1. The design parameters
Most important, the Divisionalized Form relies on the market basis for grouping units at the top of the middle line. Divisions are created accord- ing to markets served and are then given control over the operating func- tions required to serve these markets. Thus, in Figure 11-1, a typical organigram for a divisionalized manufacturing firm, each division contains its own purchasing, engineering, manufacturing, and marketing activities. This dispersal (and duplicatibn) of the operating functions minimizes the interdependence between divisions, so that each can operate as a quasi- autonomous entity, free of the need to coordinate with the others. This, in turn, allows a large number of divisions to be grouped under the head- quarters—in other words, the span of control at the strategic apex of the Divisionalized Form can be rather wide.
Figure 11-1. Typical organigram for a divisionalized man- ufacturing firm
This structural arrangement naturally leads to pronounced decentral- ization from the headquarters: each division is delegated the powers needed to make the decisions concerning its own operations. But the de- centralization called for in the Divisionalized Form is highly circum- scribed—not necessarily more than the delegation from the few managers at headquarters to the few more managers who run the divisions. In other words, the Divisionalized Form calls for decentralization of the parallel, limited vertical variety. In fact, divisionalized structures can turn out to be rather centralized in nature. The division managers can hold the lion’s share of the power, precluding further vertical decentralization (down the chain of authority) or horizontal decentralization (to staff specialists and opera- tors). As the president of one conglomerate firm—an organization that inevitably uses the Divisionalized Form—commented:
Our whole philosophy revolves around where profit responsibility is placed—the divisional general manager. I don’t want anyone in this organi- zation to have any doubts that the general manager is boss. This is where the entrepreneurial atmosphere begins, (quoted in Lorsch and Allen, 1973:55)
Of course, in theory, divisionalization does not preclude the further de- centralization of power within the divisions. But as we shall soon see, other characteristics of this structure drive the divisions to centralize more power than they would if they were independent organizations.
Were the headquarters to delegate all its power to the division manag- ers, it would cease to exist, and each division would, in effect, emerge as an independent organization. So some form of control or coordination is required between headquarters and the divisions. The question then be- comes, How can the headquarters maintain control while allowing each division sufficient autonomy to manage its own operations? And the an- swer lies in one specific design parameter: the performance control system. In general, the headquarters allows the divisions close to full autonomy to make their own decisions, and then monitors the results of these deci- sions. This monitoring is done after the fact, in specific quantitative terms—in the case of the business corporations, by measures of profit/ sales growth, and return on investment. As Ackerman notes, “Accounting reports are not immune to misinterpretation but they relieve the reviewer of the need to sift through and comprehend operating data from diverse businesses” (1975:49). So the prime coordinating mechanism in the Divi- sionalized Form is the standardization of outputs, and a key design pa- rameter is the performance control system.
This coordinating mechanism and the three design parameters so far discussed determine the basic structure: market-based units at the top of the middle line; parallel, vertical decentralization to those units (but not necessarily within them); and reliance on standardization of the outputs of these units through the use of performance control systems to effect head- quarters’ control of the divisions. These form an ideal configuration. The market basis for grouping allows for autonomy of the divisions, which encourages decentralization to them and also allows for easy identification of their outputs, which can then be coordinated through performance con- trol systems.
But other coordinating mechanisms and design parameters also have roles to play in this configuration, although they are not the primary ones.
The standardization of work processes is not used by headquarters as a coordinating mechanism because that would interfere with divisional autonomy. So little of the division’s behavior is formalized by headquar- ters. Likewise, action planning is avoided because that, too, would impose decisions on the divisions that they need to make themselves. Mutual adjustment between the divisions, as well as the liaison devices that en- courage it, are also precluded in this structure by the absence of interde- pendence between the divisions.
There is, however, a limited role for the two coordinating mecha- nisms that remain—standardization of skills and direct supervision. The Divisionalized Form is dependent for its success on the competence of the divisional managers, to whom much of the decision-making power is dele- gated. Whereas the managers at the top of the middle line of the other configurations tend to have functional orientations and limited freedom to act independently, those of the Divisionalized Form are “mini-general managers,” who run their own operations. That is why the middle line emerges as the key part of this structure: But this characteristic puts the onus on the headquarters to train these division managers as well as it can (in effect, to standardize their managerial skills). Likewise, indoctrination is used to ensure that the division managers pursue the broader goals of the headquarters instead of the parochial goals of their divisions. Division- al managers are brought back to headquarters periodically for conferences and meetings with the central administrators, and they are sometimes rotated around the different divisions to develop a broad perspective of the organization. Direct supervision serves as a backup mechanism of coordi- nation in the Divisionalized Form. When a division runs into trouble, the headquarters managers may have to step in, perhaps to replace the divi- sion manager. So some knowledge of the operations of the division is required, at least to know when to step in, as well as how. This need for direct supervision reduces the span of control of headquarters managers somewhat.
