Coordination with heterogeneous interaction constraints

Abstract

We study a coevolutionary model of network formation in minimum-effort games. Agents dynamically choose both their effort levels and their interaction partners, but face heterogeneous constraints on how many partners they can interact with. The dynamics under heterogeneity behaves very differently from the one in the homogeneous case, with gradual transitions spreading across the population stepwise instead of sudden adoptions. The long-run outcomes depend crucially on the distribution of interaction constraints. The efficient (highest-effort) convention prevails in the long run even when many agents have loose constraints, if enough agents have tight constraints. The lowest-effort convention dominates if too many agents have loose constraints. The coexistence of conventions is also possible for certain distributions of constraints.