The Latest Decision on Disparate Impact Puts Insurers Between a Federal Rock and a State-Level Hard Place
Reprinted with permission from Law360.
By Robert Helfand August 31, 2017, 12:45 PM EDT
Two years ago, in Texas Dept. of Housing & Community Affairs v. Inclusive Communities Project Inc., 135 S.Ct. 2507 (2015), the U.S. Supreme Court ruled that defendants may be liable under the federal Fair Housing Act for actions that are otherwise lawful, and which have no discriminatory intent, but which nonetheless have a disproportionately adverse effect (a “disparate impact”) on the housing rights of certain disadvantaged communities. Although the case clarified housing law in one respect, the questions of whether, and to what extent, disparate impact liability may be extended to insurers has continued to be litigated. Last week, in National Fair Housing Alliance v. Travelers Indemn. Co., No. 1:16-cv-00928-JDB (D.D.C. Aug. 21, 2017), a district court in Washington, D.C., became the third federal court to rule not only that insurers may be sued for disparate impact, but that they may be liable under the FHA for the disparate impact of someone else’s conduct.
The new decision is important for two reasons. It is the first case to rule that a disparate-impact-by-proxy theory can survive the “more stringent pleading standard for disparate-impact claims” that was announced in Inclusive Communities; it therefore calls the strength of that standard into question. Additionally, it threatens to put insurers in the middle of a crossfire with state regulators, whose recent pronouncements — declaring, for example, a hard line against the use of “any non-risk-based attributes” in either rating or underwriting — appear to prohibit the very deliberations that disparate impact liability would require.
Background: Insurers and the FHA
The FHA, 42 U.S.C. §§ 3601 et seq., prohibits housing discrimination on the basis of race, color, religion, sex, familial status, disability or national origin. The U.S. Department of Housing and Urban Development has long taken the position that the statute applies to insurers (see 24 C.F.R. § 100.70(d)(4)), and, in deference to that position, most courts have agreed. E.g., NAACP v. American Family Mut. Insurance Co., 978 F.2d 287 (7th Cir. 1992). But see Mackey v. Nationwide Insurance, 724 F.2d 419 (4th Cir. 1984).
Despite this consensus, the question of insurer liability was revived in 2013. While Inclusive Communities was making its way through the courts, and the broader issue of disparate impact liability had yet to be resolved, HUD proposed a “discriminatory effects rule” that codified its position for all participants in the housing industry. (78 Fed. Reg. 11482.)
The rule prescribes a burden-shifting framework for disparate impact litigation, in which the plaintiff has the initial burden of pleading and proving facts which show either that the defendant “caused … a discriminatory effect,” or that it “predictably … will cause” such an effect at some point in the future. If the plaintiff makes either showing, then the defendant must show “that the challenged practice is necessary to achieve one or more [of the defendant’s] substantial, legitimate, nondiscriminatory interests.” Even if the defendant carries that burden, the plaintiff can still prevail, by showing that the defendant’s interests “could be served by another practice that has a less discriminatory effect.”
The last part of the framework is critical. To avoid liability, defendants must be prepared to prove that they have chosen to pursue their business goals in the manner that has the least discriminatory impact. As a practical matter, therefore, the rule imposes a duty on prospective defendants to assess the impact on protected classes of every course of action they consider.
HUD ultimately adopted the rule (24 CFR § 100.500), but only over the strong objections of the insurance industry. Two trade groups responded with lawsuits, asserting that the rule should not apply to insurers. In American Insurance Association v. U.S. Department of Housing and Urban Development, 74 F.Supp.3d 30 (D.D.C. 2014), one court overturned the rule, on the ground that the FHA does not create disparate impact liability under any circumstances. After the Supreme Court ruled in Inclusive Communities, that decision was vacated, and the matter is now being litigated on other grounds. In Property Casualty Insurers Assoc. of Am. v. Donovan, 66 F.Supp.3d 1018 (N.D. Ill. 2014), the court held that HUD had not adequately addressed the industry’s arguments, and it remanded the rule for further consideration.
