Equilibrium Trust

We build a simple model of trust as an equilibrium phenomenon, departing from standard "selfish" preferences in a minimal way. Agents who are on the receiving end of an other to transact can choose whether to cheat and take away the entire surplus, taking into account a "cost of cheating." The latter has an idiosyncratic component (an agent's type), and a socially determined one. The smaller the mass of agents who cheat, the larger the cost of cheating suffered by those who cheat. Depending on the parameter values, the model can have a unique equilibrium level of trust (the proportion of transactions not cheated on), or two equilibria, one with high and the other with low trust. Thus, differences in trust levels across societies can reflect different fundamentals or, for the same fundamentals, a switch across multiple equilibria. Surprisingly, we find that these two possibilities are partially identifiable from an empirical point of view. Our model can also be reinterpreted as one with standard selfish preferences and an enforcement agency with limited resources that are used to catch and fine a subset of those who cheat. Lastly, we carry out a robustness exercise in which agents learn in a simple way from experience about how many agents cheat in society. Our results indicate that when there are multiple equilibria the high trust equilibrium is less robust than the low trust one.