The Payroll Illusion: Why U.S. Job Growth Is Weaker Than It Looks
I. The headline still looks acceptable
The April jobs report looked respectable on the surface. The U.S. economy added 115,000 jobs, the unemployment rate remained unchanged at 4.3%, and March was revised to a stronger +185,000. After a weak February, when payrolls were revised down to a loss of 156,000 jobs, the last two monthly prints were enough to create the appearance of stabilization.
That is the first layer of the story. The labor market is not collapsing in the monthly headline data. It is not printing the kind of immediate, disorderly losses normally associated with a fully developed recession. On a headline basis, the report can still be presented as soft, but manageable.
The problem is that the headline number is no longer the right place to look. A labor market can look stable in monthly payroll prints while the underlying employment base quietly narrows. That is exactly what appears to be happening now. The issue is not simply the number of jobs created in April. The issue is where those jobs are being created, where they are not being created, and how far the broader trend has deteriorated since 2022.
II. The 12-month trend is the real signal
Over the last 12 months, total nonfarm employment has barely moved. Payrolls stood at 158.736 million in April 2026, compared with 158.485 million in April 2025. That is a gain of only 251,000 jobs over a full year, or roughly 21,000 jobs per month across the entire U.S. economy.
This is where the picture changes. A gain of 21,000 jobs per month is not a normal expansionary profile for the U.S. labor market. It is closer to stagnation. It is also materially weaker than the previous 12-month period, when average monthly job creation was around 80,000. Before that, from January 2023 through April 2024, the economy was still adding roughly 200,000 jobs per month on average.
The comparison with the pre-COVID period is even more important. In 2017–2019, before the pandemic shock and the subsequent recovery distortions, the U.S. economy added roughly 177,000 jobs per month. The 2021–2022 period should be treated separately because it was dominated by the recovery from the multi-million job losses caused by COVID. But once that recovery phase ended, the slowdown became consistent, broad, and increasingly difficult to dismiss.
III. Revisions matter more when growth is this weak
Monthly payroll data should always be treated with caution, but that caution becomes much more important when the reported growth rate is already close to zero. When the economy is adding 200,000 jobs per month, even a meaningful revision does not necessarily change the entire macro picture. When the economy is adding only 21,000 jobs per month, a normal benchmark revision can erase the whole gain.
This is not theoretical. The 2025 benchmark revision reduced the seasonally adjusted March 2025 level of total nonfarm employment by 898,000 jobs. On a not-seasonally-adjusted basis, the downward revision was 862,000 jobs. BLS also revised the change in total nonfarm employment for 2025 from +584,000 to only +181,000.
That is why the current payroll profile should not be taken at face value. The economy may appear to be creating 100,000–150,000 jobs in some monthly reports, but the annual benchmark process has already shown that a large part of the apparent resilience can later disappear. In a labor market this weak, the margin of safety is gone.
IV. The sector map shows the problem
The most important signal is not the total number of jobs. It is the sector structure. Over the last 12 months, only a small group of sectors continued to add jobs. Health care and social assistance added about 55,000 jobs per month on average. Accommodation and food services added about 9,000. Other services, construction, and arts, entertainment, and recreation added only a few thousand each.
Together, those five sectors account for roughly 55.1 million jobs, or about 34.7% of total U.S. payroll employment. In other words, just over one-third of the employment base is still producing positive job growth, while most of the broader economy is either flat or contracting.
This is where the labor market begins to look very different from the headline. The expansion is no longer broad. It is no longer supported by most cyclical sectors. It is increasingly concentrated in defensive and service-heavy areas, especially health care and social assistance. The chart makes the issue immediately visible: one sector is doing most of the work, while much of the economy is failing to contribute.
V. Health care is carrying the expansion — but not because the economy is strong
Health care and social assistance is now the entire labor-market story. Over the last 12 months, this sector added roughly 657,000 jobs. Without it, total U.S. payroll employment would have declined by roughly 405,000 jobs, or about 34,000 jobs per month. That single adjustment changes the interpretation of the entire jobs report. The U.S. economy is still creating jobs on paper, but outside health care, the employment base is already shrinking.
