For the last few years that I did Economic Theory research (in the main this ended as I took up ed tech administration) my work was on various applications of the heavily studied principal-agent model and knowledge of that informed my approach to teaching.
The canonical example of principal and agent is “share cropping” and designing the optimal contract in that environment. The basics of the model are something like this. The share cropper is assumed to be risk averse while the land owner is risk neutral (or risk averse but less so than the share cropper). That implies, all else equal, that in a contract between the parties the land owner should absorb the bulk of the risk and in the process insure the share cropper. The output of the farm depends on two factors. First is the share cropper’s effort. More effort produces greater output but more effort also creates more disutility for the share cropper. Effort needs to be interpreted abstractly and what it means depends on the context. (I’ll say more about that below.) The other factor is a random component, “the weather” or the aggregate of other external factors, that can’t be controlled by the share cropper or anyone else.
If the land owner (and an independent third party who enforces the contract) can observe effort directly, then the optimal contract specifies the effort level where the marginal product of effort equals the marginal disutility (these are the standard conditions that determine the efficient output level) and then the sharecropper receives a flat payment, independent of the actual output, so the land owner gets the entire farm output less that payment to the share cropper, and thereby absorbs the risk. Because economists love jargon, this solution is referred to as “the first best” which really means an ideal solution we won’t see in reality because the information requirements are too high (everyone knows the effort).
The model predicts “the second best” solution where the contract can depend on the farm output, which is observable, but not on effort. In this case, under a flat payment scheme the share cropper would choose zero effort (because effort causes a disutility while the payment does not depend on effort) while under a scheme where the share cropper pays a fixed rent to the landowner and gets to keep the entire farm output less this rent, the effort level would be what it is under the first best solution. But the latter means the share cropper absorbs all the risk, and that is not efficient. The optimal share provides the correct tradeoff between inducing share cropper effort, on the one hand, and having the land owner insure the share cropper against the environmental risk, on the other.
We see elements of this type of solution to the agency problem in sales, where commissions are very important, and in a variety of other cases where one hires an agent, such in the buying or selling of a home. There are some people who like to work on commission, but most people I know would prefer to be salaried and would prefer to have their good effort acknowledged through raises and in non-financial ways that fall under the category “recognition.” In a lot of cases, this happens although the output can be measured, meaning piece rates are possible. The conclusion is that there are other, superior ways to provide incentives and those are compatible with a salary approach.
Most of those ways can be characterized by the idea that the employee has a stake in the work. In the simplest of approaches, the stake is in the form of deferred compensation. A more sophisticated view takes the work itself as potentially rewarding and part of that is how much discretion and initiative the employee is delegated, as well as how much of the work is “play.”
Thus, the employer can create a stake in the job by making the work particularly rewarding. Typically this means the employer must take a longer term view because the employee “doing his own thing” will not pay off immediately, if at all. But the engaged employee is likely to produce more interesting things if given sufficient opportunity to do so.
How do these insights translate into good instructional practice?
First, consider textbooks. How does an instructor and potential adopter of the textbook evaluate one? I know that for intermediate microeconomics books, I look first at the end of the chapter questions? Sadly, in most cases these are dull, unimaginative, an apparent after thought for the author(s). Similarly, many of the ancillary tools for the book are developed not by that authors themselves, who have the biggest stake in the success of the book, but by others. If you ask economists how students learn economics (there are reasons to be suspicious about how economists teach) they’ll invariably tell you by working problems. So in this agency area we seem to be do less well than we might and raising awareness of the issue might encourage the market to produce a better solution.
Next, consider instructors who spend a lot of time preparing their in class presentation. In analogy to the textbook situation, they might find it more effective to instead spend time on the in class “active learning” exercises and out of class assignments that the students do. Further, open ended assignments are intriguing to assign because they have the potential to seriously engage the students. But there is also the problem that students may go astray. So if the instructor does opt for more open ended inquiry, the instructor should be helping students work through those open ended assignments rather than wait for them to be turned in. This work is time consuming. How does the instructor find time for doing it? The answer is by spending less time on presentation and preparing that.
Third, instructors might reconsider what they mandate as required readings for the course. Are the readings interesting on their own account? Will the students retain any of what they’ve read after the course has concluded? Will the readings get the students to consider majoring in that area of study?
All of these ideas fall under the category of a learner centric approach to teaching and none of them are particularly novel when set under that umbrella. So why consider the agency issue at all. Why not simply preach learner centric methods.
The answer is two-fold. One is that instructors, like students, have a basic need to understand “why” one should adopt a particular methodology. What issues does the methodology address? What issues does it create? The agency framework allows the instructor who is contemplating a modification in her teaching to consider the likely consequences.
The second reason, I believe, is the more important one. In general, the first best solution and the second best solution are quite different. An instructor thinking idealistically about instruction may contemplate the first best solution, try to implement a reasonable approximation thereof and then be rudely disappointed by the outcome. The agency framework can help temper the instructor idealism and provide a more realistic way to view the issues at hand.