“College Quality As Revealed by Willingness-To-Pay for College Graduates”, John J. Green, Nicole S. Swepston2023-05-03 (economics)⁠:

This study measures college quality by the amount by which the college adds to the salary of its students above what the median market value would be for the same majors and student quality. Commonly used national rankings of colleges such as US News and World Report or Forbes are heavily biased by a college’s average salaries and the quality of the students it enrolls, and not by the actual value-added by the colleges.

Once student quality and mix of majors are controlled, salary differences between elite and non-elite schools largely disappear.

[Keywords: College quality, major, salary, revealed preference, willingness-to-pay, value-added]

…we generate an expected salary for each college based on its mix of majors and the median salary of each major. Value-added college quality is then implied by how much firms are willing to pay for each college or university’s graduating class compared to what their graduates would have been paid had they earned the median starting salary for their major. Our preferred value-added index also holds fixed the local cost of living, the demographic composition of the student body, and the performance of its students on standardized exams before attending the university. The resulting hierarchy of schools based on value-added net of student attributes differs greatly from the most commonly used university rankings. Elite private schools look much more like flagship public schools or even modestly priced liberal arts colleges once the quality of the student body is held fixed.

We are also able to generate confidence intervals for college rankings based on the sampling variation of the model parameters. We find that there is substantial variation in the ranking of colleges, especially in the middle of the quality distribution. Only at the extreme upper and lower tails of college quality is there a tight error distribution around the mean college ranking. This corroborates our conclusion that willingness to pay measures suggest that college quality is spread broadly across schools of many types, and it explains why college rankings can vary so much across rating systems or across time.

…Our study combines the Cunha & Miller2014 and Rothwell & Kulkarni2015 strategies by controlling for the value-added by the university after controlling for both the college’s mix of majors and the average ability of its students. As we show, using a national sample of colleges and universities including public, private nonprofit, and private for-profit institutions, conclusions regarding college value-added are very sensitive to whether controls for mix of majors and student aptitudes are included in the analysis. Once both of these controls are incorporated, the differences in value-added across colleges and universities become much less pronounced. As one might expect, high quality students in valued majors earn more, regardless of what school they attend. Weak students in low-demand markets earn less.