So, Where Are All Those Robots?

Lots of people think that the robots are coming to steal everybody’s jobs. I even wrote a whole thing about what would happen if they did.

But another story is emerging from several recent papers and columns by economists and economic writers. Instead of a world without work, they say, there is currently more evidence for a world with too much work—and not enough humans to do it all. Rather than high-flying investment in machines and similarly high unemployment, there is strangely low investment and happily low joblessness. How can anybody say robots are killing jobs when the killer is nowhere to be seen and the supposed victim isn’t even dead?

Economists and writers aren’t just pushing back against the robots-taking-jobs story. Some of them are downright begging for more robots to descend on the economy and bring with them the gift of productivity. (As a side note: When economists and economic writers say “robots” they rarely mean actual anthropomorphic machines like C3PO; it’s more of a catch-all for new technology that replaces work previously done by people.) Here are the four main parts of the case for embracing robots without fear.

1. The U.S. economy is in a productivity recession. Most people agree that if automation were replacing workers, there would be an enormous productivity boom coinciding with massive job losses or a long period of miserably low wage-growth. Instead, the modern economy is showing the exact opposite of that. Unemployment is low. Wages are rising even faster than productivity—an extraordinarily rare occurrence in the last four decades. This isn’t what the end of work was ever supposed to look like.

2. Companies don’t seem to be investing in technology nearly as much as they used to. The growth in capital investment—one measure of how much companies spend on new equipment and technology—is at its lowest rate in 60 years. “Capital investment in the workplace has grown more slowly since 2002 than in any other postwar period,” write Lawrence Mishel, the longtime president of the Economic Policy Institute (EPI), and Josh Bivens, the director of research there, in a compelling new paper. “This is the opposite of what we would expect with a looming robot apocalypse based on automation.” Software may be eating the world, in some people's estimation. But by Mishel and Bivens’s count, new investment in information technology and software has never been lower on record.

What’s more, perhaps software doesn’t eliminate nearly as much work as one would think. The number of ATMs grew by a factor of ten between 1990 and 2010, and still the number of bank tellers kept growing.Today, travel websites mean that anyone can comparison shop for their flights; and yet, the number of travel agents have been growing since 2010.

3. Globalization is a much bigger deal than automation for work and wages. The economists Daren Acemoglu and Pascual Restrepo recently published a paper that found some evidence that automation has already nibbled away at work and wages. They concluded that, for every new robot per 1,000 workers, six people lose their jobs and employment fell by about one percent.

But Mishel and Biven call hogwash. They point out that, in Acemoglu and Restrepo’s own paper, “job displacements caused by trade with China in the 2000s were four times as large as their estimate of job loss due to robots.” This echoes other economists, like David Autor, who have argued that globalization, and especially trade with China, has had much more to do with stagnating wage growth than anything related to technology.

4. The U.S. economy’s creative-destruction engine is broken. In an economy where automation is rendering humans obsolete in huge swaths of the economy, one would expect to see entire occupation categories getting wiped out, as people go unemployed or take jobs in other sectors. But research by the economists Robert D. Atkinson and John Wu finds that the labor market has never been calmer. By one measure known as “occupational churn,” technology is making fewer jobs obsolete than at any time in the last 165 years. Mishel and Bivens similarly find that the pace of occupational change has fallen to its lowest point since the 1950s.

This may be a part of what I’ve called America’s declining “mojo”—the decades-long decline in entrepreneurship, job switching, and between-state migration. There are several leading explanations of America’s falling dynamism, including the high cost of housing in America’s most productive cities and the emergence of new monopolies. Either way, America’s famous capacity for creative destruction is currently neither creating nor destroying much of anything.

These four arguments tell an altogether compelling story: The U.S. economy currently suffers not from too much automation, but rather from too little investment in the sort of technology that would raise the country’s lackluster productivity.

Do I buy it? I do. It’s hard to look at the economy and come away thinking that it suffers from a crisis of automation, or is on the precipice of a new industrial revolution right at this moment. I also agree that more technology, even labor-replacing technology, would be good for economic growth. And yet I still think it’s sensible to worry about the future of automation, for the following reasons.

First, there is only so much that one can say about the future of work from studying an economy many years into an expansion, since the most wrenching changes to the job force almost always occur during recessions. Just look at manufacturing jobs. They have declined by almost 40 percent since their 1960s high. But, if you only look at periods of economic growth, manufacturing employment actually grew by 4 million between 1969 and 2015. That means more than 100 percent of the decline in manufacturing jobs occurred during the brief period of time between 1970 and 2015 that the economy has been in a recession. In downturns, firms let excess workers go and lean more on labor-saving technology to maintain their profits. The next recession will tell us more about the future of the workforce than the current recovery.

Second, the last six months in retail employment provide a fine parable—not only for how technology typically creates more work, but also for it might subtly endanger jobs at the same time. Since October 2016, U.S. department stores have shed nearly 100,000 jobs—more than the number of American steel workers—yet overall unemployment remains quite low. That’s because the e-commerce sector isn’t just gutting malls. It’s also created tens of thousands of new jobs in warehousing (all that Amazon merchandise has to live somewhere) and transportation.

In the short run, the digitization of retail has created jobs. But it’s replacing in-store salespeople—not easily automated, since who wants a robot clothing assistant?—with warehousing and transportation workers. There are nearly 2 million truck drivers and 300,000 warehouse laborers and stocking clerks in the United States, and there are rather direct efforts underway to make a great number of these jobs obsolete. Amazon has expanded its armada of warehouse robots to more than 40,000, and self-driving cars have the attention of practically every auto and technology company in the world. So, e-commerce has created jobs—ones that are quite vulnerable to automation.

Finally, it’s conceivable that the ostensibly tranquil and low-turbulence economy is masking something more disruptive underneath the surface. Ryan Avent, the author of The Wealth of Humans, has thought about this question deeply and offered a plausible explanation. In his telling, automation has created an abundance of labor, including machine labor and human labor. Just as rising supply typically leads to falling prices, the oversupply of labor has put a downward pressure on wages. Companies, seeing that they have access to cheap labor in a slowly growing economy, invest less in new risky technology, which leads to less productivity growth. High employment, low productivity, low wage growth, and automation can all live together in the same story.

It’s trite yet true to point out that predicting the future is hard, and there are centuries of history to support the argument that technology never permanently increases unemployment. What the robot crowd is predicting is nothing less than an industrial revolution—a once-(or twice)-in-history event that forever alters the shape of the economy. By definition, historically unprecedented events aren’t easily predicted by extrapolating from historical record. Instead, one has to pay extremely close attention to the most up-to-date statistics. For now, I think it’s hard to make the case that automation is even a tertiary concern for the 2017 labor market. But the next recession will tell us much more.

Derek Thompson is a staff writer at The Atlantic, where he writes about economics, technology, and the media. He is the author of Hit Makers and the host of the podcast Crazy/Genius.

Connect Twitter