Risk Discrimination vs. Race Discrimination
One of those arguments is that disparate impact liability would penalize basic aspects of insurance practice, because insurers generally must avoid taking account of any factors other than loss and expense when they sell their products. An insurance rate may not be “unfairly discriminatory,” but, according to the National Association of Insurance Commissioners, that means that “price differentials” must “reflect equitably the differences in expected losses and expenses.” (NAIC, Property and Casualty Model Rating Law, § 5.A(3).)
A duty to take account of demographic differences is incompatible with this requirement. As one actuary has explained:
It is reasonable to assume a priori that no protected minority class … will be uniformly distributed throughout any given insurance risk classification plan. This assumption implies that all risk factors used to measure and assess risk are potentially in violation of a disparate impact rate standard, even though each risk factor accurately reflects expected losses and expenses. M.J. Miller, “Disparate Impact and Unfairly Discriminatory Rates,” Casualty Actuarial Society E-Forum 276, 277 (2009). Thus, the Property Casualty Insurers Association (PCI) argued that HUD’s rule conflicts with state rating laws — and, for that reason, that it violates the McCarran-Ferguson Act, 15 U.S.C. § 1012.
HUD published a new response to the insurers’ arguments in October 2016. (81 Fed. Reg. 69012.) It asserted that “nothing in the Rule prohibits insurers from making decisions that are in fact risk-based,” because “practices that an insurer can prove are risk based, and for which no less discriminatory alternative exists, will not give rise to discriminatory effects liability” (emphasis added). HUD also rejected a proposal that the rule carve out “safe harbors” for rating plans using risk-based characteristics that have been “historically allowed by state insurance regulators.” HUD contended that any such exemptions would be “overbroad, arbitrary and quickly outdated.”
HUD’s view, in other words, is that the possible existence of a less discriminatory alternative is a valid basis for attacking any rating characteristic, no matter how well-established under state law. That view would change existing insurance law, which, as a rule, does not require insurers to choose between rating options on the basis of their respective discriminatory effects.
The PCI’s litigation against the discriminatory effects rule is still pending. In the meantime, courts are treating HUD’s burden-shifting framework as authoritative — at least in the noninsurance context. E.g., Mhany Management Inc. v. City of Nassau, 819 F.3d 581, 618 (2d Cir. 2016).
The National Fair Housing Case
The National Fair Housing Alliance (NFHA) is a public-interest organization, dedicated to ending discrimination in housing. In 2015 and 2016, NFHA sent “testers” to the offices of five different insurance brokers that represent multiple carriers, including Travelers. The testers told the agents they were considering the purchase of a four- or five-unit apartment complex in Washington, D.C., and that they wanted to find out what it would take to insure the property. They also stated that tenants in the complex receive assistance under Section 8 of the Housing Act of 1937 (42 U.S.C. § 1437f). In various ways, the brokers allegedly reported to the testers that Travelers would not insure buildings whose tenants receive housing assistance. According to NFHA, they advised that the testers would have to buy nonstandard policies that might cost $1,000 more.
Citing these statements, the NFHA alleged that the insurer had a policy of refusing to sell insurance to landlords with subsidized tenants, and that this policy “makes it more difficult and expensive for landlords to obtain property insurance if they have such tenants.” (At this stage of the litigation, it is impossible to determine whether Travelers had such a policy, and, if so, what the nature of the policy might have been. Travelers has represented that it does not currently consider the level or source of tenants’ income when underwriting any commercial lines residential property insurance. It also asserts that its current policy was in place when at least some of the NFHA’s testers were making their inquiries.)
The NFHA further alleged that the alleged difficulty of obtaining insurance might “compel” landlords “to evict or reject … tenants” who participate in the Section 8 program. In Washington, D.C., 92 percent of households that receive Section 8 vouchers are African-American, and 81.5 percent are headed by women. Consequently, the NFHA alleged that the insurer’s policy is having, or will have, a disproportionate, adverse effect on the housing rights of African Americans and women-headed households.
