How Do You Think About Selection on Observables

I first met Peter Arcidiacano, professor of economics at Duke University, while I was a PhD student at the University of Georgia and I have followed his work since from a distance. I originally followed Peters work because he’d written several articles about sex from a two-sided matching perspective. I was struck by the fact that we both saw thinking about sexual relationships in terms of a matching problem. Two sided matching perspectives focus on the assignment mechanisms that bring people together, and when it comes to sexual relationships, the relative supply of possible partners and competition for those partners will in equilibrium result in pairings, some of which may become the most life sustaining and defining partnerships of those peoples lives. Peter’s work was gratifying to read, and I have often looked up to him for his successful merging of theory and econometrics to study topics I cared about.

The economic way of thinking is not about topics, nor is it is not about data, even though economists tend to have particular topics they study intensely and use data usually to do so. The economic way of thinking does though typically involve careful study of allocation mechanisms, such as prices and markets, that bring the productive capacity of communities into existence. These things are important as they animate humans to work together, produce output that manages the production itself, and increasingly towards the end of history, left surplus for humans to enjoy. Who ends up in what activities doing what types of specialized work ultimately shapes that which is made, how much and how it is distributed. The allocations end up not only shaping our lives, but our children’s lives. Starting conditions can cast a long shadow lasting centuries even causing certain groups to creep ahead as more and more of the surplus mounts and accrues to them, while others watch as a shrinking part of the growing pie flows to them.

In the United States, in the 21st century, one of the key institutions in all of this assignment of love and commerce has been the university. And within the university system, there are gradations of institutional pedigree and at the top of the pack sits elite institutions whose students seem practically destined to shape and receive the surplus. Given the path dependence in wealth, and how it has interacted with race, it is therefore no wonder that policymakers and economists have for decades sought to refine the rules by which schools can select high school applicants for admission. In many ways, our country’s fight over the use of race in selecting students into college is the old debates about capitalism and the self adjusting market system writ large.

So it’s in this broader context about work, schools, matching and allocation mechanisms that I think of Peter and his scholarship. When I review the range of topics on Peters vita, I see the signature marks of the modern 21st century labor economist. Someone interested in markets and how they work to connect people into productive cooperation. Someone interested in institutions, someone concerned about inequality and discrimination, someone versed both in economic theory and econometrics, someone at home with a bewildering array of numbers in a spreadsheet. To me, it is natural that Peter has pivoted so fluidly between topics like sex, work, discrimination and higher education because in my mind these are all interconnected topics concerning the assignment mechanisms we use in America to organize society and maintain our collective standard of living.

I invited Peter on the Mixtape with Scott as part of an ongoing series I call “economists and public policy”. The series focuses on how economics and economists think about and attempt to shape public policy. It includes people with a variety of perspectives, and even some who are critics of economics itself. Previous guests on the podcast in the “economists and public policy” series have been Sandy Darity, Elizabeth Popp Berman, Anna Stansbury, Mark Anderson, Alan Manning, Larry Katz, Jeremy West and Jonathan Meer.

Peter has not only produced academic articles in some of economics’ most impactful outlets — he has recently served as expert witness in two major discrimination cases, one of which put him on the opposite side of the stand as David Card, winner of 2021 Nobel Prize in economics. You can read about the cases here. They involve the broader topic of race and affirmative action at universities. The cases more specifically involve whether Harvard and UNC Chapel Hill admissions criteria show signs of discrimination.

One of the things about Peter’s involvement as expert witness that I want to highlight, though, is that his expert testimony was, at its core, an example of the role that econometrics can play in the shaping of public policy. It is more and more the case that economics’ role in the shaping of public policy in the 21st century will involve not merely economic theory, but also statistical analysis of complex datasets too, and I think it is worth pausing and noting that the economist shapes public policy oftentimes these days as much through interpreting data as her counterpart did using pure economic theory. I hope you find this discussion with Peter thought provoking and informative about both his work on these cases, but also about the role of economics and econometrics in forming public policy. But I also hope that the interview will give you a deeper insight into Peter and who he is.

Scott’s Substack as well as The Mixtape with Scott are supported by user subscriptions. Please share this episode to people within and outside economics that you think might be interested. I love doing these interviews and using the substack to do deeper dives into econometrics and the lives of economists and if you find this work valuable, please consider subscribing and supporting it.

Scott's Substack is a reader-supported publication. To receive new posts and support my work, consider becoming a free or paid subscriber.

Transcript

Scott Cunningham:

In this week's episode of Mixtape: the Podcast, I had the pleasure of talking with a professor at Duke University, named Peter Arcidiacono. I can never pronounce it correctly, no matter how many times I try. I first met Peter in graduate school. He was, probably then, an assistant professor at Duke, where he has spent his entire career. I was a PhD student at the University of Georgia. And he had a research paper on a topic that I was also working on, involving marriage markets. He's been an incredibly prolific producer in the area of labor economics and education, as well as affirmative action. And he uses tools in econometrics, that I largely never invested in, structural econometrics and discrete choice modeling. So when I read his work, I usually do it, both, because I'm interested in the paper and the paper topic, but also because I'm hoping that this will be a chance for me to open my mind a little bit more, and pick up on some of that econometric modeling, that I lack.

Peter is also an expert witness in a high profile case, right now, involving affirmative action and racial discrimination at Harvard University, and the University of North Carolina Chapel Hill, both of which have been combined into a single case. As I understand it, it's going for the Supreme court soon. In this interview, we walk through a lot of big and small issues around society's preferences around poverty, inequality, as well as the role that higher education is playing in both. My name is Scott Cunningham, and this is Mixtape: the Podcast.

Scott Cunningham:

Okay. This is great. I don't know if you remember. So this is an interview with professor of economics at Duke University, Peter Arcidiacono. And we're going to be talking about a range of topics. But just to give the reader and the listener a little bit of background, Peter, could you tell me a little bit about yourself, and what your involvement is with a current case, going before the Supreme court, involving University of North Carolina and Harvard University?

Peter Arcidiacono:

Certainly. And thanks for having me on. I've been at Duke now, for over 20 years. This was my first job out of grad school, and stayed here ever since. And a lot of my work has been on higher education, both with regard to choice of college major, as well as affirmative action.

And one of the really dissatisfying things about working on affirmative action, is that universities hide their data. So you can't really get a good sense of how the programs are working, because you typically don't have the data. And I think that that really matters, because to me, so much of the discussion about affirmative action, is in the binary. Either we have it, or we don't have it. But what it means to have it, is something, as economists, we would think about, that's something we would be optimizing over. And so, there's really a large space between race as a tiebreaker in admissions, and what somebody like Abraham Kennedy would advocate for, which would be more of a quota system.

And so, thinking about where you stand on that, to me, I had this opportunity to work on these two cases, two lawsuits. One brought against Harvard, and one against UNC, on the role of race in these admissions processes. And for me, it was an opportunity to look behind the veil, and see how these programs actually operated.

My intent was always to, a feeling as though, if I'm going to be an expert on affirmative action, I should know how these processes actually work. So my intent was always to use this for the purposes of research, as well. And we've written a number of papers out of the Harvard case. Four have been accepted now, and we just released a fifth one on racial preferences of both schools. And we'll see what happens with that. So those lawsuits, I testified in trial, at both those cases. My counterpart in the Harvard case was David Card, who recently won the Nobel Prize. I was wondering how I would respond to that. And actually, my response, I got to go up against a Nobel Prize winner.

Scott Cunningham:

Yeah. Right. Right.

Peter Arcidiacono:

So those experiences were somewhat traumatizing. But both experts, David Card and Kevin Hoxby, are pillars in the field, and people who have been very helpful to me, and who I have a great deal of respect for.

Scott Cunningham:

Yeah. Yeah.

Peter Arcidiacono:

So these cases have now, in both those cases, the side I was on lost at the first round. In the Harvard case, they also lost at the appellate round.

Peter Arcidiacono:

In UNC, it didn't actually go through the appellate round, because-

Scott Cunningham:

Oh, so-

Peter Arcidiacono:

... supreme court merged the cases.

Scott Cunningham:

... Both the Harvard University case and the Chapel Hill case, were already decided, but not at the Supreme Court level.

Peter Arcidiacono:

That's right.

