Internal Revenue has had all intercorporate relations between the income statement and balance sheet of the parent and the subsidiary removed, but the unconsolidated return has not. If accounts receivable and accounts payable include accounts receivable from and payable to other corporations in the same control unit, these should be deducted from the totals for the unit. But in the case of unconsolidated subsidiaries, this cannot be done. The reports to the Bureau of Internal Revenue do not distinguish between accounts payable to or receivable from a corporation in the same control unit, and those to or from any other corporation. Therefore, items subject to this type of duplication show evidence of large error (for the purposes of this report), and the following items should be read with this in mind:

Interest received from taxable investments.
Cash dividends received.

Interest paid.

Cash dividends paid.

Compiled net profits by definition includes dividends received, and is therefore subject to this type of error. The same is true of the derived items: income derived from operations, and corporate savings, each of which includes items which involve double counting.

The summation of figures for total assets does not represent complete consolidation of the balance sheet of each unit of control. Neither does the summation of total assets of all corporations reported by the Bureau of Internal Revenue represent the consolidation of control units. This discrepancy arises out of the intercorporate holdings of securities and loans within the same corporate unit which would be cancelled out in the process of consolidation. A direct comparison of the sum of the assets of the 200 largest control units and the sum of all corporations reporting to the Bureau of Internal Revenue tends to give a larger concentration ratio than would be obtained if the assets of all control units, whether in the 200 largest or not, were completely consolidated. A large part of the discrepancy arising from the lack of complete consolidation can be corrected by subtracting taxable investments, the element making the largest contribution to the discrepancy, from the figures of both the largest and all corporations. The distortion resulting from intercorporate lending within corporate units is probably small. Subtraction of the entire holdings of taxable investments overcorrects somewhat because of holdings of securities of corporations not in the same control unit. There would be no error in the concentration ratio from this source if the securities of other corporations held as investments by the 200 largest units and those held by corporations other than the 200 largest bear the same proportion to their respective total consolidated assets. This seems to be a condition

closely enough approximated to make the discrepancy arising from this source of a minor character.

The concept of "the assets of the 200 largest corporations" may now be refined. What is really meant is the arithmetic sum of all the asset items on the 200 largest nonfinancial consolidated balance sheets (and, of course, income statements). All intercompany relations within any consolidated balance sheet would be eliminated but not other intercompany relations within the 200. A distinction must be made, therefore, between this concept and the concept of a single consolidated balance sheet embracing the 200 largest consolidated balance sheets. The latter would eliminate intercorporate relations between any corporations affiliated with any of the 200 largest corporations. The latter totals would be smaller in the items mentioned above, since intercompany eliminations would have been made.

As a matter of fact, it was impossible to set up the 200 consolidated balance sheets, but departures from this desired procedure are regarded in this report as

errors.

Besides the double counting that cannot be eliminated from the sum of the parent and unconsolidated subsidiaries, other errors are likely to occur. The classification of unconsolidated subsidiaries in the industrial group of their parents may change the geographical classification from that in which they are listed in Statistics of Income.13 These shifts do not concern this report. But the industrial classifications may be, and undoubtedly often are, shifted. The unconsolidated subsidiaries were reported in their own industrial groups in Statistics of Income, but in the present tabulation they were reported in the industrial group of their parent companies. The resulting industrial misclassification has been discussed in section 1, above.

14

5. Notes on the Interpolation

Three asset items, total assets, capital assets, and total assets less taxable investments, were estimated for the 200 largest corporations for the years 1930–32 on a basis comparable with table I. The procedure and assumptions were set up in consideration of the state of the data for this period, and the limited time and clerical facilities allotted to this study. Therefore, these figures are merely estimates rather than tabulated totals and they are derived by making certain specific assumptions which are subject to partial test.

Estimates were derived for each industrial group and the four estimates were summed to obtain the estimated totals for the 200. The work was shortened by omitting from consideration all unconsolidated subsidiaries and by using all the companies in the lists of the 200 largest

13 The same applies to net income-net deficit classification as well.

14 And therefore in the denominator of the concentration ratio.

for both 1929 and 1933 where the data were available. This procedure is justified by the small number of replacements in the list a yearly change of 2 or 3 percent. So that all information available would be used, the yearly percent changes of all companies which reported in both of each two successive years were used. From these comparable data a chain index series was constructed for each item for each industrial group. These indexes were taken to represent the percent changes of all companies in the largest 200, including unconsolidated subsidiaries. 15

The chain indexes were derived from yearly percent changes of identical companies. The actual data used were not entirely from the Bureau of Internal Revenue. Where a company was missing in a single intermediate year, or where an erratic change threw one year out of line with the two adjacent years, a comparison was made with Moody's Manuals. If the behavior of the data for the preceding and following years from the two sources approximately agreed,16 the intermediate year was interpolated using the Moody's figures as an index of change.

This augmentation of the Bureau of Internal Revenue data had a tendency to stabilize the annual percent changes by increasing the size of the samples and by eliminating individual erratic observations. Hence the movements of the chain index were damped; the erratic quality of the index was understated. This modification of the originally designed procedure was intended to protect the annual percent changes from variation in the degree of consolidation in the returns of the same corporations in adjacent years.