2. The structure of the divisions
Given an understanding of the means of control of the divisions by head- quarters—through performance controls backed up by management train- ing, indoctrination, and direct supervision—we can return to the question of decentralization within the divisions. In theory, the Divisionalized Form can be superimposed on any of the other configurations. A multiversity or a national accounting firm with regional offices draws a set of Professional Bureaucracies into the Divisionalized Form; a newspaper chain does the same thing with a set of Adhocracies. And a venture capitalist with equity control of entrepreneurial firms may draw a set of Simple Structures into the Divisionalized Form. The divisions of any one organization may also exhibit a variety of structures, as, say, in the case of a municipal govern- ment with four “divisions”—a small Simple Structure antipoverty pro- gram, a Machine Bureaucracy sanitation service, a Professional Bureau- cracy police force, and an Adhocracy urban development group.
But the Divisionalized Form works best with Machine Bureaucracy structures in its divisions and, moreover, drives these structures, no mat- ter what their natural inclinations, toward the Machine Bureaucracy form. The explanation of this important point lies in the standardization of out- puts, the key to the functioning of the divisionalized structure. The only way that headquarters can retain control yet protect divisional autonomy is by after-the-fact monitoring of divisional performance. That requires the establishment of clearly defined performance standards, the existence of which depends on two major assumptions. First, each division must be treated as a single integrated system with a single, consistent set of goals. In other words, although the divisions may be loosely coupled with each other, the assumption is that each is tightly coupled within.1 Second, those goals must be operational ones—in other words, lend themselves to quan- titative measures of performance control. In the organic configurations— Simple Structure and Adhocracy, which exist in dynamic environments— such performance standards are difficult to establish. In the Professional Bureaucracy, as noted in the last chapter, the complexity of the work precludes the establishment of such standards. Moreover, the Professional Bureaucracy is not one integrated system but a collection of individuals with a wide range of goals. That leaves only one configuration that satisfies the assumptions: the Machine Bureaucracy. In other words, the Division-alized Form is best superimposed on the Machine Bureaucracy, the only structure that is integrated and has operational goals.
Now, what happens when the Divisionalized Form is superimposed on one of the other three configurations? To make that form work, the assumptions must be made to hold. That is, each division must be made to function as a single integrated system, on which one set of performance measures can be imposed. The division manager, to whom power is dele- gated from the headquarters, must be able to impose the measures on his division; in other words, he must treat it as a top-down, regulated system. For the Professional Bureaucracy and Adhocracy—in large part bottom-up and nonregulated—that amounts to a pressure to centralize. Moreover, when the division is organized on a functional basis—as it typically is in the Simple Structure, Machine Bureaucracy, and Adhocracy—the division manager is forced to use an action-planning system to ensure that division personnel pursue the performance goals. Action planning imposes ever more specific standards concerning decisions and actions on personnel down the line. That amounts to pressure to formalize (and bureaucratize) the structure of the division, especially the Simple Structure and Ad- hocracy, which are organic to begin with. So the Divisionalized Form drives the divisions to be more centralized and more formalized than they would be as independent organizations. (That is, of course, the effect predicted from Hypothesis 14, since the headquarters is a specific form of external control of the division.) And these are the two distinguishing characteristics of the Machine Bureaucracy. So we conclude that divisional- ization drives the structure of the divisions, no matter what their natural inclinations, toward the Machine Bureaucracy form. The performance con- trol system of the Divisionalized Form weakens the organic nature of the Simple Structure and the Adhocracy, and it upsets the notion of operator autonomy in the Professional Bureaucracy.2 Only in the Machine Bureau- cracy does divisionalization require no fundamental chance in structure.
Why, then, is “divisionalization” treated in so much of the literature as synonymous with “decentralization” (and implicitly with debureaucra- tization)? The answer seems to lie in the origins of the configuration. As certain Machine Bureaucracy corporations inAmerica grew and diversified their markets early in this century, they became increasingly unwieldly— too centralized and too bureaucratic. The development of the Division- alized Form—in du Pont in 1921—came as a godsend. Instead of one inte- grated functional structure, a set of them could be designed, one for each market. This eased the bottleneck at the strategic apex, allowing for less centralization and less formalization. So, compared with the Machine Bu- reaucracy structure—that is, with one overall Machine Bureaucracy for all markets—the Divisionalized Form, by creating many smaller and more focused Machine Bureaucracies, reduced the overall centralization of the structure.
