by the regulators and response by the bank. The success of the supervision therefore depends vitally upon the quality of communication between the regulated banking firm and the bank regulatory agency. This relationship is both extensive and informal. It is extensive in that bank examiners concern themselves with all manner of a bank’s affairs: Not only the classification of assets and the review of financial transactions, but also the adequacy of security systems and of internal reporting requirements, and even the quality of managerial personnel are of concern to the examiners. The supervisory relationship is informal in the sense that it calls for adjustment, not adjudication. In the process of comment and response, the bank may agree to change some aspect of its operation or accounting; alternatively, if the bank and the examiners reach impasse, then their dispute may be elevated for resolution at higher levels within the regulatory agency. It is the very rare dispute, however 382 COLUMBIA BUSINESS LAW REVIEW [Vol. 2022 than culminates in any formal action, such as a cease and desist order.289 The broader discussion of agency procedures is in the Supreme Court’s decision in FCC v. Schreiber,290 cited prominently in the Court’s decision in Vermont Yankee.291 Neither the agency involved nor the specifics of the dispute are especially apposite for determining how administrative law requirements should apply to banking supervision.292 Nor, in light of Vermont Yankee and subsequent cases, does the holding restricting lower courts’ scope for augmenting agency procedures seem surprising. For present purposes, the noteworthy feature of Chief Justice Warren’s opinion is his decidedly non-trans-substantive starting point for analysis: [A]dministrative agencies ‘should be free to fashion their own rules of procedure and to pursue methods of inquiry capable of permitting them to discharge their multitudinous duties.’ This principle, which has been upheld in a variety of applications, is an outgrowth of the congressional determination that administrative agencies and administrators will be familiar with the industries which they regulate and will be in a better position than Federal courts or Congress itself to design procedural rules adapted to the peculiarities of the industry and the tasks of the agency involved. Thus, underlying the broad delegation . . . of procedural rule-making power to the Federal 289 In re Subpoena Served upon the Comptroller of the Currency, 967 F.2d 630, 633–34 (D.C. Cir. 1992) (internal citations omitted). I cannot resist pointing out that, prior to his appointment to the bench, Judge Ginsburg taught banking regulation at Harvard Law School. His consequent familiarity with the supervisory process, rare among judges, is reflected in the opinion. 290 381 U.S. 279, 288–98 (1965). 291 Vermont Yankee Nuclear Power Corp. v. Nat. Res. Def. Council, 435 U.S. 519, 524–25, 543 (1978). 292 Schreiber affirmed the FCC’s decision to reject the request of a party to whom a subpoena had been issued that the Presiding Officer depart from the Commission’s regular practice of public hearings in order to protect the confidentiality of the materials requested in the subpoena. Schreiber, 381 U.S. at 279, 298–99. No. 1:279] BANK SUPERVISION AND ADMINISTRATIVE LAW 383 Communications Commission is a ‘recognition of the rapidly fluctuating factors characteristic of the evolution of broadcasting and of the corresponding requirement that the administrative process possess sufficient flexibility to adjust itself to these factors.’293 A good foundation for building an administrative law regime for bank supervision could be laid by combining Warren’s emphasis on “the peculiarities of the industry and the tasks of the agency involved” with Ginsburg’s insight into the supervisory relationship. Vermont Yankee extended the Schreiber holding by disallowing judicial additions to APA procedures for notice-and-comment rulemaking. Here, courts could apply Chief Justice Warren’s broad point about the distinctive characteristics of each regulatory area by allowing and expecting the banking agencies to develop internal processes for governing supervisory activities that appropriately reflect the necessarily iterative nature of bank supervision. Both participles in the preceding sentence are important. The judiciary should allow the agencies discretion to shape an internal administrative law of supervisory governance based on their knowledge of banking and of their own organizational capacities and limitations. But the judiciary should also make clear its expectation that those internal processes and standards will in fact be implemented. 3. Internal Administrative Law and Judicial Review. Two questions immediately arise from the prospect of modifying administrative law along these lines. First, how would judicial review under such an approach differ from what we would expect under current doctrine? Second, how would courts evaluate whether the agency’s internal administrative law was satisfactory? In considering whether an action by a banking agency that rested directly or indirectly on supervisory activity was arbitrary or capricious, a court would focus less on the specific 293 See Schreiber, 381 U.S. at 290 (footnote omitted) (quoting FCC v. Pottsville Broad. Co., 309 U.S. 134, 138 (1940)). 