We use cookies to distinguish you from other users and to provide you with a better experience on our websites. Close this message to accept cookies or find out how to manage your cookie settings.
Published online by Cambridge University Press: 07 November 2014
The proportion of an industry's output concentrated in a few large firms is recognized by economists as an important characteristic of industrial structure. Economic analysis agrees with the layman's judgment that high concentration is a major factor making for monopolistic practices, and there is ample confirmation in the record of antitrust investigations.
The degree of concentration is of course not by any means the only determinant of market behaviour. In the absence of collusion or government regulation, however, high concentration is a necessary condition for monopolistic practices, while low concentration is a necessary condition for the type of competition that tends to equate price and marginal costs. Moreover, with low concentration informal collusion is unlikely and the organization of formal collusion (for example, through a trade association) becomes difficult. Government regulation is frequently the result of political pressures the strength of which reflects either high concentration or close association of the business interests involved.
It is therefore important to investigate the degree of concentration and its causes. This paper presents a comparison of concentration in the manufacturing industries of Canada and the United States. We find, first, that most industries are more concentrated in Canada than in the United States. This can be ascribed to the fact that industries are typically much smaller in Canada, while firms are of about the same size as in the United States. Secondly, we find that industries with relatively high concentration in one country also tend to have relatively high concentration in the other, though the correlation coefficient is only about 0.7. It appears that cultural and technical similarities make for similar industry-size patterns in the two countries, while the technical similarities mean, further, that industries with a relatively large average size of firm in one country also have relatively large firms in the other.
This paper is based on a portion of the research undertaken by the author as a fellow of the Social Science Research Council and continued as a research associate of the National Bureau of Economic Research. The assistance of these institutions is gratefully acknowledged.
1 See for example: Edwards, C., Maintaining Competition (New York, 1949), 91–5.Google Scholar Stocking, G. and Watkins, M., Monopoly and Free Enterprise (New York, 1951), 87–91, 110–12, 292–6, and passim.Google Scholar
2 Cf. Bain, J. S., “Price and Production Policies” in Ellis, H. S., ed., A Survey of Contemporary Economics (Philadelphia, 1948), chap. IV.Google Scholar
3 Particular thanks are due to Mr. H. Marshall, Dominion Statistician, Mr. N. Keyfitz, who developed the method of compilation, and Miss B. Mercer, who supervised the compilation and provided much additional material.
The firm to which A plant belongs was found by reference to mailing lists and the financial manuals. Fully owned subsidiaries and those controlled through ownership of a majority of voting stock have been identified with the parent company. If a firm owns plants classified in different industries, it will appear in the statistics as several firms. This concept of the firm is more appropriate to a study of product markets than one which would record all activities of a multi-industry firm in a single industry.
4 United States, Department of Commerce, Concentration of Industry Report (mimeographed, Washington, 12, 1949).Google Scholar Reprinted in Hearings before the Subcommittee on Study of Monopoly Power of the Committee of the Judiciary, H. of R., 81st Congress, 1st Session, serial 14, part 2-B (Washington, 1950), 1437–56.Google Scholar
5 The Spearman rank correlation coefficient is 0.958. In 86 of the 136 industries output concentration exceeds employment concentration. Differences are generally small. See Rosenbluth, G., “Measures of Concentration” in Business Concentration and Price Policy, Special Conference Series, no. 5 (to be published shortly by the National Bureau of Economic Research).Google Scholar
The concentration indexes on which these conclusions are based are published in: National Resources Committee, Structure of the American Economy, part I (Washington, 1939), Appendix 7, Tables I and II.Google Scholar
6 See G. Rosenbluth, “Measures of Concentration.” Also Rosenbluth, G., Concentration in Canadian Manufacturing Industries (unpublished Ph.D. dissertation, Columbia University, 1953), chap. v.Google Scholar
7 Decisions regarding the homogeneity of product groups, the specialization of firms, and the regional segmentation of markets involved some personal judgment, as well as consultation with officers of the Dominion Bureau of Statistics. They were confirmed by comparing the resulting classification with a similar classification of United States manufacturing industries in National Besources Committee, Structure of the American Economy, Appendix 8.
