Cooperation via Social Networks

Sufficiently frequent interaction between partners has been identified by, a.o., Axelrod as a more-or-less sufficient condition for stable cooperation. The underlying argument is that rational cooperation is ensured if short-term benefits from opportunistic behavior are offset by the long-term costs of sanctions imposed on the culprit. This argument does not require that these sanctions are (only) imposed by the victims. In this paper, we develop a model for ‘embedded trust’ in which a trustee interacts with a number of trustors who may communicate via a social network with each other about the behavior of the trustor. We analyze the model under the assumption that a trustor uses a non-forgiving provocative strategy, i.e., withholding trust if (1) the trustee ever abused trust against herself, or if (2) she has information that he ever abused trust against someone else. The analysis reconfirms the standard predictions about how the level of trust depends on the payoffs and shadow of the future. We provide new predictions both on between-network effects (“which network is more favorable for cooperation?”) and on within-network effects (“in what network position can you trust more?”).