Five distinct types of vertical and horizontal decentralization seem to emerge from our discussion. These can, in fact, be placed along a single continuum, from centralization in both dimensions at one end to decentral- ization in both at the other. There are shown in Figure 5-4, as distortions of our logo (where, it should be noted, the inflated size of a shaded part represents its special decision-making power, not its size in membership). Each of the five types of decentralization is discussed briefly below.
Type A: Vertical and Horizontal Centralization Decisional power here is con- centrated in the hands of a single individual, the manager at the top of the line hierarchy—namely, the chief executive officer. Power bulges in Figure 5-4(a) at the strategic apex. The chief executive retains both formal and informal power, making all the important decisions himself and coordinat- ing their execution by direct supervision. As such, he has little need to share his power with staffers, middle-line managers, or operators.
Type B: Limited Horizontal Decentralization (Selective) In this type we find the bureaucratic organization with unskilled tasks that relies on standard- ization of work processes for coordination. (Here is where the experiments in democratization have been concentrated.) The analysts play a leading role in this organization by formalizing the behavior of the other members, notably the operators, who consequently emerge as rather powerless. Standardization diminishes the importance of direct supervision as a coordinating mechanism, thereby reducing the power of the middle-line managers as well, particularly at the lower levels As a result, the structure is centralized in the vertical dimension; formal power is concentrated in the upper reaches of the line hierarchy, notably at the strategic apex. (Should attempts be made to shift it to the operating core as part of a program of democratization, it immediately reverts to the strategic apex by virtue of election procedures.) Because of their role in formalizing behavior, the analysts are, however, able to gain some informal power, which means limited horizontal decentralization. Because the analysts are few relative to the other nonmanagers and their actions serve to reduce the power of the other nonmanagers, notably the operators, the horizontal decentralization turns out to be of the most limited kind. It is selective, in any event, since the analysts are involved only in the decisions concerning work formalization. Figure 5-4(b) shows power bulging at the strategic apex and slightly in the technostructure.
Type C: Limited Vertical Decentralization (Parallel) Here we find the organiza- tion that is divided into market units, or divisions, to whose managers are delegated (in parallel) a good deal of formal power to make the decisions concerning their markets. But because that power need be delegated no farther down the chain of authority, the vertical decentralization is limited in nature. Likewise, because the division managers need not necessarily share their power with staff personnel or operators, the organization can be de- scribed as centralized in the horizontal dimension. Of course, the strategic apex retains ultimate formal power over the divisions. And because it coordi- nates their behavior by the standardization of outputs, effected by perfor- mance control systems designed in the technostructure, a few high-level plan- ners retain some power as well. Thus, Figure 5-4(c) shows the major bulge well up in the middle line and minor ones in the strategic apex and at the top of the technostructure.
Figure 5-4. Five types of decentralization
Type D: Selective Vertical and Horizontal Decentralization Here we see our findings about selective decentralization in the two dimensions coming together. In the vertical dimension, power for different types of decisions is delegated to work constellations at various levels of the hierarchy. And in the horizontal dimension, these constellations make selective use of the staff experts, according to how technical the decisions are that they must make: for some, the experts merely advise the line managers; for others, they join the managers on teams and task forces, sometimes even controlling the choices themselves Coordination within as well as between the con- stellations is effected primarily through mutual adjustment. Power in Fig ure 5-4(d) bulges in various places (corresponding to Figure 5-2), notably in the support staff (especially as compared with the other four types), where a good deal of the organization’s expertise lies.
Type E: Vertical and Horizontal Decentralization Decision power here is con- centrated largely in the operating core—the only bulge in Figure 5-4(e)— because its members are professionals, whose work is coordinated largely by the standardization of skills. The organization is strongly de- centralized in the vertical dimension because this power rests at the very bottom of the hierarchy. And it is strongly decentralized in the horizontal di- mension, since this power rests with a large number of nonmanagers—name- ly, the operators. If another power center were to be identified, it would have to be shown apart, since the organization is forced to surrender a good deal of its control over decision processes to the professional schools that train its opera- tors and the professional associations that control their standards.
1. Decentralization and the other design parameters
The relationship between our two forms of decentralization and the other seven design parameters has been discussed throughout this chapter; here we need merely review these findings briefly.
Decentralization is closely related to the design of positions. The formalization of behavior takes formal power away from the operators and the managers who supervise them and concentrates it near the top of the line hierarchy and in the technostructure, thus centralizing the organi- zation in both dimensions. The result is Type A decentralization. Training and indoctrination produce exactly the opposite effect: They develop ex- pertise below the middle line, thereby decentralizing the structure in both dimensions (Type E). Putting these two conclusions together, we can see that specialization of the unskilled type centralizes the structure in both dimensions, whereas specialization of the skilled or professional type de- centralizes it in both dimensions.
