Abstract for: A Dynamic Model of Centralized vs. Decentralized Process Improvement

Vigorous debate exists in process improvement circles between the proponents of “lean” and other methods in which line workers lead process improvement efforts and advocates for more centralized approaches (such as six sigma) in which improvement is lead by experts. We build a model that simulates two kinds of process issues: those that are system-wide and those that are localized. We also simulate centralized and decentralized process improvement efforts to manage each of these two types of process issues. This allows us to explain a number of interesting phenomena found in the literature. In particular, we have captured the benefit—seen in e.g. Toyota—of beginning with a centralized period of process improvement attacking systemic problems followed, when those returns begin to diminish, by a period of decentralized process improvement that attacks local problems. Our simulations have also supported the possibility that high levels of technology change may cause a firm to prefer to remain forever in the centralized mode of process improvement. Finally, our simulations have also shown that, in environments with many local issues like healthcare, a concentration on decentralized process improvement may make more sense, even if the workers doing the process improvement are not highly trained.