A Plant Productivity Measure for High-Tech Manufacturing

A total productivity measure was developed as part of an action research productivity analysis project for a manufac turer of computer peripheral devices. The productivity meas ure had to be appropriate for a broad range of automation levels, yet resolve the long-standing methodological problems of index construction in a manner intuitive to management. The difficulties encountered in attempting to institutionalize the measure point to the need for a compromise with cost accounting. Our research objective was to de velop a measure of total plant pro ductivity which would satisfy three criteria: that it express the efficiency of the use of all key plant resources; that it meet the academic's requirements for ri gor; and that the manager find it simple, practical, and useful in a variety of set tings, including highly automated ones. The methodology used was action re search: the project was a collaborative ef fort of a university and a major US-based computer systems manufacturer whom we shall call "Hi-Tech." Partial productivity measures (of which labor productivity is the most common) are powerful tools when directed at spe cific problems [M?ndel 1983], but are often misleading when used to assess overall efficiency ? especially in auto mated settings [Eilon, Gold, and Soesan 1976]. The effort to establish overall plant pro ductivity measures is not new. Davis [1955] made one of the first systematic ef forts to popularize company productivity work [1965] still serves as a useful guide. Frameworks for the managerial use of overall productivity measurement have reached a high level of synthesis [Hayes 1982]. But while the research use of overall measures at a plant level has been sys tematically explored (see the survey in Sumanth [1984]), their practical use has barely progressed. Indeed, Sumanth and Einspruch [1980] found that "total factor" or "total" productivity measures were used by only four of the 73 manufactur ing and operations functions they surveyed in 90 large companies. The reason for the slow diffusion of these measures seems to be that they are (1) too sophisticated and therefore vulner able to subversion by threatened man agers, (2) too simple and therefore unreliable in the presence of such com mon but analytically complex phenomena as product mix changes, or (3) too diffi cult to implement for lack of available data [Stewart 1984]. The challenge lies therefore in "finding the proper balance between concept and reality" [Stewart 1984]. A growing num ber …