The NFHA’s complaint purported to assert claims under both the FHA and the D.C. Human Rights Act, D.C. Code § 2-1402.21. Like many state housing laws, the latter statute expressly prohibits any refusal to provide property insurance for a “discriminatory reason,” which might involve any of the demographic categories listed in the FHA, but which also might be based on “source of income.” Thus, unlike the federal claim, NFHA’s claim under D.C. law is based on disparate treatment — a direct act of prohibited discrimination — and not on disparate impact.
Disparate Impact By Proxy
The NFHA’s claim under the FHA contains an unusual combination of elements. Like other disparate impact claims, it asserts that the defendant insurer “predictably … will cause” African-American tenants to lose their housing, because of actions that are not otherwise discriminatory or unlawful. But it asserts that the insurer will cause those losses only at one remove: by causing someone else to make choices with a disproportionate effect on protected classes. In the past, a few courts have accepted FHA claims based on allegations that the insurers penalized landlords who rented to protected groups. In Nevels v. Western World Insurance Co., 359 F. Supp. 2d 1110 (W.D. Wash. 2004), and in Wai v. Allstate Insurance Co., 75 F. Supp. 2d 1 (D.D.C. 1999), the defendant insurers refused to insure properties with disabled tenants. But Nevels and Wai were both disparate treatment cases: because the act of discriminating on the basis of disability is expressly prohibited by the FHA (see 42 U.S.C.§ 3604(f)), the plaintiffs (landlords who complained about the increased cost of insurance) were not obligated to prove a causal connection between the insurers’ conduct and any impairment of housing rights.
That distinction is important, because the majority opinion in Inclusive Communities emphasized the necessity of firmly establishing precisely that causal relationship. Justice Anthony Kennedy, the author of the opinion, found that the FHA has a “robust causality requirement”:
[A] disparate-impact claim that relies on a statistical disparity must fail if the plaintiff cannot point to a defendant’s policy … causing that disparity. … A robust causality requirement ensures that ‘[r]acial imbalance … does not, without more, establish a prima facie case …’ and thus protects defendants from being held liable for racial disparities they did not create. …
This “causality requirement,” the court went on to say, might defeat a disparate impact claim, where the defendant’s act is only one of several causes contributing to a denial of housing access, or where the defendant’s choices are constrained by statute or regulation:
It may … be difficult to establish causation [in a disparate impact case] because of the multiple factors that go into decisions [affecting the housing market] … . And … if the [plaintiff] cannot show a causal connection between the [defendant’s] policy and a disparate impact — for instance, because [governing] law substantially limits the [defendant’s] discretion — that should result in dismissal … .
Finally, the majority opinion in Inclusive Communities emphasized that the causality requirement applies to the plaintiff’s complaint. Courts must “examine with care whether a plaintiff has made out a prima facie case of disparate impact[,] and prompt resolution of these cases is important.” Thus, in National Fair Housing, the district court agreed with the defendant that Inclusive Communities had created a “more stringent pleading standard for disparate-impact cases.” In 2015, two other federal courts denied dispositive motions in disparate impact cases based on insurers’ alleged refusal to rent to landlords who accept Section 8 tenants. Viens v. Am. Empire Surplus Lines Insurance Co., 113 F. Supp. 3d 555 (D. Conn. 2015); Jones v. Travelers Cas. Insurance Co. of Am., No. CV 5:13-02390 (N.D. Cal. May 7, 2015). But both cases were decided before Inclusive Communities. Neither applied the standard that was announced in that case.
A “More Stringent” Standard?
The court in National Fair Housing accepted the premise that the U.S. Supreme Court, in Texas Dept. of Housing & Community Affairs v. Inclusive Communities Project Inc., 135 S.Ct. 2507 (2015), has imposed a “more stringent pleading standard” for disparate impact claims. The court also added, however, that this standard “does not require courts to abandon common sense or necessary logical inferences that follow from the facts alleged.” According to the court, the common-sense, necessary inferences in this case run along the following lines: If an individual cannot obtain insurance, she might lose access to credit with which to buy a property (“lack of insurance … makes housing unavailable”). Even if she succeeds in acquiring the property, the buyer without insurance “must risk losing” it to fire or other loss. Either way, that person would have “powerful incentives” to avoid renting to tenants who make insurance unavailable. “Hence, insurance policies can create exactly the type of ‘artificial, arbitrary, and unnecessary barriers’ to housing that disparate-impact liability is suited to address.”
