So far, all this seems clear enough. But that is only because we have not yet looked inside that black box called decentralization. The fact is that no one word can possibly describe a phenomenon as complex as the distribution of power in the organization. Consider the following questions:
- Which is more centralized: a library called “centralized” because it is in one place, although most of the decision-making power is dispersed to its department heads; or a “decentralized” library system, consisting of widely scattered satellite librarigs, where the chief librarian of each guards all the power, sharing it with none of the other employees?
- How about the organization where decision-making power is dis- persed to a large number of people but, because their decisions are closely monitored by a central individual who can fire them at a moment’s notice, they make those decisions with careful assess- ment of his wishes? Or the case of the Jesuit priest or CIA agent who has complete autonomy in the field, except that he has been carefully indoctrinated to decide in a given way before he ever left the central headquarters? Are these organizations decentralized?
- In the United States, divisionalized corporations that rely on per- formance control systems for coordination are called “de-centralized,” whereas Americans are in the habit of calling the communist economies “centralized,” even though they are orga- nized like giant divisionalized corporations that rely on perfor- mance control systems for coordination. Which is it?
- Does standardization of the work process bring about centraliza- tion or decentralization? When a worker, because he is subject to a great many rules, is left free of direct supervision, can we say that he has power over his decisions? More generally, are bureau- cracies centralized or decentralized? How about the one Crozier describes, where the workers force through rules that reduce the power of their managers over them, with the result that both end up in straitjackets?
- What about the case where a line manager has the authority to make a decision, but his advisors, by virtue of their superior tech- nical knowledge, lead him into his choices? Or the case where the manager decides but, in executing the choices, his subordinates twist the outcome to their liking? Are these organizations cen- tralized by virtue of the distribution of the formal power, or de- centralized by virtue of the distribution of the informal?
- Finally, what about the organization where some decisions—say, those concerning finance and personnel—are made by the chief executive, and others—say, those in the areas of production and marketing—are dispersed to managers lower down? Is it cen- tralized or decentralized?
The answer to these questions is that there is no simple answer, that unqualified use of the term centralization or decentralization should always be suspect. Yet a great deal of the research and discussion on organization structure has used them in just that way.
So the waters of decentralization are dirty. But before spilling them away, it may be worthwhile to see if we can find a baby in there.
Our list of questions seems to indicate two major points about the concept. First, centralization and decentralization should not be treated as absolutes, but rather as two ends of a continuum. The Soviet economy is not “centralized,” just more centralized than a capitalist economy; the divisionalized firm is not “decentralized,” just more decentralized than some firms with functional structures.1 Second, much of the confusion seems to stem from the presence of a number of different concepts fighting for recognition under the same label. Perhaps it is the presence of two or even three babies in that bathwater that has obscured the perception of anyone.
Below we discuss three uses of the term decentralization and retain two for our purposes. Each is discussed at length in the body of this chapter, and together they are used in a summary section to develop a framework of five basic kinds of decentralization commonly found in organizations.
1. Three uses of the term decentralization
The term decentralization seems to be used in three fundamentally different ways in the literature:
- First is the dispersal of formal power down the chain of authority. In principle, such power is vested in the first instance in the chief executive at the strategic apex. Here it may remain, or the chief executive may choose to disperse it—delegate is a common syn- onym for this kind of decentralization—to levels lower down in the vertical hierarchy. The dispersal of formal power down the chain of line authority will be called vertical decentralization.
- Decisional power—in this case, primarily informal—may remain with line managers in the system of formal authority, or it may flow to people outside the line structure—to analysts, support specialists, and operators. Horizontal decentralization will refer to the extent to which nonmanagers control decision processes.2
- Finally, the term decentralization is used to refer to the physical dispersal of Libraries, copying machines, and police forces are “centralized” in single locations or “decentralized” to many, to be close to their users. But this “decentralization” has nothing per se to do with power over decision making (the satellite library, like the copying machine, may not make the decisions that most affect it). Thus, this third use of the term only serves to confuse the issue. In fact, we have already discussed this concept in Chapter 3, using the terms concentrated and dispersed instead of centralized and decentralized. In this book, the term decentralization will not be used to describe physical location.
