Division of Labor Through Self-Selection

Self-selection–based division of labor has gained visibility through its role in varied organizational contexts such as nonhierarchical firms, agile teams, and project-based organizations. Yet, we know relatively little about the precise conditions under which it can outperform the traditional allocation of work to workers by managers. We develop a computational agent-based model that conceives of division of labor as a matching process between workers’ skills and tasks. This allows us to examine in detail when and why different approaches to division of labor may enjoy a relative advantage. We find a specific confluence of conditions under which self-selection has an advantage over traditional staffing practices arising from matching: when employees are very skilled but at only a narrow range of tasks, the task structure is decomposable, and employee availability is unforeseeable. Absent these conditions, self-selection must rely on the benefits of enhanced motivation or better matching based on worker’s private information about skills, to dominate more traditional allocation processes. These boundary conditions are noteworthy both for those who study as well as for those who wish to implement forms of organizing based on self-selection.