Social Planning with Partial Knowledge of Social Interactions

Economists studying collective decision problems often consider how a social planner would behave. The standard exercise presumes complete knowledge of the welfare achieved by each feasible policy. However, we often have only partial knowledge of policy impacts. This paper extends my program of research on planning under ambiguity from settings with individualistic treatment to ones where treatments interact, each person's outcome depending on his treatment and on the population treatment allocation. I consider the problem in abstraction and use medical treatment of an infectious disease to illustrate.