Professors Respond to Incentives; and so do students

The C.D. Howe Institute recently published a study that shows professors are more productive at universities that have merit-based salaries. Also, students at those universities seem to perform better. [thanks to Erin Airton for the pointer].

Let's suppose that the objective function of the university has nothing to do with education, research, or any of that airy-fairy stuff that university leaders talk about. Suppose, instead, that the objective function is to maximize the wealth of the university, including the expected net present value of future donations and future grants, etc., and future funding provided by various levels of gubmnt. With this objective function in mind, what matters is not quality of education or research, but perceptions of quality of education and research. As Ms. Eclectic says so cynically, "It's a business."

E.g. if a university can attract tonnes of really REALLY smart, innovative, creative, assertive, energetic students, then, even if the value added by the university is not worth a pinch of anything, it will appear that the university has done a great job. Further, its alumni, potential donors, and the politicians providing support for the university will tend to think the place is terrific.

With this view in mind, it is reasonable to be concerned about a simultaneity bias in the CD Howe study. What if good professors and good students tend to select schools that have merit pay for their professors? Then the existence of merit pay would have an incentive effect quite different from "merit pay encourages professors to do a better job."

And what if mediocre students select schools rated less highly. Then it might be a mistake to attribute causation to the CD Howe study results. Fortunately, the authors recognize this:

The salary structures at Canadian universities appear to matter. Evidence shows that the performance of universities with merit salaries exceeds that of other universities. They perform better in a variety of research-based and student quality measures without any sacrifice in other dimensions, such as student satisfaction. While such associations by themselves do not necessarily imply causation, the findings conform to the predicted consequences of different salary arrangements.