This study has two aspects. First, the author examines the theoretical properties of the constant elasticity of substitution (CES) production function and the implications of this formulation for the properties of a structural macroeconomic model. He then seeks to determine whether Canadian macroeconomic data correlate better with a CES production function with an elasticity of substitution between labour and capital equal to one, which would be the case with a Cobb- Douglas function, or with a CES function whose elasticity of substitution is different from one.

Cobb-Douglas-type production functions have some very attractive properties, which is probably why they are so widely used in macroeconomic models. Referring to results from previous studies, the author demonstrates that it is possible to retain these properties when using a CES production function with an elasticity of substitution different from one, provided it features constant returns to scale and that technological progress only increases the efficiency of the labour factor.

In terms of empirical analysis, the estimation frameworks used in this study and applied to Canadian macroeconomic data yield an elasticity of substitution of capital for labour lying between 0.4 and 0.6, or well below one. Most of the tests reject use of the Cobb-Douglas formulation for representing Canadian data. These results suggest that capital and labour are much more complementary than is assumed by a Cobb-Douglas production function.