Hostname: page-component-cd9895bd7-gvvz8 Total loading time: 0 Render date: 2024-12-26T08:13:40.531Z Has data issue: false hasContentIssue false
  • English
  • Français

Size of Firm, Oligopoly, and Research: The Evidence*

Published online by Cambridge University Press:  07 November 2014

D. Hamberg
D. Hamberg*
Affiliation:
University of Buffalo
Get access

Extract

For many years the defence of the giant corporation and oligopoly has been rooted in the argument that both were the product of a technology that requires large-scale enterprises for efficient production. In brief, the economies of mass production are supposed to give rise to optimum-scale firms that are too large, relative to the size of markets, to permit the large numbers of sellers needed to minimize monopoly powers over prices and production.

Although still possessing many adherents, this defence of the giant corporation and oligopoly has recently fallen from favour, perhaps mainly for lack of empirical support, but probably also for other reasons that need not be examined here. In its place has come a new defence. The large oligopolistic corporations, we are now told, have become the principal sources of research and development, and thus of the technological progress which is the backbone of economic growth.

Type
Research Article
Information
Canadian Journal of Economics and Political Science/Revue canadienne de economiques et science politique , Volume 30 , Issue 1 , February 1964 , pp. 62 - 75
Copyright
Copyright © Canadian Political Science Association 1956

Access options

Get access to the full version of this content by using one of the access options below. (Log in options will check for institutional or personal access. Content may require purchase if you do not have access.)

Footnotes

*

This paper is part of a larger study of the determinants of industrial research and development. The study is being conducted under a grant from the Ford Foundation. Additional funds have been granted by the University of Buffalo, where I have also been able to utilize the services of the Computer Center. Professor Nanda K. Choudhry has patiently discussed my problem with me and I am grateful.

References

1 Some of the leading references to these various facets of the new defence of the giant corporation and oligopoly are: Schumpeter, J. A., Capitalism, Socialism, and Democracy (New York, 1942)Google Scholar; Galbraith, J. K., American Capitalism: The Concept of Countervailing Power (Boston, 1952)Google Scholar; Kaplan, A. D. H., Big Enterprise in a Competitive System (Washingtan, D.C., 1954)Google Scholar; and most recently, Villard, H. H., “Competition, Oligopoly, and Research,” Journal of Political Economy, Dec., 1958.CrossRefGoogle Scholar

2 The interested reader will find useful discussions of the counter-arguments to the “New Defence” in the following: Jewkes, J., Sawers, D., and Stillerman, R., The Sources of Invention (New York, 1958)Google Scholar; Hennipmann, P., “Monopoly: Impediment or Stimulus to Economic Progress?” in Chamberlin, E. H., ed., Monopoly and Competition and Their Regulation (London, 1954)Google Scholar; Nutter, G. W., “Monopoly, Bigness, and Progress,” Journal of Political Economy, 12, 1956 CrossRefGoogle Scholar; J. Schmookler, “Bigness, Fewness, and Research,” Ibid., Dec., 1959; and Hamberg, D., “Size of Firm, Monopoly, and Economic Growth,” in Employment, Growth and Price Levels, Hearings of the Joint Economic Committee, United States Congress, Part 7, The Effects of Monopolistic and Quasi-Monopolistic Practices (Washington, D.C., 1959).Google Scholar

3 The notable exceptions are the drug, “other chemicals,” and non-ferrous metals industries, where firms in the 1000–4999 size class play a much more important role. See National Science Foundation, Funds for Research and Development in Industry, 1959, Table A-1.

4 Industrial Research Laboratories of the United States, 11th ed. (Washington, D.C., 1960).Google Scholar

5 One firm in primary metals employed 1000 people, two in petroleum and petroleum products employed 800 and 900 people respectively, and one in food and kindred products employed 2000.

6 Based on product-line information contained in Poor's Register of Directors and Executives, 1962 (New York, 1962)Google Scholar, which gives four-digit product classes for each company, and Moody's Industrial Manual, 1960 (New York, 1960).Google Scholar

7 Positive correlation, of course, does not mean causality. Nevertheless, I think it is fair to presume that the direction of causality in this case has run from size to R and D activity, because in most cases it seems clear that the firms were large before they got into R and D, at least in a big way. I would apply the same argument to the results obtained in the analysis of the same relations in the individual industry groups.

