Centralized vs. decentralized management: an experimental study

Abstract

We introduce a new game to the experimental literature and use it to study how behavioral phenomena affect the tradeoffs between centralized and decentralized management. Our game models an organization with two divisions and one central manager. Each division must choose or be assigned a product. Ignoring asymmetric information, the underlying game is an asymmetric coordination game related to the Battle of the Sexes. In equilibrium, the divisions coordinate on identical products. Each division prefers an equilibrium where the selected products are closest to its local tastes while central management prefers the efficient equilibrium, determined by a randomly state of the world, which maximizes total payoffs. The state of the world is known to the divisions, but the central manager only learns about it through messages from the divisions who have incentives to lie. Contrary to the theory, overall performance is higher under centralization, where the central manager assigns products to divisions after receiving messages from the divisions, than under decentralization where the divisions choose their own products. Underlying this, mis-coordination is common under decentralization and divisions fail to use their information when they do coordinate. Mis-coordination is non-existent under centralization and there is a high degree of truth-telling by divisions as well. Performance under centralization is depressed by persistent sub-optimal use of information by central managers.