In February 2013, the U.S. Department of Housing and Urban Development adopted a "Discriminatory Effects Rule," which established liability under the Fair Housing Act for conduct that is otherwise lawful, but which has a disproportionately adverse effect on racial minorities or other protected classes. 24 CFR § 100.500. Insofar as it applied to insurers, this rule was promptly challenged in two lawsuits brought by trade groups. In one, Property Casualty Insurers Assoc. of Am. v. Donovan, 66 F.Supp.3d 1018 (N.D. Ill. 2014) (the "PCI Suit"), a federal court in Illinois held that the agency had given inadequate consideration to the arguments of the insurance industry, and it remanded the rule to HUD.
Now, two years later, HUD has responded with a lengthy and aggressive rejection of all the insurers' objections. The response concludes that there is no insurance practice—not even the use of "long-recognized" risk factors in pricing—that could not, under some circumstances, form the basis for a valid disparate impact claim. It finds that insurers are exposed to liability under virtually every section of the Fair Housing Act—including provisions that some courts have previously declined to apply. And it endorses application of the disparate impact theory to marketing and claims handling practices; to non-actuarial aspects of underwriting; and to claims that insurers will cause other parties to commit lawful acts with a disparate impact.
The response comes at a time when consumer advocates and government agencies are increasingly turning to disparate impact theories, under a variety of laws, to assail "price optimization," credit scoring and predictive modelling tools that rely on "big data." Thus, in addition to potentially broadening insurers' liability under the FHA, HUD's statement encourages potential litigants to test a broad range of theories under a wide array of civil rights and consumer-protection legislation.
The Discriminatory Effects Rule
The Fair Housing Act, 42 U.S.C.A. §§ 3601 et seq., prohibits several forms of discrimination, based on "race, color, religion, sex, familial status ... national origin" or disability, in connection with the sale or rental of a "dwelling." Individuals claiming past or future injuries may seek damages or injunctive relief, either in court or before the Department of Housing and Urban Development.
Intentionally discriminatory acts— "disparate treatment"—are expressly prohibited. Before last year, most federal courts also entertained claims based on "disparate impact." These claims are based on the consequences of otherwise lawful acts that make no prohibited distinctions, and which may have a wholly innocent motive, but which nonetheless produce a "disproportionately adverse effect" on the housing rights of protected classes.
In February 2013, HUD issued a rule "to prohibit housing practices with a discriminatory effect." The rule found that the FHA can impose liability on the basis of disparate impact, and it created a burden-shifting framework for litigating disparate impact claims. Last June, the U.S. Supreme Court cited parts of the rule with approval in Texas Department of Housing and Community Affairs v. Inclusive Communities Project, 135 S. Ct. 2507 (2015). Inclusive Communities definitively established the validity of disparate impact theories under the FHA.
HUD's latest statement describes the Discriminatory Effects Rule this way:
In fact, HUD's rule is broader. For one thing, it governs conduct that "caused or predictably will cause a discriminatory effect." 24 CFR § 100.500(c). That is, it can force insurers to make "an adjustment" to practices that have never caused a "discriminatory effect" of any kind. Furthermore (as discussed below), the rule can impose liability for effects suffered not only by "insureds," but also by parties with whom the defendant insurer has no relationship at all.
In litigation, under the rule's burden-shifting framework, a plaintiff's showing of either an actual or a "predictable" discriminatory effect places the burden on the defendant to demonstrate that its conduct is "necessary to achieve ... substantial, legitimate, nondiscriminatory interests." Id. Even where the defendant satisfies that burden, plaintiffs can still succeed, if they can show that the defendant's interests could "be served by another practice ... [with] less discriminatory effect." Id.
In short, defendants need evidence both that their actions had a valid, nondiscriminatory business objective, and that they pursued that objective in a way that is more efficacious than any alternative with less discriminatory consequences. In effect, therefore, the rule imposes a duty to calculate in advance the likely results that nondiscriminatory practices, as well as any possible alternatives, will have on the housing rights of protected classes.
Insurers Challenge the Rule
Most of the relatively small number of lawsuits against insurers under the Fair Housing Act have alleged disparate treatment—intentional discrimination—such as "redlining." E.g., NAACP v. American Family Mutual Insurance Co., 978 F.2d 287 (7th Cir. 1992). But when HUD first announced its proposed Discriminatory Effects Rule in November 2011, it cited "the provision and pricing of homeowner's insurance" as "[e]xamples of a housing practice that may have a disparate impact."
