Stochastic bargaining in the lab

We study novel multi-stage bargaining games that differ in the timing and efficiency of Stationary Subgame Perfect Equilibrium (SSPE) outcomes, but include the possibility of an equal split in every stage, based on the theory of Merlo and Wilson (1995). Despite this theory’s wide application to bargaining in natural environments, our study is the first to rigorously test it using laboratory experiments in both gain and loss domains. In our design, the total surplus to be split within a stage is not a fixed amount, and it can expand or shrink between stages, making delay and inefficiency possible in equilibrium. We find that players regularly agree on the largest available equal split regardless of SSPE predictions. These patterns are well captured by an “Efficient Equality Equilibrium,” built on the idea that people share lexicographic preferences for equal-split outcomes. Our results highlight the important interplay of equality and efficiency in bargaining and other conflict resolution environments. In addition, we find that men and women are equally likely to propose equal splits over gains, while male proposers are significantly more selfish over losses.