Binding arbitration is a common method of alternative dispute resolution used in resolving labor disputes. Two different forms of binding arbitration dominate in practice: conventional arbitration (CA) and final offer arbitration (FOA). In CA, the arbitrator is allowed to choose any settlement as the arbitrated outcome. Criticisms that arbitrators merely “split the difference” of the disputants’ final positions led to the arguments that FOA, in which the arbitrator is constrained. to choose one of the disputant’s final offers, might induce more negotiated settlements. A large literature has developed showing that disputants are not, however, theoretically predicted to converge towards agreement under FOA. This paper presents results from a controlled laboratory study of bargaining behavior and dispute rates under an innovative procedure called “combined arbitration” or CombA (Brams and Merrill, 1986). The rules of CombA involve a simple combination of using CA or FOA, depending on whether or not the arbitrator’s notion of a fair settlement lies between the disputants’ final offers. The potential importance of the procedure is that it is theoretically shown to induce convergence of disputants’ final offers. The result is that it theoretically predicts negotiated, as opposed to arbitrated settlements. Disputants generally prefer negotiated settlements, which would also imply substantial cost savings by not actually invoking arbitration. In our experimental environment, subjects anonymously bargain over the size of a disputed variable, X. Subject payoffs are such that bargaining is zero-sum over a $2 pie in each of 20 bargaining rounds. Different dispute resolution procedures are implemented in the event of bargaining impasse at the end of a 2-minute round. We test CombA along with two modified forms of CombA (also suggested in Brams and Merrill) and find that dispute rates are still significantly higher than when disputes are resolved by destroying the disputed monetary pie (i.e., simulating the high cost of a labor strike perhaps). On the other hand, as the theory predicts, the CombA procedure induces statistically significantly lower dispute rates than the modified CombA procedures that lower the uncertainty costs of basic CombA procedure. To the procedure’s credit, CombA is also shown to not adversely affect negotiated outcomes. The implications of these findings may be significant, and they call for direct comparisons of disputant behavior under CombA, FOA, and CA.