Three insurance trade associations—the Property Casualty Insurers Association of America, the American Insurance Association and the National Association of Mutual Insurance Companies—submitted comments on the proposed rule, arguing (among other things) that it should not be applied to insurance in general, and to underwriting in particular. When the rule was adopted over these objections in February 2013, both PCI and AIA filed lawsuits. [Full disclosure: the author represents PCI, AIA and NAMIC, but has not done so in connection with the matters discussed in this article.]
In the PCI Suit, the court held, in September 2014, that HUD's enactment of the rule had been "arbitrary and capricious," because the agency had failed to give adequate consideration to several of the commenters' arguments. The court also faulted HUD for failing to weigh in on whether the commenters' concerns might be better addressed through rule-making, rather than in piecemeal litigation. The case was remanded to HUD for consideration of those issues.
Two months later, in American Insurance Association v. U.S. Department of Housing and Urban Development, 74 F.Supp.3d 30 (D.D.C. 2014) (the "AIA Suit"), a federal court in Washington, DC, predicted that the Supreme Court would disapprove all disparate-impact claims under the FHA. That ruling was later superseded by the decision in Inclusive Communities. The court's award of summary judgment was reversed, but the lawsuit is still pending.
HUD Responds
Following the court's decision in the PCI Suit, HUD did not invite additional submissions or conduct other public proceedings. Instead, after two years, on October 5, 2016, it simply published a new set of responses to the issues the court had directed it to consider.
Those responses betray little sympathy for insurers as a class. They repeatedly accuse the industry of having a "long, documented history of discrimination," and they imply that this "history" is not yet over—asserting, for example, that creating certain exemptions to the rule "would inevitably insulate insurers engaged in ... unlawful discriminatory practices." (HUD supports its historical claims by citing Congressional testimony, all of which was delivered more than twenty years ago. Ironically, some of this testimony actually failed to persuade Congress of a need for anti-redlining legislation in the early 1990s.)
On the specific matters HUD was ordered to address, the agency also gave no quarter.
McCarran-Ferguson and Risk-Based Pricing
The insurance groups that commented on the Discriminatory Effects Rule all contended that it should exempt insurance practices—or, at least, underwriting practices—from liability for disparate impacts on housing. Failing that, the commenters suggested that the rule should at least create "safe harbors" for certain established practices, such as the use of "long-recognized" risk factors in pricing. In support of these proposals, the commenters made two closely-related arguments. (A more detailed review of the arguments is available here.)
First, they observed that insurers are legally required to make pricing decisions on the basis of risk classifications—predictions about how different pricing decisions will affect their loss experience. The analysis required by the Discriminatory Effects Rule—predictions about how decisions might affect the housing rights of different demographic groups—is incompatible with traditional underwriting. As one authority cited by AIA observed,
Michael J. Miller, Disparate Impact and Unfairly Discriminatory Rates, Casualty Actuarial Society E-Forum 276, 277 (2009).
This argument has even greater force in states where insurers are prohibited from collecting or using data about race. Judge Easterbrook once summed up this point with an aphorism: "Risk discrimination is not race discrimination." N.A.A.C.P. v. Am. Family Mut. Ins. Co., 978 F.2d at 290.
Second, under the McCarran-Ferguson Act, 15 U.S.C.A. § 1012, a state law that "regulat[es] the business of insurance" will "reverse preempt" any federal statute that conflicts with it. Some years ago, in Doe v. Mut. Of Omaha Ins. Co., 179 F.3d 557, 563-64 (7th Cir. 1999), Judge Posner found that a federal law will run afoul of state regulations, if it requires a court to decide the same issues that are addressed by state regulators—such as whether an insurer's risk-based pricing decisions "are actuarially sound":
Because the PCI Suit is pending in the Northern District of Illinois, Judge Posner's view has particular salience for HUD's deliberations.
HUD's Response
In connection with both arguments, HUD now asserts that the insurers' concerns should be resolved through case-by-case adjudication of lawsuits challenging specific practices, rather than by any comprehensive rule that HUD could devise.
According to HUD, "nothing in the Rule prohibits insurers from making decisions that are in fact risk-based," because
On the other hand, even ratemaking can incorporate an element of non-actuarial judgment based on business goals, and so
Moreover, HUD asserts, many aspects of insurance practice—such as marketing and claims processing—are not risk-based at all. Thus, broad exemptions "would immunize a host of potentially discriminatory insurance practices that do not involve actuarial or risk-based calculations."
