This is the final article of a three-part series about two recent decisions by federal courts in Connecticut and California: Viens v. America Empire Surplus Lines Ins. Co., No. 3:14cv952 (D. Conn. June 23, 2015), and Jones v. Travelers Cas. Ins. Co. of Am., No. CV 5:13-02390 (N.D. Cal. May 7, 2015). The first article discussed the legal and procedural background of the two cases. The second article discussed the issues these cases raise about the intended scope of the federal Fair Housing Act (“FHA”) and the state laws that are based on it. This article will discuss issues these cases raise about the application of federal law to underwriting decisions, and it will suggest steps that prudent insurers should now be taking.
What’s Wrong With Disparate Impact Claims Against Insurers?
As discussed in part 2, Viens and Jones raise a number of questions about what kinds of activity fair housing statutes are intended to reach—in particular, whether they were meant to impose a duty on insurers to predict and act upon the effects their decisions will have on the intervening acts of others. But the two cases also present the issue of whether the type of liability asserted in Viens and Jones is compatible with the very nature of insurance underwriting.
Risk Discrimination is Not Race Discrimination
It is an axiom of the insurance industry—and of insurance law—that insurance rates may not be “unfairly discriminatory.” E.g., Alaska Stat. §§ 21.36.090, 21.36.120; D.C. Stat. § 31-2231.13. This actuarial concept of “unfair discrimination” focuses on costs: according to the Casualty Actuarial Society’s “Principles Regarding Property and Casualty Insurance Ratemaking“:
A rate is … not … unfairly discriminatory if it is an actuarially sound estimate of the expected value of all future costs associated with an individual risk transfer.
In some states, laws against “unfair discrimination” by property insurers go further, forbidding insurers to distinguish among customers on the basis of factors that have been associated with racially discriminatory “redlining”—for example, the location of a property in a certain neighborhood. But even statutes of that kind make allowances for geographic determinations that result from “sound underwriting and actuarial principles related to actual or reasonably anticipated loss experience.” Colorado R.S.A. § 10-3-1104(f)(XIV).
Consequently, it is possible for insurers to make underwriting decisions that disfavor classes protected by civil rights laws, without engaging in “unfair discrimination” in the actuarial sense. Some years ago, in N.A.A.C.P. v. Am. Family Mut. Ins. Co., 978 F.2d 287, 290 (7th Cir. 1992) (“NAACP“), Judge Easterbrook made precisely this point with the aphorism, “Risk discrimination is not race discrimination.”
The disconnect between these two approaches to defining discrimination is especially pronounced when racial “discrimination” is defined to include (in the words of the U.S. Department of Housing and Urban Development(“HUD”) “actions that have a [racially] discriminatory effect but not a discriminatory intent.” 78 Fed. Reg. 11482. InAmerican Ins. Assoc. v. U.S. Dep’t of Housing & Urban Dev., No. 13-00966 (D.D.C. Nov. 7, 2014) (“AIA“), an industry group’s challenge to HUD’s Discriminatory Effects Rule, the plaintiffs (and, subsequently, the court) cited a 2009 article by one actuary, which argued that there is an “inevitable and irreconcilable conflict between the two standards.” Michael J. Miller, Disparate Impact and Unfairly Discriminatory Rates, Casualty Actuarial Society E-Forum 276, 277 (2009).
The article explained the conflict in this way (id.):
If applied to insurance, a risk/rate factor will potentially be said to have a disparate impact if it more adversely impacts a protected minority class than it does the majority class, regardless of its relationship to underlying costs.
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.
[But if] a risk classification plan were changed to eliminate one or more risk factors found to have a disparate impact, the resulting rates would likely be unfairly discriminatory because the rate differences would no longer be based on the underlying insurance costs.
The Case for Disparate Impact
The groups that advocate holding insurers responsible for disparate impact in underwriting decisions do not generally challenge analyses like Mr. Miller’s. Instead, they answer it by pointing out that underwriting practices incorporate many decisions which are not based on the “expected value of all future costs.” They argue that makingthose decisions the target of disparate impact litigation will not impair the legitimate, risk-based practices of the insurance industry.
Filed rates, after all, may deviate from strict actuarial models on the basis of insurers’ business goals or judgments—such as a desire to penetrate a new market or change the company’s mix of business; predictions about future legal or regulatory developments; or judgments about the credibility of the risk experience of individual insureds. (Decision-making about goals and judgments of this kind lie at the heart of the current debate over so-called “price optimization.”)
