Trends in Manufacturing in the United States
In the 1980s, manufacturing workers enjoyed a positive wage premium and earned more than workers with similar observable characteristics in other industries. In the early 1990s, the premium turned negative, meaning that manufacturing workers now earn less on average than comparable workers in the broader economy. When broken down by education level, a clear divergence in wage premium trends emerges. Excepting the pandemic period, workers with a high school diploma or less have faced negative wage premia beginning in the late 1980s. In contrast, again excepting the pandemic period, college-educated manufacturing workers have consistently received positive wage premia.
The rising role of automation in manufacturing
The manufacturing sector has been shaped, in part, by the growing role of automation in production. One way to track this shift is through the firms’ use of robots, since this use is often used as a proxy for the diffusion of automation (see, for example, Acemoglu and Restrepo, 2020; and Graetz and Michaels, 2018). The International Federation of Robotics reports a sharp increase in robot adoption beginning in the early 1990s. Measured as installations per thousand workers, robot density rose from fewer than 0.5 robots per thousand workers in the 1990s to nearly 30 per thousand by 2023.
Research suggests that this rise in automation has affected both productivity and labor demand. Acemoglu and Restrepo (2020), for instance, show that robot adoption is associated with higher value-added output and labor productivity at the industry level, but lower labor demand in the local labor markets most exposed to robots. Their estimates on the effects of local exposure to robots suggest that each additional robot per thousand workers is associated with a decline of 0.45 percentage points in the employment-to-population ratio in more exposed commuting zones relative to less exposed ones, of which 0.16 percentage points are accounted for by the manufacturing sector. Their paper also documents a 0.77 percent reduction in local hourly and weekly wages.
Evidence from other advanced economies points to similar adjustments within the manufacturing sector. Studying the German labor market, Dauth et al. (2021) show that industrial robots reduce employment in manufacturing industries even though some of these losses are offset by employment gains in other sectors of the economy. Their findings suggest that manufacturing bears the most direct labor-market effects of automation. In Germany, manufacturing job losses are largely offset by gains in services, so that the aggregate employment effect is close to zero.
Taken together, this body of evidence suggests that automation has contributed to reshaping production processes and labor demand within manufacturing.
Conclusion
The evidence reviewed in this Economic Commentary shows that the US manufacturing sector remains large in absolute terms but has undergone a profound transformation in the past few decades. While manufacturing continues to contribute substantially to US output, its shares of GDP and employment have declined amid steady shifts toward services. Once a leading factor in productivity growth, manufacturing TFP growth has lagged behind the broader private economy since the mid-2000s.
The declines in manufacturing’s employment share have stemmed mainly from fewer workers starting out in the sector at the beginning of their careers rather than from workers leaving it later in their careers. At the same time, the manufacturing workforce has grown more educated. The wage premium once associated with manufacturing has been maintained for these more educated workers, but has eroded and is now negative for less-educated workers. Automation and technological adoption appear to have contributed to these shifts. Together, these developments highlight the structural transformation of US manufacturing over the past several decades.
