A Note on the Empirical Measurement of Vertical Integration

examine some determinants of trends in vertical integration across both industries and firms, using the ratio of value added to sales as an index of vertical integration. In their industry-level analysis, T & W calculate this index from Census four-digit SIC industry data. The purpose of this note is to explain why Census data are inappropriate for this use. It is helpful first to review the theoretical rationale for the use of the ratio of value added to sales (VA/S) as an index of vertical integration. As stated byT & W: ... value added may be viewed as the difference between sales and purchased material inputs (i.e. inputs other than labor and capital). Hence, for a given firm or industry, backward integration will tend to reduce the purchases of material inputs while leaving sales of final outputs constant, with a resulting increase in the ratio of value added to sales. Similarly in forward vertical integration, sales will tend to increase more than proportionally to purchased material inputs, also resulting in an increase in the ratio of value added to sales [4, pp. 82-3]. Thus, if other economic factors affecting VA/S are constant, an increase (decrease) in VA/S implies an increase (decrease) in the extent of vertical integration. But, since vertical integration occurs within a company, the ability of