Automation, Market Concentration, and the Labor Share

Abstract

Since the early 2000s, a rising share of production has been concentrated in a small number of superstar firms. We argue that the rise of automation technologies and the cross-sectional variation of robot use rates have contributed to the increases in industrial concentration. Motivated by empirical evidence, we build a general equilibrium model with heterogeneous firms, endogenous automation decisions, and variable markups. Firms choose between two types of technologies, one uses workers only and the other uses both workers and robots subject to an idiosyncratic fixed cost of robot operation. Larger firms are more profitable and are thus more likely to choose the automation technology. A decline in the cost of robot adoption increases the relative automation usage by large firms, raising their market share of sales. However, the employment share of large firms does not increase as much as the sales share because the expansion of large firms relies more on robots than on workers. Our calibrated model predicts a cross-sectional distribution of automation usage in line with firm-level data. The model also implies that a decline in automation costs reduces the labor income share and raises the average markup, both driven by between-firm reallocation, consistent with empirical evidence.