Common agency and public good provision under asymmetric information

The provision of public goods under asymmetric information has most often been viewed as a mechanism design problem under the aegis of an uninformed mediator. This paper focuses on institutional contexts without such a mediator. Contributors privately informed on their willingness to pay non‐cooperatively offer contribution schedules to an agent who produces the public good on their behalf. In any separating and informative equilibrium of this common agency game under asymmetric information, instead of reducing marginal contributions to free‐ride on others, principals do so to screen the agent's endogenous private information obtained from privately observing other principals' offers. Under weak conditions, the existence of a differentiable equilibrium is shown. Equilibria are always ex post inefficient and interim efficient if and only if the type distribution has a linear inverse hazard rate. This points to the major inefficiency of contribution games under asymmetric information and stands in contrast to the more positive efficiency result that the common agency literature has unveiled when assuming complete information. Extensions of the model address direct contracting between principals, the existence of pooling uninformative equilibria, and the robustness of our findings to the possibility that principals entertain more complex communication with their agent.