Andrew Yang’s dystopian warnings about automation and mass unemployment have mostly been a quirky and harmless sideshow throughout this year’s Democratic primary. Like many tech types and TED Talk fans, the former businessman believes that the United States is living through an age of technological upheaval that has already devastated manufacturing work and will soon render millions of other jobs obsolete. His plan to address this grim future is to offer every American a $1,000 per month basic income—he calls it a “freedom dividend”—which he hopes will prevent society from unraveling as everyone from truckers to doctors sees their occupations devoured by intelligent machines. (I’m not exaggerating: He’s warned about street riots by the unemployed.)
The idea that a universal basic income could prevent a backlash when new tech inevitably steamrolls old industries is trendy in Silicon Valley, and bringing it to the primary trail has won Yang a devoted, very online following that has donated a significant amount of money to his campaign—more than enough to keep him in the debates, where he’s become adept at shoehorning his pitch about the impending robot apocalypse into discussions of other more mundane policy topics. But during Tuesday’s Democratic showdown in Ohio, his pet subject took center stage when CNN asked the candidates about their plans to help Americans who lose their jobs to automation. This led to a brief and unexpected clash between Yang and Sen. Elizabeth Warren, who suggested that the main reason the U.S. has bled jobs in industries like manufacturing was not technology but rather “bad trade policy,” and that she would focus on helping workers by strengthening unions and giving them seats on corporate boards so they would have a say in decision-making.
Yang responded essentially that, no, the robots are the problem:
Sen. Warren, I’ve been talking to Americans around the country about automation. And they’re smart. They see what’s happening around them. Their Main Street stores are closing. They see a self-serve kiosk in every McDonalds, every grocery store, every CVS. Driving a truck is the most common job in 29 states, including this one; 3.5 million truck drivers in this country. And my friends in California are piloting self-driving trucks.
What is that going to mean for the 3.5 million truckers or the 7 million Americans who work in truck stops, motels, and diners that rely upon the truckers getting out and having a meal? Saying this is a rules problem is ignoring the reality that Americans see around us every single day.
Following the debate, a “fact check” by the AP claimed that Yang was right and Warren wrong. “Economists mostly blame [manufacturing] job losses on automation and robots, not trade deals,” it stated. But this was incorrect. No such consensus exists, and if anything, the evidence heavily suggests that trade has been the bigger culprit in recent decades. All of which points to a broader issue: Yang’s schtick about techno doom may be well-intentioned, but it is largely premised on BS, and is adding to the widespread confusion about the impact of automation on the economy.
Yang is not pulling his ideas out of thin air. Economists have been debating whether automation or trade is more responsible for the long-term decline of U.S. factory work for a while, and it’s possible to find experts on both sides of the issue. After remaining steady for years, the total number of U.S. manufacturing jobs suddenly plummeted in the early 2000s—from more than 17 million in 2000 to under 14 million in 2007. (The Great Recession saw about 2.2 million more vanish, though they’ve bounced back a bit since.) This all coincided with China’s entry into the World Trade Organization and rapid transformation into an industrial powerhouse, which led many to assume that offshoring had caused America’s rapid industrial decline. But some economists disagreed. They pointed out that while the number of manufacturing workers had crashed, factory output was still rising, which suggested that technological advances like industrial robots were just making things much more productive and efficient. In 2015, economists from Ball State University suggested that around 87 percent of manufacturing job losses between 2000 and 2010 were due to improved productivity from automation, and just 13 percent were due to trade, claims that later appeared in the New York Times. So when Yang says that the “reason Donald Trump was elected was that we automated away 4 million manufacturing jobs in Michigan, Ohio, Pennsylvania and Wisconsin,” he’s just echoing stuff that’s been printed in the paper of record.
The problem is that the Ball State team’s findings have basically been eviscerated by other researchers. In a 2018 paper, Susan Houseman of the Upjohn Institute for Employment Research showed that the rise in manufacturing productivity after the late 1990s was largely an illusion driven by how the government measures output in the computer and semiconductor industry. Within other manufacturing sectors, productivity grew slowly, which meant industrial robots probably couldn’t explain job losses.
There are other clues that the automation story is off. America hasn’t just lost manufacturing workers; as Houseman notes, the number of factories also declined by around 22 percent between 2000 and 2014, which isn’t what you’d expect if assembly workers were just being replaced by machines. In a 2017 paper, meanwhile, economists Daron Acemoglu of MIT and Pascual Restrepo of Boston University concluded that the growth of industrial robots in the U.S. since 1990 could only explain between between 360,000 and 670,000 job losses.
