Modular Organization

Niko Matouschek (Northwestern University)
Michael Powell (Northwestern University)
Bryony Reich (Northwestern University)

Abstract: The defining products of our times are made up of modules, groups of components that are highly connected to each other but only loosely connected to those in other groups. We explore the implications of modular production for the emergence of modular organizations. To this end, we develop a team-theoretic model with costly communication. Production is described by a production network where each node represents a decision and a state, and the weight of the link between two nodes captures the need to coordinate the two decisions. Each node has an agent who observes the state, communicates with other agents in a designed communication network, and makes the decision. The optimal communication network trades off the cost of an additional link with the gain from better coordination. To explore modular production, we assume production has a community structure where each decision belongs to one community. The need for coordination is higher for decisions in the same community. We show the optimal organization will only be modular if the properties of the production function are moderate: there are few modules, none is large, and the need for coordination within modules is neither too high nor too low. We extend the model to allow for an overlapping community structure in which each decision belongs to two communities, and we generate testable predictions for the emergence of commonly observed organizational structures, such as M-form, U-form, and matrix organizations.