Given capabilities and objectives, the orthodox explanation of behavior- what firms do, given constraints-runs in terms of maxi mizing choice. The postulate that firm behavior results from maxi mizing choice leads the theorist to analyze an optimizing decision lule for the firm, a rule that maps from market conditions and other variables external to the firm to the feasible action that scores highest on the fi rm’s objective function. Both of the terms “maximizing” and IIchoice” warrant some scrutiny.
Simple textbook treatments generally presume that the actions taken by firms are truly maximizing in the sense that, given the cir cumstances, there are no better actions . However, we stressed earlier that recent sophisticated versions of the theory back off from that presumption. Lags between decision and effective action are recog nized, along with the possibility that predictions of what the market will be are not perfect: maximization becomes maximization of ex pectation. That all potentially available information may not be fully exploited at decision time also is recognized. Maximization must be understood as recognizing information costs as well as other costs.
It is not clear whether the new most complex models of decision making with limited and costly information are intended to capture, as well, the fact of limited information-processing capacity, or the possibility that firms may be wrong in their understanding of the de cision problems they face. Some economists seem to believe that models of maximizing behavior under limited information do ade quately capture these more general implications of bounded ratio nality.
We think this is a misconception, and a serious one. In orthodox decision theory1 the capacity to process information is invariably treated as costless and unlimited in amount; as Marschak and Radner explain, economic man is a perfect mathematician (Marschak and Radner, 1972, p. 315) . Among other and more consequential implica tions, this says that the actors represented in economic theory already know all the theorems (” mere” logical truths) about their behavior that theorists struggle to prove. This affront to realism is not innocuous. It opens the door to full reliance on the notion of a fUlly preplanned behavior, even in contexts where the level of com plexity involved is such as to overwhelm the aggregate capacity of Earth’s computers. At the same time, it shuts the door on the study of devices that individuals and organizations actually employ to cope with their severe information-processing constraints-devices that often have a key influence on the actions taken. And it suppresses the role of the firm’s own internal organization as a determinant of the effective level of uncertainty to which the firm’s actions are sub ject.
Perhaps the most extensive evidence on this point comes not from the realm of economic activity, but from the history of intelligence failures in international relations. A consistent theme in retrospec tive studies is that failure occurs not because the intelligence system failed to acquire warning signals but b ecause it failed to process, re late, and interpret those signals into a message relevant to available choices.6 Difficult conceptual issues are involved in judging the ex tent to which such failures may be explained by “mistakes,” “dere lictions of duty,” or “irrational behavior. ” But nothing could be more plainly relevant to their explanation than the fact that intel ligence analysts and decision makers have only a limited amount of time each day, limited communication channels to connect their systems, and limited assistance in the task of organizing, analyzing, and thinking about the available information. Sometimes, highly “obvious” and emphatic signals get lost in the noise as a result of these limitations . We see no reason to think that economic decision making is any different in this regard.
There is similarly a fundamental difference between a situation in which a decision maker is uncertain about the state of X and a situa tion in which the decision maker has not given any thought to whether X matters or not, between a situation in which a prethought event judged of low probability occurs and a situation in which something occurs that never has been thought about, between judging an action unlikely to succeed and never thinking about an action. The latter situations in each pair are not adequately modeled in terms of low probabilities. Rather, they are not in the decision maker’s considerations at all. To treat them calls for a theory of atten tion, not a theory that assumes that everthing always is attended to but that some things are given little weight (for objective reasons) .
In short, the most complex models of maximizing choice do not come to grips with the problem of bounded rationality. Only meta phorically can a lllimited information” model be regarded as a model of decision with limited cognitive capacities . It is inadequate in many contexts because it does not explain or predict how a decision maker actually will behave: the metaphor is then nearly devoid of content. In fact, in most formal theorizing, the simple unsophisti cated version of maximization is employed, perhaps augmented by partial recognition of limits on predictive capacities . The firm is vi sualized as truly optimizing its choices, given constraints and uncer tainty.
