taking the period 1939 to 1946 as a whole, the smaller corporations gained considerable ground over their larger rivals.

IV. BUSINESS SIZE AND PROFITABILITY

One of the most important functions which profit should perform is to act as a natural selector, choosing for survival those business units which are fittest for serving the public interest. This suggests two questions for study. First, does capital earn the higher return when invested in large business units or in small business units? Second, does the rate of profit favor the particular size of enterprises which is best for the general welfare?

A certain amount of more or less satisfactory statistical evidence is available for answering these questions. But before examining this evidence it is well to straighten out certain logical points which are involved in its interpretation. In the first place it would not be proper to start with the assumption that business profitability is a measurement of productive efficiency. Such an approach would assume away the most important question at issue, namely, Is the size of business which is favored for survival by the profit rate the size which is most desirable from the standpoint of the public interest? It is necessary to accept, at least as a working hypothesis, the possibility that a high profit rate may be due to exploitation of a monopoly position rather than to attainment of high efficiency. Strangely, it has been the economists whose conclusions are most unfriendly to big business who have insisted that the profit rate is a measure of efficiency. In several instances such economists have come to the conclusion, supposedly on statistical grounds, that large enterprises earn a smaller return on their capital than their smaller rivals. They point to this conclusion as proof that large units are less efficient and therefore less desirable socially. Actually, such a conclusion should be reassuring from the standpoint of these investigators. The rate of profit favors for survival the more efficient small units. It seems impossible to believe that those who create large enterprises do so through a deliberate intention to behave antisocially, despite the (alleged) fact that they could earn a higher rate of return by investing their capital in smaller businesses.

Another logical point to be kept in mind is that statistical records of business profits are available only for firms which have already met the test of survival, at least for a certain length of time. Presumably, each concern has been developed to its present magnitude with some regard to the probable profitability of

different possible sizes. Statistical studies may indicate whether existing large business units are more, or less, profitable than existing small business units. They cannot give any information as to whether existing units would increase or decrease their earning rates if they became smaller or larger.

The analysis of the present section is based on three published statistical studies of the relationship between business size and profitability. The three reach radically different conclusions. The emphasis here is on the factual evidence presented rather than on the conclusions drawn by the authors of these analyses. The three studies were chosen because their statistics are enlightening, rather than because their conclusions are valid. The three studies are as follows: 1. The H. B. Summers study. This was published in the Quarterly Journal of Economics for May 1932, under the title "A Comparison of the Rates of Earning of Large-Scale and Small-Scale Industries."

2. The TNEC study.―This appeared as a part of Monograph No. 13 of the "Investigation of the Concentration of Economic Power" (the TNEC investigation). Monograph No. 13 was entitled "The Relative Efficiency of Large, Medium-Sized, and Small Business." It was prepared by the Federal Trade Commission in 1940, for the use of the Temporary National Economic Committee.

3. The Crum study. This was prepared by Prof. W. L. Crum, of Harvard University, and published as a book in 1939 under the title "Corporate Size and Earning Power."

The use of this ready-made material on the subject is a matter of convenience. There is no intention of accepting ready-made conclusions. However, the purpose is not merely to criticize these studies, but to derive from them such valid and positive conclusions as seem justified.

The H. B. Summers study 10

Mr. Summers' study has special value because his statistical measurements of profitability cover a long-time period. Studies of profits in a single year may be badly biased by the peculiar economic characteristics of that year. Profitability can be accurately appraised only by averaging the results of a series of years covering a variety of economic conditions.

The study is based on published financial statements of a sample of companies of varying size. The quantity measured is "the rate of return on permanent investment." Permanent investment, the denominator of this fraction, is also used as the criterion of size. It includes capital stock, surplus, and funded debt. The "net earnings" used as the numerator in computing the rate of return includes interest on funded debt as well as profit. The period covered is the years 1910 through 1929, inclusive.

Table 17 sumarizes the results of the Summers study. In the case of "all companies," and in five of the nine industry groups, there is a pronounced tendency for the companies in the very smallest group (less than a million dollars of permanent investment) to earn a higher return than larger companies. From these facts Mr. Summers draws the conclusion:

66* * * With respect to any 'positive advantages' accruing to large-scale production, the study offers nothing whatever in support of the idea that size in itself brings greater earning power. On the contrary, so far at least as the 1,130 companies studied is concerned, it is apparent that the opposite holds true; and with certain exceptions, heavy investment is apparently a disadvantage, rather than an advantage, in securing high rates of earnings."

