Biases in empirical estimates of the elasticities of substitution of C.E.S. production functions *

 1. Already well-born, the C.E.S. production function made its debut to a wide audience in an article by Arrow, Chenery, Minhas and Solow (1963). Minhas subsequently renamed the function and used it for further analysis in his book (Minhas, 1963). ¹ The particular empirical findings which led to its debut were the close association, as confirmed by the appropriate regressions, between the logarithms of labour productivity (value added per unit of labour used) and money wage rates in the same industries in different countries. (See Arrow et al, 1963, pp. 227–8; Minhas, 1963, chapter 3.) If the values added and labour inputs used in the regression analysis are assumed to be observations from a C.E.S. production function, the regression coefficient, say b, can be shown to be an estimate of the elasticity of substitution between capital and labour. (See Arrow et al., 1963, pp. 228–9; Minhas, 1963, chapter 2.) It was shown in both studies that the value of b was usually less than one and differed as between industries.