July 13, 2021
Housing

Urging the Biden Administration to Address Technology’s Role in Housing Discrimination

Harlan Yu, Aaron Rieke, and Natasha Duarte

Letter
Table of Contents keyboard_arrow_down

Upturn, ACLU, The Leadership Conference on Civil and Human Rights, and a coalition of other organizations urge the White House Office of Science & Technology Policy to fully incorporate the Biden administration’s commitment to racial equity into its AI and technology priorities.

This is our memo on technology's role in housing discrimination. We sent this memo to:

  • Secretary Marcia L. Fudge, U.S. Department of Housing and Urban Development (HUD)

  • Acting Director Dave Uejio, Consumer Financial Protection Bureau (CFPB)

  • Chair Lina Khan, Federal Trade Commission (FTC)

  • Assistant Attorney General Kristen Clarke, Civil Rights Division, U.S. Department of Justice (DOJ)

  • Chief Sameena Shina Majeed, Housing & Civil Enforcement Section, U.S. Department of Justice

Upturn was joined by the following signatories and contributors to this letter:

  • American Civil Liberties Union

  • Center for Democracy & Technology

  • Center on Privacy & Technology at Georgetown Law

  • Lawyers’ Committee for Civil Rights Under Law

  • National Consumer Law Center (on behalf of its low-income clients)

  • National Fair Housing Alliance

This is one of three memos on technology and discrimination that we sent to the Biden administration — see our letter to the White House Office of Science & Technology Policy; and our other two memos on hiring and financial services discrimination.


RE: Addressing Technology’s Role in Housing Discrimination

Too often, technology amplifies and exacerbates racial, gender, disability, economic, and intersectional inequity in our society. Governments and corporations, at the national, state, and local level, are using computer software, statistical models, assessment instruments, and other tools to make important decisions in areas such as employment, health, credit, housing, immigration, and the criminal legal system. In light of these developments, policymakers must take steps to ensure non-discriminatory and equitable outcomes in housing.

We offer the following proposals for the Biden-Harris administration for addressing the role of technology in perpetuating discrimination in housing. We urge all agencies to engage with a diverse range of stakeholders, including impacted communities, tenant organizers and housing advocates, reentry advocates, direct services providers (including housing court advocates and those assisting with finding housing and housing subsidies), and civil rights organizations, in order to receive ongoing input and feedback on these important issues. We also encourage agencies to prioritize transparency, both by sharing their data, models, decisions, and proposed solutions so that all of the stakeholders can stay apprised of and comment on the potential impact of proposed actions, and by encouraging users of housing-related technologies to share with the public as much information as possible regarding their technologies and assessments of those technologies. Additionally, we encourage agencies to actively monitor and audit the use of housing-related technologies, and to take investigation and enforcement actions to ensure compliance with the Fair Housing Act and civil rights laws. We’d be pleased to discuss these ideas further in the weeks and months ahead.

A. HUD and its partner agencies should proactively monitor and investigate housing technologies for discriminatory effects.

Housing discrimination has evolved along with new and expanding uses of technology, such as machine learning-based lending models, algorithms for determining eligibility for and allocating housing services, online advertising, and tenant screening scores. It is often difficult or impossible for impacted people to learn about housing discrimination in these systems, and HUD cannot rely on individual complaints alone for its fair housing enforcement. HUD should work with its partner organizations and other stakeholders to develop new testing methods to uncover discrimination in digital systems. HUD and its partners and funding recipients should publicly report the methodologies, including the specific types of data used in testing, and results of their testing.

HUD should also issue guidance recommending that covered entities design, test, and audit their models to prevent discriminatory effects and affirmatively further fair housing. These practices should include regularly auditing models for discriminatory effects throughout their conception, design, implementation, and use; proactively looking for and adopting less discriminatory alternatives; assessing whether data used in training technologies is representative and accurate, and that the technologies measure lawful and meaningful attributes and seek to predict valid target outcomes; retaining and documenting in-depth information about the technology, its development and internal auditing sufficient to allow for third party auditing; and publicly releasing internal and external audit reports. Audits should be designed to assess impacts on all protected classes — including disability, which is the basis for the largest percentage of fair housing complaints — as well as intersectional discrimination.