Scott Cunningham:

Okay.

Peter Arcidiacono:

So the decision was appealed. It's now before the Supreme Court.

Scott Cunningham:

Yeah.

Peter Arcidiacono:

I think the Supreme Court scheduled here, arguments in October, and then, we'll see when they release a decision.

Scott Cunningham:

Okay. So, and these are both cases involving affirmative action and racial discrimination amongst particular groups of people? Is that groups of students, is that right?

Peter Arcidiacono:

That's right. Though, in the UNC case, there's actually no claim of Asian American discrimination. So that actually, you only see that at Harvard. You don't see that at UNC. That doesn't mean, I think that Asian discrimination is unique to Harvard.

Scott Cunningham:

Right.

Peter Arcidiacono:

I think it has to do with the fact of there not being that many Asian Americans in North Carolina.

Scott Cunningham:

North Carolina, right.

Peter Arcidiacono:

It's always been a bigger issue at the very top schools.

Scott Cunningham:

And you were called in, as an expert witness, for the plaintiff in both of those cases.

Peter Arcidiacono:

That's right.

Scott Cunningham:

Right. So David Card is the expert witness for Harvard, representing Harvard, against an accusation of, well, what exactly is the accusation against both of these institutions, and who brought these accusations against them?

Peter Arcidiacono:

So the group is called Students for Fair Admissions. And they basically got groups of students to, as their plaintiffs. Though, it's not about those particular students, in terms of remedies. And in Harvard, there's three claims. One, whether or not they're discriminating against Asian American applicants, relative to white applicants.

Scott Cunningham:

Mm-hmm.

Peter Arcidiacono:

Two, whether the size of the preferences given for underrepresented groups, is constitutional.

Scott Cunningham:

Yeah.

Peter Arcidiacono:

And three, whether there were race-neutral alternatives that they could have used. So the Supreme Court has said, "If there is a race-neutral alternative, you should use that."

Scott Cunningham:

Okay.

Peter Arcidiacono:

I'm not really involved at the race-neutral part. We had a different expert for that aspect.

Scott Cunningham:

Yeah.

Peter Arcidiacono:

Though, in both cases, Card and Hoxby actually did the race-neutral part, as well.

Scott Cunningham:

What exactly does the constitution say a admissions committee can use, when drawing up a student cohort?

Peter Arcidiacono:

Well, so I'm not sure what the constitution has to say on it, but I can say what the history of this of the court challenges have been.

Scott Cunningham:

Okay.

Peter Arcidiacono:

So I think, it's Title VI of Civil Rights Act said, "You're not supposed to use race-"

Scott Cunningham:

Race.

Peter Arcidiacono:

"... in these types of things." And there are other categories too.

Scott Cunningham:

Okay.

Peter Arcidiacono:

But race is the focus of this one. Now, the reason they had that, was because of the history of ill treatment of African Americans.

Scott Cunningham:

Right.

Peter Arcidiacono:

And this is obviously going in the other direction-

Scott Cunningham:

Mm.

Peter Arcidiacono:

... with regard to African Americans receiving preferences in the admissions process.

Scott Cunningham:

Mm. Mm.

Peter Arcidiacono:

So, but then, the history was that the original decision, the Bakke case, said, "Look, you can't use race in admissions, because of reparations. You can only use it because of the benefits of diversity." So the state can have an interest in diversity. And that was a compromised position to get that swing justice, to sign onto it.

Scott Cunningham:

Mm.

Peter Arcidiacono:

Since then, there have been a number of cases. I think the ones that are most relevant right now, are the ones that came out of the Michigan cases.

Scott Cunningham:

Yeah.

Peter Arcidiacono:

And there was one at the undergraduate level, which they found that you could not use race as part of an explicit point system.

Scott Cunningham:

Mm.

Peter Arcidiacono:

So you can get points for having a good SAT score, points for being a particular race, you add them up together, then you could rank the-

Scott Cunningham:

I see.

Peter Arcidiacono:

... applicant.

Scott Cunningham:

So there were schools that were doing a point system based on individual characteristics, like race. And that was, at that moment, it was unclear whether that would be legal. It was, I guess, or was it something that schools were, potentially, in a legal, bad situation, when they were using it? Or was it just not known?

Peter Arcidiacono:

I don't think it was clear. And that's where the court ruled. You cannot use it in that way.

Scott Cunningham:

Got it. Okay.

Peter Arcidiacono:

At the same time, there was a case against Harvard's Law School. And on that one, they said that you could use race, holistically. As an economist, I can express anything as a formula. And then, the question is, whether you see all parts of the formula or not.

Scott Cunningham:

Yeah.

Peter Arcidiacono:

So it gets a little tricky. And I think that, from my perspective, I would've rather had the ruling go in the exact opposite way.

[inaudible 00:11:59] on if we're going to find in favor of one or the other.

I would prefer a point system to a holistic one, because then, everything's clear.

Scott Cunningham:

Clear. Yeah. It seemed really precise-

Peter Arcidiacono:

[inaudible 00:12:09], to hide their data.

Scott Cunningham:

... Yeah. It seems like lots of times with the law, the imprecision of this language, as though it's a solution to the problem, is really challenging for designing policy.

Peter Arcidiacono:

I totally agree. Yeah.

Scott Cunningham:

So, okay. I want to set up the reader a little bit, oh, the listener, to know who you are before we dive into this, because I'm loving this thread, but I want people to know who you are. So before we get more into the case, can you tell me where you grew up, and why you got into economics? Your first, what was the touchstone that brought you into this field?

Peter Arcidiacono:

So I grew up in the Pacific Northwest. My first set of years were actually in Ellensburg, Washington, which is a town of 13,000. My dad was a math education professor.

Scott Cunningham:

Oh, okay. What university was he a professor at?

Peter Arcidiacono:

Central Washington University.

Scott Cunningham:

Okay. Okay.

Peter Arcidiacono:

And then-

Scott Cunningham:

Hey, but what'd you say it was? What was it again?

Peter Arcidiacono:

... It was math education.

Scott Cunningham:

Math education.

Peter Arcidiacono:

Yeah. So he was teaching teachers how to teach math.

Scott Cunningham:

Oh. So you've always been, it's in the family to be interested in education?

Peter Arcidiacono:

Yes. And-

Scott Cunningham:

And even this math education part. That's another way of describing an economist that studies education.

Peter Arcidiacono:

... Right.

Scott Cunningham:

Math education.

Peter Arcidiacono:

Well, my parents actually met in linear algebra class, so.

Scott Cunningham:

Oh, that's romantic.

Peter Arcidiacono:

And I've got two brothers, and they were both math majors.

Scott Cunningham:

Oh, wow. Okay.

Peter Arcidiacono:

I was the only non-math major.

Scott Cunningham:

Okay. Okay.

Peter Arcidiacono:

But I came into college, and started out in chemistry. I think, Econ PhD programs are filled with former, hard science majors.

Scott Cunningham:

No joke. Yeah, yeah. They hit organic chemistry, and then, they changed their major.

Peter Arcidiacono:

Right. And I just couldn't stand the lab. It was too social. And one of my good friends, a guy who ended up being the best man at my wedding, was a couple years ahead of me, told me I should take an economics class.

Scott Cunningham:

Mm.

Peter Arcidiacono:

And it was amazing. I think that just the way of thinking, just worked naturally for me.

Scott Cunningham:

Well, so when you say way of thinking, the way of thinking that was, can you tell me what your 19 year old self would've been jarred by? What are the specific things, that economic way of thinking, that he was noticing?

Peter Arcidiacono:

Well, it just fit with a lot of how I operated. So I view economics as a great model of fallen man.

Scott Cunningham:

Uh-huh.

Peter Arcidiacono:

Fundamentally, I was the guy who always looked for the loopholes. So responding really well to incentives. I had a keen eye for how I could game the system.

Scott Cunningham:

Right. Right.

Peter Arcidiacono:

And so, I think a lot about what economics is doing, is the dismal science, right? The reigns on the parade of well-intentioned policies.

Scott Cunningham:

Right. Right.

Peter Arcidiacono:

How are people going to get around the policies? Well, that's where I lived, was figuring out how I could game the system.

Scott Cunningham:

Right. Right. Right. So you were, this idea of that rational choice paradigm, is that what you mean?

