The puzzle of strong GDP growth and weak employment

Stephen Stanley | January 9, 2026

This material is a Marketing Communication and does not constitute Independent Investment Research.

One of the biggest puzzles of 2025 was that real GDP growth outperformed consensus while the labor market weakened much more than projected. Several Fed officials have noted this divergence and are waiting to see how the inconsistency is resolved. Economic growth and job gains presumably will come back into closer alignment in 2026. If not, it would suggest that productivity increases have accelerated sharply, which would open another set of questions.  If efficiency gains have risen that much, then the FOMC is not necessarily supposed to cut rates in response to relatively soft labor market data.

Strong growth, weak employment

Entering 2025, the consensus of forecasters called for 2% growth for real GDP, not only for the year but in each individual quarter.  As it turns out, the quarter-to-quarter volatility last year was immense, owing to a large degree to trade flows associated with Liberation Day.  Imports poured into the U.S. in the first quarter, as firms rushed to get supplies and goods into the country before tariffs were imposed.  Real GDP contracted in the first quarter, driven by a massive widening in the trade gap.  Net exports reversed in the second and third quarters, which, along with solid gains for consumer spending and business fixed investment, drove hefty real GDP advances of 3.8% in the second and 4.3% in the third.  Real final domestic demand—GDP minus trade and inventories—has been less volatile, posting quarterly annualized increases of 1.4%, 2.4%, and 2.9%, respectively, over the first three quarters of the year, an average of 2.2%, modestly firmer than the consensus forecast coming into the year.

The third quarter GDP results, which were delayed by nearly two months by the federal government shutdown in the fall, were especially robust.  Nominal GDP growth surged to an 8.2% annualized pace, the best performance since the post-COVID rebound in 2022.

In contrast, labor market data steadily deteriorated over the course of the year.  Payrolls rose by well over 200,000 a month in the last four months of 2024. To be fair, those readings will be revised downward noticeably when the new benchmark adjustment is incorporated in early 2026. A string of substantial downward revisions in the late spring and summer dramatically altered the view of the health of the labor market.  Jobs barely rose at all, on balance, in the middle four months of the year with gains of 13,000 a month, though, to be fair I believe that there is some downward seasonal bias to the readings for the summer months, and then rebounded modestly late in the year.  When all is said and done, including the forthcoming benchmark adjustments that will apply to the first three months of the year, job gains for 2025 will likely be less than 50,000 month.

Parallel to the payroll figures, the unemployment rate backed up over the course of 2025, moving up by close to half a percentage point.  This took the jobless rate up from slightly below the FOMC’s estimate of the economy’s long-run equilibrium to somewhat above.

A pressing question for 2026 is how this divergence will be reconciled.

Measurement error

One possibility is that the gap between strong growth and soft employment is merely a function of measurement error.  That is, GDP growth could get adjusted noticeably lower or employment could get revised higher, bringing the two sets of data into closer alignment.

It seems unlikely that employment gains will be revised substantially higher.  The last two annual benchmark adjustments have been negative and historically large.  Chair Powell has repeatedly argued, leaning on Fed Board staff analysis, that the next annual benchmark revision is also likely to be negative and massive.  I am skeptical that this will be the case, but, even so, it seems unlikely that employment will be bumped up by enough to match the solid 2025 GDP growth figures.

Conversely, GDP could be revised lower. This seems more plausible, as there are a tremendous number of assumptions and preliminary data points used in real-time GDP estimates that are eventually replaced by more complete data that are compiled with a significant lag.  That being said, the annual GDP revisions over the past few years have been minimal, so there is no presumption that the numbers, as they currently stand, are grossly inaccurate.

Productivity acceleration?

If not measurement error, the most logical explanation for the divergence between strong economic activity and tepid job gains would be a sharp acceleration in productivity growth.  This is certainly a plausible hypothesis.  There has been much talk about an AI-driven jump in efficiency.  I believe the excitement over an AI-focused productivity boom is premature, but the idea is worth exploring.