But is the Divisionalized Form inherently decentralized, or, to be specific, more decentralized than the other configurations? Than the Sim- ple Structure, with all power concentrated in a single office, certainly. Than the Machine Bureaucracy (operating in one market, as it is designed to do), not clear. Who is to say which structure distributes its power more wide- ly—the one with limited horizontal decentralization, where the few analysts of the technostructure share power with the managers of the strategic apex, or the one with limited vertical decentralization, where the few man- agers at the top of the middle line share that power? And than the Profes- sional Bureaucracy or Adhocracy, with their extensive decentralization deep into the line structure and out to a large number of operating or staff specialists, certainly not.
Moreover, there is another, more logical alternative to the Division- alized Form: complete fragmentation of the organization. And that is also more decentralized. It is a rather small step from quasi-autonomous divi- sions controlled by one central headquarters to fully autonomous organiza- tions, each controlled by its own board of directors. In fact, the Division- alized Form often emerges not from the “decentralization” of a Machine Bureaucracy operating in many markets, but from the “centralization” of a set of independent organizations operating in different markets. They con- solidate themselves into a single “federation” with a Divisionalized Form configuration, in the process surrendering some of their powers to a new central headquarters.
Ironically, this is what happened in the most famous example of divisionalization, the one most frequently touted as “decentralization”— Alfred P. Sloan’s restructuring of General Motors in the 1920s. It was this example that set off the first waves of divisionalization among the Fortune 500. Yet no example better illustrates the fallacy of the “divisionalization means decentralization” relationship. For although Sloan may have divi- sionalized General Motors, by no stretch of the imagination did he de- centralize it. As a well-known student of his actions commented, “If any one word is needed to describe the management structure of General Motors as it was recast by Sloan and the brilliant group around him, then that word is not decentralization, but centralization” (Harold Wolff, quoted in Perrow, 1974:38). As Chandler (1962) and even Sloan (1963) himself tell it, William C. Durant put General Motors together as a holding company, but failed to consolidate it into a single entity. Sloan was brought in to do that job. He instituted central controls, which reduced the power of the unit managers by subjecting their performance to headquarters control. In other words, Sloan consolidated the structure to the Divisionalized Form, and thereby centralized it. (Later in this chapter we shall see that this process of centralizing power in General Motors apparently continued throughout this century to the point where the current structure of the automotive component of the company can no longer be called divisionalized.)
3. The powers of the divisions and the headquarters
Both communication and decision flows in the Divisionalized Form reflect one central fact: There is a sharp division of labor between the headquar- ters and the divisions. Communication between the two is circumscribed and largely formal, in good part restricted to the transmission of perfor- mance standards down to the divisions and of performance results back up. This is supplemented by personal interchanges between the manag- ers at the two levels, but that is carefully limited. Too much detailed knowledge at the headquarters level can invite meddling in the decisions of the divisions, thereby defeating the very purpose of divisionaliza- tion—namely, divisional autonomy.
In the Divisionalized Form, the divisions are given the power to run their own businesses. They control the operations and determine the strategies for the markets that fall under their responsibility. What powers then are retained by the headquarters? We shall discuss six in all. The first is the formation of the organization’s overall product-market strat- egy. Whereas the divisions determine the strategies for given product mar- kets, the headquarters decides which ones will be given. In effect, the headquarters manages the strategic portfolio, establishing, acquiring, sell- ing, and closing down divisions in order to change its mix of products and markets. This, in fact, is one of the main reasons for using the Division- alized Form and, according to studies of du Pont in the 1920s, explains why it evolved in the first place:
Unencumbered by operating duties, the senior executives at the general of- fice now had the time, information, and more of a psychological commitment to carry on the entrepreneurial activities and make the strategic decisions necessary to keep the over-all enterprise alive and growing and to coordinate, appraise, and plan for the work of the divisions. (Chandler, 1962:111)
Second, headquarters allocates the overall financial resources. Only pooled coupling exists among the divisions. That is, they do not pass their work back and forth but do share common financial resources. It is clearly the responsibility of the headquarters to manage these resources—to draw excess funds from the divisions that do not need them, to raise additional funds in the capital markets when necessary, and to allocate available funds among the divisions that do need them. Headquarters’ power over resource allocation also includes the authorization of those divisional capi- tal projects large enough to affect the overall capital budget of the organiza- tion. The need to seek such authorization may constitute some interference with the autonomy of the divisions, but that is an interference necessary to ensure the balanced allocation of funds. In general, however, the assess- ment by headquarters of divisional capital projects is purely financial in nature—concerned only with questions of risk and availability of funds, not with those of product-market strategy.