Peter Arcidiacono:

Yeah.

Scott Cunningham:

And that-

Peter Arcidiacono:

Yeah.

Scott Cunningham:

... that people would just simply, if they have goals, those goals don't just go away with a policy. They might just continue to try to achieve those goals at lowest cost, even then.

Peter Arcidiacono:

Exactly.

Scott Cunningham:

Right. Right.

Peter Arcidiacono:

And the other studying thing, which I think, really affected why I ended up doing the research that I did, was, for me, the chemistry classes were just way harder-

Scott Cunningham:

Uh-huh.

Peter Arcidiacono:

... than economics classes.

Scott Cunningham:

Yeah.

Peter Arcidiacono:

And I'm not trying to say that any classes are easy.

Scott Cunningham:

Yeah.

Peter Arcidiacono:

But there is definitely large differences-

Scott Cunningham:

Mm-hmm.

Peter Arcidiacono:

... in every university, and what the expectations are-

Scott Cunningham:

Mm-hmm.

Peter Arcidiacono:

... across fields.

Scott Cunningham:

Yeah.

Peter Arcidiacono:

And that distorts people's behavior.

Scott Cunningham:

Mm.

Peter Arcidiacono:

So I view it, that most colleges are subsidizing students, to go into low paying fields. And how do they subsidize them to do that? They offer higher grades-

Scott Cunningham:

Mm.

Peter Arcidiacono:

... and lower workloads, smaller class sizes. All those things work, so that lots of people come in wanting to major in well-paying fields, and switch in, and switch out.

Scott Cunningham:

Right. Right. Right.

Peter Arcidiacono:

And they do so because of the incentives the universities provide.

Scott Cunningham:

Yeah. So you got interested in economics, and that's like, you're describing some sort of price theory, microeconomics. But you've also have made a career out of being such a strong econometrician in this area of structural econometrics and discrete choice modeling. How did you get interested in those topics? What was your first reaction to econometrics?

Peter Arcidiacono:

I had a very strange econometrics background. So my first year econometrics, was taught by Chuck Manski.

Scott Cunningham:

Oh.

Peter Arcidiacono:

The whole year. And so, it was lots of bounds.

Scott Cunningham:

Uh-huh.

Peter Arcidiacono:

And then, my second year, it was all John Rust.

Scott Cunningham:

Mm.

Peter Arcidiacono:

So a complete swing, right? So you go from the non-parametrics, what can you identify under the smallest number of assumptions?

Scott Cunningham:

Mm.

Peter Arcidiacono:

To what can you identify, if you want an answer something really big.

Scott Cunningham:

Right.

Peter Arcidiacono:

You got to make a lot of assumptions to make that.

Scott Cunningham:

Oh, boy. That's an interesting journey, right there.

Peter Arcidiacono:

So I actually never had the mostly harmless econometric-

Scott Cunningham:

Right. Right.

Peter Arcidiacono:

... at all.

Scott Cunningham:

Yeah. Yeah. Yeah. Yeah.

Peter Arcidiacono:

And the econometrics has always been-

Scott Cunningham:

This was Wisconsin?

Peter Arcidiacono:

... That's right.

Scott Cunningham:

Yeah.

Peter Arcidiacono:

That's right.

Scott Cunningham:

What year was this?

Peter Arcidiacono:

In the econometrics, the advances were always more, because I needed to do something to estimate my models.

Scott Cunningham:

Right. This was mid nineties? This would've been the mid nineties, or late nineties?

Peter Arcidiacono:

I'd like to say late nineties. Yeah-

Scott Cunningham:

Late nineties? Okay. Yeah-

Peter Arcidiacono:

... [inaudible 00:19:10].

Scott Cunningham:

... Yeah. Yeah. Yeah. Yeah. Okay, keep going. Sorry.

Peter Arcidiacono:

So I was thinking about my own experience, in terms of choosing a college major, and thinking about, Well, people are learning over time. They start out those STEM classes, and figured out, wow, this is a little bit harder than I expected.

Scott Cunningham:

Yeah. Yeah.

Peter Arcidiacono:

And then, moved through.

Scott Cunningham:

Right.

Peter Arcidiacono:

So I had a mind, I actually had the idea for my job market paper, my first year. And had this idea of a forward looking model, of how people choose their college major.

Scott Cunningham:

Mm-hmm.

Peter Arcidiacono:

And so, then, I go into John Rust's office, because he's my second year econometrics professor, and was describing this problem to him, that people are making decisions today, giving expectations about the future.

Scott Cunningham:

Mm-hmm.

Peter Arcidiacono:

And he says, "Yeah, I think I can help you with that." And I was like, "No, you don't understand. This is a really hard problem." And of course, John Rust had written the [inaudible 00:20:13] paper about how to estimate these types of models-

Scott Cunningham:

Right.

Peter Arcidiacono:

... And he was fantastic with me. [inaudible 00:20:20]. He didn't say idiot. You could at least look at what I do, before you come to my office. He was fantastic with me.

Scott Cunningham:

Yeah. Yeah.

Peter Arcidiacono:

And actually, the funny story about that too, is he's actually the only reason I'm an economics professor, because-

Scott Cunningham:

Oh, yeah?

Peter Arcidiacono:

... I only got into one grad school. Got rejected from much worse places in Wisconsin. It was the only place that accepted me.

Scott Cunningham:

Mm.

Peter Arcidiacono:

And the joke was that that was the year John rusted everybody in. So there were 53 of us to [inaudible 00:20:57].

Scott Cunningham:

That's awesome.

Peter Arcidiacono:

17 got PhDs.

Scott Cunningham:

Wow.

Peter Arcidiacono:

And if you look at another guy, one of my friends, I just actually found out we were actually at a conference in honor of John Rust, this past weekend.

Scott Cunningham:

Yeah.

Peter Arcidiacono:

And it turns out, that was the only place that admitted him, as well. And he's been incredibly successful too.

Scott Cunningham:

The John Rust fixed effect is filled with stories.

Peter Arcidiacono:

That's right.

Scott Cunningham:

That's really cool. That's really cool. I'm curious, thinking about what your, I want get to the Harvard and the Chapel Hill. But before we move on, you could imagine, had you gone to Princeton, or MIT, and worked with, or Berkeley, and worked with these, the treatment effects guys, like Imbens, and Angrist, and Card, and Kruger, and O'Reilly, and all these people. It's not just that your knowledge of econometrics would've been slightly different. Even the kinds of questions, that you would be asking, might be different. So I'm curious, what do you think your training and structural, under Manski and Rust, how has that shaped, not just the way you do your work, but even the types of questions that you ask, that you imagine, you might not have asked? For instance, just even thinking, modeling choice-

Peter Arcidiacono:

[Inaudible 00:22:40].

Scott Cunningham:

I'm sorry. I don't know. Did I lose you?

Peter Arcidiacono:

You froze on me.

Scott Cunningham:

Ah, I froze? Okay-

Peter Arcidiacono:

You're still frozen.

Scott Cunningham:

... I'm still frozen? Okay. There. Okay.

Peter Arcidiacono:

Now, you're back. So you're asking about what types of questions.

Scott Cunningham:

Yeah. What kinds of questions do you think you ended up being really interested in, and working on? Not just the model that you wrote down, but even the actual topics. Because I'm curious, I'm wondering if listeners could really frame their understanding of this structural, versus this causal inference, tradition. Not just in terms of the technical pieces, but like this is practically how, the work a person ends up, that you think you ended up doing, versus if you had got Angrist as an advisor.

Peter Arcidiacono:

Oh, I think it has shaped me quite a bit. I am certain that if I'd gone to a place like Chicago, I would've probably ended up working with Steve Levitt. I am naturally attracted to some of those topics, that are more of a freakaconomics-type nature. And if you look at it, we actually had competing papers-

Scott Cunningham:

Yeah.

Peter Arcidiacono:

... on discrimination in the Weakest Link game show.

Scott Cunningham:

Uh-huh. Yeah.

Peter Arcidiacono:

And I've written a couple of sports papers. So I have that in me, to think about those types of things. If I'm-

Scott Cunningham:

Topics, right? Right.

Peter Arcidiacono:

... Yeah.

Scott Cunningham:

Yeah.

