Models for distinguishing innovative and noninnovative small firms

Abstract

The importance of innovation is underscored by the fact that some authors blame U.S. industry's competitive decline on the excessively short-term focus of managers as opposed to longer-term innovative product or process development. At the same time the contribution of small firms to research and innovation continues to rise. Thus from 1945 to 1980 they introduced an average 48% share of innovations, a figure that rose to 56% for the 1980–1985 period. This article studies innovation in small manufacturing firms in Texas. The study sample was drawn from the 1985 Directory of Texas Manufacturers and covered SIC codes 34–39. These include Metal Fabrication, Nonelectrical Machinery, Electrical and Electronic Machinery, Transportation Equipment, Instrumentation, and Miscellaneous Manufacturing. A total of 50 usable responses were received of which the CEO was the responding executive in 38. Prior literature on innovation provided the relevant variables. In broad groupings, these covered strategy, structure, function, environment, the firm, and the responding executive plus his role in innovation. Product differentiation and risk taking were also included. Each broad group often consisted of several variables. For example, the environment grouping comprised dynamism, heterogeneity, and hostility. In all there were 31 independent variables. Their data values were based on either the responses to individual questions or the means of responses to groups of questions. The dependent variable, product-service innovation, averaged the responses to questions on technological leadership and quantity and quality of innovations. A stepwise regression procedure yields an eight-variable model with an R 2 of 0.66. A runs test on the residuals confirms the randomness of errors. The computed discriminant function classifies the sample firms correctly in 43 of 50 cases. While the modeling process is successful, the models have not been retested on fresh data because of the small sample and so the caveat of sample specificity remains. The recurring variables in the two models are integrated decision making, environmental heterogeneity, percentage of research expenditure to cost of goods sold, and the responding executive's role in technical development of innovations. For the latter, the answer seems to be that less is more or “hands-off” is best. The coefficients for the others are all positive. For investors or lenders the consequence of being able to evaluate the innovative potential of small firms is likely to attenuate the risk. Secondarily, it is also likely to enhance the efficient allocation of resources. The data obtained lacked sufficient variation in structural variables like centralization—the sample firms tended to be centralized. Thus future research exploring centralization and its effect on innovation in small firms would be most interesting, particularly as prior results are somewhat ambiguous.