The Translog cost function extimation approach has been used in this paper, to estimate important industrial policy parameters. The translog approach differs from the more frequently used techniques viz, the generalized Cobb-Dauglas production function (C0D) approach and the Constant Elasticity of Substitution (CES or ACMS) approach, in that restrictive assumptions about the elasticity of substitution between factors do not have to be made. The translog approach also has the additional advantage of providing cost shares for each factor, pair-wise elasticity of substitution, and own and cross price elasticities for the inputs. Thus, a whole range of policy parameters can be simultaneously estimated.

While estimates of elasticity of substitution between capital and labour, and capital and materials are not significantly different from one, the parameter estimate is significantly below unity for substitution between labour and materials. The cross price elasticities show that the three inputs capital, labour and materials are complements in production. The own price elasticities have the appropriate signs and are all significantly below one. As expected the cost share of raw materials is the largest (78.9%) followed by labour’s share (14.6%) and capital’s share (6.5%). Alternative hypotheses regarding scale economies have also been tested. Data from the manufacturing sector of Bangladesh for the period 1969/70 to 1978/79 have been used.