Peter Arcidiacono:

I think that the Manski Rust combination did have a big effect on me, and, in the types of questions that I asked. Which is what structural brings to the table, is thinking about mechanisms.

Scott Cunningham:

Mm.

Peter Arcidiacono:

So when you think about the effect of affirmative action on outcomes, understanding why the effect is what it is, matters.

Scott Cunningham:

Yeah.

Peter Arcidiacono:

How it affects application behaviors. How is affects what majors issues. What would be those counterfactuals? And for that, I think you need some of these structural approaches.

Scott Cunningham:

Right.

Peter Arcidiacono:

Now, one of the things about those structural approaches, to say, typically involve making some pretty big assumptions.

Scott Cunningham:

Yeah.

Peter Arcidiacono:

And I think that that's where the Manski influences had on me, because I also have papers that use subjective expectations data. And I think that that is actually an incredibly promising area of work.

Scott Cunningham:

Mm.

Peter Arcidiacono:

It's quite clear that people don't know as much as they should know, when they make important decisions. Certainly, higher education being a prime example of that.

Scott Cunningham:

Yeah.

Peter Arcidiacono:

COVID really makes that clear, you know? How can it be that the people who are unvaccinated, are least likely to wear a mask? Clearly, they're operating under very different beliefs about-

Scott Cunningham:

Right.

Peter Arcidiacono:

... what's going to happen.

Scott Cunningham:

Right. Right. Right. Right. Okay. So let's move into this Harvard Chapel Hill project. So setting it up, tell me, what is the first event that happens, that makes this a case against Harvard? Not counting alleged discrimination, but the actual historical event, that leads to a need for an expert witness.

Peter Arcidiacono:

Well, I think the need for the expert witness came about, because Harvard had to release their data, in the context of the trial.

Scott Cunningham:

Mm.

Peter Arcidiacono:

So in the context of the lawsuit, the claim was there were some smoking guns that suggested the possibility, for example, of Asian American discrimination

Scott Cunningham:

That would not fit this holistic criteria, that you mentioned earlier?

Peter Arcidiacono:

Well, so, it's an interesting question, right? So you can't have with the holistic criteria, you can take race into account, but the question is whether you could take race into account, in a way that penalizes a group, relative to white applicants?

Scott Cunningham:

Yeah.

Peter Arcidiacono:

So it might be one thing to say, "We're going to give a bump for African Americans, relative to whites."

Scott Cunningham:

Yeah.

Peter Arcidiacono:

Maybe another thing to say, "We're going to give a bump for whites, relative to Asian Americans."

Scott Cunningham:

Yeah. Yeah. Right. Right. Okay. So they've had a lawsuit brought against Harvard. Harvard's had a lawsuit filed against them. What year is-

Peter Arcidiacono:

[inaudible 00:27:32]. Sorry, say it again.

Scott Cunningham:

What year would that have been?

Peter Arcidiacono:

Oh, man. I think it was back in 2015, or something like that.

Scott Cunningham:

2015. Did anybody see that coming? Or was this odd, this is just inevitable?

Peter Arcidiacono:

I think that, they were advertising for plaintiffs, students who had been rejected. So certainly, there was an intent to file such a lawsuit, for sure. And then, they had to weigh what universities to file it against. And they chose Harvard, because of the patterns on what were going on with Asian Americans. And I think UNC had more to do with the, there was some evidence in the record, from past cases, that race-neutral alternatives would work there.

Scott Cunningham:

Mm-hmm. Okay. So you get involved. How do you get selected as the expert witness? And what's your job, exactly, in all this?

Peter Arcidiacono:

So I think I get selected, I've written a couple of survey articles on affirmative action. And I view it that there are lots of nuances. So the fact that I would actually say there are nuances, as opposed to it being always good-

Scott Cunningham:

Right.

Peter Arcidiacono:

... made it attractive for them, I think.

Scott Cunningham:

Mm-hmm.

Peter Arcidiacono:

And back in 2011, there was actually a protest here, at Duke, over one of my studies.

Scott Cunningham:

Oh, really?

Peter Arcidiacono:

Yes. So that one, we were actually using Duke data, and confronting a tough fact, which is lots of black students at Duke came in, wanted to major in STEM and economics, but switched out. In exploring why they were switching out at such a higher rate, relative to white applicants.

Scott Cunningham:

Mm-hmm.

Peter Arcidiacono:

So for men, it was very extreme.

Scott Cunningham:

Yeah.

Peter Arcidiacono:

8% of white men switched out of STEM and economics-

Scott Cunningham:

Mm-hmm.

Peter Arcidiacono:

... to a non-STEM, non-economics major. Over 50% of black males switched out. And you look at that, you think, that's a problem.

Scott Cunningham:

Mm-hmm.

Peter Arcidiacono:

And once you account for the differences in academic background, prior to Duke, all those racial gaps go away. And I think what, the path to the protest to serve in the long run. So I won't get into all details of that, but I think that they didn't believe the fact at first.

Scott Cunningham:

And what was the fact exactly, that the racial discrimination, the racial bias, the racial differences vanished, once you conditioned on what, exactly?

Peter Arcidiacono:

I conditioned on academic background.

Scott Cunningham:

Oh, I see. Okay.

Peter Arcidiacono:

Course and such like that. But I think even the original effect, they were surprised by, which was that the switch out rates were so different.

Scott Cunningham:

Yeah.

Peter Arcidiacono:

And at that time-

Scott Cunningham:

But why is that a protest against you? What does that have to do with you, if you're just documenting facts?

Peter Arcidiacono:

Well, I think that the negative press headline said, potentially racist study says black students are taking the easy way out. And so-

Scott Cunningham:

Potentially racist study.

Peter Arcidiacono:

Potentially racist study. Yes.

Scott Cunningham:

This study was racist.

Peter Arcidiacono:

That's right. And I think that the issue, it actually makes a lot more sense now, than it did to me at the time. And economists thought this was crazy at the time. It's actually interesting, because I got attacked from people all over the country. It didn't make a major news flash, but within certain circles, it did. And actually, one of the people who wrote about it at the time, was Abraham Kendi. This was before he changed his name. He's not the, he wasn't famous in the same way that he is now. But the fact that I wasn't pointing the finger at the departments, I was pointing the finger, I think it was interpreted as victim blaming. It's their fault that they're switching out because they're not prepared. That's never how I would want to frame it. I would want, to me, this is, the issue is that you're not prepared-

Scott Cunningham:

You think you framed it?

Peter Arcidiacono:

No, I don't think so. But the way economists talk about things is different.

Scott Cunningham:

I know. I think that something, I think we're, a generous view is that we can't, we don't know what we sound like or something. I get into this a lot with my work on sex work, and I've, I work really hard to try to be very factual. And it, the use of words can be so triggering to a group of people. And I can never, I still can't quite articulate what exactly it is, in hindsight, that I, what word I used that was so wrong. But you feel like you would write that paper differently now?

Peter Arcidiacono:

Knowing that non economists would read it? Yes.

Scott Cunningham:

What would you do differently?

Peter Arcidiacono:

Well, I think, you have to be much more, when I say, it counts for the differences in switching behavior.

Scott Cunningham:

Okay.

Peter Arcidiacono:

The way other people hear that, is I'm able to explain why every single person switches their major, and has nothing to do with other factors. That's the reductionist claim against economists, as opposed to, on average, this is occurring.

Scott Cunningham:

Mm.

Peter Arcidiacono:

So I did a radio interview at the time, and one of the people on the show was a blogger from Racialicious, who was a regular on the show. And-

Scott Cunningham:

Yeah.

Peter Arcidiacono:

... I didn't really know anything about the show, going in.

Scott Cunningham:

Yeah.

Peter Arcidiacono:

And she spent, so she got to go first, and she talked about how problematic my study was. And the way she described it, were ways that I did not think was consistent.

Scott Cunningham:

With what the the study was.

Peter Arcidiacono:

Right. And so, my response to that, really, by grace-

Scott Cunningham:

Yeah.

Peter Arcidiacono:

... was to say, if I thought that was what the paper was saying, I'd be upset too. And then, was able to pivot into, look, we're actually on the same side on this. We want black students at Duke to succeed in the majors that they're interested in. And to that point, we need to identify the barriers that are affecting that, and what resources we can provide, to make it so that that would not be the case.