There are several gaps in this chain of reasoning. Most importantly, the court found that Travelers had created a “lack of insurance” so severe as to “compel” landlords to turn away Section 8 tenants, but it failed to consider whether NFHA’s allegations actually support that finding. NFHA did not allege that any of its testers had tried and failed to obtain insurance. It asserted only that they were told their premiums might be $1,000 higher than the cost of a policy from the one carrier (Travelers) that was the target of their investigation. The court’s opinion did not address the question of whether that allegation supports a plausible inference that Travelers had made it impossible, or even difficult, for landlords to obtain property coverage at an acceptable price.
Moreover, even if it were the case that Travelers increased the cost of Section 8 housing singlehandedly, the court also failed to consider whether that “incentive” was really “powerful” enough to persuade landlords to abandon Section 8 tenants. Under Washington’s Human Rights Act, housing discrimination on the basis of “source of income” is expressly prohibited, and victims may bring actions for both compensatory damages and civil penalties. D.C. Code §§ 2-1403.13 and 2-1403.16. The court found, in essence, that a possible $1,000 annual increase in the cost of insurance “predictably … will cause” a significant number of landlords to expose themselves to administrative proceedings and civil actions under D.C. law — but the court made no effort to justify that finding.
Finally, the court dismissed the relevance of the fact that landlords’ participation in the Section 8 program is voluntary under federal law. Two courts have held that landlords who refuse to participate may not, under any circumstances, be subject to disparate impact claims under the FHA. Salute v. Stratford Greens Gardens Apts., 136 F.3d 293 (2d Cir. 1998); Knapp v. Eagle Prop. Mgmt. Corp., 54 F.3d 1271 (7th Cir. 1995). A third has speculated that a disparate impact theory might possibly be used against a landlord who withdraws from the program, but held not that is not available against one who has never participated at all. Graoch Assocs. #33, L.P. v. Louisville / Jefferson City Metro Human Relations Comm’n, 508 F.3d 366 (6th Cir. 2007).
Consequently, the court in National Fair Housing implicitly endorsed an interpretation of the FHA that provides no remedy for families whose landlord chooses to avoid Section 8 tenants for independent business reasons, but which does create a remedy where the landlord is motivated by a desire to save money on insurance. That interpretation would be odd under any circumstances. It is particularly hard to defend in light of Justice Anthony Kennedy’s admonition that it is “difficult to establish causation” where “multiple factors” go into a decision about housing policy or practice, because the law “protects defendants from being held liable for racial disparities they did not create.”
For all these reasons, it is significant that the court’s reasoning expressly followed two cases from the 1990s — NAACP v. American Family Mut. Insurance Co., 978 F.2d 287 (7th Cir. 1992), and Wai v. Allstate Insurance Co., 75 F. Supp. 2d 1 (D.D.C. 1999). The court found that the arguments in those cases “apply with full force to Travelers.” NAACP and Wai, however, were both disparate treatment cases. In NAACP, the defendant’s “redlining” allegedly prevented the plaintiff from obtaining homeowners’ insurance; in Wai, the plaintiffs were landlords who complained that discrimination against the disabled constrained their own choices about their properties. In neither case did the court have to rule on whether there was a plausible causation claim, based on nonparties’ intervening acts. By adopting the arguments from those cases, the court in National Fair Housing effectively eliminated the “more stringent” and “robust” causality requirement it purported to apply.
