BUS603: Managing People

The M-form, or multidivisional form, is widely utilized in industry. It is structured by product, customer, or country. Early on, Chandler, in his study of large American corporations, described how the M-form worked and how it was efficient as adopted by General Motors and DuPont. Both corporations were able to coordinate production and customer needs more efficiently and effectively. Later, Williamson presented an economic-reasoned argument in support and formalized the M-form hypothesis:

The organization and operation of the large enterprise along the lines of the M-form favors goal pursuit and least-cost behavior more nearly associated with the neoclassical profit maximization hypothesis than does the U-form (or functional form) organizational alternative.

The M-form is structured around demands in the market, where the U-form is structured around production, sales, finance, human resources, and perhaps other specializations.

In the 80s, Burton and Obel did a series of simulation studies to investigate contingency theory concepts related to the M-form hypothesis with the purpose of developing a more elaborate set of design rules. Using parsimonious models, they conducted an experimental design to test performance consequences of various coordination mechanisms, including the budget versus pricing planning systems and different organizational information systems. Then they investigated the M-form hypothesis (vs. U-form) for different levels of decomposable technologies using the same experimental design. The results of the experiments confirmed the M-form hypothesis: The M-form is generally more efficient than the U-form, but with some conditions. Comparing the M-form and the U-form, the M-form does better for both low and high interdependency tasks, but the coordination mechanism is an important factor. The M-form allocation works very well for less interdependency and an iterative price coordination mechanism, while the U-form works well for a centralized coordination mechanism. Steer and Cable found that the U-form can be more efficient when centralized control is utilized, which is more prevalent in small firms. Size is not the main issue, but control and coordination are. For an organizational design, the choice of the structure and the coordination mechanism are not independent. The M-form design hypothesis is a contingent statement depending upon the task interdependency and the coordination mechanism, not a universal one. The experiment included two that might be alternatives. The M-form did not perform well for a less decomposable task, and the U-form did not perform well for a highly decomposable structure. We seldom observe these designs as natural experiments in the real world, and for good reason as indicated by the experiment. This experiment included both an examination of the M-form hypothesis, but also investigation of what might be design alternatives.

Further, in the M-form, a non-optimal allocation of resources to a division does not yield large opportunity losses if these resources are used well within each division, i.e., the intra-divisional coordination is done well. In the U-form, if the coordination is not done well, e.g., when the sales department and the production department do not use the same quantities, the opportunity losses are quite high. A transfer pricing mechanism can yield large losses if the price is not optimal. Stated differently as implications for organizational design, good coordination is more important than good structure or resource allocation.

In a later laboratory experiment, Burton and Obel included opportunism, which Williamson defines as self-interest seeking with guile. In the experiment, opportunism arises when an M-form division or U-form function manager can request non-optimal quantities from the headquarters that enhance her own divisional or functional performance at the expense of other divisions or functions and the overall firm performance.

For the U-form design alternative, the experiment found incentive schemes based on functional profit to be inefficient, while the U-form worked better for an incentive scheme based on corporate profit. The M-form alternatives were, however, both slightly better than the best U-form alternative, given the particular situation. The discussions about opportunistic behavior and different kinds of hierarchies are an ongoing debate.

What are the implications for the M-form design hypothesis? First, individuals quickly understood that opportunism is possible, and if it was to their personal advantage to report non-optimal quantity to manipulate the prices and thus enhance their own profits at the expense of others and the firm overall. Second, individuals understood in which direction to change the quantities in their own favor. Thirdly, would they behave opportunistically? Not all did; most did, but more cheated in the U-form than in the M-form. In short, an individual will understand the opportunity to cheat; he or she will know how to cheat; and most individuals will cheat, but not all. Some are altruistic. Thus, the M-form design is less compromised than the U-form design when opportunism is invoked.

Under opportunism, the M-form performed much better than the U-form for corporate profits. The opportunity losses in the U-form were more severe than in the M-form - confirming the M-form design hypothesis. Again, coordination losses were more serious than loss due to structure, which confirms that good coordination is more important than good structure.

Putting the various experiments together, the M-form design hypothesis is supported without opportunism, and it is strengthened when opportunism is present. First, the M-form is confirmed parsimoniously using minimal information-processing mechanisms. Second, when the manipulation of information is possible, the M-form hypothesis is strengthened. Resource allocation is less important than coordination under decentralization with opportunism and without opportunism. Experimentation allowed a finer grained set of design rules than the rules only developed out of economic theory and empirical observations.

Design rules are "if-then" heuristics, which guide thoughtful guidance for what might be. Burton and Obel developed design rules based upon the M-form hypotheses and the experiments above:

If the task is nearly decomposable, then the divisional or M-form is superior to the unitary or functional form,

The degree of decomposability is the primary contingency for this rule. A corollary design rule is then:

If the task is not decomposable, then coordination is the main issue and can be realized with a unitary or functional form with a centralized coordination.

A design rule, which emerges directly from the opportunism lab experiment is:

If the organization has a unitary or functional form, then a unit profit scheme should not be used.

In the experiment, this incentive scheme is very likely to yield cheating with very high opportunity losses. What are the risk implications, if coordination is not achieved as desired? The M-form suffers less opportunity loss than the unitary form. That is, less risk is incurred with the M-form than the unitary form.

If the designer is risk averse, then the M-form organization is preferred.

There are many other design rules to state and develop.

A formal rule-based multi-contingency theory of organization was developed based on empirical observations, the information-processing theory, and simulation and laboratory experiments. The multi-contingency theory model was tested on an empirical basis using the survey methodology, and it was shown that misfits in the model led to significant loss in performance. Not only the main effects but also the interaction effects were shown to be important. Further studies then investigated specific interaction effects between leadership, climate, and strategy. Not only are the interactions important, but there are asymmetric effects such that the misfit effects in dynamic situations are much more important than in non-dynamic situations. Examining business-unit adaptation through new-product introductions in the global mobile device industry, Gaba and Joseph find that business-unit responses to poor performance lead to greater new product introductions. Taken together, these results were further expanded into a diamond model of contingency fit.