Multidimensional Discrimination in the Online Rental Housing Market: Implications for Families With Young Children

ABSTRACT

A half century after passage of the federal Fair Housing Act, studies continue to document racial discrimination in the housing market, which serves to reproduce racial inequality and residential segregation. Building on this work, we examine housing discrimination experienced by individuals belonging to multiple disadvantaged groups. Employing an online field experiment in 31 U.S. cities over 20 months, we investigate patterns of discrimination against female rental housing applicants at the intersections of race, ethnicity, family structure, and Section 8 housing voucher receipt. Consistent with prior work, we find discrimination against Black women and Section 8 recipients. We also find that only Black women and Latinas are penalized for being parents and for being single mothers to young children. Finally, examining the relevant policy landscape, we find evidence that state and local laws barring discrimination against Section 8 recipients may not be sufficient to protect voucher holders and their families and may instead prompt landlords to engage in subtler forms of discrimination (i.e., increased nonresponse). These findings reveal a dynamic pattern of multidimensional discrimination and support arguments for an intersectional approach to understanding and combatting inequality.

KEYWORDS:

• vouchers
• race
• discrimination
• families

Acknowledgments

Both authors contributed equally to the production of this article. Our work was supported by the Russell Sage Foundation (Award 83-17-24) and funding from New Yorl University’s Robert F. Wagner School of Public Service. The authors thank Matthew Desmond, Ingrid Gould Ellen, Patrick Fowler, S. Michael Gaddis, Maria Krysan, Brian McCabe, Katherine O’Regan, David Pedulla, Vincent Reina, Eva Rosen, Zawadi Rucks-Ahidiana, and Margery Turner for their extensive feedback on early drafts of this project. We are also very grateful for the contributions made by Sarah Aita, Claudia Babcock, Roberta Barnett, Shengdi Chen, Rediet Demissie, Arman Hirose-Afshari, Alimatou Juwara, Peter Mattingly, Kahinee Shah, and Haowen Zheng.

Disclosure Statement

No potential conflict of interest was reported by the authors.

Notes

1. To obtain a voucher, families apply to a local PHA, which certifies that their income is below the eligibility threshold of 80% of the area median income (Ellen, 2018). Recipients pay 30% of their income toward rent, whereas the federal government subsidizes the remainder, up to the local maximum payment standard set between 90% and 110% of the Fair Market Rent (FMR). The FMR is defined as either the 40th or the 50th percentile of rent in the metropolitan area.

2. Table A2 displays response rates for each name used in our study.

3. Assessing compounding discrimination against families with male adolescents relative to families with younger children, for example, would require additional signals for older children and sex. We were concerned that additional experimental variations would limit statistical power.

4. Table A3 displays the number of observations for each combination of experimental signals.

5. This variation allows us to assess the impact of unit size on bias against various family structures. Families may be refused a unit if the landlord believes it is too small (Turner, 2015), including refusal to rent a one-bedroom apartment to a family with children because of housing code restrictions. Landlords may also simply prefer smaller households because of decreased wear and tear on the unit.

6. The age of a post was unrelated to discriminatory patterns (i.e., there was no difference in outcomes between newest posts and 100th newest posts). There were also no observed differences between ads that had been updated at least once and those without an update.

7. Models include indicators for whether the advertisement included a specific address (28% of listings), a street intersection (9%), or latitude and longitude (59%).

8. Models include indicators for the month our message was sent and a continuous measure of the number of messages previously sent in each market to control for idiosyncrasies, in part because of national holidays and because of occasional issues with data collection (e.g., research assistant hiring cycles). The number of previous messages sent to landlords within the same city was not a significant predictor of response, nor were month-by-year interactions. Together, these null results suggest that detection by landlords was likely not a source of bias.

9. We found no variation in family structure effects across the Section 8 signal.

10. Full results are available in Table A4.

Additional information

Funding

This work was supported by the Russell Sage Foundation [Award 83-17-24].

Notes on contributors

Jacob William Faber

Jacob William Faber is an associate professor of sociology and public service at New York University. He leverages observational and experimental methods to study the mechanisms responsible for sorting individuals across space and how the distributions of people by race and class interact with political, social, and ecological systems to create and sustain economic disparities. His scholarship highlights the rapidly changing roles of numerous institutional actors in facilitating the reproduction of racial and spatial inequality.

Marie-Dumesle Mercier

Marie-Dumesle Mercier is a doctoral student in the Department of Sociology at New York University. Her research interests include the relationship between law, social change, and racial inequality. She holds an A.B. from Princeton University and a J.D. cum laude from the University of Michigan Law School.