Furthermore, individual underwriters sometimes exercise broad discretion in applying the rate structure to real-life circumstances. In a second challenge to the Discriminatory Effects Rule, Property Casualty Insurers Assoc. of Am. v. Donovan, No. 1:13-cv-08654 (N.D. Ill. Sept. 3, 2014) (“PCIA“), amici curiae supporting the rule cited an author who contended that “half or more of the underwriting decisions may be ultimately made . . . by human underwriters,” rather than in strict accordance with the filed rates. Donald Light, “Transforming Underwriting: From Risk Selection to Portfolio Management,” at 7 (Celent, March 2004).
Consumer groups also cite the distinction between filed rates and underwriting guidelines—the standards provided to agents and brokers, which they use to make an initial decision about whether (and at what price) to offer coverage. (These offers are typically subject to review by the company’s underwriters before the risk is accepted.) In most states, underwriting guidelines are not publicly filed, and some of them include rules that are not part of the filed rate.
For example: the Texas Office of Public Insurance Counsel conducted a survey of underwriting guidelines in 2007.Seventy percent of participating homeowners insurers reported making adverse underwriting decisions “based on whether the property shows ‘Pride of Ownership.'” According to the survey, that kind of “pride” is established by
vegetation that is well manicured, watered and cared for , with any dead vegetation removed. Premises free of any debris, clutter, disabled or unusable vehicles, disabled or unusable appliances, discarded lumber or scattered trash.
Critics say criteria like these make it possible for the biases of individual agents and underwriters to enter into the process of selling insurance—and to do so in ways that negatively affect protected classes without any actuarial justification.
A Test Case: Jones v. Travelers
Between these two poles is the reality that many of the non-actuarial decisions involved in pricing, marketing and selling insurance are both objectively reasonable and integrally connected to the risk-based calculations of the insurance business. The evidence developed in Jones v. Travelers helps illustrate this point.
The policy at issue in Jones was a product called “Apartment Pac,” which Travelers describes as a “high-volume, low risk, simplified insurance product” that is sold “with minimal involvement of Travelers’ underwriters.” According to the insurer, the relatively lower price of the product reflects its “highly automated, … ‘low touch’ nature”—one aspect of which is that agents are given authority to bind policies for exposures that satisfy underwriting guidelines, without consulting underwriting personnel, and without a physical inspection of the premises.
The underwriting guidelines exclude properties that the company considers specialized risks—risks that are “not within Travelers’ core knowledge and experience.” “Subsidized, public or government-funded [apartment] complexes” are excluded, along with correctional facilities and halfway houses. But the excluded risks also include properties that are not clearly associated with protected classes—such as waterfront complexes with marinas, resort communities and certain buildings over six stories tall.
The exclusion of these risks is not based on actuarial assessments per se. But Travelers produced evidence that insurance for subsidized housing (and for other specialized risks) is generally provided by surplus lines insurers and brokers, and that these companies emphasize the special expertise this market segment requires. Travelers showed, in other words, that it had an objective, reasonable basis to believe that the “Apartment Pac” guidelines allow it toavoid underwriting complexities that are unique to individual market segments, and whose analysis would raise the overall cost of this particular product. In the company’s words, “the fact that a specialized market exists for insurance for subsidized housing is evidence that the risk itself is unique.”
The plaintiffs, on the other hand, asserted that they have been targeted in Travelers’s underwriting guidelines on the basis of “stereotyping or assumptions about landlords and their tenants”—such as a belief, reported in social science research, that subsidized tenants “don’t take care of their property.” Travelers described the “Apartment Pac” product as “geared toward well-run, well-maintained properties,” but a deposition witness testified that propertieswithout subsidized tenants enjoyed a “presumption” of being “well-run.”
Ruling that the issue could go to a jury, the court in Jones focused on the narrow question of the sufficiency of the evidence: it found that the jury could reasonably conclude that subjective biases had crept into the business decisions about whether “Apartment Pac” should cover buildings with subsidized tenants. The court did not discuss more basic questions, such as: Can an insurer establish the “necessity” of its course of conduct without a full-blown, actuarial assessment? Does the FHA limit an insurer’s discretion over whether to enter a market segment in which it does not currently operate?
