Are factor biases and substitution identifiable? The Canadian evidence

Revised productivity accounts recently released by Statistics Canada are used to estimate a KlumpMcAdamWillman (KMW) normalized CES supply-side system for the half-century 19612012. The model permits distinct rates of factor-augmenting technical change for capital and labour that distinguish between short-term versus long-term effects, as well as a non-unitary elasticity of substitution and time-varying factor shares. The advantage of the Canadian data for this purpose is that they provide a unified treatment of measurement issues that have had to be improvised in the US and European data used by previous researchers. In contrast to previous results, we find that an elasticity of substitution and distinct factor biases of technological progress are not well determined by the model. For the Canadian data, the KMW model does not appear to provide a framework that overcomes the classic DiamondMcFaddenRodriguez non-identification result. That impossibility theorem is manifested in our findings, not overcome by them.