Click to view a PDF (Adobe Acrobat PDF) file.
September 2000
I study self-enforced incentive contracts under imperfect observability, using a model of repeated agency that allows for both common and private performance monitoring. When performance measures are commonly observed, stationary contracts that use the same incentive scheme in every period are optimal. When performance measures are privately observed, self-enforced agreements involve the possibility of separation on the equilibrium path, but optimal contracts still take a basic "termination" form. Using these results, one can characterize optimal long-term contracts as the solution to a static contracting program with an endogenous limit on transfers. With hidden information, optimal contracts call for effort distortion by all types and always involve at least some pooling. With moral hazard, optimal contracts are "one-step" --- a uniform discretionary transfer is made to the agent any time performance is above some cut-off