Robots are pushing low-skilled workers off the career ladder, making it harder for them to move up, move on, or make more money, a new study finds.

The co-authored study from Wharton marketing professor Pinar Yildirim also reveals the provocative effects of career mobility on politics. As automation worsens job prospects in an area, the long-term investments in housing and education decrease, and those residents increasingly vote for candidates with populist agendas, such as Donald Trump.

“Looking particularly at the 2016 elections, in regions that were heavily impacted by manufacturing and experienced a decline [in career values], we see a turn towards the GOP candidate,” Yildirim said during an interview with Wharton Business Daily. (Listen to the podcast.)

The paper, “Automation, Career Values, and Political Preferences,” was written with Maria Petrova, economics professor at Pompeu Fabra University; Gregor Schubert, finance professor at the University of California, Los Angeles’ Anderson School of Management; and Bledi Taska, vice president and head of analytics at SkyHive.

The scholars analyzed the resumes of approximately 16 million people to track their careers over 16 years — from 2000 to 2016. They looked at how the individuals transitioned to future jobs and the earning potential of those jobs, two factors they combined into a “lifetime career value.” Although wage growth slowed during that period in general, the researchers determined that the average career value went down in part because of the reduced upward mobility toward better-paying jobs, caused by automation and robotization.

“In regions that were heavily impacted by manufacturing and experienced a decline, we see a turn towards the GOP candidates.”— Pinar Yildirim

Robots Are Taking Over Jobs and Displacing Careers

The numbers are significant: One additional robot per 1,000 workers decreased the average local market career value by $3,900 between 2004 and 2008, and by $2,480 between 2008 and 2016, corresponding to 1.7% and 1.1% of the average career values from the year 2000. In commuting zones that had more automation, the average career value declined even further.

The decline was more pronounced for low-skilled workers who, when displaced by automation, had difficulty finding similar or better jobs at the same rate of pay or higher. In contrast, high-skilled workers with a college education were able to mitigate the effect of automation because they had better career prospects.

“It could be that some of the lower-skilled individuals, as their wages start to go down or their occupational mobility starts to go down, lose their bargaining power,” Yildirim said. “They don’t necessarily have better outside options, so they’re perhaps willing to accept a lower wage in another industry.”

Automation has negative impact on career mobility beyond the immediate sectors where robots are adopted. The researchers found spillover to other industries, such as service and retail. As workers make less money, they have less money to spend in their communities.

“They cannot go to the restaurants, they cannot get a haircut as comfortably as they used to,” Yildirim said.

“Our findings highlight that it is essential to think about the effects of automation from a broader perspective.”— Pinar Yildirim

Automation’s Impact on Jobs Could Change Voter Behavior

Curious about how automation influences socioeconomics, the researchers also analyzed investments in housing and education, and any changes in voting patterns in the regions most affected by it, including the Rust Belt. They found that as career values increased, there was a 1.1% increase in people getting a higher education. But as career values declined, there was a 13% drop in housing permits as the demand for housing was expected to downshift.

But another salient finding was how career value changed political behavior. Where career values increased, there was a decrease in votes for Trump in 2016. Where career values decreased, there was a slight increase in votes for Trump. Automation did not increase votes for Republicans Mitt Romney and John McCain in previous elections, suggesting that other factors, such as choice of candidate, also play a role.

The scholars said the findings align with the theory that economic shocks prompt changes in political preferences. Right-wing and populist candidates have become more prevalent in recent years as financial crises, identity politics, and hostility toward immigrants have grown.

“The hypothesis that economic changes affect voting is almost as old as the modern political economy literature. However, this hypothesis implies important heterogeneity: People who are the most affected by the shock should be people who change their voting behavior in response. This is a sanity check that the literature rarely tests for,” the authors wrote.

The study was conducted using data before the 2022 release of ChatGPT, which has ushered in the widespread availability of artificial intelligence. Yet the scholars think updated research might impact broader groups of industries and workers as white-collar jobs are also replaced by AI, reducing the need for middle managers and eroding the middle class.

“Overall, our findings highlight that it is essential to think about the effects of automation from a broader perspective — focusing on industries beyond those that are most immediately impacted and beyond the immediate occupation of workers,” the authors wrote.