Voting rules in multilateral bargaining: using an experiment to relax procedural assumptions

Experiments can be used to relax technical assumptions that are made by necessity in theoretical analysis, and further test the robustness of theoretical predictions. To illustrate this point we conduct a three-person bargaining experiment examining the effect of different decision rules (unanimity and majority rule). Our experiment implements the substantive assumptions of the Baron-Ferejohn model but imposes no structure on the timing of proposals and votes. We compare our results to those obtained from an earlier experiment which implemented the specific procedural assumptions of the model. Our results are in many ways very similar to those from the more structured experiment: we find that most games end with the formation of a minimum winning coalition, and unanimity rule is associated with greater delay. However, the earlier finding of "proposer power" is reversed. While some important patterns are robust to the less stringent implementation of procedural assumptions, our less structured experiment provides new insights into how multilateral bargaining may play out in real world environments with no strict procedural rules on timing of offers and agreements.