Innovation in industry and the diffusion of technology
J M Utterback
DOI: https://doi.org/10.1126/science.183.4125.620
IF: 56.9
1974-02-15
Science
Abstract:The varied definitions used in the sources that have been discussed make any aggregate analysis difficult. A simple three-stage analysis of flows to, from, and within the firm was used to facilitate comparisons. Even so, each of the generalizations is drawn from relatively small and unrepresentative samples. Case studies may continue to be a source of ideas and hypotheses for further research, but do not appear to offer a means for deeper understanding of the innovation process. The retrospective nature of nearly all of the sources discussed probably means that the process has been viewed as much more rational and well-ordered than it is in fact. This failing is partially overcome in firsthand accounts such as those of Suites and Bueche (63) and Frey and Goldman (64). Each of these accounts involves a successful innovation according to technical or commercial criteria, or both. However, many of the characteristics of innovations that have failed commercially (10) appear to be similar to those of successful cases. The few longitudinal studies, and studies comparing more and less successful cases, do support the main conclusions drawn above (10, 32, 38). More serious problems are raised by the distinctly nonrepresentative nature of the samples used. There are few cases (17, 33, 65) in which the contributions of more than one organization, or details of interactions over a significant period of time, are discussed. There is a wide variation in the importance of the innovations included, ranging from those affecting the economy as a whole to cases involving production in a single firm, albeit with significant commercial results (66). In addition to questions of comparability and sampling, a central problem for further research on innovation will be to devise an operational model to account for interfirm and interindustry differences. Polar definitions used in past studies, "high technology" and "mature industry," for example, are insufficient. One possibility is to use the strategy for growth or competition evident in a firm or an industry, such as sales maximization (automotive), cost minimization (transportation, communications), performance maximization (aircraft, chemicals), or control of materials resources (mining, petroleum), as a basis for drawing distinctions (67). For example, in an industry that seeks to maximize sales, one would expect innovations that would be highly visible to consumers to be developed rapidly (68). In a cost-minimizing situation, production, as opposed to product technology, would be a major source of uncertainty, while the reverse might be the case in a performance-maximizing situation. Greater uncertainty arising from technical sources would imply greater sophistication in effective firms' product planning approaches, while a more stable technology would imply greater sophistication in market research and market-oriented strategies for innovation, and so forth. Much more work is needed along these lines if outcomes of interventions in the innovative process are to be predicted with any accuracy. Some implications for providing incentives and reducing barriers do seem clear from the work to date. Effective directions for federal action lie in strategies such as creating new markets through purchases or procurement policies; aggregating or focusing markets through regulation and other means; providing for market entry by contracts to smaller firms, venture capital, stronger patent protection, and so on; and providing for mobility and informal contacts within the technical community. Technology "push" strategies (such as tax incentives) to increase most research spending, prizes for new technology, and documentation and information retrieval systems would probably be less important in stimulating innovation. Definitive answers will require the most difficult kind of research-experiments in the field. Since the interventions required are difficult and expensive in most cases, they will not be under the researcher's control. Nor will the effect of policy changes be visible over a short period. Thus it seems imperative to take advantage of interventions that occur fortuitously to construct "quasi-experiments" (69) with as great a degree of control over other factors as possible. For example, have recent changes in policy regarding federally held patents increased the commercial use of these patents? Have changes in the capital gains laws retarded the development and growth of "spin-off" enterprises? Has the identification of technology gaps (3) and competitive opportunities stimulated innovation? The effects of such actions on technical innovation could be carefully observed with a modest but sustained research effort, which promises to yield valuable information beyond that available from largely historical sources.