Pairwise imitation and evolution of the social contract

This paper studies the evolution of conventions in Stag Hunt games when agents’ behaviour depends on pairwise payoff comparisons. The results of two imitative decision rules are compared with each other and with those obtained when agents myopically best respond to the distribution of play. These rules differ in terms of their rationale, their requirements, and the extent to which they make individuals learn from others. Depending on payoffs and the interaction process being considered, best response learning can cause either the rewarding All Stag equilibrium or the inefficient All Hare equilibrium to emerge as the long-run convention. In contrast, pairwise imitation favours the emergence of the Pareto-inferior equilibrium. This result is robust to assuming assortative matching and some heterogeneity in decision rules.