Who Discriminates in Hiring? A New Study Can Tell.

Twenty years ago, Kalisha White performed an experiment. A Marquette University graduate who is Black, she suspected that her application for a job as executive team leader at a Target in Wisconsin was being ignored because of her race. So she sent in another one, with a name (Sarah Brucker) more likely to make the candidate appear white.

Though the fake résumé was not quite as accomplished as Ms. White’s, the alter ego scored an interview. Target ultimately paid over half a million dollars to settle a class-action lawsuit brought by the Equal Employment Opportunity Commission on behalf of Ms. White and a handful of other Black job applicants.

Now a variation on her strategy could help expose racial discrimination in employment across the corporate landscape.

Economists at the University of California, Berkeley, and the University of Chicago this week unveiled a vast discrimination audit of some of the largest U.S. companies. Starting in late 2019, they sent 83,000 fake job applications for entry-level positions at 108 companies — most of them in the top 100 of the Fortune 500 list, and some of their subsidiaries.

Their insights can provide valuable evidence about violations of Black workers’ civil rights.

The researchers — Patrick Kline and Christopher Walters of Berkeley and Evan K. Rose of Chicago — are not ready to reveal the names of companies on their list. But they plan to, once they expose the data to more statistical tests. Labor lawyers, the E.E.O.C. and maybe the companies themselves could do a lot with this information. (Dr. Kline said they had briefed the U.S. Labor Department on the general findings.)

In the study, applicants’ characteristics — like age, sexual orientation, or work and school experience — varied at random. Names, however, were chosen purposefully to ensure applications came in pairs: one with a more distinctive white name — Jake or Molly, say — and the other with a similar background but a more distinctive Black name, like DeShawn or Imani.

What the researchers found would probably not surprise Ms. White: On average, applications from candidates with a “Black name” get fewer callbacks than similar applications bearing a “white name.”

This aligns with a paper published by two economists from the University of Chicago a couple of years after Ms. White’s tussle with Target: Respondents to help-wanted ads in Boston and Chicago had much better luck if their name was Emily or Greg than if it was Lakisha or Jamal. (Marianne Bertrand, one of the authors, testified as an expert witness in the trial over Ms. White’s discrimination claim.)

This experimental approach with paired applications, some economists argue, offers a closer representation of racial discrimination in the work force than studies that seek to relate employment and wage gaps to other characteristics — such as educational attainment and skill — and treat discrimination as a residual, or what’s left after other differences are accounted for.

The Berkeley and Chicago researchers found that discrimination isn’t uniform across the corporate landscape. Some companies discriminate little, responding similarly to applications by Molly and Latifa. Others show a measurable bias.

All told, for every 1,000 applications received, the researchers found, white candidates got about 250 responses, compared with about 230 for Black candidates. But among one-fifth of companies, the average gap grew to 50 callbacks. Even allowing that some patterns of discrimination could be random, rather than the result of racism, they concluded that 23 companies from their selection were “very likely to be engaged in systemic discrimination against Black applicants.”

There are 13 companies in automotive retailing and services in the Fortune 500 list. Five are among the 10 most discriminatory companies on the researchers’ list. Of the companies very likely to discriminate based on race, according to the findings, eight are federal contractors, which are bound by particularly stringent anti-discrimination rules and could lose their government contracts as a consequence.

“Discriminatory behavior is clustered in particular firms,” the researchers wrote. “The identity of many of these firms can be deduced with high confidence.”

The researchers also identified some overall patterns. For starters, discriminating companies tend to be less profitable, a finding consistent with the proposition by Gary Becker, who first studied discrimination in the workplace in the 1950s, that it is costly for firms to discriminate against productive workers.

The study found no strong link between discrimination and geography: Applications for jobs in the South fared no worse than anywhere else. Retailers and restaurants and bars discriminate more than average. And employers with more centralized personnel operations handling job applications tend to discriminate less, suggesting that uniform rules and procedures across a company can help reduce racial biases.

An early precedent for the paper published this week is a 1978 study that sent pairs of fake applications with similar qualifications but different photos, showing a white or a Black applicant. Interestingly, that study found some evidence of “reverse” discrimination against white applicants.

More fake-résumé studies have followed in recent years. One found that recent Black college graduates get fewer callbacks from potential employers than white candidates with identical resumes. Another found that prospective employers treat Black graduates from elite universities about the same as white graduates of less selective institutions.

One study reported that when employers in New York and New Jersey were barred from asking about job candidates’ criminal records, callbacks to Black candidates dropped significantly, relative to white job seekers, suggesting employers assumed Black candidates were more likely to have a record.

What makes the new research valuable is that it shows regulators, courts and labor lawyers how large-scale auditing of hiring practices offers a method to monitor and police bias. “Our findings demonstrate that it is possible to identify individual firms responsible for a substantial share of racial discrimination while maintaining a tight limit on the expected number of false positives encountered,” the researchers wrote.

Individual companies might even use the findings to reform their hiring practices.

Dr. Kline of Berkeley said Jenny R. Yang, a former chief commissioner of the E.E.O.C. and the current director of the Office of Federal Contract Compliance Programs, which has jurisdiction over federal contractors, had been apprised of the findings and had expressed interest in the researchers’ technique. (A representative of the agency declined to comment or to make Ms. Yang available.)

Similar tests have been performed since the 1980s to detect discrimination in housing by real estate agents and rental property owners. Tests in which white and nonwhite people inquire about the availability of housing suggest discrimination remains rampant.

Deploying this approach in the labor market has proved a bit tougher. Last year, the New York City Commission on Human Rights performed tests to detect employment discrimination — whether by race, gender, age or any other protected class — at 2,356 shops. Still, “employment is always harder than housing,” said Sapna Raj, deputy commissioner of the law enforcement bureau at the agency, which enforces anti-discrimination regulations.

“This could give us a deeper understanding,” Ms. Raj said of the study by the Berkeley and Chicago researchers. “What we would do is evaluate the information and look proactively at ways to address it.”

The commission, she noted, could not take action based on the kind of statistics in the new study on their own. “There are so many things you have to look at before you can determine that it is discrimination,” she argued. Still, she suggested, statistical analysis could alert her to which employers it makes sense to look at.

And that could ultimately convince corporations that discrimination is costly. “This is actionable evidence of illegal behavior by huge firms,” Dr. Walters of Berkeley said on Twitter in connection with the study’s release. “Modern statistical methods have the potential to help detect and redress civil rights violations.”