384 COLUMBIA BUSINESS LAW REVIEW [Vol. 2022 action being challenged and more on the process that produced it. So, in the example discussed in Part III, courts would not try to determine whether a banking agency’s supervisory guidance on the use of AI models crossed some elusive line into the realm of “practically binding.” Supervisory communications are supposed to have influence over bank practices, though they would need to be supplemented with additional formal steps before they become legally binding and enforceable. In considering the consequences of ignoring specific supervisory guidance or an MRA on the overall supervisory relationship, a bank’s executives will usually conclude they need to be at least somewhat responsive. A court would instead examine whether governance mechanisms for supervision established and maintained within the agency have appropriately channeled the supervisory action within the iterative supervisory relationship. Courts would generally defer to supervisory actions where appropriate internal procedures governing that relationship were in place. Specific actions that, when considered in isolation, could be argued to run afoul of at least some interpretations of administrative law procedural requirements might thus be upheld.294 Conversely, a court would intrude more into the process and substance of significant supervisory action where the internal administrative processes were found wanting. It is thus possible that judicial deference to specific actions taken by banking agencies—I have ratings decisions particularly in mind—would depend on their emanating from a more developed internal process than is required by current administrative law doctrines. How courts would evaluate the sufficiency of an agency’s internal administrative law is obviously a delicate issue. One 294 The proposal advanced here is addressed to supervisory actions as such—ratings, MRAs, guidance, and the like. It does not cover judicial review of other banking agency actions, such as decisions on applications for mergers. As noted earlier, supervisory information may be an input into agency consideration of one or more statutory factors relevant to these decisions. But there seems no reason to modify the current approach to judicial review of these actions. No. 1:279] BANK SUPERVISION AND ADMINISTRATIVE LAW 385 can readily imagine judicial scrutiny moving closer to de facto judicial imposition of a court’s own preferred set of procedures—an outcome that would violate the premise on which my proposal rests and, possibly, Vermont Yankee. However, similar possibilities exist in any doctrinal context in which the judiciary says it has established a deferential standard of review of agency action—whether procedural, substantive, or interpretive. As in those other contexts, fidelity to the stated standard will be determined by some combination of judicial self-discipline and the attention paid by higher courts when lower courts drift away from it. The specification and elaboration of relevant norms for assessing internal processes would need to evolve over time. As a starting point, let me suggest five. First, internal administrative law systems would ensure that agency principals—that is, the presidentially appointed and Senate confirmed members of the boards of the Federal Reserve and the FDIC, as well as the Comptroller of the Currency—set the overall structure and direction of the supervisory function. As explained by then-Professors Barron and Kagan in the somewhat different context of Chevron deference, requiring involvement of the statutory recipients of the authority delegated by Congress promotes the aims of agency accountability and discipline that lie at the center of much administrative law.295 The policy discretion granted by Congress should be actively exercised by those to whom it has 295 David J. Barron & Elena Kagan, Chevron’s Nondelegation Doctrine, 2001 SUP. CT. REV. 201, 204 (2001). This article, written just after United States v. Mead Corp., 533 U.S. 218 (2001), presciently argued that the test for application of Chevron deference laid out there would prove unworkable. Id. at 212–25. In an introductory comment of particular relevance to banking supervision, Barron and Kagan observed that the Mead Court’s “preference for formality in administration, even in cases when not statutorily required, fails to acknowledge the costs associated with the procedures specified in the APA, which only have increased in significance since that statute’s enactment.” Id. at 203–04; see also Gillian B. Metzger, The Constitutional Duty to Supervise, 124 YALE L. J. 1836, 1846 (2015) (“Systemic administration and internal oversight are becoming increasingly central mechanisms for ensuring accountability in government operations and programs.”). 386 COLUMBIA BUSINESS LAW REVIEW [Vol. 2022 been granted. Agency principals not only have that statutory responsibility. Sitting atop their agencies, they are also best positioned to prioritize policy and resource demands. In the context of bank supervision, a reasonable application of this norm might entail two elements. One is that agency principals would themselves consider and approve the framework for formulating and applying supervisory policies—that is, the internal administrative law that would at least presumptively validate the resulting specific supervisory actions. The other is that principals should pass on the main elements of their agencies’ supervisory programs such as priority areas for supervisory emphasis, significant horizontal reviews,296 and major pieces of written guidance. The organizational means by which principals fulfil these responsibilities would be left to each banking agency. There may be variation on the engagement of agency principals. For example, the Vice Chair for Supervision at the Federal Reserve would presumably assume a disproportionate share of Board oversight, as at present, though there should be regular involvement of the full Board, which is the specified delegee of statutory authority for prudential regulation.297 296 In supervisory parlance, a “horizontal review” is an examination of a specific issue across one or more categories of banks, such as commercial real estate underwriting and exposures. Depending on their scope and whether they involve more than one supervisory portfolio, a horizontal review can occupy substantial supervisory resources. The Shared National Credit Review Program is a longstanding interagency review of large and complex credits across a range of institutions. The banking agencies annually publish a joint report of their overall findings. See Shared National Credit Program, BD. OF GOVERNORS OF THE FED. RSRV. SYS. (Mar. 9, 2017), https://www.federalreserve.gov/supervisionreg/snc.htm [https://perma.cc/RQU4-VLPD] (providing annual reports). Oral and/or written communication of specific supervisory findings are regularly communicated to the banks. 