The problems of how business firms ought to makedecisions — as contrasted with how they do — form the basis for an extended,growing, and sophisticated literature. Substantial parts of such analyticaltools as modern statistical decision theory, linear, quadratic, dynamic,heuristic, and integer programing, various kinds of direct search methods,game theory, and theory of teams have been developed in the context of generating normative solutions to problems of the firm. Few of themajor economic decisions of the firm have escaped assault by some newtechnique of operations research or management science. It is natural,therefore, to ask whether positive models of organizational decision makingbased on a behavioral theory of the firm have any implications for a normativetheory.
On the whole, we think an honest answer must emphasize boththe severe limitations on the theory as a normative model and themodest amount of effort that has been directed at considering normative implications. The main positive thing that can be said is that the theoryoffers some glimmers of possible approaches to a few problems withwhich current normative theory has difficulty. Specifically, we willconsider four kinds of problems: (1) finding an analytically useful way ofdescribing one’s own organization, (2) building an operations research model that considers explicitly both the organizational structure and the implementationof “optimal” decision rules, (3) predicting the behaviorof other organizations within the environment, and (4) managing an organization rationallyfrom the point of view of the manager.
Organization description. Traditionally, organizations aredescribed by organization charts. An organization chart specifies theauthority or reportorial structure of the system. Although it is subject tofrequent private jokes, considerable scorn on the part of sophisticatedobservers, and dubious championing by archaic organizational architects, the organization chart communicates some of the most important attributes ofthe system. It usually errs by not reflecting the nuances ofrelationships within the organization; it usually deals poorly with informalcontrol and informal authority, usually underestimates the significance ofpersonality variables in molding the actual system, and usually exaggeratesthe isomorphism between the authority system and the communication system. Nevertheless, the organization chart still provides a lotof information conveniently — partly because the organization usuallyhas come to consider relationships in terms of the dimensions of the chart.
Alternative ways of describing organizations have been suggestedfrom time to time. The most common alternatives focus on dimensions ofthe organization that are poorly reflected by the usual organizationchart, but they retain the basic representation in terms of a set of relationsbetween pairs of individuals. Thus, we conceive the organization as a communication system and describe it in terms of some dimension ofcommunication (channels actually used, channels available, and so forth). Or, weview the system in terms of sociometric choices and describe it as asociometric network. In either case, we use a simple linear graph or a matrix representation of the system.
The kinds of models presented in this book provide anotherpossible alternative descriptive view of an organization. We can describethe organization as a decision- making process. Such a description tendsto be quite different — in terms of the organizational features ithighlights — from an organization chart or the more common alternatives to it. Byopening for examination a different set of attributes, it is likely tostimulate the consideration of a different set of alternative changes in thesystem. Beyond this, a description in terms of the decision process hassomething to recommend it for a primary position in a system having decisionsas a primary output. In either case, however, a description does notnecessarily solve problems. It abstracts from a complicated organization some attributes of the organization. In order to be useful, the persons usingthe description must be able to do something with it.
Supplement to operations research models. Much of modern effortin operations research and management science is directed towarddeveloping decision rules and strategies for making the classic decisionswithin business firms. These decisions — pricing, production, inventory,advertising, investing — overlap considerably with the decisionsconsidered in the previous chapters from a positive point of view. Quite literally,therefore, widespread shifts in the decision rules used by firms would requireus to reconsider the details — if not the basic framework –of our theory.
The converse is also true. Two of the persistent difficulties ingenerating and implementing management science recommendations seem tostem from implicit acceptance of the classical model of an omniscientlyrational firm. First, despite considerable sophistication gained by hardexperience, few operation research workers are fully satisfied with thecriterion functions provided them by the organizations with which they work. It is onething to start a textbook discussion with the qualifier, “Given awell-defined preference order….” It is quite another thing to findan organization that will give you one. Second, management science modelsfrequently experience difficulty in moving from “acceptance” bythe organization to full implementation. Both of these difficulties are typicalmanifestations of an adaptive, multiple-goal system.