With respect to McCarran-Ferguson, HUD observes that the issue of reverse preemption depends on "a host of case-specific variables." It contends that a uniform rule would fail to take account of the fact that insurance laws and insurance practices vary widely—both between states and over the course of time. It notes, for example, that courts in different jurisdictions have reached different conclusions about whether McCarran-Ferguson bars disparate impact claims that challenge credit scoring under the FHA. Consequently, HUD concludes that "safe harbors" or other broad exemptions to the Discriminatory Effects Rule for specific underwriting practices would necessarily be "overbroad":
Is HUD Right?
What HUD's analysis fails to consider is the practicality of asking insurers to "prove" to juries and other fact-finders that their practices are risk-based (i.e., actuarially justified), and that no less discriminatory, "risk-based" alternative exists. How could an insurer establish those facts (or, alternatively, how could a plaintiff establish that the insurer's underwriting policies are "not purely risk-based") without asking the jury or the court "to determine whether" the practices and the proposed alternatives "are actuarially sound and consistent with principles of state law"—in other words, without "stepping on the toes of state insurance commissioners," in violation of McCarran-Ferguson? Doe v. Mut. Of Omaha, 179 F.3d at 564.
Indeed, HUD specifically cited state regulators' recent responses to "price optimization" as evidence that states are apt to change their rules in response to new practices—and so that there should be no broad exemption based on McCarran-Ferguson. But the first two lawsuits challenging price optimization under state law were both recently dismissed under the "primary jurisdiction" doctrine—based on findings that the courts could not determine what the insurers' actual practices were without invading the province of a state insurance department:
Harris v. Farmers Ins. Exch., No. BC579498 (Cal. Super. Ct. Jan. 25, 2016). See also Stevenson v. Allstate Ins. Co., No. 15-cv-04788 (N.D. Cal. March 17, 2016). (Because these cases were brought under state law, McCarran-Ferguson was not an issue.)
The practicality of the litigation solution HUD advocates is also affected by another matter that its responses entirely failed to address: the way in which its rule has been limited by Justice Kennedy's majority opinion in Inclusive Communities. That opinion stated that the FHA forbids discriminatory effects that are "artificial" and "arbitrary," and it admonished courts to "avoid interpreting disparate-impact liability to be so expansive as to inject racial considerations into every housing decision." To prevent that result, the opinion stressed that disparate impact claims are subject to a "robust causality requirement":
The Court suggested several ways defendants might successfully challenge claims about causation—for example, by showing that the defendant's choices merely contributed to a situation that can be traced to more than one cause, or that defendant's choices are constrained by statute or regulation:
In the AIA Suit, the trade group has recently filed a motion for summary judgment, arguing that these aspects of Inclusive Communities foreclose any disparate impact claims against insurers. (As of this writing, the motion is still pending.) Even if that argument were to fail, however, it would still be appropriate to consider whether, in light of these limitations, HUD can credibly maintain that plaintiffs will "inevitably" prevail in all the cases that might be eliminated by "a one-size-fits-all exemption."
After all, by asserting that any exemption would be "overbroad," HUD is arguing that there is no insurance practice which could not, under the right circumstances, form the basis for a valid disparate impact claim. But, given the constraints of McCarran-Ferguson, and given, as well, the "robust causality requirement" imposed by Inclusive Communities, that argument probably goes too far; it is possible to imagine at least some cases that will have virtually no chance of succeeding. How plausible is it, for example, that a plaintiff might be able to prove (or even be allowed to prove) that there are equally efficacious, actuarially sound, and less discriminatory "alternatives" to rating factors which a state regulator has been approving for decades? How reasonable would it be to place that decision in the hands of a state-court jury?
HUD's response does not address these questions—in part, because HUD has not accepted any comments on the Discriminatory Effects Rule since 2013, long before Inclusive Communities was decided. But the issues are likely to surface if the rule comes before the court hearing the PCI Suit for a second time.More Issues: Expanding Liability Theories
HUD's arguments based on those aspects of insurance practice which are not "risk-based" raise additional concerns.
First, HUD used its response to express the view that "discriminatory insurance practices can violate ... [Sections] 3604(a), (b), (c), (f)(1), (f)(2), 3605 and 3617" of the FHA.
The FHA declares that several different actions constitute "discriminatory housing practices": not only "mak[ing] [a dwelling] unavailable" to a person because the person is a member of a protected class (42 U.S.C. § 3604(a)), and "discriminat[ing] ... in the provision of services ... in connection [with]" the "sale or rental of a dwelling" (§ 3604(b)), but also "discriminat[ing] against any person in making available" a "real estate-related transaction[]," or "in the terms or conditions of such a transaction" (§ 3605(a)).
