By Justin Steil (Click here to view entire P&R issue)
The struggle for civil rights in housing policy has been a long and continuing effort. Congress enacted the Civil Rights Act of 1866 over 150 years ago to ensure that all people in the United States shall have the same right “to inherit, purchase, lease, sell, hold, and convey real and personal property” “as is enjoyed by the white citizens” and, just over a century later, it passed the Fair Housing Act, to provide “for fair housing throughout the United States” (42 U.S.C. § 1981(a); 42 U.S.C. § 3601). The fight to realize these guarantees continues today. Two recent cases from the Supreme Court have made important advances in civil rights in housing, affirming the importance of broad standing and of disparate impact liability under the Fair Housing Act, allowing victims of discrimination to challenge policies that have the effect of denying access to housing if there is a less discriminatory alternative policy available (Texas Dep’t of Hous. & Cmty. Affairs v. Inclusive Communities Project, Inc., 2015; Bank of Am. Corp. v. City of Miami, Fla., 2017). Yet, over the past 18 months, the Department of Housing and Urban Development (HUD) has undertaken several deeply problematic moves backward, including an effort to undermine the agency rule that governs these disparate impact claims.
In order to contextualize these developments related to disparate impact in housing, I introduce below the distinction between an anticlassification and antisubordination approach to civil rights, review the positive recent developments from the Supreme Court, present the recent HUD proposal, and then discuss the implications of the antisubordination approach for local governments and local housing policy. The durable racial caste structure of the United States is intertwined with the spatial organization of inequality (Sharkey, 2013). Conceptions of Black threat were central to the making of modern metropolitan areas in the United States, as white homeowners, white landlords, and white real estate brokers sought at times to exclude and at others to exploit nonwhite homeseekers. Americans continue today to live in neighborhoods that are largely separate by race and ethnicity and also highly unequal, for instance in access to high performing schools or exposure to environmental hazards (Ellen and Steil, 2019). These separate and unequal neighborhoods are associated with negative effects on health, education, employment, and a host of other outcomes, including socioeconomic mobility (see Chetty et. al, 2014). The continuing challenge for public policy and urban planning is how to address these spatial dimensions of inequality, entrenched as they are in custom, jurisdictional boundaries, zoning codes, municipal financing structures, subsidized housing policy, and actual housing units and infrastructure.
Antisubordination
Addressing these inequalities requires a sea change in public policy and regional planning to shift from an antidiscrimination approach to an antisubordination approach to civil rights. An anticlassification or antidiscrimination approach prohibits the use of categories based on seemingly arbitrary characteristics and looks to intent to prevent actions that are consciously motivated by racial or other prohibited animus, what is known as a disparate treatment case. With an anticlassification theory, a policy that adversely impacts a disadvantaged group is permissible so long as it does not distinguish explicitly on the basis of a protected characteristic (such as race, sex, religion, or national origin) and so long as no discriminatory intent can be proven (see Village of Arlington Heights v. Metropolitan Housing Development Corporation, 1977). State policies that explicitly distinguish on the basis of a protected characteristic in order to benefit disadvantaged groups are generally prohibited, except to remedy instances where there is convincing evidence of recent intentional discrimination by the governmental unit involved in the policy itself (see Parents Involved in Community Schools v. Seattle School District No. 1, 2007).
An antisubordination or antisubjugation theory of equal protection, by contrast, looks not at intent, but at the collective effects of an action and seeks not just to prohibit arbitrary classifications, but to actually address persistent group disparities in a social system in which some are systematically disadvantaged (Balkin and Siegel, 2003; Fiss, 1976). The antisubordination perspective focuses on identifying and prohibiting practices that perpetuate the subordinate position of a disadvantaged group, defined as a social group, that has been in a position of historical subordination, and whose political power is circumscribed (Fiss, 1976). From the perspective of an antisubordination theory, policies or practices that have a disproportionate adverse impact on a disadvantaged should be prohibited, regardless of intent or classification.
The fundamental premise of an antisubordination approach is that equal citizenship cannot be fulfilled in a context of pervasive social stratification and that true equal protection of the laws requires the reformation of institutions or practices that reproduce subordinate social statuses. In the United States, it is essential to take into account the historically specific circumstances that continue to inscribe durable, categorical inequalities along ascriptive, arbitrary lines that have been given social significance, such as sex, gender, sexuality, race, or ethnicity (Tilly, 2009). The dismantling of these asymmetrical structures of power requires more than just an anticlassification approach to planning—it requires an antisubordination approach.
