Essays on Economics of Postsecondary Education
Akmanchi, Suchitra, Economics - Graduate School of Arts and Sciences, University of Virginia
Turner, Sarah, Economics Department, University of Virginia
An individual's postsecondary enrollment, completion, and post-completion wages are subject to labor demand, education supply, and individual-specific factors. Each chapter of this dissertation looks each of these factors in turn to determine what relationship each has to postsecondary education. In the first chapter, I study how participation in newly added community college programs is related to employment and wage outcomes. The introduction of new programs provide a policy option for community colleges to connect their students to employment opportunities in growing local industries. I use administrative data on students in the Virginia Community College System, as well as program introduction dates to address this question. Using a generalized difference-in-differences approach, I find preliminary evidence that new program participation increases quarterly earnings. I do not find clear evidence of an effect on employment. Several factors limit my analysis including: the variety in length and subject area of newly added programs; differential selection into new programs; and focus on non-traditional students.
In the second chapter, which is co-authored with Benjamin L. Castleman (University of Virginia) and Kelli A. Bird (University of Virginia), we apply machine learning methods to compare the performance of an algorithm to advisors in predicting college enrollment for high school seniors. In the context of a national college advising program that focuses on high-achieving, lower-income students, we compare the performance of advisor predictions to a trained logistic regression algorithm. We find that college advisors slightly outperform a prediction algorithm. Advisors predictions are more accurate for students with whom the advisor had more interactions, and are similar in accuracy to the algorithm for students with fewer advisor interactions. These results are indicative of the potential of algorithms to provide efficient predictions in contexts where resources to fund human prediction may be scarce.
I explore the effect of local negative labor shocks on community college enrollment and graduation in the final chapter of the dissertation. In economic downturns, established evidence shows that displaced workers invest in their human capital by enrolling in postsecondary education. I investigate whether workers respond to negative local labor market shocks by enrolling and earning credentials at their local community college. In Virginia, every community college has a policy-defined service region which both defines the local labor market and is an approximate catchment area for the college. I apply a fixed effects regression with a region-specific linear trend to publicly available mass layoffs data as well as counts of community college enrollment and awards. My results show that enrollment and Associate's degrees increase in response to local mass layoffs. The effect on sub-Associate's degrees is dependent on the sample years included-- positive in the full sample, and negative when I restrict to years preceding the Covid-19 pandemic.
PHD (Doctor of Philosophy)
community college, postsecondary education, mass layoffs, prediction algorithm
English
All rights reserved (no additional license for public reuse)
2024/07/29