Network-based peer monitoring design

We study a team incentive design problem where multiple agents are located on a network and work on a joint project. The principal seeks the least costly mechanism to incentivize full effort, by choosing the work assignment sequence and the rewards to the agents upon success. Whereas the agents' actions are hidden to the principal, they may be observed among the agents given the internal information that is determined by the network and the sequence. Under effort complementarity, the transparency of the agents' actions can reduce their incentive costs, but exhibits diminishing marginal effectiveness. This gives rise to the desire to balance internal transparency when it is infeasible to uniformly enhance transparency. For several typical network topologies, we derive explicit properties of an optimal assignment sequence, and propose two new measures, total cost and stability, for the principal to rank these networks.