A new study from the National Bureau of Economic Research suggests that systemic racial discrimination during the job application process is still happening.

Researchers sent more than 80,000 fake job applications for entry-level openings to Fortune 500 firms and found that, on average, applications with distinctively Black names were about 10% less likely to get a call back than comparable applications with distinctively white names.

A significant portion of the hiring discrimination documented in the study was associated with a small portion of the 108 companies, according to Evan Rose, a Saieh Family Research Fellow at the University of Chicago and a co-author of the study.

“We find that the top 20% of the firms that we studied here explained about half of the total discrimination against Black applicants in our study,” Rose said during an interview with “Marketplace Morning Report’s” David Brancaccio. “That’s heavily concentrated in a small share of employers.”

Below is an edited transcript of Rose’s conversation with Brancaccio, with details on which sectors are seeing hiring discrimination, how bias also targets gendered names and more.

David Brancaccio: How many of these fake applications did your team send out?

Evan Rose: By the end, we’d sent about 84,000 applications to about 100 large U.S. employers. So you can think about it as sending a pair of resumes where everything on the resumes was equivalent, their job experience or education but one resume might have a name like Emily or Greg, and another resume might have a name like Leticia or Jamal, and we measure how much less likely is the employer to call back the Leticia or Jamal than the Emily or Greg, as well as, by the way, the difference in callback rates between Emilys and Gregs. So this way, we can try to measure both race and gender discrimination, signaling those characteristics to the employers with these distinctive first names.

Brancaccio: And when all was said and done, you found evidence of what many would call systemic discrimination?

Rose: Yeah, that’s right. So on average, there’s a penalty that’s pretty meaningful, about 10%, to applying to a job with a distinctively Black name. But what’s really interesting is that we find that that penalty varies tremendously across the 100 or so large U.S. employers in our study. And in fact, we find that the top 20% of the firms that we studied here explained about half of the total discrimination against Black applicants in our study. That’s heavily concentrated in a small share of employers. And that’s actually even more true for gender, where, on average, we find no difference in the contact rates for distinctively male and female names. But we find that masks a bunch of between-firm differences. Some firms have very strong preference to call back the male names. Other firms have quite strong preference to call back the female names. So gender discrimination is also highly concentrated in a small share of firms.

Brancaccio: And just so we rest on this just for another moment, so people didn’t miss it, the worst actors, according to your data, account for a whole lot of the problem.

Rose: Absolutely, yes, it’s this top quintile of firms, this relatively small share of firms that explain the bulk of the discrimination that we measure in the experiment. And these are, again, large U.S. employers. You know and love them, at least I shop with them regularly. And unfortunately, it seems that there’s widespread patterns of discrimination across their establishments. And one thing we show in the paper is that, in fact, it looks like at least 20% of the actual jobs that we applied to at these firms are discriminating on the basis of race. So, it’s absolutely what the EEOC, or the [Equal Employment Opportunity Commission], would call a systemic pattern or practice of discrimination.

Brancaccio: So I’m on the edge of my seat here. Name some names. Who’s doing poorly, who’s doing well?

Rose: So we’re thinking about what to do with the identities of those firms. But we want to make sure that we go through academic peer review, engage with our partners in the civil rights community and government to try to make sure we can do the most good with this information.

Brancaccio: All right, so you’re thinking about it. That’s a process. Are you gonna tell the worst-offending companies?

Rose: That’s a possibility. We could be sharing this information directly with the public or with the government or with firms themselves. But again, you know, I think that’s sort of step two – step one here in this research was just figuring out this basic scientific question of whether or not firms matter at all. And our answer to that is definitely yes. And now you’re right. And we’re excited to think about what you could do with the actual information on specific companies’ discrimination.

Brancaccio: Are you seeing patterns in terms of industry, like certain types of companies are doing better or worse?

Rose: Yeah, industry turns out to be quite important and much more important than other factors that we thought might matter, like geography and job title. Interestingly, firms in the auto sector – firms that both sell cars and sell car parts – many of the firms we studied in those sectors exhibit very large race gaps, as well as firms in sort of more general retail and food services. Firms in engineering services, health services and accommodation, interestingly, seem to discriminate much less on the basis of race. And for gender, what we find is that the industry patterns are again very strong and sort of correspond closely to what you might think of as more stereotypically gendered industries. So in heavy industries, like wholesale durables, and construction materials, firms seem to have a preference to call back our male applicants. In apparel, we see a very strong preference to call back our female applicants. So there’s a strong industry skew there as well.

Brancaccio: Is it your hope that this type of data could be used to quell this form of discrimination?

Rose: Yeah, I mean, what we’re showing here is that the identity of the firm matters, and even within an industry like the auto sector, some firms seem to exhibit much more or much less discrimination than others. And that suggests to us that firm policies, practices, structure [and] organization could potentially matter. And if we started looking more closely at what these firms are doing differently, you know, we might be able to identify some of those policies and practices that could help prevent or remedy this kind of discrimination in the future.