Job churn has been at historic lows. AI could change that.

How often have you heard business leaders, management consultants and think tank gurus say something along the lines of “We’re living in an age of unprecedented technological change and labor market upheaval?”

Yet economic history suggests the U.S. labor market over the past several decades has been unusually stable. But there are gathering signs that a big shift in the labor market is coming. Senior economics contributor Chris Farrell spoke to Marketplace host Sabri Ben-Achour to discuss more. The following is an edited transcript of their conversation.

Sabri Ben-Achour: The job market of today and of the recent past, it doesn’t seem particularly stable. I mean, taxi drivers getting displaced by ride-hailing apps, factory workers losing jobs to automation, tollbooth collectors replaced by electronic tolls — that doesn’t seem that stable.

Chris Farrell: So the U.S. economy is dynamic, and there’s plenty of job churn. The internet, software advances and other innovations have pushed companies to overhaul the way they do business. That said, the shifts in job categories in recent decades, you know, aren’t extraordinary by historical standards. Far from it.

Ben-Achour: So how do we know this counts as stable historically?

Farrell: OK, so there’s a recent study by three economists, including former Treasury Secretary Larry Summers. And they use a measure of occupational titles, and they found the years 1990 to 2017 were the most stable period in the history of the U.S. labor market — and that’s going back 150 years. And their results echo an Information Technology and Innovation Foundation report from 2017, and it looked back at 165 years of U.S. history and highlighted that the U.S. labor market wasn’t experiencing particularly high levels of occupational churn, you know, defined as new occupations being created while older occupations were destroyed. Instead, levels of occupational churn were at historic lows.

Ben-Achour: Yeah, that’s wild. I mean, what we think of as unstable and change, people had it worse back in the day. What was going on in the not-stable times?

Farrell: OK, so key is the emergence of general-purpose technologies, which is something like steam power and electricity. General-purpose technologies, they drive these massive economic changes and is a technology that’s widely used and encourages complementary innovations across many industries. So you take the period 1880 to 1960: In 1880, 41% of all workers were employed as farmers or farm laborers. By 1960, only 6% of employment was in agriculture. And the three-decade period 1940 to 1970, that was the most volatile labor market period in U.S. history. Agricultural employment was still disappearing. Manual labor was increasingly moving to factories, and clerical and administrative work — boy, that expanded rapidly.

Ben-Achour: So what kind of lessons can we draw from that for the current and future job markets?

Farrell: Summers and his co-authors suspect that AI, artificial intelligence, is the next general-purpose technology. Now, if they’re right, the labor market disruption would be far greater than anything we’ve witnessed since the 1970s. And they highlight four job market trends that reflect the impact of technology, including AI, and their calculations show that the post-pandemic labor market churn has been greater than anything that we’ve seen since the 1970s.

Ben-Achour: Wow. So we could be entering a period where job insecurity gets worse?

Farrell: So here’s my takeaway: The risk is real that too many employees will feel the downside risk of AI-led creative destruction in coming years. The need is for leaders to focus on developing a better safety net that allows, you know, to exploit the opportunities being generated, but at the same time helping displaced workers find new, rewarding work with good wages and benefits.