AUTOMATION, SIZE, AND THE LOCUS OF DECISION MAKING: THE CASCADE EFFECT.

This paper began with the problem of the effects of size and automation on the decentralization of decision making. On the basis of previous theory and re- search I expected to find that increases in size force decision making to become decentralized, while automation allows recentralization to occur, since the rapid feedback of information to top management enables top management to make decisions which formerly had to be made at lower levels. However, the data showed exactly the opposite set of relationships. Automation had a strong positive effect on decentralization, and size had a weak negative effect. This led to reconsideration of the original hypothesis, which proved to have two obvious weaknesses. First, office automation, at least in the state employment agencies, is used only for the most routine tasks, so that any effect that it has on the decentralization of hiring decisions must be very indirect. Second, the argument assumed that centralization is always and inherently desirable and that top executives decentralize only out of necessity. This assumption is certainly open to question, especially in large organizations, where the decentralization of some decisions may be quite desirable but impossible unless there are mechanisms which routinize the internal organizational processes, thereby increasing the capacity of people at all levels of the organization to take on new responsibilities and to delegate others. Automation has this effect in that it reduces routine problems at the lowest level of the organization. This frees first-line supervisors from some of their day- to-day problems and allows them to accept the responsibility for decisions which had been forced back up to higher levels in the hierarchy. Each level is, in turn, freed from its more routine tasks; and the process repeats itself on up to the director, who sets the flow of authority in motion by delegating some of his own authority so that he can take on more appropriate executive functions such as long-range planning and the maintenance of interorganizational contacts. The pressure created by the new responsibilities which thus devolve on the second level force personnel on that level to further delegate some of their more routine responsibilities, and authority cascades down through the hierarchy to fill the vacuum which automation has created. The use of this model enabled us to explain why there should be such a strong correlation between the two decentralization variables, assuming that one of them is a decision on a higher level than the other. It also showed that the effect of automation on lower-level decisions was indirect, since it travelled entirely through higher-level decisions. The plausibility of the model was further strengthened by the fact that an alternative model which assumed that the correlation between automation and decentralization was spurious, due to the effects of size and formalization of personnel procedures, could not explain away the correlation between automation and decentralization. [ABSTRACT FROM AUTHOR]