Scott Cunningham:

So what are you going to say to your old, let's say you could go back in time, 10 years to that young economist, writing this paper. Without telling him exactly what specifically to say, you can only say a general principle. As you think about writing this, I want you to think about writing it in a different way. What exactly should you be? I guess, what I'm getting at, is how would you pause, what is, what pedagogically should we be communicating to young economists, about language and audience, that we haven't been doing historically, so that we are not unnecessarily tripping people up and creating confusion?

Peter Arcidiacono:

Yeah. Yeah. I think it's really tricky, because on a lot of things, it's just very hard to have a discussion where the emotions are not involved.

Scott Cunningham:

Yeah. Yeah.

Peter Arcidiacono:

So when you speak about things related to race, and you talk about things in a very matter of factual way-

Scott Cunningham:

Yeah.

Peter Arcidiacono:

... that can be heard as you don't care.

Scott Cunningham:

Yeah.

Peter Arcidiacono:

You are not interested in fixing the problem at all. You're just explaining away why we don't need to do anything.

Scott Cunningham:

Right. Right.

Peter Arcidiacono:

And that's how, there's actually this marriage book, I really like, which is, again, I'm going to say this, it's going to come across as stereotyping. This is obviously distributions overlap, but it's called Men Are Like Waffles, Women Are Like Spaghetti. And the ideas is that men compartmentalize everything. So we're talking about this specific issue, not seeing how it relates to the broader picture.

Scott Cunningham:

Yeah.

Peter Arcidiacono:

The advice, the marriage advice I always give now, is don't try to solve your wife's problems. That's always a mistake.

Scott Cunningham:

Yeah.

Peter Arcidiacono:

And, but that's effectively, as economists, exactly what we do. We are working in the little waffle box.

Scott Cunningham:

Right.

Peter Arcidiacono:

Focused on this particular problem.

Scott Cunningham:

Yeah.

Peter Arcidiacono:

And I don't know how to change that with regard to economics papers. I really try to be very nuanced in my language and such.

Scott Cunningham:

Yeah. Yeah.

Peter Arcidiacono:

Maybe in how you motivate the paper, recognizing the racial inequities and the historical discrimination.

Scott Cunningham:

Yeah.

Peter Arcidiacono:

But there is a sense in which it will not be enough.

Scott Cunningham:

Yeah. Yeah. There's this, I can't, I just now drew a blank on the, I teach it all. I can see the slide in my deck, but there's a famous computer scientist. And he says, this principal about writing code, and he says, "Be conservative in what you do, and be liberal in what you accept from others." And it's this principle of code writing, which I guess is like, he's basically saying, "When you write code, it needs to be, the noise to signal ratio needs to be very, very low. You need to be very clear in what it's doing, in a very efficient choice to minimize this, these unnecessary errors." But when you're receiving the code, either from your earlier part of the code, or for some other foreign source, you have to change your viewpoint in that sense, because really, the goal, when you're on the receiving end of the code, it seems like your goal is to be this antenna.

Scott Cunningham:

And this antenna is trying to extract information from any meaningful information from the noise. And so, you have to have, as a listener, a certain amount of grace that tolerates that this other person may make mistakes, doesn't say it all right, goes really, really to great lengths to try to, you go to great lengths, to try to figure out exactly what the message is, and what it isn't. And it does seem like, successful communication is a, about a sender who is being clear, and a receiver that is being charitable in what they're going to allow the sender to say, unless the goal is conflict.

Peter Arcidiacono:

That's right.

Scott Cunningham:

If the goal is conflict, then obviously, you don't do that. What you do with conflict, is you find the most bad, then, it's just bad faith. It's just like, trap a person, win the debate. And sometimes, many of us don't realize who we're talking to. We don't know if we're talking to a good faith or a bad faith person. But there's limits, I think, to what an economist or anyone can do, if the person they're talking to really is not interested in connecting.

Peter Arcidiacono:

That's right. And it's interesting, because I think when I either speak publicly, or even giving seminars to economics audiences, the first part is building trust.

Scott Cunningham:

Totally.

Peter Arcidiacono:

We have the same goals.

Scott Cunningham:

Yeah.

Peter Arcidiacono:

We may have different views about how to get there. And I've got some information that may change your mind on this.

Scott Cunningham:

Yeah.

Peter Arcidiacono:

And the issue is whether they can hear the information I say, or if it's going to be ruled out because I'm a bad person.

Scott Cunningham:

Right. Well, let me ask you something. So these tests for, okay, so you correct me where my thinking is wrong. Testing for racial discrimination in admissions. I could imagine econometrics one, I get the data set from Harvard, and I run a regression of admit onto a race dummy.

Peter Arcidiacono:

Right.

Scott Cunningham:

And then, I interpret the statistical significance on the race dummy. And then, I add in more observables. In what sense is this, philosophically, what we are trying to do in the United States, legally, to detect for whatever it is that's violating the constitution. And in what sense is it a big fat failure, that's not what we're trying to do? Can you elaborate that as a multivariate regression-

Peter Arcidiacono:

Yeah. So I think, how to interpret that beginning coefficient, I don't think that coefficient has much of an interpretation, particularly in admissions, because of who applies. And that was, one of the papers that we published on this, is about Harvard's recruiting practices.

Scott Cunningham:

... Mm.

Peter Arcidiacono:

And Harvard, they recruit a lot of people. And particularly, African Americans, who simply have no chance of admission. And so, you could make it. And that could be part of the reason, right, would be, we want to appear as though when you do just that one regression with that one variable-

Scott Cunningham:

Mm.

Peter Arcidiacono:

... through affecting my applicant pool, I can always make it so that coefficient-

Scott Cunningham:

So what happening? So if I've got a university, just in real simple sense, let's say a university, if they're white, they span their, they basically task to the university, to whoever, and they say, "Get a pool of white applicants, use this rule. Get a pool of black applicants, use this rule." And it's just very, very different rules.

Peter Arcidiacono:

That's right.

Scott Cunningham:

Okay. If I then run a regression, how in the world am I going to detect racial preference in admission, when racial preference was used in the drawing up of the application in the first place?

Peter Arcidiacono:

So I think that's where, I think one of the principles that, it's not randomization for sure.

Scott Cunningham:

Right.

Peter Arcidiacono:

But one of the key principles, is how do you think about selection on observables versus unobservables?

Scott Cunningham:

Yeah. Right, right.

Peter Arcidiacono:

And so, if you can account, in the case we just described, if it was differences in test scores alone, once you account for test scores, then you could see how they were treated differently. Conditional on those test scores.

Scott Cunningham:

Yeah.

Peter Arcidiacono:

And typically, the way that works, is that when you add controls, the coefficient on the discriminated group typically goes down, because there was, because of history discrimination, that there was going to be differences in those things. That was why you had the program in the first place.

Scott Cunningham:

Right. Right.

Peter Arcidiacono:

But what's interesting in the case of Asian Americans, is it tends to go in the opposite direction. Right? So they're stronger on a lot of the observables.

Scott Cunningham:

Right.

Peter Arcidiacono:

You add controls, it looks like the coefficients becomes more negative. For African Americans-

Scott Cunningham:

The coefficient, as in, the, so if I did a regression of admit onto an Asian dummy, nothing else, it'll be positive?

Peter Arcidiacono:

Well, it depends. So it would be positive if you had nothing else, and you excluded legacies-

Scott Cunningham:

Legacies.

Peter Arcidiacono:

... and athletes.

Scott Cunningham:

Okay. So I dropped the legacy and the athletes. I regress admit onto an Asian dummy. Asians are more likely to... So when does the, so what-

Peter Arcidiacono:

When it's slightly positive and insignificant.

Scott Cunningham:

... Okay.

Peter Arcidiacono:

As soon as you add anything related to academic background-

Scott Cunningham:

So then, I put in high school GPA and zip code, and I start trying to get at these measures of underlying academic performance, observable. And that's when it flips?

Peter Arcidiacono:

Oh yeah. Yeah. This is something I just did not appreciate before the Harvard case, is how incredibly well Asian Americans are doing academically.

Scott Cunningham:

Mm.

