The claim that automation is responsible for massive job losses has been made in almost every one of the Democratic debates. In the last debate, technology entrepreneur Andrew Yang told of automation closing stores on Main Street and of self-driving trucks that would shortly displace “3.5 million truckers or the 7 million Americans who work in truck stops, motels, and diners” that serve them. Rep. Tulsi Gabbard (Hawaii) suggested that the “automation revolution” was at “the heart of the fear that is well-founded.”
When Sen. Elizabeth Warren (Mass.) argued that trade was a bigger culprit than automation, the fact-checker at the Associated Press claimed she was “off” and that “economists mostly blame those job losses on automation and robots, not trade deals.”
In fact, such claims about the impact of automation are seriously at odds with the standard data that we economists rely on in our work. And because the data so clearly contradict the narrative, the automation view misrepresents our actual current challenges and distracts from effective solutions.
Output-per-hour, or productivity, is one of those key data points. If a firm applies a technology that increases its output without adding additional workers, its productivity goes up, making it a critical diagnostic in this space.
Contrary to the claim that automation has led to massive job displacement, data from the Bureau of Labor Statistics (BLS) show that productivity is growing at a historically slow pace. Since 2005, it has been increasing at just over a 1 percent annual rate. That compares with a rate of almost 3 percent annually in the decade from 1995 to 2005.
This productivity slowdown has occurred across advanced economies. If the robots are hiding from the people compiling the productivity data at BLS, they are also managing to hide from the statistical agencies in other countries.
Furthermore, the idea that jobs are disappearing is directly contradicted by the fact that we have the lowest unemployment rate in 50 years. The recovery that began in June 2009 is the longest on record. To be clear, many of those jobs are of poor quality, and there are people and places that have been left behind, often where factories have closed. But this, as Warren correctly claimed, was more about trade than technology.
Consider, for example, the “China shock” of the 2000s, when sharply rising imports from countries with much lower-paid labor than ours drove up the U.S. trade deficit by 2.4 percentage points of GDP (almost $520 billion in today’s economy). From 2000 to 2007 (before the Great Recession), the country lost 3.4 million manufacturing jobs, or 20 percent of the total.
Addressing that loss, Susan Houseman, an economist who has done exhaustive, evidence-based analysisdebunking the automation explanation, argues that “intuitively and quite simply, there doesn’t seem to have been a technology shock that could have caused a 20 to 30 percent decline in manufacturing employment in the space of a decade.” What really happened in those years was that policymakers sat by while millions of U.S. factory workers and their communities were exposed to global competition with no plan for transition or adjustment to the shock, decimating parts of Ohio, Michigan and Pennsylvania. That was the fault of the policymakers, not the robots.
Before the China shock, from 1970 to 2000, the number (not the share) of manufacturing jobs held remarkably steady at around 17 million. Conversely, since 2010 and post-China shock, the trade deficit has stabilized and manufacturing has been adding jobs at a modest pace. (Most recently, the trade war has significantly dented the sector and worsened the trade deficit.) Over these periods, productivity, automation and robotics all grew apace.
In other words, automation isn’t the problem. We need to look elsewhere to craft a progressive jobs agenda that focuses on the real needs of working people.
First and foremost, the low unemployment rate — which wouldn’t prevail if the automation story were true — is giving workers at the middle and the bottom a bit more of the bargaining power they require to achieve real wage gains. The median weekly wage has risen at an annual average rate, after adjusting for inflation, of 1.5 percent over the past four years. For workers at the bottom end of the wage ladder (the 10th percentile), it has risen 2.8 percent annually, boosted also by minimum wage increases in many states and cities.
To be clear, these are not outsize wage gains, and they certainly are not sufficient to reverse four decades of wage stagnation and rising inequality. But they are evidence that current technologies are not preventing us from running hotter-for-longer labor markets with the capacity to generate more broadly shared prosperity.
National minimum wage hikes will further boost incomes at the bottom. Stronger labor unions will help ensure that workers get a fairer share of productivity gains. Still, many toiling in low-wage jobs, even with recent gains, will still be hard-pressed to afford child care, health care, college tuition and adequate housing without significant government subsidies.
Contrary to those hawking the automation story, faster productivity growth — by boosting growth and pretax national income — would make it easier to meet these challenges. The problem isn’t and never was automation. Working with better technology to produce more efficiently, not to mention more sustainably, is something we should obviously welcome.
The thing to fear isn’t productivity growth. It’s false narratives and bad economic policy.
This column first appeared in the Washington Post.