Asymptotic Efficiency in Dynamic Principal-Agent Problems

In a seminal paper, B. R. Holmstro m and P. R. Milgrom (1987, Econometrica 55, 303 328) examine a principal-agent model in which the agent continuously controls the drift rate of a Brownian motion. Given a stationary environment, they show that the optimal sharing rule is a linear function of aggregated output. This paper considers a variant of the Brownian model in which control revisions take place in discrete time. It is shown that no matter how ``close'' discrete time is to continuous time, the first-best solution can be approximated arbitrarily closely with a random spot check and a suitably chosen sequence of step functions. Journal of Economic Literature Classification Numbers: D82, J33. 2000 Academic Press