The role of race in the labor market
A conversation with Professor Dan O’Flaherty
When talking about economic inclusion, we are thinking about what it takes for more people to participate more fully in the economy. One of the clearest indicators of economic inclusion is whether someone can freely pursue work—or, as economists would say, whether one can freely participate in the labor market. While we can look across a number of groups and find gaps in people’s participation, some of the greatest gaps fall along racial lines. We have outlawed some of the most egregious discriminatory barriers to participation in the labor market, but by many measures, racial inequality in the labor market has improved very little over recent decades.
Why is this still happening?
While self-reported racially discriminatory views have fallen tremendously over recent decades, does this mean that discrimination in hiring is no longer a key driver of labor market inequality? Gaps in test scores between Black children and white children persist despite decades of school integration efforts. What do these gaps indicate about the quality of schools that Black and white children attend and the accompanying differences in preparation for the labor market they receive? When employers hire workers using signals of talent that are influenced by inequality of opportunity and discrimination, how can they sort through the noise effectively?
I recently spoke with Professor Dan O’Flaherty about what the economics literature can help us learn about these topics. He’s an economist at Columbia University who has studied racial inequality, homelessness, and crime, among other topics. In 2015, he published a book called The Economics of Race in the United States.
Professor O’Flaherty told me about audit studies, which use actors, resumes, or contact information from two different groups, such as Black applicants and white applicants, to test for discrimination in the labor market. While these studies provide some key insights, they are limited in what they can tell us about the extent of harm resulting from discrimination. We also talked about how residential segregation results in segregating job referral networks, a situation which can curtail a company’s access to the full talent pool. This access stands out in cases in which employers are forced to look harder to hire diverse talent; evidence shows that employers benefit when they keep using their new methods to search for talent going forward.
The following conversation has been edited and condensed. Listen to an extended version of this conversation.
Evidence of racial discrimination in the labor market from audit studies
One of the ways that we've tried to detect discrimination in economics is to look at audit studies. I'm wondering if you could describe audit studies generally, some of the big results recently. I'm thinking of Bertrand and Mullainathan (2004), but also since your book was published, Kline, Rose, and Walters (2021).
What audit studies do is that audit studies can tell us about process. They cannot tell us about outcomes. What audit studies started out being was sending pairs of people who were trained to act in the same way to employers, one was Black, one was white. They apply. They see what happens. The results there have generally been almost always, the employers treat both the same in that they don't get the job or get a callback.
To the extent that we see them getting callbacks, generally, the white auditors get called back more. This was criticized by Jim Heckman with one very good objection, which I'll come back to, and one serious objection, which was met, which was that the auditors know what's going on. They're going to play. You can't trust them.
Bertrand and Mullainathan had this wonderful idea of sending out resumes with Black sounding names and white sounding names and looking at what the response was about calling back. Once again, they found that very few people get called back. To the extent that there are callbacks, white-sounding names do better than Black-sounding names.
Now, this moved toward meeting one of Heckman's objections, in that there are no auditors, so they can't play. It moved further from Heckman's other objection, which I would say is more definitional in that you can't imagine anyone being worse off because of this because there's no one to be worse off. In legal terms, I think what the auditors do are looking for disparate treatment. They're looking for bad motive.
The other thing about the audit studies is they find that there are asymmetries in both directions. They don't find that no employers ever call back Blacks and not whites. There are some who call back Blacks and not whites and some in other direction. What they find is the preponderance goes in the direction of helping whites.
That's why I think it's important. I think a lot of the discussion of discrimination has gotten hijacked on the minutiae, where the preponderance of it is what matters.
As an example, I believe very firmly that you treat your kids better than you treat me. If one of your kids and I were having birthday parties the same day, and I invited you and they invited you, you would show up at their birthday party.
Yeah. I'm sorry to confirm, that's probably what would happen.
Yep. And I am no more deserving than your kids or your kids are no more deserving than I am. And one, I don't take it personally, but you show an unreasonable preference for your kids, and I hope you do.
What I would say is that we have this kind of discrimination going on all the time. And there are barbershops that will not cut my hair. I've managed to go do pretty well in life without them. I have never gotten an offer of employment from North Korea, although I'm probably a better economist than anybody that they have.
But don't you think this whole question though gets back to the issue of power, but also how resources are already distributed throughout the economy? I mean, you could think about it in terms of a job ladder, but you could think about it in lots of ways where it could be the case that not being able to get that job means you don't get that opportunity. And there's more of a consequence than for example, me not showing up at your birthday party.
If it were the case that, for example, let's say the best birthday present that you were going to get was coming from me and then I choose to go to my daughter's, that could be more of an issue.
I'm not saying that discrimination isn't an issue. I'm saying that there is lots of discrimination going on that really doesn't matter. There is some that does matter.
And it's hard for us to know.
Well, what we know is that when it's pervasive, if nobody comes to my birthday party, then I'm in trouble. I think, the evil or the harm comes from pervasive discrimination. The other kind of studies, the disparate impact studies, the wage regressions, they show that, okay, the preponderance matters. That in fact, all sorts of bad things happen to white people. All sorts of bad things happen to Black people, but on net, it is worse off for Black people.
Test scores contain information about experiences before entering the labor market
Thinking about this issue of the role of education in the labor market and racial disparities, I'm thinking both in terms of attainment and achievement. I'm thinking about, for example, Keane and Wolpin (1997), where they're finding that a large share of lifetime labor market outcomes can be explained by what's happened to you by age 16 or 18. How do you think about that and the connection with education?
