Operations Systems with Discretionary Task Completion

摘要

In this paper, we have presented analytical models that capture one important characteristic of many operations systems, particularly those involving service and professional work—discretionary task completion. Our main insight is that, in operations systems with discretionary task completion times, it is attractive to adjust processing time (hence job quality) in response to system congestion. In factory physics terminology, quality is a variability buffer (along with capacity, inventory, and time). The presence of this fourth type of variability buffer causes systems with discretionary completion times to exhibit some very different characteristics from systems with nondiscre-tionary completion times. The work in this paper represents a first step toward modeling and understanding work systems with discretionary process times. To develop a more comprehensive set of principles concerning design and management of such work systems, further modeling work is needed. For instance, it would be useful to extend the single-station model to network models of multistage DTC work systems. This raises questions about interactions and impact of discretionary task completion at the different stages. Another useful area to explore for DTC systems would be the setting where the objectives of the employee and the firm do not align. This could be modeled as a version of the classic principle-agent problem and used to investigate interesting questions relating to incentives and utility characterization. The long-term goal of research on DTC work systems is to gain insights into the best ways to design, manage, and improve operations in which workers have discretion in task completion. By understanding the interactions between capacity, variability, information, and customer service (determined by both quality and responsiveness), DTC workers can make better time management decisions, while managers can make better decisions regarding cross training, information sharing, and collaboration policies.