297 Unlike the Board of Governors, at which executive and managerial responsibilities have traditionally been allocated among Board members, the Chair of the FDIC has generally reserved to himself or herself nearly all such responsibilities. So there the Chair would likely individually conduct much of the oversight but, again, there should be regular opportunities for No. 1:279] BANK SUPERVISION AND ADMINISTRATIVE LAW 387 It is important, though, that a court not slip into the implicit expectation that principals will have in some fashion directly approved all consequential supervisory actions.298 The same banking characteristics of opaque balance sheets, heterogenous activities and often fast-changing risk profiles that explain the utility of the supervisory function also explain why line supervisors must regularly exercise their own judgment in assessing bank practices. The point of mandating a meaningful role for principals is to ensure that they have set the overall policies for supervisory action and a framework for its consistency with those policies, not to retract necessary discretion from line supervisors.299 Moreover, many second order issues common to one or more the entire FDIC Board to review the supervisory program. At the single- principal OCC there is obviously no sharing of responsibility. 298 Cf. Kisor v. Wilkie, 139 S.Ct. 2400, 2416 (2019) Justice Kagan wrote for the Court, in the context of authoritative agency interpretations when an agency interprets its own regulations: Of course, the requirement of “authoritative” action must recognize a reality of bureaucratic life: Not everything the agency does comes from, or is even in the name of, the Secretary or his chief advisers. So, for example, we have deferred to “official staff memoranda” that were “published in the Federal Register,” even though never approved by the agency head. . . . But there are limits. The interpretation must at the least emanate from those actors, using those vehicles, understood to make authoritative policy in the relevant context. Id. (quoting Ford Motor Credit v. Milhollin, 444 U.S. 555, 566, n.9, 567, n. 10 (1980)). 299 Issues of delegation to line supervisors are not always best understood as implicating straightforward agency cost problems, though some certainly are. In thinking about appropriate delegations to supervisors, one cannot rely exclusively, or perhaps even principally, on whether the principal can ex ante anticipate that line officials will have similar policy predispositions to his or her own. The views of principals will frequently take shape only through briefings and other discussions with staff, since the issues are often complex and may vary from bank to bank. Hence the very interesting approach to internal agency delegations taken by Professor Nou may not be fully applicable in the supervisory context, though her point that courts need to better understand internal agency dynamics surely is. See Nou, supra note 276, at 511–25. 388 COLUMBIA BUSINESS LAW REVIEW [Vol. 2022 supervisory portfolios could be decided by senior career officials within the general directions set by principals. The next three norms pertain to features that would be required of each agency’s organizational framework for supervision, though again the particulars would be up to agency principals. The second norm is that the framework for disseminating supervisory policies should contain proactive measures for monitoring effectiveness and consistency. Each agency should have its own governance mechanisms for ensuring that the exercise of supervisory discretion is neither arbitrary nor demonstrably at odds with overall supervisory policies. It would be especially important for courts not to impose their own views on this element of the supervisory framework, since it will be most effective if it is integrated into the varying organizational structures in the three agencies. The third norm is that the agencies be reasonably transparent in their supervisory activities. While the agencies should have considerable discretion in determining what “reasonable” transparency is, there are at least two features that should be reflected in an agency’s framework. One is that supervisory communications—whether broadly applicable guidance or more bank-specific messages—should generally be in written form. As has usually been the case, broadly applicable agency guidance should be published. The other transparency feature is that, to the degree an agency has generated internal guidelines for interpreting regulations or evaluating applications for mergers and other actions, they should presumptively be available to the public.300 Although consideration of these applications takes place under required statutory procedures, and outcomes are legally binding in the 300 The Federal Reserve is already taking steps along these lines. In its 2020 changes in its regulation for determining when a company has “control” over a bank for purposes of the Bank Holding Company Act, the Board of Governors has provided substantially more information on the types of relationships that the Board will consider to reflect control. Control and Divestiture Proceedings, 85 Fed. Reg. 12,398, 12,401–02 (Mar. 2, 2020) (to be codified at 12 C.F.R. pts. 225, 238). Vice Chair Quarles has further indicated that the Federal Reserve may create a word-searchable database of regulatory interpretations by Board staff. See Quarles, supra note 214. No. 1:279] BANK SUPERVISION AND