B. HUD should update its Fair Housing Advertising Guidelines to provide examples of online advertising practices that can violate the Fair Housing Act.

Advertising platforms conduct ad auctions that determine the actual audience to which an ad will be delivered. These auctions can result in advertisements for economic opportunities being systematically delivered to audiences with significant skews on the basis of race, gender, or other protected group. This can happen even when the advertiser intentionally targets its ad toward a broad audience, such as an entire geographic region. HUD should clarify that advertising platforms can face liability for violating the Fair Housing Act when their conduct in delivering advertisements — including the design and function of their ad delivery algorithms — disproportionately steers housing opportunities away from protected groups.

2. Help local jurisdictions evaluate and affirmatively further fair housing.

A. When reinstating the Affirmatively Furthering Fair Housing (AFFH) Rule, HUD should provide guidance to program participants on addressing the impacts of technology on fair housing.

Examining the design, use, and outcomes of housing-related technologies must be part of a holistic assessment of fair housing. These technologies include tenant screening systems, risk assessment models, and housing services allocation models, which embed criteria for determining who has access to housing, as well as surveillance technologies used to monitor tenants, often in low-income housing.

HUD should promulgate AFFH guidance clarifying steps local jurisdictions should take to address the fair housing impacts of systems such as tenant screening, surveillance of tenants, and assessments for determining risk or distributing housing and homeless services. These systems fall under the “contributing factors” that HUD identified in its 2015 AFFH guidance. For example, HUD should advise jurisdictions to affirmatively further fair housing by strictly limiting the criteria and information housing providers can use to screen tenants, shielding court records that housing providers unjustifiably use to deny housing (including criminal, landlord-tenant, and other civil court records), and eliminating digital surveillance of tenants.

B. HUD should continue to update and expand its AFFH datasets to help program participants evaluate access and barriers to fair housing.

To supplement the data HUD already provides for fair housing assessments, HUD should partner with local organizations and agencies to make more local data available to its program participants. This data should include demographics and patterns of segregation; housing data such as property characteristics and affordability; zoning and other land use data; environmental burdens and hazards; disability access and services; and data on access to services and opportunities such as education, employment, social services, childcare, transportation, and healthcare. HUD should broadly share datasets provided by local jurisdictions.

3. End discriminatory background screening as a barrier to housing.

A. HUD should expand its 2016 guidance on the use of criminal records in tenant screening under the Fair Housing Act.

HUD’s 2016 guidance acknowledged that criminal record screening tends to have a disparate impact on Black and Latinx renters, and clarified that blanket rejections of potential tenants based on the existence of a criminal record are not justifiable under the Fair Housing Act. HUD’s guidance states that “[a] housing provider must [] be able to prove through reliable evidence that its policy or practice of making housing decisions based on criminal history actually assists in protecting resident safety and/or property,” and that a tenant screening policy that fails to consider the nature, severity, and recency of the alleged conduct is unlikely to satisfy this standard. However, criminal records remain a formidable barrier to housing, and several jurisdictions have “crime-free housing” ordinances requiring housing providers to conduct criminal background checks and reject or evict tenants for alleged criminal conduct.

HUD should revise its 2016 guidance to further clarify the steps housing providers must take to ensure that any criminal record screening is necessary and narrowly tailored to serve a substantial, legitimate, nondiscriminatory interest. In revising its guidance, HUD should look to the growing number of states and cities that have adopted fair chance housing policies limiting housing providers’ ability to request and consider criminal records. Fair chance housing laws often prohibit housing providers from inquiring into an applicant’s criminal history before extending a conditional offer of housing, limit the types of conviction records housing providers can consider, and require the housing provider to consider several factors in evaluating the applicant’s record, such as the applicant’s age at the time of the alleged offense, whether the alleged offense arose from an applicant’s disability, and the degree to which the alleged offense, if it re-occurred, would impact the safety of other tenants. These laws also require housing providers to consider other mitigating factors, such as recommendations from community members or participation in education, employment, or other programming. 

