Joint ABA, CBA, HPC Comments on HUD’s Proposed Rule on Fair Housing Act Disparate Impact

Ladies and Gentlemen:

This comment is submitted by the American Bankers Association (ABA), the Consumer Bankers Association (CBA), and the Housing Policy Council (HPC) in response to the August 19, 2019 proposed rule of the U.S. Department of Housing and Urban Development (the Department or HUD). The Proposed Rule “follows a June 20, 2018, advance notice of proposed rulemaking, in which HUD solicited comments on the disparate impact standard set forth in HUD’s 2013 final rule.” The Department “proposes to amend HUD’s interpretation of the Fair Housing Act’s disparate impact standard to better reflect the Supreme Court’s 2015 ruling in Texas Department of Housing and Community Affairs v. Inclusive Communities Project, Inc.”

Prequel

ABA, the CBA, and the HPC vigorously support both the letter and spirit of the Fair Housing Act, and the associations and their members devote substantial resources on an ongoing basis to ensure that credit decisions for all loan applicants are made without regard to race or other prohibited bases. The issues faced by the Department in promulgating rules and enforcing the Fair Housing Act are complex, and this comment is intended to help ensure that HUD codifies a standard that is fully consistent with Supreme Court precedent and effectuates the Fair Housing Act’s requirements and goals in a clear and transparent manner.

The Supreme Court has recognized disparate impact liability under the Fair Housing Act as the law of the land, and we believe that application of the proper standard of disparate impact advances the Act’s objectives and purposes by providing a mechanism to further the goal of removing artificial, arbitrary, and unnecessary barriers to fair housing. At the same time, HUD’s reopening and proposing amendments to the 2013 Rule properly recognizes that the disparate impact standard must accurately reflect Supreme Court precedent and provide necessary guidance regarding the application of the law.

As described in detail below, we appreciate HUD’s analysis of the issues and support the amendments in HUD’s Proposed Rule, with some suggested modifications.

Discussion

A.Brief Background Regarding the 2013 Rule’s Incorrect Standard for Disparate Impact Under the Fair Housing Act

HUD promulgated its interpretation of the Fair Housing Act’s standard of disparate impact in 2013, but in 2015 the Supreme Court decided Inclusive Communities, which held that disparate impact claims were cognizable under the Fair Housing Act and discussed standards for, and limitations on, such claims. To understand the legal necessity of the amendments in the Proposed Rule, it is important to recognize a flaw in the 2013 Rule, which is briefly summarized below.

When HUD issued the 2013 Rule, significant Supreme Court precedent existed to both direct HUD and limit the reach of its rulemaking—precedent arising largely from the Supreme Court’s Wards Cove decision interpreting the standards of disparate impact claims in the Title VII employment context. There is a break-point in time, however, after which Title VII jurisprudence no longer guides the interpretation of the applicable standard for disparate impact liability under the Fair Housing Act. The break-point occurred when Congress passed the Civil Rights Act of 1991 and amended Title VII to abrogate—for future Title VII claims only—the disparate impact standard articulated by the Supreme Court in Wards Cove. The 1991 Title VII amendments were intended by Congress to make the disparate impact standard for employment discrimination more plaintiff friendly and more difficult for a defendant to rebut, and they superseded Wards Cove on issues including, among others, the burden of proof and requirement to isolate specific practices being challenged. While displacing Wards Cove and relaxing the standard, the 1991 amendments did not allow money damages for future Title VII disparate impact claims.

Significantly, Congress took no similar action to amend the Fair Housing Act to displace Wards Cove, and the Supreme Court has expressly held that, because Congress acted only to amend Title VII, Wards Cove remains binding precedent for disparate impact claims brought under other statutes. SeeGross v. FBL Fin. Servs., Inc., 557 U.S. 167, 174 (2009) (“[w]e cannot ignore Congress’ decision to amend Title VII’s relevant provisions but not make similar changes to the ADEA. When Congress amends one statutory provision but not another, it is presumed to have acted intentionally”); Smith v. City of Jackson, Miss., 544 U.S. 228, 240 (2005) (“While the relevant 1991 amendments expanded the coverage of Title VII, they did not amend the [Age Discrimination in Employment Act] ADEA or speak to the subject of age discrimination. Hence, Wards Cove’s pre-1991 interpretation of Title VII’s identical language remains applicable to the ADEA”). Yet, HUD erroneously rejected Wards Cove as supplying the governing standard for Fair Housing Act disparate impact claims in its 2013 Rule, stating that “HUD does not agree ... that Wards Cove even governs Fair Housing Act claims” and referencing Wards Cove as “superseded.” 2013 Rule at 11,473. Compounding the error and reaching far beyond its regulatory authority, HUD chose the “Title VII discriminatory effects standard codified by Congress in 1991” as the standard to govern disparate impact claims under the Fair Housing Act. Id. at 11,474. Because HUD lacks legislative authority and authority to overrule Supreme Court precedent, the 2013 Rule is flawed.

The 2013 Rule’s legal error in rejecting Wards Cove in the context of FHA and adopting the standard of the 1991 Title VII amendments was confirmed in 2015 when the Supreme Court decided Inclusive Communities and continued to rely on Wards Cove in the Fair Housing Act context, noting only that Wards Cove was “superseded by statute on other grounds”—i.e., superseded by the 1991 amendments for claims arising under Title VII. 135 S. Ct. at 2523. The Proposed Rule makes clear that the amendments will correct the 2013 Rule’s legal error, and HUD specifically cites Wards Cove with approval as providing binding guidance regarding the Fair Housing Act’s standard of disparate impact. E.g., Proposed Rule at 42,858 & n.43. The Proposed Rule adopts an approach that would be, if adopted, aligned with the Supreme Court’s emphasis on the “cautionary standards,” “safeguards,” and “limitations” on disparate impact claims brought under the Fair Housing Act.

B.The Proposed Rule’s Implementation of the Fair Housing Act’s Standard of Disparate Impact Aligns With Supreme Court Precedent and Operates to Provide Clarity to Entities Seeking to Voluntarily Comply with the Law

1.Available Remedies in Cases Where Disparate Impact Liability Is Proven

We offer several suggestions and observations regarding the language of, and supplementary information concerning, Proposed Rule Section 100.7 (“Liability for Discriminatory Housing Practices”), which addresses remedies in disparate impact cases. The Department’s proposed provision states as follows:

The remedy in an administrative discriminatory effect case should concentrate on eliminating or reforming the discriminatory practice so as to eliminate disparities between persons in a particular protected class and other persons through neutral means, and may include equitable remedies, and, where pecuniary damage is proved, compensatory damages or restitution. Punitive or exemplary damages shall not be available as a remedy.

Proposed Rule Section 100.7(c). HUD correctly states that punitive damages are not available as a remedy in administrative cases finding disparate impact liability. However, administrative law judges (“ALJs”) do not have statutory authority under the Fair Housing Act to award punitive damages in any proceeding (whether presenting a claim of disparate impact or disparate treatment), and Proposed Rule Section 100.7(c) should be modified to make that clear. HUD should also clarify in Proposed Rule Section 100.7(c) that civil penalties, which ALJs have statutory authority to award, are not available as a remedy in administrative disparate impact cases. In response to the specific question presented, at the final stage the Proposed Rule should also be modified to state that punitive damages and civil penalties are not proper remedies in Fair Housing Act court cases where disparate impact liability is proven. The bases for these suggestions are described in detail below.

HUD’s guidance regarding remedies available in disparate impact cases properly reflects the Supreme Court’s focus on what happens “when courts do find liability under a disparate-impact theory.” Inclusive Communities, 135 S. Ct. at 2524. The Supreme Court directs that “[r]emedial orders in disparate-impact cases should concentrate on the elimination of the offending practice that arbitrarily operates to discriminate on the basis of race.”Id. (emphasis added here and throughout unless otherwise indicated) (alterations omitted). HUD properly addresses the directive with respect to administrative cases establishing disparate impact liability by stating that remedies “should concentrate on eliminating or reforming the discriminatory practice so as to eliminate disparities between persons in a particular protected class and other persons through neutral means.” Proposed Rule Section 100.7(c).

Given the terms of the statute itself, however, the Proposed Rule’s list of available remedies would benefit from more precise language. The Fair Housing Act authorizes administrative proceedings in 42 U.S.C. § 3612 (“Enforcement by Secretary”), but ALJs do not have statutory authority to award punitive damages in any proceeding. Section 3612(g)(3) lists the forms of “relief” that ALJs are authorized to order, including “actual damages,” “injunctive or other equitable relief,” and “a civil penalty.” Thus, the language of Proposed Rule Section 100.7(c) stating that “[p]unitive or exemplary damages shall not be available as a remedy” in administrative disparate impact cases might be misread as suggesting that punitive damages are available in administrative disparate treatment cases.

