JUE Insight: College Student Travel Contributed to Local COVID-19 Spread

Hi all! @Paul_Niekamp and I have a new paper using SafeGraph data. We investigate whether travel over spring break by university students contributed to local COVID-19 spread. Due to the rapid growth of COVID-19 and the subsequent suspending of in-person classes, students with earlier spring breaks traveled normally but students with later spring breaks effectively had their spring break canceled and didn’t return to campus. We show with SafeGraph data that devices residing on early spring break campuses traveled extensively and returned to campus after break but later spring break devices left campus and did not return. When we split counties into more/less early spring break students, we see a noticeable difference in the growth rate of confirmed COVID-19 cases and mortality. First, in the week after students returned (primary infections) and then two weeks after students return (secondary spread). For mortality, we don’t see any noticeable difference until around 3 weeks after students return, but then mortality growth rates remain higher for the next month in early spring break counties than in late spring break counties. We also find that the effect of university student travel was over 50% larger for students who traveled via air, to New York City, or to Florida while cruise travel was no riskier than that of the average student traveler.

You can find the paper here: https://papers.ssrn.com/sol3/papers.cfm?abstract_id=3606811
Tweet summary with more graphics: https://twitter.com/danielmangrum/status/1263513039397101568?s=20

Great use of the data!

@Daniel_Mangrum_Vanderbilt_University @Paul_Niekamp I enjoyed reading your paper. Would you be willing to share the code you used?

Wow that’s a really interesting article thanks for sharing.

It strikes me that at the time there were lots of news reports of college students on break saying things like “I’m not worried about the virus” … and commentators saying it wasn’t about the students health but the health of others. Seems like your research supports that framing.

@Daniel_Mangrum_Vanderbilt_University @Paul_Niekamp Thanks for sharing your paper. It is a pretty convincing case that reducing college student travel would help contain community spread. I noticed that in your map Figure A1, the late break universities are more concentrated close to the east coast, would this correlate with the political inclinations of those students, which would be a factor in how they behave on campus and in the communities around those campuses? Why do you think that cruiseliner travels were not as impactful as air travel? I assume air travel would be much more frequent than cruiseliner travel, or traveling to highly risky destinations such as NYC and Florida ports.

@Ruowei_Yang_UM_Baltimore Thanks for the comments. That’s a really good point about political affiliation. We check to see whether early versus late counties are different along a number of dimensions but we haven’t checked political affiliations. We should add that to the variable list. I’m sure 2016 presidential vote share would be a good starting point.