Costly Communication in Groups: Theory and an Experiment

I develop a novel model of group-based deliberation in which communication is costly in two directions: agents must pay separate costs to send and to receive messages. Equilibrium strategies have an intuitive characterization - those with the best information send, those with the worst information receive. But free-riding leads to less information exchange than is optimal. Testing the model's predictions with an experiment I find that subjects overcommunicate when costs are high, but fail to benefit from this as much as they should. In welfare terms the experiment finds that listening costs are more harmful to welfare, in contrast with theory, which indicates sending costs. The experiment also suggests that the existence of costly communication channels can reduce total welfare.