Search strategies and topographies: Elements of an Enriched Search Model

Our proposal for a more general model of search has several major elements. The first is a set of not yet discovered or invented tech­ nologies. Any technology can be described under two different headings. One involves economic parameters, such as input coeffi­ cients or certain product attributes. If these are known, then for any given set of product demand and factor supply conditions, one can d irectly calculate the economic meri t of the technology-unit pro­ duction costs (at various levels of output) or the price at which the product can profitably be sold . A technology also can be described in certain “technological” dimensions, such as size, chemical composi­ tion, or thermodynamic cycle employed. While these are not of eco­ nomic interest in themselves, knowledge of them may be very im­ portant in R&D decision making.

The economic attributes of members of the set are noC in general, known to the R&D decision maker. What he does know includes some of the technological att ributes of the technologies (these may provide the “name” or the description of the technological alterna­ tive in question), and also some general stochastic relationships between technological attributes and economic attributes . Thus, it is known that a plane with a higher-pressure and higher- temperature engine will fly faster and that this offers certain advantages to a po­ tential purchaser of the plane, but the engine will cost more to pro­ duce. It is known, too, that certain classes of chemicals are much more likely to include effective pain-killers than o ther classes of chemicals. However, these relations are not known sufficiently well so that economic attributes can be perfectly predicted from tech­ nological ones.

There is a set of activities that may be used for fi nding out more about the technological and economic attributes of a technology. This “finding out” can be considered as “doing research, ” “testing,” or IImaking a study.” There is a related but different set of activities that is involved in working out the details and developing a technol­ ogy so that it can be employed in practice. The decision maker can predict, to some degree but not perfectly, the outcomes of con­ ducting these various activities at various leve ls of input utilization . The R&D decision maker is viewed as having a set of decision rules that guide the employment of the above activities; these rules determine the direction of “search,” in the general sense in which we are using the term, and may be termed a “search strategy. ” A strategy may be keyed to such variables as the size of the firm, its profitability, what competitors are doing, assessment of the payoff of R&D in general and of particular classes of proj ects in particular, evaluation of the ease or difficulty of achieving certain kinds of tech­ nological advances, and the particular complex of skills and experi­ence that the firm possesses.

The outcome of the actions taken by any firm can be described sto­ chastically in terms of two variables. One of these corresponds to in­ vention; probabilis tically there will be certain previously undiscov­ ered or uninvented technologies that become known and certain previously undeveloped technologies that have been developed suf­ ficiently to permit their implementation. But there will be, as well, a change in the knowledge possessed by the firm that in general will involve information much more broadly useful than merely the knowledge about the particular new technology. Something will be learned about a class or “neighborhood” of technologies (not merely the particular one developed) that may involve the technological or economic attributes associated with that class. In principle, the dis­ tributions of these stochastic outcomes can be deduced if one knows the other three components of the model: the topography, the search activities, and the decision rules. In practice, unless the model is kept quite simple, deductions may be difficult or impossible.

The search model of Chapter 9 can be regarded as a very simple truncated special case of the above broader model, involving regular topography, very simple search procedures, naIve decision rules, and no explicit advance of knowledge except for the new technologies found. The “neighborhood” search characterization was a simple formalization of the idea that knowledge at a given technology is par­ tially transferable to related ones. Many of the contempora ry neoclas­ sical models of invention also can be reconciled with this framework. We diverge from the neoclassical formulation in rejecting the suppo­ sition that the decision rules employed by the firm can be literally optimal, and in placing emphasis on the uncertain, stochastic nature of the processes involved. As usual, the problem with the neoclas­ sical metaphor here is not that it connotes purpose and intelligence, but that it also connotes sharp and objective definition of the range of alternatives  confronted  and  knowledge  about  their  properties. Hence, it misleadingly suggests an inevitability and correctness in the decisions made, represses the fact that interpersonal and in­ terorganizational differences in judgment and perception matter a lot, and ignores the fact that it is not at all clear ex ante what is the right thing to do.

In the following subsections we shall employ this broad frame­ work to consider certain common decision rules, strategies, and paths taken by technological advance that have been revealed by empirical research . These “facts” about i nvention or R&D will be given an interpretation within the model.

Source: Nelson Richard R., Winter Sidney G. (1985), An Evolutionary Theory of Economic Change, Belknap Press: An Imprint of Harvard University Press.

Leave a Reply

Your email address will not be published. Required fields are marked *