It is significant, in this regard, that the plaintiffs in Jones had originally obtained their “Apartment Pac” policy by mistake—because they failed to disclose facts that would have disqualified them under the guidelines. The basis for the disparate impact claim against Travelers, therefore, was not that landlords would stop accepting Section 8 participants to avoid losing their policies, but that they would do so to take advantage of a low-cost product that was previously unavailable.
If, like Travelers, an insurer decides to direct a product only to non-specialty risks, does the FHA impose a duty both (i) to review demographic information about the specialty markets, so as to anticipate a possible disparate impact; and (ii) to conduct an actuarial analysis of those markets, to confirm that the exclusion is justified? If so, it might be impossible to create and sell a “low touch” product of the kind Travelers offers. And what if, once the insurer conducts such an analysis, the data suggest that it might safely enter the market after all? Under the FHA, does the company then have an obligation to do so?
Jones was settled before the court could address those questions in a written opinion (and before the case got to a jury). But the court’s preliminary ruling shows that cases of this type—cases in which an insurer allegedly causes its customers to “discriminate” through disparate impact—could seriously burden the conduct of basic functions of the insurance business.
What About McCarran-Ferguson?
These theoretical questions about how deeply the FHA can reach into insurer’s business decisions inevitably fuel disputes involving the McCarran-Ferguson Act, 15 U.S.C. § 1012.
Under McCarran-Ferguson, a state law that “regulat[es] the business of insurance” will “reverse preempt” any federal statute that conflicts with it. More than 30 years ago, the U.S. Court of Appeals for the Second Circuit found, with respect to McCarran-Ferguson:
Congress … had no intention of declaring that subsequently enacted civil rights legislation would be inapplicable to any and all of the activities of an insurance company that can be classified as ‘the business of insurance.’
Spirit v. Teachers Ins. & Annuity Ass’n, 691 F.2d 1054, 1065 (2d Cir. 1982). Following this reasoning, a majority of courts has held that McCarran-Ferguson does not provide insurers with a generalized exemption from federal antidiscrimination laws, including the FHA. E.g., Dehoyos v. Allstate, 345 F.3d 290 (5th Cir. 2003); Moore v. Liberty National Life Insurance Co., 267 F.3d 1209, 1222-23 (11th Cir. 2001); Nationwide Mut. Ins. Co. v. Cisneros, 52 F.3d 1351 (6th Cir. 1995); Mackey v. Nationwide Ins., 724 F.2d 419 (4th Cir. 1984).
Nevertheless, the Supreme Court has never addressed that question, and there remains authority to the contrary. E.g.,Saunders v. Farmers Ins. Exch., 537 F.3d 961, 967-68 (8th Cir. 2008).
Reverse Preemption for Actuarial Judgments
The case for preemption is strongest where resolution of claims under the federal law would require courts to make judgments that are usually made by state insurance regulators. In NAACP, supra, the U.S. Court of Appeals for the Seventh Circuit rejected a reverse preemption argument against a claim under the FHA. But several years later, in Doe v. Mut. Of Omaha Ins. Co., 179 F.3d 557 (7th Cir. 1999), the same court held that the plaintiffs could not assert a claim against their insurer under the Americans with Disabilities Act, because (id. at 564):
[I]f federal courts are now to determine whether … [policy terms] are actuarially sound and consistent with principles of state law they will be stepping on the toes of state insurance commissioners.
In Doe, the plaintiffs challenged provisions in health insurance policies that capped lifetime benefits for AIDS or AIDS-related conditions at levels that were below the limits for other conditions. For tactical reasons, the insurer had stipulated that it could not show the caps to be “consistent with sound actuarial principles, actual or reasonably anticipated experience, bona fide risk classification, or state law.” Judge Posner nevertheless held that preemption was proper:
[McCarran-Ferguson does not permit] federal courts to determine whether limitations on coverage are actuarially sound and consistent with state law. Even if the formal criteria are the same under federal and state law, displacing their administration into federal court—requiring a federal court to decide whether an insurance policy is consistent with state law—obviously would interfere with the administration of state law. …
It is true that we are not being asked in this case to decide whether the AIDS caps were actuarially sound … . But if the McCarran-Ferguson Act does not apply, then we are certain to be called upon to decide such issues in the next case …. .
