29 Pages Posted: 6 Jul 2013 Last revised: 28 Feb 2015
Date Written: December 1, 2014
In this paper I examine a T-period agency model with imperfect public monitoring between a risk-neutral principal and a risk-averse agent where signals can depend on the agent's past actions and exhibit serial correlation. In this general environment, I show that near-efficiency obtains when T is large if the monitoring technology satisfies two basic properties: concentration of measure and informativeness. The tension between these properties determines the boundary at which asymptotic efficiency obtains in agency models with frequent actions, unifies and extends various efficiency results in the agency literature, quantifies the value of knowing detailed features of signal processes and solves a large class of incentive problems with highly persistent monitoring technologies.
Keywords: dynamic agency; efficiency; robust incentives
JEL Classification: D86
Suggested Citation: Suggested Citation