Day 13: Athletics
A series following my new class: The Economics of Higher Education
On to Athletics!1 Why do small schools have DIII athletics? Why do some have DI programs? Why not more of them? How much are students really willing to pay for athletics at their college? And do athletes do better or worse than non-athletes on the job market?
Today’s readings
Moody, J. (2024) Boosting the bottom line through athletics. Inside Higher Ed.
Jacob, B., McCall, B., & Stange, K. (2018). College as country club: Do colleges cater to students’ preferences for consumption? Journal of Labor Economics, 36(2), 309-348.
Amornsiripanitch, N., Gompers, P., Hu, G., Levinson, W., & Mukharlyamov, V. (2023). No Revenge for Nerds? Evaluating the Careers of Ivy League Athletes (No. w31753). National Bureau of Economic Research Working Paper.
Moody, J. (2024) Boosting the bottom line through athletics. Inside Higher Ed.
This article is meant to set the stage for a conversation about how athletics fits into the business of liberal arts colleges. Collegiate athletics both cost money and generate revenue. The explicit costs are obvious (facilities, coaches salaries, travel budgets, etc.) and there are probably smaller implicit costs too (e.g. if athletes are admitted with lower academic standards, the institution may have to add remedial classes and/or academic supports to match). Similarly there are obvious explicit revenues (ticket sales, TV deals) and less obvious but sometimes large implicit revenues too (most notably, athletes who enroll and pay tuition; also, non-athletes who enroll because they want to enjoy watching college sports even if they are not playing themselves).
At large universities with lots of D1 sports, some of those D1 sports are net explicit revenue positive. Most are not. This article talks about how big schools are starting to cut some D1 sports. It doesn’t discuss how big schools are thinking about how changing their sports atmosphere might change demand for their “college experience.”
At small colleges with lots of D3 sports, many D3 sports are often net revenue positive because of the tuition revenue that athletes are willing to pay so they can play their sport in college. D3 colleges realize this even if they can’t quantify it, and thus many are adding D3 sports at the same time big schools are dropping D1 sports.
Jacob, B., McCall, B., & Stange, K. (2018). College as country club: Do colleges cater to students’ preferences for consumption? Journal of Labor Economics, 36(2), 309-348.
Ok, so one reason why colleges have athletics is because it draws in students. But just saying that qualitative principle isn’t enough for economists. We want to quantify the effect. That’s what this paper does!
This paper uses survey data of incoming college first years in 1992 (National Educational Longitudinal Study) and 2004 (Educational Longitudinal Survey) to estimate students’ willingness-to-pay for different aspects of the college experience. The surveys ask students where they got in and where they decided to go, so they can compare the attributes of the colleges the students did go to compared to the ones they didn’t.
These authors don’t directly look at athletics alone, but instead lump athletics in together with other “amenities.” They find that students are willing to spend 0.14-0.28% more on a college that spends 1% more on amenities like student services, dorms, food, and sports. By comparison, students are willing to pay 0.7-1.1% more for every 1 percentage point of improvement in the average SAT scores of the student body (there’s those peer effects again). The results show that before controlling for reputation, students are willing to spend 0.24% more for every 1% additional spending on instruction; after controlling for reputation this estimate turns negative, which means that academic reputation is more important than year to year variation in spending on instruction.
What this says to me is that while the “athletes fill seats” effect is observable, it’s relatively small. Academic reputation and peers with high test scores matter a lot more. And that’s when sports is lumped in together with other things that students care about like the quality of the food. Presumably the effect of sports alone is even smaller. At the end of the day, to me this means that in order to maintain demand for enrollment, it’s much more important to maintain academic quality than athletic.
Amornsiripanitch, N., Gompers, P., Hu, G., Levinson, W., & Mukharlyamov, V. (2023). No Revenge for Nerds? Evaluating the Careers of Ivy League Athletes (No. w31753). National Bureau of Economic Research Working Paper.
A friend sent me this paper after I wrote the post about my draft syllabus, saying that people at their school found it interesting because it “challenged a number of their priors.” It is an interesting paper! That I ended up having a LOT of problems with. Or rather, some problems and many obvious questions that the authors did not even try to answer. Which makes me additionally worry that they made other mistakes I didn’t catch. Reputation effects, kids! They matter!