Based on the way that the statistics are put together, there are two ways in which productivity growth can pick up. Productivity is a residual, calculated by comparing the increase in economic activity with the change in hours worked.  The first scenario for a productivity pickup is that businesses economize on workers.  Consider the example where a firm loses an employee and is unable or unwilling to replace them.  In that case, management may ask the remaining employees to work harder to cover for their departed former colleague.

Technically, this scenario can yield a rise in productivity, but it is likely to be unsustainable over time.  The remaining employees may be willing and able to buckle down and work harder for a time, resulting in a statistical bump to productivity (output divided by hours worked) but over an extended period, the workers would likely burn out.

In the current context, this dynamic seems quite plausible.  Businesses have frequently discussed that they have been operating cautiously in light of policy-related uncertainty, postponing decisions on investment and hiring until there is greater clarity.  Thus, it would not be surprising to find that firms were asking their workforce to temporarily close ranks and do more with less. If this scenario is the main driver of productivity gains in 2025, then one should expect to see improved employment figures in 2026 and a commensurate moderation in productivity.

Alternatively, productivity growth can accelerate because innovation is driving permanent efficiency gains.  Of course, this discussion in the current context would revolve around AI. I see this coming over the next several years, but the adoption of AI as a way to transform business operations appears to be in the very early stages.  I expect productivity growth to pick up on a trend basis over the next five to 10 years, but I doubt that the real impact in 2025 was sufficient to explain the wide gap between solid GDP growth and tepid job gains.  In the 1990s, the last big technology boom, it was several years after the tech stock frenzy began before the aggregate productivity figures started to show a discernible pickup.

How should the FOMC respond?

Governor Waller was one of the first Fed officials to point out the discrepancy between healthy data on activity and weaker readings for the labor market.  He initially framed the divergence as likely a fleeting anomaly that would correct in relatively short order.  He felt that the data would break one way or the other, either validating the strength in activity, which would mean less need to cut rates, or underscoring the softness in the labor market, which would justify continued easing. Of course, the economic data calendar has been distorted since the government shutdown, so it is still possible that the activity and labor figures will reconcile relatively soon, but it seems increasingly likely that this divergence could prove more than a brief data anomaly.

If in fact the gap persists, the FOMC will have to decide the proper policy stance for such a confusing environment.  In the short term, confusion generally argues for a wait-and-see approach.  Now that Chair Powell and a number of Fed officials have assessed that the current policy stance is within the plausible range of neutrality, there is less of a compelling push to cut rates in the near term.  The unusual and inconsistent array of economic data likely is contributing to the FOMC’s inclination to pause in early 2026, as officials hope that the picture becomes clearer over the next several months.

If the divergence persists, suggesting that productivity growth has picked up on a trend basis, the implications for monetary policy are not entirely obvious.  In that scenario, the labor market could be persistently weak at the same time that efficiency gains hold down inflation.  Doves, most notably Governor Miran, have argued emphatically that this scenario would justify lower interest rates.

The counter to this argument is that a period of structural weakness in the labor market caused by AI-driven innovation cannot be effectively addressed by easier monetary policy.  Indeed, not only is the Fed not equipped to “fix” a period of structural softening in labor demand, but consensus economic theory dictates that it is not even supposed to try.  In effect, this scenario amounts to a rise in the long-run equilibrium unemployment rate, and the Fed should adjust its sights accordingly.

In any case, there is a far more upbeat narrative.  Higher productivity growth should translate directly into faster potential economic growth, which would presumably be correlated with higher interest rates.  Put another way, the sustained pickup in investment demand that would likely accompany a rise in trend productivity growth would, all else equal, drive interest rates higher to balance the savings/investment equation.

Empirically, the most recent analog to this scenario would be the 1990s, when a productivity boom propelled the economy forward.  In that timeframe, interest rates were quite high, as Minneapolis Fed President Kashkari pointed out on January 5.  The fed funds rate target spent the entire period in the range of 5%, with longer-term Treasury yields even higher.  It is also worth noting that the booming economy managed to generate ample demand for workers, even with productivity growth unusually high.  Of course, the late 2020s could play out much differently than the late 1990s, but my sense is that if we see the sort of high productivity scenario being widely discussed, the implications for the FOMC may not necessarily be as dovish as some are currently assuming.