When a university increases expenditures in an attempt to improve quality, a natural question for social policy is raised: is the improvement worth the cost? One partial measure of "worth" in this context is given by earnings of graduates: will those earnings rise by enough to pay a reasonable return on the investment in higher quality?
The question is hard to answer by statistical means: universities of higher quality have better reputations which allow them to select students who are themselves of higher quality; so in trying to measure the relation between earnings and university quality, quality of the school may be confounded with quality of the students. Recently, however, several large data sets have been made available which give sufficient information on the students' own backgrounds, and on their later earning, that we can hope to disentangle the effect on earnings of the school from that of the student himself.
This article summarizes the results of several such studies, including our own work. The verdict to date is that the relation between "quality", however measured, and earnings is significant, and that expenditures to improve quality are worth the cost, even when "worth" is measured narrowly as higher earnings for graduates.