However, a careful consideration of the statistical evidence casts serious doubt on this neat conclusion. First of all, notice that it is based solely on the high profit rate of the very smallest size group. If the results in this size group are set aside momentarily, there appears to be very little tendency for rate of return to vary systematically with size. Taking all companies together, those in the 2-5 million dollar class earned 9.2 percent on investment, whereas those in the over 100 million-dollar-size group earned 9.5 percent. (See the upper half of chart VII.)

A second fact to be noted in table 17 is that it is in the industries that are known to be in fields dominated by large enterprises that the small units seem to have the greatest degree of superiority in earning power: Iron and steel, automobiles, chemicals, and petroleum. It is in textiles and foodstuffs, where large enterprises are least prevalent, that the smaller units seem to have no particular advantage in respect to profitability.

10 A Comparison of the Rates of Earning of Large-Scale and Small Scale Industries, by H. B. Summers Quarterly Journal of Economics, May 1932.

96347-50-pt. 2B- -46

If Professor Summers' results are taken at face value, these odd conclusions are difficult to interpret. Apparently the organizers of large-scale units in the petroleum industry made a colossal mistake, since they could have earned a better return by spreading their investment over numerous smaller units. (This leaves out of consideration the fact that it would be virtually impossible to go into the oil-refining business with an investment of less than $2,000,000.) If they wanted to establish a large enterprise, they should have done so in the textile industry since in that field, large size is not so disadvantageous. But this is one of the fields in which large units were not formed.

However, Professor Summers' description of his procedure in picking the companies included in his sample reveals what is probably the cause of these incomprehensible results. The companies were chosen from those who published the necessary report in Moody's and Poor's Manuals. Such manuals do not cover all companies in existence, but they are likely to give more complete coverage to the larger companies. Among the smaller companies they tend to include only those which have achieved some prominence, and prominence is likely to be associated with high profitability.

As already noted, the population of small-business enterprises contains many units which last but a short time and end with the loss of all the capital invested in them. But these moribund small enterprises are seldom listed in corporate financial manuals, and, therefore, would necessarily be omitted from Professor Summers' study. He would tend to get in his sample only the more successful small companies which survived long enough to be mentioned in the manuals.

Thus it seems reasonable to conclude that the Summers' sample of very small companies is biased in favor of the more profitable firms. If account is taken of this bias among the small companies in interpreting table 17, and the class with less than $2,000,000 left out of consideration, the only conclusion left is that there is, in general, no tendency for profitability to vary with the size of the business unit.

The discussion of the economic significance of this conclusion can best await its confirmation (or contradiction) by the other two statistical studies to be examined.

The TNEC study

The TNEC analysis of the relation of business size to business profits was made in connection with the study of the "relative efficiency of large, medium-sized, and small business." As already explained, there is a logical inconsistency involved in the assumption that profitability in relation to size can be used as a measurement of efficiency in relation to size.

Systematic criticisms of the TNEC study have been made elsewhere," and will not be attempted here. The present purpose is to glean from the statistical material whatever may be of use in this inquiry.

The TNEC study measures profitability in terms of "rate of return on invested capital"; that is, profit as a percent of stockholder's equity. The analysis consists of a series of "tests," each such test being a comparison of rates of return for companies of various size in one industry in 1 year. The companies are classified as large, medium-sized, and small, not by any quantitative rule, but by the judgment of the authors of the study.

The results of these "tests" are summarized as follows in the monograph: "In the 84 tests made for the rates of return on invested capital earned by individual companies in 18 industries, the largest company showed the highest rate of return only 12 times.

"In 2 of the 84 tests a large company, although not the largest, showed the highest rate of return.

"In 57 of the 84 tests a company classified as medium-sized showed the highest rate of return.

"In 13 of the tests a company classified as small showed the highest rate of return.

"On the average about one-third of the total number of companies in each test showed higher rates of return than the largest company.'

12

Notice that each statement in this quotation, except the last, is concerned with the size of the particular company which happened to show the highest

11 See Fact and Fancy in the TNEC Monographs, compiled by John Scoville and Noel Sargent, p. 19 et seq.

19 TNEC Monograph No. 13, The Relative Efficiency of Large, Medium-Sized, and Small Business,

p. 13.

0

rate of return for the test. This is a very limited use of the data. The last statement is the only one which gives any attention to the question of whether there is any general tendency, within each test comparison, for size to be connected with rate of return. This treatment of the data makes its evaluation extremely difficult.

