Covid-19 prevalence and empty college seats

ABSTRACT

Using the National Association for College Admission Counselling’s annual list of colleges with open seats and Covid-19 cases and deaths at the county level, this paper provides a first look at how Covid-19 has altered enrolment at four-year colleges. I find that a one standard deviation increase in deaths per 100,000 is associated with a 61% increase in the probability that a school reports available seats, housing, and aid for first-year students. For a one standard deviation increase in cases per 100,000, schools are 53% more likely to report openings. For a one standard deviation increase in the growth rate of deaths and cases, schools are 45 and 56%, respectively, more likely to report openings. These openings are not driven by the fact that many Covid-19 hot spots in March and April of 2020 are also home to schools with higher tuition and fees, a high share of out-of-state or international students, and more progressive political leanings.

KEYWORDS:

• Coronavirus
• Covid-19
• SARS-CoV-2
• college admissions
• college enrolment

JEL CODES:

• I23
• I22
• I18

Acknowledgments

All errors are mine.

Declarations of interest

none

Disclosure statement

No potential conflict of interest was reported by the author.

Supplemental material

Supplemental data for this article can be accessed here.

Notes

¹ The American Council on Education surveyed 192 Presidents. When asked about the most pressing issues they faced, summer/fall Enrolment was 22 percentage points higher than the second most pressing issue: long term financial viability at 64%.

² The focus of this study is on four-year institutions. These college openings data are limited to the roughly 1,200 U.S. NACAC member four-year colleges out of approximately 2,200 four-year colleges in the United States. Members are predominantly selective institutions and include national universities, such as Baylor, Berkley, Boston College, Clemson, Georgia, Harvard, Iowa, MIT, Stanford, Texas, and Vanderbilt, national liberal arts colleges, such as Amherst, Austin, Berry, Furman, Grinnell, Pomona, Rhodes, Stonehill, University of the South, and Williams, and regional universities and college such as Belmont, Bryant, Butler, Eastern Oregon, North Georgia, Providence, and Western Carolina. The NACAC membership directory can be found here: https://hub.nacacnet.org/institutionmemberdirectory

³ This later became a reality when schools, such as the University of North Carolina at Chapel Hill and East Carolina University, witnessed a large number of cases and moved their classes online.

⁴ For examples see, Montgomery (2002); Long (2003), (Long 2004); Avery and Hoxby (2004); Niu, Tienda, and Cortes (2006); Lipman Hearne Inc. (2006); Niu and Tienda (2008); Bruce and Carruthers (2014); Marx and Turner (2018); and Carruthers and Welch (2019).

⁵ A few examples include Cabrera and La Nasa (2000); Freeman and Thomas (2002); Hossler, Hu, and Schmit (1999); Niu and Tienda (2008). Following Soares (2013), others such as Iloh (2018) have sought to extend college choice modelling to include non-traditional college searches and choice.

⁶ These case studies cover several states: Georgia (Alm and Winters 2009), Texas (Jones and Kauffman 1994), Washington (Ullis and Knowles 1975), and West Virginia (K. Ali 2003).

⁷ For some students, particularly those from lower-income households, some argue that the rising cost of education has made location more important today than ever before (Roscigno, Tomaskovic-Devey, and Crowley 2006; Shaw et al. 2009; Tate 2008).

⁸ One notable exception is that NACAC members voted to eliminate some provisions in their Code of Ethics in September of 2019. Striking these provisions meant that members could now offer exclusive incentives to early decision applicants and could aggressively recruit students who had made deposits at other schools before and after the traditional May 1st deadline (Jaschik 2019). However, as noted by the President of the NACAC, Jayne Caflin Fonash, most schools continued to abide by the basic principles of the Code of Ethics, and the governing board voted in March 2020 to recast these as best practices instead of mandatory code (Fonash 2020).

⁹ Unfortunately, institutions do not report the number or share of open seats.

¹⁰ Because the focus is on first-year students, I do not use information about openings for transfer students. I also construct an alternative measure that looks at whether a school reports having first-year openings or available housing, or financial aid. The estimation results are similar.

¹¹ The estimation does not include year effects because these data include only two years, 2019 and the treatment year 2020.

¹² There has been debate over the accuracy of death and case rates (Brown, Reinhard, and Thebault 2020). This is especially true of case rates given test availability, the protocol used to determine whether tests are administered, and conflating diagnostic tests and anti-body test results (Ketchum and O’Connor 2020; Madrigal and Meyer 2020). However, what is essential for this study is the information available when students are selecting whether and where to enrol.

¹³ An alternative approach is to exclude available housing as a necessary condition. Using this less binding condition results in qualitatively similar estimated coefficients.

¹⁴ All results tables report the estimated coefficients. However, throughout the remainder of the paper, I will discuss the change in the probability of reporting openings associated with a one standard deviation change in the Covid-19 variable of interest. All probabilities are found by multiplying the standard deviation of the Covid-19 variable by the estimated coefficient. For example using the coefficient on deaths in column 1 of Table 2 and noting that a one standard deviation increase in deaths is 3.718, I find that the probability of reporting openings increases by 82.5% or 3.718 (or 372/100)* 0.222 = 0.825 or 82.5%.

¹⁵ To perform his estimation, I assume all schools’ decision deadlines in 2019 were May 1^st. This assumption may not be valid for all institutions, especially those with rolling and open admissions. However, it is highly unlikely that schools would report deadline extensions in 2020 if they are a school that traditionally offers rolling or open admissions.

¹⁶ The elimination of some NACAC recruiting rules may have resulted in more aggressive recruiting by some institutions. However, more aggressive recruiting would reduce the likelihood of open seats. Moreover, given that schools near Covid-19 hotspots were facing more empty seats, these schools would be more likely to recruit aggressively. Aggressive recruiting by schools in or near Covid-19 hotspots would, ceteris paribus, downward bias the estimated coefficients.

¹⁷ According to MyPlan.com, 11,627 MyPlan.com users rate the political leanings of the college they attended. MyPlan.com only reports political leanings of a school if they have enough observations to report statistically significant results. More details are available at https://www.myplan.com/education/colleges/college_rankings_14.php

¹⁸ The summary statistics are in Tables A1 of Appendix A.

¹⁹ Tables A2 through A4 report the results.

²⁰ See Tables A5 through A7 for complete results.

²¹ See Tables A8 through A10 for detailed results.

²² See Tables A11 – A13 for the complete results.