Putting aside the statutory authority of ALJ, the issue that HUD raised is important in the context of the Inclusive Communities Court’s directive that “elimination of the offending practice” is the proper remedy in disparate impact cases. The very nature of a disparate impact violation is such that the defendant had no intent to discriminate, therefore punitive damages that punish and deter bad conduct are not appropriate remedies for these claims. As the Supreme Court has held, “[t]he purpose of punitive damages is to punish the defendant for his willful or malicious conduct and to deter others from similar behavior.” Memphis Cmty. Sch. Dist. v. Stachura, 477 U.S. 299, 307 n.9 (1986). Disparate impact liability arises from the unintended effects of a race-neutral policy or practice, not the defendant’s intentional treatment of individuals in a discriminatory manner. Thus, under the circumstances presented by a pure disparate impact violation, the goals of punishment and deterrence cannot be well-served by an award of punitive damages because the defendant had no intention to violate the Fair Housing Act in the first place.

These same concepts warrant further amending Proposed Rule Section 100.7(c) to clarify that civil penalties are not appropriate remedies in disparate impact cases. While ALJs have statutory authority to order, where “appropriate,” civil penalties in administrative proceedings, see 42 U.S.C. § 3612(g)(3), “the remedy of civil penalties is similar to the remedy of punitive damages,” the difference being that the former is recoverable only by the government, while the latter is recoverable by private plaintiffs, Tull v. United States, 481 U.S. 412, 423 n.7 (1987). Just like punitive damages, awards of civil “penalties, which go beyond compensation, are intended to punish, and label defendants wrongdoers.” Gabelli v. S.E.C., 568 U.S. 442, 451–52 (2013); see also Hudson v. United States, 522 U.S. 93, 102 (1997) (addressing civil penalties in context of federal banking statutes and noting that “all civil penalties have some deterrent effect”), and Tull, 481 U.S. at 423 n.7 (an “important characteristic of the remedy of civil penalties is that it exacts punishment”). Existing Fair Housing Act jurisprudence confirms that civil penalties should not be awarded in disparate impact cases. For example, the legislative history of the 1988 amendments confirms that that civil penalties “are not automatic in every case”—courts should focus on “the nature and circumstances of the violation” and “the goal of deterrence” in awarding civil penalties. House Comm. On the Judiciary, Fair Housing Amendments Act of 1988, H.R. REP. No. 711, 100th Cong., 2d Sess. 18, reprinted in 1988 U.S.C.C.A.N. 2173, 2201. Also, appellate courts have determined that a Fair Housing Act civil penalty “is especially appropriate when a defendant is found to have intentionally discriminated,” Smith & Lee Associates, Inc. v. City of Taylor, Mich., 102 F.3d 781, 798 (6th Cir. 1996), and have reversed civil penalty awards lacking a focus on serving the goal of deterring future discriminatory conduct, e.g., Morgan v. Sec'y of Hous. & Urban Dev., 985 F.2d 1451 (10th Cir. 1993).

Finally, the Proposed Rule is designed to provide guidance to courts as well as administrative officials and thus the Proposed Rule should be modified at the final rule stage to state that punitive damages and civil penalties are not appropriate remedies in disparate impact cases filed in courts. Punitive damages and civil penalties should not be awarded by courts in disparate impact cases for the same reasons as those described above regarding the complete disconnect between the punishment and deterrence justifications for those remedies, on the one hand, and the unintentional nature of a disparate impact violation, on the other. This guidance would apply to lawsuits filed under Section 3612(o) (“election” lawsuits) and Section 3613 (“Enforcement by Private Persons”), because those provisions of the Fair Housing Act authorize courts to award punitive damages (but not civil penalties). The guidance would also apply to lawsuits filed by the Attorney General under Section 3614, since that provision of the Fair Housing Act authorizes a court to award both punitive damages and civil penalties.

2.The Proposed Rule Defines a Prima Facie Case of Disparate Impact in a Manner Consistent with Supreme Court Precedent

We provide the comments below in response to the language of, and supplementary information concerning, Proposed Rule Section 100.500(b) (“Prima facie burden”). Proposed Rule Section 100.500(b) properly describes the requirement at the initial, pleading stage of a case for a plaintiff to assert “facts plausibly alleging each of the [] elements” of a disparate impact claim. There is no requirement for a plaintiff to prove a case at the pleading stage, and HUD’s formulation requiring plausible allegations is not only consistent with the Supreme Court’s rules of pleading, but also comports with the specific instruction of Inclusive Communities regarding the application of pleading requirements to disparate impact claims under the Fair Housing Act.

The Supreme Court directs lower courts considering the sufficiency of allegations at the pleading stage to “begin by taking note of the elements a plaintiff must plead to state a claim.” Ashcroft v. Iqbal, 556 U.S. 662, 675 (2009). Each “element of a cause of action [] must be adequately alleged at the pleading stage in order for the case to proceed” and “[i]f a plaintiff's allegations, taken as true, are insufficient to establish [an element of a cause of action], then the complaint must be dismissed; if they are sufficient, then the plaintiff is entitled to an opportunity to prove them.” Lexmark, 572 U.S. at 134 n.6 (2014).

Under Inclusive Communities, a “plaintiff who fails to allege facts at the pleading stage ... cannot make out a prima facie case of disparate impact.” 135 S. Ct. at 2523. The Supreme Court also instructed that “[c]ourts must [] examine with care whether a plaintiff has made out a prima facie case of disparate impact” because “prompt resolution of these cases is important.” Id.Effective, clear, and consistently applied standards for gatekeeping of disparate impact claims are important because, “[w]ithout adequate safeguards at the prima facie stage, disparate impact liability might cause race to be used and considered in a pervasive way and would almost inexorably lead governmental or private entities to use numerical quotas, and serious constitutional questions then could arise.” Id. Thus, federal appellate courts applying the “Inclusive Communities framework” at the “motion to dismiss stage” consider whether “plaintiffs sufficiently alleged a prima facie case of disparate impact.” Reyes, 903 F. 3d at 429 (concluding that “Plaintiffs have made a prima facie case” at the pleading stage).

Proposed Rule Sections 100.500(b)(1)-(5) also properly identify and explain the “elements”—i.e., essential components—of a prima facie case of disparate impact. In response to the first question presented in the Proposed Rule, and as explained in detail below, HUD’s articulation of the elements closely tracks the language of controlling Supreme Court precedent. In sum, the Proposed Rule strikes the appropriate balance envisioned by Inclusive Communities to provide a framework that promotes “heartland” uses of the theory while adhering to the “cautionary standards,” “safeguards,” and “limitations” on disparate impact that the Supreme Court described as “necessary” to protect against “abusive” claims and preclude “serious constitutional questions.” 135 S. Ct. at 2522, 2524. Below we discuss each element in detail.

i. Specific, Identifiable Policy or Practice.

Proposed Rule section 100.500(b) confirms that the starting point of a disparate impact claim is a “specific, identifiable policy or practice,” and subsection 100.500(b)(ii) reiterates that disparate impact plaintiffs must point to a “specific practice” at issue. This tracks Supreme Court precedent describing a specific policy or practice as the sine qua non of a disparate impact case. See Inclusive Communities, 134 S. Ct. at 2523 (disparate impact claim “must fail if the plaintiff cannot point to the defendant’s policy or policies”); Wards Cove, 490 U.S. at 657 (plaintiff must isolate the “specific or particular ... practice that has created the disparate impact”); Meacham v. Knolls Atomic Power Lab., 554 U.S. 84, 100-101 (2008) (“[i]dentifying a specific practice is not a trivial burden” and “the requirement has bite”). This concept parallels how federal regulators examine disparate impact risk in fair lending reviews. See Interagency Fair Lending Examination Procedures, Appendix A at 26, 28 (examiners review a “specific neutral policy or criterion” and “the policy or criterion suspected of producing a disproportionate adverse impact on a prohibited basis should be clear enough that the nature of action to correct the situation can be determined”).

