Skip to main content
Faculty Viewpoints

Understanding the Economics of Education

College isn’t for everybody—but who is it for? School choice systems are too complicated—so how can they be improved? Yale SOM’s Seth Zimmerman takes on real issues facing students, parents, and schools and uses the tools of economics to offer data-driven answers.

Q: What project are you undertaking with your academic career?

I study the economics of education. My goal is to understand how education shapes big-picture outcomes like inequality, upward mobility, and economic growth. I’m looking across educational levels from early childhood through college to think about the value of education both as instrumental to our economic lives and as an intrinsic good. I want to help provide evidence for which education policies are effective in promoting the ambitious goals that we set for our education system.

Q: A number of people reading this will be asking, is college worth it?

On average, yes. It’s not worth it for everybody, but it’s a very good bet. I think people underrate it. If you look at the data, in 2019, the wage gap between college grads and those who didn’t graduate college was near a 100-year high. So while it’s true costs are going up, on average the return is really big.

Q: Given that college is expensive, the “on average” is a significant qualifier for those who are making choices as individuals.

That’s a real concern. For governments looking at the total student population, the average return of a college degree really is the relevant thing, and the data show it clearly has real value. But for an individual, what matters isn’t the average experience; it’s their own experience.

There’s a zeitgeist that says, ‘Let’s face it: college just isn’t worth it for most people. You’re better off training in a trade.’ While I agree that college isn’t for everyone, I wanted to address it as an empirical question.

Inevitably, some people will see returns outside the average—both below and above. And for individuals who accept the risk of taking on debt and something happens so that they don’t get a degree, “debt and no degree” is what’s relevant.

It used to be the states funded public higher education to a great extent. Since the 1990s, the cost of college has shifted so that it’s borne by students rather than state taxpayers. I’m concerned that we’re expecting individuals and families to take on risks they’re not well-equipped to bear. Even so, when asked whether more people should go to college—given the way we have our educational system organized now—my answer is yes.

Q: The other qualifier is that “college is not for everybody.” Would you explain how your work tries to put some data behind who does benefit?

There’s a zeitgeist that says, “Let’s face it: college just isn’t worth it for most people. You’re better off training in a trade.” While I agree that college isn’t for everyone, I wanted to address it as an empirical question. I looked at students who didn’t do incredibly well in their high school grades or on their standardized tests. Does it make sense for people who aren’t that well-prepared academically to go to college?

You can’t answer that question just by looking at students who go to college and comparing their earnings to those who don’t. You need to find a way to look specifically at the outcomes for minimally qualified students who decide to go to college. I looked around to find a circumstance that provided a natural experiment. I found it in Florida.

The least selective campus within the state system—which at the time was Florida International University—has a sharp grade cutoff. I was able to compare students admitted with grades just above the cutoff with students just below the cutoff.

What I found surprised me, at the time. Students who barely met the admissions criteria graduate at rates similar to the overall population of students at that university. Students who were just admitted saw their earnings rise by 22%, or about 11% for every additional year of college they attended. This is similar to what we see on average in the broader population, and more than enough to outweigh the costs of attending college.

It changed my mind about who benefits from college. I think it’s maybe changed some other people’s minds, too. Added to that, there have been a number of other papers since that found similar results.

It got me thinking, OK, marginally prepared students, that’s one important parameter, but it’s one of many in the higher education system.

Q: You’ve also looked at who benefits the most from attending elite colleges.

I was at Yale, so it was on my mind. Graduates of elite institutions end up in top roles, but what’s attributable to the students selected and what’s the causal effect of the college itself?

I was curious to know whether we could apply a similar quasi-experimental approach. It’s tricky in the U.S., because the top schools have admissions processes that aren’t typically amenable to isolating any single variable. But in Chile, there’s a set of elite schools that produced a big share of CEOs, government leaders, and very high-income earners, and the admissions process is based on a set of scores that are publicly observable. I was able to get insight into the role of elite higher education in producing top outcomes.

Q: What did you find?

