Theory of the Growth of the Firm

by Edith Penrose

  241 The finding by Butters, Lintner, and Cary that merger was a more important source of growth for small firms as a group than for large firms (ibid., p. 267) has been challenged by John Blair and M. F. Houghton in the Review of Economics and Statistics, Vol. XXXIII, No. 1 (Feb. 1951), pp. 63–7 and by John Blair in Vol. XXXIV, No. 4 (Nov. 1952), pp. 34364. The central issue of the controversy turns on the amount of unreported mergers undertaken by small firms. Lintner and Butters effectively sustained their position in a rejoinder to Blair’s second reply, ibid., pp. 364–67.

  242 In one study it was found, for example, that the younger and newer of the major firms in the United States steel industry did not expand less and sometimes expanded more in depression than did the largest firms. Gertrude Schroeder, The Growth of Major Steel Companies, 1900–1950 (Baltimore: The Johns Hopkins Press, 1952), pp. 207–8.

  243 See Ch. X.

  244 See the excellent discussion of the subject in Gideon Rosenbluth, ‘Measures of Concentration’, Business Concentration and Price Policy, op. cit., pp. 57 ff.

  245 The United States Federal Trade Commission for the most part presents its estimates of concentration in terms of the per cent of the value of total shipments in given industries which can be attributed to the four largest companies, but it also uses alternative measures, in one of which concentration is expressed in terms of the number of firms controlling 60 per cent of output. Rosenbluth’s study of concentration in Canada related primarily to the number of firms required to account for 80 per cent of output. Gideon Rosenbluth, Concentration in Canadian Manufacturing Industries, op. cit.

  246 For an excellent discussion of the statistical and conceptual problems relating to the measurement and meaning of concentration, see the papers by Conklin and Goldstein, Rosenbluth, and Scitovsky, and the comments thereon in Business Concentration and Price Policy, op. cit.

  See also the discussion of the possible uses of the logarithmic normal curve for measuring concentration (in this case inequality or relative concentration) in P. E. Hart and S. J. Prais, ‘The Analysis of Business Concentration: A Statistical Approach’, Journal of the Royal Statistical Society, Series A. Vol. 119, Part 2 (1956); also reprinted by the National Institute of Economic and Social Research, London, Reprint Series No. 8.

  247 In a valuable survey both of the methods of measuring concentration and of the empirical evidence regarding concentration, Adelman concluded that concentration in terms of total assets was about 60 per cent higher for the largest 200 firms in the United States in 1948 than concentration when measured in terms of employment. M. A. Adelman, ‘The Measurement of Industrial Concentration’, Review of Economics and Statistics, Vol. XXXIII, No. 4 (Nov. 1951), p. 278.

  248 As we have shown earlier, the fact that there is a limit on the rate at which a firm can expand its productive operations may be responsible for an increase in its financial investments in other firms. Thus if we measure the expansion of large firms in terms of their total assets we may very much overstate their expansion as producing units, for the decline in the rate of growth of large firms as producing units may be marked by an increase in their financial size. Consequently, concentration of the kind we are discussing here—neglecting purely financial as contrasted to organizational relationships—is not a measure of total economic and financial power.

  249 See Simon Kuznets, ‘Quantitative Aspects of the Economic Growth of Nations: Levels and Variability of Rates of Growth’, Economic Development and Cultural Change, Vol. 5, No. 1 (Oct. 1956) Table 9, pp. 38–40, where calculations of changes in the average rates of growth of a number of countries are presented for three different periods since around the last quarter of the 19th century.

  250 Federal Trade Commission, Report on Changes in Concentration in Manufacturing, 1935 to 1947 and 1950 (Washington D.C., 1954). Between 1935 and 1950 the Commission, using Census data, found that the percentage of total value of product accounted for by the five largest companies rose from 10.6 to 11.4; by the 50 largest, from 26.2 to 26.6; by the 100 largest, from 32.4 to 33.3; and by the first 200 companies, from 37.7 to 40.5 (p. 17). See also page 19 of this report for a discussion of other attempts to measure concentration in the same period.

  One of the more comprehensive surveys of the data concerning concentration in the United States since the turn of the century is to be found in M. A. Adelman, ‘The Measurement of Industrial Concentration’, loc. cit. After examining the various measures and various studies of concentration, he concluded that since 1901 ‘The odds are better than even that there has actually been some decline in concentration. It is a good bet that there has at least been no actual increase; and the odds do seem high against any substantial increase’. Professor Adelman’s methods and conclusions were criticized by Corwin Edwards, George Stocking, Edwin George, and A. A. Berle, Jr., but none of the discussion seriously upset his basic contention regarding the long-run trend. See ‘Four Comments on “The Measurement of Industrial Concentration” With a Rejoinder by Professor Adelman’, Review of Economics and Statistics, Vol. XXXIV, No. 2 (May 1952), pp. 156–78. Adelman’s conclusions were also attacked by John Blair ‘The Measurement of Industrial Concentration: A Reply’, Review of Economics and Statistics, Vol. XXXIV, No. 4 (Nov. 1952), pp. 343 ff. The controversy became rather heated and confusing, but by and large the evidence offered does not seem to be sufficient to challenge the conclusion as stated by Lintner and Butters that ‘Even though imperfections and gaps in the data counsel caution, the best available evidence establishes a rather strong presumption that there has been no increase in over-all concentration over the last fifty-year period and indicates that there probably has been some decrease in concentration over this period, at least so far as manufacturing is concerned’. John Lintner, and J. Keith Butters, ‘Effects of Taxation on Concentration’, Business Concentration and Price Policy, op. cit., p. 239.