We support this “specific, identifiable policy or practice” language as a critical part of the Proposed Rule, and believe that it will clarify the disparate impact standard for all stakeholders. The Department observes that “many parties have failed to identify a ‘specific, identifiable practice.’” Proposed Rule at 42,858. This observation is unfortunately correct—federal appellate courts are regularly required to review and dispose of purported disparate impact claims that fail to properly identify a specific policy subject to challenge. See, e.g., Khan v. City of Minneapolis, 922 F.3d 872, 875 (8th Cir. 2019) (rejecting “disparate-impact claim under the FHA” because the “allegation does not even begin to describe a city policy”); Ellis v. City of Minneapolis, 860 F.3d 1106, 1114 (8th Cir. 2017) (plaintiffs “have not pleaded sufficient facts to plausibly support the existence of such a policy” and thus “have not pleaded a prima facie case of disparate impact under the FHA”); City of Los Angeles v. Wells Fargo & Co., 691 F. App'x 453, 455 (9th Cir. 2017) (rejecting disparate impact claims premised on a theory that the defendant lenders “failed to adequately monitor [their] loans for disparities,” concluding that a failure-to-adequately-monitor “is not a policy at all”); City of Los Angeles v. Bank of Am. Corp., 691 F. App'x 464, 465 (9th Cir. 2017) (same); Frederick v. Wells Fargo Home Mortg., 649 F. App'x 29, 30 (2d Cir. 2016) (rejecting plaintiffs’ claim that “defendants conduct had a disparate impact” because they “failed to identify any specific policy or practice of the defendants that had such an effect”); City of Joliet, Illinois v. New W., L.P., 825 F.3d 827, 830 (7th Cir. 2016) (affirming trial verdict in defendant’s favor on a Fair Housing Act disparate impact claim because plaintiff challenged “a specific decision, not part of a policy”).

The appellate cases cited in the paragraph above should not have proceeded beyond the initial stages in the lower courts, but the lack of clarity regarding the policy-identification requirement has resulted in lower court decisions that improperly permit claims to proceed even where no specific policy has been identified. For example, Fair Housing Act disparate impact cases have improperly been permitted to proceed past the pleading stage based on a purported identification of “mortgage lending and servicing policies; pricing and marketing policies; various underwriting policies; loan servicing and loss mitigation policies; and foreclosure-related policies.” Cty. of Cook v. HSBC N. Am. Holdings Inc., 314 F. Supp. 3d 950, 967 (N.D. Ill. 2018). Focusing on even one of these broad categories of “policies” exposes the flaw in the plaintiff’s approach. Lenders often have hundreds of “underwriting policies,” but which of those myriad underwriting policies is being challenged under a disparate impact theory? Consideration of the Supreme Court’s directive that “remedial orders in disparate impact cases should concentrate on the elimination of the offending practice” serves to further undercut the improper amalgamation approach—if a plaintiff purports to challenge entire business operations like “mortgage lending and servicing policies,” it is hard to imagine what “remedial order” could be fashioned beyond the elimination of the business in its entirety. The misguided approach stands in stark contrast to disparate impact claims that have been upheld by appellate courts, such as those upheld by the Fourth Circuit in Reyes v. Waples Mobile Home Park Limited Partnership, where the plaintiff identified the defendant’s specific “Policy of evicting occupants that are unable to provide documentation of legal status in the United States [as] caus[ing] a disproportionate number of Latinos to face eviction from the Park compared to the number of non-Latinos who faced eviction based on the Policy.” 903 F.3d 415, 428 (4th Cir. 2018).

The 2013 Rule improperly omits the threshold step of requiring the identification of the specific policy or practice, and thus, under the regulation as written, differentials observed in data may prompt disparate impact lawsuits or investigations, even though socio-economic differences among different groups—and not “a defendant’s policy”—most often explain the disparities inoutcome. Inclusive Communities, 135 S. Ct. at 2523. We therefore support HUD’s proposal to update the 2013 Rule to comport with Supreme Court precedent and with the direction that the federal financial institutions regulators have provided to their fair lending examiners. SeeInteragency Fair Lending Examination Procedures, Appendix A at 26 (“[g]ross HMDA denial or approval rate disparities are not appropriate for disproportionate adverse impact analysis because they typically cannot be attributed to a specific policy or criterion”). The Proposed Rule helpfully provides much needed clarification that plaintiffs must challenge a “specific, identifiable, policy or practice” and that it “is insufficient to identify a program as a whole without explaining how the program itself causes the disparate impact as opposed to a particular element of the program.” Proposed Rule at 42,858.

In response to HUD’s fourth question, the Proposed Rule’s guidance will decrease costs and economic burdens associated with litigating disparate impact claims. Specifically, the guidance will reduce the costs of discovery for both plaintiffs and defendants by encouraging discovery targeted to the specific policy cited as the source of the claim. Additionally, the Proposed Rule’s language will help eliminate improper challenges based not on policies, but based solely on statistical disparities, which the Department properly describes as inappropriate. See Proposed Rule at 42,860 (disparate impact claims cannot proceed based on “statistical imbalances or disparities alone”).

ii. Element 1: Challenged Policy Is Artificial, Arbitrary, and Unnecessary

The first element of a disparate impact claim is set forth in Proposed Rule Section 100.500(b)(1), and it requires factual allegations at the pleading stage that the “challenged policy or practice is arbitrary, artificial, and unnecessary to achieve a valid interest or legitimate objective such as a practical business, profit, policy consideration, or requirement of law.” The term “arbitrary, artificial, and unnecessary” articulated in the Proposed Rule comes directly from Inclusive Communities, where the Supreme Court specifically held that “[d]isparate impact liability mandates the ‘removal of artificial, arbitrary, and unnecessary barriers,’ not the displacement of valid governmental priorities” and “[g]overnmental or private policies are not contrary to the disparate-impact requirement unless they are ‘artificial, arbitrary, and unnecessary barriers.’” Inclusive Communities, 135 S. Ct. 2522, 2524. This element must be alleged at the “pleading stage” because, as the Supreme Court explains, “[w]ere standards for proceeding with disparate-impact suits not to incorporate at least the safeguards discussed here, then disparate-impact liability might displace valid governmental and private priorities, rather than solely 'removing artificial, arbitrary, and unnecessary barriers.’” Id. at 2523-2524(alterations omitted).

Although the Supreme Court uses the ‘artificial, arbitrary, and unnecessary’ language for the first time in the Fair Housing Act context in 2015, this limitation on the use of disparate impact has been recognized for over 45 years. The language traces back to Griggs v. Duke Power Co., 401 U.S. 424 (1971), a case that recognized disparate impact claims under Title VII and upon which the Supreme Court heavily relied in recognizing disparate impact liability under the Fair Housing Act. See Inclusive Communities, 135 S. Ct. at 2516-2517 (in-depth analysis of facts and legal holdings of Griggs), 2518 (“the logic of Griggs and Smith provides strong support for the conclusion that the FHA encompasses disparate-impact claims”), 2519 (examining the Fair Housing Act’s “results-oriented language” and “comparison to the antidiscrimination statutes examined in Griggs and Smith is useful”), (considering argument that “the phrase ‘because of race’” in Fair Housing Act precludes disparate impact, but holding “Griggs and Smith, however, dispose of this argument. Both Title VII and the ADEA contain identical ‘because of language,’ and the Court nonetheless held those statutes impose disparate-impact liability”).

An examination of Griggs is necessary to fully understand the “artificial, arbitrary, and unnecessary” safeguard on disparate impact claims and the Supreme Court’s reasons for imposing it in cases arising under the Fair Housing Act. In Griggs, the plaintiffs challenged a power company’s policy of requiring employees of the Coal Handling, Operations, Maintenance, or Laboratory departments to complete high school or pass a general intelligence test. Griggs, 401 U.S. at 427-28. The criteria were adopted “without meaningful study of their relationship to job-performance ability” and the evidence showed that “employees who have not completed high school or taken the tests have continued to perform satisfactorily and make progress in departments for which the high school and test criteria are now used.” Id. at 431–32. In other words, there was no relationship between ability to perform the jobs and the test criteria. The Supreme Court noted the defendant’s “lack of discriminatory intent,” but held that Title VII allowed claims challenging “the consequences of employment practices, not simply the motivation” and found disparate impact liability. Id. at 432.

As recognized by the Supreme Court in Inclusive Communities, however, the Griggs Court “put important limits on its holding: namely, not all employment practices causing a disparate impact impose liability under [Title VII].” 135 S. Ct. at 2517 (interpreting Griggs). Rather, Congress requires the “removal of artificial, arbitrary, and unnecessary barriers to employment when the barriers operate invidiously to discriminate.” 401 U.S. at 431. The high school completion requirement and intelligence tests at issue in Griggs were “artificial, arbitrary, and unnecessary barriers” that gave rise to disparate impact liability because they did not measure capability to perform the specific jobs in question, and “[w]hat Congress has commanded is that any tests used must measure the person for the job and not the person in the abstract.” Id. at 436.

