An army of robots now scrub floors, grow microgreens and flip burgers. Due to advances in artificial intelligence, computers will supposedly take over much more of the service sector in the coming decade, including jobs in law, finance and medicine that require years of education and training.
Will automation-induced job loss tear society apart? The question has even influenced the US presidential race. Candidate Andrew Yang blames automation for a long-simmering crisis of underemployment. He plans to hand out free money to every American citizen in the form of a monthly “dividend” of $1,000.
Poor job quality and stagnant wages are major problems in America and across much of the world, but it is wrong to blame these problems on an accelerating pace of automation, which is hardly in evidence.
Automation Cassandras often point to the manufacturing sector as the precedent for what will happen to the rest of the economy. It is true that, for the manufacturing industry, a jobs apocalypse has already taken place.
And this process is occurring across the world: according to the UN, the share of all workers employed in manufacturing is falling globally, even as industrial production per person continues to rise. This is the case in wealthy and poor countries. Yet it is hasty to ascribe these trends to accelerating automation.
While machines now make everything from shoes and shirts to cars and computers, there has been no significant uptick in the pace of labor-saving productivity growth in industry in recent decades.
On the contrary, industrial efficiency has been improving at a sluggish pace for decades, leading the Nobel-prize-winning economist Robert Solow to quip, in 1987: “We see the computer age everywhere except in the productivity statistics.”
Our collective sense that the pace of labor-saving technological change is accelerating is an illusion. It’s like the feeling you get when looking out of the window of a train car as it slows down at a station: passing cars on the other side of the tracks appear to speed up. Labor-saving technical change appears to be happening at a faster pace than before only when viewed from across the tracks – that is, from the standpoint of our ever more slow-growing economies.
That is the real problem: a pervasive and increasingly global economic stagnation – affecting industry especially – that is marked by low rates of investment, low rates of economic growth and hence low rates of job creation.
In the context of economic stagnation, even small increases in productivity are enough to destroy more manufacturing jobs than are created.
The best explanation for this worsening economic stagnation is that, since the 1970s, more and more countries adopted export-led growth strategies, built up manufacturing sectors and began to compete in global markets.
That led in turn to heightened competition, making fast-paced industrial expansion – and fast-paced economic growth – much more difficult to achieve.
In this context, countries with high levels of robotization are not necessarily the ones that have lost the most industrial jobs. On the contrary, Germany, Japan and South Korea have some of the highest levels of robots per manufacturing worker but also boast higher manufacturing employment shares.
In Germany and Japan, automation helps firms preserve jobs in manufacturing in the face of intense international competition. Chinese firms have been investing heavily in robotics in the past few years, to preserve jobs as domestic wages rise and competition from even lower-cost countries intensifies.
Meanwhile, no other sector has replaced industry as a major economic growth engine. In country after country, slowing industrial expansion has been accompanied by falling rates of economic growth.
Some services like wholesale trade have seen spurts of rapid productivity growth, but these fail to coalesce into sustained, sector-wide efficiency gains like those endemic to manufacturing over the history of its development.
The wider environment of slowing growth explains the prevailing low labor demand largely by itself. Once again, the major problem in labor markets is a slowing pace of job creation, associated with this sluggish economic expansion, rather than an accelerating pace of automation-induced job destruction.
From the 1970s, when industrial competition began to heat up and economic growth rates deteriorated, unemployment levels in many countries first rose and then stubbornly refused to fall. Politicians began to push for weakened job protections and scaled-back unemployment benefits.
Workfare came to serve as the main response to job loss. Outside of a few countries that still provide generous benefits to the unemployed, such as France, few workers can afford to remain unemployed for long. Job-losers tend to join young labour-market entrants in part-time, temporary or no-contract work.
In most countries, these “non-standard” workers have few legal protections and are economically precarious. They are forced to moderate their demands in slow-growing economies. Workers who are not protected by powerful unions or labor laws find it difficult to pressure employers to raise their wages or improve working conditions.
As long as underemployment persists, inequality is likely to intensify. An expanding gap between the growth of real wages and productivity levels has contributed to a 9% shift from labor income to capital income in the G20 countries over the past 50 years.
In sluggish economies periodically racked by economic crisis and austerity, it is easier to blame the resulting social deterioration on robots, or on vulnerable sections of the workforce such as immigrants, women, and racial or religious minorities, than to face its true causes.
Given the winding down of the industrial-growth engine – which has accompanied the spread of productive capacities around the world – restoring previously prevailing rates of economic growth will prove difficult if not impossible. Unless we find some way to share the work that remains, beggar-thy-neighbor politics really will tear our societies apart.
Aaron Benanav is a researcher in the social sciences at the University of Chicago. He is writing a book about the global history of unemployment. This is an abridged version of his New Left Review article on automation