Academic journal article Economic Inquiry

Mixups in the Warehouse: Centralized and Decentralized Multi-Plant Firms

Academic journal article Economic Inquiry

Mixups in the Warehouse: Centralized and Decentralized Multi-Plant Firms

Article excerpt

LYNN HUNNICUTT (*)

I. INTRODUCTION

All firms receive information and use it to make decisions. Multi-plant firms receive information at scattered locations and must choose how this knowledge will flow and how decisions will be made. The firm might have a single decision maker, which I call centralization, or it might allow each location to make decisions based on any information available to them, a situation I call decentralization.

This article considers issues affecting the choice between one and several decision makers, first through a simple model of a multi-plant firm and then through two case studies that illustrate many of the model's conclusions. The model assumes, and the case studies show, that a single decision maker often has trouble handling all information available from widely scattered locations. Mixups (shipping the wrong bundle to an outlet) are an unavoidable consequence of having a single "boss." (1) When the firm has more than one decision maker, mixups are avoided, but overhead costs are larger. One of the model's key results, which is confirmed by the case studies, is that the centralized firm often "standardizes" as a way to simplify the problem it faces. That is, it ships similar (and less-than-optimal) bundles to all outlets, even though resizing or reforming bundles could increase the profitability of any one outlet.

There is a large literature in which delay (and its minimization) is the driving force behind organizational form. Radner (1993), Van Zandt and Radner (1995), and Van Zandt (1999) analyze the time required to process a batch of information under different organizational structures. Bolton and Dewatripont (1994) extend Radner (1993) by explicitly including both processing and communication delay, and differentiating between the two types. The firm minimizes delay when agents specialize in processing particular types of information. Specialization also occurs in the model of Geanakoplos and Milgrom (1991), in which the firm may dictate which bits of information managers attend to. The firm's problem then is to coordinate the different bits of information its managers hold. Communication delay is at issue in Marschak and Reichelstein (1998), who relate organizational structure to the size of communication costs and message complexity. The issue of time to transmit information is also considered in Govindarajan ( 1986), who notes that communication delay in a rapidly changing environment or in a firm where coordination is crucial may favor centralization. A key assumption of these models is that no worker below the top of the hierarchy is empowered to act without direction from above.

A second function of organizational form is to prevent the firm from undertaking substandard projects. Sah and Stiglitz (1986; 1991) show that when overall project quality is low, hierarchy is preferred, as it is more profitable when individual managers hold potential projects to high standards. This work was extended by Koh (1992), who showed that the firm minimizes its costs if managers undertaking the first review of a project reject it when they work in a hierarchy and accept it when they work in a polyarchy. Thus, managers working in a hierarchy should have a higher standard for acceptance than those working a polyarchy. Sali (1991) describes managerial ability as also affecting the choice of organizational form, since a single manager's influence is larger in a centralized organization than under decentralization. When the manager is not able, centralization may lead to markedly lower profits than a decentralized firm would achieve.

Discovery and implementation of optimal policies may also be affected by the organization's structure. A single location may search for the optimal policy to be implemented everywhere (centralization), or all units may search and may or may not adopt policies selected by others (decentralization). Better policies are discovered when the organization employs multiple searchers, and/or when the single searcher becomes more sophisticated. …

Search by... Author
Show... All Results Primary Sources Peer-reviewed

Oops!

An unknown error has occurred. Please click the button below to reload the page. If the problem persists, please try again in a little while.