Multidimensional Arbitration

Most analyses of arbitration consider disputes over a single dimensional issue; however, in many situations more than one issue can be in dispute. In a labor context, the parties may disagree on the hourly wage, number of paid holidays, and fringe benefits as well as other conditions of employment. In addition to conventional arbitration (CA), two variants of multidimensional final-offer arbitration (FOA) are employed in practice. Under one version of FOA, the arbitrator must pick the entire package submitted by one of the disputants. Under the other version of FOA, the arbitrator can select elements from each disputant’s final offer on an issue-by-issue basis. Theoretically, we show that when disputants have different relative values between dimensions, package FOA leads to expected welfare improvements for both parties as compared to CA and issue by issue FOA. In a controlled laboratory experiment, we examine how asymmetry in relative values between issues impacts proposals and welfare under both package FOA and issue-by-issue FOA. For package FOA, we observe that asymmetry leads to proposals that are more aggressive in the more valuable dimension and more generous in the less valuable dimension, as predicted by theory. As a result, payoffs are higher under package FOA than issue-by-issue FOA or CA, even though observed treatment effects are not as large as predicted.