Match choice and Ghettoization in evolutionary games

I study how having a choice about who to play affects the conventions that arise in a population playing a 2×2 common interest game. Match choice allows agents playing “fragile but efficient” strategies to isolate themselves, raising their returns but making it harder for outsiders to duplicate their success. When agents are myopic, the second effect dominates: long run play can shift toward either the risk-dominant equilibrium (with common interests in matching) or toward the inefficient equilibrium (with opposing interests in matching). In contrast, when agents are patient, supra-Nash payoffs can be sustained.