Inclusive Communities itself helps distinguish “artificial barriers to housing” from the “heartland of disparate-impact suits.” 135 S. Ct. at 2522. As an example of a “heartland” case, the Supreme Court cites a “post-Hurricane Katrina ordinance restricting the rental of housing units to only ‘blood relatives’ in an area of the city that was 88.3% white and 7.6% black.” Id. Consider: What legitimate governmental objective could possibly be served by imposing, in the wake of a catastrophic destruction of housing, a policy requirement that apartment tenants share a genetic relationship? The local government may have some response, but a disparate impact claim challenging the blood-relation ordinance easily meets the “arbitrary, artificial, and unnecessary” element at the pleading stage. The circumstances ofEllis v. City of Minneapolis provide a contrast. 860 F.3d 1106 (8th Cir. 2017). There, a city enforced its habitability code, precluding rodent “infestation” and illegal electrical “wiring,” against a low-income housing provider. Id. at 1108. The housing provider argued that complying would disparately impact its residents, and the city should apply lower habitability standards. The Eighth Circuit found no “facts plausibly demonstrating that the housing-code standards complained of are arbitrary and unnecessary under the FHA,” and rejected the disparate impact claim. Id. at 1112.

To be sure, like any other element of any other cause of action, the ‘arbitrary, artificial, and unnecessary’ element of a disparate impact claim must be plausibly alleged—not proved—at the pleading stage before the case can proceed to the merits. SeeLexmark Int'l, Inc. v. Static Control Components, Inc., 572 U.S. 118, 134 n.6 (2014) (an “element of a cause of action [] must be adequately alleged at the pleading stage in order for the case to proceed” and “[i]f a plaintiff's allegations, taken as true, are insufficient to establish [an element of a cause of action], then the complaint must be dismissed; if they are sufficient, then the plaintiff is entitled to an opportunity to prove them” (citing Ashcroft v. Iqbal, 556 U.S. 662, 678–679 (2009))). Inclusive Communities emphasizes the specific importance of this principle of law in the context of disparate impact claims under the Fair Housing Act, instructing that “at the pleading stage,” courts “must [] examine with care whether a plaintiff has made out a prima facie case of disparate impact” because “prompt resolution of these cases is important.” 135 S. Ct. at 2523.

While the “arbitrary, artificial, and unnecessary” element must be plausibly alleged with facts at the pleading stage, the discussion in the Supplementary Information section of the Proposed Rule confirms that the Department has engaged in a considered analysis of the pleading requirements and has crafted additional, appropriate guidance to facilitate application of this aspect of the framework. Specifically:

HUD recognizes that plaintiffs will not always know what legitimate objective the defendant will assert in response to the plaintiff’s claim or how the policy advances that interest, and, in such case, will not be able to plead specific facts showing why the policy or practice is arbitrary, artificial, and unnecessary. In such cases, a pleading plausibly alleging that a policy or practice advances no obvious legitimate objective would be sufficient to meet this pleading requirement. However, in cases where a policy or practice has a facially legitimate objective, the plaintiff must allege facts at the pleading stage sufficient to support a plausible allegation that the policy is arbitrary, artificial, and unnecessary.

Proposed Rule at 42,858. Applying this guidance to the two contrasting examples of disparate impact claims described above—one challenging the blood-relation ordinance referenced by the Supreme Court in Inclusive Communities and the other challenging the habitability code at issue in Ellis—demonstrates the operation of the pleading requirement in practice. The blood-relation ordinance appears unrelated to any legitimate governmental objective relating to regulation of residential landlord and tenant relationships, and thus, “the arbitrary, artificial, and unnecessary” element would be satisfied with factual allegations that the ordinance advances no obvious legitimate objective. In contrast, the habitability code provisions at issue in Ellis serve the obvious legitimate objective of ensuring sanitary and safe housing conditions, and thus the plaintiffs would be required to plead facts alleging why the habitability code provisions are artificial, arbitrary and unnecessary.

HUD’s articulation of the element balances the concerns of plaintiffs with the Supreme Court’s mandate that “[g]overnmental or private policies are not contrary to the disparate-impact requirement unless they are ‘artificial, arbitrary, and unnecessary barriers’” and “[w]ere standards for proceeding with disparate-impact suits not to incorporate at least the safeguards discussed here, then disparate-impact liability might displace valid governmental and private priorities, rather than solely 'removing artificial, arbitrary, and unnecessary barriers.’” Inclusive Communities, 135 S. Ct. at 2524. Although this element must be plausibly alleged with facts at the pleading stage, there is no requirement for a plaintiff to prove at the pleading stage that a policy is “artificial, arbitrary, and unnecessary.” This is consistent with the general rules of pleading discussed above. See supra at 8-9.

Beyond the “arbitrary, artificial, and unnecessary” language, this element of a prima facie case of disparate impact recognizes the right of businesses to pursue a profit motive and to adopt credit-worthiness standards in accordance with the businesses’ assessment of and tolerance for risk. Inclusive Communities confirms that “disparate-impact liability must be limited so ... regulated entities are able to make practical business choices and profit-related decisions that sustain a vibrant and dynamic free-enterprise system” and “[e]ntrepreneurs must be given latitude to consider market factors.” 135 S. Ct. at 2523; see also DOJ Fair Lending Enforcement Position Statement at 4 (“we would not condemn any facially neutral policy or practice the absence of which would demonstrably jeopardize the safety or soundness of any financial institution” and “we do not expect lenders to adopt any particular credit-worthiness requirements”). Consistent with Inclusive Communities, the Proposed Rule’s recognition that disparate impact liability will not result from pursuit of “a valid interest or legitimate objective such as a practical business, profit, policy consideration, or requirement of law” is critical to a properly-functioning regulation, and we strongly support implementation of this language at the final rule stage. Proposed Rule Section 100.500(b)(1).

iii.Element 2: Robust Causal Link that Shows the Specific Practice Is the Direct Cause of the Alleged Discriminatory Effect.

The second essential element of a disparate impact claim is set forth in Proposed Rule Section 100.500(b)(2). It requires facts alleging “a robust causal link between the challenged policy or practice and a disparate impact on members of a protected class that shows the specific practice is the direct cause of the discriminatory effect.” Codification of the requirement of a “robust causal link” at the final rule stage is essential to ensuring that all interested parties operate under a consistent and legally correct framework.

The Inclusive Communities decision mandates that a Fair Housing Act plaintiff must meet a standard of “robust” causality, linking the challenged practice with a disparity, as an element of a prima facie case of disparate impact. 135 S. Ct. at 2523. The Supreme Court expressly holds that “[i]f a statistical discrepancy is caused by factors other than the defendant’s policy, a plaintiff cannot establish a prima facie case” and “a disparate-impact claim that relies on a statistical disparity must fail if the plaintiff cannot point to a defendant’s policy or policies causing that disparity.” 135 S. Ct. at 2514, 2523. The “robust causality requirement” is “necessary” to “protect[] defendants from being held liable for racial disparities they did not create.” Id. at 2523 (quoting Wards Cove, 490 U.S. at 653). And “robust” causation has long been the rule promulgated by Supreme Court disparate impact jurisprudence applicable in the Fair Housing Act context. Indeed, the Inclusive Communities causal standard is entirely consistent with the Supreme Court’s prior ruling in Wards Cove. There, the Supreme Court upheld a “specific causation requirement,” whereby a plaintiff must “show[] that each challenged practice has a significantly disparate impact on [] opportunities for whites and nonwhites,” and explained that “[t]o hold otherwise would result in [defendants] being potentially liable for the myriad of innocent causes that may lead to statistical imbalances.” Wards Cove, 490 U.S. at 658. HUD properly recognizes the application of Wards Cove and specifically relies on the decision (see Proposed Rule at 42858 n.43 & n.44), and thus corrects a lingering legal error that pervades the 2013 Rule (see supra at 3-4).

Thus, in response to HUD’s third and fifth specific questions, a decision not to amend the 2013 Rule will perpetuate existing conflict between the regulation and the Supreme Court’s binding legal standards. In fact, courts have recently examined the conflict between Inclusive Communities and the 2013 Rule and have expressly noted: “We read the Supreme Court’s opinion in [Inclusive Communities] to undoubtedly announce a more demanding test than set forth in the HUD regulation. ... [T]he HUD regulation contains no ‘robust causation’ requirement; rather it requires only a showing that ‘a challenged practice caused or predictably will cause a discriminatory effect.’” Inclusive Communities Project, Inc. v. Lincoln Prop. Co., 920 F.3d 890, 902 (5th Cir. 2019); see also Reyes, 903 F.3d at 424 n.4 (“Without deciding whether there are meaningful differences between the frameworks,” the Fourth Circuit found “that the standard announced in Inclusive Communities rather than the HUD regulation controls our inquiry”).

