How Does Income Affect Where Students Go to College?

Tim Lloyd: Welcome to the Timely Topics podcast from the St. Louis Fed. I’m your host, Tim Lloyd.

And with me today is St. Louis Fed economist Ricardo Marto. Ricardo, thanks for joining the show.

Ricardo Marto: Thank you for having me.

Lloyd: You’re here to preview some really interesting research that you’ll be publishing soon on how income doesn’t only affect where students go to college, but if they even apply in the first place. And before we get into some of your findings, I do want to ask, what prompted you to start looking into this topic?

Marto: I started to be interested in the functioning of the college market around 2017. Why 2017? There was a very popular New York Times article published on college mobility. In that article, they basically had really nice graphs of who was going to where across U.S. colleges. And the article was based on research done by Raj Chetty, John Friedman, Emmanuell Saez, Nicholas Turner and Danny Yagan, economists at top U.S. institutions. And they show that essentially elite colleges in the U.S. were filled with very, very wealthy students, and very few of the lower income students were actually enrolled in those top colleges. So, I wanted to understand why. Why was that the case? Was it because of funding this course? Was it because colleges were not admitting them in the first place? And so that’s the background for why we started this project.

Lloyd: That’s a huge question. And I’m sure you needed lots of data to begin to try and even answer some of those questions. So where do you start looking for answers?

Marto: Well, like anybody else, we googled. And we did find that the U.S. Department of Education has a center called the National Center for Education Statistics. It’s a federal agency that essentially aggregates all the statistics related to education. And that was our starting point. They have different data sets tracking children, high schoolers or college graduates. We focused on in this project one particular data set, which is the High School Longitudinal Study. The survey followed, essentially, about 24,000 ninth graders from more than 900 high schools in the U.S. And they followed them through time, through their high school, right before enrolling in college or applying to college, while they’re in college. And now there’s a new wave where we know a bit more about their early career experiences.

The good thing of this data set is that there’s a huge focus on the college decision making. So, we know essentially where students are applying, where they got in, where they ended up enrolling. And so, it gives us a very good perspective of the students. The good thing of this data set is that it’s nationally representative. And the data comes from a mix of administrative records. So, we know a lot about their high school GPAs, their test scores, and so on, but also information about students themselves, how they perceive the entire process. And also information about their parents, what’s their wealth, how much income they have. We have this huge data set which allows us to look at parental income, test scores, high school GPAs, and so on. We had all the ingredients for us to really start digging into this topic.

Lloyd: Well, let’s start getting into your findings. What did you discover about why low-income students are maybe not attending top colleges?

Marto: The first step was to essentially look at this huge number of colleges in the U.S. We focus essentially on four-year colleges and we classify them in essentially two buckets: One, which we’re calling the highly selective ones. These are essentially the Ivy League colleges, the flagship universities. And then the less selective ones, which probably easier to get in. So, what we do is then look at the statistics across who applies where across these two buckets of colleges. And what we find is that top students, those who scored in the top deciles of the SAT score distribution who were wealthier are more likely to attend a highly selective college than their peers in the same decile of SAT distribution, but whose parents earn significantly less. And that holds even after controlling for other observable such as high school GPA, gender, race, the regions where the students leave, and so on. So, this was a bit of a surprise, since we expected that essentially SAT scores would dictate essentially where you would effectively get in. And in contrast, parental income shouldn’t matter that much. So, that was the first surprise that we got.

Then we dig a little bit further. There are three reasons for why students are not attending a highly selective college. First, they have to apply. If the don’t apply they won’t enroll. That was the first hypothesis, look at where students are applying.

Then the second one is well, maybe they apply, but they don’t really get in there. They’re unlucky or they don’t meet the threshold. And then the third one is, well, maybe they are admitted by these highly selective colleges, but for one reason or another, they prefer not to enroll and enroll in less selective colleges.

Lloyd: Well, let’s keep digging into those findings. You mentioned the three big questions you’ve got. Either they don’t apply in the first place, they apply and they don’t get in. Or maybe they were admitted, but for one reason or another decide not to enroll. So, what do we know about that?

Marto: So, what we find is that among the top scoring students. So again, students at the very top of the SAT score distribution were less likely to apply to selective colleges relative to their peers who were in the same score percentile, but from parents who will make a lot, were essentially wealthier. So, to give you a bit of an idea, for instance, an increase in family income of about $80,000 is associated with an increase of 10% in the probability of a student applying to a highly selective colleges. And that takes care of all other potential funding factors that I mentioned before. In contrast, what we find is that there are very little differences in terms of admission rates. So, it seems that really the problem starts not at the college level, but at the student themself not really applying. What we see in terms of admissions is that if anything, they appear to be U-shaped.