There have recently been several high profile complaints that athletes are getting into elite schools with lower academic credentials than non-athletes (see that Chetty et al. 2023 paper we talked about on Day 7, Students for Fair Admission v. Harvard 2023, and the Varsity Blues scandal). The complaints are based on the idea that getting into college should be about academics only, because academic training is the only important part of human capital, and therefore what is valuable about college. This paper takes a different tack by pointing out that athletes at the Ivy League have better career outcomes than non-athletes, so maybe it’s ok or even good that they get into college with lower academic profiles because their human soft skills more than make up for their lack of academic achievement.
The authors combine records about who participated in a varsity sport at 8 Ivy League institutions from 1970 through 2021 with data from their resumes to compare outcomes between athletes and non-athletes from those elite schools. They gather information on sports participation by scraping the university/college website, and they use resume data from Lightcast, which so far as I understand it is basically scraping the information that people voluntarily post on LinkedIn. They show that on average, athletes are more likely to get an MBA, more likely to go into finance, more likely to report more advanced job titles, and therefore have higher estimated earnings both cumulatively and at the peak of their careers.
My biggest question is about the Lightcast/LinkedIn data and how we should think about selection into the sample and selective accuracy of the data. Not everyone even has a LinkedIn profile. Like for example Ivy League graduates who aren’t particularly proud of their careers. By contrast, some people are more likely to have LinkedIn profiles, like for example athletes who receive specific coach-required training on how to set up a LinkedIn profile during their senior year of college so that their teammates and more importantly the fundraising office can find them more easily in the future. People don’t always update their LinkedIn profile, especially when they’re not actively looking for a new job. Other people update their LinkedIn profile often, like for example competitive people who may be more likely to play sports. And of course not everyone tells the exact literal truth on their LinkedIn profile; people who are actively looking for a job and/or updating their LinkedIn profile often are probably more likely to fudge a bit.
The authors do not discuss these issues AT ALL, which is a huge red flag to me.
Another red flag: 54% of their sample is male. Men make up only 46% of Harvard’s Class of 2025. Men haven’t made up more than 50% of the overall college student population since the 90s. Selection into this sample is a problem.
I also have questions about how they define a “varsity athlete.” Their summary statistics list 7 male and 5 female varsity cricket players in the Ivy League. Cricket? In the US? Really? They also list 261 female varsity football (not soccer, American football) players, a full 2.71% of all female varsity athletes. More even than the 64 female baseball players and 53 female wrestlers in their sample. That all seems, um, suspect to me.
Then we get into the technical issues. The authors present two different types of evidence: 1) raw unadjusted means and 2) regression estimates. In their abstract and introduction they switch back and forth between talking about these types of evidence, as if both types are equally credible. This is another red flag. There are very good reasons why we don’t engage in “eyeball econometrics” and treat differences in raw unadjusted means as if they are good indicators of Truth. Confounding variables matter. For example, in this study while athletes are more like to get an MBA from an elite school when you look at raw unadjusted means, the effect is negative and statistically significant in the regression estimates. The intro only talks about the positive difference in raw means. I know. You’re shocked too.
And then there are big problems with the regressions themselves, so I really don’t know what to make of any of this evidence. What problems? 1) They condition on outcomes, which is a huge no-no (i.e. they use length of career, total number of jobs reported, major choice, and first job industry as control variables, when really they are outcome variables; this can bias estimates up or down, so we can’t even say if these estimates are likely too big or too small); 2) they adjust for heteroskedasticity in the standard errors but don’t cluster by institution, which will very likely overstate statistically significance; 3) they run several dozen regressions with no correction for multiple hypothesis testing or mention of a pre-analysis plan, which again will likely lead to the overstatement of effect sizes and statistical significance due to specification searching and possibly p-hacking.
Bottom line: I have scrubbed my mind of anything I “learned” from this paper. I don’t trust it. Sad! I really do think they are asking an interesting question. Too bad they didn’t answer it well.
On Day 14 we’re wrapping up the reading and discussion portion of the class. We’ll close with a discussion of grades, learning, and graduation rates.
We were very lucky to have a conversation with our Head Men’s Basketball Coach, and he sent along several readings about the House v. NCAA litigation so we could talk about how these current events might affect a small school like ours. Unfortunately I’m not going to be able to write up a post about them here. I caught the flu at one of the graduation parties and that’s put me behind on everything. So I’m just going to go with the draft post I wrote about the articles I would have assigned had Coach not jumped in. Sorry and thank you Coach!