Not a Typical Firm: The Joint Dynamics of Firms, Labor Shares, and Capital–Labor Substitution

Abstract

While the US labor share has declined, especially in manufacturing and retail, the labor share of a typical firm has increased. This paper introduces a model where firms incur fixed costs to automate tasks. In response to lower capital prices, the model reproduces the labor share dynamics observed in the data: large firms automate more tasks, reducing the aggregate labor share; while the median firm continues to operate a labor-intensive technology with a rising labor share. Using our model, we decompose the labor share decline and the rise in concentration into a part driven by lower capital prices and a part driven by sales reallocation to more productive higher-markup firms. Reallocation played a minor role in explaining the labor share decline in manufacturing but an important role in retail and other sectors during 1982-2012. These conclusions are in line with estimates of markups and capital elasticities from Compustat.

New Evidence on the Effect of Technology on Employment and Skill Demand

“…The technology adoption events in our data are almost entirely product rather than process-type changes. But both types may occur in reality, and some studies report examples of the latter when it comes to automation (e.g., Acemoglu and Restrepo 2020;Restrepo and Hubmer 2021). We explain next why our findings are distinctive but logical-and applicable to other settings where similar incentives for process vs. product type technology adoption prevail.…”

“…Manufacturing subsidies are widespread (see, e.g., Gruber and 63 These reasons include: 1) externalities, e.g., in the product market, the intermediate input market, the factor market, or due to technological externalities, 2) compositional effects, e.g., through expansion and contraction of firms and industries, 3) technologies creating new areas in the economy, e.g., video games, the Apollo program, or Google, and 4) technologies directly inducing macro-level changes, e.g., self-booking platforms displacing travel agents, the internet changing job search, or technologies inducing broad organizational and cultural changes. The papers addressing externalities and compositional effects include Acemoglu et al (2020b); Aghion et al (2020); Humlum (2019); Koch et al (2021); Restrepo and Hubmer (2021), and Oberfield and Raval (2021).…”

Frictions and the diffusion of automation

“…Following Acemoglu and Restrepo (2018b), labor productivity is given by γ N (j) = e χj , for χ > 0 capturing the growth rate in labor productivity from one task to the subsequent one. Hubmer and Restrepo (2022) and Guerreiro et al (2021) assume similar functions. Acemoglu and Ziliboti (2001) and Acemoglu and Autor (2011)…”

Robots, Exports and Top Income Inequality Evidence for the U.S.

“…1 places big companies one step ahead. Indeed, existing evidence shows that the adoption of automation technologies concentrates on very large firms (Acemoglu, Lelarge, and Restrepo, 2020;Acemoglu et al, 2022), that access to export markets stimulates productivity-enhancing investments within firms (Aw, Roberts, and Xu, 2011;Koch, Manuylov, and Smolka, 2021) and that the production processes of the largest manufacturing firms have become more capital intensive over the last decades (Hubmer and Restrepo, 2022).…”