Working Paper: NBER ID: w24871
Authors: David Autor; Anna Salomons
Abstract: Many technological innovations replace workers with machines, but this capital-labor substitution need not reduce aggregate labor demand because it simultaneously induces four countervailing responses: own-industry output effects; cross-industry input–output effects; between-industry shifts; and final demand effects. We quantify these channels using four decades of harmonized cross-country and industry data, where we measure automation as industry-level movements in total factor productivity (TFP) that are common across countries. We find that automation displaces employment and reduces labor's share of value-added in the industries in which it originates (a direct effect). In the case of employment, these own-industry losses are reversed by indirect gains in customer industries and induced increases in aggregate demand. By contrast, own-industry labor share losses are not recouped elsewhere. Our framework can account for a substantial fraction of the reallocation of employment across industries and the aggregate fall in the labor share over the last three decades. It does not, however, explain why the labor share fell more rapidly during the 2000s
Keywords: Automation; Labor Displacement; Productivity Growth; Labor Share
JEL Codes: D33; J23; O33; O57
Edges that are evidenced by causal inference methods are in orange, and the rest are in light blue.
| Cause | Effect |
|---|---|
| Total Factor Productivity (TFP) growth (O49) | Employment decline (J63) |
| Total Factor Productivity (TFP) growth (O49) | Labor share decline (J39) |
| Total Factor Productivity (TFP) growth (O49) | Employment gains in customer industries (L89) |
| Productivity growth (O49) | Labor demand across all sectors (J23) |
| Uneven productivity growth across industries (O49) | Aggregate labor share (E25) |