Optimal Contracting with Costly State Verification, with an Application to Crowdsourcing

We study incentive design in an agency setting when monitoring is costly, and this cost is endogenously determined by the contract structure. A firm employs multiple workers to obtain costly unverifiable information about an unknown state (a task) and report it to the firm. The firm uses peer-monitoring to incentivize workers by hiring multiple workers and conditioning payment upon how a worker’s message compares to the messages of other workers. We place two key restrictions on the contract space: workers are protected by limited liability and the firm cannot commit to truthfully reveal whether a worker was monitored (partial commitment). We derive the firm’s optimal contract under both full and partial commitment. The optimal contracts exhibits three key features: (i) the structure of the optimal monitoring technology depends crucially on the commitment power of the firm – virtual monitoring, or monitoring with arbitrarily small probability, is optimal under full commitment while stochastic monitoring, or monitoring with strictly positive probability, is optimal under partial commitment; (ii) bundling – simultaneously assigning a worker multiple tasks – reduces worker rents and monitoring inefficiencies; and (iii) approximate efficiency is achievable under full commitment but not under partial commitment. We conclude with an application to contract design on crowdsourcing platforms such as Mechanical Turk, and characterize the optimal contract for simple tasks found on these platforms, such as moderating the content of user-generated images or advertisements.