Separability and capital aggregation in sectoral models of US production

Separability plays a fundamental role in applied studies of production inasmuch as it provides for consistent aggregation and allows multi-stage estimation of models with many inputs. This paper examines whether two types of capital, structures and equipment, are weakly separable from labour and energy and materials in sectoral models of US production, and provides an assessment of the effects of maintaining capital separability. The study focuses on the elasticity of substitution as a measure of input association and associated standard errors and confidence intervals based on bootstrapping. Elasticity estimates and tests indicate that capital separability is generally inconsistent with data representing sectoral US production. Elasticity sign, however, appears robust to capital aggregation and performs especially well as an indicator of input association.