What happens when the big data economy meets the fast-moving and extremely high-stakes rental market? That’s the question that The Markup and The New York Times sought to answer in our investigation into the automated tenant screening industry.
It started with a tip from an attorney, who told us that some background screening companies used automated algorithms to create “risk scores” for people applying for housing, similar to how some algorithms are used in the criminal justice system to make crucial determinations about who is eligible for bail. We wanted to learn more about the tenant screening process and what happened when it got things wrong.
View our work
To learn how this process might affect the most vulnerable people—those applying to public housing authorities for low-income units or for vouchers that partially subsidize rent—we requested screening-service contracts from 88 public housing authorities across the country. We focused on those in larger cities and counties, like New York, Chicago, and Los Angeles, and then looked into other, smaller institutions that our reporting suggested might also be using screening companies in their admission processes.
We received responses from 67 housing authorities. Of those, 46 told us they used for-profit screening services, and most, but not all, sent us the contracts. We then asked these agencies for samples of screening reports for housing applicants who had been accepted and others who had been rejected.
Some housing authorities—in New York, Boston, Cincinnati, and Oakland, for example—don’t use third-party screening companies at all. Instead, they run checks on their applicants directly with local police departments and other government agencies. It’s unclear whether those checks lead to the same kinds of misreported information. We did not investigate that, but rather focused on for-profit screening companies.
To find the most commonly used for-profit companies and how they operate, we interviewed experts, advocates, attorneys, and screening company employees. We read publications about consumer reporting agencies put out by the Consumer Financial Protection Bureau and others. We used this information to compile a list of about two dozen tenant screening services that seemed to be the most frequently used and reported on those more deeply.
In the course of our reporting, we learned that when people find out a screening company misreported information about them, one of their options is to sue the company under the Fair Credit Reporting Act. These cases almost always wind up in federal court, so the records are included in the online federal court records system, PACER. We searched the system for civil cases that had been brought against these companies, narrowing our search to cases that had been filed against them since 2010 and where PACER had coded the “nature of suit” as 480 (for “consumer credit”) or 890 (for “other statutory action”).
Some of the bigger companies have been named in hundreds or thousands of court cases that fall into these categories. For those, we programmatically parsed the dockets and complaints to narrow them down to cases that were relevant to this story. This involved writing a program that automatically downloaded each case’s docket and court complaint, converting the complaint PDF into searchable text and highlighting the documents that contained the search terms landlord, tenant, apartment, house, housing, or rent.
The results were several hundred lawsuits. We read the court complaints from all of them. We found that some involved other kinds of background screening, such as employment, and we eliminated those. That left us with approximately 150 cases that were relevant for our story: a qualitative rather than a quantitative database. It is not a comprehensive list of all cases filed against all tenant screening companies in that time period.
We reached out to dozens of lawyers and housing applicants who had filed these lawsuits to learn more about their circumstances. Some agreed to be interviewed. In most cases, only the lawyers would speak to us. In some cases, no one got back to us. Nearly every lawsuit we reviewed that’s been resolved has ended in a settlement—and we learned that in most of those, the companies required the people who sued to sign an agreement not to talk about the case.