B. HUD should extend its 2016 guidance to cover the use of eviction records, credit reports, and other unjustifiably discriminatory records to screen applicants for rental housing.

Landlord-tenant court records and credit reports also represent pervasive barriers to housing and reflect systemic patterns of discrimination. For example, housing providers evict Black tenants, and Black women in particular, at staggeringly disproportionate rates. The vast majority of eviction records do not represent a proven allegation of failure to pay rent or other lease violation. About three million evictions are filed in the U.S. each year, but few end in a judgment in favor of the landlord at trial. Cases often end in dismissals for lack of good cause to evict the tenant, or agreements in which the tenant stays and continues paying rent. Eviction filings also commonly result in default judgments because tenants lack the notice and/or resources to defend themselves. Thus, much like arrest records, the vast majority of eviction records are not reliable indicators of tenant suitability or risk. Yet, landlords often use the existence of any eviction record, including simple filings, as a reason to disqualify potential tenants.

HUD should clarify that any screening policy or practice based on eviction records, other civil court records, or credit reports must be demonstrably necessary and narrowly tailored to achieve a substantial, legitimate, nondiscriminatory interest. In revising its guidance, HUD should consider the growing number of jurisdictions that have adopted limits on the consideration of eviction records and credit reports in tenant screening. For example, some policies prohibit housing providers from taking adverse action based on an eviction record that did not result in a judgment in favor of the plaintiff at trial; prohibit the denial of housing based on a credit score; and/or prohibit the consideration of credit history for applicants using rental assistance, such as vouchers.

C. The FTC and CFPB should publish updated guidance for tenant screening companies on complying with the Fair Credit Reporting Act (FCRA).

The guidance should elaborate on practices that constitute failure to reasonably assure accuracy under the FCRA, including but not limited to:

  • Failing to verify the accuracy of public records matched to an applicant;

  • Using unreliable or overly loose matching criteria, such as name-only searches, to match applicants with public records;

  • Furnishing court records that do not include an accurate date and disposition (or incorporating such records into a score or recommendation); and

  • Reporting sealed or expunged records.

D. The FTC should conduct a 6(b) investigation into the tenant screening industry to study how companies obtain, match, and report on information such as criminal records, eviction and other civil court records, and credit and financial information.

The study should also investigate how companies compile tenant screening reports and scores. The Commission should seek to provide guidance on tenant screening practices that may be unfair or deceptive under Section 5 or that constitute unlawful discrimination. HUD, the FTC, and the CFPB should collaborate to ensure that the findings of this study support and are incorporated into HUD’s guidance and enforcement activities.

4. Proactively enforce against discriminatory uses of technology and records in housing decisions.

HUD should develop systems to ensure full and consistent compliance with, and implementation of, the recommendations set forth in this memorandum by its nationally administered programs in the Office of Housing (including the Federal Housing Administration), the Office of Public and Indian Housing, and other programs under HUD’s authority. HUD should review guidance and other rules and policies issued by these offices and programs to ensure that they reflect a meaningful commitment to protecting against discrimination through background screening, online advertising practices, and other housing-related technologies. 

The DOJ Housing and Enforcement Section should also use its authority to intiate investigations and file suit where evidence of discrimination resulting from use of housing-related technologies is uncovered. Similarly, this should include situations where there is evidence of unjustifiable use of criminal records, eviction filings and other civil court records, credit reports and other criteria that may have a discriminatory impact.

Thank you for your attention to these matters. For any questions or further discussion, please contact Natasha Duarte, Senior Policy Analyst, Upturn, at 202-677-2359 or natasha@upturn.org.

Download


1

See, e.g., Jennifer Miller, Is an Algorithm Less Racist than a Loan Officer?, N.Y. Times, Sept. 18, 2020, https://www.nytimes.com/2020/09/18/business/digital-mortgages.html?searchResultPosition=1.

