Authority in a Simple Screening Model

I study a simple principal agent screening model where the agent's type determines hispayoff as well as his preferences over activity levels. The agent's activity inflicts a payoffexternality on the principal, and she proposes a revelation mechanism to the agent.When the agent needs the principal's consent to undertake the activity (Principalauthority), the optimal contract has standard properties and stipulates inefficientlylow (efficient) activities for the low-type (high-type) agent. When instead the agenthas 'authority' over the activity so that his participation constraint becomes typedependent, the principal (1) implements the efficient activity as long as the payoffexternality is not too large; and (2) if it is large, outcomes depend on the sign of theexternality. While agent authority often yields a larger total surplus (always so whenexternalities are positive), principal authority can be more efficient when the activitygenerates large negative externalities. Finally, uner the optimal mechanism all agents'work harder' under agent authority than under 'principal authority'