ADMINISTRATIVE LAW 389 traditional sense, prior and contemporaneous supervisory evaluations of the banks involved are often important factors in these determinations. In the context of their larger argument that would bring supervision much closer to more formal administrative adjudication, some representatives of banking interests have complained about the confidentiality of a substantial amount of supervisory information and some supervisory policies.301 While my suggestion for a norm of transparency would likely produce some of the same results as they advocate,302 the justification for this norm is nearly the opposite. It is precisely because formal administrative processes are but a small part of the supervisory relationship that more transparency about supervisory policies is needed. The fourth norm is that the supervisory framework provides for meaningful internal review of supervisory actions. As suggested in the discussion of ratings in Part III, this is an area where a court might reasonably expect more than current administrative law doctrine would find procedurally necessary. While the fact of an iterative supervisory relationship argues for moving away from doctrines such as the “practically binding” test for requiring notice-and-comment rulemaking, it also argues for augmenting procedures where needed to make an option for internal review realistically available, given the nature of that relationship. The nature and formality of the review for supervisory actions would obviously vary with the practicalities and 301 See, e.g., Tahyar Testimony, supra note 9, at 41, 44. 302 Some, though by no means all. For example, Ms. Tahyar suggests that banks themselves should have the option of deciding whether to make their supervisory ratings public. Id. at 45. That kind of transparency would have no direct effect on the supervisory relationship itself, but would allow banks to leverage good ratings with investors or the public. There are certainly strong arguments that banking agencies should release more information about their ratings on an aggregate basis, and some plausible (though weaker) arguments that individual bank ratings should be released publicly. But these issues are more about the desirability of public monitoring of banking agency practice. 390 COLUMBIA BUSINESS LAW REVIEW [Vol. 2022 significance of those actions.303 It seems especially desirable that agencies provide a mechanism for review of supervisory ratings decisions beyond the skeletal statutory requirements. The annual ratings process both summarizes prior supervisory communications and sets a path for future supervisory activity. It is already more formally structured within the banking agencies than much supervisory work and, as such, the object of considerable internal administrative law.304 Especially for smaller banks, but also to some degree for large ones, it is the pressure point of the supervisory relationship. For all these reasons, the ratings process is an appropriate focus for both agency principals and, potentially, courts in monitoring the supervisory process. For senior agency officials, a well-structured review system could provide insight into supervisory laxity, as well as overreach. 303 In addition to the various well-established modes of supervisory communication such as guidance or MRAs, supervisors may less formally— sometimes in writing, but often orally—communicate expectations for an individual’s bank practices. These communications may reflect the supervisors’ understanding of how published guidance or regulations apply to the circumstances of a specific bank or it may relate solely to supervisors’ concerns about a bank’s condition or practices that are not specifically addressed in regulations or published guidance. For larger banks with dedicated supervisory teams, the frequent interactions between bank personnel and line supervisors mean that many such communications occur. Indeed, they lie at the heart of the iterative supervisory relationship. However, they also can be the means by which line supervisors deviated from a reasonably well-articulated agency supervisory policy. The second norm of proactive oversight of the supervisory process should limit the number of what is in any case a relatively small universe of cases relative to the volume of supervisory communication. However, a norm of effective review of supervisory decisions would argue for some mechanism for recourse in the unusual, but potentially significant, case in which a bank believes it is receiving supervisory direction that may not reflect agency policy. 304 There have been modest efforts in this direction already. For example, the FDIC has recently proposed replacing the FDIC staff from the Office of Supervisory appeals with outside experts. Press Release, Fed. Deposit Ins. Corp., FDIC Proposes Changes to its Supervisory Appeals Process (Aug. 21, 2020), https://www.fdic.gov/news/press- releases/2020/pr20093.html [https://perma.cc/CE35-DV7J]. No. 1:279] BANK SUPERVISION AND ADMINISTRATIVE LAW 391 There are many ways in which agencies might afford effective review of ratings decisions. One obvious route would be to fill out the current internal review processes to ensure that they were not pro forma, or otherwise structured to predictably affirm the initial determination.305 But an agency might also opt for something more interactive. For example, an agency might introduce into its initial process for assigning ratings an opportunity for comment by the bank. That is, before a recommended set of ratings from the relevant supervisory team was ratified or modified through the internal review process, the bank could add information or perspective. Ex ante review might be more effective than ex post review, since the bank’s arguments would be combined with other questions and discussion engendered by the internal process, and thereby not require a “reversal” of a ratings decision.306 