Harvard Business Review January-February 197 Exhibit Ill Effect of vertical integration on investment/sales ratio Vertical Market share Under 10% 30%-40% Over 40% High 男m ExhibitⅣv Purchase-to-sales ratio corrected for vertical integration Market share Over 40% 28 46 On the surface then, higher investment turnover on sales exhibits a strong, smooth, upward trend as does not appear to be a major factor contributing to market share increases higher rates of return. However, this observation is subject to some qualification. Our analysis of the Why do profit margins on sales increase so sharply pims data base shows that investment intensity (in- with market share? To answer this, it is necessary vestment relative to sales tends to vary directly with to look in more detail at differences in prices and a business's degree of vertical integration. operating expenses (The degree of vertical integration is measured as 2 the ratio of the total value added by the business to The biggest single difference in costs, as related to its sales. Both the numerator and denominator of market she n the purchases-to-sales ratio. As the ratio are adjusted by subtracting the pretax in- shown in IL, for large-share businesses-those come and adding the PIMS average ROI, multiplied with shares over 40%-purchases represent only 33% by the investment. J of sales, compared with 45% for businesses with shares under Io% Vertical integration thus has a strong negative rela tion to the ratio of purchases to sales. Since high How can we explain the decline in the ratio of pur- market-share businesses are on the average some- chases to sales as share goes up? One possibility, as what more vertically integrated than those with mentioned earlier, is that high-share businesses tend smaller shares, it is likely that investment turnover to be more vertically integrated-they"make/rather increases somewhat more with market share thanthan"buy, and often they own their own distribu- the figures in Exhibit II suggest. In other words, as tion facilities. The decline in the purchases-to-sales shown in Exhibit Ill, for a given degree of vertical ratio is quite a bit less see Exhibit IV) if we control integration, the investment-to-sales ratio declines for the level of vertical integration. A low purchases significantly, even though overall averages do not. to-sales ratio goes hand in hand with a high level of vertical integration Nevertheless, Exhibit ll shows that the major reason for the ROI/market-share relationship is the dra- Other things being equal, a greater extent of vertical matic difference in pretax profit margins on sales. integration ought to result in a rising level of manu- Businesses with market shares under Io% had aver- facturing costs. For the nonmanufacturing business- age pretax losses of oI6%. The average ROI for busi- es in the PIMS sample, "manufacturing" was defined nesses with under Io% market share was about 9%. as the primary value-creating activity of the busi Obviously, no individual business can have a nega. ness. For example, processing transactions is the tive profit-to-sales ratio and still earn a positive ROI. equivalent of manufacturing in a bank. But the The apparent inconsistency between the averages data in Exhibit II show little or no connection be- reflects the fact that some businesses in the sample tween manufacturing expense, as a percentage of incurred losses that were very high in relation to sales, and market share. This could be because, de sales but that were much smaller in relation to pite the increase in vertical integration, costs are investment In the PIMS sample, the average return offset by increased efficiency