Peter Arcidiacono:

If you did admissions based solely on academics, over half would be Asian American. That is a stunning number. All groups would go down, and Asian Americans would be the only group that went up.

Scott Cunningham:

Okay. Say that again. I didn't quite follow. So what will astound me? What would it?

Peter Arcidiacono:

So Asian Americans, they're in the low twenties, in terms of their share of admits, or something like that.

Scott Cunningham:

Yeah.

Peter Arcidiacono:

When you look at typical applicants.

Scott Cunningham:

Yeah.

Peter Arcidiacono:

If you had admissions based solely on academics-

Scott Cunningham:

Mm-hmm.

Peter Arcidiacono:

... with some combination of test scores and grades-

Scott Cunningham:

Mm-hmm.

Peter Arcidiacono:

... they would be over half of Harvard's.

Scott Cunningham:

I see. Got it. They're just, it's just such an incredibly selective group. Selective, in terms of the measures of probable performance and success, and all these things. They are, as a group, high... What's the right word? How do you, this is one of these things, we're using the languages, is really careful. I was going to say, I know economists, we have models that say high type, low type. And obviously, it's like, what's the right way to start talking about these young people? These are young people at the beginning of their, everybody comes at a difference. So what's the right, what's the loving, charitable, honest way of talking about people with these underlying differences?

Peter Arcidiacono:

Well, I think that, what happened to them before college, was such, that on average, you see tremendous differences-

Scott Cunningham:

Yeah.

Peter Arcidiacono:

... in the skills that have been accumulated-

Scott Cunningham:

Right.

Peter Arcidiacono:

... prior to college.

Scott Cunningham:

Right. Right. So there appears to be, one way you could describe it, is to say, there appears to be differences in human capital.

Peter Arcidiacono:

That's right. But I think human capital, I guess-

Scott Cunningham:

Unobservable human capital appears to be different, but it's like showing up on these observable dimensions.

Peter Arcidiacono:

... That's right.

Scott Cunningham:

Got it.

Peter Arcidiacono:

And for me, that doesn't, in any way, point the finger, and say there's something wrong-

Scott Cunningham:

Right.

Peter Arcidiacono:

... with the groups that aren't doing well on that.

Scott Cunningham:

Yeah. No.

Peter Arcidiacono:

And in fact, there's some people who argue, look, the differences in test scores, the reason African American score worse on the tests, is because of stereotype threat.

Scott Cunningham:

Mm-hmm.

Peter Arcidiacono:

And that idea is that everybody expects them to do poorly. And so, they do poorly.

Scott Cunningham:

Yeah.

Peter Arcidiacono:

To me, that's giving the K through 12 education system a pass. There are real differences-

Scott Cunningham:

Right.

Peter Arcidiacono:

... in the K through 12 education experience-

Scott Cunningham:

Mm-hmm.

Peter Arcidiacono:

... for African Americans.

Scott Cunningham:

Yeah.

Peter Arcidiacono:

That's what we need to fix.

Scott Cunningham:

Right.

Peter Arcidiacono:

We can't shy away from the real issue. And that's actually one of my big concerns with places like the UC system, saying, "We don't want standardized tests anymore." We're just going to ignore that there's a serious deficiency.

Scott Cunningham:

Yeah.

Peter Arcidiacono:

Not that the people are deficient, that the educational system was deficient-

Scott Cunningham:

Yeah. Yeah.

Peter Arcidiacono:

... for these students.

Scott Cunningham:

It's interesting. It's like, one of the papers I teach a lot, is, I know you're familiar with, is Mark Hoekstra's review of economics and statistics article, on the returns to attending the state flagship school. I've always thought-

Peter Arcidiacono:

Yeah.

Scott Cunningham:

... that this really interesting study, it feels relevant to what you're working on with Harvard and UNC, because it's about, I feel like when I was in graduate school, I came away from my labor courses, just realizing attending college is crucial. College is an anti-poverty program, as far as I can tell. You could see it in my work on crime, with the, you and I actually have some similar backgrounds. We're both interested in sex ratios and marriage markets.

Peter Arcidiacono:

Yeah.

Scott Cunningham:

But you could see the incarceration rate of African American men just plummeting, with college attainment, levels of college enrollment. But so, it's like, I graduated thinking, "Oh, well, the returns to college are important." But then, it's like, Mark's paper highlighted that there was this heterogeneity, even there. Even in these, in terms of the flagship school and Harvard.

Peter Arcidiacono:

Right.

Scott Cunningham:

And the reason why this stuff is important, I feel like it gets into these complicated things with regards to how we've decided to organize America, because the United States, we purchase goods and services using, goods and services go into the utility function. In many ways, that's the, trying to get utility functions that are virtuous and correlated with a life that's worth living, is the big goal. But we buy those goods and services at market prices, using labor income. And so, then, it always wraps back into this issue about something like Harvard or Chapel Hill, which is, some of these schools have imbalanced returns that affect labor income and quality of life, or might arguably, subjective wellbeing, as it's measured by utility. And I guess I'm just sitting here thinking to myself, if you have a group of people who are just for historical, it's not even historical accident, because they were historically discriminated against in the United States.

Scott Cunningham:

But at this point, it's a stock. African Americans have come to the table with this different kind of human capital, that's going to end up shaping all of their labor income. It's going to have massive impacts on labor income, where they go to college. It's like, I don't see how you can separate out the fact that there, we've got to decide, collectively, what exactly is the goal for these different groups of people that live here in the United States, and that one of the existing mechanisms for income, is college. And it all wraps back into this whole issue, about what exactly should the composition of the student body be, given these ridiculously imbalanced returns to each of these individual schools?

Peter Arcidiacono:

That's right. But I think that some of those things could be balanced more, if we were doing the things that were actually successful in changing the human capital-

Scott Cunningham:

Yeah. Yeah.

Peter Arcidiacono:

... upfront. And so, one of the most, it was really disappointing, in my mind, when, after Floyd, I think KIPP Charter schools decided that their motto was no longer appropriate. Be nice, work hard. And I say that, mainly because no excuse charter schools, which no excuse, that's something that you can't really say quite the same way now.

Scott Cunningham:

Right.

Peter Arcidiacono:

These schools were incredibly successful at closing the achievement gap.

Scott Cunningham:

Yeah.

Peter Arcidiacono:

They were actually very successful.

Scott Cunningham:

Right.

Peter Arcidiacono:

We could be doing that. That's where the resources ought to go.

Scott Cunningham:

Right, right.

Peter Arcidiacono:

Instead, what you see in California now, is they're getting rid of advanced classes. There's two ways to deal with an achievement gap, right? You can bring the people who aren't doing as well, up.

Scott Cunningham:

Right.

Peter Arcidiacono:

Or you could bring the people who are doing well, down. The getting rid of the advanced classes, is not bringing, in my mind, those students up.

Scott Cunningham:

Mm-hmm.

Peter Arcidiacono:

And if anything, it's providing huge advantages to people of means, because you cripple the public education system, take the path out for them to develop that human capital.

Scott Cunningham:

Right. Right.

Peter Arcidiacono:

And then, the people with resources send their kids to private schools, so that stuff isn't going to go on.

Scott Cunningham:

Right. Right. Right.

Peter Arcidiacono:

And that's where I think a lot of the discussion, we can talk about affirmative action at Harvard. At the end of the day, that's really about appearances. The people are going to Harvard are all, most of them are coming from an incredibly rich backgrounds.

Scott Cunningham:

Right. Right.

Peter Arcidiacono:

Regardless of what race. There are differences across the races. But that's where the action is.

Scott Cunningham:

Right. Right.

Peter Arcidiacono:

And that's what we typically focus on in education. But where we really need to be doing more, is for the lower income kids.

Scott Cunningham:

Mm-hmm.

Peter Arcidiacono:

And COVIDs is going, we're starting to see that that's going to be a train wreck. Our education for this kids who went to-

Scott Cunningham:

Yeah. Yeah, yeah, yeah.

Peter Arcidiacono:

... public schools.

Scott Cunningham:

Yeah. Yeah. Yeah. There's certain elasticities, that I think COVID highlights, which is that there's a, there are groups of students who, probably, their ability to substitute to the best case scenario in a very difficult situation, was really, they had a very high, they were able to do it. It may not have been, it wasn't a perfect substitute. They were able to continue to do it. And I think for some groups of students, it was a train wreck.