For men who went to one of a relatively small set of fancy private schools, getting into one of the top programs of study at an elite university gave a big boost to their chances of rising to the very top of the business leadership and income levels. For example, into the highest 0.1% of incomes.

However, for women and for men who didn’t go to one of those private schools, graduating from the top programs at the top universities didn’t lead to the very top jobs or incomes.

To say it slightly differently, these schools are causally important in attaining top positions, but, at least in Chile and for the business-focused positions I studied, those benefits only go to some. Rather than reducing inequality, elite schools tend to expand gaps by baseline socioeconomic status. This is really quite a different finding than you see at the bottom part of the college distribution, where it seems like public universities create a leveling effect. It’s also different from what I found when I looked at other fields of study, such as medicine, which can lead to high incomes but typically not to the very top of the distribution or to top corporate roles.

Q: Why the difference?

There’s suggestive evidence that people are meeting at college and forming bonds that are valuable later. If it’s networking and social interactions at elite universities that generate these top roles, that would be an important thing to know because it would shape what we think the value proposition is for universities.

It’s a common trope—what’s valuable about attending an elite school is the Rolodex as much as what you learn. But again, is there evidence to back up this broadly held belief? In Chile, there wasn’t much in the data to tell us precisely what was happening. Or why we saw this differential effect by group.

Trying to understand, I worked on a project with Valerie Michelman, who now is a post-doc at Brown University, and Joseph Price of Brigham Young University.

Historically, Harvard kept great records of who was in what club. So we examined the impact of being chosen to join the school’s most prestigious social clubs in the 1920s and 1930s. Did differences in who you socialized with and what clubs you belonged to shape long-term outcomes? We focused on careers in finance as an outcome of interest, along with earnings. What we found both confirmed and gave context to the previous work in Chile.

Joining exclusive social clubs in college led to big differences in earnings. And students from a small number of fancy private schools were more likely to join these clubs.

It added to the evidence that in certain contexts the social element of college is important. And it may be differentially available to people from different parts of the socioeconomic status distribution.

Q: You have another line of research connected to the New Haven Public Schools.

I got my PhD at Yale. I started working with the New Haven Public Schools early in graduate school. At the time there was a massive school construction program. Everyone saw the buildings going up. I was curious whether investments in capital improvements led to better educational outcomes.

There wasn’t a compelling answer in the literature, so I worked on finding an answer with Chris Nielsen, who was a fellow doctoral student then and is now a professor in the economics department at Yale and my frequent co-author.

We discovered that school construction paid off with improved standardized test scores, increased school enrollment, and neighborhood home price gains.

In doing that research, we got to know people inside the New Haven Public Schools—practitioners and families. We came to understand the challenges they face and that has led to a long-running and deep collaboration.

Q: That’s how the work on the mechanics of school choice came about?

Right. This is work with Chris Neilson and also with Adam Kapor, who was another PhD classmate at Yale and is now a professor at Princeton, as well as with Felipe Arteaga, a PhD student at Berkeley.

School choice is a confusing process. When we started, nobody had done the work to see whether families participating in the choice process had the information they needed to put together their best applications to the school choice system. The approach the schools used asked families to solve a complicated strategic problem in order to submit their best application. They needed to understand not just which schools they liked, but also their chances of getting into each school given their application, which in turn depended on knowing the system rules and who else was applying to what school. We conducted surveys of participating families and found they generally did not know all of this. The system was frustrating and expected too much.

Our research suggested it made sense to change to an approach that didn’t demand an informed strategic method of families. We worked with New Haven to improve the algorithm. But once we got the back end—the algorithm—right, then it became clear we needed to make the front end work better.

The problem is that learning enough about a school to want to send your kid there can be hard, and you may not want to spend the time to do it if you think you’re going to get into a school you already know about and like. But what if you’re wrong about your chances? Then you might not apply to enough schools, and you might not get into any school on your application. We were able to implement a system that warned parents that they were submitting applications where the student was unlikely to get a spot. It created an opportunity to revise the list, and many parents did.

