Economics researchers have long studied how local workers respond when an industry such as steel manufacturing is squashed by obsolescence or competition. Is the region able to regenerate with new industries and workers to fill them? Do displaced workers migrate to where new jobs appear, helping the economy at large absorb adverse economic shocks? Or does the local population decline, stuck in economic decay?
The answers to those questions are vital for economists, policymakers, and local governments to decide as new waves of technology, such as artificial intelligence and robotics, sweep across the business environment. The subject is also a hot political potato, from Republican President Donald Trump's dramatic trade approach to proposals from Democrats on automation, robots, and job loss.
A new study looks at worker migrations caused by two powerful economic shocks over the last 25 years that have caused deep manufacturing job losses in the United States: the appearance of robots/automation, and increasing import competition from China.
“THE APPEARANCE OF ROBOTS CAUSED A SIZABLE REDUCTION IN THE LOCAL POPULATION, WHEREAS MANUFACTURING JOBS LOST TO CHINESE EXPORTS DID NOT.”
The paper, The Impact of Technology and Trade on Migration: Evidence from the US (pdf), was co-authored by Marco E. Tabellini, Harvard Business School; Marius Faber, Basel University; and Andres Sarto, Princeton University.
It turns out, say the researchers, that both shocks are similar in their initial impact in the economy, and in line with previous findings by Daron Acemoglu and Pascual Restrepo and David Autor, David Dorn, Gordon Hanson and Jae Song (pdf). Both shocks significantly reduced the demand for manufacturing workers in areas historically specialized in these industries.
But by focusing on the migration response induced by these shocks, the authors document very different patterns. In short, the appearance of robots caused a sizable reduction in the local population, whereas manufacturing jobs lost to Chinese exports did not trigger such a response.
"Results are striking: Despite the similar, negative effect on manufacturing employment, the two shocks have a strongly different impact on migration" of American workers in and out of a region, according to the paper. "In particular, while robots lead to a significant reduction in population growth, Chinese imports have no detectable effect on population size."
Impact of robots more widespread
Tabellini and his colleagues compiled data from 722 US commuting zones (CZs), or areas where people live to commute to work. They split 25 years into three sections: 1990 to 2000, 2000 to 2007, and 2007 to 2015, the last year where full data on such shifts is available. They also controlled for conditions during the Great Recession, to see if the pattern held even in recession. It did.
Three fewer people move into a region for each robot added versus an area where local manufacturing jobs shifted to China, the study found. For every one robot installed per 1,000 humans, some 370,000 fewer people migrated to a region inside the US. Between 1993 and 2015, the number of robots in the US rose by almost 190,000. That implies a drop in people moving to local areas by 570,000 working-age individuals, the researchers estimate.
"The negative effect of robots is not confined to manufacturing employment, and you can see that pretty much everywhere," Tabellini says. "This can have very different implications, of course, and we find that this is entirely by people not willing to move in."
Generally, competition from Chinese exports is bad for manufacturing in the US on a local level, Tabellini points out. But over time, professions outside manufacturing adjusted in local economies, new industries developed, and more people decided to move to areas once dominated by production. The population in those commuting zones didn't shrink, according to the research. One reason might be that costs are lower for companies that moved production offshore, and the resulting, more efficient firms developed higher-skilled service work, like computer science, to fill the employment gap. But that idea needs to be studied further, Tabellini says.
“THREE FEWER PEOPLE MOVE INTO A REGION FOR EACH ROBOT ADDED VERSUS AN AREA WHERE LOCAL MANUFACTURING JOBS SHIFTED TO CHINA.”
Automation, meanwhile, has a more widespread negative effect throughout all types of sectors, Tabellini says. In regions where robots killed jobs, new industries didn't develop as they had in areas where jobs were lost to China. Without new jobs requiring higher skills, like computer science, people didn't move to the area, population didn't rise, and the economy stagnated.
Non-manufacturing jobs in commuting zones with "high service intensity" like nursing, law, and computer science, actually emerge in regions that import less expensive goods from China. Service jobs like retail and restaurants fall slightly, but not in a statistically significant way, the researchers found.
Robots on the march
Changes in the American economy from outsourcing and automation aren't likely to slow anytime soon. The number of robots will double this year from 2016 and rise at least three-fold by 2025, researchers estimate, citing figures including those from the International Federation of Robotics. And US workers are moving—or migrating inside the country—less often overall, in part because the population is aging.
A key question for policymakers centers on whether local labor markets can adjust or whether a population drop tied to robots will lead to persistent levels of unemployment and growing regional inequality if workers decide not to move into zones once dominated by factories, the study concludes.
More research coming
To understand what this means for the US economy longer term, Tabellini is undertaking new research.
"If this migration response comes from the fact that fewer high-skilled individuals are coming to our local labor market, is it really a good thing?" Tabellini asks. "Or is it something that eventually might, because of this multiplier effect, harm the local labor market even more, slowing down its recovery relative to what would have happened?"
Tabellini, who specializes in migration, also plans additional research on the differences in worker adjustments between Europe and the US together with his co-authors, Marius Faber and Andres Sarto. In Europe, people are far less likely to move, and labor market institutions are very different from their American counterparts, so the impact on local economies from robots replacing people may differ from that in the United States.