Washington University in Saint Louis
September 1, 2014
In dynamic agency models with imperfect public monitoring over finite T instances where signals can depend arbitrarily on past actions and exhibit moderate serial correlation, one can implement near-efficient outcomes when T is large if there exists a performance test statistic that has sufficiently concentrated measures and conveys enough information about the agent's true profit contribution asymptotically. Conversely, in a large class of agency models with frequent actions, asymptotic near-efficiency obtains only if this condition is satisfied. To show these results, I construct test contracts which attain robust performances for each T even if details of the signal process are not exactly known to the contract designer.
Number of Pages in PDF File: 41
Keywords: dynamic agency; robust incentives
JEL Classification: D86working papers series
Date posted: July 6, 2013
© 2014 Social Science Electronic Publishing, Inc. All Rights Reserved.
This page was processed by apollo3 in 0.250 seconds