The Multi-unit Assignment Problem: Theory and Evidence from Course Allocation at Harvard

This paper uses data consisting of students' strategically reported preferences and their underlying true preferences to study the course allocation mechanism used at Harvard Business School. We show that the mechanism is manipulable in theory, manipulated in practice, and that these manipulations cause meaningful welfare losses. However, we also find that ex-ante welfare is higher than under the strategyproof and ex-post efficient alternative, the Random Serial Dictatorship. We trace the poor ex-ante performance of RSD to a phenomenon specific to multi-unit assignment, "callousness". We draw lessons for the design of multi-unit assignment mechanisms and for market design more broadly.