We now turn our attention to the presumption that behavior is the result of choice. Contemporary appreciative theory is comfortably vague about what “choice” means, and the vagueness signals a problem with the concept. Sometimes “choice” refers to a process in volving deliberation. But sometimes choice is understood to be in volved in the following of a preassigned decision rule without delib eration, the decision rule itself in this usage presumed to be the re sult of ancestral deliberation. And in some of the more careful de-fenses of the theoretical use of optimization assumptions, there even is an admission that the firms may never go through any explicit cal culating deliberation.
It seems useful to distinguish between processes for taking action that do involve a considerable amount of deliberation, and processes that involve more or less mechanical following of a decision rule. One might question whether the latter processes involve much real choosing using the everyday sense of that term. But, more impor tant, if one knew that a certain class of action was the result of indi viduals following a prescribed decision rule, this would seem to be an interesting fact in itself, regardless of the provenance of the rule. Such information might lead the analyst to study, and perhaps model, the decision rule being employed. Indeed, if it is not as sumed that the decision rule is truly a maximizing one, or one that is maximizing within the particular model of the firm being employed by an economist, this would seem the only way to proceed. The ana lyst might go on to analyze why the decision rule is what it is, the analysis involving some theory of decision rule creation and change. And, from this perspective, it would be interesting to go on to ana lyze the adequacy of prevailing decision rules and rule-change pro cesses in terms of how well they enable the firm to cope with the cir cumstances it faces. That is, the decision rules employed by a firm ought to be regarded as an important part of its overall capabilities, in the same sense as the production activities in its production set. In our reading, this is not the perspective that orthodox theory takes regarding, for example, the pricing policies or advertising policies of firms.
As we shall elaborate in the next two chapters, a co nsiderable portion of what is treated as “choice” in traditional theory indeed largely involves following prescribed decision rules. But this is not to deny that in many cases there is a certain-perhaps considerable-amount of deliberation involved. Again, if this is known, it is useful information. It is not useful as evidence in support of a theory that presumes that firms truly maximize something; the difficulty with this theory is the fact that, even if firms explicitly try to maximize, they cannot truly maximize. Rather, it is useful because it calls attention to the processes of deliberation. An analyst aiming to explain or predict action that is known to have come from pro cesses involving considerable deliberation might want to exploit known aspects of deliberation processes in organizations. It is useful to list several of these.
First, deliberative choice reflects a lack of complete preplanning adequate to the state of affairs. One deliberates about a choice be cause one has not thought through in advance what one would do under such circumstances, or, if such predeliberation has gone on, because for some reason the particular context has made the preplan ning incomplete or inadequate for the present purposes . Delibera tion signals problems or opportunities of the present status quo that were at least partly unanticipated.
Second, deliberative choice is co ntingent: its outcome depends on the special circumstances of the situation in which choices are made. In general, it is particular unanticipated problems or opportunities that trigger deliberation, and the deliberation is focused at least ini tially on these. But deliberative choice is likely to be influenced also by a broader set of particular circumstances.
Third, deliberative choice is fragmented. The temporal aspect of its fragmentation has already been noted, but in large organizations it is likely to be fragmented as well along lines of organizational authority and responsibility. A variety of differing information bases and organizational interests impinge on different aspects of the same interrelated decision problem. Commitments to a course of action may be made in one group or set of meetings, while crucial informa tion on the risks or costs of that action resides, untapped, in another. The timing or compatibility of intendedly complementary actions may be deficient because responsibility is divided along functional or input-category lines, and within each such area of responsibility there are competing concerns that pull attention and effort away from the joint task. Warnings of unfavorable developments may suffer delay or distortion in communication to higher authority, because they may seem to reflect adversely on the performance of those charged with responsi bility in the area in which the problems arise. These and similar categories of difficulties are the classic manifesta tions of the fragmentation of choice in large organizations, described by organization theorists from a variety of dis ciplinary perspectives. Theorists of optimal organization have made some progress in mod eling informational fragmentation, less progress in modeling in traorganizational conflict, negligible progress in representing the realities of personal power and reputation-and have done nothing that departs from the basic assumption of the choice monad: the simultaneous confrontation of all constraints.
