Although, as indicated above, there has been extensive discussion in the economic literature about the motivational aspect of the theory of the firm, there has been startlingly little examination of the implicit theory of the capabilities of business firms that is employed as a key building block in orthodox theory.
The orthodox mode of formal representation of what an organiza tion can do res ts on the concept of a production set. The elements of the set are vectors of input and output quantities; to say that a vector is in the production set is to say that it corresponds to a productive transformation that the organization can accomplish. Or, as Debreu put it, “A given production y may be technically possible or techni cally impossible for the jth producer. The set Y of all production pos sible for the jth producer is called the production set” (Debreu, 1959, p. 38) . Depending on the purpose of the inquiry, the fact that pro duction processes take time may or may not receive explicit attention in the formal representation. Also, the basic formalism can, but need not, be elaborated to include detailed representation of the internal structure of the production process-for example, by including intermediate products in the list of commodities and by identifying pro duction II activities” with particular stages in the production process.
The production set idea is very general, but traditionally, at least, the capabilities so described related to production of goods in the everyday sense of that term. A long tradition in economic writing suggests that “production” is the sort of thing that happens either on a farm (corn) or in an establishme nt in the metalworking branch of manufacturing (pins, widgets). In recent years, however, the range of capabilities to which economists have applied the production set idea has increased greatly. While it may be “obvious” that concepts introduced for corn and widget production are readily and appropri ately transferable to furniture storage, haircuts, and vending ma chine services, it does seem that some anxiety might be j ustified con cerning the extendability of the same apparatus to, for example, the services produced by attorneys, educators, psychiatrists, and parents. We shall attempt to articulate this anxiety later on. But for the most part in this volume we adhere to tradition: when we speak of production capabilities, we have manufacturing prominently in mind.
What determines a firm’s production set? Why is it what it is? On the surface, at least, orthodoxy is relatively clear about this. It is a state of knowledge that the production set is supposed to characterize-not, for example, the ultimate limits imposed by physical law, or the limits imposed by the actual conditions of input availability. Arrow and Hahn are quite explicit: “The production possi bility set is a description of the state of the firm’ s knowledge about the possibilities of transforming commodities” (1971, p. 53).
What is the nature of this knowledge? Here the orthodox position is less plain. Considering the weight that this conceptualization of productive knowledge must bear in the overall structure of economic theory, the literature contains surprisingly little discussion intended to motivate and defend the approach . However, the connotation clearly is of knowledge “of a way of doing something” or “tech nological knowledge.” Technological knowledge often is identified with a “book of blueprints” or with the knowledge of engineers and scientists. The latter is at least consistent with the view that specific operational knowledge exists in the context of theoretical under standing, while the “blueprints” metaphor suggests that knowledge is unitized, organized in packages labeled “all you need to know about X.” Implicit in both metaphors, and in other discussions, is the view that technological knowledge is both articulable and articu lated: you can look it up. At least, you could if you had the appropri ate training.
Consistent with the notion of a book with a finite number of blue prints, in some treatments the producti on set is viewed as being gen erated by a finite number of activities or techniques that a firm knows how to operate. In the formal statement of models of this kind, certain assumptions generally are made about the character istics of these individual activities – fixed input coefficients, con stant returns to scale, and independence of other activities. The firm’s production set then is defined as the input-output combina tions achievable with all possible levels and mixes of the activities known to the firm. In o ther treatments economists simply assert cer tain characteristics of the set-for example, that the frontier of the set is described by a Cobb-Douglas production function. From either perspective, one important feature of the production set concept as it is employed is that, using our terms, a producer either has a capabil ity or he does not. He knows how to run an activity or he does not; he has the blueprints or he does not. There are no fuzzy edges to the set, in fact or in mind.
The identification of a firm’s production set with a “state of knowledge” could be interpreted as inviting consideration of a range of further questions. Why is the state of knowledge what it is? How does it change over time? Is it the same for all firms at a given time? For the most part, orthodoxy has declined to examine these issues.
In the standard treatment, the production set is simply taken as given. Issues of its change over time are not considered . The ques tion of whether di fferent firms have different production sets is not treated in a uniform way in orthodox models, but neither is it much discussed . In general, it appears that the most natural assumption within the orthodox framework is that all firms’ production sets are identical-the blueprint file is a matter of public information. To make the sets different is implicitly to postulate positive costs of in formation transfer from firm to firm-a plausible view. But to make them different and immutable, as orthodoxy does when it takes this path, is implicitly to postulate that such costs are indefinitely large -an assumption that is clearly not in the spirit of the usual orthodox treatment of information.