The 1933 figure of this chain index on a 1929 base gave the percent change for the period 1929 through 1933. But, as a matter of fact, this percent change was accurately known, for it could easily be derived from the two tabulations for 1929 and 1933. This gave a correct index number for 1933. The crude 1933 chain index number was therefore adjusted to equal the correct index number and the earlier years were adjusted by an increment based on an assumed linearity of the drift for the period 1929-33.17 After the indexes were corrected they could be multiplied by the base figure, yielding a complete set of estimates of the three asset items by industrial groups for the three interpolated years.

These estimates can be tested in various ways. If

15 No independent determination of the lists of the 200 largest for the interpolated years was made.

16 i. e., had approximately the same level and the same percent change for the 2-year interval.

17 Hence the difference between the 1933 relative of the complete tabulation for each industrial group (1929=100) and the corresponding chain index for 1933 (1929=100) was allocated evenly over the four annual intervals. This adjustment was made necessary by at least two definable causes: (1) The number of the 200 corporations in each industrial group was not the same in 1929 and 1933 in every case, and (2) the index was not perfectly representative of the 200 parents, plus their consolidated subsidiaries, plus their unconsolidated subsidiaries of all sizes.

the percent changes are truly representative of the whole 200 largest corporations, including unconsolidated subsidiaries, and the industrial percent distributions are accurate, the sum of the totals for the various industrial groups should be the same as the estimates for the 200 largest taken as a unit without regard to industrial classification. This comparison was made and is presented in table IX. The index numbers made up without regard to industrial classification were derived as described above. The absolute totals for the various industrial groups were summed, and the totals for all groups reduced to index series for comparison.

TABLE IX.-Estimates made (1) with and (2) without regard to industrial groups

This test is one-directional in nature. For the estimated figures for industrial groups to be accurate, it is a necessary, but not a sufficient condition, that the two estimates be close. If the estimates made by the two methods are close, the industrial group estimates might be accurate and they might not. But if the two estimates are not close, the industrial group estimates cannot be accurate. By definition, the closeness of the estimates for the total, with and without regard to industrial groups, is evidence in the direction of accuracy of that total. It can be seen directly that the two indexes yield the same result within rounding errors except for the middle year. The discrepancies for total assets are 0.0, 0.1, 0.2, 0.1, 0.0, and those for capital assets are 0.0, 0.1, 0.5, 0.1, 0.0. The assumption of a linear adjustment makes the end years correct. The years next to the end years are least likely to suffer from this procedure, with the middle year the least accurate. This is what is actually found. The largest errors indicated here are smaller than other errors known to exist for these estimates, so it is evident that the interpolation meets this test of consistency.

A second test of the accuracy of the interpolation may be made by further examining the internal consistency of the results. Total assets should be larger than capital assets and larger than total assets less taxable investments. The difference between total assets and total assets less taxable investments should yield a reasonable result for taxable investments. Similarly, the difference between total assets less taxable investments and capital assets should yield a reasonable result for liquid assets.is These estimates for taxable

18 Inventories, cash, notes and accounts receivable, and miscellaneous assets.

investments and liquid assets, which are implied in table IV, raise no serious problems

A third test of the accuracy of the interpolation depends on the obvious condition that any asset item for the 200 largest must be smaller than the same item for all nonfinancial corporations. This must apply to taxable investments and liquid assets as well as total assets, capital assets, and total assets less taxable investments. In every case the figures for the 200 largest were smaller than the figures for all corporations.

From the three tests described above there would seem to be little objection to the figures in table IV that

could arise from the interpolation.

Taxable investments were poorly reported and the chain index intended for interpolation had to be based on a small and somewhat erratic sample. Therefore the same index as was used to interpolate for total assets was also used to interpolate for total assets less taxable investments by applying it to the 1929 and 1933 figures for total assets less taxable investments. No inconsistency was detected in the resulting figures. Total assets and capital assets were interpolated from chain indexes made up of yearly percent changes of total assets and capital assets, respectively.

A crude investigation was made of the degree of concentration of financial corporations in 1933. The list examined was composed of the 50 largest financial corporations in 1933, excluding unconsolidated subsidiaries. Of the 50, 24 were banks, 17 were insurance companies,19 and the remaining 9 were "other financials". The list of the 50 largest was selected after examination of the returns of all financial corporations with total assets over 50 million dollars. The 50 corporations with the largest total assets (considered independently of their unconsolidated subsidiaries) which were independent, according to Moody's, and which were classified by Moody's as financials, were listed. This last requirement eliminated three companies (holding companies) classified by the Bureau of Internal Revenue as financials, but classified by Moody's in the Utilities or Railroad Manuals. One company included in the list actually was not independent at the end of 1933. Since the company was independent through most of the year, since the assets of its parent were too small for the parent to get on the list, and since the corporate structure of its parent was so involved that the Moody analysis in no way corresponded with the situation found by the Securities and Exchange Commission, the company was included in the list as independent.