Inclusive Communities and City of Miami
Two Supreme Court decisions regarding the Fair Housing Act over the past five years open up some room for a renewed focus on antisubordination in housing.
In Texas Department of Housing and Community Affairs v. Inclusive Communities Project, the Court confirmed that the Fair Housing Act prohibits policies or practices that, regardless of intent, have a discriminatory effect on a protected class, if there is a less discriminatory alternative policy available (Texas Dep’t of Hous. & Cmty. Affairs v. Inclusive Communities Project, Inc., 2015). The Court noted with approval that disparate impact liability “permits plaintiffs to counteract unconscious prejudices and disguised animus that escape easy classification as disparate treatment” (Id.). Research in psychology has established the ubiquity of implicit biases and the significant effects they have on members of historically excluded groups (Id.). Courts, by contrast, have been slow to recognize the importance of these unconscious prejudices in perpetuating discrimination and have instead increasingly limited antidiscrimination laws to a narrow focus on aberrant actors who manifest an explicit intent to discriminate. The recognition of the importance of implicit biases in Inclusive Communities creates the potential for renewed attention to the broader antisubordination principle that is the foundation of civil rights statutes such as the Fair Housing Act.
In Bank of America v. City of Miami, the plaintiff, the City of Miami, alleged that Bank of America and Wells Fargo issued riskier mortgages on less favorable terms to African American and Latino borrowers than they issued to similarly situated white borrowers, in violation of the Fair Housing Act. The City of Miami further alleged, among other claims, that this discrimination “impaired the City’s goals to assure racial integration and desegregation,” “frustrate[d] the City’s longstanding and active interest in promoting fair housing and securing the benefits of an integrated community,” and disproportionately “cause[d] foreclosures and vacancies in minority communities in Miami,” decreasing the City’s property tax revenues and forcing the City to spend more on municipal services to address unsafe conditions at properties that were foreclosed because of the banks’ discriminatory lending (Bank of Am. Corp. v. City of Miami, Fla., 2017). The Court confirmed that these injuries fall within the zone of interests that the Fair Housing Act protects. The decision left lower courts to consider the question of whether there was a sufficiently direct relation between Miami’s injuries and the banks’ lending practices.
In both cases, the Court focused on the “results-oriented language” of the Fair Housing Act, recognizing the legal significance of the actual effects that policies have, in addition to the stated intent of the individuals who enacted those policies (Texas Dep’t of Hous. & Cmty. Affairs v. Inclusive Communities Project, Inc., 2015). The Court confirmed that housing providers, lenders, and state and local agencies crafting housing policies should consider potential adverse impacts of their policies by race, color, sex, family status, national origin, religion, or disability.
Efforts to Amend HUD’s Disparate Effects Standard
At the same time that the Supreme Court’s recent decisions have affirmed the importance of disparate impact liability under the Fair Housing Act, however, the current administration is taking steps to eviscerate the disparate impact rule that the Department of Housing and Urban Development issued in 2013.
The 2013 rule requires plaintiffs to meet the initial burden of proving that a challenged practice caused or predictably will cause a discriminatory effect, either a disparate impact on a group of persons protected by the Fair Housing Act or a disparate impact that reinforces, or perpetuates segregated housing patterns because of race, color, religion, sex, handicap, familial status, or national origin (24 C.F.R. § 100.500(a); 24 C.F.R. § 100.500(c) (1)). The burden of proof then shifts to the defendant, who has the burden of proving that the challenged practice is necessary to achieve a substantial, legitimate, nondiscriminatory interest. The plaintiff then bears the burden of proving that the substantial, legitimate, and nondiscriminatory interest could be served by another practice with a less discriminatory effect
Despite the Supreme Court’s citation with approval to HUD’s disparate impact rule and its decision consistent with that rule in Inclusive Communities, HUD in June of 2018 proposed to amend the rule (83 Fed. Reg. 28560), and the formal proposed rule was released this month (84 Fed. Reg. 42854).
