Horizontal Decentralization in Organization

Now we turn to the question of horizontal decentralization—namely, to the shift of power from managers to nonmanagers (or, more exactly, from line managers to staff managers, analysts, support specialists, and opera- tors). An assumption in our discussion of vertical decentralization was that power—specifically formal power, or authority—rests in the line structure of the organization, in the first instance at the strategic apex. Vertical decentralization dealt with the delegation of that power down the chain of authority, at the will of the top managers.

When we talk of horizontal decentralization, we broaden the discus- sion in two regards. First, in discussing the transfer of power out of the line structure, we move into the realm of informal power, specifically of control over information gathering and advice giving to line managers and the execution of their choices, as opposed to the making and authorizing of these choices. And second, in discussing horizontal decentralization, we drop the assumption that formal power necessarily rests in the line struc- ture, in the first instance at the strategic apex. Here formal power can rest elsewhere—for example, with operators who are empowered to elect the managers of the strategic apex.

Assuming the presence of managers, analysts, support staff, and operators, we can imagine a continuum of four stages of horizontal de- centralization, listed below:

1. Power rests with a single individual, generally by virtue of the office he occupies (i.e., a manager).

2. Power shifts to the few analysts of the technostructure, by virtue of the influence their systems of standardization have on the decisions of others.

3. Power goes to the experts—the analytic and support staff special- ists, or the operators if they are professional—by virtue of their knowledge. 
4. Power goes to everyone by virtue of membership in the organization.

Thus, in the most horizontally centralized organization, one person holds all the power, typically the top manager. Of course, even here, there can be variations according to how open that person is to advice. There is a difference between the “omnicompetent, aloof, imperial ruler,” such as the Byzantine emperor, and the “omnicompetent but very accessible and responsive leader,” such as a John F. Kennedy (Kochen and Deutsch, 1973:843). Hereafter, we find different degrees of horizontal decentraliza- tion, first to a few analysts whose systems control the behavior of others, then to all the experts with knowledge, and finally to everybody just be- cause everybody is a member of the organization. The first case requires no further discussion; let us therefore consider the other three in turn.

1. Power to the analysts

When an organization relies on systems of standardization for coordina- tion, some power must pass out from the line managers to the designers of those systems, typically the analysts of the technostructure. How much power, of course, depends on the extent and the kind of standardization. Obviously, the more the organization relies on systems of standardization for coordination, the greater the power of the analysts. Soviet government planners have more power than their American counterparts; the work- study analysts of an automobile company are more influential than those of a hospital. And the tighter the kind of standardization, the more power- ful the analysts. By that token, job designers and work-study analysts— those who tell workers how to produce by standardizing their work pro- cesses—should typically have more power than production schedulers and planners—those who only tell them what and when to produce by standard- izing their outputs. And trainers—those who teach people to produce by standardizing their skills—should have less power still. Thus, the factory worker would normally perceive the work-study analyst as the greatest threat to his autonomy, followed by the production scheduler and then the trainer.⁴

Who surrenders power to the analysts? Obviously, those whose work is standardized, such as the operator who loses the power to choose his work process, or the manager who loses the power to decide on his unit’s outputs. But so, too, do the managers of these people; as noted earlier, their jobs became institutionalized, technocratic standardization replacing their power of direct supervision.

This leads us to two important conclusions. First, power to the ana- lysts constitutes only a limited form of horizontal decentralization. Only a few nonmanagers—these designers of the technocratic systems—gain some informal power, and that at the expense of the many operators and others whose behavior and outputs are standardized. And second, this kind of limited horizontal decentralization in fact serves to centralize the organization in the vertical dimension, by reducing the power of the lower-line managers relative to those higher up. In other words, organiza- tions that rely on technocratic standardization for coordination are rather centralized in nature, especially in the vertical dimension but also some- what in the horizontal.

Are bureaucracies centralized? This has been a controversial question in the research literature. As we have seen, the research has not been conclusive. Some researchers have argued that bureaucratic work stan- dards, by limiting the power of the manager to exercise direct supervision, thereby give more power to the workers. The work of Crozier suggests quite a different conclusion: that both end up in a straitjacket, with deci- sion-making power flowing up to a remote central headquarters.

We can sort out much of this confusion by discussing centralization in terms of our five coordination mechanisms. Those who see work rules as giving rise to decentralization seem to equate centralization with direct supervision: an organization is centralized if direct supervision is close; to the extent that work standards replace direct supervision, the organization becomes decentralized. But calling a bureaucracy decentralized because work rules instead of managers control the workers is like calling puppets purposeful because computers instead of people pull their strings.