8 No comparisons are made in this study on the basis of fifty firms.
9 The average size of the x largest firms in the first country (measured as a percentage of industry size) is y ÷ x per cent, and the z − x firms next in the size array cannot be larger. Hence, in this country z firms cannot account for more than zy ÷ x per cent.
10 The weighted average is the hypothetical percentage accounted for by the x largest firms in the “combined” United States industry which would be obtained if the largest firms of the component industries were under the same ownership (and, hence would be “merged” in the combined industry) and the same were true of the second largest firms, and so on, for all x firms. This is clearly the maximum possible value of this percentage for the combined industry.
11 The same relation between concentration, inequality, and number of firms holds when various other indexes of concentration and inequality are used. See G. Rosenbluth, “Measures of Concentration.”
12 Comparisons are possible in cases corresponding to types “1” and “2” described in the text above for comparisons of concentration. In addition, where one industry in Canada is matched by two or more in the United States, a comparison can be made if the maximum possible inequality in the combined United States industry is less than in Canada, or the minimum greater. Let United States component industry i have ni firms, while the largest y firms in this industry account for xi per cent of the combined industry. Then, inequality in the combined industry cannot be as great as control of ∑ xi per cent of the industry by y ÷ (∑ ni) per cent of the firms and it cannot be as low as control of Maximum xi per cent by y ÷ (Maximum ni) percent of the firms.
The upper limit represents maximum concentration and the maximum total number of Arms in the industry, while the lower limit represents minimum concentration and minimum number of firms.
13 See for example, Marshall, H., Southard, F. A. Jr., Taylor, K. W., Canadian-American Industry (New Haven, 1936).Google Scholar
14 Dominion Bureau of Statistics, The Canadian Balance of International Payments, 1926–1948 (Ottawa, 1949), 185 Google Scholar, and Dominion Bureau of Statistics, United States Direct Investments in Canada (Ottawa, 1949).Google Scholar
15 Ignoring differences of less than one percentage point.
16 United States, Department of Commerce, Concentration of Industry Report, Table IV, p. 24.Google Scholar
17 G. Rosenbluth, Concentration in Canadian Manufacturing Industries, chap. II.
18 See for example, Reynolds, L. G., The Control of Competition in Canada (Cambridge, Mass., 1940), 186–99.CrossRefGoogle Scholar On the other hand, it should be noted that the tariff tends to reduce the concentration of domestic output by stimulating the establishment of foreign branch plants in Canada. There are some indications that as a result Canadian manufacturers do not regard protective customs duties as an unmixed blessing. (See Reynolds, , The Control of Competition in Canada, 194–5Google Scholar; Southard, Marshall, and Taylor, , Canadian-American Industry, 275, 281 ff.Google Scholar)
19 See Canada and International Cartels, Report of the Commissioner, Combines Investigation Act (Ottawa, 1945), 2–16.Google Scholar Also Southard, Marshall, and Taylor, , Canadian-American Industry, 267–70.Google Scholar
20 Reynolds, , The Control of Concentration in Canada, xii.Google Scholar
21 For method of estimation, see G. Rosenbluth, Concentration in Canadian Manufacturing Industries, Appendix B.
22 Table V and the correlation coefficient are based on the 41 industries that can be compared without “merging” two or more of the component industries in the United States.
23 The distributions of these variables are extremely skew, but become symmetrical when logarithms are used. Scatter diagrams also indicate that the relation between United States and Canadian variables is linear in terms of the logarithms. Hence, logarithmic values were used for the correlations.
24 The degree of inequality of firm size has been omitted from the above analysis. For a uniform index of inequality the ratio of the average size of the largest four firms to the average size of all firms may be used (this index, multiplied by four and divided by the number of firms, gives the concentration index). The logarithms of these indexes for the Canadian and United States industries yield a correlation coefficient of 0.68. It is difficult to account for this correlation and for the fact that inequality is generally greater in the United States. The problem deserves further study.