Unlike the first four norms, the fifth centers on the substance of supervisory activity, rather than the processes through which it is executed: The supple dynamic of the supervisory relationship should be used only in fulfillment of the function I have suggested it serves within the regulatory system established by Congress. As described in Part II, this function is essential to complement the framework of statutes, regulatory rules, and adjudicatory processes that defines that 305 Since 1994 each banking agency is required by statute to create the position of an ombudsman, whose role is to intermediate between any party—obviously importantly including regulated banks—and the agency “with respect to any problem such party may have in dealing with the agency resulting from the regulatory activities of the agency.” Riegle Community Development and Regulatory Improvement Act of 1994, Pub. L. 103-325, § 309, 108 Stat. 2160, 2219 (codified at 12 U.S.C. § 4806). While an ombudsman may serve a useful purpose with respect to some issues, the position has not proven helpful in what might be termed routine matters of ratings decisions or consequent supervisory determinations of other sorts. As suggested in the text, an effective internal mechanism will almost surely require the expertise and involvement of those involved in the supervisory function itself, but with sufficient organizational distance from the original decision-makers to ensure an independent perspective. 306 Of course, even if such an alternative approach were adopted, each agency would need to maintain a formal, independent ex post review process to satisfy statutory requirements. 392 COLUMBIA BUSINESS LAW REVIEW [Vol. 2022 system. Because of the heterogeneity, opaqueness, and sometimes fast-changing activities of banks, supervision is more effective and efficient in achieving an optimal level of prudential constraint than a mix of very blunt and very detailed rules would be. The multitude of potentially unsafe and unsound banking practices means that the supervisory function may sometimes be fairly broad-ranging. But it is not without limit. So, for example, it would be inconsistent with the supervisory function to use guidance in place of notice-and- comment rulemaking as a way to introduce a new form of prudential regulation, such as quantitative liquidity ratio requirements. However, as the hypothetical on risk management of AI models in Section III.B. showed, guidance on bank practices posing safety and soundness concerns will often be entirely appropriate even where there is no regulatory rule governing the practices in question. The issue, then, will be whether the agency has continued to use guidance, even after it has formulated more specific requirements such as quantitative liquidity ratio requirements. Another example would be the use of supervisory guidance either to restrain banks from, or require them to continue, lending to certain industries. Readers may recall that following the mass shooting at a Parkland, Florida school in early 2018, Citigroup announced that it would cease doing business with retail sector customers or partners who did not observe restrictions on sales of guns to minors or those who had not passed background checks.307 Citi’s decision on social responsibility grounds not to lend to gun manufacturers was criticized by gun rights groups308 and at least two United 307 Ed Skyler, Announcing Our U.S. Commercial Firearms Policy, CITIGROUP (Mar. 22, 2018 11:30 AM), https://blog.citigroup.com/2018/03/announcing-our-us-commercial- firearms-policy (on file with the Columbia Business Law Review). 308 Katanga Johnson, U.S. Gun Lobby Takes Aim at ‘Gun-Hating’ Banks Citi, BofA, REUTERS (May 18, 2018, 12:09 PM), www.reuters.com/article/us-usa-guns-banks/u-s-gun-lobby-takes-aim-at- No. 1:279] BANK SUPERVISION AND ADMINISTRATIVE LAW 393 States Senators.309 Suppose banking agencies wished to be responsive to these political pressures by issuing guidance instructing banks not to discriminate against otherwise lawful industries in their lending decisions. The justifications for this guidance would probably be that the bank practice was unsafe or unsound because it exposed the bank to “reputational risk” and that it ran afoul of the OCC’s mandate to ensure “fair access to financial services.”310 While reputational risk can sometimes be quite significant if the result of the controversial bank practice is a substantial loss of customer or counterparty business, it can also be used as a pretext for extending banking agency policies into areas only tenuously related to safety and soundness. Indeed, since many bank practices unrelated to lending or other financial gun-hating-banks-citi-bofa-idUSKCN1IJ260 [https://perma.cc/W8CT- 9JSM]. 309 Alan Rappeport, Banks Tried to Curb Gun Sales. Now Republicans Are Trying to Stop Them, N.Y. TIMES (May 25, 2018), https://www.nytimes.com/2018/05/25/us/politics/banks-gun-sales- republicans.html [https://perma.cc/C3L8-9CM2]. 310 12 U.S.C. § 1(a) (2018). In a similar situation involving the reluctance of banks to lend to companies engaged in oil drilling in the Arctic—presumably at least partially because of pressures from environmental groups—the Acting Comptroller of the OCC responded favorably to the suggestion by a senator from Alaska that the refusal of the banks might constitute a violation of this section. Letter from Brian Brooks, Acting Comptroller of the Currency, to Dan Sullivan, Sen., (July 24, 2020), https://bankingjournal.aba.com/wp-content/uploads/2020/07/brian-brooks- occ-letter-sen-dan-sullivan.pdf [https://perma.cc/R2BL-J9VK]. A look at that first section of the National Bank Act suggests this is a tenuous argument at best: There is established in the Department of the Treasury a bureau to be known as the “Office of the Comptroller of the Currency” which is charged with assuring the safety and soundness of, and compliance with laws and regulations, fair access to financial services, and fair treatment of customers by, the institutions and other persons subject to its jurisdiction. 