8 The statistics reported above may be compared with those reported by Ira Horowitz in “Firm Size and Research Activity,” Southern Economic Journal (Jan., 1962). Horowitz employed data on 700 companies surveyed by the National Association of Manufacturers in 1947 and on 4,800 surveyed by the Harvard Business School in 1951 and 1952. For 18 two-digit industries in the NAM survey and 29 two- and three-digit industries in the Harvard survey, Horowitz ranked industries on the basis of average employment size per establishment and value added per establishment. Then, also ranking the industries on the basis of percent of respondents maintaining research organizations, he obtained rank-order correlations between this measure of “research inclination” and each of the size variables, using the Kendall rank-order correlation method. The Harvard data yielded correlation coefficients of .31 for both size variables, with both results significant at the .05 level. The NAM data yielded coefficients of .51 with employment as the size variable and .55 with value added as the size variable, and both these statistics were significant at the .05 level. Evidently, research inclination is also positively associated with size of firm, but the association is not particularly strong.

9 There is evidence of an increase in the influence of total assets on R and D employment between 1950 and 1960. Compare the results of Table III with those reported by Worley, J. S., “The Changing Direction of Research and Development Employment Among Firms,” in Universities-National Bureau Conference, The Rate and Direction of Inventive Activity: Economic and Social Factors (Princeton, 1962), 245.Google Scholar

10 A numerical example may make this point clearer: if 60 per cent of 20 firms (or 12) spends 1 per cent of sales on R and D and 60 per cent of 30 firms (or 18) also spends 1 per cent of sales on R and D, total industry spending on R and D in both cases will equal 0.6 per cent of sales.

11 See National Science Foundation, Funds for Research and Development in Industry, 1959, Tables A-20 and A-22.

12 But see below, Section III, where statistical analysis indicates a weak relation between R and D and oligopoly.

13 To see whether better results would obtain with the use of an R and D/sales ratio, the least squares correlations were also run for the ratio of R and D employment/sales against total employment and assets. The respective correlation coefficients were .037 and –.024, and neither was significant at the .05 level.

14 Using R and D employment/sales as the measure of research intensity yields little difference in the results. Ranking industries in the same way described in note 8, and using R and D expenditures/sales as the measure of research intensity, Horowitz (“Firm Size and Research Activity”) obtained a rank correlation coefficient of .31 (not significant at the .05 level) with average employment as the industry size variable and .40 with value added per establishment as the size variable (significant at the .05 level), on the basis of the NAM data.

15 Worley, J. S., “Industrial Research and the New Competition,” Journal of Political Economy (04, 1961).CrossRefGoogle Scholar

16 Inadvertently, regressions were run also for the textiles and apparel and the professional and scientific instruments industries; so for what they are worth, we report the estimated values of b with their levels of significance in parentheses. With employment as the size variable, b = .017 (.10) and .515 (.10) in the textiles and instruments industries, respectively In the same order, with assets as the size variable, b = .480 (.30) and .664 (.10).

17 I am indebted to one of my students, John Murphy of Canisius College, for carrying out these correlations—in connection with a paper presented to my seminar in the Economics of Technological Change. The data on absolute R and D expenditures are from National Science Foundation, Funds for Research and Development in Industry, 1959, Table A-8; the data on R and D/sales ratios are from idem., Funds for Research and Development in Industry, 1958, Table A-26. The data on industry concentration are from Concentration Ratios in Manufacturing Industry, Report by the Bureau of Census for the Subcommittee on Antitrust and Monopoly of the Committee on the Judiciary, United States Senate (Washington, D.C., 1962).

Briefly, Murphy obtained his “weighted average concentration indicator” for each industry by weighting the percentage of shipments in each (five-digit, SIC) product class accounted for by the four largest companies and summing the results for each (two- and three-digit) industry group; each sum was then divided by the corresponding sum of weights to obtain the weighted average concentration indicator for each industry group. These were correlated with the R and D data for 20 two- and three-digit manufacturing industry groups. Further details on Murphy's procedure will be available in a forthcoming paper.

18 Ranking industries on the basis of the concentration ratio for sales of the four largest firms in an industry, and correlating these ranks with ones based on the per cent of respondents maintaining research organizations, Horowitz (“Firm Size and Research Activity”) obtained rank correlations of .40 with the Harvard data and .60 with the NAM data, both significant at the .05 level. On the basis of well-known critiques of the (common) approach to the measure of industrial concentration used by Horowitz, critiques which Murphy (see note 17) tried to meet with his measure, I am inclined to place more credence in the latter's results, although both leave too much variance in industry R and D spending unexplained by industrial concentration to attach much importance to the latter variable as a determinant of R and D.

19 With the same measure of industrial concentration as that described in footnote 18, and with R and D/sales as his intensity measure, Horowitz (“Firm Size and Research Activity”) obtained a rank correlation of .38 (significant at .05) with the NAM data.