APPENDIX 5

MATERIAL RELATING TO THE TESTIMONY OF

DR. WILLIAM G. SHEPHERD

TABLE 1.-Changes in the share of industry value-added in the largest size plants-Selected industries, 1947 to 1958

TABLE 2.-Changes in the share of the largest size plans—2-digit industry groups

1 Industries in group 19 (ordnance) were included in the 1958 census data, and so the patterns for 1947 and 1958 are not meaningfully comparable.

TABLE 3.-The main industries with shifts toward larger plant sizes

NOTE.-Figures in parentheses are estimated from the patterns in the 2-digit industry groups.

EXCERPTS FROM: "CHANGES IN BRITISH INDUSTRIAL CONCENTRATION, 1951-58"* (By William G. Shepherd)

Studies of concentration in several countries have found little tendency for concentration to increase in the postwar period, or indeed since the turn of the century. In at least one case, there may well have been a decline.1 Furthermore, a number of downward pressures on concentration have recently been talked of, such as industry growth. Moreover, many economists argue that a broadening of markets and interindustry competition has been bringing about a genuine reduction in monopoly power and behavior.

The evidence in this paper suggests that, by contrast, industrial concentration in Britain has increased during the 1950's. This shift has been far from uniform, and its causes cannot be pinpointed thoroughly in the space of this paper. The upward trend nonetheless has implications for policies affecting market competition.

I

Data on concentration in the top three firms in each industry are available for 1951 and 1958.3 These are arrayed in table 1 to show the distribution of industrial activity (shown by net output and employment) according to the degree of threefirm concentration. For example, in 1951, in 10 industries accounting for 3.6 percent of manufacturing net output and 2.8 percent of employment, the 3 top firms had 70 to 79 percent of net output. In 1958, there were 5 industries in the same 70 to 79 concentration decile, but they now accounted for 7.5 percent of net output and 6.6 percent of employment.

The Board of Trade's industry definitions changed somewhat during the period, so that the manufacturing sector was divided into 139 separate "industries" or trades in 1951 and only 120 in 1958. The industries were in 1958 therefore more broadly drawn than in 1951. For obvious and well known reasons, the measured

1 See my "Trends of Concentration in American Manufacturing Industry, 1947-58." Review of Economics and Statistics, May 1964, pp. 200-212, and the references cited there. On Britain, see especially R. Evely and I. M. D. Little, "Concentration in British Industry," Cambridge University Press, 1960; P. E. Hart and S. J. Prais, "The Analysis of Business Concentration: A Statistical Approach," Journal of the Royal Statistical Society, series A, 1956, pp. 150-191; and my "A Comparison of Industrial Concentration in the United States and Britain," Review of Economics and Statistics, 1961, pp. 70-75. E. Rotwein offers a new, reduced estimate of the extent of concentration and monopoly power in Japan in his "Economic Concentration and Monopoly in Japan," Journal of Political Economics. 1964, pp. 262–277.

2 See Shepherd. "Trends of Concentration in American Manufacturing Industry." op. cit.. pp. 203-209, and Ralph Nelson, "Concentration in American Industry," Yale University Press, 1963, pp. 50-56.

To appear in Oxford Economic Papers, October 1965.

I am indebted to the Institute of Public Administration and to the Ford Foundation for supporting the research for this paper, and to Shorey Peterson and John Lansing for their comments on an early draft.

For 1951, in R. Evely and I. M. D. Little, op. cit.; for 1958 in the Board of Trade, "The Report on the Census of Production for 1958," pt. 133, summary table 5, London, Her Majesty's Stationery Office, 1962.

degree of concentration increases with the fineness of industry definition. Accordingly, the 1958 data undoubtedly understate-across the board-1958 concentration compared to 1951 concentration. It is not possible to measure the degree of this understatement, although it could possibly range above 10 percent on average. In any event this factor further strengthens the conclusion that concentra. tion generally increased during the period.

