While automation and AI are often thought to broadly replace jobs, an MIT economist’s study highlights distinct patterns in the U.S. since 1980. Rather than maximizing productivity, companies have frequently used automation to target workers with a “wage premium,” those earning more than peers with similar qualifications. This has predominantly impacted non-college-educated workers who had secured higher salaries.
This trend has two major consequences. First, automation has contributed more to U.S. income inequality than many realize. Second, it has provided only modest productivity improvements, as firms focus on cost-cutting rather than leveraging technology for growth. “There has been an inefficient targeting of automation,” says MIT’s Daron Acemoglu, co-author of the study.
The research estimates that automation accounts for 52% of the increase in income inequality from 1980 to 2016, with 10 percentage points due to replacing higher-earning workers. This selective automation has negated 60-90% of potential productivity gains. “It’s one of the possible reasons productivity improvements have been relatively muted,” Acemoglu notes.
The paper, “Automation and Rent Dissipation,” was published in the May issue of the Quarterly Journal of Economics. Co-authors include Acemoglu of MIT and Pascual Restrepo from Yale University. Their analysis of data from various sources reveals that higher-earning workers, typically in the 70th-95th salary percentiles, are most affected by automation.
Acemoglu highlights that about one-fifth of the increase in income inequality is due to firms targeting wage premiums. “Automation is an engine of economic growth, but it creates significant inequalities,” he comments. The study underscores a fundamental choice for companies between profitability and productivity.
Acemoglu emphasizes that reducing costs does not equate to increased productivity. “Those two things are different,” he says. The study suggests that productive and non-productive automation are often mixed, and the implications of this need more recognition.
The research does not argue against automation itself, which can enhance productivity and lead to more employment. Acemoglu believes understanding the history of U.S. automation can clarify its tradeoffs. “The important thing is whether it becomes incorporated into people’s thinking,” he says, hoping the study influences perspectives on automation’s role in inequality and productivity.
Original Source: news.mit.edu
