Directed Technological Change: A Quantitative Analysis

We explore the role of skill non-neutral productivity differences and barriers to technology adoption in explaining cross-country income differences. Using a model of directed technological change modied to include physical capital, technology diffusion and barriers to entry, and a new data set on output and labor force composition we construct measures of productivity by worker skill type for a large cross-section of countries going back as far as the 19th century in some cases. Additionally, the equilibrium conditions of the model allow us to back out the measure of barriers to adoption of technology. We use these measurements to (1) study the historical patterns of directed technological change and (2) evaluate the contribution of non-neutral technology and barriers to cross-country income differences. We find that allowing for non-neutrality of productivity increases the importance of human capital differences in explaining income variation across countries. Additionally, we find that barriers to adoption of technology are important for understanding productivity differences but this importance has been declining in recent decades. Importantly, in the presence of technology diffusion, barriers are only important if the elasticity of substitution between skill types is well above 2. We argue that other predictions of the model, especially about the cross-country wage distributions, are also consistent with data only for large values of elasticity of substitution. We provide some auxiliary empirical evidence using EU KLEMS data which suggest that the elasticity of substitution between skill types may indeed be considerably greater than 2. Finally, we find that historical patterns of skill bias in technological change which we measure are largely consistent with priors.