Peter Arcidiacono:

Yeah.

Scott Cunningham:

Just their ability to make those substitutions to whatever was required, could be anything ranging from the access to physical resources, like computing, computers, and wifi that's stable, and all these things, to, just simply, the way your brain works. Just being able to be present. I definitely think that COVID cut a mark through the students, that, it did in our family, completely cut a mark through students in weird jagged way, for sure.

Peter Arcidiacono:

But within your family, you're able to substitute in ways that other families cannot.

Scott Cunningham:

Yeah. Yeah.

Peter Arcidiacono:

And that's the catch. And I think that, I don't work a lot in the K through 12 space, so this is a non-expert opinion on that. But if my read on the studies, is if you find positive effects of, say, charter schools, Catholic schools, smaller class size, if you're going to find positive effects for anyone, it's going to be inner city African Americans. And I think that the reason that you see that, is the way family substitutes, that they're not, their families are not in as good of a position to substitute-

Scott Cunningham:

Yeah. Yeah.

Peter Arcidiacono:

... the way my family is. My kid has a bad teacher, we're going to do the bad effects.

Scott Cunningham:

Right. Right. Right.

Peter Arcidiacono:

So you're going to think, "Oh, the teacher's fine." But no, we even did the effects of that teacher, in ways that other families cannot.

Scott Cunningham:

Right. Right. Right. So what do you think is the smoking gun evidence, that that Harvard University has to... What's the smoking gun fact, that's evidence for, that's the most damning evidence for racial discrimination in admissions, that-

Peter Arcidiacono:

So racial discrimination against Asian Americans, I think that there's a, there's so many damning facts. Well, I'll start with the first one, which is Harvard's own internal offices. They have their own internal research teams. They estimated models of admissions, and consistently found a penalty against Asian Americans.

Scott Cunningham:

... Mm.

Peter Arcidiacono:

You could look at that. You'll hear people say, "Well, those are simplistic models." The fit of those models was incredibly high.

Scott Cunningham:

Yeah. Yeah.

Peter Arcidiacono:

I think. So they were explaining-

Scott Cunningham:

I think people underestimate the shoe leather sophistication that goes on in these admissions office, with developing their own internal models.

Peter Arcidiacono:

... Well, and what was striking, is Harvard's defense of this was, "Well, we really didn't understand the model."

Scott Cunningham:

Mm.

Peter Arcidiacono:

Well, what was interesting, is that those models also had whether or not you were low income, in it.

Scott Cunningham:

Mm.

Peter Arcidiacono:

And they were confident that those models, the same model, showed that they were giving a bump to low income students.

Scott Cunningham:

Mm.

Peter Arcidiacono:

It's like, you're going to interpret the coefficient one way when it's the result you like, and another way, when it's the result you don't like.

Scott Cunningham:

Right. Right. So their own models showed, so what was the penalty? What was it? It was a dummy, a coefficient on a binary indicator for Asian American, or Asian?

Peter Arcidiacono:

That's right. That's right.

Scott Cunningham:

How big was it?

Peter Arcidiacono:

And then, also, it even had stuff on the personal rating. You can see, there was charts from their office that shows, what do you know, Asian Americans on all of Harvard's ratings, are scoring either much better than whites-

Scott Cunningham:

Mm.

Peter Arcidiacono:

... or the same as whites, even on the alumni personal rating. So Harvard has these alumni interview, the students, and even on that, Asian Americans are doing similarly to whites.

Scott Cunningham:

Yeah.

Peter Arcidiacono:

And then, you see their own personal rating, based, not on meeting with the applicants. They do much, much worse.

Scott Cunningham:

Yeah. Yeah, yeah, yeah. Yeah. Well, so what does Harvard have to prove?

Peter Arcidiacono:

Well, I think typically, in something like discrimination cases, well, what they have to prove, probably depends on the judge, I suppose-

Scott Cunningham:

Yeah. Right.

Peter Arcidiacono:

... is the catch. What they were able to say at trial, were things like, "Well, the teachers must be giving them poor ratings. We don't think that Asian Americans are deficient on personal qualities, but maybe the teachers are scoring them poorly." How that is an excuse. I don't-

Scott Cunningham:

Yeah. I don't see what they're trying to... This is, I guess, where it's frustrating, because I'm struggling to know exactly what the objective function for Harvard is, in their own stated goals. What is their objective function? To create a particular kind of cohort? What is the cohort?

Peter Arcidiacono:

... Well, I think you'd get a lot of gobbledygook when it comes to that-

Scott Cunningham:

That's what I was wondering. Yeah. Okay.

Peter Arcidiacono:

... Yeah. So, but I think it is also interesting to think about the counterfactual of, if this case was not associated with affirmative action at all-

Scott Cunningham:

Yeah.

Peter Arcidiacono:

... would it have played out the same way? And to me, I think the answer is no. Honestly, I don't think Card even takes the case.

Scott Cunningham:

Mm.

Peter Arcidiacono:

I think it would've been a much better... Your worse look for Harvard than it was. I think that it was a bad look for Harvard as it was, but because of who brought the case, and because of its ties to affirmative action, that gets back to that waffle analogy, right? If you look at it in the context of the waffle, there's just simply no argument in my mind, for the way they're treating Asian Americans.

Scott Cunningham:

Mm-hmm.

Peter Arcidiacono:

It's a clear cut discrimination case.

Scott Cunningham:

Mm.

Peter Arcidiacono:

And if you just put it in a different context, it would just be completely unacceptable. Imagine Trump Towers having a discrimination suit brought against them by black applicants. And the defense being, "Look, it's not that we're discriminating against black applicants. They just happen to score poorly in our likability rating."

Scott Cunningham:

Mm-hmm.

Peter Arcidiacono:

That would be outrageous.

Scott Cunningham:

Yeah.

Peter Arcidiacono:

There would be protests. This is because it's tied to that third rail of affirmative action.

Scott Cunningham:

Yeah.

Peter Arcidiacono:

But to me, the judge could have ruled, "Look, you can have affirmative action, but you got to stop discriminating against Asian Americans relative to whites.

Scott Cunningham:

So then, if you could fill up half of Har... So is this what the thing is? Harvard, as a university, collectively, however this ends up being decided, collectively, they have a preference over their student composition.

Peter Arcidiacono:

Right.

Scott Cunningham:

And that preference is discriminatory.

Peter Arcidiacono:

Their preference, I think, lines up with Kendi's in some sense. They would like to have their class look like the population.

Scott Cunningham:

They would like to have it look like, that they would like 13% African American, whatever percent, what is it, Asian American is what, five, is single digit?

Peter Arcidiacono:

Yeah.

Scott Cunningham:

Yeah. And they would like to have a balanced portfolio of Americans.

Peter Arcidiacono:

And, but even that, I think, is giving Harvard too much credit, in the sense that, what we choose to balance on, we choose to balance on skin color.

Scott Cunningham:

Right.

Peter Arcidiacono:

You're not balancing on income.

Scott Cunningham:

Right.

Peter Arcidiacono:

You're not balancing on parental education.

Scott Cunningham:

Yeah. Yeah.

Peter Arcidiacono:

A whole bunch of other things you could've balanced on. Why-

Scott Cunningham:

Yeah. There's like an infinite number of character. Every person is a bundle of, just almost an infinite number of characteristics. And it's not practically... Yeah. Yeah.

Peter Arcidiacono:

... If you really want a representative class, then you do a lottery among high school graduates.

Scott Cunningham:

Yeah, yeah, yeah. Exactly.

Peter Arcidiacono:

That would be the only way.

Scott Cunningham:

That would be the only way, the only way it would be to have a randomized student body. Okay.

Peter Arcidiacono:

Do you feel like ask this about, was by somebody from a class at Duke, about how would you make the admissions process more equitable?

Scott Cunningham:

Uh-huh.

Peter Arcidiacono:

And I'm like, it's a selective admissions process. I don't even know what that-

Scott Cunningham:

Right.

Peter Arcidiacono:

... means. Even a process where you did the lottery, why is that equitable, because you've got the winners and the losers? The lottery. We're not equalizing outcomes for everybody. We're equalizing X anti.

