Optimal Incentive Contract with Costly and Flexible Monitoring

Recent advances in IT and data science give firms more flexibility to process, store and communicate the growing volume and variety of performance data at a reduced and yet significant cost. In this paper, we study the impact of this cost and flexibility on employee monitoring and the internal organization of firms. In an otherwise standard principal-agent model with moral hazard, we allow the principal to adopt any monitoring technology that classifies the agent's raw performance data into finite categories, at a cost that increases with the fine-grainedness of the information that the induced performance signal carries. Through analyzing the trade-off between the compensation cost and the monitoring cost, we obtain characterizations for the optimal monitoring technology, such as information aggregation, strict MLRP, the fine-tuning of monitoring intensity across tasks according to the agent's tendency to shirk, and the coexistence of individual and group evaluation systems among technologically similar firms. We apply these results to human resource management and suggest new explanations for long-lasting puzzles