More Robots, Fewer Jobs

By Mira Rojanasakul and Peter Coy

May 8, 2017

Are you about to be replaced by a robot? The question has broad implications for the U.S. economy, especially the manufacturing sector. Industries that robotize tend to increase output. But robots can have dire consequences for workers.

Two economists recently concluded that both jobs and wages fall in parts of the U.S. where more robots are installed. The March 2017 study by Daron Acemoglu of Massachusetts Institute of Technology and Pascual Restrepo of Boston University shows the commuting zones—i.e., local labor markets—where robot installations have grown the most.

The upper Midwest, particularly Michigan, was ground zero for the robot explosion from 1990 to 2007. That makes sense, since the automobile industry uses more robots than any other. The other hot spots also make sense on closer inspection. In Beaumont, Texas, lots of workers are employed in the plastic, chemicals and pharmaceuticals industry, another big user of robots. Wilmington, Delaware, has a big chunk of workers in that industry and others in car manufacturing, according to Restrepo, one of the researchers.

Those increases tended to mean fewer jobs. Of course, lots of factors weigh on employment. Foreign competition, overvaluation of the dollar and rising productivity all play a big part, too. But even after taking all those other factors into account, Acemoglu and Restrepo still found that additional robots in an area reduces workers and cuts local wages. Their research included any machine that was autonomous and capable of performing multiple tasks. (So: not coffee makers.)

The period covered by the study ended right before the recession of 2007–09. Acemoglu and Restrepo said the recession introduced too many variables that affected employment, which would have made it more difficult to isolate the impact of robots.

So what’s happened since the recession, and what does it mean for manufacturing jobs? We gathered and plotted data from the Bureau of Labor Statistics for various manufacturing sectors to try to get a better sense. Obviously, this type of analysis doesn’t let you draw the direct conclusions Acemoglu and Restrepo were able to make about robots. But you do see some interesting trends when it comes to manufacturing employment.

For starters, U.S. industry still likes robots. Orders for new robots slumped during and just after the recession. Since then, they’ve shot back up.

Automation doesn’t end with robots, however. BLS data shows that “equipment”—machinery, furniture, vehicles and computer software, along with some other components—is increasingly important in production processes compared to labor. Within manufacturing industries, which have historically been the leading edge of robotics usage in production, there were increases across the board.

That seems to coincide with increased productivity in many sectors. Often, when there’s an increase in equipment use, there’s also an increase in output. And output is growing faster than employment, which means that output per worker is rising.

But there are huge disparities between sectors. In some, like the apparel sector, output has fallen along with employment, even as the industry has become more automated. Some of the sectors with the biggest job losses were ones that didn't automate intensively, like textiles and paper products. That goes to show that keeping out robots won't necessarily protect your job.

Bottom line: Robots do replace workers. On the other hand, some industries that don't automate end up losing workers anyway, because their costs are too high and their customers go elsewhere. For workers, robots are only part of the problem.