Decomposing Outcome Differences between HBCU and Non-HBCU Institutions

Mels de Zeeuw, Sameera Fazili, and Julie L. Hotchkiss
Working Paper 2020-10
July 2020

Full text Adobe PDF file format

Abstract: This paper investigates differences in outcomes between historically black colleges and universities (HBCU) and traditional college and universities (non-HBCUs) using a standard Oaxaca/Blinder decomposition. This method decomposes differences in observed educational and labor market outcomes between HBCU and non-HBCU students into differences in characteristics (both student and institutional) and differences in how those characteristics translate into differential outcomes. Efforts to control for differences in unobservables between the two types of students are undertaken through inverse-probability weighting and propensity score matching methodologies. We find that differences in student characteristics make the largest contributions to each outcome difference. However, some hope in identifying policy levers comes in the form of how characteristics translate into outcomes. For example, whereas HBCUs appear to be doing a better job helping female graduates parlay their education into higher earnings, non-HBCUs are doing a better job in helping graduates in science, technology, engineering, and mathematics translate their training into higher earnings. Patterns and importance of regressors are similar at different points of the distributions of outcomes.

JEL classification: I24, I26, C21

Key words: HBCU, decomposition, student debt, returns to education, propensity-score matching, inverse-probability weighting, quantile regression

https://doi.org/10.29338/wp2020-10Off-site link


The views expressed here are not necessarily those of the Federal Reserve Bank of Atlanta or the Federal Reserve System. The authors thank Kalee Burns and Wenhua Di for helpful comments and suggestions and Fernando Rios-Avila for his support, comments, and Stata expertise. The authors declare that they have no relevant or material financial interests or conflicts that relate to the research in this paper. Any errors are the authors' responsibility.

Please address questions regarding content to Julie L. Hotchkiss.

To receive e-mail notifications about new papers, subscribe. Under "Publications" select "Working Papers."