The National Fair Housing case involved an underwriting decision, rather than a rating decision, but the scope of the U.S. Department of Housing and Urban Development’s rule is not limited in that way. In fact, the HUD’s most recent statement argues that its rule applies to all phases of insurance operations, including “marketing and claims processing and payment.” Nothing now prevents a plaintiff from asserting that a particular rating classification (for example, crime rates at the insured property’s location) will impose disproportionately higher rates on minority homeowners. Under the HUD’s rule, the defendant insurer might have to prove to a jury that no less discriminatory alternative to this classification has an equal level of actuarial validity. In light of that fact, it might be prudent for property insurers to incorporate demographic considerations into their decision-making. But it’s unclear that state insurance laws will permit them to do so — especially in light of the recent furor over “price optimization.”
“Price optimization” is a term used to describe certain kinds of modifications to insurance rates. Although the classifications and relativities that make up a rating plan must be risk-based, insurers often adjust rates for individual classes to achieve business goals — for example, by capping an indicated price hike, in an effort to retain existing customers. In recent years, some insurers have based those adjustments on sophisticated predictive models, which consider (among other things) data about the specific consumers who make up the insured population — including data about their “price elasticity of demand.” It is the use of those models that is most often labeled “price optimization.”
Over the last three years, more than 20 state insurance departments have issued bulletins or otherwise spoken out about price optimization. Many of those bulletins use a broad argument to justify the principle of restricting this practice: They contend that current law prohibits consideration of any factor other than risk or cost in connection with any aspect of a rating decision. For example:
While insurers may employ judgment in setting their rates, judgmental adjustments … may not be based on non-risk-related factors … . (Rhode Island Bulletin No. 2015-8; Vermont Bulletin No. 186.)
[P]ractices that adjust premiums, whether included or not included in the insurer’s rating plan, are not allowed when the practice cannot be shown to be cost-based. (Connecticut Bulletin PC-81.)
To the extent price optimization involves gathering and analyzing data related to … characteristics … unrelated to risk of loss or expense, insurers may not use price optimization to rate policies … . (Delaware, Bulletin No. 78.)
Earlier this year, Nevada’s Department of Business and Industry adopted the same position, declaring that models which are used to change a base rate or relativity “may not utilize anynon-risk-based attributes.” (Nevada, Bulletin No. 17-001; emphasis in original.) But Nevada’s bulletin goes farther than the others, in that it appears to extend its prohibition beyond rating factors and into underwriting decisions, such as company placement or eligibility. The bulletin announced that “any mathematical model used in underwriting or rating of any personal line of property and/or casualty insurance” must now “be filed … for prior approval.” Examples of models that must be submitted for approval include not only “‘price optimization’ models,” but also any models that place insureds in different tiers or with particular writing companies, and any scoring models that affect rates or eligibility.
Any model that uses a mathematical algorithm to calculate a score or index for eligibility purposes, and that is capable of being used for rating, is … considered a rating model[,] since the decision to reject a risk based on a score … is considered to be a more extreme variant of a decision to surcharge that riskbased on the same score … (emphasis added).
These pronouncements appear to be on a collision course with housing law — at least to the extent that the latter now requires insurers to choose the least discriminatory alternative among possible business strategies. The HUD’s rule, as applied in National Fair Housing, provides that consideration of potential discriminatory impact must be decisive when a company chooses between two rating or underwriting choices — even where neither choice involves a prohibited classification. Yet Nevada’s bulletin (for example) suggests that information about discriminatory impact must be excluded from any mathematical model which assists the insurer’s decision.
Insurance regulators are not indifferent to the concerns that animate disparate impact theories. They proved that in their reactions to the use of credit scores for automobile insurance. But state officials have, for the most part, identified those concerns as issues of availability and affordability — problems of public policy that need to be addressed by statute or regulation. They generally have not tried to cure social inequalities by making insurers liable in damages for “discrimination” they did not intend.
For the moment, then, it appears that insurers must try to adhere to two incompatible visions of how to address inequality.
Robert D. Helfand is a member at Pullman & Comley LLC's Hartford, Connecticut, office. The opinions expressed are those of the author(s) and do not necessarily reflect the views of the firm, its clients, or Portfolio Media Inc., or any of its or their respective affiliates. This article is for general information purposes and is not intended to be and should not be taken as legal advice.