  Adelman’s analysis was based largely on total assets and he gave strong reasons for rejecting net capital assets. For his purposes his reasons are good, but in the framework of this study, total assets would be an inappropriate measure. However, there is no reason to presume that the trend of concentration would be towards substantially higher levels over the long haul if measured in terms of capital assets; rather I should expect long-run concentration to show even more of a decline.

  251 For example, the fears of Berle and Means: ‘The corporate system has done more than evolve a norm by which business is carried on. Within it there exists a centripetal attraction which draws wealth together into aggregations of constantly increasing size, at the same time throwing control into the hands of fewer and fewer men. The trend is apparent; and no limit is as yet in sight. Were it possible to say that circumstances had established the concentration, but that there was no basis to form an opinion as to whether the process would continue, the whole problem might be simplified. But this is not the case. So far as can be seen, every element which favoured concentration still exists, and the only apparent factor which may end the tendency is the limit on the ability of a few human beings effectively to handle the aggregates of property brought under their control’. A. A. Berle Jr. and Gardiner C. Means, The Modern Corporation and Private Property (New York: Macmillan, 1932), p. 18. It is the contention of our analysis that the process cannot be expected to continue, at least in the form of the expansion of the industrial firm, which, to be sure, Berle and Means were not specifically concerned with, but not for the reasons they advanced. We have found no evidence that the limit to concentration lies in the inability of men to administer large units.

  252 P. E. Hart and S. J. Prais, ‘An Analysis of Business Concentration’, loc. cit.

  253 Ibid., p. 175.

  254 Gideon Rosenbluth, Concentration in Canadian Manufacturing Industries, op. cit., pp. 80 ff.

  255 Rosenbluth’s comparison was between concentration of employment in Canada and of output in the United States because these were the figures available. Concentration when measured by output t
ends to be higher than when measured by employment, and consequently his figures probably overstate concentration in the United States relative to that of Canada. Ibid., p. 76. Furthermore, his study dealt only with concentration in specific industries and his general statements about concentration in the economy as a whole are based on a comparison of industry concentration in the two economies.

  256 ‘Growth of the market is therefore revealed as an important factor, which a theory of concentration cannot afford to neglect. Further growth of the Canadian economy can be expected to exercise a continued moderating influence on the level of concentration. Should concentration nevertheless increase, the theory that such increase is necessary for technological efficiency should be treated with distrust’. Ibid., p. 108. This statement was made with specific reference to plant concentration, but it is equally valid for firm concentration.

  257 Furthermore, concentration in particular industries and concentration in the economy as a whole can move in different directions, the one increasing and the other remaining constant or even decreasing. See Federal Trade Commission, op. cit., p. 19.

  258 In the United States in 1947, for example, ‘Among manufacturing industries, four industries—steel, automobiles, chemicals, and oil—account for nearly two-thirds of the assets of the over-$100 million group’, Adelman, loc. cit., p. 276.

  259 I am making no attempt here to distinguish ‘industry’ from ‘product’. We can treat the interstices as opportunities to produce new products within or outside existing industries, or as opportunities to enter old or establish new industries, whichever seems most appropriate. The amorphous boundary lines between ‘industries’ and ‘products’ provide one of the reasons why it may be misleading to estimate concentration in the economy as a whole from a weighted average of ‘industry’ concentration. The narrower the definition of industry or product, the higher is concentration is likely to be.

  260 Cf. the following comment: ‘In the early stages of the development of a new industry, entry very often is easy, not only (or even mainly) because the design of the product and the technique of production are primitive by the standards of later developments, but because as the product “takes on” the aggregate demand expands rapidly of its own momentum, so that custom is easily acquired. Hence the “mushroom” of small-scale, high-cost producers is characteristic of the early development of a new industry. In time, however, as the rate of expansion slackens and the process of production is consolidated in order to realize the full potentialities of the production function, a price equal to the cost of production in the most efficient manner—the “right” price—is possible. It is at this stage (“settled” conditions), and if such a price prevails, that new entry is very difficult’. H. R. Edwards, ‘Price Formation in Manufacturing Industry and Excess Capacity’, Oxford Economic Papers, (New Series), Vol. 7, No. 1 (Feb. 1955), p. 101.