1

See, e.g., Catriona Wilkey et al., Coordinated Entry Systems: Racial Equity Analysis of Assessment Data, Oct. 2019, https://c4innovates.com/wp-content/uploads/2019/10/CES_Racial_Equity-Analysis_Oct112019.pdf.

1

See, e.g., Charge of Discrimination, HUD v. Facebook, https://www.hud.gov/sites/dfiles/Main/documents/HUD_v_Facebook.pdf; Muhammad Ali et al., Discrimination Through Optimization: How Facebook’s Ad Delivery Can Lead to Skewed Outcomes, 2019, https://arxiv.org/abs/1904.02095.

1

See, e.g., Kaveh Waddell, How Tenant Screening Reports Make it Hard for People to Bounce Back from Tough Times, Consumer Reports, March 11, 2021, https://www.consumerreports.org/algorithmic-bias/tenant-screening-reports-make-it-hard-to-bounce-back-from-tough-times/.

1

See, e.g., Claudia Aranda, Urban Institute, Fighting Housing Discrimination in 2019, https://www.urban.org/urban-wire/fighting-housing-discrimination-2019.

1

See, e.g., Claudia L. Aranda, Urban Institute, Statement Before the Subcomm. on Transportation, Housing & Urban Development of the H. Comm. on Appropriations, Feb. 27, 2019, [https://www.urban.org/sites/default/files/publication/99836/housing_discrimination_in_america_-claudia_aranda.pdf](https://www.urban.org/sites/default/files/publication/99836/housing_discrimination_in_america-_claudia_aranda.pdf).

1

In developing guidance, HUD should review existing literature and frameworks such as Dillon Reisman et al., AI Now, Algorithmic Impact Assessments: A Practical Framework for Public Agency Accountability, Apr. 2018, https://ainowinstitute.org/aiareport2018.pdf; Wilkey et al., supra note 2; Inioluwa Deborah Raji, et al., Closing the AI Accountability Gap: Defining an End-to-End Framework for Internal Algorithmic Auditing, 2020, https://arxiv.org/abs/2001.00973; Timnit Gebru et al., Datasheets for Datasets, 2020, https://arxiv.org/abs/1803.09010; Margaret Mitchell et al., Model Cards for Model Reporting, 2019, https://arxiv.org/abs/1810.03993; Office of the Comptroller of the Currency, Supervisory Guidance on Model Risk Management, Apr. 2011, https://www.occ.gov/news-issuances/bulletins/2011/bulletin-2011-12a.pdf.

1

See Congressional Research Service, The Fair Housing Act: HUD Oversight, Programs, and Activities 10, Apr. 7, 2021, https://fas.org/sgp/crs/misc/R44557.pdf.

1

See, e.g., Brief for Upturn, Inc. as Amicus Curiae 2, Opiotennione v. Facebook, 3:19–cv–07185, June 26, 2020, https://www.courtlistener.com/recap/gov.uscourts.cand.350804/gov.uscourts.cand.350804.44.1.pdf; Facebook, Business Help Center, About Optimization for Ad Delivery, https://www.facebook.com/business/help/355670007911605?id=561906377587030.

1

See Muhammad Ali et al., supra note 3; Basileal Imana, Aleksandra Korolova & John Heidemann, Auditing for Discrimination in Job Ad Delivery, Proceedings of the Web Conference 2021, https://ant.isi.edu/datasets/addelivery/.

1

See Muhammad Ali et al., supra note 3.

1

See generally, e.g., Valerie Schneider, Locked Out by Big Data: How Algorithms and Machine Learning May Undermine Housing Justice, 52 Columbia Human Rts. L. Rev. 251, 2020, http://hrlr.law.columbia.edu/files/2020/11/251_Schneider.pdf.

