Working Papers

In this paper, we study the conditions under which termination is a useful incentive device in the canonical dynamic principal-agent moral hazard model of Sannikov (2008). We find that temporary suspension of the agent after poor performance dominates termination if the principal's outside option is low and the agent's outside option is moderate. In suspension, the agent performs tasks free of moral hazard and receives no compensation, which rebuilds his "skin in the game" and allows for incentives to be restored without terminating. If the agent's outside option is low, suspension is ineffective because it rebuilds the agent's skin in the game too slowly. If the agent's outside option is high, the profitability of the relationship with the agent is low, so the principal prefers to terminate rather than extend the relationship through temporary suspension. Because the optimal use of suspension versus termination after poor performance can be highly sensitive to the principal's and agent's outside options, similar jobs can have vastly different average job durations, purely for incentive reasons.

*This paper was previously published under the title "Optimal Contracts with Reflection."