Automation and Task Allocation Under Asymmetric Information

Abstract

A firm can complete the tasks needed to produce output using either machines or workers. Unlike machines, workers have private information about their preferences over tasks. I study how this information asymmetry shapes the mechanism used by the firm to allocate tasks across workers and machines. I identify important qualitative differences between the mechanisms used when information frictions are large versus small. When information frictions are small, tasks are substitutes: automating one task lowers the marginal cost of other tasks and reduces the surplus generated by workers. When frictions are large, tasks can become complements: automation can raise the marginal cost of other tasks and increase the surplus generated by workers. The results extend to a setting with multiple firms competing for workers.