Monitoring, Transparency and Accountability

Pierre Fleckinger (Mines Paristech, CERNA)
Nicolas Roux (Max Planck Institute)

Abstract: A monitor is hired by a principal to oversee an agent's activity and to report on the agent's result. The monitor first collects information about the agent so as to develop an expertise in the agent's activity. This expertise makes her more likely to learn the result's value in the future. The agent observes the level of expertise of the monitor and makes a productive effort. Once the result is realized, the monitor decides whether to report its value to the principal (if she knows it). The principal uses the report to form an expectation of the result's value and subsequently rewards or punishes both the monitor and the agent as a function of his expectation. The monitor then is accountable for the result in the sense that she is punished if the agent's result is bad. In this model, the monitor faces a commitment issue, which undermines incentive provision. Ex ante, she would like to disclose the result's value any time she observes it so that her oversight disciplines the agent. Ex post however, she wants to inflate the principal's expectation and therefore hides bad results. This paper studies the monitor's incentives and asks whether incentive provision benefits from the principal observing the expertise level of the monitor (transparency). We show that the answer involves a trade-off between making expertise valuable to the monitor and making expertise deterrent for the agent.