12 U.S.C. § 1 (2018) (emphasis added). Bootstrapping a general charge of the OCC into an obligation by individual banks to lend to all comers seems more than a bit of a stretch. 394 COLUMBIA BUSINESS LAW REVIEW [Vol. 2022 activities—from its philanthropic decisions to its human resources policies—can elicit public criticism from some interest groups and elected officials, virtually anything can be denominated a reputational risk. And it would be a considerable stretch to bootstrap a general statutory charge that the Comptroller ensure fair access to financial services into supervisory guidance pushing banks to lend to specific borrowers or industries. While there may sometimes be good reason for a banking agency to act in these kinds of circumstances, it should not be able to exert influence through the iterative supervisory relationship to avoid justifying the agency’s authority to act and taking comments on the merits of its actions.311 Just as the line between nascent safety and soundness concerns and specific regulatory requirements may in some cases be blurry, so the line between prudential concerns and non-prudential areas may sometimes not be clear. There will of course be close cases. But these two distinctions are a good starting point for delimiting the supervisory function. And, of course, once a regulatory rule is in place, supervisory evaluation of compliance would be wholly proper, as is the case today with anti-money laundering and consumer protection regulations. 311 Whatever the limits of OCC’s legal authority to require banks to lend to specific borrowers or classes of borrowers, it is noteworthy in the current context that, in response to the controversy around bank lending to oil companies, the Acting Comptroller began a notice-and-comment rulemaking, albeit an eleventh-hour one following President Trump’s defeat in the November 2020 election. See Fair Access to Financial Services (Fair Access Rule), 85 Fed. Reg. 75,261, 75,264 (proposed Nov. 24, 2020) (to be codified at 12 C.F.R. pt. 55). Following the resignation of the Acting Comptroller in January 2021, the agency paused publication of the rule in the Federal Register to “allow the next confirmed Comptroller of the Currency to review the final rule and the public comments the OCC received, as part of an orderly transition.” News Release, Off. of the Comptroller of the Currency, OCC Puts Hold on Fair Access Rule (Jan. 28, 2021), https://www2.occ.gov/news-issuances/news-releases/2021/nr-occ- 2021-14.html [https://perma.cc/5XBW-R9J9]. No. 1:279] BANK SUPERVISION AND ADMINISTRATIVE LAW 395 4. Some Qualifications. The approach just suggested would require a significant change in contemporary administrative law, one that would likely be more consequential than adjusting review of administrative actions to take due account of relevant substantive statutory provisions. There are certainly arguments that could be offered against this change. Some rest on a preference for trans-substantive administrative law. Others could be directed at the specifics of the alternative approach. As noted earlier, some banking interests advocate strong versions of current administrative law doctrines so as to severely constrain the supervisory function. Much of my earlier analysis has been directed to rebutting these arguments, both doctrinally and normatively. However, there are also trans-substantivity arguments that do not begin from self-interested anti-regulatory premises. Indeed, as reflected in the earlier-noted views on guidance of some academics, one version of trans-substantivity could cut back dramatically on judicial review of supervisory action. A proponent of continued adherence to trans-substantivity might argue that it is a more economical approach to administrative law. While agencies and courts may reasonably contemplate some variation in the details of procedural requirements, the preservation of common doctrinal starting points will relieve courts of having to master the organization and procedures of dozens of agencies in order to determine the appropriate scope for more agency- specific processes. Further, it arguably gives more direction to agencies themselves, since they can draw on precedents decided in other regulatory contexts while conducting their business. Banking agencies developing their own procedures would need to await judicial review of their own actions before knowing if they are valid. This kind of trans-substantive position is not without force, especially if the alternative would actually be the development of a separate administrative law for each agency. However, as I have already argued, the unusual characteristics of banking supervision—from its historical 396 COLUMBIA BUSINESS LAW REVIEW [Vol. 2022 roots to its association with the regulation of the sovereign function of money creation to its strong statutory foundations—make a strong case for an adjustment of trans- substantive doctrine. As a former regulator, I am especially sensitive to doctrines that would unduly hamper the supervisory function. But, as indicated in my discussion of ratings, the implications of the unique character of bank supervision should be taken into account in both directions. To emphasize once again the core of my argument—even though supervisory actions are not legally binding, they are supposed to influence bank conduct, and generally do. The norms of consistency, fairness, and accountability that inform the APA and judicially created administrative law are equally relevant to bank supervision as to other areas, but they need to be realized in a way that takes account of that unique character.312 There are two significant arguments that might be directed at the specifics of the approach proposed here: transition problems