Direct supervision may be the tightest coordinating mechanism, and therefore close control by managers may constitute the tightest form of horizontal centralization. Any move the individual makes can bring a rap on the knuckles from the boss: “That is not the way I expected you to do it.” And standardization of work processes may provide the employee with more autonomy, since he knows what he can and cannot do. But that does not mean that it is a loose coordinating mechanism. Of course, if the rules are few, the employee has considerable discretion. But we are here discussing organizations where the rules are many—bureaucracies that rely on such rules for coordination, and so proliferate them. The important point is that reliance by the organization on any of the other coordinating mechanisms would yield its employees more freedom still in their work. That would happen if their outputs were standardized and they were allowed to choose their own work processes. Better still, if their work was coordinated by the standardization of skills, they would be trained and indoctrinated before they started to work and thereafter would be left alone to choose their work processes and determine their outputs as they saw fit. And best of all would be the absence of standardization and direct supervision altogether; the employees would be completely free to work out their own coordination by mutual adjustment.

Figure 5-3.     The coordinating mechanisms on a continuum of horizontal decentralization

In other words, as shown in Figure 5-3, the coordinating mecha- nisms form a continuum, with direct supervision the most horizontally centralizing and mutual adjustment the least, and with the three forms of standardization—first work processes, then outputs, finally skills—fall- ing in between. And because standardization of work processes falls next to direct supervision as the second most centralizing coordinating mecha- nism, we conclude that organizations that rely on this mechanism for coordination are relatively centralized. Specifically, such organizations give a certain amount of power to their analysts to design the standards, and as we have just concluded, such power to the analysts means vertical centralization coupled with only limited horizontal decentralization.

But to tie up a loose end, we cannot say that all bureaucracies are centralized. These particular bureaucracies are—the ones that rely on the standardization of work processes to coordinate the work of their unskilled operators. But earlier we came across a second kind of bureaucracy, one with professional operators who coordinate their work by the standardiza- tion of their skills. And because this coordinating mechanism falls near the decentralization end of our Figure 5-3 continuum, we can conclude that this second kind of bureaucracy is relatively decentralized in the horizontal dimension. We shall return to it below.

2. Power to the experts

In this stage of horizontal decentralization, the organization is dependent on specialized knowledge. So it must put its power where its knowledge is—namely, with the experts, whether they be in the technostructure, support staff, operating core, or, for that matter, middle line. “In the world of blind men, the one-eyed man is king.” The surgeons dominate the operating rooms, the Wernher von Brauns rule the space agencies. In the previous discussion, there was only one recognized expert—the analyst— and his power was informal. But here the organization draws on the knowledge of a wider array of experts and begins to formalize more and more of the power it gives to them. The experts do not merely advise; they come to participate actively in making decisions.

How dependent the organization is on its experts and where they are found in its structure determine how much power they can accumulate. We can identify at least three types of expert power.

• Informal expert power superimposed on a traditional authority structure. In the least horizontally decentralized type, the system of formal authority remains intact; that is, formal power remains in the hierarchy of line man- But to the extent that the organization has need of specialized knowledge, notably because certain decisions are highly technical ones, certain experts attain considerable informal power. Thus, the mainte- nance men ruled the tobacco factories Crozier studied because only they could handle the one major source of uncertainty.

These experts made choices. Others gain informal power by virtue of the advice they give managers before choices are made, especially technical choices that the managers do not understand. The authorization step of decision making, often carried out as part of a capital budgeting process, lends itself to the manipulation of managers by experts. The sponsor of a decision or project, that person who first decided to proceed with it, has the expert knowledge of it but also has a strong commitment to see it authorized. The manager above, who must do the authorizing, can be more objective in his assessment of the project, but he lacks the detailed knowledge of it and the time to get it. So the situation is ripe for manipula- tion. In effect, systems of capital budgeting often fail because they cannot put the formal power for authorization where the required knowledge of the project is.

• Expert power merged with formal authority. As expertise becomes in- creasingly important in decision making, the distinction between line and staff—between the formal authority to choose on the one hand and the expertise to advise on the other—becomes increasingly artificial. Eventually, it is done away with altogether, and line managers and staff experts join in task forces and standing committees to share decision- making A good example is the new-product group that brings together marketing, manufacturing, engineering, and research personnel from the technostructure, middle line, and support staff. Power within the group is based not on position but on expertise; each person participates according to the knowledge he can bring to the decision in question. This situation of expert power merged with formal authority amounts, there- fore, to selective decentralization in the horizontal dimension, the experts having power for some decisions but not for others. In fact, reference back to Figure 5-2, where various functional work constellations were overlaid on our logo, suggests a link to selective decentralization in the vertical dimension. In other words, selective decentralization seems to occur con- currently in both the horizontal and vertical dimensions.

