Human beings have a natural fear of being replaced by machines. Folk tales like the legend of John Henry, the mighty railroad worker who out-hammered a steam engine and then died of exhaustion, are forever depicting the rise of technology as a desperate battle by humanity against obsolescence. In the early 1800s, a disorganized movement of British workers known as Luddites expressed the revolutionary sentiment of the day by smashing industrial machinery.
This fear is natural, and it’s far from crazy. But it’s easy to overdo it. In a recent op-ed in The New York Times, Thomas Edsall suggested industrial robots were an important factor behind the election of President Donald Trump. Edsall draws a causal line between robots, job loss in the industrial Midwest and the shift of white, working-class voters from Democrats to the Republicans. It’s a compelling idea, but there are reasons to be skeptical.
First of all, there’s the question “Why here?” In terms of industrial robots, the U.S. is a laggard — as of 2015, Asian and European nations were way ahead.
If political unrest is being caused by robots displacing manufacturing workers, why is it that the U.S. and U.K., two countries with relatively low numbers of robots, saw the biggest populist upheavals in 2016?
Proponents of Edsall’s thesis might retort that even if robots aren’t taking over the U.S., they’ve had an outsized impact in Michigan, Ohio, Wisconsin, Indiana and western Pennsylvania, where Trump benefited from big Republican gains and flipped crucial Democratic-voting states.
The problem is that the Rust Belt has also seen so many other big changes lately. As a manufacturing-dependent region, it took the brunt of Chinese competition during the 2000s. Economists David Autor, David Dorn, Gordon Hanson and Kaveh Majlesi have linked Chinese imports to a shift toward Republicans in majority-white industrial regions. The opioid crisis has also hit Michigan, Ohio and western Pennsylvania hard, and studies have found that places most affected by opioid abuse tended to support Trump.
The Midwest has also been undergoing rapid demographic change — although the region has few immigrants, the number has increased recently. As Edsall himself has reported, polls show that fear of immigrants was a strong motivating factor behind support for Trump. A Wall Street Journal analysis found that a rapid increase in diversity spurred small Midwestern towns to abandon the Democrats in 2016. Political science studies have found that fear of increasing racial diversity was strongly correlated with the likelihood of vote-switching from Democrats to the GOP in 2016.
It’s still possible that robots contributed to a long-term economic decline that made Midwesterners more vulnerable to the scourge of opiates and the lure of xenophobia. But here, too, there is cause for skepticism. Edsall’s thesis relies on research by economists Daron Acemoglu and Pascual Restrepo that links robots to job losses and wage declines. But as Lawrence Mishel and Josh Bivens of the Economic Policy Institute have shown, the conclusions are far more ambiguous than many journalists have been reporting.
Industrial robots represent only one type of automation. Acemoglu and Restrepo find that information-technology investment has a positive effect on employment. In other words, even if robots replace human workers, IT overall tends to complement humans. The manufacturing industry alone spends tens of billions of dollars on information technology each month:
That’s a lot more than the sector spends on robots. Someday, robots may take a larger share, but for now, it looks like most technology investment has a benign effect on workers.
As Mishel and Bivens point out, estimates by Acemoglu and Restrepo imply that the effect of Chinese competition on U.S. manufacturing-job losses has been three times the effect of robots. So even researchers who are alarmed about robots think that so far, trade has been a much bigger shock to U.S. workers.
Robots are, at most, a tiny part of the Trump story. When looking for long-term factors behind the Midwest’s shift toward Trump, we should first think about trade with China, the opioid epidemic and fear of demographic change. The Trump movement was no Luddite rebellion.
Noah Smith is a Bloomberg View columnist.