Centralization vs. decentralization in a multi-unit organization: A computational model or a retail chain as a multi-agent adaptive system

A computational model of a retail chain is developed in which store managers continually search for better practices. Search takes place over a rugged landscape defined over the space of store practices. The main objective of this research is to determine how the amount of discretion given to store managers as to how they run their stores influences the rate of innovation at the store level. We find that greater decentralization enhances firm performance when stores' markets are sufficiently different, the horizon is sufficiently long, and markets are sufficiently stable.