The Effects of Precedent on Arbitration

This paper presents a new view of the arbitration process in which the arbitrator is depicted as a random device to generate arbitration decisions. The conflicting parties must decide whether to send their dispute to the arbitrator, based on their subjective probability beliefs concerning the arbitration’s outcome. If their beliefs are “sufficiently divergent,” we can expect both to agree to arbitration. As similar disputes are arbitrated and precedent is set, however, these divergent beliefs can be expected to vanish. Using some game theoretical results of Aumann and Rosenthal, we demonstrate that such convergence will ruin the incentive to arbitrate for conflicts that are zero (or constant) sum in nature, and for games that are non-constant sum but “best responsive equivalent” to zero sum games. It will not do so in general for conflicts that are non-zero sum. We then examine the welfare implications of these results and point out a paradox that arises in this context.