Designing Incentives under Multidimensional Performance Measures

We attempt to shed some light on the problem of providing incentives to service providers such as teachers and doctors. Often, outcomes of such services are not verifiable, and this has been cited as a reason for lack of incentive provision. We derive the contract offered by a principal if, in addition to moral hazard, the output of the agent may not be verified. We show that verified high attainment may not be rewarded. In this case, the bonus is increasing in the ability to verify output. We also show that an increase in potential nonverifiability may lead to efficiency gains and be beneficial for the principal.