This is not a small technical detail. A labor market that creates jobs because manufacturing, transportation, finance, information, construction, retail, and professional services are broadly expanding is one thing. A labor market that creates jobs because one defensive sector keeps hiring while most cyclical sectors stall or contract is something very different. The first profile points to broad economic strength. The second points to narrowing resilience.
Health care also has its own structural demand cycle. The sector can expand even when the broader economy weakens because its demand is driven less by discretionary spending and more by demographics, aging, chronic conditions, insurance coverage, public programs, and basic care needs. The U.S. population aged 65 and older continues to rise, and BLS expects health care and social assistance to remain one of the strongest employment-growth areas over the next decade, largely because of aging and chronic-care demand.
That makes health-care employment both important and misleading. It is important because it is genuinely supporting the headline payroll number. But it is misleading because it does not necessarily confirm a healthy expansion in the productive or cyclical parts of the economy. A hospital, clinic, nursing facility, home-health provider, or social-assistance organization may need more workers because the population is aging, not because the economy is accelerating.
In that sense, health care is acting less like a growth engine and more like a demographic stabilizer. It is preventing the headline labor market from looking as weak as the rest of the sector map already looks. That is precisely why the payroll headline is so deceptive: the economy appears to be adding jobs, but the job creation is increasingly concentrated in a sector whose expansion can reflect structural pressure rather than broad economic momentum.
VI. The weakness is broad, not isolated
The negative side of the sector table is more important than the positive side. Government employment declined by about 22,000 jobs per month over the last 12 months. Information lost about 8,000 per month. Financial activities lost about 7,000. Transportation, warehousing, and logistics lost about 6,000. Manufacturing also lost about 6,000. Education, professional and business services, and mining were negative as well.
This is not one weak pocket. It is a broad labor-market slowdown across sectors that normally matter for business-cycle momentum. Manufacturing speaks to industrial demand. Transportation and warehousing speak to goods movement and logistics. Financial activities and professional services speak to corporate confidence, capital formation, and white-collar employment. Information speaks to technology, media, communications, and digital infrastructure.
Education deserves special attention. Private educational services lost about 38,000 jobs over the last 12 months, or roughly 3,200 jobs per month. That is not a large number compared with government, information, finance, or transportation, but it carries a different message. Education is not only a cyclical sector. It is part of the country’s human-capital infrastructure.
A decline in education employment does not automatically mean that the future population will be less educated. The signal is more subtle. It may reflect lower enrollment, school consolidation, pressure on private colleges, affordability problems, demographic shifts, online delivery, or changing demand for traditional education. But even with those caveats, the direction is concerning. A growing population should normally require a larger education system, or at least one that is not shrinking in employment terms.
The longer-term comparison makes this more important. Over the last decade, private educational services employment has grown only modestly. Over the last 20 years, it has expanded, but not in a way that suggests a strongly growing education base relative to the size and complexity of the economy. At the same time, undergraduate enrollment has been under pressure since the 2010 peak, and public school enrollment is projected to decline into the next decade.
This turns education into a structural warning, not just another negative line in the payroll table. If health care is expanding because the population is aging, while education is shrinking or stagnating because the younger and student-facing part of the system is under pressure, the labor-market map begins to tell a deeper demographic story. The economy is adding workers where the population is older and care needs are rising, while losing workers in areas tied to training, productivity, and future labor quality.
The comparison with 2017–2019 makes the broader deterioration clear. Professional and business services added about 30,000 jobs per month in that period. Transportation and warehousing added about 18,000. Construction added about 20,000. Financial activities added about 13,000. Manufacturing added about 12,000. Education added about 4,000. Today, most of those engines are either flat or negative, and even education has moved from modest growth to contraction.
