Racial discrimination in rental markets is alive and well.
In a new working paper from the National Bureau of Economic Research, researchers found rampant racial discrimination in American rental markets — specifically, that property managers are less likely to respond to prospective Black and Hispanic tenants when they inquire about open listings.
Using a software bot, the economists sent inquiries from fake renters to 8,476 property managers in the 50 largest US metropolitan housing markets. The bot assigned names to fictitious renters that would indicate whether the race of the inquirer was white, Black, or Hispanic.
The bot found that names perceived to be white got a response 5.6 percentage points more than Black-sounding names, and 2.8 percentage points more than Hispanic-sounding names.
Though the economists were using fabricated identities to test for discrimination, they also followed up to see what happened in the properties in real life. In what could become a major contribution to the field, the researchers find that a non-response to an inquiry from a Black or Hispanic renter “lowers the probability that a renter of color will ultimately inhabit a given property by 17.3 percent.” This is “the first available evidence on the relationship between disparate treatment and subsequent rental housing outcomes.”
The researchers, University of Illinois Urbana-Champaign’s Peter Christensen, University of Los Andes’ Ignacio Sarmiento-Barbieri, and Duke University’s Christopher Timmins, found that discrimination isn’t the same everywhere. The researchers find that for Black would-be renters, the most discriminatory region is the Midwest and the most discriminatory individual cities are Chicago, Los Angeles, and Louisville. For Hispanic would-be renters, the most discriminatory region is the Northeast and the most discriminatory individual cities are Louisville, Houston, and Providence.
In a handful of cities, researchers found that Black and Hispanic names receive more responses than white renters. Notably, Jacksonville (31 percent Black) and Columbus (29 percent Black) for Black names and Phoenix (42.6 percent Hispanic), Sacramento (23.6 percent Hispanic), and Tampa (26.4 percent Hispanic) for Hispanic names. It’s possible that the large Black and Hispanic populations in these cities make it more difficult to discriminate without losing out on good renters, or, as Christensen noted, it could be an outlier, “from a statistical perspective, 5-10 outliers [in] a sample of 100 cities would be expected by chance.”
You might be familiar with résumé studies where researchers will send in identical résumés with just one thing changed, such as a 2003 study by economists Marianne Bertrand and Sendhil Mullainathan that showed résumés with names perceived as Black received 50 percent fewer callbacks than those with white-sounding names.
The new rental market discrimination study, the largest of its kind, was able to capitalize on the increasing migration of the rental market to the internet. That enabled the creation of software to make it a lot quicker to see how discriminatory rental markets are. It also allowed researchers to compare across all 50 of the top metropolitan areas simultaneously, something that would be much more difficult without advances in technology.
“The objective is not just to develop this software for our own work but to actually make it available to other researchers and enforcement agencies,” Christensen told Vox. “HUD’s only out there running a study every 10 years or maybe there’s some civil rights organization in your market or maybe there isn’t and [these studies have traditionally been] very expensive to run. And so nobody thinks that ... there’s monitoring really happening.”
Historically, the Department of Housing and Urban Development (HUD) has conducted much more time-consuming research called a “paired study” or an “audit” where they send two trained testers (one white and one person of color) to ask about housing units in 28 metropolitan areas across the country.
The testers are trained to present themselves as identically as possible outside of their differences in race and then record their experience. In their most recent 2012 study, HUD found that while “minority renters who call to inquire about recently advertised homes or apartments are rarely denied appointments that their white counterparts are able to make” people of color “are told about and shown fewer homes and apartments than whites” and “agents also quote slightly higher rents to blacks and Hispanics than to whites.”
One of the problems with traditional audit studies is that they involve real people — who are going to act somewhat differently not only from each other but with different landlords or property managers. That means it could be difficult to isolate the exact effect of race over perhaps other differences exhibited during these interactions.
David Neumark, a University of California Irvine economist who has investigated the use of correspondence and audit studies to test for discrimination in the labor market said in an email that both types have their uses: “Correspondence studies allow much larger samples. Audit studies are far harder to do, and get smaller samples. But they do allow collection of more information — importantly including whether the final transaction (getting an apartment, a job, etc.) actually happens.”
Researchers before Christensen, Sarmiento-Barbieri, and Timmins have adapted these studies for the internet age, since emails and other written missives can reduce the variability of in-person interactions. One study by Georgia State University economists Andrew Hanson and Zackary Hawley finds a “net level of discrimination of 4.5 percentage points against African American sounding names” where they observe interactions over email. In Sweden, researchers looked at discrimination against Muslims on a rental website via an audit study and found a 24.8 percent bias in favor of Swedish male names when compared to Muslim male names.
None of this research can capture the totality of discrimination in the housing market. Some people might not even find out about potential housing options if information is disseminated differently for different groups — for example, info that spreads through word of mouth.
Even if Black or Hispanic renters do know about an open unit and do get a response from the landlord, they could still face discrimination when a property manager is deciding between different tenants or during the application process.
Despite the supposed protections of the Fair Housing Act, which bans discrimination on the basis of race, gender, and other protected classes, civil rights groups have traditionally borne the brunt of responsibility for detecting this discrimination. It’s an expensive and time-consuming endeavor, limiting the frequency of oversight.
The development of this new tool could reduce some of that oversight burden — making it easier to consistently monitor at least one aspect of discrimination in the housing market.
Correction, December 8, 10 am: A previous version of this article incorrectly attributed the tool’s development. It was developed by University of Illinois Urbana-Champaign’s Peter Christensen, University of Los Andes’s Ignacio Sarmiento-Barbieri, and Duke University’s Christopher Timmins.