Bilateral and community enforcement in a networked market with simple strategies

We present a model of repeated games in large buyer-seller networks in the presence of reputation networks via which buyers share information about past transactions. The model allows us to characterize cooperation networks - networks in which each seller cooperates (by providing high quality goods) with every buyer that is connected to her. To this end, we provide conditions under which: [1] the incentives of a seller s to cooperate depend only on her beliefs with respect to her local neighborhood - a subnetwork that includes seller s and is of a size that is independent of the size of the entire network; and [2] the incentives of a seller s to cooperate can be calculated as if the network was a random tree with seller s at its root. Our characterization sheds light on the welfare costs of relying only on repeated interactions for sustaining cooperation, and on how to mitigate such costs.