Learning in Multi-issue Bargaining

Abstract: How does learning about proposer power affect agents’ ability to compromise? We study a dynamic multi-issue bargaining game between a proposer and a responder. Issues arrive at random, and with each new issue, the proposer makes a proposal, which the responder either accepts or rejects. In case of rejection, the proposer can attempt to force the issue and implement his ideal and will succeed with some probability that is a function of the proposer’s unobserved ability. Both players learn about the proposer’s ability over time as new issues arise. We show that there is conflict when the belief about the proposer is either high or low, but that compromise can occur in an intermediate region of beliefs. This is driven by both players’ incentives to avoid learning in that region. We extend the model to include the possibility of difficult issues arising in which no compromise is possible. Dif- difficult issues can disrupt a previously established compromise, forcing conflict for issues in which compromise was previously possible (easy issues). The reason conflict ensues is that the responder learns the proposer is weak and no longer has an incentive to compromise, even on easy issues.