Accounting rates of return are frequently used as indices of monopoly power and market performance by economists and lawyers.' Such a procedure is valid only to the extent that profits are indeed monopoly profits, accounting profits are in fact economic profits, and the accounting rate of return equals the economic rate of return. The large volume of research investigating the profits-concentration relationship uniformly relies on accounting rates of return, such as the ratio of reported profits to total assets or to stockholders' equity as the measure of profitability to be related to concentration.2 Many users of accounting rates of return seem well aware that profits as reported by accountants may not be consistent from firm to firm or industry to industry and may not correspond to economists' definitions of profits. Likewise, they recognize that accountants' statements of assets, hence also stockholders' equity, may fail to correspond to economically acceptable definitions, because accounting practices do not provide for the capitalization of certain activities such as research and development and do not incorporate allowances for inflation. This is to say they are well aware of certain measurement problems which arise in using available accounting information to measure profitability. They seem, however, totally unaware of a much deeper conceptual problem, namely, that accounting rates of return, even if properly and consistently measured, provide almost no information about economic rates of return.3 The economic rate of return on an investment is, of course, that discount rate that equates the present value of its expected net revenue stream to its initial outlay. Putting aside the measurement problems referred to above, it is clear that it is the economic rate of return that is equalized within an industry in long-run industry competitive equilibrium and (after adjustment for risk) equalized everywhere in a competitive economy in long-run equilibrium. It is an economic rate of return (after risk adjustment) above the cost of capital that promotes expansion under competition and is produced by output restriction under monopoly. Thus, the economic rate of return is the only correct measure of the profit rate for purposes of economic analysis.4 Accounting rates of return are useful only insofar as they yield information as to economic rates of return.5 *Fisher is professor of economics, Massachusetts Institute of Technology. McGowan was Vice-President, Charles River Associates. He died on April 7, 1982. This paper is based on work done for Fisher's testimony as a witness for IBM in U.S. v. IBM (69 Civ. 200, U.S. District Court, Southern District of New York). We are indebted to Larry Brownstein, Steven Hendrick, and especially Karen Larson and Leah Hutten for computational and programming assistance. Any errors are our responsibility. 'Aside from U.S. v. IBM, see, for example, Joseph Cooper, p. 15; the various industry studies in Walter Adams; and the discussion in Philip Areeda and Donald Turner, Vol. II, pp. 331-41. 2See the comprehensive reviews of this literature by Leonard Weiss and more recently by F. M. Scherer, pp. 267-95. Additional accounting problems raised by attempting to measure profitability by line of business are discussed extensively in George Benston. 3A referee suggests that even the crudest accounting information tells us IBM is more profitable than American Motors (AMC), but we disagree. Surely accounting information tells us IBM generates more dollars of profits per dollar of assets than does AMC but, as the examples below demonstrate, that information alone does not tell us which firm is more profitable in the sense of having a higher economic rate of return. 4This is literally true only if the cost of capital is first subtracted. In what follows below, we follow the usual empirical practice of measuring all rates of return before such subtraction. sThe existence of a uniquely defined economic rate of return-which we now assume for the theoretical analysis below and which occurs in all the examples-is
Stored only when both papers are in the corpus — 772 citations exist upstream.