I am unable to support this revamping of the rule implementing the Community Reinvestment Act (the “CRA”).
First, the agencies have not made a case for the rule’s unspoken premise that a significant portion of banks are not doing enough to meet the credit needs of their entire community.1 That premise is made clear by the design of the retail lending test. Because banks are graded on a forced curve against the market benchmarks, a bank that meets the market in a low- or moderate-income market segment—i.e., a bank that performed as well as all lenders did on average in that market segment—will actually be deemed “low satisfactory” under the applicable market benchmark.2 Over the 2018-2020 evaluation period, around 90 percent of banks’ facility-based assessment areas would have received a “needs to improve” or “substantial noncompliance” under the community benchmarks for mortgages to low-income borrowers, and almost 80 percent of banks’ facility-based assessment areas would have received a “needs to improve” or “substantial noncompliance” under the community benchmarks for small farm loans to low-income geographies.3 As a result of the stringent calibration of the community and market benchmarks, the agencies estimate that almost half of banks would have received a bank-level “low satisfactory” under the retail lending test over the 2018-2020 evaluation period.4
This premise is simply a significant change in policy from 45 years of practice since the CRA’s enactment and should have received at least some discussion. More generally, we should have endeavored to offer a more explicit calibration framework that rationalizes the design decisions adopted in the rule. While the CRA itself is not especially prescriptive as to how the agencies should assess each bank, we should base our decisions on more than a “gut sense” as to the right outcome.
Second, I cannot support a rule without some confidence that its benefits will exceed its costs, but it is impossible to hazard even a rough guess as to this rule’s likely effects. The approximately 60,000 words of rule text (including appendices), which contains more than 40 benchmarks and 20 metrics, are enough to preclude anyone from comprehending the rule as a whole.5 More problematically, big chunks of the rule remain unfinished works in progress, leaving big questions unanswered. The community development financing test (40 percent of a rating), the retail services test (10 percent), and the community development services test (10 percent) amount collectively to at least 17 benchmarks and 7 metrics—but without any methodology at all for converting all those many inputs into a conclusion. So that tells us what counts, but nothing about how it counts.6
Third, while the retail lending test does offer more transparency, the test introduces conflicting incentives that could have significant unintended consequences. For example, due to cliff effects relating to retail lending assessment area and major product line triggers, a bank will have two options, and a mix of conflicting incentives, with respect to a locality or product line in which its performance is weak from a CRA perspective. The bank could invest more resources to improve its CRA performance or, conversely, scale back its activities to avoid triggering a retail lending assessment area or major product line. It is hard to imagine that banks always, or even typically, will choose to invest more resources, particularly where bank capital requirements or other factors give nonbanks a competitive advantage over banks in lending to low- and moderate-income communities.7
Fourth, I have yet to be convinced that the regulators have statutory authority to prescribe important aspects of the rule. The CRA requires each agency to assess a bank’s record of “meeting the credit needs of its entire community.”8 I have not seen a convincing argument that we have the authority to consider lending activities outside a bank’s facility-based assessment areas. Similarly, our arguments in support of the consideration of deposit products, free-checking accounts, and other non-credit products are not persuasive to me.
Given the significant changes in lending since the CRA’s enactment, there could be merit to modernizing the CRA framework to consider lending activities outside facility-based assessment areas or non-credit products. However, the CRA was enacted at a time of facility-based banking, and legislation would appear to be necessary to achieve that modernization.
Finally, several more specific aspects of the rule also concern me:
- Large bank threshold. Just reading—let alone complying with—this 60,000 word rule will be a very expensive and time-consuming effort. That compliance effort will be a particular problem for our community banks, and so in my view, the $2 billion asset threshold for “large bank” compliance with substantially all of the rule’s requirements has been set too low.
- “Beat the market” risks. Given the potential public stigma associated with a “low satisfactory” under the retail lending test, banks generally are likely to push hard to beat the market so as to achieve a “high satisfactory” under the market benchmarks. That will particularly be the case where the community benchmarks are impractical. Efforts to beat the market will necessarily shift the market, driving a continuing evolution in pricing and underwriting as banks compete against each other for low- and moderate-income market share. Over time that could result in looser underwriting standards and potentially even risks to safety and soundness and consumer protection.9
- Tailored multipliers. The rule applies the same multipliers to set the community performance ranges across product lines and income segments. There are of course meaningful differences across product lines in demand by low-income consumers. There likely are also meaningful differences in nonbanks’ market presence in each product line. These differences tend to weigh in favor of more tailored community multipliers.
