The Robustness of Laboratory Gift Exchange: A Reconsideration

Abstract

We report a gift exchange experiment in which we systematically vary the following experimental design and implementation characteristics: the choice of equilibrium (interior versus corner point), the extent of potential efficiency gains, and the choice of frames (abstract versus employer worker). We also employ a matching mechanism that had been shown to best preserve the nature of one-shot interactions (rotation). Much of the observed play of our participants, especially responders, is at or close to equilibrium. Our results therefore stand in stark contrast to much of what has been reported in literature. Specifically, we find little evidence for positive reciprocity but substantial evidence for negative reciprocity. Our results suggest that laboratory gift exchange is highly sensitive to the parameterization of the model and implementation characteristics and question the common belief that trust and reciprocity are robust phenomena.