Multidimensional communication mechanisms: cooperative and conflicting designs

This paper investigates optimal communication mechanisms with a two-dimensional policy space and no monetary transfers. Contrary to the one-dimensional setting, when a single principal controls two activities undertaken by his agent (cooperative design), the optimal communication mechanism never exhibits any pooling and the agent's ideal policies are never chosen. However, when the conflicts of interests between the agent and the principal on each dimension of the agent's activity are close to each other, simpler mechanisms that generalize those optimal in the one-dimensional case perform quite well. These simple mechanisms exhibit much pooling. When each activity of the agent is controlled by a different principal (non-cooperative design) and enters separately into the agent's utility function, optimal mechanisms under private communication take again the form of simple delegation sets, exactly as in the one-dimensional case. When instead the agent finds some benefits in coordinating actions, a one-sided contractual externality arises between principals under private communication. Under public communication instead, there does not exist any pure strategy Nash equilibrium with continuous and piecewise differentiable communication mechanisms. Relaxing the commitment ability of the principals restores equilibrium existence under public communication and yields partitional equilibria. Compared with private communication, public communication generates discipline or subversion effects among principals depending on the profile of their respective biases with respect to the agent's ideal policies.