U.S. News: Why the GDP-Jobs Split? Look to AI

“The economy is either booming or on the brink of recession. Honestly, you could make the case for either.

 

The broadest measure of economic output, inflation-adjusted gross domestic product, grew at a blistering annual rate of 3.8% in the third quarter, a Federal Reserve Bank of Atlanta model predicts.

 

Yet employment and hours worked barely grew in the three months through August. September data weren't released Friday because of the government shutdown. Payroll processor ADP estimates private payrolls shrank in September.

 

GDP and employment occasionally go in different directions, but rarely by this much.

 

There are several possible explanations. First, consumption, investment and trade data that go into GDP, and employment data, for past months might be wrong and could be revised.

 

Second, the gap is anomalous, and will probably close -- through slower GDP growth, more jobs, or both. Neil Dutta of Renaissance Macro Research said consumer spending has been driven by an unsustainable drop in savings and, according to credit-card data, has already softened in September.

 

The third possibility is the most intriguing: A productivity boom might be in the works, which means the economy could grow faster with less inflation.

 

Increased output with no increase in hours means labor productivity grew around 3.5%, annualized, a robust pace.

 

Because the divergence between GDP and employment likely won't be sustained, neither will that productivity performance. Even so, including that third-quarter estimate, productivity has grown about 2% annualized for two years now. That's a decent pickup from the 1%-to-1.5% rate that prevailed in the decade before the pandemic and hints that something is going on.

 

That something very likely has to do with technology. One candidate is generative artificial intelligence, embodied in large language models, or LLMs, such as ChatGPT, released less than three years ago. Adoption has been remarkably fast: In June, Gallup found that 19% of employees used AI a few times or more a week. Walmart recently said it would keep employment flat in the next three years as AI would transform "literally every job."

 

Still, it took decades for electricity and computers to transform business and the economy, and the same might be true of AI.

 

One widely cited report published by Nanda, an AI initiative at the Massachusetts Institute of Technology, found that 95% of 52 organizations surveyed had a zero return on their AI initiatives. This suggests AI has yet to make enough impact to raise the economy's aggregate productivity.

 

But LLMs are building on existing investments in e-commerce, software, big data, cloud computing and machine learning. A recent Goldman Sachs analysis found that the pickup in productivity has been concentrated in technology and related sectors such as scientific research, engineering and consulting. It's also more evident among the fastest-growing "superstar" companies.

 

"While it is probably premature to ascribe all of the acceleration to AI, these industries have some of the strongest use cases for AI technology," Goldman noted.

 

The vast sums spent by companies such as Oracle, Amazon.com and Meta Platforms on the microprocessors, data centers and power generation needed to run AI look to many like a bubble. But bubbles can still generate genuine productivity benefits.

 

That was certainly true of the internet-tech boom a generation ago. Economists at Citigroup recently compared the two episodes. The prior boom began in 1995 with the initial public offering of Netscape and the full commercialization of network infrastructure developed by the National Science Foundation.

 

In the first five years of the boom, annual internet-related investment rose by a massive 1.25% of GDP, Citi found. Annual productivity growth averaged 2.9% from 1995 to 2004, double the rate of the prior two decades. At that rate, per capita incomes double in 24 instead of 47 years. The productivity boost lifted annual economic growth to 3.3%.

 

Citi estimates annual spending on AI equipment has risen by 0.9% of GDP a year since 2023, a bigger surge than at the equivalent point of the 1995-2004 boom.

 

Nathan Sheets of Citi said GDP growth has been boosted by the demand-side fuel of all that AI investment, not by its supply-side productivity benefits. Still, mapping today's investment to the internet era suggests a productivity boom "within the next few years," Citi said.

 

Several academics have estimated AI's boost to productivity by measuring how many tasks AI could automate and the resulting cost savings. Citi concludes from those estimates that AI could raise productivity growth by 0.5 to 1.5 percentage points a year -- less than the internet except in the most bullish scenario.

 

That would not only boost growth, it could put downward pressure on inflation by reducing companies' labor costs. But a recent report by economists at Deutsche Bank throws cold water on that prospect.

 

They note that the tech-driven 1990s productivity boom wasn't the only thing holding down inflation. Baby boomers and women were still flooding into the labor force. The end of the Cold War, the North American Free Trade Agreement and the Uruguay Round of world trade negotiations had lowered trade barriers, squeezing goods' price inflation.

 

"The shift towards de-globalization of the U.S. economy will reverse this earlier disinflation," Deutsche Bank economists predict. Meanwhile, "an aging global population along with a turn away from immigration flows. . .will lead to tighter US labor markets and structurally higher cost pressures from labor."

 

That shift could explain another part of today's data puzzle: A weak labor market and rising productivity haven't done much to reduce inflation.” [1]

 

1. U.S. News: Why the GDP-Jobs Split? Look to AI. Ip, Greg.  Wall Street Journal, Eastern edition; New York, N.Y.. 04 Oct 2025: A2.