and the potential unwieldiness of judicial review of an agency’s internal administrative governance. First, it may not be easy to exit the highway of evolving administrative law onto the less travelled alternate route I discuss here, even if courts were potentially receptive to the idea. No court is going to offer sua sponte a kind of advisory opinion that it would adapt its approach to APA review of supervisory actions were the banking agencies to reorganize themselves along the lines I have described. From the agencies’ perspective, even with leadership more sympathetic to supervision, there may be little legal impetus to change practices. Should significant legal challenges of the sort envisioned by some banking interests be filed and meet with some success, the agencies would probably scramble to change 312 Although this Article does not address the problems of supervisory under enforcement, it is worth noting that a framework such as that proposed here would at least incrementally enhance the ability of agency principals, congressional oversight committees, and the public to monitor, and perhaps correct, supervisory laxity of the sort that contributed to the global financial crisis. No. 1:279] BANK SUPERVISION AND ADMINISTRATIVE LAW 397 their practices so as not to run afoul of current, generally applicable administrative law doctrines. We can conjure up scenarios that might be more favorable. One is that a banking agency leadership committed to the supervisory function would make changes in this direction on its own, with the aims of achieving more efficiency, consistency, and fairness. Another, perhaps less likely, is that a series of bank-friendly decisions in administrative law cases that undermined the core of supervision would be met with blunter regulations forthcoming from agencies headed by officials committed to rigorous regulation. In that event, both banks and regulators might have an interest in courts taking a different approach to supervision. A third would be action by Congress to validate such an approach, though prospects for any banking legislation that does not become highly divisive are at present not good. Second, assuming a viable means of transitioning to a form of judicial review that evaluates an agency’s overall governance of the supervisory function when a supervisory action is challenged, some potentially non-trivial issues could arise. An obvious one is that an evaluation of the procedures and safeguards around the entire supervisory function could be a burdensome undertaking. It might not be necessary to look at all aspects of that function in order to conclude, for example, that a guidance document and subsequent supervisory communications that contributed to a ratings downgrade had been developed consistently with the kinds of norms explained earlier. Still, in at least some cases the inquiry might prove quite open-ended. If so, at least in the early going, the investment of judicial resources could be considerable. Another potential problem is that different courts could reach different conclusions on the key question of whether the governance process that produced the supervisory action at issue in the case was in fact consistent with applicable norms. In that event, the banking agencies might be presented with unhelpfully mixed messages as to what refinements, if any, were needed in its procedures. Of course, even if courts did have different views of the agency’s supervisory governance, 398 COLUMBIA BUSINESS LAW REVIEW [Vol. 2022 the immediate practical effect would be limited to whether a court would conduct a more (or less) stringent review of the supervisory actions as part of an arbitrary and capricious review. And differences of views among the circuits may not be materially more troublesome for agencies (and regulated entities) than they are today with respect to other doctrinal questions. Still, it is likely that the very novelty of this approach would create some problems, unanticipated as well as anticipated. It remains to be seen how realistic this or other routes may be for reaching a different administrative law framework for bank supervision. Governance mechanisms of the sort discussed here would be better policy than what we have now. For that reason alone, they are worth discussing and implementing. If they were to be implemented just because they are better policy, there would be at least some opening for courts to assess differently the supervisory actions emanating from them. Were the courts to do so, the agencies would have external reinforcement for fostering the integrity of those processes. At the least, this outline of an alternative that could better and more efficiently achieve the goals of the APA provides a starting point for considering how to achieve greater congruence of administrative law doctrine, banking law, and the realities of agency supervision. V. CONCLUSION The subject of this Article has been the relationship of administrative law to banking supervision, more specifically, the degree to which administrative law constrains supervisory oversight of banking organizations. Maintaining a focus on this already broad topic has meant that important questions suggested by this analysis have been addressed only in passing: What are the best supervisory policies within the constraints of administrative law? To what degree would the reorientation of administrative law suggested in Part IV address the problem of supervisory under-reach? How relevant is the case of banking supervision for the application of administrative law doctrine to other areas? In concluding, I offer a few thoughts on each of these topics. No. 1:279] BANK SUPERVISION AND ADMINISTRATIVE LAW 399 As to the first question, an obvious but still useful point is that not every supervisory practice that could be fairly criticized as unnecessary or ill-considered is thereby inconsistent with administrative law. I