First, the proposal will make it virtually impossible for plaintiffs to pass the very first burden of establishing a prima facie case of discrimination. The proposal erects multiple obstacles to making out a prima facie case by creating five new elements plaintiffs must establish: 1) that “the challenged policy is arbitrary, artificial, and unnecessary to achieve a valid interest”; 2) that there is a “robust causal link between the challenged policy or practice and a disparate impact on members of a protected class which shows the specific practice is the direct cause of the discriminatory effect”; (3) that the disparity has an adverse effect on members of a protected class; (4) that the disparity caused is significant; and (5) that there is “a direct link” between the disparate impact and the injury. Although the Inclusive Communities opinion strongly affirms the importance of disparate impact liability, Justice Kennedy in dicta expressed some concerns about possible “abuse” and, quoting the employment case, Griggs v. Duke Power, 401 U.S. 424, 431 (1971), wrote that “[d]isparate impact liability mandates the ‘removal of artificial, arbitrary, and unnecessary barriers,’ not the displacement of valid governmental policies.” HUD seeks to twist these the words in Inclusive Communities to undermine disparate impact liability. Justice Kennedy’s quotation from Griggs implies that disparate impact liability appropriately requires the removal of artificial policies, arbitrary policies, and unnecessary policies that have an adverse effect on members of a protected class. The safeguard in the existing HUD rule requiring a plaintiff to establish a less discriminatory alternative that accomplishes the same legitimate ends already ensures that disparate impact liability can only be used to challenge policies that are either arbitrary, artificial, or that have an unnecessary adverse impact and does not displace valid governmental priorities. Having to establish, as HUD now seeks to require, that a challenged policy is “arbitrary, artificial, and unnecessary” (emphasis added) defeats the purpose of disparate impact liability because the only cases that are likely to be able to meet this are truly disparate “treatment cases,” in which a defendant intentionally created a policy that is arbitrary, and artificial, and unnecessary in order to discriminate. The rule then, would essentially collapse all disparate impact cases into disparate treatment cases, exactly the opposite of what the Court in Inclusive Communities actually held. In Inclusive Communities, which HUD cites as the justification for this rule change, the Court states that disparate impact liability “permits plaintiffs to counteract unconscious prejudices and disguised animus that escape easy classification as disparate treatment” and that “disparate impact liability may prevent segregated housing patterns that might otherwise result from covert and illicit stereotyping” (Texas Dep’t of Hous. & Cmty. Affairs v. Inclusive Communities Project, Inc., 2015). As the Court suggests, unconscious prejudices escape easy classification as disparate treatment exactly because they are embedded in policies that appear reasonable and necessary. When those seemingly reasonable, necessary policies advancing valid interests have a disparate impact on the basis of race and there is a less discriminatory alternative that could accomplish the same valid interests, however, the Fair Housing Act requires that those policies be struck down. This proposed additional element would frustrate the purposes of the Fair Housing Act that the rule is supposed to implement. Further, the vague and ambiguous additional requirements of pleading a “robust causal link” and establishing that the challenged practice is the “direct cause” of the disparity, as well as the requirement that there be “a direct link” between the disparity and the injury, both threaten to raise the requirements necessary to meet a prima facie case so much that many or most victims of actual discrimination will be unable at the filing stage to pass these tests.
Second, the proposal seeks to place many forms of algorithmic bias and discrimination above the law. Where a victim of discrimination alleges that the cause of the discrimination is “a model used by the defendant, such as a risk assessment algorithm,” and the defendant “[p]rovides the material factors which make up the inputs used in the challenged model and shows that these factors do not rely in any material part of on factors which are substitutes for or close proxies for protected classes under the Fair Housing Act,” the defendant is protected and the victim is without recourse. If that defense is unavailing for the defendant, the defendant can alternatively show that the model is either produced, or maintained, or distributed by a third party “that determines industry standards” and again evade responsibility. If both of those defenses fail, the defendant can instead show “that the model has been subjected to critical review and has been validated by an objective and unbiased neutral third party.”
