Economists have long argued that automation, not trade, is responsible for the bulk of the six million jobs shed by the manufacturing sector over the last 25 years. Now, they have a put a precise figure on some of the losses.
The number is stunning on the face of it, and many have interpreted the study as an indictment of technological change — a sign that “robots are winning the race for American jobs.” But the bigger takeaway is that the nation has been ill-equipped to deal with the upheaval caused by automation.
The researchers estimate that half of the job losses resulted from robots directly replacing workers. The rest of the jobs disappeared from elsewhere in the local community. It seems that after a factory sheds workers, that economic pain reverberates, triggering further unemployment at, say, the grocery store or the neighborhood car dealership.
In a way, this is surprising. Economists understand that automation has costs, but they have largely emphasized the benefits: Machines makes things cheaper, and they free up workers to do other jobs. For instance, 41 percent of Americans were farmers a century ago, but thanks to tractors and mechanical harvesters, only 2 percent work in the agriculture today. The rest of us now can now aspire to be programmers or anesthesiologists or DJs or drone pilots.
The latest study reveals that for manufacturing workers, the process of adjusting to technological change has been much slower and more painful than most experts thought. “We were looking at a span of 20 years, so in that timeframe, you would expect that manufacturing workers would be able to find other employment,” Restrepo said. Instead, not only did the factory jobs vanish, but other local jobs disappeared too. Acemoglu and Restrepo say that every industrial robot eliminated about three manufacturing positions, plus three more jobs from around town.
If we are to make it through the next wave of automation, which is predicted to upend even more industries, we may have to rethink our policies about work and education — and learn from the industries that have coped the best.
Their research from Acemoglu and Restrepo joins the work of David Autor, David Dorn and Gordon Hanson, who have shown that the harms of trade with China were similarly concentrated in certain communities. The laid-off manufacturing workers couldn’t quickly find new jobs, so the economic pain lingered in their neighborhoods. Experts still believe that trade and automation can benefit Americans overall, contributing to lower prices and creating new kinds of jobs. But this evidence draws attention to the losers — the dislocated factory workers who just can’t bounce back.
The United States does have a program to retrain workers who lost their jobs to overseas competition, but research shows that most of them turn to other parts of the government safety net, such as Social Security, disability benefits and Medicaid. None of these efforts, though, seem to be doing enough for communities that have lost their manufacturing bases, where people have reduced earnings for the rest of their lives.
Perhaps that much was obvious. After all, anecdotes about the Rust Belt abound. But the new findings bolster the conclusion that these economic dislocations are not brief setbacks, but can hurt areas for an entire generation.
Acemoglu and Restrepo’s paper is also notable for its specificity. It has been difficult to pinpoint the impacts of technology on employment, in part because the effects have been so widespread. “When economists talk about automation, we’re actually talking about a bunch of stuff — we’re talking about capital, software, machinery, robots, artificial intelligence,” Restrepo said.
Many of these changes are invisible, or at least taken for granted, which is why false narratives persist, like the idea that trade with China caused the vast majority of job losses in the past decade. It’s harder to villainize Microsoft Word, or the robotic welders that have quietly replaced humans in many car factories.
How do we even know that automation is a big part of the story at all? A key bit of evidence is that, despite the massive layoffs, American manufacturers are making more stuff than ever. Factories have become vastly more productive. Many factors contributed to these changes and Acemoglu and Restrepo focused on one in particular — the rise of the industrial robot.
These are what people typically envision as robots — the autonomous sleds that carry parts across the factory floor, or the programmable arms that can weld, paint and even operate heavy machinery. The researchers obtained new data on the spread of this new technology, which is what enabled them to estimate how many jobs it displaced. Theirs is not a full accounting of the costs of automation, but a precise look at one component of this trend.
Since industrial robots still represent just a fraction of what we think of as automation, the claim that they caused 670,000 lost jobs is all the more surprising. As the researchers mention, some consultants believe that the number of industrial robots will quadruple in the next decade, which could mean millions more displaced manufacturing workers.
Restrepo is the first to concede that his research focuses on mostly the debit column. The benefits from robots — and from technological advancement in general — are even harder to measure, and it’s a matter that many economists are still sanguine about.
In the past, machines did automate many jobs out of existence, but new technology always created new opportunities — new kinds of desires, and new kinds of jobs to fulfill them. The recent anxieties about technological change are hardly new: Writing in the 1930s, the economist Maynard Keynes counseled patience, promising that any jobs lost to technology marked only a “temporary phase of maladjustment.”
The question, now, is what to do if the period of “maladjustment” that lasts decades, or possibly a lifetime, as the latest evidence suggests. Some say the time is nigh for a universal basic income. Bill Gates recently offered another provocative suggestion: Perhaps robots should pay taxes to compensate the workers that they replace.
Another lesson from history is that humans may have to become more flexible. America’s transformation from an agricultural nation to a manufacturing nation didn’t happen by accident, says Michael Chui, a partner at the McKinsey Global Institute who studies automation trends. “For people to go from working on the farm to working in factories, we greatly increased the educational attainment of the country over that time,” he said. “Our leaders made intentional decisions that made those changes possible.”
Some workers have weathered the strains of automation better than others. The number of jobs in finance, for instance, has continued to climb in recent decades, despite computers taking over many tasks, from filing papers, conducting research or even executing trades. Computerization displaced some people, but also created new kinds of work — jobs for programmers and people who sift through terabytes of financial data. In this case, automation amplified opportunities for people with advanced skills and talents.
Even in auto manufacturing, an industry that has been the poster child for robots displacing workers, there are signs of new opportunity. Ron Harbour, an analyst at Oliver Wyman, says that many of the most automated factories these days actually require more human labor to produce a car. That’s because cars themselves have become more complex, with things like powered seats and side airbags, and entertainment systems and backup cameras.
“There’s actually more work required of a plant today than ever before, so the labor hours have actually gone up a little bit,” he said. “The plants have made significant productivity improvements, but that has been offset by increasing complexity of the process. There’s just more work.”
The latest auto jobs are not the same as the old auto jobs, of course. These days, plants are seeking more robot technicians than assembly line welders. But this illustrates the hope that someday there will be more than enough work for both humans and robots and artificial intelligence routines, as long as we are prepared for it to look different than we're used to. We just have to muddle through the meanwhile.