Scott Cunningham:

Yeah. It's like, this is all this comp, this is this deep collective choice, social preferences questions about... And it's weird. I guess we're talking about this at Harvard, because we believe that Harvard University will literally change a kid's life, more than going to University of Tennessee, Knoxville, or something like that. Right? That's why we're having this conversation.

Peter Arcidiacono:

Yeah. I think that that's the perception, that it will literally change their kids' lives.

Scott Cunningham:

Yeah. Yeah.

Peter Arcidiacono:

I'm not totally convinced that of there being massive gains-

Scott Cunningham:

Right.

Peter Arcidiacono:

... relative to the counterfactual for-

Scott Cunningham:

Yeah.

Peter Arcidiacono:

... at that level.

Scott Cunningham:

Right.

Peter Arcidiacono:

I think, when you're at the margin of going to college or not-

Scott Cunningham:

Yeah.

Peter Arcidiacono:

... that's the big margin.

Scott Cunningham:

That's the big margin. Yeah. Yeah.

Peter Arcidiacono:

College quality effects, I think get undone a little bit by college major effects.

Scott Cunningham:

Right, right, right.

Peter Arcidiacono:

So yeah. I think that's a real valid question, about whether it's worth it to be paying the huge sums of money to go to just a slightly better school.

Scott Cunningham:

Okay. I want to conclude with this. So you've now spent many years working on this, going deeper into administrative data, about questions that, about phenomenon that we had only speculated about, possibly. What are the top two things that you learned, that, being a career economist that had worked on topics in education and affirmative action, what are the top two things that you learned, that you just literally, it was, you just, would've never learned, had you not been involved in these two cases, that are of real, that you think that matter for other people too?

Peter Arcidiacono:

Right. So I think one of the things, is that the Supreme Court rulings in the Michigan cases, were really a handout to elite private schools.

Scott Cunningham:

Oh, interesting.

Peter Arcidiacono:

Then, when you think about doing holistic admissions, a place like Harvard can do that in a way that UNC cannot.

Scott Cunningham:

Mm.

Peter Arcidiacono:

UNC is effectively, and you can see this in our models, they are more formulaic.

Scott Cunningham:

Mm.

Peter Arcidiacono:

I don't even know how Berkeley's doing admissions now, without test scores. Think about how many applications they get, with the resources that they have. I just don't know. The other part to-

Scott Cunningham:

Wait, [inaudible 01:08:39], Peter, are you basically saying that Harvard and elite universities get, first of all, a profound amount of information about each student, and they get this incredibly right tail applicant, that they don't even need test scores, to find the students that are going to be successful?

Peter Arcidiacono:

Well, I think they can get information in a way that the public schools can't, because of the resources. So you won't have to take those test scores, but they're going to see, you might have things like winning a science fair-

Scott Cunningham:

Right.

Peter Arcidiacono:

... in other proxies.

Scott Cunningham:

Right.

Peter Arcidiacono:

And that's information that the public schools aren't even going to really collect-

Scott Cunningham:

Yeah, yeah.

Peter Arcidiacono:

... because they don't have the resources to do that. You can see that, just in the number of letters Harvard requires versus UNC.

Scott Cunningham:

The fact that any university would ever voluntarily say, "We're not going to collect this informa, or make decisions based on this anymore," you should already just automatically think, the only people that would ever voluntarily do anything, is because they don't need that thing.

Peter Arcidiacono:

Right.

Scott Cunningham:

Right. They're endogenously sorting into something that, probably, they incur almost no cost of doing it. So an elite university that drops some sort of admission criteria, probably, they've got something just as good sitting right there.

Peter Arcidiacono:

Well, I think, I'm not sure if that's something just as good. I think it's also a protection against an Asian American discrimination suit, because Asian Americans are just doing so well there.

Scott Cunningham:

Right.

Peter Arcidiacono:

You need to take it out of the criteria. And that probably brings me up to the second point, which I think is a huge one. I think holistic admissions favors people from privileged backgrounds. And I'd say that even more so, after going through these cases.

Scott Cunningham:

Yeah.

Peter Arcidiacono:

We think about test scores as being unfair, because of coaching, and they're correlated with income. And that, of course, is true.

Scott Cunningham:

Yeah.

Peter Arcidiacono:

But the other things are even more unfair. Harvard has an athletic rating for non recruit athletes, and the people who score best on that are white legacies.

Scott Cunningham:

Mm.

Peter Arcidiacono:

And why is that? Because part of that, is are you likely to walk on to one of Harvard's sports teams? Well, Harvard sports, normally we think about sports as being more of an equal opportunity thing. That's not true at Harvard. Harvard offers more varsity sports than any school in the country.

Scott Cunningham:

Mm.

Peter Arcidiacono:

And what's a marginless sport? It's sailing. So sailing, who does sailing? People who are coming from really rich families.

Scott Cunningham:

Yeah. Right.

Peter Arcidiacono:

That was another stunning thing, is that I think we really need to rethink college athletics. It's one thing to think about Duke's basketball team-

Scott Cunningham:

Right.

Peter Arcidiacono:

... as a pathway where, sure, of course, we're going to admit you. You're... But it's another thing, in my mind, to think about a sailing team as being, I don't have to be particularly strong academically, as long as I'm on the sailing team.

Scott Cunningham:

Right.

Peter Arcidiacono:

That seems like a pathway for the rich.

Scott Cunningham:

Mm. Mm. Well, how much longer do you think you'll be involved in this? This is, you've got now a treasure trove of a data too, right? You have a lot of questions that you're going to be probably mining for a long time, just even your scholarly career, outside-

Peter Arcidiacono:

I wish. If I had access to the raw data.

Scott Cunningham:

Oh, you don't?

Peter Arcidiacono:

I don't.

Scott Cunningham:

What do you, so what do you use? How can you be an expert witness without the raw data?

Peter Arcidiacono:

Well, I had the raw data at the time I wrote my reports.

Scott Cunningham:

Oh.

Peter Arcidiacono:

So all my reports have tables from that.

Scott Cunningham:

Got it.

Peter Arcidiacono:

And so, we've published four papers, effectively, out of my reports and out of other things revealed in the trial.

Scott Cunningham:

Yeah.

Peter Arcidiacono:

They actually tried to get the database admitted into evidence, and not surprisingly, Harvard objected.

Scott Cunningham:

Yeah, yeah, yeah. That would've been, once it's in evidence, anybody can have it.

Peter Arcidiacono:

That's right. My fear with a lot of this stuff, is universities will selectively release their data to people who will get them the answers that they want.

Scott Cunningham:

Mm.

Peter Arcidiacono:

And no one else will be able to look at it.

Scott Cunningham:

Yeah. Yeah. Yeah.

Peter Arcidiacono:

It's a serious problem. It's like pharmaceutical companies being the only ones allowed to evaluate the trials of their drugs.

Scott Cunningham:

I bet you, the Asian American community was, has been, I bet this has been very troubling.

Peter Arcidiacono:

Oh. And confirming for them. So that, I didn't really know about this. But then, I talked to some of my former Asian grad students, and they're like, "Yeah, this is, we've known this has been going on for a long time." And I think what, to me, what's probably more troubling to them, is not that it's going on, but the fact that now that it's been exposed, how comfortable people are with it. That's disappointing.

Scott Cunningham:

Yeah. Yeah. Yeah. It's so nice to talk. We met a long time ago. You came and presented your terms of in, I think it was, no, I don't think you presented the terms of endearment paper. I think that you presented something else, but you were going to. But Chris Cornwell had invited you to Georgia while I was a graduate student. And we, I was in Chris's office with you and, because we had this marriage market. I had a marriage market dissertation, and-

Peter Arcidiacono:

Yeah.

Scott Cunningham:

... and you had this other thing. And it's nice to meet again after probably 16 years.

Peter Arcidiacono:

Yeah. It's been a long time.

Scott Cunningham:

Been a long time. Yeah. Well, good luck with everything. Thanks for sharing everything about this. This has been really interesting for me. I hope for other people too.

Peter Arcidiacono:

Oh, I love doing it. I really appreciate you having me on. When I look at the other people you have on, it's like, wow. This guys, which one of these doesn't belong?

Scott Cunningham:

All right. You have a great day.

Peter Arcidiacono:

You too. Talk to you later.