These capacious defenses against disparate impact claims essentially creates safe harbors for discrimination for those using algorithms to set prices for housing-related products, including homes, home loans, home insurance, and more. The proposed defenses are also contrary to the holding in Inclusive Communities and the text and intent of the Fair Housing Act. Including in algorithm variables to identify protected class members and adversely impact them is already prohibited under a disparate treatment analysis, and close proxies are arguably already prohibited by disparate treatment as well. As above, HUD is again creating limitations that limit disparate impact cases to those that could be brought under a disparate treatment theory, which again would be inconsistent with the holding of Inclusive Communities. Creating a defense for a policy or practice simply because it is an “industry standard” is inconsistent with the decision in Inclusive Communities and would surely have permitted many of the white supremacist industry standards that the Fair Housing Act was enacted to prohibit, such as denying home loans because of the race of the borrower or the racial composition of the borrower’s neighborhood. Creating defenses specifically for the increasingly common use of algorithms or models is highly likely to calcify the “unconscious prejudice” that the Court sought to dislodge in Inclusive Communities (Texas Dep’t of Hous. & Cmty. Affairs v. Inclusive Communities Project, Inc., 2015). It is widely acknowledged that using unrepresentative seed data for a machine learning model will lead to biased results, yet HUD seeks to give that kind of systemic bias a free pass. Disparate impact liability is so important exactly because systemic discrimination, whether created by algorithmic bias or not, is almost impossible to challenge without an analysis of the differential effects by sex or disability or race or other protected characteristics of a policy or a model. Without disparate impact analyses, substantial discrimination, by being written in computer code would suddenly be exempt from the legal code. HUD’s effort to effectively use revisions to the disparate impact rule in order to overturn the Supreme Court’s decision in Inclusive Communities would undermine crucial protections in the Fair Housing Act that are essential for creating a minimum of equal access to housing, to credit, and to insurance.
Groups such as the NAACP Legal Defense Fund, the National Fair Housing Alliance, the National Community Reinvestment Coalition, the Poverty & Race Research Action Council, and others have all written letters opposing this dangerous rule change. As the proposed rule changes have now moved from review by the Office of Information and Regulatory Affairs to the proposed rulemaking stage, widespread public comment and mobilization in opposition to the undermining of this essential rule is necessary.
While the federal government continues to try to erode our basic civil rights protections, however, some cities are enacting local policies that operationalize what an antisubordination approach to policymaking could look like on the ground.
Municipal Racial Equity Initiatives
Groups such as the Center for Social Inclusion and the Government Alliance on Race and Equity have made significant efforts in recent years to push public agencies towards a focus on equity, consistent with an antisubordination approach. These collaborations have influenced a growing number of urban policies. For instance, the City of Seattle, Washington created a Race and Social Justice Initiative in 2005 to bring together city department heads to create and track annual work plans on advancing racial equity in Seattle’s policies, programs, and budget decisions. Seattle’s Race and Social Justice Initiative describes itself as a “citywide effort to end institutionalized racism and race-based disparities in City government” (City of Seattle, 2015). The City of Seattle now requires all city departments to conduct a racial equity analysis of every budget request and to post online annual work plans to advance racial justice so that the public can view goals and progress by city department and by neighborhood
One outcome of the Race and Social Justice Initiative was the Seattle 2035 Growth and Equity report, analyzing impacts on both displacement and opportunity related to Seattle’s growth strategy and development policies. The Growth and Equity Report asked (1) whether the intensity of development in neighborhoods across the city would be likely to have an impact on the displacement of marginalized populations; (2) whether the intensity of development would be likely to affect marginalized populations’ access to key determinants of physical, social, and economic well-being; and (3) what strategies and investments can mitigate the impacts of development and maximize equitable outcomes (Seattle 2035). Seattle’s Office of Planning and Community Development created a Displacement Risk Index and an Access to Opportunity Index to analyze how different alternative strategies for updating Seattle’s Comprehensive Plan would affect groups in the city that had historically been marginalized and used that analysis to shape the ultimate Comprehensive Plan to guide future development. Other cities, such as Minneapolis, MN; Madison, WI; and Portland, OR have subsequently established their own racial equity initiatives. In New York City, several City Council members and the Public Advocate are advocating for bills that would require environmental impact statements to include analyses of how proposed development or land-use changes would affect the population by race and ethnicity and assess whether the proposed action would affirmatively further fair housing.
These efforts to incorporate attention to racial equity into municipal policy are important steps towards an antisubordination approach to urban planning and policy. Such an approach should include community-engaged planning that foregrounds a focus on disparities in access to opportunity across durable categories of inequality, that recognizes institutionalized asymmetries of power and unconscious biases that perpetuate those disparities, and that prioritizes policies that reduce inequality even if they may have higher costs than policies that exacerbate it (Steil, 2018).
Justin Steil is an Associate Professor of Law and Urban Planning at MIT, and a member of PRRAC’s Board of Directors.