We have also seen a number of relationships between decentraliza- tion and the design of the superstructure. The use of market grouping leads to limited vertical decentralization of a parallel nature (Type C): a good deal of power rests with the managers of the market units. No such definitive conclusion can be drawn for functional grouping. Types B and D are both typically functional structures, the first bureaucratic and rather centralized in both dimensions, the second organic—that is, reliant on mutual adjustment—and selectively decentralized in both dimensions. Similarly, Types A and E, at the two ends of our continuum, are often described as functional. Thus, we are led to the conclusion that functional structure is possible with almost any degree of decentralization, in either dimension.
The same conclusion can be drawn for unit size, or span of control. Too many other factors intervene. For example, large unit size may reflect extensive use of behavior formalization, in which case the structure is rather centralized in both dimensions (Type B). But it may also reflect extensive use of training and indoctrination, in which the structure is de- centralized in both dimensions (Type E). It may also indicate the presence of market-based grouping, which results in limited vertical decentraliza- tion (Type C). Likewise, small unit size may indicate close supervision and centralization (of Type A), or the presence of small autonomous work teams and selective decentralization (of Type D).
As for the lateral linkages, we have seen that performance control systems are used primarily to control quasi-autonomous market units, and so are related to limited vertical decentralization (Type C). Action planning enables the strategic apex to control the important organizational deci- sions, although it must surrender some of its power to the staff planners, which results in Type B decentralization. In general, therefore, planning and control systems emerge as design parameters to effect modest or ex- tensive centralization. And finally, the liaison devices are used primarily to coordinate the work within and between the selectively decentralized work constellations (Type D).
2. Decentralization by part of the organization
We have so far had little difficulty discussing each of the other design parameters by part of the organization. The same will not be true for the two kinds of decentralization, since the distribution of power is an organi- zationwide phenomenon. Nevertheless, some conclusions can be drawn. By definition, vertical decentralization involves only the chain of au- thority—that is, the strategic apex and middle line. And here all kinds of patterns are possible. In some organizations, power remains at the strate- gic apex; in others, it is delegated to various levels in the middle line, sometimes selectively, sometimes in parallel; and in still other cases, power passes right to the bottom of the middle line, and perhaps beyond, to the operating core. If one generalization is in order, it is that classic authority patterns continue to dominate organizational power systems. That is, for- mal power resides in the first instance with the chief executive at the top of the hierarchy. From there it is delegated at his will. And formal power, vis- a-vis the informal, still matters a great deal in organizations. Thus, struc- tures may tend to be more centralized in the vertical as well as the horizon- tal dimension than their situations call for. In other words, there may be a tendency to retain somewhat more power than is necessary in the line structure, especially at the strategic apex.
Horizontal decentralization, by definition, brings the other three parts of the organization—the technostructure, support staff, and operat- ing core—into the power system. Again, we have seen all kinds of power distributions, from negligible staff groups to powerful ones, from weak operating cores to dominant ones. But one point is clear. All have informal power to the extent that they contain expertise. Staff groups do more than just advise when they have the knowledge needed to make technical deci- sions; operators accumulate power when they have the expertise needed to execute managerial decisions and when they are professionals—that is, when they perform jobs based on complex knowledge and skills. As a final point, we might note that within the technocratic units and the higher-level support units, where the work is essentially professional, we would expect to find a good deal of decentralization, from the staff managers to the staff specialists themselves.
We have now discussed our design parameters in some detail. We have seen the various forms each can take in the structure as well as the relation of each to the coordinating mechanisms. Direct supervision is effected through the design of the superstructure, notably the grouping into units, which creates the hierarchy of managerial positions. It is also strongly influenced by the design of the decision-making system—that is, by horizontal and vertical decentralization. Standardization of work pro- cesses is achieved through the formalization of behavior, standardization of skills through the establishment of training and indoctrination pro- grams, and standardization of outputs through the use of planning and control systems. Finally, mutual adjustment is encouraged by the use of the liaison devices.
We have also begun to see some fundamental interrelationships among the design parameters. Some are mutually exclusive. For example, an organization may rely on prejob training or else it may formalize behav- ior through the use of on-the-job rules; it seldom does a great deal of both. Other design parameters are clearly used concurrently—for example, per- formance control systems and market-based grouping, or the liaison de-vices and organic structure. But more important, we have seen a good deal of indication that it is the clustering or configuring of many of these design parameters, not the interacting of any two, that seems to hold the key to understanding the structuring of organizations. But before we can discuss this clustering, we must put our design parameters into the context of the organization’s situation.
Source: Mintzberg Henry (1992), Structure in Fives: Designing Effective Organizations, Pearson; 1st edition.