• Expert power with the operators. In this third and most decentralizec case of expert power, the operators themselves are the experts. And this expertise vests in them considerable power, which in turn decentralizes the organization in both dimensions: power rests in the operating core, at the bottom of the hierarchy with nonmanagers. Of course, expert opera- tors are professional ones, which leads us to a rather important relation- ship, one that is well supported in the research: the more professional an organization, the more decentralized its structure in both dimensions. This brings the issue of bureaucracy and centralization into sharper focus We can now see the two kinds of bureaucracy emerging clearly, one rela- tively centralized, the other decentralized. The first is bureaucratic by vir- ute of the work standards imposed by its own technostructure. Its operat- ing work is specialized but It is relatively centralized both vertically and horizontally, because most of its decision-making power rests with its senior managers and the small number of analysts who for- malize the behavior of everyone else. In the second, the operating core is staffed with professionals. It is bureaucratic by virtue of the standards imposed on it from the outside, by the professional associations that train its operators and later impose certain rules to govern their behavior. But because the professionals require considerable autonomy in their work, and because coordination is effected primarily by the standardization of skills—a coordinating mechanism shown near the decentralization end of the Figure 5-3 continuum—this second bureaucracy is rather decentralized in both dimensions. That is, power rests with the operators at the bottom of the hierarchy.

3. Power to everyone

The theme of our discussion so far has been that power in the hands of the managers constitutes horizontal centralization; that bureaucratization through the formalization of behavior puts some power into the tech- nostructure and thereby constitutes a limited form of horizontal decentra- ization; and that the more that power is attributed to knowledge as op- posed to position, the more the structure becomes horizontally decentralized, culminating in the professional organization whose opera- tors control much of the decision making.

But, in theory at least, that is not the ultimate case of decentralization Professional organizations may be meritocratic but they are not demo-cratic. As long as knowledge is not uniformly dispersed, so too will power not be evenly distributed. One need only ask the orderlies (or even the nurses) of the hospital about their status vis-a-vis the doctors.

Decentralization is complete when power is based not on position or knowledge, but on membership. Everyone participates equally in decision making. The organization is democratic.

Does such an organization exist? The perfectly democratic organiza- tion would settle all issues by something corresponding to a vote or con- sensus. Managers might be elected to expedite the members’ choices, but they would have no special influence in making them. Everyone would be equal. Certain volunteer organizations—such as Israeli kibbutzim or pri- vate clubs—approach this ideal, but can more conventional organizations? “Industrial democracy” has received considerable attention in Eu-rope recently. In Yugoslavia, workers own many of the enterprises and elect their own managers. In France, there has been much talk of “autoges- tion” (self-management). In Germany, half the seats on the boards of direc- tors of the larger corporations are by law reserved for workers’ representatives.

The evidence from these efforts suggests, however, that these steps do not lead to pure democratization, or anything close to it. Thus, in their excellent review of worker participation in eight countries of Europe, Asia, and the Middle East, Strauss and Rosenstein conclude:

1. Participation in many cases has been introduced from the top down as a symbolic solution to ideological contradictions;
2. Its appeal is due in large part to its apparent consistency with both socialist and human relations theory;
3. In practice it has only spotty success and chiefly in the personnel and welfare rather than in the production areas;
4. Its chief value may be that of providing another forum for the resolution of conflict as well as another means by which manage- ment can induce compliance with its (1970:171)

These reviewers and others suggest that workers are not really in- terested in issues that do not pertain directly to their work. Most surpris- ing, participation has been shown in some studies to strengthen the hand of top management at the expense of other groups, “to bypass middle management, to weaken the staff function, and to inhibit the development of professionalism” (p. 186). Paradoxically, industrial democracy seems to centralize the organization in both the vertical and horizontal dimensions. (A probable reason for this will be discussed in the next chapter.)

