Robots and automation will destroy American jobs — it’s a worrying refrain we hear echoed by prominent figures in academia, the private sector, and the media alike. The Brookings Institution, the management consulting firm McKinsey & Company, labor leader Andy Sternand, most recently, CNN, all have issued serious, well-researched reports and written books saying so. Silicon Valley entrepreneur and Carnegie Mellon University scholar Vivek Wadhwa goes even further, predicting that thanks to automation, unemployment will top 70% within the next century.


Thanks to these reports, policymakers across the political spectrum have called for a vast array of solutions to deal with a radically different future of work. A look at the facts, however, shows a very different reality: Jobs are not currently disappearing due to automation in a way that throws people out of work — and haven’t done so historically. In fact, a major problem facing the American economy is not too much job displacement, but too little.

The most important salient fact is this: There is no new technology now or on the horizon that will allow the same amount of economic production to occur with vastly fewer people. The Bureau of Labor Statistics has found that average productivity growth since the Great Recession ended in 2007 has been the slowest of any long economic expansion of the post-World War II era, even as technological progress marched on. Furthermore, although productivity growth has picked up in the last two years, unemployment has remained near historical lows. Either way, robots are not robbing Americans of their jobs in large numbers.

History shows that they probably never will. Except during drawdowns of war production (the U.S. economy shrank 11.6% the year after World War II ended) and total economic collapses like the one happening in Venezuela, entire categories of jobs have never vanished overnight. Instead, these changes occur much more gradually.


Take, for example, coal mining. In 1919, about 700,000 Americans, nearly 2% of the labor force, earned their living mining coal. They did so mostly with hand tools. Today, about 50,000 people work in coal, about 3/100ths of 1% of the labor force, and all of them use highly automated equipment. But the decline did not occur suddenly, norhas there been a sharp decline in coal industry jobs caused by any particular technology. Instead, a group of factors ranging from better coal technology to cheaper natural gas has slowly but surely reduced the ranks of miners over an entire century. What’s more, the number of people working in the coal industry has grown slightly in the past few years even though almost nobody believes coal has a bright future.

The same pattern repeats itself across other declining industries, ranging from travel agents to farmers to textile workers. The jobs that are the easiest to automate will always be those that are most repetitive and dull, and the fact that a machine is good at something does not mean that all human jobs related to that activity will be eliminated.

A $50 tablet sold in an airport vending machine has the processing power to do more mathematical calculations more quickly than any human being, but the wide availability of great computer processing power has actually increased the need for math teachers, mathematicians, scientists, engineers and others involved with technology. The need for vastly fewer people on farms (about 2% of the labor force today compared with over a third a century ago) hasn’t destroyed jobs associated with food, but instead helped to create millions of new jobs in restaurants, food processing, and related industries.

Abrupt productivity growth that did eliminate categories of jobs would probably be quite good for the economy and workers. Wages can rise most quickly in a non-inflationary way when workers produce more. Relatively slow wage growth (although accelerating in the past few years) is a direct result of this slow productivity growth. Faster productivity growth, through automation or otherwise, would translate into faster wage growth.

In short, there is no real evidence that robots are about to take you and your colleagues’ jobs on a massive scale.

Image credit: Abdul Razak Latif

Featured Publications