The key to the control of the divisions in this configuration is the performance control system. Hence, as its third major power, the head- quarters designs the performance control system. The managers there, with the aid of their own technostructure, set up the system. They decide on performance measures and reporting periods, establish formats for plans and budgets, and design an MIS to feed performance results back to headquarters. They then operate the system, setting targets for each re- porting period, perhaps jointly with the divisional managers, and review- ing the MIS results.
What happens when the MIS signals that a division has run into trouble, that it can no longer meet its performance targets? The manage- ment at headquarters must first decide whether the problem lies in condi- tions beyond the control of the division or in it. If the former—the problem being an economic downturn, the arrival of new competition, or what- ever—headquarters basically has the choice of divesting itself of the divi- sion or carrying it financially to ride out the trouble. In other words, it acts in terms of one of its first two powers, the management of the strategic portfolio or the allocation of financial resources. But if the problem is per- ceived to lie in the division, then headquarters draws on its fourth major power. The headquarters replaces and appoints the managers of the divi- sions. This is a crucial power in the Divisionalized Form, because the structure precludes direct interference by the headquarters managers in the operating affairs of the divisions; the closest they can come is to deter- mine who will run the divisions. To an important extent, therefore, success in the Divisionalized Form depends on this fourth power, on selecting the right people—general managers with the ability to run quasi-autonomous operations effectively, yet in accordance with the goals of the overall organization.
The performance control system may signal a problem in a division, but it is of little help in determining whether that problem is rooted in adverse conditions or incompetent management. Moreover, there are times when the performance control system fails to do a proper job of reporting problems. Being dependent on- quantitative historical data, the MIS sometimes misses the nuances that signal imminent problems. The MIS can also be manipulated by the divisional management, as when an advertising or research budget is cut to show better short-term profit at the expense of long-run profitability. So, although the headquarters depends on the MIS to monitor divisional behavior, it cannot rely exclusively on that system. This leads to the fifth function. The headquarters monitors divi- sional behavior on a personal basis. Here coordination reverts partly back to direct supervision as a supplement to the standardization of outputs. Headquarters managers—sometimes called “group executives” and given charge of a number of divisions—visit the divisions periodically to “keep in touch,” to get to know them well enough to be able to foresee problems. Such knowledge also enables the headquarters managers to assess re- quests by divisions for large capital expenditures, and it gives them knowl- edge of the people in the divisions when replacements must be made.
But, as noted earlier, too much direct supervision defeats the purpose of the Divisionalized Form—the provision of autonomy to the units in the middle line. So in normal times, the headquarters managers stand on a tightrope between being ignorant of division problems and becoming so familiar with them that they are tempted to interfere in their solution. Some divisionalized organizations try to achieve the right balance by re- stricting the size of the headquarters. In Textron in 1970, for example, with sales of more than $1.5 billion from thirty different divisions, the headquar- ters staff numbered only thirty executives and administrators, and the group vice-presidents had no assistants or private technocratic staff, just one secretary each.
As its sixth and final power, the headquarters provides certain sup- port services common to the divisions. The location of support services— their concentration in headquarters or dispersal to the divisions—is a major design issue for the Divisionalized Form. Services that must be geared to the needs of single divisions, those that must be located in physically convenient places, and those that are relatively easy to duplicate—as in the cases of a marketing research group, a cafeteria, and a public relations unit, respectively—are typically dispersed to the divisions (and are sometimes duplicated at headquarters as well). But coordinated services that must be offered across the range of divisions, or those that must be provided at the common strategic apex, are concentrated in single units at headquarters. Thus, a central finance unit supports the headquarters role of resource allocation; looks after income tax, insurance, pension matters, and the like common to the different divisions; and may also house the technocratic staffers concerned with the performance control system. Again, however, any organization that wishes to be divisionalized must severely limit the number of support services it provides at headquarters. Each one imposes decisions on the divisions, thereby curtailing their autonomy.
Figure 11-2. The Divisionalized Form
To conclude our discussion of the basic structure, Figure 11-2 shows the Divisionalized Form represented symbolically in terms of our logo. Headquarters is shown in three parts: a small strategic apex of top manag- ers; a small technostructure to the left, concerned with the design and operation of the performance control system as well as some of the man- agement-development programs; and a slightly larger staff support group to the right. Four divisions are shown below the headquarters, with a bulge put in at the level of division manager to indicate that the middle line is the key part of the organization. All four divisions are represented as Machine Bureaucracies to illustrate our point that divisionalization encour- ages the divisions to use this configuration.
Source: Mintzberg Henry (1992), Structure in Fives: Designing Effective Organizations, Pearson; 1st edition.