Are college return-on-investment calculations meaningful?

With the rapid increases in costs of higher education, there has been a related effort to calculate whether college is worth the cost. One well-known site, built by, has created a database of 20-year earnings for graduates of a vast array of U.S. colleges and universities and compared these to costs. From these data, they calculate the return-on-investment (ROI) for different schools.

One well-known limitation of these ROI numbers is that they don’t account for the properties of the students who attend each school. The average MIT or Harvard student may earn more than the average student at a second-tier college, but that probably has a great deal more to do with the student than the school. In fact, there is no distinguishable difference in economic outcomes between schools once you correct for the attributes of the incoming students (also see this article for a less academic tone).

Another limitation of college ROI numbers is that they look only at the average incomes of all graduates, without controlling for the percentage of enrolled students studying fields that tend to have higher incomes vs. lower incomes. Not surprisingly then, schools at which the vast majority of students study a STEM field have much higher ROI’s than universities with broader populations. At the college ROI site, you can sort colleges on the basis of a specific major and that provides more insight, but the comparison is still hindered by the lack of information on student attributes. Even given this limitation, the results still give some interesting insight. The average graduate of CU-Boulder in computer science is estimated to reap $940,000 in additional earnings over twenty years (in 2018 dollars), as compared to a high school graduate and accounting for the cost of the degree.

Attending Harvard, a much more selective school, reaps a twenty-year gain of about $133,000 in current dollars as compared to Boulder grads. Computer science graduates at Northwestern ($1,054,000) and Cal Tech ($1,059,000), both highly selective schools, make a bit less than Harvard graduates but more than CU graduates. To me, these seem like small economic gains (about $7,000 per year) given that there is no control on the differences in the average attributes of the students when they arrive on campus. If, for example, we were able to compare the economic outcomes for the best incoming CU students to the average Harvard student (a fairer comparison), this gap would probably be a lot smaller or might even disappear, as previous citations suggest.

There are, ultimately, many confounding issues in understanding college ROI. I am quite certain that the average student at a school like Georgia Tech (where I did my undergraduate degree) has a far more vocational perspective on how to plan an education than the average student at many highly-selective institutions. On the other hand, for families weighing the economic costs of education, these statistics may be worth considering as one element of college choice. The perception that going to a highly-selective school is crucial for professional success is certainly thrown into question by these data, assuming that one views economic metrics as some measure of professional success.