of 3,752 individual companies appeared as one of the leading 4 producers of at least 1 of the 1,807 products included in the study. About three-fifths of these companies were classified by the Bureau of the Census as independents or single-plant companies, while the remaining concerns were central-office companies or multiple-plant enterprises. Although the number of central-office companies made up only 3.8 percent of all concerns reporting to the Bureau of the Census, these multiunit concerns accounted for 38 percent of the total number of companies appearing as leading manufacturers of products covered in T. N. E. C. Monograph No. 27.

One company 14 appeared among the leading 4 producers of 99 different products, a second company made 82 appearances, and a third made 72 appearances. By far the greater proportion of the companies appeared as leading producers in only a few products. There were 2,656 companies which manufactured a sufficient value of an individual product to place them among the leading 4 producers of that item. Approximately a fifth of the leading producers made two or three appearances, there being 547 companies and 222 companies at these levels, respectively.

There was a marked tendency for a company making many appearances to have those appearances in first or second place, while those companies making only a few appearances tended to have those appearances in third or fourth place. Of the 2,656 companies which appeared as one of the leading producers of a single product, 488 were in first place, and 807 in fourth place. At the opposite extreme the 1 company appearing as a leader in the manufacture of 99 products made 65 of its appearances in first place and only 5 in fourth place.

Since central-office companies tended to be the largest producers,15 a considerably larger number of first-place appearances were made by them than by independent concerns at each appearance level with the exception of the 1-appearance level. At the 1-appearance level, 35 percent of the total number of company-appearances were made by central-office companies and 65 percent by independents; at the 11appearance level, however, the comparable ratios were 86 and 14 percent respectively.

Relation of concentration to various product characteristics.-That a large number of products are manufactured under conditions of relatively high concentration has already been emphasized. A further step in the inquiry is concerned with the character of the commodities that are produced under such circumstances and the varying degrees of control associated with different economic attributes. 16 In order that these relationships might be appraised, the 1,807 census products which formed the nucleus of the study of product concentration were classified on the basis of (1) type of immediate purchaser (producer or consumer); (2) type of ultimate user (producer or consumer); (3) degree of durability (nondurable, semidurable, or durable); (4) degree of fabrication (semimanufactured or finished); (5) type of market

14 This company actually manufactured 302 census products but only 142 of them were included in the 1,807 items analyzed in the study.

15 Central-office companies represented 3.8 percent of the number of manufacturing concerns in 1937, but they accounted for 61.1 percent of the value of all manufactured products (see Temporary National Economic Committee Monograph No. 27, Part II, The Integration of Manufacturing Operations).

16 For a fuller discussion of this subject and an explanation of the methods of classification of products, see ibid, Part V, ch. III, and appendix D.

(national or regional); (6) source of raw material (agriculture, mines, or forests); (7) construction materials; and (8) producers' supplies. Certain general observations predicated upon the foregoing type of analysis may be made. Products purchased by producers tended to be manufactured under conditions of appreciably higher concentration than goods purchased by consumers. As a matter of fact, in 1937 approximately 80 percent of the number of items destined to be purchased by producers had concentration ratios above 50 percent, whereas only 66 percent of the consumer items were in these high concentration brackets. In terms of value, however, the concentration patterns for the two classes of products were of the same general conformation, and in both cases the more important products valuewise had relatively lower concentration ratios. Products to be used ultimately by producers were likewise manufactured under conditions of considerably higher concentration of control than goods to be used ultimately by

consumers.

At this point it is important to realize that these broad generalizations obscure the varying degrees of concentration among commodities which represent the output of different types of industries but which otherwise possess the same economic characteristics. For example, only 9 percent of the value of tomatoes canned in 1937 was accounted for by the leading four producers of this product, whereas the leading four producers of passenger-car tires manufactured 77 percent of the total value of such goods. Both of these products are consumer items but are products of distinctly different industrial groups and have distinctly different marketing patterns.

Concentration in production varies with the degree of durability of the product. Goods whose use extends over a number of years were produced under conditions of higher concentration than the semidurable and nondurable products, the levels of concentration tapering off as the products are less durable in nature. The durable goods in the high concentration ranges consist largely of items which are purchased by producers.

No striking differences are apparent in the concentration patterns of semimanufactured and finished goods. If raw materials were taken into account, considerable variations would undoubtedly be noted. This analysis, however, relates only to products which have undergone transformation in one or more manufacturing processes.

With reference to the relation of type of market and to concentration, it is apparent that various factors tend to fix market limits for certain commodities. Perishable goods or goods which cannot be transported with facility cannot be marketed on a national scale. In the case of a product marketed regionally, the output may be highly controlled, but this fact would not be revealed in concentration calculations based upon Nation-wide production data. The drawing of inferences with respect to the type of market in relation to concentration is, therefore, hardly justified in these summary findings. Despite the understatement of actual market concentration, however, well over a third of the 231 "regional" products included in the basic study of product concentration had national concentration ratios greater than 60 percent.

When attention is focused upon the source of raw material entering the manufactured goods, quite different patterns of concentration are evident. Products derived from agricultural sources tended to bunch

heavily in the low concentration classes, whereas those produced from minerals had high concentration ratios. Products which were predominantly manufactured from forest materials were more evenly distributed among the concentration classes. Fifty-seven percent of the value of products from agricultural sources and only 31 percent of the products fabricated from minerals had concentration ratios in 1937 of less than 50 percent.

