Will robots cause the factory to go the way of the farm? Many economists seem to think so. “Rising levels of productivity benefit manufacturing,” declares a recent paper from the Federal Reserve Bank of St. Louis, “but also naturally lead to declining employment—much as the agricultural sector has experienced declining employment and rising output in the twentieth century.” Since the end of World War II, technology has caused agriculture’s share of U.S. employment to drop from nearly 10% to just 1%.
This analysis misses a point that becomes obvious once you think about it: One can eat only so much. The average person in a wealthy country consumes only about twice as many calories as someone on a subsistence diet. Demand for consumer goods is entirely different: Not only can it grow far faster than demand for food can, but innovators also create entirely new demands.
True, technology has other effects on agriculture. It can improve quality and safety, make more efficient use of land and water, improve distribution, and genetically engineer flavors, nutrients and disease-resistance. But none of that does much to accelerate the growth in overall agricultural consumption.
Manufacturing is different. In the past half-century, U.S. agricultural consumption grew around 80%—roughly the same rate as the population—while industrial output rose about 300%.
As productivity creates more wealth while reducing the costs of goods, demand for manufactured goods accelerates. We’ve gone from one TV and car per household, to a half-dozen and a couple respectively. Vacations are far more common, which among other things means more demand for new aircraft. Some 75% of airline passengers travel for leisure.
The biggest wild card is the creation of new appetites. Until cars, chemicals, computers, pharmaceuticals and smartphones were invented, there was no demand for them—or for the hardware and materials that make their production possible. Many fascinating products have yet to be put into widespread use—drones, virtual-reality systems, biocompatible electronics, robots and the computers that make artificial intelligence possible.
Tons of materials—petroleum, metals and other minerals—must be extracted, moved and processed to make a single automobile. Tons more must be extracted and processed to produce an equal weight of, say, lithium batteries, smartphones or robots. Each year the world fabricates a half-million tons of portable electronics alone.
Meeting the world’s untapped appetite for existing things—never mind what’s yet to be invented—would cause global consumption to rise by as much as 100 times in the coming decades. Emerging economies average one car per several hundred people, while in wealthy countries there is nearly one car per person.
More automation, including AI and robots, will be critical for meeting such huge prospective demands. Such rapidly expanding demand will ameliorate the job-killing effects of rapidly improving labor productivity. Between 1980 and 2000 U.S. factory output nearly doubled, with only a 10% loss in factory employment.
On the other hand, domestic factory employment has dropped about 30% since 2000. Not from automation, however; output has also stagnated since 2000, battered by a combination of a long recession, hostile tax and regulatory policies and so-called bad trade deals.
Translation: Growth in output matters. To be fair, the St. Louis Fed acknowledged the importance of output. Since future demand for manufactured goods is insatiable, don’t expect America’s factories to follow the trajectory of farm employment.
Mr. Mills, a senior fellow at the Manhattan Institute and faculty fellow at Northwestern University’s McCormick School of Engineering, is author of “Work in the Age of Robots,” out this month by Encounter Books.
Appeared in the June 26, 2018, print edition.