• Prose and Purpose

    After 25 years on the job, a former provost examines the world on campus and in higher ed.



Going back to my days in graduate school, I have always had a fondness for economic models.

October 4, 2015

Going back to my days in graduate school, I have always had a fondness for economic models. Starting with actual data and past experience, and assuming some things change going forward, it is possible to project what the result will be of the change. If investment increases or consumer spending decreases, or auto sales accelerate, or oil prices decline, it is possible to predict the consequences, often with real accuracy.  In higher education, this modeling often takes the form of predicting the impact on yield of changes in the discount rate. What discount rate will yield what student profile and student numbers, all other things remaining the same. Often the result of these calculations is a class size, a class profile (both GPA and SAT) and the costs involved. Models can also be used to predict the success of students based once again on past data and experience

Models in higher education often predict a specific number, not a range and so with an investment of x dollars, you yield students with a z profile. And of course if the result is a shortfall, there is almost a sense of what did we do wrong to not bring in the number of students we wanted.

But wait, all these models are based on the premise that all others things remain the same or perhaps that they change only in a way that we can predict. What if we assume that oil prices remain stable and instead there's some shock to the system that disrupts the flow of oil? Or, what if we assume that all other discount rates stay stable and instead the discount rate for our competitors increases by 1 percent?  Clearly the result will likely be quite different than we projected. The reality is that periodically or even more often, all other things remaining the same will not happen.

Regardless of the external shocks that we cannot predict and regardless of possible wrong assumptions, models are still extremely valuable. If we expect our modeling to be the equivalent of a perfect crystal ball, we will doubtlessly be disappointed. If achieving an exact number or a precise result is the goal, we will also be disappointed. Perfection really is not under our control. If we instead view our models as providing a solid prediction of what the future holds, we are likely to be pleased and tremendously helped by the support provided.


Back to Top