Automation is Driving Inequality
What’s Driving Inequality? Automation, BU Researcher Says
Pascual Restrepo on robots’ role in the rich-poor gap
Any number of suspects have been blamed for the growing gap between rich and poor Americans: the decline of unions. Trade. Jobs being shipped offshore. But a recent paper coauthored by BU economist Pascual Restrepo, says the primary driver is something else altogether: automation.
The paper—by Restrepo, a College of Arts & Sciences assistant professor of economics, and MIT’s Daron Acemoglu—says blue-collar workers in jobs vulnerable to automation are losing employment and income. They compete for fewer jobs with other blue-collar workers, pushing down wages. Meanwhile, better-skilled and educated workers whose jobs can’t be automated will see higher wages.
“We looked at the data, and that’s exactly what you see,” says Restrepo. “Think about the welder in car manufacturing or think about clerks in the service sector. Both of these types of occupations can be automated, the first one using robots and the second one by software. These are exactly the groups where you find wages going down in relative terms. You don’t see wages going down for the managers, the CEOs, the designers.”
BU Today spoke with Restrepo about the implications of his findings.
With Pascual Restrepo
BU Today: You downplay the role of unions’ decline, but might it be related? If unions were more robust, might they not have protected some workers’ jobs from automation, or secured retraining or better severance pay for workers replaced by machines?
Pascual Restrepo: I agree a hundred percent. The line we take in the paper is to argue that unions, if they had an effect, were mostly mediating the effects of technology. For example, [in] Germany, what unions did was agree with managers [to] a way in which robots were going to be deployed in plants: as workers started to retire, the plants could start phasing in the adoption of robots. Unions can also foster retraining or create an occupational ladder, where workers move to other occupations inside the firm. I think [this] is one of the reasons why the same technologies might be having a more unequal impact in the United States compared to other countries.
The point is not to say that unions do not matter—and these are empirical findings that might change in the future—but the big factor is technology.
BU Today: What about the role of competition?
Pascual Restrepo: We studied the role of offshoring. Offshoring conceptually is similar to automation. In automation, we are taking the welder and replacing them with a robot. With offshoring—for example, a US seamstress—you are replacing them with a seamstress in another country. Offshoring also contributed to increasing inequality, but it had a smaller role. Uneven exposure to automation [among workers] explains about 50 percent of the increase in inequality, whereas offshoring only explains 10 to 15 percent.
Automation concentrates in manufacturing, retail, mining. Not as much in the service sector, with some exceptions. There’s some automation in legal services; now you use software for document discovery, and you don’t rely as much on paralegals.
BU Today: The United States has worse inequality than other wealthy nations. Do you have any sense of whether automation plays a similar role in fostering inequality abroad?
Pascual Restrepo: Automation is not unique to the United States. If you look at countries spearheading the adoption of robots, it’s not the United States. Japan, Germany, Italy, and many European countries are ahead of the United States. But it is having different effects on wages. Part of the reason could be unions. Maybe the same technology is unfolding differently because of the way institutions protect workers in different countries. In some European countries, you see more unemployment and less changes in wages—if institutions are such that you cannot change wages, then you would expect firms to fire workers or stop hiring them, rather than lower wages. The Europeans would be ahead of the United States [in] how good they are at retraining workers.
BU Today: What should policymakers do?
Pascual Restrepo: If you give unions too much power, they’re going to push up wages and generate incentives for automation. You have to strike the right balance. When you go to European restaurants, they would have self-serving kiosks, and that’s in part because wages are so high, firms find it profitable to use these automated devices.
The question is how much can we retrain people in practice, but if we could retrain people, that would be very useful. In the longer run, we can expand education. Many of those people who were going to become welders, they’re going to become designers, and those are the tasks that are not being automated.
But again, can we educate people at the same rate as technology is advancing? Between 1950 and 1980, we managed to increase the rate at which new cohorts entering the labor market were completing college or high school degrees. But that stopped. Among people between 25 and 30 years of age, only 30 percent are completing college, and that number has remained flat for men and increased slightly for women. What that’s telling you is the US educational system has failed to respond to these waves of technology.
One alternative is to direct technology—what are the technologies that we want to invest in? Technologies that generate automation create inequality. But other technologies [generate] improvements that do not involve automation. For example, the personal computer—that’s a fantastic technology that has made all of us more productive. We’re starting to think about artificial intelligence. The invitation for scientists would be, can we develop artificial intelligence technologies that augment humans instead of automating them. Maybe you want more [public] subsidies to technologies that have the potential to create prosperity or augment workers, especially workers without a college degree.
Our point is not that technology is bad. We are not radical Luddites.
Comments & Discussion
Boston University moderates comments to facilitate an informed, substantive, civil conversation. Abusive, profane, self-promotional, misleading, incoherent or off-topic comments will be rejected. Moderators are staffed during regular business hours (EST) and can only accept comments written in English. Statistics or facts must include a citation or a link to the citation.