- Loan purchases. Section __. 22(g)(1) allows the agencies to adjust a recommended conclusion if a bank purchased loans to “inappropriately” enhance its performance. Staff has assured me that the intent of this provision is to take into account loan churning, which is an understandable objective.10 The rule text, however, does not equate “inappropriately enhancing” with loan churning only; in fact, the rule text defines churning to be but one non-exhaustive example of inappropriate enhancement. So there remains some question about the extent to which banks may rely on loan purchases (without churning) to enhance CRA performance.
- Fairness concerns. Banks would not know some of the retail lending test’s benchmarks in advance. Besides undermining their ability to plan and allocate resources, this seems to raise basic fairness concerns.
1 While perhaps unspoken in the final rule, agency leadership actually has been quite explicit about this premise in other contexts. See John Heltman, OCC’s Hsu: ‘Lending has to go up’ under new CRA rule, Am. Banker (October 20, 2023, 11:31 a.m. EDT) (Speaking about this forthcoming final rule, Acting Comptroller Hsu said: “[A]t a high level, there’s got to be more, it has to be better and it’s got to be faster. In very simple terms, that’s it: The amount of CRA investments and lending has to go up.”); Urban Institute, Modernizing the CRA: Ensuring Banks Meet the Credit Needs of Their Communities, YouTube (June 6, 2022) (Speaking about the proposal, Chairman Gruenberg said: “Fundamentally, what this rule is going to do is broaden the reach to capture more lending activity than is being captured now for CRA evaluation and raise the bar for performance, meaning what a bank did before to earn an outstanding or high satisfactory rating will not be sufficient. They’re going to have to engage in more lending activity to earn the recognition. So . . . from a broad perspective, we’re going to broaden the reach and raise the bar with the goal of more lending to more low- and moderate-income communities. I think fundamentally that’s what this is [trying to do].”).
2 See Final Rule, App. A, paragraphs V.b.2 & V.d.2. In the preamble, the agencies explain they “determined that the ‘High Satisfactory’ market multiplier should result in a calibrated market benchmark that is at least slightly above the market benchmark, rather than equal to the market benchmark. In making this determination, the agencies decided that in an area where the performance ranges are based on the market benchmark, bank performance that is exactly equal to the market average, or only marginally above the market average, should correspond to a ‘Low Satisfactory.’” Final Rule at 611.
3 These estimates are based on loan distributions in facility-based assessment areas for banks reporting both HMDA and CRA loans.
4 Based on a set of intermediate and large banks that are both CRA and HMDA reporters, the agencies estimate that more than 44 percent of those banks would have had a retail lending test conclusion of “low satisfactory” or worse during the 2018-2020 evaluation period. Final Rule, Table 33 to § __.22, at 654.
5 Efforts to improve transparency and predictability—which are laudable goals of both the proposal and this final rule—can lead to more prescriptiveness and thus complexity. But the agencies could have taken steps to achieve these goals while reducing complexity, such as by using safe harbors or presumptions to foster simplicity.
6 The agencies note that data limitations precluded a methodology for assigning a community development financing test rating. The agencies do suggest an openness to revisiting the test after collecting data. Final Rule at 734 (“While the agencies also believe that consistency could be improved using thresholds in the Community Development Financing Test, current data limitations preclude the agencies’ ability to explore including thresholds in the test at this time. The agencies note that they could consider thresholds in a future rulemaking once they have accumulated data and have experience applying the metrics and benchmarks. For now, the agencies intend to issue guidance to further clarify how they will apply the Community Development Financing Test.” (footnote omitted)).
7 The agencies have proposed increases to large banks’ regulatory-capital requirements for mortgage loans to borrowers who cannot afford a 20% down payment. That could lead to an increase in interest rates for low- and moderate-income and other historically underserved borrowers who cannot always afford a 20% down payment.
8 See 12 U.S.C. §§ 2903(a)(1) & 2906(a)(1) (emphasis added); see also id. § 2901(b).
9 The agencies acknowledge but dismiss this risk, pointing to the community benchmarks as an alternative avenue toward a “high satisfactory” or “outstanding” that would eventually check this evolution in the market. See Final Rule at 606. But, again, some of the community benchmarks are impractical, especially with respect to low-income borrowers and localities, so these benchmarks will sometimes be of little, if any, help in mitigating these risks.
10 See Final Rule at 632–34.