The element of robust causality may be challenging to plead and prove in “novel” cases far from the “heartland” of disparate impact under the Fair Housing Act, but the Supreme Court requires the application of this “safeguard at the prima facie stage” to prevent the “serious constitutional questions” that may otherwise arise. Inclusive Communities, 135 S. Ct. at 2523. Indeed, appellate courts have upheld the sufficiency of prima facie allegations of “robust” causation in appropriate cases. See, e.g., Reyes, 903 F.3d at 429 (“Plaintiffs satisfied the robust causality requirement by asserting that the specific Policy requiring all adult Park tenants to provide certain documents proving legal status was likely to cause Latino tenants at the Park to be disproportionately subject to eviction compared to non-Latino tenants at the Park”).

iv.Element 3: Alleged Disparity Caused by the Specific Practice Adversely Effects a Protected Class

The third element of a prima facie case of disparate impact is that “the alleged disparity caused by the policy or practice has an adverse effect on members of a protected class.” Proposed Rule Section 100.500(b)(3). The question of whether a particular outcome is “adverse” to a protected group may sometimes have an easy answer (such as, for example, when a person is evicted from a dwelling), but in many circumstances the answer is not always clear. Inclusive Communitiescautions against challenging policies in circumstances where, “from the standpoint of determining advantage or disadvantage to racial minorities,” it “seems difficult to say as a general matter” that one outcome “is discriminatory, or vice versa.” 135 S. Ct. at 2523. The Department provides helpful guidance on this point, explaining that this “element would require a plaintiff to explain how the policy or practice identified has a harmful impact” on a protected group. Proposed Rule at 42,858.

v.Element 4: Alleged Disparity Is Significant

Proposed Rule Section 100.500(b)(4) articulates the fourth element of a prima facie case of disparate impact, and requires facts showing “[t]hat the alleged disparity caused by the policy or practice is significant.” This element flows from longstanding Supreme Court precedent in the Title VII context that controls the analysis of disparate impact claims under the Fair Housing Act (see discussion supra at 3-4) and holds that the disparate impact theory requires a significant impact on a protected group. See Wards Cove, 490 U.S. at 657 (requiring a “showing that each challenged practice has a significantly disparate impact”); Washington v. Davis, 426 U.S. 229, 246–47 (1976) (disparate impact challenge to “practices disqualifying substantially disproportionate numbers of black” applicants). C.f. Int’l Bhd. of Teamsters, 431 U.S. at 339 (“approv[ing] the use of statistical proof, where it reached proportions comparable to those in this case”). HUD’s formulation of the element is consistent with the directions provided by federal regulators for assessing disparate impact risk—namely, that examiners should focus on “whether there is reason to suspect a significant disproportionate adverse impact.” Interagency Fair Lending Examination Procedures, Appendix A at 26; see also id. at 28 (guidance regarding testing for “the existence of the significant disparity”).

While the term “significant” has a specified mathematical definition in the world of statistics and quantitative data analysis, in the context of a prima facie case of disparate impact, “significant” means “substantial” or “meaningful.” In other words, “significant” must extend beyond mere statistical significance in that the disparity must have practical, real-world significance. The Proposed Rule correctly refrains from imposing an arbitrary cutoff, as the particular factual context is important to the evaluation.

For example, as the majority of an Eleventh Circuit panel recently observed in a concurrence, “under a disparate-impact theory of liability, proof of a violation requires the plaintiff to establish that the challenged policy produced a significant statistical disparity,” and a claim “comes up short” with a “tendentious assumption that the two loans identified by [plaintiff] suffice to establish ‘a disproportionately adverse effect on minorities.’”City of Miami Gardens, 931 F.3d at 1297; see also Schaw v. Habitat for Humanity of Citrus Cty., Inc., --- F.3d ---, 2019 WL 4458370, at *10 (11th Cir. Sept. 18, 2019) (rejecting Fair Housing Act disparate impact claim and noting “it’s not enough to show that a few people are affected by a policy—rather, the disparity must be substantial enough to raise an inference of causation”), andWaisome v. PortAuth. of New York & New Jersey, 948 F.2d 1370, 1376 (2d Cir. 1991) (applying Wards Cove and rejecting Title VII disparate impact claim based on finding that “though the disparity was found to be statistically significant, it was of limited magnitude”). As explained by a district court rejecting a disparate impact claim in an opinion that was later confirmed by the Ninth Circuit, “[t]he difference between 0.0033 percent and 0.0008 percent does not create a genuine dispute such that a jury must decide this issue” because “the evidentiary disparities are negligible. The [plaintiff] must provide evidence of a significantly disproportionate effect on minorities, and comparing thousandths of a percentage fails to meet the minimum threshold of Inclusive Communities.” City of Los Angeles v. Wells Fargo & Co., 2015 WL 4398858 (C.D. Cal. July 17, 2015), aff’d, 2017 WL 2304375 (9th Cir. May 26, 2017).

The Proposed Rule provides guidance to explain the meaning of the element requiring a “significant” disparity, with the Department properly clarifying that a “negligible disparity” is not enough, but rather it must be shown “that the statistical disparity identified is material and caused by the challenged policy or practice, rather than attributable to chance.” Proposed Rule at 42,858-859. And while Department’s guidance provides the necessary explanation that is presently missing in the context of adversarial claims, the Proposed Rule’s language tracks the disparate impact guidance provided by federal regulators in the supervisory context, which directs fair lending examiners to consider the “magnitude of the impact” and specifies that the “difference between the rate at which prohibited basis group members are harmed or excluded by the policy or criterion and the rate for control group members must be large enough that it is 19 303831007 v6unlikely that it could have occurred by chance.” Interagency Fair Lending Examination Procedures, Appendix A at 27-28.

vi.Element 5: Direct Link between Disparate Impact and Alleged Injury

The fifth element of a disparate impact claim requires facts alleging “a direct link between the disparate impact and the complaining party’s alleged injury.” Proposed Rule Section 100.500(b)(5). The Department explains its intent to “codify the proximate cause requirement under the Fair Housing Act.” Proposed Rule at 42,859. The requirement stems from a Supreme Court decision issued two years after Inclusive Communities was resolved, in which the Supreme Court held that Fair Housing Act plaintiffs must satisfy a “requirement” of pleading “direct” proximate causation. City of Miami v. Bank of America Corp., 137 S. Ct. 1296, 1305-06 (2017). Codification of the proximate cause element is a proper exercise of the Department’s authority to “make rules” interpreting the Fair Housing Act, see 42 U.S.C. § 3614a, and is essential to the implementation of a cohesive regulation that reflects all binding precedent.

More specifically, the Supreme Court confirmed that proximate cause is an “element” of a cause of action under the Fair Housing Act. City of Miami, 137 S. Ct. at 1305. And “proximate cause” functions “like any other element of a cause of action,” in that “it must be adequately alleged at the pleading stage in order for the case to proceed.” Lexmark Int’l, Inc. v. Static Control Components, Inc., 134 S. Ct. 1377, 1391 n.6 (2014). Thus, all Fair Housing Act plaintiffs must satisfy the proximate cause requirement, regardless of whether the claim proceeds under a disparate impact theory or disparate treatment theory.

The Department’s codification of the proximate cause requirement is important given the reality that “[t]he housing market is interconnected with economic social life. A violation of the [Fair Housing Act] may, therefore, be expected to cause ripples of harm to flow far beyond the defendant’s misconduct. Nothing in the statute suggests that Congress intended to provide a remedy wherever those ripples travel.” City of Miami, 137 S. Ct. at 1306 (citations and quotations omitted). Indeed, the Inclusive Communities decision echoes the considerations at the forefront of the proximate cause jurisprudence that the Supreme Court cited in City of Miami as supplying the directness principles applicable to claims under the Fair Housing Act. Compare Inclusive Communities, 135 S. Ct. 2523-24 (“[i]t may also be difficult to establish causation because of the multiple factors that go into investment decisions”), withHemi Grp., LLC v. City of New York, 559 U.S. 1, 15 (2010) (cited by Supreme Court in City of Miami) (finding no proximate cause due to presence of “independent factors,” including “independent actions of third and even fourth parties”).