1

For example, some housing providers use background check services, cameras, and other devices to monitor their tenants’ conduct, which can result in eviction or law enforcement intervention for lawful conduct, such as having non-resident visitors. See, e.g., HUD, Office of the General Counsel Guidance on Application of Fair Housing Act Standards to the Enforcement of Local Nuisance and Crime-Free Housing Ordinances Against Victims of Domestic Violence, Other Crime Victims, and Others Who Require Police or Emergency Services, Sept. 13, 2016, https://www.hud.gov/sites/documents/FINALNUISANCEORDGDNCE.PDF; Samantha Michaels, Hundreds of Cities Have Adopted a New Strategy for Reducing Crime in Housing. Is it Making Neighborhoods Safer—or Whiter?, Mother Jones, 2019, https://www.motherjones.com/crime-justice/2019/10/crime-free-housing-making-neighborhoods-safer-or-whiter/; Ginia Bellafante, The Landlord Wants Facial Recognition in Its Rent-Stabilized Buildings. Why?, N.Y. Times, Mar. 28, 2019, https://www.nytimes.com/2019/03/28/nyregion/rent-stabilized-buildings-facial-recognition.html.

1

HUD’s previous guidance has acknowledged that contributing factors can include “admissions and occupancy policies and procedures, including preferences in publicly supported housing;” lending discrimination; source of income discrimination; “displacement of residents due to economic pressures;” and “barriers faced by individuals and families when attempting to move to a neighborhood or area of their choice.” Dep’t of Housing & Urban Development, AFFH Rule Guidebook 206–19, 2015, https://www.nhlp.org/wp-content/uploads/2017/09/AFFH-Rule-Guidebook-2015.pdf. Tenant screening, underwriting, and surveillance technologies can contribute to these factors, for example, by keeping previously evicted tenants from being able to stay in their neighborhoods or cities. The guidance also states that “relevant local data and local knowledge that may assist a regional analysis include: demographic data from neighboring PHAs and policies and procedures concerning admissions and residency preferences for PHAs in the area.” Id. at 91.

1

See, e.g., Urban Institute, Data and Tools of Fair Housing Planning, https://datacatalog.urban.org/dataset/data-and-tools-fair-housing-planning.

1

See Leah Hendey & Mychal Cohen, Urban Institute, Using Data to Assess Fair Housing and Improve Access to Opportunity: A Guidebook for Community Organizations, Aug. 3, 2017, https://www.urban.org/research/publication/using-data-assess-fair-housing-and-improve-access-opportunity/view/full_report.

1

HUD, Office of General Counsel Guidance on Application of Fair Housing Act Standards to the Use of Criminal Records by Providers of Housing and Real Estate-Related Transactions, Apr. 4, 2016, https://www.hud.gov/sites/documents/HUD_OGCGUIDAPPFHASTANDCR.PDF.

1

Id.

1

See, e.g., Housing Equality Ctr. of Penn., Nuisance and Crime-Free Housing Ordinances, https://www.equalhousing.org/fair-housing-topics/nuisance-and-crime-free-housing-ordinances/#:~:text=Crime%2Dfree%20ordinances%20may%20define,Housing%20Act%20in%20several%20circumstances; Am. Civ. Liberties Union, I Am Not a Nuisance: Local Ordinances Punish Victims of Crime, https://www.aclu.org/other/i-am-not-nuisance-local-ordinances-punish-victims-crime; Michaels, supra note 12.

1

See, e.g., Fair Chance in Housing Act, S.B. 250, 219th Leg. (N.J. 2021), https://www.njleg.state.nj.us/bills/BillView.asp?BillNumber=S250; Fair Criminal Record Screening for Housing, D.C. Code § 42–3541, https://code.dccouncil.us/dc/council/code/titles/42/chapters/35B/; Seattle Mun. Code § 14.09, https://library.municode.com/wa/seattle/codes/municipal_code?nodeId=TIT14HURI_CH14.09USSCREHO; Ronald V. Dellums & Simbarashe Sherry Fair Chance Access to Housing Ordinance, Oakland Mun. Code § 8.25, https://library.municode.com/ca/oakland/codes/code_of_ordinances?nodeId=TIT8HESA_CH8.25ROV.DESISHFACHACHOOR; Bill no. 210330–A (Philadelphia 2021), https://phila.legistar.com/LegislationDetail.aspx?ID=4915398&GUID=EBAAF6DA-BB11-4ED4-8BD9-6A5224A55347.

