Final-offer arbitration with reference-dependent preferences

When negotiators fail to reach agreements, arbitration is used to resolve disputes. This study examines final-offer arbitration (FOA), in which an arbitrator selects one of the negotiators’ offers as the binding outcome. FOA encourages pre-arbitration settlements by increasing the risks associated with arbitral uncertainty. Nevertheless, negotiators often proceed to arbitration in practice. To address this, we characterize the conditions under which negotiators either reach an agreement or escalate to FOA. Our model incorporates reference-dependent preferences and the uncertainty over arbitral decisions. Negotiators incur costs of losing in FOA that depend on their offers, and an arbitrator is drawn from a probability distribution. We reveal a necessary and sufficient condition for agreements under a general class of distributions. FOA facilitates settlements only when negotiators’ costs are sufficiently high relative to the odds of “gambling” in arbitration. We demonstrate that the condition becomes distribution-independent under linear utilities.