Research: Firms often use automation to control the wages of certain employees | MIT News

When we hear about automation and artificial intelligence replacing jobs, it may seem like a technological tsunami will wipe out workers in general, in the name of greater efficiency. But a study co-authored by an MIT economist shows a markedly different dynamic in the US since 1980.

Rather than using automation to pursue higher productivity, firms often use automation to hire workers who earn “wage pay,” who earn higher wages than other similar workers. In practice, that means that automation tends to reduce the wages of workers without a college education who were earning better wages than most workers with the same degree.

These findings have at least two major implications. First, automation has affected the growth of US income inequality even more than most observers realize. At the same time, automation has produced lower productivity, apparently due to companies’ focus on managing wages rather than finding more technology-driven ways to improve efficiency and long-term growth.

“There was a misdirection of automation,” said Daron Acemoglu of MIT, co-author of a published paper detailing the study’s findings. “The higher the salary of a worker in a particular industry or job or occupation, the more attractive automation is to companies.” In theory, he notes, firms can automate. But they did not, by insisting on it as a tool to spend wages, which helps their internal numbers in the short term without creating the right way to grow.

The study estimates that automation is responsible for a 52 percent increase in wage inequality from 1980 to 2016, and that nearly 10 percent of that point is attributable primarily to firms replacing wage-earning workers. This inefficient management of certain workers removed 60-90 percent of the productivity gains from automation over a period of time.

“It’s one of the reasons that product development is probably a little quieter in the US, except that we’ve had an incredible amount of new patents, and an incredible amount of new technology,” Acemoglu said. “Then you look at the production numbers, and they’re sad.”

The paper, “Automation and Rent Dissipation: Implications for Wages, Inequality, and Productivity,” appears in the May issue of Quarterly Journal of Economics. The authors are Acemoglu, Institute Professor at MIT; and Pascual Restrepo, associate professor of economics at Yale University.

Implications of inequality

Dating back to the 2010s, Acemoglu and Restrepo have teamed up to conduct a number of studies on automation and its effects on employment, wages, productivity and firm growth. In general, their findings suggest that the effects of automation on the workforce after 1980 are more significant than many other scholars have believed.

To conduct the current study, researchers used data from multiple sources, including US Census Bureau statistics, data from the bureau’s American Community Survey, industry numbers, and more. Acemoglu and Restrepo analyzed 500 demographic groups, sorted by five levels of education, as well as gender, age, and ethnic background. The study links this information to an analysis of changes in 49 US industries, to look at the granularity of how automation has affected workers.

Ultimately, the analysis allowed the researchers to estimate not just the total number of jobs eliminated due to automation, but how much of that involved firms specifically trying to eliminate the wage bill that accrues to some of their workers.

Among other findings, the study shows that in the groups of workers affected by automation, the biggest results come from workers in the 70-95 percent of the salary range, indicating that the highest paid workers are bearing the brunt of the process.

And as the analysis shows, almost one-fifth of the overall increase in income inequality is due to this factor alone.

“I think that’s a huge amount,” said Acemoglu, who shared the 2024 Nobel Prize in economics with his longtime collaborators Simon Johnson of MIT and James Robinson of the University of Chicago.

He adds: “Automation is an engine of economic growth and we’re going to use it, but it creates huge inequality between capital and labor, and between different labor groups, and that’s why it could be a major contributor to the increase in inequality in the United States over the last few decades.”

The paradox of production

The study also sheds light on the fundamental, but often overlooked, choices of corporate executives. Think about the type of automation – call center technology, for example – that might not work well for the business. Still, corporate executives have incentives to accept, cut wages, and oversee a low-productivity, high-profit business.

On balance, another version of this appears to have occurred in the US economy since the 1980s: Capital gains are not the same as productivity increases.

“Those two things are different,” Acemoglu said. “You can reduce costs while reducing productivity.”

Indeed, the current study by Acemoglu and Restrepo is reminiscent of the observation of the late MIT economist Robert M. Solow, who wrote in 1987, “You can see the computer age everywhere but in productivity statistics.”

In that sense, Acemoglu notes, “If managers can reduce productivity by 1 percent but increase profits, many of them may be happy with that. It depends on their priorities and values. So another important implication of our paper is that good automation at the margins is combined with not-so-good automation.”

To be clear, the research does not mean that less automation is always better. Certain forms of automation can increase productivity and feed a virtuous cycle where a company makes more money and hires more workers.

But at the moment, Acemoglu believes, the complexity of automation has not been seen clearly enough. Perhaps seeing the broad historical pattern of US automation, since 1980, will help people better understand the trade-offs involved – not just economists, but hard managers, workers and professionals.

“The important thing is whether it is included in people’s thinking and where we come in terms of a comprehensive assessment of automation, in terms of inequality, productivity and labor market outcomes,” Acemoglu said. “So we hope this study will move the dial there.”

Or, as he concludes, “We could be missing out on potentially even better productivity gains by measuring the type and level of automation carefully, and in a way that improves productivity. All choice, 100 percent.”