Finally, the occasions of choice are often opportunities for the clar ification and elaboration of goals. Questions of “what we are trying to accomplish here” often come in for active consideration, not in the mode of logical deduction from premises accepted in the past, but rather in a mode that recognizes the specifics of the choice situ ation as posing issues of general direction, balance, and tradeoff that had not hitherto been confron ted. Since issues of this kind are raised and partially resolved in a sequential, contingent process of choice, there is a sense in which the objectives of an organization are a “path dependent” historical phenomenon . Even if the underlying motiva tional picture is constant and starkly drawn-such as “We are in business to make money”-the delineation of objectives in terms sufficiently precise to inform choice is ordinarily deferred to an actual choice situation.
All of these facets indicate that deliberation is a form of economic activity in its own right, constrained by the scarcity of inputs and by the existing state of the “technology” of deliberation. Although the new sophisticated interpretation of maximizing behavior recognizes information costs, it remains committed to a sharp distinction between having and operating an activi ty or capability, and choosing an action. This fact accounts for some s trikingly paradoxical features of orthodoxy’S perspective on economic organization and economic change. An improvement in information-processi ng tech niques that is linked to a metal-shaping device-for example, a nu merically controlled machine tool- clearly falls under the “tech nological change” rubric and is quite typical of the sort of thing economists have in mind when they seek to measure technological change. By contrast, an information-processing i mprovement that is linked to a deliberative process- such as an econometric model of the firm’s output market, or a linear programming procedure to help decide which factories should ship to which warehouses-is theo retically invisible to orthodoxy because it is part of the choice process. Similarly, orthodoxy seems incapable of recognizing that different firms may have different ways of making choices. These differences in the processes of deliberating ought to be a central part of the explanation of why firms make different choices.
Similarly, there is a process of implementation that follows real choice and is also a form of economic behavior in its own right, shaped by input scarcity and the state of i mplementation technol ogy. For example, the choice of a price policy or pricing rule does not actually suffice to get the proper prices into the catalogs, onto the goods, and into the billing system. Sometimes, implementation costs may constitute a major factor i,n the choice of the price policy itself. The exercise of an organizational capabili ty is involved in imple-menting a newly decided pricing policy for goods, just as much as in producing them. Similarly, specific capabilities are exercised in the actual carrying out of market transactions, in the processes of in ternal control, in record keeping, and so on. That these aspects of business behavior go virtually unnoticed in theoretical economics is certainly not attributable to inhibitions about broadening the scope of the production concept: applications of that concept made in the analysis of health, education, and child rearing testify to the weakness of those inhibitions. Neither, certainly, is it the case that the issues involved are so trivial as to make explicit attention by managers or theorists unnecessary-consider, for example, the com plexities of the problem of preventing embezzlement by computer. Rather, the reason the production-like aspects of implementation re main virtually hidden from orthodox eyes is that implementation, like deliberation, is so intimately related to choice-and choice is simply something done optimally.
The above discussion suggests that ability to deliberate and im plement are elements of a fi rm’s capabilities, just as is its command over a particular technical production process. But if this is so, the sharp separation in orthodox theory between capabilities and choosing becomes suspect. The processes of economic choice, like technical capabilities in a narrower sense, can undergo technological progress or regress. And the questions we have raised about the knowledge that underlies capabilities are as relevant to capabilities for choosing as they are to capabilities for producing. In particular, the proposition that the limits of a firm’s capabilities are not sharply defined is relevant to both. A firm may be uncertain of its judgmen tal and deliberative competence in a given area of activity just as it may be uncertain about its technical competence, and a variety of ways of improving its capabilities are open to it.
Source: Nelson Richard R., Winter Sidney G. (1985), An Evolutionary Theory of Economic Change, Belknap Press: An Imprint of Harvard University Press.