Clarifying the application of the Supreme Court’s proximate cause ruling would provide guidance regarding the framework for analyzing a type of disparate impact claim that has been improperly been brought against our members. In particular, lenders have been forced to defend purported “disparate impact” claims of “pricing discrimination” that challenge the lender’s so-called “policy” of allowing discretion to independent third-party businesses (such as mortgage brokers) to set fees that are charged to customers. The City of Miami decision recognizing the proximate cause requirement confirms that any attempt to bring a similar claim must be dismissed: a lender cannot be held liable under the Fair Housing Act for a disparate impact where the independent actions of independent third parties break the proximate causal chain.

3.The Proposed Rule Establishes Proper Methods for Rebutting and Defending Against Disparate Impact Claims

Proposed Rule Section 100.500(c) provides three methods by which a “defendant, or responding party, may establish that a plaintiff’s allegations do not support a prima facie case of discriminatory effect.” In response to HUD’s first question, the proposed inclusion of defenses to liability comports with the overarching directive of Inclusive Communities that it is “necessary” to apply “cautionary standards,” “safeguards,” and “limitations” to disparate impact claims brought under the Fair Housing Act. Indeed, after recognizing the applicability of disparate impact, the Supreme Court’s articulation of the limitations on disparate impact claims, and the rationale for those limitations, constitute a major portion of the Inclusive Communities decision. We support amending the regulation to include the defenses articulated in the Proposed Rule as reflective of the design of the Inclusive Communities decision.

i.Demonstrate Plaintiff’s Failure to Allege Sufficient Facts to Support Each Element of a Prima Facie Case of Disparate Impact

As described in detail above, Proposed Rule Section 100.500(b) correctly articulates the “elements” of a “prima facie case” of disparate impact and properly requires a Fair Housing Act plaintiff to “state facts plausibly alleging each of the [] elements.” In Proposed Rule Section 100.500(c)(3), the Department provides a method for disposing of a disparate impact claim if “[t]he defendant demonstrates that the plaintiff has failed to allege sufficient facts under paragraph (b) of this section.” In other words, the provision permits a defendant to rebut a disparate impact claim at the pleading stage by showing that the plaintiff has failed to allege any of the essential elements of a prima facie case. The proposal aligns with the language of Inclusive Communities instructing courts and regulators to carefully scrutinize disparate impact allegations at the pleading stage to weed out meritless claims, but it also complies with the standards promulgated by the Federal Rules of Civil Procedure governing all litigation in federal courts and Supreme Court precedent setting the requirements a plaintiff must meet to state a plausible claim for relief.

Federal Rule of Civil Procedure 12(b)(6) allows defendants to dispose of lawsuits at the pleading stage by showing the plaintiff’s “failure to state a claim upon which relief can be granted.” As the Supreme Court has explained, courts considering whether a claim can survive a Rule 12(b)(6) motion must “begin by taking note of the elements a plaintiff must plead to state a claim.” Ashcroft, 556 U.S. at 675. “Threadbare recitals of the elements of a cause of action, supported by mere conclusory statements, do not suffice.” Id. at 679. Rather, courts must identify “well-pleaded factual allegations, ... assume their veracity and then determine whether they plausibly give rise to an entitlement to relief.” Id. “If a plaintiff's allegations, taken as true, are insufficient to establish [an element of a cause of action], then the complaint must be dismissed.” Lexmark, 572 U.S. at 134 n.6 (2014).

As explained above, Inclusive Communities provides instruction regarding the application of the general rules of pleading to disparate impact claims under the Fair Housing Act (see supra at 8), and holds that a “plaintiff who fails to allege facts at the pleading stage ... cannot make out a prima facie case of disparate impact.” 135 S. Ct. at 2523. The Supreme Court also instructed that “[c]ourts must [] examine with care whether a plaintiff has made out a prima facie case of disparate impact” because “prompt resolution of these cases is important.” Id.

Consistent with the Inclusive Communities instruction to “carefully” review and achieve “prompt resolution” of disparate impact cases, and in line with the Supreme Court’s pleading-stage plausibility standards, the Proposed Rule explains that “HUD proposes to provide that the defendants may raise [the] defenses in paragraph (c)” by filing “a rule 12(b)(6) motion to dismiss, [where] the defendant can make an argument under the paragraph (c) defense that the facts alleged in the complaint fail to allege sufficient facts to support a claim.” Proposed Rule at 42,859. This requires the plaintiff to allege plausible facts to proceed past the pleading stage, but it does not require the plaintiff to prove the elements of a disparate impact claim to survive the pleading stage. The Proposed Rule provides proper instruction regarding the gatekeeping role of courts and agencies presiding over disparate impact claims at the pleading stage, and we strongly support implementation of the defense at the final rule stage.

ii.Establish that a Third Party Materially Limits Defendant’s Discretion

The defense codified in Proposed Rule Section 100.500(c)(1) provides that a defendant may rebut a disparate impact claim at the pleading stage by showing “that its discretion is materially limited by a third party such as through: (i) A Federal, state, or local law; or (ii) A binding or controlling court, arbitral, regulatory, administrative order, or administrative requirement.” Lenders’ discretion is frequently limited by government-sponsored entities. The Department acted properly in proposing to amend the rule to include a defense that recognizes that adherence to governmental standards should not create disparate impact liability, and the proposed language accurately codifies the Supreme Court’s holding that “if the [plaintiff] cannot show a causal connection between the [defendant’s] policy and a disparate impact—for instance, because federal law substantially limits the Department’s discretion—that should result in dismissal of th[e] case.” 135 S. Ct. at 2524; see also Meyer v. Holley, 537 U.S. 280, 286-89 (2003) (limiting vicarious liability under the Fair Housing Act to agency relationships). For the reasons described below, we strongly support this defense as consistent with Inclusive Communities, but would suggest a slight modification to remove the “such as” language to clarify that the defense is not available absent a governmental or quasi-governmental limit.

Finalizing this defense is essential to properly implementing the standard of disparate impact announced by the Supreme Court, which held that disparate impact claims under the Fair Housing Act must not put regulated entities “in a double bind of liability.” Inclusive Communities, 135 S. Ct. at 2523. At present, the 2013 Rule fails to incorporate these principles, and opens the door to possible disparate impact liability for entities complying with contractual obligations set by the federal government. For instance, a loan servicer applying loss-mitigation criteria set by Treasury may face liability under the 2013 Rule for purported disparate impact caused by the criteria which the servicer is bound to apply. As another example, the purpose and design of several Dodd-Frank Act revisions to ability-to-repay and underwriting standards is to improve safety and soundness by causing fewer loans to be made to borrowers with weaker credit profiles, and to increase the cost of loans to those borrowers if they are made. In these circumstances, lenders may be faced with an impossible choice: violate government-imposed safety-and-soundness standards to avoid disparate impact liability, or comply and face disparate impact claims challenging disparities that may resulting from following governmental standards.

Yet, the Supreme Court is clear that adherence to government standards, policies, or programs should not create disparate impact liability, and the defense articulated in the Proposed Rule properly creates an exemption for lending and servicing practices that are undertaken in compliance with governmental or policies or programs. The critical element of the revision to the Proposed Rule is that it ensures the availability of disparate impact will not create a Catch-22 “double bind of liability” for lenders or servicers who are otherwise complying with government rules or assisting in effectuating the government’s residential housing policies. Inclusive Communities, 135 S. Ct. 2523.

iii.Defenses to Claims Challenging Automated, Algorithmic Models

Automated models are being used in all aspects of the mortgage lending process, representing a seismic shift in the industry since the passage of the Fair Housing Act in 1968. To take a few examples, algorithmic modeling technology is widely used to underwrite applications, price loans, manage accounts, and market to prospective borrowers, and the trend is only increasing. A fundamental purpose of a rule is to provide meaningful guidance regarding the application of the law to regulated entities, and the Department’s formulation of the Proposed Rule properly “recognize[s] that additional guidance is necessary in response to the complexity of disparate impact cases challenging these models.” Proposed Rule at 42,859. Not only does the proposal to establish defenses to claims challenging algorithmic models represent a proper exercise of the Department’s rulemaking obligation to provide guidance and clarity, but it also conforms to the key directives of Inclusive Communities requiring “adequate safeguards.”