What does this mean? It means that along the income distribution, top scoring students who are poorer are more likely to be admitted to a highly selective college than a peer with middle income. And also, it’s also true that wealthier students were top scorers are also more likely to be admitted to a highly selective college than appears with simply middle income with a similar test score.

Lloyd: What about low-income students who apply and get in to highly selective schools? Are they equally likely to enroll?

Marto: That was another surprise for us. Actually, what we find is that lower income students were admitted to a highly selective college are indeed less likely to enroll than their wealthier peers. What we see is that they instead end up enrolling in a less selective college despite being accepted into a top college. And the reason for why they end up choosing the less selective college relates to their expectations about the cost of attendance. So either, they got very good financial offers from these less selective colleges, or they thought at the beginning when they were applying that they would get a lot more financial aid at the highly selective colleges then they effectively got. In the end, they look at the pros and cons and prefer to enroll in the less selective college.

There’s also another, smaller fraction of students who decide to delay entry or not even attend the highly selective college despite being admitted. And those students were effectively from a lower income background, tend to instead go to the military or have some family issues that they have to take care of before making a choice about where they want to enroll.

Lloyd: You just brought up some financial reasons that might be driving some of this decision making. So, what do we know about financial aid?

Marto: To study this question we built an equilibrium model of the college market that essentially resembles the one we have in the U.S. So, students will differ in terms of parental income have to choose whether to apply to college or not. And if they do, they have to choose essentially which college they want to apply to. At that stage, there is uncertainty about admissions and the cost of attendance. And in response colleges have to choose whom they want to enroll and how much financial aid they want to offer.

Students in return, then, are informed about their admission status and financial aid packages, and then decide where to go. What we do in the model is we simulate an increase in Pell Grants. Pell Grants are federal grants targeted at low-income students. At the period we’re looking at the Pell Grants were about $6,000. Which is not a lot if you consider the cost of attendance, at least at top colleges. Even though now if you’re poorer and you attend, Harvard or other elite universities, you get almost a free ride. But at the time of when the students were survey that was still not the case. So, what we do in this quantitative model we increase the Pell Grants to about $25,000, which is close to the average tuition, reported by the students. What we see is that these low-income students benefit the most from this policy. They’re more likely to apply to a highly selective college in the beginning.

There is a caveat, however, which is that colleges see that more students are applying. And in response they end up raising their tuition fees in response.

Lloyd: We were just talking about financial concerns, but I want to kind of switch gears because you also study what effect removing standardized test scores as a screening mechanism might have in these admission rates. What did you find there?

Marto: Absolutely. That’s another important topic that we address in this project, which is even though test scores are somewhat imperfect and often times are associated with preventing low-income students from attending colleges. We find that removing them effectively lowers applications from low-income  students who are top students. And the reason is that by removing these test scores, colleges prefer to take fewer risks with lower income students. And instead, they start admitting wealthier students who can pay the sticker price who need less financial aid. So, removing these test scores has this perverse effect of harming those they intended to help.

And so, removing them, might not be, the best policy. In fact, what we find is that having them in place allows top scoring students to signal their ability and in fact get ahead of the pack. Even though their test scores might be slightly lower than their wealthier peers, it’s still an instrument for them to show they are they are committed to succeed in college.

Lloyd: All these questions and all these findings bring up more questions. Right. That’s how research works. So, what do you have planned in the future on this topic?

Marto: That’s an excellent question. So still within the scope of this project, I think there’s another big issue that we’re now trying to address or understand, which is whether low-income students, who are also top students, have a harder time believing their own abilities.

In some sense it might be they’re not applying not only because they think it’s too expensive to attend college, but maybe because they think they will not fit in or not be successful or eventually drop out.  And so, they have this misperception that college might not be for them, or that elite colleges might be outreach for them. And so that’s the next step in this project is to really to look at whether that plays a role in explaining the application sorting that we see across the income and ability distribution.

Lloyd: This has been fascinating. And whenever you’re done with that part of your research, you just got to promise to come back and talk to us again.

Marto: Sounds good, I will.

Lloyd: If you’re interested in the work of Ricardo and other St. Louis Fed economists you can find it at stlouisfed.org. A reminder that you can subscribe to Timely Topics on Spotify, Apple Podcasts or anywhere you get your podcasts.

And from the St. Louis Fed, you’ve been listening to Timely Topics.