The summary quoted above seems designed to create the impression that larger companies make a poor profit showing in comparison with smaller firms. (Note the use of the word "only" in the first quoted sentence.)

However, the statement that about one-third of the total number of companies earned a better return than the largest, contradicts this conclusion. If there were no tendency whatever for the largest firm to earn a better return than the general average, then about one-half of the firms should earn a better return than the largest. The actual record shows the largest firm doing a little better than this, thus indicating that there may be some slight advantage, profitwise, in large size.

The authors of this TNEC monograph seem to have been obsessed with the single problem of discovering the size of the company which showed the best profit record in each test. Thus their results merely show that the most profitable company is not invariably the largest company, it is not invariably a large company, it is not invariably a medium-sized company, and it is not invariably a small company. This result does not contradict the general conclusion derived from critical analysis of the H. B. Summers study-i. e. that profit is not systematically affected by business size-although it cannot be said to give that conclusion much positive support.

The Crum study

The study prepared by Prof. W. L. Crum has the virtue that it covers all corporations in the country, rather than any selected group. It has the disadvantage that it analyzes the data only for the years 1931 to 1936, a period of descent into, and incomplete emergency from, the serious depression of the 1930's. For purposes of the present discussion, the Crum results have been extended, using the same methods, to the years 1939 and 1943.

Professor Crum measures profit as a percentage of average stockholders' equity. The standard of size is the total assets of the corporation. The figures are derived from the corporate income tax returns, as compiled and published by the Bureau of Internal Revenue.13

Professor Crum's most general conclusion can best be presented in his own words:

"I believe my findings clearly show that, on the average, large enterprise-in all or nearly all broad lines of industry, and in different stages of the economic cycle is more profitable than small enterprise, especially very small enterprise." 14 Notice that this is the reverse of the conclusion reached by Professor Summers. This conclusion is supported by a large mass of data relating to different industries. The facts in regard to all corporations taken together are reproduced from the Crum study as table 18 of this report.

Since losses predominate in the period covered by these figures it would be more accurate to say that they indicate that larger units have a smaller rate of loss on equity, rather than a higher rate of profit. This suggests the desirability of duplicating the computations for a more profitable period. This is done for the years 1939 and 1943, in table 19.

In the year 1939 there seemed to be a tendency for the profit rate to increase with size up to about $500,000 of assets. Beyond that point, while there was some variation in the profit rate, there was no systematic increase in the rate of return with increasing size of assets.

In 1943, the very small corporations (less than $50,000 of assets) earned a slightly lower return than the remaining corporations. However, among the nine size classes above $50,000 of assets, the profit rate showed very little variation with corporate size.

This suggests, although it does not prove, that Professor Crum's conclusions may be due to the special character of the period he studied. There seem to be better grounds for believing that there is in general no particular association between business size and business earning power.

13 See the annual volume Statistics of Income.

14 Corporate Size and Earning Power by Prof. W. L. Crum, p. 7.

SOURCE:

CCMPUTED BY MAM FROM DATA PUBLISHED BY THE U. S. BUREAU OF INTERNAL REVENUE

Size and profitability-conclusions

Prof. H. B. Summers asserted, on the basis of his statistical studies, that in general large business units are less profitable than smaller enterprises. W. L Crum, also on the basis of statistical evidence, came to the opposite conclusion. The present examination of their evidence has led to the conclusion that both were wrong, and that, in general, big business is neither more or less profitable than small business.

The first reaction to this conclusion is to be surprised, and even shocked. Doesn't it destroy any possible justification for big business? Why do businessmen form large enterprises if they are not more profitable than smaller concerns? Actually the statement that large, medium-sized and small business are about equally profitable makes more economic sense than any other possible conclusion. First it must be emphasized that the empirical facts considered in reaching this result apply to the firms, large and small, which have been formed and are actually operating. Presumably there are economic reasons for each such firm to have assumed its present size. The statistical examination indicates that there is no relationship between the size and the profitability of these existing firms. It does not prove that any given firm's profits would not be affected if it were either reduced or increased in size. Our conclusion, although it seems to deny any connection between size and profitability, is perfectly consistent with the view that in the cases where large size is advantageous, large size has been reached, and in cases where small size is most profitable, the units tend to be small. Had it been discovered that large units are in general more profitable than small units, it would necessarily have followed that the business structure was out of balance. If capital can earn more when it is collected in large organizations, then this must be a deterrent to investing it in the less profitable small enterprises. Small business would have to be viewed as an economic anachronism, whose present survival is due merely to a social lag.