Our models do not suggest an obvious solution to the problem ofdealing normatively with such a system. However, they may help inunderstanding the nature of the problem. If our description of the objectives ofan organization is accurate, it is not surprising that operationsresearch efforts occasionally stumble on inconsistent or ambiguous criterionfunctions. It is even less surprising that they frequently encounter problems in implementation. The “goals” of an organization are subjectto the decentralization and attention factors with which we have been concerned. Asa result, it is virtually impossible to specify them a priori.It is also virtually impossible for a “responsible executive” to specifythem accurately. Recent efforts to determine criterion functions are a response toone form of this difficulty. They attempt to avoid the intrapersonalproblems of defining a utility function. To be more successful in anorganization, however, the management scientist must expose a wider part ofthe organizational system to the implications of the decisions. The”responsible executive” can at best predict the situations in which hisorganization (or a part of it) will protest. Even the best executive makes errors inprediction — especially if he is accustomed to keeping the organizationviable by responding to feedback rather than relying exclusively onpredictions of trouble.
If our analysis is correct, a behavioral theory of the firm hasimplications for operations research models at two different levels ofgenerality. First, a model of a specific decision-making process could form a basisfor identifying organizational constraints on a decision rule. Thus, it wouldbe one basic device for defining the precise problem facing theorganization. Second, the theory seems to indicate that efforts to improve anorganization as an adaptive system might be more relevant than efforts togenerate some kind of optimizing decision rule. In the final accounting, it willprobably be the first of these implications that will prove the moresignificant. Effective normative treatment of organizational decision making isparticularly dependent on a precise specification of the institutional andbehavioral limits on organizational design. In the short run, however, thetheory may also be useful as a guide to developing adaptive procedures usingthe organization’s limited capacities more effectively.
Predicting the behavior of other organizations. Business organizations exist in an environment consisting largely of otherorganizations. Effective management requires knowledge of the probable behaviorof suppliers, customers, competitors, governmental regulatory bodies,labor unions, and a host of other complex organizationaldecision-making systems. As in every case in this section, we have no directevidence that the models are useful for such a purpose. It should be obvious,however, that if we can predict the ordering behavior of a retail departmentstore, we should be able to provide some useful information to amanufacturer selling to such a store. If we can understand the search behaviorof the department store, we can adopt a sales strategy that will increasethe likelihood that our products will be visible to the store when itis searching. Similarly, if we can predict the price, output, and sales strategyreactions of a class of firms to certain kinds of situations, we should be able to provide some help to a firm operating in an industry composed of firms ofthat class.
Use of such models will scarcely eliminate the use of other moreclassic devices of tactical intelligence — spying, trial balloons,analysis of public information, and direct observation. Classical intelligence devicesare (as we have noted) means for avoiding uncertainty by usingshort-run feedback. An organization will continue to need such feedback evenif the models are much more successful than we as yet have any reason tohope they will be. An organization will continue to use such feedbackeven if it does not need it. For special purposes and in special situations,however, it should be possible to provide some additional aids to decisionmaking involving other organizations by developing models of those other organizations and testing alternative strategies against the models.
In order for behavioral models of organizations to be useful for predicting organizations within the environment, we require a knowledge ofthe internal operation of those organizations. Such knowledge may bedifficult to obtain; few competitors will knowingly provide detailedinformation for such a purpose. Effective normative use, therefore, will depend onthe extent to which the development of the theory will limit the numberof organization-specific characteristics that have to be determinedand on the extent to which the standard devices for gathering informationabout other organizations can be used to estimate the key parameters. Atpresent, the theory is not well enough developed to offer much immediate hopeexcept in a few areas.
The rational manager. We have argued that the business firmis basically a coalition without a generally shared, consistent set ofgoals. Consequently, we cannot assume that a rational manager can treatthe organization as a simple instrument in his dealings with theexternal world. Just as he needs to predict and attempt to manipulate the “external” environment, he must predict and attempt to manipulate his ownfirm. Indeed, our impression is that most actual managers devote muchmore time and energy to the problems of managing their coalition thanthey do to the problems of dealing with the outside world.”Optimal” allocation to a particular subunit cannot be determined independent of theeffect of an allocation on the goals (and thus the demands) of subunits inthe organization and on the development of organizational slack. Thesemore or less elementary lessons in the administrative facts of life havenot as yet been adequately considered by management science. Since O. E. Williamson has pursued (in Chapter 9) some of the possibleimplications of using the present theory as a guide to a rational manager, we will notattempt to suggest any further detail.
Source: Skyttner Lars (2006), General Systems Theory: Problems, Perspectives, Practice, Wspc, 2nd Edition.