By comparison, the proof placing blame on trade and China is much stronger. Justin Pierce of the Federal Reserve Board and Peter Schott of Yale have found evidence that the U.S.’s decision to grant the People’s Republic permanent normal trade relations in 2000 led to declines in American jobs. And in a widely celebrated line of papers, economists David Autor, David Dorn, and Gordon Hanson have concluded that imports from China eliminated almost 1 million American manufacturing jobs from 1999 to 2011, and that the knock-on effects within local communities led to a total loss of as many as 2.4 million U.S. jobs overall. Other research has suggested that the rising value of the dollar may have hurt American manufacturing during the 2000s by making our exports more expensive.
In an interview, Autor told me that the approach in his paper was “conservative,” and that the actual number of jobs lost to Chinese imports may well have been higher. “If we talk about the economic trauma of the 2000s, that’s not primarily due to automation,” he told me. “Nobody can tell you what great invention happened in 1999 that wiped out 20 percent of manufacturing jobs. So common sense suggests trade was the main factor.”
What makes this whole conversation a bit complicated is that, over the very long term, automation probably has played a role in limiting manufacturing’s share of employment in the U.S. Factories really have become more advanced and efficient since, say, the 1970s. Some of Autor’s own recent research suggests that automation has shifted the power balance between workers and capital owners, leading labor’s share of national income to decline. It’s also true that, to some extent, trade and tech may have worked together to eliminate jobs: Some factory owners may have automated their production lines to save labor costs because they were under pressure from Chinese competition. But robots alone simply don’t explain the tornado that hit the industrial Midwest at the turn of the century.
Of course, that’s all in the past. The future could be different, and the past few years have spawned an entire genre of articles speculating about whether technology like self-driving cars and machine learning algorithms are about to make vast swaths of today’s workforce obsolete. This deluge of pieces has been fueled by academics, think tanks, and consulting firms that churn out dramatic claims about the share of workers whose jobs could be terminated. During Tuesday’s debate, moderator Erin Burnett referred to a recent study that found about “a quarter of American jobs could be lost to automation in just the next 10 years.”
Could is the operative word there, because in the end, this is speculation—a bunch of McKinsey analysts licking a finger and pointing it up to the wind. What makes it all a bit surreal is that, as of now, one big problem with the U.S. economy appears to be that there isn’t enough automation. Productivity growth has been weak for years, suggesting that businesses either aren’t investing enough in new technology, or that their investments are paying off with great efficiency gains.
Some of the big, headline-dominating narratives about how tech and software are already devouring the world aren’t everything they seem, either. Take the fabled retail apocalypse: Many have blamed Amazon for the death of brick-and mortar stores across the country. But several high-profile bankruptcies, like Toys R Us, seem to have been the result of unsustainable debts borrowed by their private equity owners as much as competition from the internet.
Meanwhile, the overall job market is doing fine. Employment among working-age adults is back above pre-recession levels, and pay is actually growing fastest among low-wage workers and industries. The labor market isn’t perfect, by any means. But the fact that local malls have turned into ghost towns doesn’t seem to have created a wave of desperate former Forever 21 clerks.
New technology will change the economy and the way people work. It already is. But those shifts will be more complex than Yang admits and probably won’t look like the wave of mass unemployment that he and his like-minded supporters tend to envision. Take one of the candidate’s favorite topics: self-driving cars. Just as the rise of Uber has changed the taxi industry forever and raised concerns about how drivers are treated, experts have worried that the rise of self-driving trucks could remake the freight hauling industry. In a 2018 paper, UC–Berkeley sociologist Steve Viscelli suggested that in the most likely scenario, long-haul truckers, who tend to make middle-class wages, will be replaced by poorly paid drivers tasked with steering autonomous vehicles through tricky city streets, which onboard navigation systems handle less well than highways. That’s not all jobs disappearing. It’s jobs changing. To address that change, Viscelli suggests that new labor regulations will probably be needed to make sure companies don’t exploit this new class of employees. This is a far different approach than Yang’s, which is to just hand Americans a monthly allowance so they don’t start throwing Molotovs.
And therein lies the real problem with Yang’s outlook. It’s not just unrealistic. It’s lazy. When you buy the sci-fi notion that technology is simply a disembodied force making humanity obsolete and that there’s little that can be done about it, you stop thinking about ideas that will actually prevent workers from being screwed over by the forces of globalization or new tech. By prophesying imaginary problems, you ignore the real ones.