There is general consensus that discrimination in the sale of property insurance can violate § 3604. See, e.g., N.A.A.C.P. v. American Family Mut. Ins. Co., supra; Fuller v. Teachers Ins. Co., No. 5:06-CV-00438-F (E.D.N.C., Sept. 19, 2007). But cases recognizing claims against insurers under § 3605 or § 3617 are extremely rare. See National Fair Housing Alliance v. Prudential Ins. Co., 208 F.Supp.2d 46, 57, 58 (D.D.C. 2002) (§ 3605); Nevels v. Western World Ins. Co., Inc., 359 F.Supp.2d 1110, 1122 (W.D. Wash. 2004) (§ 3617). Some cases have expressly found that the sale of property insurance is not a "real estate-related transaction," and so that insurers may not be found liable under Section 3605. E.g., N.A.A.C.P. v. American Family Mut. Ins. Co., 978 F.2d at 297.
Thus, HUD's responses take the broadest possible view of how the FHA might be applied to insurance practices.
Second, HUD's responses expressly endorse the theory that disparate impact liability might apply to "claims processing." To support that idea, HUD cites two cases in which insurers were accused of intentionally denying individual claims for property damage because of the race of the claimants. Franklin v. Allstate Corp., No. C-06-1909 (N.D. Cal. July 3, 2007); Burrell v. State Farm, 226 F.Supp.2d 427 (S.D.N.Y. 2002). The agency did not try to postulate a scenario in which a claims handling practice might predictably have a disparate impact on the housing rights of insureds.
Other critics of the insurance industry have been less circumspect. For example, this past August, at a hearing of the Big Data (D) Working Group of the National Association of Insurance Commissioners, Consumer Advocate Birny Birnbaum testified that predictive models, of the kind insurers are now using for fraud detection and claim benchmarking (among other uses), "may reflect historic discrimination because of biased data, biased assumptions or faulty model specifications," and may therefore have a disparate impact on historically disadvantaged communities. In the future, systems that rely on such models might be the basis for claims that minority insureds are more likely to be investigated for fraud, are subjected to delays in claim adjudication (either because of fraud investigations or for other reasons), or are relegated by benchmarking programs to less experienced claims personnel.
HUD's discussion of its own Discriminatory Effects Rule does not address those potential claims. But HUD's maximalist interpretation of the scope of disparate impact liability under the FHA puts the agency in line with this growing trend.
Finally, recent cases have found that insurers may be liable under the FHA, not only for "discriminatory effect[s] on [their] insureds," but for "predictably" causing their insureds to act in a way that has a disparate impact on other parties—parties who are strangers to the insurance contract. Viens v. America Empire Surplus Lines Ins. Co., 113 F.Supp.3d 555 (D. Conn. 2015); Jones v. Travelers Cas. Ins. Co. of Am., No. C-13-02390 (N.D. Cal. May 7, 2015). (Another, similar case, National Fair Housing Alliance v. Travelers Indemnity Co., No. 1:16-cv-00928 (D.D.C.), is currently pending.)
Both Viens and Jones involved insurers that declined to underwrite properties with tenants who receive housing assistance from the federal government's "Section 8" program. Both cases were settled after the courts denied dispositive motions, and both left serious questions unanswered. For example, Viens involved a property in Connecticut, where state law (unlike the FHA) requires landlords to refrain from "source of income" discrimination. The court found that the insurer's policy "create[d] an incentive for [policyholders] to not comply with this requirement." But the court did not suggest that the goals of the FHA can actually be advanced by imposing liability for nondiscriminatory conduct that "causes" another to commit an act for which the law already provides a remedy.
Jones involved a property in California, where courts have held that landlords may choose to reject Section 8 tenants without violating the FHA. E.g., Doe v. WCP I, LLC, No. 05-438885 (Cal. Super. Ct. June 4, 2009). Under the ruling in Jones, therefore, the FHA would give the tenants no remedy if the landlords acted on their own initiative, but it would provide a remedy, if the landlords' decisions had been "caused" by their insurer. (A detailed discussion of these cases can be found here.)
Both Viens and Jones were decided before the Supreme Court, in Inclusive Communities, announced a "robust causality requirement" for disparate impact claims. Nevertheless, because both cases are cited with approval in HUD's response to the PCI Suit, it appears that HUD wishes to throw its support behind this extended theory of disparate impact liability.
Conclusion
As a practical matter, HUD's response does little more in the short term than preserve the possibility of lawsuits based on practices that could have been protected by reasonable "safe harbor" provisions. But the response is only one of several calls to hold insurers responsible for alleged discriminatory effects. It will be surprising if multiple disparate impact theories are not tested in lawsuits in the next few years.
The content of this article is intended to provide a general guide to the subject matter. Specialist advice should be sought about your specific circumstances.