You better get educated.
Yeah, a large part of outcomes can be explained by pre-market factors. But one of the things that has been coming out since the book, or at least coming to my attention since the book, is that a large part of what goes on probably is determined in the social and behavioral, as well as what used to be called cognitive parts. And the sleeper effects there, where we have all these interventions that happen earlier in life, they do nothing to test scores. But then you find out at age 20, they make all the difference in outcomes in your life that really matter.
There are several interesting questions about test scores: When are they informative? Or are they informative at all? What are they actually measuring? Could you talk a little bit about some of the research using the National Longitudinal Surveys of Youth 1979 wave, and the AFQT variable in there. Can you speak a little bit about that?
Sure. So there's this important paper by Neal and Johnson in 1996 that has an interesting framework. They imagine life as a two-stage process, in which you're a kid and that after ceasing to be a kid, you're a worker. And there can be discrimination in both stages.
Their question is, is there second stage discrimination? And what they propose is that if you look at test scores around age 18, Armed Forces Qualifying Test scores (AFQT), these particular cognitive tests summarize everything relevant that you have learned up until age 18, and it is what you bring to the labor market. Now, the interesting thing to me, since I've been thinking about this recently in a police homicide context, is that the overlap between Black AFQT scores and white AFQT scores is very small. And they give themselves degrees of freedom, and they find that for everyone except Black men, for Hispanics, and Black women, the AFQT difference explains the difference in earnings in the late 20s. And instead of a 20% difference for Black men, it becomes a seven point difference. This has inspired a wonderful amount of research and a wonderful amount of thought, and I don't know if the authors of the study believe it anymore.
So, how do you think about it? How do you think about the contribution of these, let's call them cognitive skills, to labor market outcomes?
My own views are that even if it's not as large as they find, I think what it's saying to me is that education does matter, and as you said, it's not the panacea. It's not going to solve all the other forms of discrimination in the labor market. It's not going to necessarily solve residential issues.
But I guess the way I see it is, what this is saying is that cognitive skills, or whatever it is that's measured in these tests, do matter for the labor market. And the thing that I find encouraging is that that is something we can work on. That's something that I think we can change.
I definitely agree with that.
Employers can learn other ways to judge pre-market experiences and potential
I think what I've realized recently is that there is a very active literature in policing which is the exact same problem. So you have one stage in which a police officer arrests somebody and a second stage in which a district attorney makes a charging decision. The question that was a very active debate two summers ago was, “What can you say about the second stage, given that you have a really hard time observing the first stage? Because you cannot observe the people whom the officer did not arrest.” How do you think about the second stage? Nobody in this literature said that it is not important whom they arrest.
Here, let's go to the employment case. I think employers, under law, employers do have obligations to applicants. In some cases, if I'm a little guy, I'm hiring people for my pizza store. And I can see somebody's resume. I can see they went to a school which I never heard of. I'm not going to hire them, because I really want smart pizza delivery people. But if I'm a large employer, then I ought to question how good the information is that I'm getting from the first stage.
So, you're drawing an analogy with the district attorney receiving people that have been arrested and saying, "You need to investigate that process a little bit." And now you're saying, “Okay, let's think about large employers. They should also be investigating a little bit the process by which someone's presenting to them with whatever credentials.”
And if you look at how affirmative action works in large employers, the stories that we have and actually some data on this, what does it do? Coming under affirmative action requirements forces employers to go back and look in the first stage in the sense that they find sources of referral. They find schools, they learn about them. They improve their ability to judge the first stage.
That is, they improve their ability to judge an applicant’s potential.
And then we've got Conrad Miller's result, that they keep using it once they've made the investment. We saw in Conrad's paper that it seems to be profit maximizing, once you've set it up.
I want to turn to another paper by Conrad Miller, this one coauthored with Ian Schmutte, where they look at data from Brazil that includes the race of the entrepreneur who started a business.
In this paper, they find that businesses that are started by white founders end up hiring Black employees at the same rate as businesses started by Black founders. But that it takes time. And so they're able to interpret this in terms of a story about job referral networks. And essentially, you get the chain of causality from a job referral network being segregated because social networks are segregated, which are segregated because of residential segregation.
I think this helps to show how important job referral networks are in the hiring process and how that connects with the issue of residential segregation.
Yeah. I think job referral networks are important. The first Conrad Miller paper also was about job referral networks, in the sense that one of the reasons for the continued hiring of Black employees was that the job referral networks were set up. And that's again, what I would think about as due diligence. Like the district attorney checking on what goes on in the first stage, if I'm not getting any referrals from this neighborhood or this group of people, maybe I ought to find out why and take steps to get referrals.
The opinions expressed in this article are those of the participants and do not necessarily represent the views of the Federal Reserve Bank of Cleveland or the Board of Governors of the Federal Reserve System.
Professor of Economics
Professor Dan O’Flaherty is a professor of economics at Columbia University. He is the author of the books The Economics of Race in the United States and Shadows of Doubt: Stereotypes, Crime, and the Pursuit of Justice, and the textbook City Economics; and he is co-editor of the book How to House the Homeless. Dr. O’Flaherty has extensive experience advising government and community groups in Newark, New Jersey. Dr. O’Flaherty earned his bachelor’s degree in 1973 and his PhD in economics in 1980, both from Harvard University.