I’m often asked about the course bidding system used at Kellogg to assign students to classes. Can it be gamed? Could it be improved? My advice  is don’t think hard about it because its largely cosmetic. Certain classes at Kellogg are scarce. They are scarce in spite of multiple sections being offered, because  the faculty member is not substitutable (like my colleague David Besanko who can make the problem of counting sheep riveting). Or, the course is scarce because the combination of subject and faculty is not substitutable (Kraemer on leadership). A scarce resource has to be rationed.

One could ration with money. Genuine coin of the realm. This would mean students in the same program end up paying different amounts for their degree. If one were to `think bravely’ for a moment, why not? Imagine one charges a flat fee plus a separate per course fee that depends on the scarcity of that course. Thus a student who cares very much about which professor delivers which course would pay more than one who did not care as much. One could argue that a program consisting entirely of Allon, Besanko, Chopra, Diermeier, Kraemer, Petersen, Rogers, Rucker, Stern, Zettlemeyer is very different from one consisting of Schmo, Schmo, Schmo and Vohra.

One could ration by lottery. This what we used to do. It imposed a tax on the administration because students who did not win would complain. After all, through no fault of their own they were denied the chance to take Professor X’s class.

One could use some combination of funny money and auctions. This is what we do. Students get a budget of points. The same budget for all students. A smaller number of classes are still oversubscribed as many bid essentially all their points for these classes. In these cases they are rationed by lottery. The only real effect is that complaints to administration drop. Now when a student does not get into a class, its their responsibility because they did not bid enough. It is true that there are simple modifications that could remove some of the anxiety associated with this system, but, again, this is cosmetic.

The vast majority of instructor-course combinations are substitutable. This is reflected in the bid prices for these combinations. In these cases, what drives preferences is time of day; 8.30 vs. 10.30 vs. 1.30 etc. The number of classrooms limits the number of 10.30 classes that can be offered. Packing bums into seats is easy. Ration and allow trade to sort out the preferences over time slots. There is little reason to agonize over getting an 8.30 slot when one wanted a 10.30 slot.  A more important issue  is what one thinks a Kellogg experience is supposed to be. Is it, for example, vital that each student have a chance to be at least in one class taught by a Kellogg faculty member generally acknowledged to be, for want of a word, WOW? If so, it is easy to determine which ones. Hand each member of the entering class a ticket to a subset of these course that guarantees them a seat if they choose to register for that class. Allow them to trade the tickets to sort out time and subject preferences.