have no doubt, for example, that in the wake of the Financial Crisis, some supervisors overused MRAs. As a result, supervisory priorities may have been obscured for both the banks receiving bundles of MRAs and for the supervisors themselves. I suspect that with more attention to internal governance mechanisms such as those discussed in Part IV, senior agency officials would have more quickly rationalized MRA practices, whatever their specific policy orientation. But, once the legitimacy under administrative law of the MRA as a supervisory tool is understood, the extent to which that tool is used should generally be a policy issue, not a legal one. A more interesting question is what, exactly, the banking interests arguing for defanging supervision hope to gain. My suggestion that success of this effort could result in a sub- optimal outcome of less efficient regulatory rules would imply that the strategy could ultimately be self-defeating. If banking agencies are deprived of important supervisory tools and respond with some combination of blunter and inflexibly detailed regulatory rules, the result could be a higher regulatory cost for many forms of intermediation. Of course, banking interests might be seeking incapacitation of robust supervision in the expectation that the bank regulatory agencies will not dynamically respond by strengthening regulatory rules. This may have been a reasonable assumption when the leadership of federal banking agencies had a deregulatory bent. Indeed, the agency leaders appointed by President Trump voluntarily made many, though not all, of the changes proposed by these interests. With capital and other regulatory rules having been relaxed rather than tightened, the changes to supervision looked to be another form of deregulation. Following President Biden’s election, agency leadership will likely be replaced by officials with a more rigorous regulatory orientation as their terms expire. If these new officials were then deprived of a strong supervisory tool 400 COLUMBIA BUSINESS LAW REVIEW [Vol. 2022 because of litigation, their recourse will be the more stringent regulatory rules. Accordingly, unless these same banking interests are counting on the courts applying a resurrected non-delegation doctrine to invalidate Congressional grants of prudential authority to the banking agencies, successful administrative law attacks on supervisory discretion could ultimately prove a pyrrhic victory. As to how much impact the proposals in Part IV would have on the problem of supervisory underreach, the answer is probably only at the margin. While current administrative law doctrines may be an obstacle to realizing a sensible approach to procedural requirements and substantive review of supervisory action, they are closer to an impregnable wall confronting any effort to counter agency inaction. This issue runs much deeper than bank supervision, implicating as it does generally applicable standing doctrine and traditional judicial deference to agency decisions not to take enforcement actions. Still, internal administrative law of the sort suggested in Part IV could mitigate somewhat the problem of supervisory underreach, especially as it pertains to line supervisors. Expectations of principal involvement in setting supervisory directions and priorities, proactive monitoring systems, increased transparency, and well-conceived processes for reviewing supervisory decisions should all help make supervisory practice more consistent across banks and closer to the policies of agency principals. However, when those principals are themselves inclined to light touch supervision, even the best internal governance mechanisms will not prevent supervisory laxity. Finally, as to the relevance of my analysis to other regulatory areas, one implication seems reasonably clear: In determining procedures that the APA requires of an agency fulfilling its regulatory or programmatic charge, courts should be attentive to the statutory provisions through which Congress has delegated these responsibilities. Courts universally recognize specific procedural provisions in enabling legislation as either supplementing or superseding APA requirements. But, as the doctrinal analyses in Part III No. 1:279] BANK SUPERVISION AND ADMINISTRATIVE LAW 401 reveal, substantive features of regulatory legislation are also relevant. The question, of course, is how relevant. The answer will presumably vary both with the specifics of the substantive legislation and with the priors of the judge or commentator parsing the question. Banking supervision is clearly an unusual administrative function, defined as it is by its explicit statutory grounding, its role in overseeing the money-creating function of banks, and its importance in guarding against financial crises that inflict long-lasting harm on the economy. But is it essentially an outlier, or just toward the end of a spectrum of agency responsibilities created by Congress that contemplates or requires relationships between agency officials and regulated entities? One reading of Professor Parrillo’s evaluation of the use of guidance across agencies suggests that, at least as the practice has evolved over the decades, the latter is the case. But the question bears considerable further discussion. Everyone interested in administrative law presumably believes both that some measure of trans-substantivity is desirable and that more tailored applications of APA provisions are sometimes warranted. If administrative law scholarship is any indication, there is a broad range of positions on the appropriate, if not required, mix of the general and the particular. There may be considerable room for debate on how much weight to give substantive statutory provisions in applying administrative law doctrines in other areas, much less whether those areas warrant exploration of alternative approaches grounded in internal administrative law.