This paper focuses on the scientific output of small and medium sized firms in high tech industries in the U.S. The indicators for technological output have been defined differently by different researchers. Economists have used patents as the measure of research and development (R&D) output, partly because they are easier to delineate than other indicators. Behavioral researchers interested in the R&D management process have relied on publications as a surrogate measure for R&D output, while research in science policy and innovation has focused on products and processes as indicators of innovation and technical output. There has not been any consensus about the validity of the different measures. It has been argued that the propensity to patent differs from industry to industry. The incentive to patent is dependent on how well patent can be used to protect the interests of the firms. This paper has focused on all three measures as indicators of R&D output and investigated the interrelationship among them as well as the correlates for these output. Except for the chemical industry, R&D expenditure is closely related with size as measured by annual sales, rather than the net income. It is observed that a growing firm is more likely to spend proportionately more money in R&D. The three output measures, patents, papers, and new products, are correlated, but the pattern of relationship is different in different industries. This means that technical behavior of the firms in different industries differs significantly. Correlates of patents, papers, and new products also differ in the different industries. It is interesting to note that a firm's growth is linked with new products in only some industries. Patent does not seem to have any effect on sales growth. The growth in R&D is not a factor leading to a higher level of scientific productivity. The important factor is the average level of R&D spending. The policy implication for this observation is that one ought to maintain a steady funding level of R&D instead of sudden changes in R&D budget. Consistency in R&D is important in determining higher level of scientific output. © 1990 IEEE