The guidance in the Proposed Rule regarding disparate impact challenges to automated models is consistent with the trend of increased federal agency focus on modeling techniques used by the lending industry. Meaningful guidance on this topic that promotes voluntary compliance and provides defenses to frivolous or unnecessary disparate impact claims is appropriate and important—as HUD observes, automated models can be an “invaluable tool in extending access to credit and other services to otherwise underserved communities.” Proposed Rule at 42,859. Similarly, the Bureau has noted that “modeling techniques have the potential to benefit consumers,” in that they “could increase access to credit” for Americans with insufficient, stale, or no credit history; could “lower lenders’ costs” and result in cost savings to consumers “in the form of lower prices or in lenders’ ability to make smaller loans economically”; and “could allow lenders to better assess the creditworthiness of consumers who are already scored.” Bureau Modeling RFI at 11,186. Use of automated models may specifically further the goals of the Fair Housing Act by, for example, “reduc[ing] any discretionary judgments that may sometimes lead to discrimination.” Id. at 11,187. In fact, the Bureau recently released an update on a case study of a lender’s “use of alternative data and machine learning for its underwriting and pricing model.” The lender’s model expanded access to credit for all borrowers, and from the “fair lending” perspective, the “expansion of credit access reflected in the results provided occurs across all tested race, ethnicity, and sex segments resulting in the tested model increasing acceptance rates by 23-29% and decreasing average APRs by 15-17%.” Id.

The Department explains that the “first defense” in Section 100.500(c)(2)(i) allows a defendant to “show[] that the model is not the actual cause of the disparate impact alleged” by reviewing the model “piece-by-piece” to “demonstrate that each factor considered could not be the cause of the disparate impact and to show how each factor advances a valid objective.” Proposed Rule at 42,859. Of course, use of a variable in an automated model should present no disparate impact concern if the variable is predictive of credit risk and the variable does not serve as a proxy for a prohibited characteristic. See Board of Governors of the Federal Reserve System, “Report to Congress on Credit Scoring and Its Effects on the Availability and Affordability of Credit” (Aug. 2007) (“[u]nder court and regulatory agency interpretations, the test for disparate impact requires that a practice both have a disproportionate effect on a protected population and lack a sufficient business justification. An empirically derived, demonstrably and statistically sound credit-scoring model is likely to have a sufficient business rationale for the characteristics that constitute the model”); see also Interagency Fair Lending Examination Procedures, App. A at 26 (directing examiners evaluating disparate impact risk to consider “if the policy or criterion is obviously related to predicting creditworthiness and is used in a way that is commensurate with its relationship to creditworthiness,” and “examples are reliance on credit reports or use of debt-to-income ratio in a way that appears consistent with industry standards and with a prudent evaluation of credit risk”). A plaintiff could rebut by identifying an input or model that exists and is known or reasonably known to the defendant that would have a lesser impact on the protected class and would serve the defendant’s identified interest in an equally effective manner without imposing greater costs on, or creating other burdens for, the defendant. This defense properly reflects the principle that disparate impact liability must be “limited so employers and other regulated entities are able to make the practical business choices and profit-related decisions that sustain a vibrant and dynamic free-enterprise system.” Proposed Rule at 42,859 (quoting Inclusive Communities, 13 S. Ct. at 2518). While supporting the first defense, we would propose modifying the wording to remove the reference to “close proxies,” and thus eliminate possible confusion as there is no difference in meaning as compared to “substitutes ... for protected classes.”

The “second defense” in Section 100.500(c)(2)(i) applies in situations where the “challenged model is produced, maintained, or distributed by a recognized third party that determines industry standards, the inputs and methods within the model are not determined by the defendant, and the defendant is using the model as intended by the third party.” The defense meets one of the most significant challenges facing lenders in connection with possible claims challenging automated models; namely, that lenders commonly use models and algorithms that are developed by and proprietary to third parties. For example, the government-sponsored enterprises Fannie Mae and Freddie Mac require lenders to evaluate credit risk pursuant to automated underwriting systems containing models proprietary to those entities. Yet, lenders have no ability to alter the models used by Fannie Mae and Freddie Mac, and lenders are not in a position to “justify” each element of such a model, much less the relationships among all the variables. As HUD notes, “[i]n these situations, the defendant may not have access to the reasons these factors are used or may not even have access to the factors themselves, and, therefore, may not be able to defend the model itself.” Proposed Rule at 42,859.

The second modeling defense ensures that disparate impact claims are brought against the proper party—i.e., the party responsible for the model allegedly causing the disparate impact. Thus, the defense properly embraces the “robust causality requirement,” which precludes “defendants from being held liable for racial disparities they did not create.” Inclusive Communities, 135 S. Ct. at 2523; Proposed Rule at 42,859 (“a successful defense under this section would demonstrate the lack of a robust causal link between the defendant’s use of the model and the alleged disparate impact”). HUD correctly observes that even if a disparate impact is established, “the plaintiff would only remove the model from use by one party, whereas suing the party that is actually responsible for the creation and design of the model would remove the disparate impact from the industry as a whole.” Id. HUD’s observation is directly on point with the Supreme Court’s admonition that “[r]emedial orders in disparate-impact cases should concentrate on the elimination of the offending practice that arbitrarily operates invidiously to discriminate on the basis of race.” Inclusive Communities, 135 S. Ct. at 2524.

These defenses articulated in the Proposed Rule are consistent with a Fair Housing Act challenge to algorithmic models recently filed by the National Fair Housing Alliance (NFHA) against the operator of an advertising platform that displays advertisements concerning housing and real-estate related transactions. See Nat’l Fair Hous. Alliance, et al., v. Facebook, Inc., Case No. 1:18-02689 (S.D.N.Y.). NFHA’s complaint alleged that “data-analyzing algorithms” developed and proprietary to the defendant “empower advertisers to seek out potential audiences with incredible specificity” and that the “algorithms consider protected characteristics such as sex, familial status, disability, race, and national origin.” These allegations are consistent with and would be sufficient to meet the modeling defenses articulated in the Proposed Rule at the pleading stage—the first defense would not succeed in defeating the claim on the pleadings, as the complaint alleges that the algorithm expressly considered protected characteristics, and the second defense would not apply, as the complaint alleges that the defendant itself (and not a third party) is responsible for developing and operating the algorithmic model.

As HUD observes, the “third defense is similar to the first.” Proposed Rule at 42,859. As this observation suggests might be the case, we and our members had some difficulty with distinguishing the defense and its application. For purposes of promoting the highest level of clarity, we would request additional clarification and guidance from HUD regarding the application of the third defense in Proposed Rule Section 100.500(c)(2)(iii). As noted above algorithmic models have significant potential to further the goals of fair lending, but expectations regarding model validation of third parties may discourage smaller community lenders from engaging in innovative partnerships, and the third defense may be designed to help address this issue while ensuring that disparate impact claims can still be brought in appropriate circumstances against the party with the ability to “eliminate[e] the offending practice.” Inclusive Communities, 135 S. Ct. 2525.

Finally, as the use of automated models remains an evolving issue that requires continued analysis, we would urge HUD to continue to consult with other federal financial institutions regulators and agencies on the topic. It may be helpful and promote further clarity for regulators to consider issuing joint guidance, including possible updates to the Interagency Fair Lending Examination Guidelines to address discrimination claims challenging automated models under both disparate treatment and disparate impact legal theories.

4.The Proposed Rule Adopts the Correct Burdens of Proof Applicable to Claims Not Resolved at the Pleading Stage.

Proposed Rule Section 100.500(d) (“Burdens of proof for discriminatory effect”) sets out the “burden of proof to establish that a specific, identifiable policy or practice has a discriminatory effect” “if a case is not resolved at the pleading stage.” Because the 2013 Rule improperly rejected the Supreme Court’s Wards Cove standard in favor of Congress’s 1991 standard for Title VII (see supra at 3-4), the 2013 Rule incorrectly assigns the burden and standard of proof that must be applied to disparate impact claims under the Fair Housing Act. The Proposed Rule effectuates the amendments necessary to correct these errors and bring the regulation into conformity with governing law, and we strongly support implementing the Proposed Rule’s articulation of the burdens of proof at the final rule stage.

First, Proposed Rule Section 100.500(d)(1)(i) correctly establishes that the “plaintiff must prove by the preponderance of the evidence, through evidence that is not remote or speculative, each of the elements” of a prima facie case of disparate impact set forth in Proposed Rule Section 100.500(b). If a plaintiff does not meet its burden of proving each element of a prima facie case by a preponderance of the evidence, or if the defendant can “demonstrate that the plaintiff has not proven by a preponderance of the evidence an element” of a prima facie case (see Proposed Rule Section 100.500(d)(2)(ii)), then the matter ends.