1

See, e.g., Peter Hepburn, Renee Louis & Matthew Desmond, Racial and Gender Disparities Among Evicted Americans, Dec. 16, 2020, https://evictionlab.org/demographics-of-eviction/; Community Legal Services of Philadelphia, Breaking the Record: Dismantling the Barriers Eviction Records Place on Housing Opportunities, Nov. 2020, https://clsphila.org/wp-content/uploads/2020/12/Breaking-the-Record-Report_Nov2020.pdf; Sophie Beiers, Sandra Park & Linda Morris, Am. Civ. Liberties Union, Clearing the Record: How Eviction Sealing Laws Can Advance Housing Access for Women of Color, Jan. 10, 2020, https://www.aclu.org/news/racial-justice/clearing-the-record-how-eviction-sealing-laws-can-advance-housing-access-for-women-of-color/.

1

See, e.g., Brian J. McCabe & Eva Rosen, Eviction in Washington, D.C.: Racial and Geographic Disparities in Housing Instability 8–14, Fall 2020, https://georgetown.app.box.com/s/8cq4p8ap4nq5xm75b5mct0nz5002z3ap.

1

See, e.g., Josh Kaplan, Thousands of D.C. Renters are Evicted Every Year. Do They All Know to Show Up to Court?, dcist, 2020, https://dcist.com/story/20/10/05/thousands-of-d-c-renters-are-evicted-every-year-do-they-all-know-to-show-up-to-court/. Research shows that providing a right to counsel in housing court in New York City has led to large drops in the rates of eviction filings, default judgments, and executed evictions. See Nat’l Coalition for a Civil Right to Counsel, All About the Right to Counsel for Evictions in NYC, Dec. 14, 2020, http://civilrighttocounsel.org/major_developments/894. In Washington, D.C., twenty landlords filed nearly half of all evictions in 2018, often “serially” filing multiple evictions against the same tenants. Brian J. McCabe & Eva Rosen, Eviction in Washington, D.C.: Racial and Geographic Disparities in Housing Instability 6, Fall 2020, https://georgetown.app.box.com/s/8cq4p8ap4nq5xm75b5mct0nz5002z3ap.

1

See Kathryn A. Sabbath, Erasing the “Scarlet E” of Eviction Records, The Appeal, Apr. 12, 2021, https://theappeal.org/the-lab/report/erasing-the-scarlet-e-of-eviction-records/.

1

See, e.g., Bill no. 210330–A (Philadelphia 2021), https://phila.legistar.com/LegislationDetail.aspx?ID=4915398&GUID=EBAAF6DA-BB11-4ED4-8BD9-6A5224A55347; Housing Stability & Tenant Protection Act of 2019, S.B. 6458, 2019–2020 Assemb. (N.Y. 2019), https://www.nysenate.gov/legislation/bills/2019/s6458; Minneapolis Mun. Code § 244.2030, https://library.municode.com/mn/minneapolis/codes/code_of_ordinances?nodeId=COOR_TIT12HO_CH244MACO_ARTXVIREDWLI_244.2030APSCCRPRTE.

1

See supra note 25.

1

For existing guidance, see FTC, What Tenant Background Screening Companies Need to Know About The Fair Credit Reporting Act, Oct. 2016, https://www.ftc.gov/tips-advice/business-center/guidance/what-tenant-background-screening-companies-need-know-about-fair.

1

Section 6(b) of the Federal Trade Commission Act empowers the Commission to require an entity to file “annual or special . . . reports or answers in writing to specific questions” to provide information about the entity’s “organization, business, conduct, practices, management, and relation to other corporations, partnerships, and individuals.” 15 U.S.C. § 46(b).