America’s Productivity Boom Is Getting Hard To Ignore
A year ago, economists were carefully drawing comparisons between today’s U.S. economy and the mid-1990s, when a surprise burst of productivity growth allowed the economy to ignite without triggering inflation. They were hopeful but hedging their bets.
They can be done as a cover.
New data released Thursday by the Bureau of Labor Statistics show that nonfarm business productivity grew at an annualized rate of 2.8% in the fourth quarter of 2025, well above expectations. More importantly, the revisions from previous quarters painted a picture of a more efficient economy for longer than anyone imagined. Third-quarter productivity growth was revised upward to 5.2 percent – the strongest quarterly gain in five years – and hours worked were revised downward in several previous quarters, meaning workers produced more per hour than previous data showed.
Once the dust settled, the annualized productivity growth rate for the entire current business cycle – dating back to the fourth quarter of 2019 – was revised upwards from 2.0% to 2.2%. This matches the long-term average dating back to 1947 and far exceeds the 1.5% rate of the slow cycle from 2007 to 2019 that produced a decade of productivity pessimism.
“Are we finally seeing AI in productivity data? asked Jason Furman, a Harvard economist and former chairman of the Council of Economic Advisers under President Obama, on Thursday. He noted that productivity is now 2.2% higher than the Congressional Budget Office’s pre-pandemic forecast, meaning the economy has not only regained lost ground but also surpassed its former trajectory. Furman, who was skeptical of productivity gains, admitted he was wrong.
“The question of productivity”
By the late 1990s, Federal Reserve officials had a name for the phenomenon that confounded their models. They called it “The Productivity Thing” – a mixture of fear and uncertainty about forces they could observe but couldn’t fully explain.
Edward Kelley, then the longest-serving Fed governor, captured the mood in a 1999 interview with Barron’s. “Here we are, after all this time, with a very strong growth rate, a very low unemployment rate, strong job creation and very low inflation,” he said. “There’s a new dynamic there.” He described economics as having moved “from a rather traditional and well-understood era into new and uncharted territory.”
What made the end-of-cycle productivity surge so striking then – as now – is that it wasn’t supposed to happen. Conventional economic theory holds that productivity growth slows in the later stages of an expansion, as the easiest efficiency gains have already been made. Eight years after the 1990s expansion began, productivity jumped 4.6 percent in the fourth quarter of 1998. Fed officials marveled publicly. We are in a similar situation today.
The boom of the 1990s had a clear catalyst: computers and the early Internet, which had been integrated into workplaces for years without generating measurable efficiency gains—something economist Robert Solow captured in his famous 1987 observation that “the computer age is visible everywhere except in productivity statistics.” Around 1994, the winnings arrived. Semiconductor manufacturing improved, companies that had spent years learning how to use information technology began to reap the rewards, and productivity soared for the better part of a decade.
Today’s equivalent is artificial intelligence. Like computers in the late 1980s, AI tools have been widely available for several years without generating obvious gains in overall productivity data. The question economists are asking – and which the new data is beginning to answer – is whether we are now at the 1994 moment: the point where adoption becomes broad enough and deep enough to show up in the numbers.
Tariffs and immigration restrictions boost productivity
This is where the current story unexpectedly deviates from the 1990s storyline.
Critics of the Trump administration’s trade and immigration policies predicted that tariffs would make the economy less efficient, distorting the allocation of resources and raising costs. Tighter immigration enforcement, the argument goes, would reduce labor supply and slow growth. So far, productivity data tells the opposite story.
Companies facing higher production costs due to tariffs have had a strong incentive to seek efficiencies wherever they can. A tighter labor market, shaped in part by reduced immigration, has pushed companies to invest in technological and process improvements that allow them to produce more with fewer workers. When labor is expensive and goods cost more, getting more output from each hour worked is not a choice: it is a survival strategy.
Alan Greenspan identified a version of this dynamic during the inflation decline of the early 1990s, when companies losing pricing power were forced to cut costs. “We will necessarily tend to get an increase in productivity because it will be imposed on the system,” he theorized at a Federal Reserve meeting. The mechanism today is different – it is input costs rather than loss of pricing power that determine the situation – but the logic is the same.
This is not the story most economists tell when implementing policies. However, this may be the story the current data tells.
The Fed may have to abandon its fear of rapid growth
If the productivity boom is real and lasting, it will have significant implications for the Federal Reserve.
A more productive economy can grow faster without generating inflation. This means that strong nominal growth, rather than being a harbinger of overheating, could simply reflect a higher speed limit imposed on the economy. And that creates the possibility of faster wage growth without driving up prices.
This is precisely the situation Greenspan found himself in in the late 1990s. After aggressively raising interest rates in 1994 and 1995 to anticipate inflation – which ran into enormous political pressure – Greenspan became convinced by the mid-1990s that productivity data meant the economy could sustain faster growth without triggering price increases. It resisted further tightening and was vindicated. Fed Chairman Jerome Powell spoke admiringly of Greenspan’s “courage” in navigating this period.
As 2024 began, Powell was still cautious about the productivity story, saying she thought it would “collapse and get back to where we were.” The new data suggests that this assumption may have been too conservative. The Fed will be reluctant to officially revise its estimates of potential output. The central bank tends to act slowly on such judgments. But the direction the evidence points is clear.
There are caveats. Productivity data is notoriously fuzzy and heavily revised, a fact that this report illustrates, since upward revisions are themselves part of the story. Skeptics argue that AI’s gains could be more limited than those of computers, focused on office work rather than the broader economy. And as Kelley warned in 1999, even in a strong new era, “there are limits somewhere.”
But it wasn’t until 1999 that economists widely accepted that the productivity boom of the 1990s was real, even though it had been going on for years. The current boom has grown just as quietly – and the data continues to come in force and be revised more forcefully. At some point, caution becomes a form of error in itself.
