Automation, Human Task Innovation, and Labor Share: Unveiling the Role of Elasticity of Substitution

This paper investigates the elements contributing to the change in labor share, with a specific focus on the roles of ‘automation’ and ‘innovation in human tasks.’ We construct a general equilibrium model that distinctly incorporates both robot and non-robot capital to derive an econometric specification. Using task data from O*NET and employing the most recently developed sentence embedding tools to match tasks and patents, we construct a novel ‘innovation in human tasks’ variable for multiple countries. This allows us to empirically evaluate the impact of innovation in human tasks on labor share across countries for the first time in the literature. Our accounting analysis suggests that the positive influence of human task innovation outweighs the adverse effects of automation in most of countries we study. From our regression analysis, we estimate the elasticity of substitution between labor and non-robot capital to be less than one, while the elasticity of substitution between tasks is greater than one. With these estimates, we elucidate the direct and indirect effects of automation and innovation in human tasks on labor share.