This leaves us with two essential design parameters: vertical and horizontal decentralization. Conceptually, they can be seen to be distinct. Power can be delegated down the chain of authority and yet remain with line managers; the ultimate case of this vertical decentralization with hori-zontal centralization would give all the power to the first-line supervisors Alternatively, senior staff people could hold all the power. Centralizatior of both types occurs when the strategic apex keeps all the power; de- centralization of both sees power pass all the way down the chain of authority and then out to the operators.
But power over all decisions need not be dispersed to the same place. This gives rise to two other kinds of decentralization. In selective de- centralization, the power over different kinds of decisions rests in differ- ent places in the organization. For example, finance decisions may be made at the strategic apex, marketing decisions in the support units, and production decisions at the bottom of the middle line, by the first-line supervisors. Parallel decentralization refers to the dispersal of power for many kinds of decisions to the same place. For example, finance, market- ing, and production decisions would all be made by the division managers in the middle line.
But before we can begin our discussion of the kinds of decentraliza- tion found in organizations, we need to consider one more issue. Even within a single decision process, the power wielded by different people can vary. We need a framework to understand what control over the decision process really means.
What matters, of course, is not control over decisions per se but ultimately control over actions—what the organization actually does, such as marketing a new product, building a new factory, hiring a new mechan- ic. And actions can be controlled by more than just making choices. Power over any step in the decision process, from initiating the original stimulus to driving the last nail in the final execution of it, constitutes a certain power over the whole process.
Paterson provides us with a useful framework for understanding this issue. He depicts the decision process as a number of steps, as shown ir modified form in Figure 5-1: (1) collecting information to pass on to the decision maker, without comment, about what can be done; (2) processing that information to present advice to the decision maker about what should be done; (3) making the choice—that is, determining what is intended to be done; (4) authorizing elsewhere what is intended to be done; and (5) doing it—that is, executing what is, in fact, done. The power of an individual is then determined by his control over these various steps. His power is maximized—and the decision process most centralized—when he controls all the steps: when he collects his own information, analyzes it himself, makes the choice, need seek no authorization of it, and then executes it himself. As others impinge on these steps, he loses power, and the process becomes decentralized.
Figure 5-1. A continuum of control over the decision pro- cess (similar to Paterson, 1969:150)
Control over input information enables another person to select what factors will—and will not—be considered in the decision process. When information is filtered extensively, such control can be tantamount to con- trol over the choice itself. More important still is the power to advise, since it directs the decision maker down a single path. Classical line/staff distinc- tions notwithstanding, there are times when the separation between giv- ing advice and making the choice is fine indeed. History tells us of kings who were virtual figureheads, while their advisors—a Richelieu in France, a Rasputin in Russia—controlled the affairs of state. Control over what happens after the choice has been made can also constitute power. The right to authorize a choice is, of course, the right to block it or even change it. And the right to execute a choice once made often gives one the power to twist or even distort it. Newspapers carry accounts every day of how the “bureaucrats” misdirected the intentions of the politicians and ended up doing what they thought best in the first place. In effect, the decisions ended up being theirs.
And so, a decision process is most decentralized when the decision maker controls only the making of the choice (the least he can do and still be called decision maker): In the organizational hierarchy, he loses some power to the information gatherers and advisors to his side, to the author- izers above, and to the executers below. In other words, control over the making of choices—as opposed to control over the whole decision pro- cess—does not necessarily constitute tight centralization. With this in mind, let us now look at vertical and horizontal decentralization.
Source: Mintzberg Henry (1992), Structure in Fives: Designing Effective Organizations, Pearson; 1st edition.