Crozier describes another kind of organizational democracy, which seems to have a similar effect. In this case, as noted earlier, the workers institute rules that delimit the power their superiors have over them. That renders the two equal—superior and subordinate are locked into the same straitjacket (except for the maintenance men of the tobacco factories, who exploited that last remaining bit of uncertainty). Power for decision making in turn reverts up to the organization’s headquarters. The resulting struc- ture is, in a sense, doubly bureaucratic—there being the usual rules to coordinate the work as well as special ones to protect the workers. And doubly bureaucratic in this case means, in the same sense, doubly cen- tralized. So what results is a perverse kind of democracy indeed, the orga- nization emerging as more bureaucratic and more centralized than ever, its extreme rigidity rendering it less able to serve its clients or to satisfy the higher-order needs of its workers.

These movements in organizational democracy have barely touched the United States. What has received considerable attention there instead is “participative management.” In discussion of this concept, two of its propositions should be clearly distinguished. One, of a factual—that is, testable—nature, is that participation leads to increased productivity: “In- volve your employees and they will produce more,” management has been told by a generation of industrial psychologists. The other, a value proposi- tion and so not subject to verification, is that participation is a value worthy in and of itself: “In a ‘democratic’ society, workers have the right to partici- pate in the organizations that employ them.” The American debate over participative management has focused almost exclusively on the first, fac- tual proposition (although the proponents seem really to be committed to the second, value position). In the light of this focus, it is interesting that the factual proposition has not held up in much of the research. Studies by Fiedler (1966) and other have indicated that participation is not necessarily correlated with satisfaction or productivity. Those relationships depend on the work situation in question.

In any event, participative management can hardly be called democ- ratization, since it is based on the premises that the line manager has the formal power and that he chooses to share it with his employees. He calls on them for advice and perhaps to share in the making of choices as well. But democracy does not depend on the generosity of those who hold formal power; instead, it distributes that power constitutionally through- out the organization.

So far, we have found little to encourage the proponents of organiza- tional democracy. It may work in volunteer organizations, but attempts to achieve it in more conventional ones seem only to foster more centralization.

Before leaving the subject, we might mention another body of re- search that has shed light on the question. Social psychologists have con- ducted a number of “communication net” studies in which they have put a few subjects (often five) into networks of more or less restricted channels of communication, given them simple tasks to perform, and then watched what happened. In some networks, all the members had to pass their messages through one person (this was the hierarchical one); in others, they formed a circle and could communicate only with members to either side of them; in some, everyone could communicate freely with everyone else (the closest equivalent to democracy); and so on. Many of the results were expected—for example, that the hierarchical networks organized more quickly and made fewer errors, but that their members at the periph- ery enjoyed the task less than did the ones at the center. An unexpected finding, however, at least in one study (Guetzkow and Simon, 1954-55), was that the open-channel networks developed hierarchies by themselves (in 17 of 20 cases).

These findings suggest some interesting conclusions about horizontal decentralization. For one thing, the centralized organization may be more efficient under certain circumstances, particularly at early stages of the work. In contrast, the horizontally decentralized organization—the demo- cratic one—seems better for morale. But the latter may sometimes be un- stable, eventually reverting to a more hierarchical—and centralized—struc- ture to complete its tasks. This, in fact, is exactly what the field studies indicate: that democratization leads, paradoxically, to centralization.

So the answer to our question about democracy seems to be negative. Attempts to make centralized organizations democratic—whether by hav- ing the workers elect the directors, encouraging them to participate in decision making, instituting rules to delimit the power of their managers, or establishing unrestricted communication channels—all seem to lead, one way or another, back to centralization. Note that all the experiments have taken place in organizations that do simple, repetitive, unskilled tasks. A laboratory group cannot be asked to design a thermonuclear reac- tor, let alone deliver a baby. Likewise, organizational democracy has not been a burning issue in research laboratories or hospitals; the attention has been focused on automobile plants, tobacco factories, and the like, organi- zations staffed largely with unskilled operators. Here is where the workers have had the least decision-making power and have been the most alien- ated. And here, unfortunately, is where attempts to tamper with the power system—to make it more democratic—seem to have failed the most dramatically.

Other organizations come closer to the democratic ideal—namely, those with professional operators, such as research laboratories and hospi- tals. They distribute their power widely. But not because anyone decided that participation was a good thing. And not so widely that every member shares power equally. Power follows knowledge in these organizations, which itself is distributed widely but unevenly. Thus, it seems that, at best,; we shall have to settle for meritocracy, not democracy, in our nonvolun- teer organizations, and then only when it is called for by tasks that are professional in nature.

Source: Mintzberg Henry (1992), Structure in Fives: Designing Effective Organizations, Pearson; 1st edition.