Organizational Structure and Perpetual Innovation: A Computational Model of a Retail Chain

Historically, Sears, Roebuck has been a relatively decentralized organization (Worthy, 1984) which, it has been argued, is a source of their success (Emmet and Jeuck, 1950). Analogous comments have been made to explain the superior performance of Wal-Mart (Collins and Porras, 1994). While anecdotes are many, there has been no formal analysis as to how organizational structure influences performance. Is there one organizational structure that is best? Or does it depend on a chain's environment? If so, what are the pertinent features of the environment and how do they influence performance? The objective of this research is to provide some theoretical insight into these questions by constructing a computational model of a retail chain in which corporate headquarters (HQ) and stores are constantly adapting to their respective environments. Central to our work is the perspective that innovation is an act of information creation and that the performance of a retail chain depends critically on the way organizational structure influences the effectiveness of communicating and utilizing this information. The dimension of organizational performance that we focus on is the dynamic one alluded to in the description of Wal-Mart: the rate of improvement in store practices as the result of adopting useful innovative ideas. Given this conceptual framework, we model a retail chain as a two-level hierarchy consisting of a corporate headquarters (HQ) and a number of local stores, each of which is in a distinct market. Generation of innovative ideas is distributed within the organization: they are generated at the HQ as well as at the local stores. The chain's organizational structure determines where authority lies for deciding whether these ideas are adopted or discarded. A centralized organization is one in which the authority to implement new practices (ideas) rests at high levels in the hierarchy. As such, all new ideas that are deemed useful must receive the HQ's approval for the ultimate adoption at the local store level. A decentralized organization allows the store managers to independently implement new practices without the HQ approval. Our task is to characterize how organizational form influences the profit performance of a retail chain in both the short-run and long-run attained through dynamic adoption of new discoveries. Our model of a retail chain begins with the view that a store, at any given point in time, is characterized by its current operating practices or what Nelson and Winter (1982) would refer to as "routines." The practices of a retail store determine, for example, merchandising, marketing, pricing, and personnel policy. The merchandising dimension encompasses issues of product selection, inventory, and contractual arrangements with vendors while marketing includes displays and advertising. Pricing may concern the decision of offering everyday low prices or sales. The personnel dimension entails the allocation of employees (for example, how many workers are on the floor), training programs, and compensation schemes. The profit of a store is presumed to be the result of the interaction between its current routines along these dimensions and the practices desired by the consumers in its market. Given the set of embedded routines, an innovation is viewed as a new way of running a store as represented by a change in a routine. The dynamic of innovation consists of two primary processes: 1) idea generation the process by which new ideas are created; and 2) idea adoption the process by which new ideas are sifted through and either