Construction materials and producers' supplies exhibited somewhat similar concentration patterns, both in terms of number and of value. In each case most of the items were produced under conditions of relatively high concentration, although for both types of products the items produced under high concentration were somewhat less important on the basis of value than those produced under lesser degrees of concentration. It should be kept in mind that construction materials are distributed in considerable measure in regional markets. Thus, the limitations mentioned in connection with the distributions of products by type of market apply in this instance as well.

The concentration pattern in 1935 and in 1937.-A particular phase of the study of concentration centers upon changes in the pattern of the distribution of control from one period to another. In order that further light might be thrown on this subject an investigation was made of changes in concentration, in quantity produced, and in the average realized price of products from 1935 to 1937.1 To what extent did the concentration picture change over the period? If shifts in concentration ratios of products occurred, were the increases or decreases in certain types of products more pronounced than in others, or, again, were the changes in certain lines of activity more significant than in others? Were changes in concentration paralleled by variations in quantity produced and by changes in average realized prices?

A study of changes in the concentration ratios for 392 products which were identical in 1935 and 1937 provides factual information on the actual courses of concentration in the 2-year period. The overall picture of the shifts in concentration experienced by these products between 1935 and 1937 indicates no "wholesale" movement toward an increase or decrease in concentration ratios. The pattern of change is shown in chart VIII. Almost two-thirds of the products showed "no change" or an increase or decrease of less than 10 percent in their concentration ratios. Approximately 8 percent of the products registered more than a 20-percent increase and about 7 percent experienced a decrease of more than 20 percent. Thus it may be said that the changes in concentration ratios are essentially random in nature. This was certainly true for the particular interval under study, but it probably holds true for longer periods of time. The reasons underlying this position center in the fact that each product appears to have its own particular set of distinct and separate factors which determine its behavior. Many of these factors may be formulated in general terms and may be applied to all sections of the manufacturing process. Nevertheless, they affect each product at different times in varying degrees, so that for a given period they form a unique set of conditions for each product or for a group of closely related products. Products of like nature did not exhibit a similar degree or direction

17 For basic data, see ibid., Part V, chart 14, and appendix D, table 8D. 18 Ibid., Part V, ch. IV.

of change in their concentration ratios between 1935 and 1937. An examination of a selected list of products reveals a wide variety of changes among products in the same industry, among competing products, among products fulfilling the same types of wants, among products purchased by the same group of users, etc. Further more, products which increased in concentration did not show materially different characteristics from those products which decreased in concentration.

CHART VIII.-Relation between concentration ratios in 1935 and 1937.

When attention is directed toward the various characteristics of products, it is apparent that products of the same type experienced both increases and decreases in concentration of about the same degree. Thus it cannot be said, for example, that consumers' goods tended to increase in concentration while producers' goods tended to decrease or vice versa. Nor did products, which in many cases were manufactured by the same companies, show consistent changes in their concentration ratios.

All these investigations point to the fact that over the 1935-37 period, during which there was a marked increase in output of manufactured products, the distribution of the changes in concentration was symmetrical. There is, therefore, no indicated conclusion that

reas

the product structure of manufacturing was subject to wide shifts over the short run. Furthermore, the changes which did occur were random in nature.

An investigation of changes in concentration and changes in the volume of output and in the average realized price 19 of products from L 1935 to 1937 reveals no significant correlation between the variables. Products for which the quantity produced increased or decreased were equally likely to have been products for which the percentage of supply controlled by the largest four companies either increased or decreased. Likewise, the increases and decreases in average realized price over the 2-year period were evenly divided between products with positive and negative changes in concentration. This randomness may be due to the shortness of the period under study. The Concentration of Production in Mining.

The discussion of product concentration up to this point has been concerned solely with manufactured products. Production in the field of mining is likewise characterized by a high degree of concentration. This topic will be commented upon here only briefly, since basic data relating to the subject are of a more or less fragmentary nature. The available material is sufficient, however, to justify the observation that in many lines mineral output is predominantly controlled by a relatively small number of producers.20 The statistics quoted below are, for the most part, in terms of the proportion of the total value of output of a specific mineral accounted for by the leading four producers of that mineral. Thus, they are comparable to the concentration ratios widely used in the preceding analysis of manufactured products. They relate to the year 1935, however, and not to 1937.

Among the products of mines for which concentration data are readily available, potash and bauxite give evidence of the highest degree of concentration. In the case of potash, four producers accounted for the entire output in 1935. More than one-half the supply for domestic use in that year, however, came from imports. In the same year, the leading three producers of bauxite were responsible for 93 percent of the total value output of that product. The production of ingot aluminum, however, was even more concentrated since the reduction of bauxite to aluminum was under the control of a single company. Other minerals having a high degree of concentration in 1935 were marble with a concentration ratio of 84 percent; gypsum, 80 percent; copper, 78 percent; mercury, 71 percent; and iron ore, 64 percent. Low concentration was apparent in the production of sand and gravel, limestone, bituminous coal, basalt, and granite, the concentration ratios for these items ranging from 10 to 18 percent.

The importance of the mineral fuels, petroleum and coal, is indicated by the fact that in recent years they have constituted approximately half of the total value of mineral production in the United States. In the production of these important items, concentration

19 The nature and limitations of the average-realized-price measure are set forth in detail in ibid., Part V, ch. IV.

20 Basic concentration data for selected mineral products, according to reports prepared for the Temporary National Economic Committee by the Bureau of Mines, are presented in ibid., Part V, Appendix F.