The specialized literature on technical change forms, of course, a major exception to the proposition that production sets are viewed as constant over time. There, the typical model views the technological knowledge underlying the production set as changing over time as a result of “technological progress. /I In turn, technological progress may be viewed as exogenous, or as the consequence of a costly activ ity called “research and development./I In effect R&D expenditure is treated as if it were purchases of an infinitely durable, indivisible fixed input (“knowledge”) whose presence enhances the productiv i ty of other inputs. Such formulations typically assume a total separa bility of R&D from actual production, in the sense that the expansion of the production set could take place even if production itself did not. This, of course, is consistent with the interpretation, noted above, that technological knowledge is articulated knowledge. It is the sort of thing that can be recorded, stored at negligible cost, and referred to when needed . The small group of “learning by doing” models depart from this tnidition, but they remain an unconnected and unexplored annex to orthodox doctrine about production capa bilities.
To the extent that different firms do different R&D and to the ex-tent that there exist secure patent rights, or industrial secrecy, models that assume endogenous technological advance logically ought to admit that firms almost surely will differ in terms of their production sets. Strangely enough, however, virtually no extant model makes such an admission.
Consideration of the production set concept, as it is employed, seems to us to raise three critical questions. If “technological knowl edge” is what defines a firm’s capabilities, where in the firm does that knowledge reside? What rationale can be given for the pres ump tion that there is a sharp boundary line between what a firm can and cannot do? How does the knowledge possessed by one firm relate to that possessed by others, and to the “state of knowledge” in the soci ety generally? We consider these q uestions in turn.
Where does the knowle dge reside ? As we have already noted, two metaphors dominate the meager discussions in orthodox literature that seek to explicate the basic idea of technological knowledge pos sessed by a firm . One is the “symbolic records” metaphor-for ex ample, the notion that the knowledge is stored in a blueprint file. The other is the “knowledge specialist” metaphor- for example, the idea that there is a “chief engineer” to whom the “entrepreneur” looks for a succinct account of the economically relevant aspects of the array of technical possibili ti es. Although both of these metaphors are suggestive of aspects of the real phenomenon of possession of capabilities by a firm, it seems clear that they are merely suggestive and fall far short of being an adequate account of the matter.
Engineering blueprints, and symbolic design records more gener ally, do not contain an exhaustive account of the methods involved in the actual exercise of a productive capability. As a matter of fact, blueprints often are q uite gross descriptions of what to do, and seldom define a detailed job breakdown, much less provide “how to do it” instructions at the job level. As a matter of logical principle, it seems clear that a symbolic record could not provide an exhaustive account of the methods required for its own interpretation; rather, the use of such records presumes the availability of intelligent inter preters drawing on knowledge not contained in the records them selves. And as a matter of economics, cost considerations cle arly limit the extent to which organizations maintain records of their methods and activities, and the records actually maintained are much less complete than they logically might be.
Similarly, the “chief engineer” metaphor is not viable. It seems inescapable that, in the typical and significant cases, the “knowl edge” possessed by a firm is not possessed by any single individual within the firm. In the case of a manufacturing establishment of some size and sophistication, it would certainly be unusual if any single individual knew how to perform each and every task in the entire process. This is true even if the “tasks” involved are produc
tive tasks in a narrow’ sense, and becomes more emphatically so if the tasks include control functions, maintenance, purchasing and mar keting, and so on. Furthermore, the notion of a collection of describ able ” tasks” obviously falls far short of characterizing what the firm as a functioning entity “knows.” What it “knows” includes’ the system of coordinating relations among the tasks-the relations that combine the tasks into a productive performance.
Thus, the possession of technical “knowledge” is an attri bute of the firm as a whole, as an organized entity, and is not reducible to what any single individual knows, or even to any simple aggregation of the various competencies and capabilities of all the various indi viduals, equipment, and installations of the firm. This observation conforms to the accounts in orthodox textbooks, which rarely men tion the “chief engineer” or any other approach to the issues consid ered here. The usual textbook treatment ascribes the ability effec tively to combine inputs to the firm itself, as an actor, and character izes that ability by the production set. But this approach goes im plausibly far: it abstracts the possession of capability entirely from the inputs. It postulates a latent capacity to organize that, being totally disembodied from that which is organized, resides in nothing. It would have us believe that there is such a thing as an au tomobile firm that owns no plant, hires no workers, and produces no automobiles, yet retains the capability to produce automobiles and is ready to do so at the whim of the market. 5 To provide a plausible ac count of the relations between the capabilities of an organization and the capabilities of individual organization members, giving both the “reductionist” and the “holistic” viewpoints their due, is a major conceptual undertaking-and one that orthodoxy has not yet seriously attempted.
