Cooperation and Community Responsibility

I consider markets in which participants have very little information: for instance, agents are anonymous, cannot verify each other's identities, or have little information about each other's past transactions. I ask whether it is possible to prevent opportunistic behavior in such settings in the absence of contractual enforcement. I model such markets as repeated anonymous-random-matching games and show that cooperation is sustainable if players are sufficiently patient and can announce their name (though unverifiable) before every transaction. Cooperation is achieved by "community responsibility": if a player deviates, her entire community is held responsible and punished by the victim. Sustaining cooperation involves partial authentication of identities by checking players' knowledge about past transactions.