Organizational Theory:
Determinants of Structure

The objective here is to understand why organizations have the structure that they do. By "structure" I mean things like degree and type of horizontal differentiation, vertical differentiation, mechanisms of coordination and control, formalization, and centralization of power. See handouts page for more information on organizational structure.

According to Taylor, Fayol, Weber and other classical theorists, there is a single best way for organization to be structured. Yet organizations vary considerably on structural attributes. The objective of much research has been to understand what determines these variations. Is it random or systematic? Are some organizations simply less perfect than others, or are different designs better for different situations?

Contingency Theory

In contrast to the classical scholars, most theorists today believe that there is no one best way to organize. What is important is that there be a fit between the organization's structure, its size, its technology, and the requirements of its environment. This perspective is known as "contingency theory" and contrasts with the perspective of classical theorists like Weber, Taylor, Fayol, etc. who thought that there probably was one way to run organizations that was the best.


This refers to capacity, number of personnel, outputs (customers, sales), resources (wealth).

Blau's studies show that differentiation (# of levels, departments, job titles) increases with size, but at a decreasing rate. In contrast, the % of the organization that is involved in administrative overhead declines with size, leading to economies of scale.

Increasing size is also related to increased structuring of organizations activities but decreased concentration of power.

Managerial practices, such as flexibility in personnel assignments, extent of delegation of authority, and emphasis on results rather than procedures, are related to the size of the unit managed.


Consider check processing at a bank. This activity is usually performed by a business unit that is highly formalized, has a great deal of specialization and division of labor, and high centralization of decision-making. In contrast, the creative section of an ad agency is usually not formalized at all, the division of labor is often blurry, and it is highly decentralized.

It appears that certain activities naturally "go with" certain structures. Joan Woodward found that by knowing an organization's primary system of production, you could predict their structure:

Unit production/small batch. Companies that make one-of-a-kind custom products, or small quantities of products (e.g., ship building, aircraft manufacture, furniture maker, tailors, printers of engraved wedding invitation, surgical teams).

Mass production/large batch. Companies that sell huge volumes of identical products (e.g., cars, razor blades, aluminum cans, toasters). Make heavy use of automation and assembly lines. Typically,

Continuous Production. Primarily companies that refine liquids and powders (e.g., chemical companies, oil refineries, bakeries, dairies, distilleries/breweries, electric power plants). Machines do everything,  humans just monitor the machines and plan changes.

Chick Perrow '67 looked at how the frequency and type of exceptions that occurred during production affected structure. Two types of exceptions: (a) can be solved via orderly, analytic search process (like mechanic fixing car), (b) no analytic framework, rely on intuition, guesswork (like advertising, film-making, fusion research).

  Few Exceptions Many Exceptions
Un-analyzable pottery, specialty glass, motel room artwork; plumbing; computer technical support (craftwork)
routine work, but when problems crop up, it is hard
to figure what to do
film making; aerospace; (non routine research)
tasks that no one really knows how to do: work on intuition, implicit knowledge
Analyzable routine, like screws; (routine manufacturing)
the few problems that occur are usually easy to understand
custom machinery, building dams; (engineering production)
the application of well-known principles and technologies to lots of new and different situations

It turns out that bottom left organizations (analyzable and few exceptions) tend to be highly centralized and formalized -- in short, bureaucracies. Bureaucracies are the best possible organizational form when the task is well-understood, and how to best execute it can be specified in advance.

At the other extreme, the top right organizations (unanalyzable and many exceptions) are not well handled by bureaucracies. There are so many exceptions and new situations that having a set of formal procedures which specify how to handle every situation is out of the question. Organizations in this box tend to be highly decentralized and use informal means of coordination and control. The reasons have to do with human bounded rationality. (Bounded rationality refers to the fact that since humans have limited brain capacity, we cannot always find the absolute optimal solution to a given problem -- we only have the time and capacity to consider a few possible solutions, and choose the best among those. But we can't consider all possible solutions.) Really complex systems are difficult to pre-plan: there are too many contingencies. We simply can't figure it all out. Need to allow for real-time, flexible adjustment.



Organizations actively adapt to their environments. For example, organizations facing complex, highly uncertain environments typically differentiate so that each organizational unit is facing a smaller, more certain problem. for example, if Japanese tastes in cars are quite different from American tastes, it is really hard to make a single car that appeals to both markets. It is easier to create two separate business units, one that makes cars for the Japanese market, and the other that makes cars for the US market.

Natural Selection

Organizations whose structures are not fitted to the environment (which includes other organizations, communities, customers, governments, etc.) will not perform well and will fail. Most new organizations fail within the first few years.

If the environment is stable, this selection process will lead to most organizations being well-adapted to the environment, not because they all changed themselves, but because those that were not well-adapted will have died off.


The economy is a giant network of organizations linked by buying and selling relationships. Every company has suppliers (inputs) and customers (outputs). Every company is dependent on both their suppliers and their customers for resources and money. To the extent that a company needs it's suppliers less than they need it, the company has power. That is, power is a function of asymmetric mutual dependence. Dependence is itself a function of the availability of alternative supply. A depends on B to the extent that there are few alternatives to B that are available to A. Dependence is also a function of how much A needs what B has got. If the Post It's company starts to play hardball with you, and there are no good alternatives, it's still not a big deal because Post It's are just not that important.

Organizations that have power over others are able to impose elements of structure on them. For example, GM is famous for imposing accounting systems, cost controls, manufacturing techniques on their suppliers.

The sets of entities in an organization's environment that play a role in the organization's health and performance, or which are affected by the organization, are called stakeholders. Stakeholders have interests in what the organization does, and may or may not have the power to influence the organization to protect their interests. Stakeholders are varied and their interests may coincide on some issues and not others. Therefore you find stakeholders both cooperating with each other in alliances, and competing with each other.

Figure 1. Unconnected stakeholders.

When stakeholders are unconnected to each other (as in Figure 1), the organization usually has an easier time of playing the different parties off one another. For example, it can represent its goals and needs differently to each stakeholder, without fear of being found out. Or, such competitive stakeholders into outbidding each other (e.g., a university can tel one alumnus that another alumnus is about to give a huge donation). Furthermore, when the stakeholders are unconnected, they cannot coordinate their efforts, and so have trouble controlling the organization.

Figure 2. Well-connected stakeholders.

In contrast, when the stakeholders are well-connected (as in Figure 2), the organization cannot represent itself differently to each one, or it will be found out. Furthermore, if the bonds among the stakeholders are closer than the bonds with the organization, the stakeholders may side with each other against the organization, and won't act in ways that negatively affect other stakeholders.


Under conditions of uncertainty, organizations imitate others that appear to be successful. In other words, if nobody really knows what makes a movie successful, and then somebody has a blockbuster hit, everybody else copies the movie, and the organizational structure that produced the movie, hoping that they will get the same results. This can cause whole industries to adopt similar structural features.

One reason why this happens is the fear of litigation or simply blame. If several well-known, successful companies start adopting some new management style -- say, self-governing teams -- and you don't because you know its not appropriate for your company, and then things start to go wrong for your company, people will say 'see? you should have adopted self-governing teams. we told you so'. So to avoid that, if the top companies in a field all adopt some new style, then all the others do to to avoid being blamed.

In addition, diffusion of ideas due to personnel transfer and professional school training can create uniformity as well.

Copyright 1996 Stephen P. Borgatti Revised: October 08, 2001 Go to Home page