What real considerations could p roduce a sharp boundary between “tec hnically possible’! and “technically impossible” p roduction activities ? Certainly, there is no problem with saying that there are some things a firm can do and some it cannot. As an example of the former, we could point to something that th e firm is actually doing, and for an example of the latter we could refer to some hypothetical process whose characteristics violate physical law. However, as we have noted, standard usage of the production set concept contemplates a set of intermediate size, a set including (in most cases) more than what is actually done, and (certainly) less than the full range of the physically possible. The boundary is the boundary of knowledge.
Whatever “knowledge” means in the organizational context, the state of knowledge is certainly subj ect to change. It is subject to change by deliberate choice, as when effort is exerted to discover the answer to a specific question, and it is subject to change by un chosen and unwelcome processes, as when an explosion or break down signals the infeasibility of an attempted course of action. It is subject to increase, as when production workers learn “by doing” to do their jobs more efficiently, and to decrease as workers forget the details of tasks they have not recently performed. It may be increased by means trivially cheap, such as a look at the Yellow Pages, or by ex pensive research and development, as in the design of a new com puter system. It may be expanded by drawing on what others already know, as by reading reports or directly observing o thers’ p:ractice, or there may be an expansion of the limits of what is perceived to be physically possible. An attempt to improve it may be a matter of looking up the answer in a source known to contain the answer, or an extended search for a problem solution that may not exist.
Where, in all of these dimensions, are the discontinuities that could plausibly give rise to production sets with sharp boundaries? The production set approach seems to rest, albeit implicitly, on a claim that such discontinuities exist. Only on that assumption is it legitimate to consider the firm’s position at the “knowledge margin” fixed while exploring the way changing conditions affect its adjust ment at other margins. Only on that assumption does the logic of the firm’s choice among known techniques, on which so much effort has been expended, relate to a real subject matter.
How does the knowledge possessed by one firm relate to that of others, and to the knowledge environment generally? As we have noted, the standard orthodox response to this question is simply to ignore it, and to take each firm’s production set as “given .” This position con stitutes a powerful labor-saving device built into the structure of orthodox theory. In standard competitive models, it leaves market prices as the sole channel of causal influ ence linking the actions of different firms. It thus makes possible the decomposition of the problem of price and output determination into an optimization ex ercise at the firm level, with prices given, followed by an equilibrium analysis at the market level, with firm supply and demand schedules given. To recognize that nonprice information flows among firms are an important phenomenon is to forgo the intellectual economies af forded by this decomposable structure. But it is also to face reality.
The discussion above of the indefinite boundaries of a firm’s knowledge touched briefly on some obvious ways in which firms can augment their own knowledge by reaching out into the environment- into their industry or into society more broadly. In formation about the activities and methods of other firms can be ob-tained by a variety of means-by buying and studying their prod ucts; by hiring away their technically expert employees; by reading accounts of their activities in trade journals, reports of securi ties ana lysts, and their mandatory filings with government agencies; by hiring consultants who work with the other firms of the indus try as well; by reading copies of their patents or the publications of their research scientists; by overt purchase or exchange; or by covert schemes of industrial espionage . None of these methods are so cheap and effective as to make it plausible to assume that anything known to one firm is known to all . None are so expensive or ineffective as to justify an assumption that each individual firm is an island of tech nological knowledge, complete unto itself. And all of these methods are actually used.
Similarly, the firm can reach out through its R&D activity and oth erwise, to the knowledge resources of the society at large. Its research scientists can read the publications of academic and government sci entists, as well as those of other industrial researchers. It can learn from its suppliers and its customers. Performing R&D under govern ment contract may provide an opportunity to learn things useful in its market-oriented activities. Acquisition of or merger with another firm can bring whole packages of capability under unified control. And again, these op tions vary widely in cost and effectiveness, and none are neglected.
Presumably there is no room for dispute concerning the existence of these phenomena, and little room for disputing their importance. Yet in orthodox economic modeling, they are either absent entirely, or, in discussions that admit technological change, treated in an awkward and inhibited fashion. We argued in Chapter 1 that the source of the inhibition is largely to be found in the orthodox com mitments to optimization and equilibrium, but perhaps it derives also from an understandable reluctance to confront the complexities of a dynamically evolving, imperfectly defined state of knowledge that changes in response to the behavior of actors throughout the society. Our own efforts in this direction are set forth in Parts IV and V.
Source: Nelson Richard R., Winter Sidney G. (1985), An Evolutionary Theory of Economic Change, Belknap Press: An Imprint of Harvard University Press.