The items tabulated for the financial corporations were selected on the basis of their importance, and are not intended to give as complete a picture as the items for the 200 nonfinancial corporations. In particular, the complete asset side of the balance sheet is not presented for the financial corporations. Life insurance companies file a special type of income-tax return, on which no item corresponding to receipts is reported, so that the receipts tabulated for the 50 largest financial corporations are not a reliable measure of the quantity of business done by them.

19 While the definition included all types of insurance companies, all of these were, in fact, life insurance companies.

No attempt was made to adjust the totals for the 50 for missing subsidiaries, as all unconsolidated subsidiaries, of whatever size, had been omitted from the tabulation.

In comparing the 50 largest to all financial corporations, the balance-sheet items for all financial corporations were adjusted for corporations not submitting balance sheets, using the same procedure as was used for nonfinancials. Since none of the 50 financials were real estate corporations, it was felt that to compare them with all financials, including real estate companies, would distort the concentration ratios for certain items, particularly capital assets. However, adjustment for missing balance sheets could not be made directly for financials excluding real estate, since the Bureau of Internal Revenue did not separate returns with balance sheets from returns without balance sheets for subgroups of financial corporations. Consequently, the same adjustment factors as were used for all financials were applied to the totals for financials less real estate. The error thus introduced is insignificant.

Table X shows the totals for the 50 largest financial corporations, the totals for all financial corporations, and all financial corporations excluding real estate, with their adjustments, and the concentration ratios derived therefrom.

This part of the study is very crude, so a few words of caution are in order. The 50 largest financials exclude unconsolidated subsidiaries, so the unit of control is not the same as the unit in the nonfinancial corporation statistics. A "financial corporation" is therefore not comparable to a nonfinancial corporation in the terminology of this study.

The 50 largest financials are not to be regarded as an "equally important" or "the same" proportion of the total for all financials as the 200 nonfinancials are of the total for all nonfinancials. Fifty was merely a convenient number of financial corporations chosen to show a significant amount of concentration when compared

This study of interlocking directorates covers the directors of the 200 largest nonfinancial corporations and the 50 largest financial corporations in 1935. For the nonfinancial corporations, the list of 200 largest nonfinancial corporations with their assets which appears in Appendix 10 was used. The 50 largest banks and financial companies (30 banks, 20 financial companies) are listed in table I.2

The names of directors for each corporation were compiled from the lists of corporation directorates in Poor's Register of Directors, 1936. In some few cases, where Poor's omitted a corporation, Moody's Manuals were used.

A summary of the results of this study has been given in chapter IX, charts I and II. The following tables present these results in more detail.

Only 25 of the 250 corporations have no interlocks with each other. These companies are relatively small, in terms of assets, as compared with the interlocking companies. Although they constitute 10 percent of the number of companies, their assets amount to only 4 percent of the total assets of the 250 companies. The names of the 25 noninterlocking companies are given in table II. They comprise 16 industrials, 8 utilities, and 1 railroad. There are no banks and no financial companies among them. The absence of interlocking directorates between these 25 companies and others in the list of 200 largest nonfinancial and 50 largest financial corporations does not necessarily mean that these companies are free from other types of links. They include companies which are relatively free from outside control as the Crane Co., which is owned in large part by the Crane family, and companies such as Atlantic Refining and Ohio Oil which are successor firms to the old Standard Oil Co., and are members of the Rockefeller interest group.3

The 225 corporations which show interlocks with each other are classified in tables III, IV, and V, and the assets represented in each class are shown.

The interlocking directorates between specific companies have been shown in charts I and II of chapter IX. Chart I, however, shows the complete picture of interlocks only for the 100 companies with the greatest number of interlocks. The interlocks among the remaining 125 companies, which appear at the top

1 Appendix 12 was prepared by Eleanor Poland.

2 Note that this list differs from that used in Appendix 13 since the latter contains the largest 50 banks and includes no other financial companies.

3 See Appendix 13.

Bankers Trust Co..

First National Bank (Chicago).

Central Hanover Bank & Trust Co.

First National Bank (Boston)

Irving Trust Co..
Manufacturers Trust Co.

Chemical Bank & Trust Co...
Security First National Bank
First National Bank (N. Y.).
Bank of the Manhattan Co.

J. P. Morgan & Co., Drexel & Co.--
Philadelphia National Bank..
New York Trust Co.___
National Bank of Detroit_
Cleveland Trust Co.__
Mellon National Bank....
Union Trust Co. __

Northern Trust Co.___.

Corn Exchange Bank Trust Co.-
American Trust Co..

Wells Fargo Bank & Trust Co..
First National Bank (St. Louis).
Pennsylvania Co. for Insurances, etc..
Anglo-California National Bank
Harris Trust & Savings Bank__

OTHER FINANCIALS

Assets (millions)

2,350. 5

1, 880. 7

1, 847. 4

1,277. 4 1, 141. 1

1, 031. 7

925. 4

914. 8

729. 6

720. 0

673. 0

625. 2

591. O

584. 2

548. 3

537.9

432. 8

419. 7

395.9

337.7

337. 6

334. 5

320. 7

317. 4

271. 8

248. 6

235.5

235. 3

214. 3

207.6