Courts have also applied reverse preemption where a federal civil rights law would impose liability for an underwriting practice that state law clearly permits. E.g., Ojo v. Farmers Group, Inc., 356 S.W.3d 421 (Texas 2012). On the other hand, courts reject McCarran-Ferguson arguments in cases, such as Jones, where they conclude that state law “complements” the FHA. See also Nevels v. Western World Ins. Co., Inc., 359 F.Supp.2d 1110, 1122 (W.D. Wash. 2004).
The reasoning of these cases seems particularly relevant to disparate impact claims, because, under HUD’s Discriminatory Effects Rule, insurers defending those claims must show “that the challenged practice is necessary to achieve … legitimate … interests,” 24 CFR § 100.500(c)(2), and that showing will often be based on actuarial judgments. The court in AIA, supra, struck down HUD’s rule, based on its belief that no disparate impact claims may be maintained under the FHA. (The Supreme Court later reached the opposite conclusions in Texas Dept. of Housing v. Inclusive Communities, No. 13–1371 (U.S. June 25, 2015).) In dicta, the court also opined that disparate impact claims against insurers would inevitably run afoul of McCarran-Ferguson:
[E]xpansion of the FHA to include disparate-impact liability [against insurers] would … have a wide-ranging disruptive effect on the pricing and provision of homeowners’ insurance … . Indeed, recognition of disparate-impact liability under the FHA … raises serious concerns regarding widespread federal encroachment upon state insurance regulation.
Essentially the same McCarran-Ferguson argument was rejected in Viens—but the court’s discussion of the issue focused on whether state law preempts civil rights statutes in general, rather than on the features of plaintiffs’ particular claim. (In a footnote, the court did find an “indication” in the CFHA that the FHA is “complementary” to state law.) Thus, it is likely that courts will continue to be divided on this issue. In PCIA (the second industry challenge to HUD’s Discriminatory Effects Rule), the court held that the plaintiffs’ McCarran-Ferguson arguments were not yet ripe for adjudication, and that the issues should be resolved on a case-by-case basis. It also identified some considerations that will distinguish future cases:
While some states require insurers to use risk-based pricing, other states merely permit [it] …. Accordingly, some insurance practices in some states may rest on business justifications rather than actuarially sound principles or state law requirements. Under [Doe, supra], the McCarran-Ferguson Act would not necessarily preclude claims based on these practices because it is not clear that such claims would raise the question of whether the insurer’s practices are actuarially sound and consistent with state law.
As a result of the ruling in PCIA, HUD is currently reviewing the McCarran-Ferguson arguments against its Discriminatory Effects Rule. Even if HUD sticks to its guns, those arguments are likely to be litigated for at least the next several years.
Some of Justice Kennedy’s language in Inclusive Communities might end up making it harder for plaintiffs to succeed in disparate impact claims against insurers—especially claims, like Viens and Jones, based on third-party transactions. But Inclusive Communities is still a shot in the arm for disparate impact litigation in general, and the issues presented by third-party cases are too novel and complex to be resolved definitively any time soon. Insurers should expect cases like Viens and Jones to be around for years.
That means insurers should be reviewing their underwriting guidelines now. The point is not necessarily to conduct a race- or gender-based analysis (which might be unlawful in many cases), but to ensure that the policies and choices which complement purely actuarial analyses reflect business judgments that are clear, rational and defensible.
Vague terms connoting value judgments—such as “pride of ownership” or “clutter”—should be eliminated, because plaintiffs can attack those terms (sometimes with justice) as coded invitations to unlawful discrimination. For the same reason, agents’ discretion to interpret the operative language should be limited, and insurers should consider offering literature or training about how to apply guidelines in a non-discriminatory way.
Insurers should also apply an understanding of antidiscrimination laws to the way they use consumer data—especially data from outside sources—in marketing and pricing decisions. The FHA prohibits discrimination on the basis of race, color, religion, sex, familial status, national origin or disability. State laws further bar distinctions based on such characteristics as source of income, sexual orientation, gender expression and genetic information. Cal. Gov. Code § 12955. If insurance decisions are based on data points that can be characterized as surrogates for any of these prohibited categories, those decisions might expose the company to complex civil rights claims.
By building a record that shows a sound “business justification” for their policies, insurers can prepare to defend future cases on the facts. That record will also improve the chances that some of the legal arguments discussed in this article will succeed.