We also built a relationship with education officials in Chile and worked with them to apply the approach in Chile’s national choice system. A crucial step here was that Chris founded an NGO to provide school choice services, which let us do all of this at scale.

Q: This research tackles questions schools are struggling with in the moment.

It’s an approach that works for finding important questions. Talk to people involved in school districts. Listen to the issues they are grappling with. See what the scientific literature has to say. If there aren’t good answers—and in my experience, there’s rarely already a compelling answer—it’s a good place to make a contribution, both to the scholarship and to evidence-based guidance for schools.

And if you go back to them with an answer, they’ll often be receptive because they were involved in the process of generating the question itself.

Q: Education has been around for a long time. Why aren’t there good answers available?

For many reasons. These are hard things to learn about. One challenge is data and measurement. It’s only relatively recently that we began to assemble the kinds of large-scale data sets that permit us to answer these kinds of questions.

Even aside from the data, this is a world where there are so many policies that interact with each other. It’s really hard to unpack the underlying structure of causal relationships. Is this one policy causing this one outcome? Beyond that, context matters. Things that work in one place may not work in another. Knowledge is contingent—place to place and over time. The challenges that education policymakers face in 2024 are in some ways similar to the ones in 2004 or 1984, but in some ways they’re different. Things that worked 20 or 40 years ago may or may not work now.

And after all that, when research offers evidence for a certain policy, as schools move to implement, choices made by administrators, teachers, families, and students likely modify the recommended policy. The result may augment or diminish the desired outcome. That requires additional study to figure out.

There’s this very challenging process of knowledge creation, coupled with a constantly moving target that makes it difficult for people trying to confidently make education policy. So even as we’re making progress, the questions we’re facing are always evolving.

Q: Is that exciting, daunting, frustrating…?

I feel very lucky to get to do this work. Curiosity is a driving force for me. It’s rewarding to come to a point where I understand one little part of the terrain in a way that I didn’t before. Maybe in a way that nobody has before, at least for this exact application.

At the same time, it’s very hard to translate research into policy. You can do good research and for various reasons it may not actually shape the world very much. That’s reality. But it’s so meaningful when policymakers do act on knowledge I helped create. If I can hit that benchmark a few times in my career, that’s pretty good.

Q: Education policy isn’t just about the curriculum for a given subject area or the books that go into the school library. How do you think about the context of your work?

I think the purpose of education is to promote human flourishing through a variety of means—economic opportunities, enjoyment of life, enjoyment of the educational process itself, through participation in civil society, and so on.

So much of what we see happening in society is, in some way, downstream of what happens in schools. That’s what makes this such an important area to study. Research on education can be a lens on many other elements of what’s happening in the broader world. There’s a lot to be done. There always will be.

Q: Why did you choose Yale SOM as the place to do the work?

Business school students tend to be practical. They want to go out there and do stuff, change things. That’s what I want to do too.

On top of that, The Broad Center, which is dedicated to strengthening public education and fostering career development for educational administrators, makes SOM a great place to do education research.

Q: What’s next for your work on higher education?

My previous work at elite institutions suggested that it’s hard to dislodge entrenched power structures. My more recent work, which is still in process, suggests that over multiple generations, efforts to ensure widespread opportunity pay off.

Again, the earlier work in Chile found that those who reached elite universities without attending elite private schools didn’t become CEOs or top earners at the same rate as those who did attend private schools. However, the children of those who hadn’t attended private schools are more likely to attend the fancy private high school and have friends who are high status. It seems that those kids are better equipped to be part of the higher social status and better able to integrate in elite universities.

In some cases, the answer to reducing inequality is that you have to be patient. And that’s hard because we only live once.

Q: What about the work on public-school systems?

I’ve been working to understand how our education system impacts not just students but their families. One question I’ve been looking at is how access to early childhood education affects labor outcomes for parents.

And I was recently appointed to the Connecticut State Board of Education. As a researcher, I believe in implementing evidence-driven policy. My hope is to shape policy for the better, and also that this volunteer role will help me understand what evidence is relevant and helpful in making policy.

Department: Faculty Viewpoints