Second, the Proposed Rule properly reflects that the plaintiff bears the burden-of-proof at all stages. See Proposed Rule Section 100.500(d)(1)(i) (“plaintiff must prove by the preponderance of the evidence ... each of the elements” of a prima facie case); Proposed Rule Section 100.500(d)(1)(ii) (“plaintiff must prove by a preponderance of the evidence that a less discriminatory policy or practice exists that would serve the defendant’s identified interest in an equally effective manner without imposing materially greater costs on, or creating other material burdens for, the defendant”). Proposed Rule Section 100.500(d)(1)(ii) correctly states that, if a plaintiff satisfies its burden of proving each element of a prima facie case, then the defendant may “rebut” by “producing evidence showing that the challenged policy or practice advances a valid interest (or interests).” The Department’s amendments to the Proposed Rule conform to the Supreme Court’s directive that “the ultimate burden of proving that discrimination against a protected group has been caused by a specific [] practice remains with the plaintiff at all times.” Wards Cove, 490 U.S. at 659 (emphasis in original). Once a plaintiff puts forth sufficient evidence of all elements of a prima facie case of disparate impact, the defendant “carries the burden of producing evidence of a business justification for [the] practice,” and the Proposed Rule tracks the Supreme Court’s language regarding the requirement of “producing evidence.” Id.

Third, under Wards Cove, if the plaintiff puts forth evidence sufficient to prove a prima facie case of disparate impact, the defendant can justify the challenged policy by articulating a legitimate business goal that the policy serves. 490 U.S. at 658-59 (“at the justification stage of ... a disparate-impact case, the dispositive issue is whether a challenged practice serves, in a significant way, the legitimate employment goals of the employer”). Consistent with Wards Cove, Inclusive Communities states that “[a]n important and appropriate means of ensuring that disparate-impact liability is properly limited is to give housing authorities and private developers leeway to state and explain the valid interest served by their policies.” 135 S. Ct. at 2522 (emphasis added). The business interest must be “legitimate,” meaning “valid,” but the Supreme Court expressly disclaimed any requirement that the defendant show that its policy was “essential” or “indispensable.” The Department accurately describes this requirement in the Proposed Rule.

Finally, having articulated a legitimate business goal, the defendant should prevail under Wards Cove unless the plaintiff can prove “that ‘other tests or selection devices, without a similarly undesirable racial effect, would also serve the ... legitimate [business] interest[s]’” in an equally effective manner. Id. at 660 (emphasis added). Under Wards Cove, the plaintiff cannot prevail by merely showing that a less discriminatory alternative could serve the defendant’s business interest. Inclusive Communities reiterates this standard, holding that “before rejecting a business justification ... a court must determine that a plaintiff has shown that there is an available alternative practice that has less disparate impact and serves the entity’s legitimate needs.” 135 S. Ct. at 2518 (internal alternations and quotations omitted.)

The Proposed Rule properly requires that the plaintiff prove that “a less discriminatory policy or practice exists that would serve the defendant’s identified interest in an equally effective manner without imposing materially greater costs on, or creating other material burdens for, the defendant,” (Proposed Rule Section 100.500(d)(1)(ii)), and that the defendant may, “as a complete defense” to liability, “demonstrate that the policy or practice identified by the plaintiff ... would not serve the valid interest identified by the defendant in an equally effective manner without imposing materially greater costs on, or creating other material burdens for, the defendant,” (Proposed Rule Section 100.500(d)(2)(iii)). The standard in the Proposed Rule makes clear that the plaintiff cannot offer a hypothetical alternative about which a defendant may not have any knowledge or capacity for implementing. Once again, this is fully consistent with Inclusive Communities, which holds that a plaintiff must “show[] that there is an available alternative practice that has less disparate impact and serves the [defendant’s] legitimate needs.” 135 S. Ct. at 2518 (internal alternations and quotations omitted) (emphasis added).

C.Specific Suggestions for Modifications to the Language of the Department’s Proposed Rule to Further Increase Clarity

1.Collection of Data on Ethnicity, Race, Sex, and Age

Proposed Rule Section 100.5(d) provides that “[n]othing in this part requires or encourages the collection of data with respect to race, color, religion, sex, handicap, familial status, or national origin.” HUD explains that the goal of this provision is to adhere to the Supreme Court’s “warning against the use of racial quotas. The absence of any such collection efforts shall not result in any adverse inference against any party.” Proposed Rule at 42,857 (citing Inclusive Communities, 135 S. Ct. at 2512). We support the intent of this provision and agree that, under Inclusive Communities, the disparate impact standard must guard against racial quotas. Other laws, however, may require institutions to collect demographic information. For example, institutions subject to the Home Mortgage Disclosure Act are required to collect information about the ethnicity, race, sex, and age of mortgage loan applicants. See 12 C.F.R. § 1003.4(b). Accordingly, to eliminate any possible confusion, we suggest modifying Proposed Rule Section 100.5(d) to account for instances in which collection of information covering prohibited bases is required by law.

2.The Department Should Further Amend the Proposed Rule at the Final Rule Stage to Provide a Definition of Disparate Treatment

While the Proposed Rule provides a precise identification of a disparate impact claim and its required elements, we believe that the regulation would benefit from inclusion of a definition of an intentional discrimination claim. Regulators have long observed confusion regarding the foundational differences between claims of disparate impact (unintentional discrimination) and disparate treatment (intentional discrimination). See DOJ Fair Lending Enforcement Position Statement at 3 (“lenders sometimes believe that neutral practices are having only a disparate impact, when in fact the lender's employees have been applying them differentially, resulting in disparate treatment”). Courts continue to encounter this confusion between claims of disparate impact and disparate treatment. See, e.g., Ellis v. City of Minneapolis, 860 F.3d 1106, 1112 & n.3 (8th Cir. 2017) (applying Inclusive Communities and rejecting argument that different application of a neutral policy could support a disparate impact claim, holding that “[i]nconsistent application of housing standards, of course, may be the basis for a disparate treatment claim under the FHA,” but “an FHA disparate-impact claim may not be used ... merely because housing standards are inconsistently applied”).

As Inclusive Communities observes, in a “disparate-treatment case ... a plaintiff must establish that the defendant had a discriminatory intent or motive.” 135 S. Ct. at 2513. “Proof of discriminatory motive is critical.” Int'l Bhd. of Teamsters v. United States, 431 U.S. 324, 335 n.15 (1977). In some circumstances, discriminatory intent is shown with direct or overt evidence of discrimination, while in other circumstances discriminatory intent can “be inferred from the mere fact of differences in treatment.” Id. This later method of proving discriminatory intent with circumstantial evidence most frequently follows the burden shifting framework established by the Supreme Court in McDonnell Douglas v. Green, 411 U.S. 792 (1973). A plaintiff proceeding under this framework bears a burden, at the prima facie stage, to allege that she was treated differently and less favorably than other persons who are similarly situated in all material respects but for the prohibited characteristic such as race, national origin, gender, religion, disability, or familial status. If that burden is satisfied, the defendant must articulate some legitimate nondiscriminatory reason for its treatment of the plaintiff. The Plaintiff can prevail in the claim by showing that the defendant’s stated reason was in fact a pretext for discrimination. See, e.g., Lewis v. City of Union City, Georgia, 918 F. 3d 1213 (11th Cir. 2019) (en banc).

At the final rule stage, the Proposed Rule would benefit from additional guidance and discussion in the preamble contrasting disparate impact with disparate treatment. One source of confusion concerns policies of allowing the exercise of discretion, such as that rejected by the Eighth Circuit in Ellis. It is true that policies of allowing discretion could be misused to result in discrimination—i.e., by treating consumers differently because of race in exercising discretion. In such instances, however, it is not the policy of allowing discretion that leads to rogue results, but rather it is some act of intentional discrimination. Discretion should never serve as a pretext to discriminate, but allowing discretion cannot be defined as a practice that can be challenged under a disparate impact theory. HUD, at minimum, should amend the Rule to clarify that a plaintiff cannot establish disparate impact liability by challenging the exercise of discretion.

D.Economic Impact of the Proposed Rule

The Proposed Rule provides clarity to all parties regarding the framework for disparate impact claims under the Fair Housing Act—for prospective plaintiffs, the Proposed Rule will promote the ability to ensure that contemplated disparate impact complaints meet all requirements articulated by Supreme Court precedent; and for defendants, the Proposed Rule will promote the ability to defend and rebut such claims. While the 2013 Rule maintains that each situation needs to be decided on a “case-by-case” basis, the Proposed Rule provides much-needed guidance and clarity regarding the practical application of the law—the primary purposes and requirements of agency rules—which will promote heartland uses of the theory and minimize frivolous or unnecessary claims. Perhaps even more importantly for furthering the Fair Housing Act’s purpose of eradicating discriminatory housing practices from our nation, the Proposed Rule will enhance voluntary efforts of all regulated entities to proactively ensure that policies and practices comply with the Act.

Sincerely,
American Bankers Association
Consumer Bankers Association
Housing Policy Council