¹ See, for example, Averitt, Robert, The Dual Economy: The Dynamics of American Industry Structure (New York: W. W. Norton, 1968)Google Scholar; Galbraith, J. K., Economics and the Public Purpose (New York: Houghton Mifflin, 1972)Google Scholar; O'Connor, James, The Fiscal Crisis of the State (New York: St. Martin's Press, 1973)CrossRefGoogle Scholar; and Richard Edwards, David Gordon, and Michael Reich, “Labor Market Segmentation in American Capitalism” (mimeo).

² Baran, Paul and Sweezy, Paul, Monopoly Capital (New York: Monthly Review Press, 1966), pp. 43, 47Google Scholar.

³ One of the earliest works was Moody's, JohnThe Truth about the Trusts (New York: Moody Publishing Co., 1904)Google Scholar; Berle, A. A. and Means, Gardiner C., The Modern Corporation and Private Property (New York: Macmillan Co., 1932)Google Scholar presented the first time series; recent work has been spurred by Kaplan's, A. D. H.Big Enterprise in a Competitive System (Washington, D.C.: The Brookings Institution, 1954Google Scholar; revised edition, 1964) and the annual (since 1956) appearance of the Fortune “500” lists. Other studies are Norman Collins and Lee Preston, “The Size Structure of the Largest Industrial Firms, 1908–1958,” American Economic Review (December 1961); Seymour Friedland, “Turnover and Growth of the 50 Largest Industrial Firms, 1906–1950,” Review of Economics and Statistics (February 1957); Thomas R. Navin, “The 500 Largest Industrials in 1917,” Business History Review (Autumn 1970); “Management,” Forbes Magazine (September 15, 1967); David Mermelstein, “Large Industrial Corporations and Asset Shares,” American Economic Review (September 1969); and Chandler, Alfred D. Jr., “The Structure of American Industry in the Twentieth Century: A Historical Overview,” Business History Review, XLIII (Autumn 1969)Google Scholar.

Most authors have not felt compelled to offer a justification for focussing on the largest corporations, provoking George Stigler to remark that “the statistical universe of the hundred or two hundred largest corporations is inappropriate to studies of monopoly and competition.…” in “The Statistics of Monopoly and Merger,” Journal of Political Economy (February 1956). Most authors, especially Kaplan, intend their studies to be relevant to the monopoly issue on the assumption that high turnover is a sufficient condition for the existence of competition. David Mermelstein, among others, has attempted to be more explicit in relating a corporation's overall size to its power, which presumably affects its exercise of power in product markets.

See “Large Industrial Corporations and Asset Shares: Reply,” American Economic Review (March 1971). Nevertheless, for studies of concentration per se, industry studies would appear to be more appropriate. The focus here is directly on corporate power, for which the largest corporations are precisely the correct “statistical universe,” though of course there is nothing magical about any particular cutoff point (e.g., the top 100).

⁴ All of the studies concerning “Big Business” take the size of the firm's assets as the means of determining “bigness.” The choice of assets as basis for categorization represents an unfortunate but necessary compromise. Whether or not it is theoretically the appropriate variable, its measurement is sufficiently difficult to cause unease. The assets of a firm pass through a market—and hence are properly valued—only when the firm is actually sold. At all other points, the asset values must be estimated. In this respect, annual sales (on which the Fortune list is based), number of employees, or value added would provide a much more accurate measurement.

The “errors in variables” problem for assets can be simply illustrated. For extractive industries, a considerable part of each firm's assets consist of unmined minerals, oil, coal, etc. still in the ground. Not only is it difficult to measure the quantity of such stores, the value of those quantities obviously depends on such variables as the future costs of extracting them and future product prices. Thus none of the consequent estimates can be said to be based at all directly on market values—the only “true” test. A different form of the problem emerges most dramatically from the steel industry. Early estimates of the assets of U.S. Steel rely on the gross stock capitalization at its formation. Yet the merger of Carnegie Steel and several other firms to form U.S. Steel resulted in the new firm having stock worth, at face value, more than twice the combined assets of the merged companies. Some increase in the capitalized value of future earnings could legitimately be expected as a result of cost savings, increased monopoly power, and the like, and the larger value was what the J. P. Morgan promoters estimated tie stock market would bear. The subsequent decline of the stock value indicated their mistake, but it does not help the historian attempting to calculate asset values.

It might be argued that the value of the firm can be estimated from the stock prices or publicly traded shares, which prices should reflect the present value of the discounted stream of expected future net earnings. Even if this hypothesis on stock price behavior is accepted, two considerations argue against its application: (1) for many firms early in the century, the shares were not publiclv traded, so this method does not provide a general approach; and (2) for many firms whose shares were traded, large blocks of stock were held off the market by individual families (e.g., the Mellons with Gulf stock), upwardly biasing the market price of those shares traded.

These methodological problems are not trivial, but the meagerness of historical data sources allows no alternative. Evidence for asset size, though scanty, inaccurate, and not always comparable, exists; evidence for other variables does not.

⁵ Kaplan, Big Enterprise, p. 152–153, notes 13 and 37. On the basis of the “majority of assets” rule given below, Dodge is here treated as parent of the modern Chrysler Corporation and Maxwell is counted as an “exit;” see Appendix Table V.

⁶ Since specific points in time are required for the analysis, the earlier period is defined as 1903 to 1919, the latter as 1919 to 1969. Operationally I take 1919 as an acceptable compromise for the start of the latter period in order that my results can be compared directly with earlier work, but some later year, perhaps 1923, would be a better cutoff point (see footnote 8). Since these are continuous economic processes, an “exact” cutoff point is not possible nor important.

⁷ These are not, strictly speaking, entirely independent sources. All begin with the Moody's manuals. However, each supplements this basic source with investigation of company records and reports, correspondence with surviving firms, corporate histories, and similar material. Moreover, each uses a slightly different definition for what constitutes “industrial” firms. Finally, treatment of assets and estimates differ somewhat.

⁸ The choice of 1919 rather than, say, 1923 as the starting point of the later period may result in a conservative bias to the evidence for differences in stability between the two periods: 7 of the 40 “failures” listed in the bottom row of Table 2 had already occurred by 1923 (see Appendix Table V); the rate of “failure” during these four years (1.75 failures per 100 top firms per year) is very close to halfway between the failure rates for the two periods on either side. This and, as an anonymous referee has suggested, other evidence (e.g., consolidation in the auto and steel industries) make 1922 or 1923 a more plausible cutoff date.

⁹ For a discussion of early anti-trust suits, see Jones, Eliot, The Trust Problem in the United States (New York: Macmillan, 1921)Google Scholar.

¹⁰ Tarbell, Ida, The Life of Elbert. H. Gary (New York: Appleton, 1926), pp. 257–258Google Scholar. Although the strategy proved to be in vain as far as avoiding prosecution, since U.S. Steel was prosecuted anyway, the company later obtained a favorable judgment.

¹¹ These results thus broadly conform to those given by Alfred Chandler in “The Structure of American Industry.”

¹² These data do not show die several large utility consolidations which were created in the 1920's nor the dissolutions under the Public Utility Holding Company of 1935. The persistence of the original companies is hence all the more remarkable.

¹³ For a more complete discussion of monopoly capitalism see the sources cited in footnotes 1 and 2 above.