TABLE 1.—United Kingdom: Distribution of industries according to concentration of net output in the 3 largest firms, 1951 and 1958

Those who have watched the increasing concentration in the airplane and motorcar industries will not be surprised at the general upward trend. Indeed these two industries account for much of the overall increase, though not by any means all of it. Tables 1 and 2 bring out the extent of the shift. Thus (even apart from the understatement caused by broader industry definitions in 1958), the share of the net output from industries concentrated more than 60 percent rose from 6.5 to 16.8 percent, above 80 trom 2 to 6.4 percent.

Of the 120 industries listed for 1958, a large proportion (47) could not be matched with industries in the 1951 list. Of the 73 industries which could be matched, no less than 49 showed increased concentration, while 13 showed decreases and 11 had no change. This suggests a fairly broad upward trend, though one would like to be able also to evaluate the 47 unmatched industries. All matched industries showing large increases or decreases of concentration (by more than 5 points) are listed in table 2. The breadth and variety of the upward shift, and its predominance over the few downward changes, can be seen clearly. The rises took place in industries in all ranges of concentration, not just in those with low concentration. Conversely, the few declines suggest that there has been no strong tendency for concentration in oligopolistic, highconcentration industries to decline.

Table 2 also shows the 1935–51 change in concentration for most of the industries, for comparison with the 1951-58 shifts. Though the pattern is not distinct, there was some tendency for the changes to reverse from one period to the next.

The most one can say is that almost no industries continued a downward trend in concentration from 1935 to 1958; though many continued an upward trend. By main industry sectors, rising concentration is most evident in food, drink, and tobacco (rises in 11, declines in 1, same or unknown in 3 industries), in vehicles and in some chemicals, textiles, and engineering industries. Also there has been increasing concentration in newspapers and other publishing trades. In no sector have the declines outnumbered the increases, although the unmatched industries may include a variety of changes.

This is clearly shown in the trade and subtrade data in R. Evely and I. M. D. Little, op. cit., pp. 296-312, and in American four and five digit standard industrial classification data.

Two adjustments were made in deriving table 1. In both years, concentration in a number of industries was given by four, five, six, or eight firms ratios, rather than the usual three firms. These were adjusted down to estimated 3-firm ratios by the use of approximate proportions for the four and higher firms. In almost all cases there was little doubt into which decile the industry should be put. The other adjustment placed aircraft in the top decile, to reflect its amalgamation in 1959.

Similar persistence of concentration in oligopolistic industries in the United States is shown in Shepherd, "Trends of Concentration in American Manufacturing Industry, 1947-58," op. cit., pp. 209-212.

TABLE 2.-Large changes of concentration 1951-58 compared with changes 1935-51 in industries which could be matched

1 In some cases, the published ratios are based on more than 3 firms. For these cases the figures given here (in parentheses) are estimates.

2 Not available.

NOTE.-Total net output of those whose concentration during 1951-58:

There are several possible causes of the general shift in concentration, and we may now consider several of them. The optimum size of firms may have increased. Highly concentrated industries may simply have grown more rapidly than the rest. Or, since rapid growth is said to cause a decline in concentration, low or negative industrial growth rates may explain the rises in concentration. Policies against restrictive practices may have encouraged merger and concentration, in place of joint action by independent firms. Each of these (and other) possible explanations leads in turn to different conclusions for public policy. About optimum sizes of firms there continues to be little reliable evidence." Undoubtedly optimum size has increased in many industries since 1951, and this has contributed to the rise in concentration. Evely and Little mention it prominently as a cause of rising concentration in numerous industries during 1935–51. But many concentration rises have no connection with optimum size changes at all. Many other industries have already been concentrated well beyond what

7 The main source of information and methodological discussions are J. S. Bain, "Barriers to New Competition," Harvard University Press, 1956; T. R. Saving, "Estimation of Optimum Size by the Survivor Technique," Quarterly Journal of Economics, 1961, pp. 569-607; and, concerning British industries, P. J. D. Wiles, "Price Cost and Output,' Oxford. Blackwell, 1961.

8 R. Evely and I. M. D. Little, op. cit., pp. 181-183.