  261 Our analysis of the growth of firms has been primarily concerned with the growth and size of the total organization of a firm measured in terms of capital assets. Consequently it is not entirely adapted to the analysis of concentration measured in terms of output or employment. Increases in productivity, for example, may permit increases in output (and of concentration of output) without corresponding increases in capital investment and hence of size in terms of capital assets; a change in the capitallabour ratio may cause a decrease in the size of a firm measured in terms of employment and an increase measured in terms of capital assets. Asset concentration tends to run higher than output concentration largely because a high capital-output ratio facilitates concentration; output concentration tends to run higher than employment concentration for much the same reason—a higher capitallabour ratio usually implies higher labour productivity. See, for example, the discussion of mechanization in particular industries in Federal Trade Commission, op. cit., p. 29.

  262 Compare the following suggestion: ‘Since established firms operating in other industries frequently have the least disadvantage of all potential entrants to a given industry in acquiring the requisite capital, we should perhaps go slow in frowning officially on expansion of large firms via diversification to enter new fields. To discourage or prohibit this sort of diversification may well tend to raise the barriers to entry to industries generally, with adverse effects on competition. This disadvantage of an anti-diversification policy vis-á-vis large firms must be weighed carefully against the alleged dangers of the growth of gigantic firms per se. Joe S. Bain, op. cit., p. 215.

  263 Incidentally it should be noted that concentration of assets is not the appropriate measure for the analysis of the effect of diversification. Not only is it extremely difficult to allocate in any satisfactory manner the total capital assets of a large diversified firm among specific products, or to analyse the increase in the output of particular products with reference to the factors limiting the expansion of the firm as a whole, but it does not make much sense to do so from any of the points of view from which we might be interested in concentration. Only a measure of output is appropriate for this purpose.

  264 The Marxian analysis of capitalism implies, of course, a theory of increasing concentration; the Marshallian theory of the rise and fall of firms implies a theory of a constant degree of concentration with a continual change of the individual population of firms; and the theoretical analysis of the question of increasing returns to scale is also an analysis of one aspect of the problem of increasing concentration of industry. What I have called the ‘theory of the headstart’ has been floating around in a nebulous form for a long time, its most recent statement, perhaps, being that of J. K. Galbraith, op. cit., esp. pp. 33–5.

  265 A. D. H. Kaplan, Big Enterprise in a Competitive System (Washington D.C.: Brookings Institution, 1945), pp. 145 ff.

  266 United States Federal Trade Commission, A List of 1,000 Large Manufacturing Companies, Their Subsidiaries and Affiliates, 1948 (Washington, D. C., 1951).

  267 Kaplan draws the conclusion from his tables that ‘industrial leadership at the big business level is precarious’ (p. 141) and that the turnover among the leaders is great. But the data, if examined carefully, do not support this conclusion. The way in which merger among the group of large firms is handled by Kaplan is inaccurate, and, as Carl Kaysen has shown in a review (Explorations in Entrepreneurial History, Vol. 7, No. 4, April 1955, pp. 237–9) the data as presented show a substantial decline in the turnover of the identity of the firms among the 100 largest in the 30-year period, the number of firms dropping out of the list in each decade declining steadily and significantly from 1909 to 1935. It rises from 1935 to 1948, but even apart from the fact that several mergers were neglected by Kaplan, five of the firms that he had ranked among the 100 largest at some time and which do not appear in his list of the 100 largest in 1948, are listed among the 100 largest in 1948 by the Federal Trade Commission. Both lists relate to total assets, Kaplan’s referring to ‘industrials’ and the Commission’s to ‘manufacturing firms’. There are apparently differences between the two universes, but it is hard to account for the extent of the variation, which at least throws doubt on the ‘precarious’ nature of leadership as deduced from the shifting position of firms in Kaplan’s tables. Furthermore, if one makes allowance for the growth in the total number of large firms in the period (and Kaplan admits that to confine the analysis to 100 is arbitrary), one will find that remarkably few firms really suffered a significant loss of ‘leadership’ position. Kaplan’s fundamental point—that there is a good deal of competition ‘at the top’ and that the relative rank of firms constantly shifts—is sound enough, but to deduce ‘precarious’ leadership from this is going too far.

  268 One economist, at least, has wholeheartedly endorsed the case: ‘There can be little doubt that oligopoly, both in theory and in fact, is strongly oriented toward change. There can be no serious doubt at all that the setting for innovation, which is so favorable in this market structure, disappears almost entirely as one approaches the competition of the c
ompetitive model’. J. K. Galbraith, op. cit., p. 90. Galbraith has stressed the importance of what he called ‘countervailing power’, as a means of preventing the exploitation of the weaker groups in the economy by the stronger. The power of big buyers can offset that of big sellers, the power of organized business can be matched by organized labour, etc. In some areas, e.g., agriculture, the government may have to support the less powerful in order to achieve a balance. Even if ‘countervailing power’ is effective in preventing exploitation, however, it is not a means for insuring that the interstices are kept open for the smaller firms.

  269 I am making no attempt to explain here the difference between the economies of size alone, the economies of size that are also economies of growth, and the economies of growth that are not economies of size. I would ask the all-too-common busy reader who attempts to grasp the nature and import of this book from the few conclusions re-stated here to turn to Chapter VI, a short chapter dealing specifically with these distinctions, and of fundamental importance for the understanding of our entire argument.