One of the easiest things to do in policy is to advocate for policy X, so as to change effect Y.  One of the hardest things to do is to get people to explain clearly their theory of change.  That is, what are the steps by which changing X actually affects Y?

Take performance-based funding.  It’s easy to get hot for the idea that organizations can be steered by offering incentives: if you pay schools for students, they’ll raise enrolment.  If you pay them for graduates, they might spend a bit more effort and money on academic support service.  And so on.  By this theory, all you need to do to get universities to change their behaviour is to offer the right financial incentives.

But here’s the problem: that theory works a lot better for individuals than for organizations.  If what you are trying to do is force a change in organizational culture (e.g. get them to shift to a more student-centred focus), you have to remember that individuals inside an organization aren’t necessarily going to face the same incentives as the institution.  Just because an organization is incentivized doesn’t mean everyone in it is incentivized.

In extremely hierarchical organizations, it’s possible for management to pass incentives on to staff in various ways.  But universities are not particularly hierarchical institutions.  Outside of terrorist cells, universities are about the most loosely-coupled organizations on earth.  Some of the larger among them, to quote Kevin Carey, are more like holding companies for a group of departments, which are themselves holding companies for professors’ research interests.

So let’s get back to the example of a government that hopes to get universities to pay more attention to student success.  Say the government comes up with a funding formula that potentially allows an institution to access a couple million dollars more if it increases its graduation rate.  What happens?

Well, it’s certain that university leadership will try to grab the money.  That’s their job.  Then they’ll think about how to achieve the goal.  Pretty much every authority on retention will tell you that it is a institution-wide exercise.  The key is identifying students that are having trouble, and then making sure they get appropriate assistance, either from instructor(s), or from some kind of centralized suite of academic services.  But while it’s easy enough to invest money in new centralized services, the key to such an approach still rests on professors (some more than others) altering the way they behave in class, so as to spend more time/effort identifying strugglers early, and then doing something about it (talking to the students themselves, sending their name to a counsellor who can then contact the student and offer assistance, etc.)

The question is: how do you get the professor to make those changes?  The promise of more money to the institution is a pretty weak one.  First, while many people’s behaviour will need to change in order to get the money, not everyone’s does, so there’s a rational reason to try to free ride on the process.  Second, even if the institution does get the money, it doesn’t follow that the money will be distributed in such a way that all individual profs  benefit.  A prof’s behaviour is not incentivized in the same way as the institution’s.  And if that’s so, why would we expect the prof to alter his or her behaviour?

I’m not saying it’s impossible steer universities by using money as an incentive; I’m saying that success in doing so requires the incentives to be aligned in such a way that everyone’s behaviour down the chain is incentivized.  And in a university, where every professor is, to an extent, a free agent, that’s really hard to do.  It works where the incentive aligns with career goals or professional norms (e.g. do more research).  But when it pushes against professional norms, it’s a lot more difficult.

Fundamentally, people trying to steer system reforms need to ask themselves: how will this incentive alter what individuals on the ground actually do on a day-to-day basis?  If there’s no good answer to that question, chances are the incentive isn’t likely to work.