CONCENTRATION IN BRITISH MANUFACTURING INDUSTRY: A REPLY

Abstract

IN his comment2 on my paper,3 Professor Shepherd reaches four conclusions. I propose to discuss these in turn, and then discuss some of the more general points that he makes. Before this, I will dispose of some of his comments which appear to arise from a misreading of my paper. The Census of Production divides firms into size classes, which vary between industry and between years. The number of firms in the largest size-class also varies; so some adjustment has to be made before the concentration ratios can be calculated. The method I adopted was to make calculations on the maximum and minimum values of the four-firm concentration ratios; the method was described in Appendix III of my paper, and the results given in Appendix II. These calculations were also made for all n-firm concentration ratios, as n varied from 3 to 20. In calculating these limits on the concentration ratio, using a computer, it was necessary to place some upper bound on the unbound size-class. The actual figure used was 1O'?1, so that as far as the results obtained were concerned, it was equivalent to using infinity as the upper limit. The range between the maximum and minimum limits on the concentration ratios was, as expected, in general, smaller for eight-firm ratios than for four-firm ratios. All the regressions were run, using the average of the limits, for both four-firm and eight-firm ratios, and the results were consistent. Shepherd appears to have misunderstood what I was doing. So, when he says (p. 432) that 'in 73 of 234 cases in the two years, it appears that no upper bound estimate (of four-firm concentration ratio) is made', he has not realized that no entry is made in the upper bound estimate column when the four-firm ratio is known exactly, as is explained in the notes attached to Appendix II. It should be noted that these calculations do give a maximum value to the concentration ratio, and not 'an estimate '. I just do not understand his comment (p. 432) that 'in 37 of 117 cases in 1963, (see his (i.e. my) Appendix II), largest-firm-size data were lumped together for eight or more firms'. I now come to the first of Shepherd's conclusions (p. 436) that 'British industrial concentration is rather different from what he suggests, especially compared with U.S. patterns'. This conclusion is based on a comparison of his own work (Shepherd, 1972b) and my own. Clearly the acceptability of this conclusion depends upon the relative validity of his 1 I am grateful to Sam Aaronovitch for discussion on a number of the points raised in this reply; the usual disclaimer applies. 2 Shepherd (1972a). 3 Sawyer (1971).