VII. A second early recession warning
The current data do not prove that the U.S. economy is already in recession. That would require a broader set of indicators: output, income, spending, industrial production, credit conditions, corporate profits, and final demand. But the labor-market structure is now consistent with at least a recession-like profile. It is not the headline number that matters most. It is the breadth.
This is also the second early warning signal we have seen in recent months. In March, we highlighted the abrupt downward reversal in the 10-year/2-year Treasury spread. At that point, the concern was not simply the level of the spread, but the character of the move. The spread had started to reverse downward in a way that suggested a high probability of moving back toward negative territory over the next one to two years.
That matters because the 10Y–2Y spread is one of the most watched early recession indicators. The latest available readings still show the spread positive, but the risk we flagged was about the direction and structure of the move, not only the current level.
The labor-market data now add a second layer to that warning. The bond-market warning came from the rates structure. The payroll warning comes from employment breadth. These are different signals, but they point to the same underlying issue: the economy may still look stable in headline data, while its internal structure is weakening.
The key point is breadth. When roughly two-thirds of major sectors are no longer creating jobs, the labor market is not healthy, even if the headline number remains positive. The economy can still report job gains because one large defensive sector continues to expand, but that does not mean the employment base is strong.
The same logic applies to the yield curve. A positive spread does not automatically mean the risk is gone, just as a positive payroll number does not automatically mean the labor market is healthy. In both cases, the structure matters. The 10Y–2Y spread warned that the bond market was moving back toward a more recession-sensitive configuration. The payroll-sector map now shows that employment growth has narrowed to a recession-like profile.
This is how early warnings usually appear. They do not start with a collapse. They start with narrowing, reversals, compression, and deterioration under the surface. In March, the warning came from the Treasury curve. In May, it came from the labor market. Neither signal alone is conclusive, but together they deserve attention. The economy is not yet sending a single decisive recession signal. It is sending a cluster of early warnings.
VIII. The historical warning
The labor market has been slowing since 2022. At first, the slowdown looked like normalization after the post-COVID hiring surge. That interpretation made sense for a while. The economy had recovered millions of jobs, labor shortages were extreme, and a slower pace of job creation was both expected and necessary.
But the current situation has moved beyond normalization. The trend is no longer simply slower than 2021–2022. It is now weaker than 2017–2019, weaker than the previous 12-month period, and close to flat over the last year. The end of 2025 even produced the first negative six-month moving-average episode since 2008, excluding 2020 for obvious reasons.
Unlike 2008, that negative trend did not immediately become entrenched. The last two monthly reports improved the headline picture. But that is not enough to remove the concern. In the current setup, the labor market does not need to collapse to send a warning. It only needs to remain this narrow.
That is why the present setup should be treated with caution. If the economy were creating 150,000–200,000 jobs per month across a wide range of sectors, the story would be very different. Instead, the economy is producing weak aggregate growth, relying heavily on health care, and showing contractions across several cyclical and structurally important sectors. This is not a classic expansionary labor-market profile.
IX. Conclusion — the payroll illusion
The U.S. labor market is not collapsing in the headline data. That is precisely why the risk is easy to underestimate. April showed a positive payroll number, March was revised higher, and the unemployment rate remained stable. On the surface, the report looked soft, but not alarming.
Under the surface, however, the picture is much weaker. The economy added only about 21,000 jobs per month over the last 12 months. Excluding health care and social assistance, it lost about 34,000 jobs per month. Most major sectors are no longer contributing to employment growth. Several key cyclical sectors are already contracting.
The structure of the labor market now matters more than the headline. Health care is expanding, but its growth is supported by powerful demographic and structural forces: aging, chronic-care demand, insurance coverage, public programs, and basic care needs. That makes it a stabilizer, not necessarily a confirmation of broad economic strength. At the same time, education, manufacturing, transportation, finance, information, and professional services are no longer showing the kind of employment momentum normally associated with a healthy expansion.
This is the payroll illusion. The headline still shows job creation, but the job creation is no longer broad. The labor market is being held together by one defensive sector while much of the rest of the economy has already stalled. Combined with the March warning from the 10Y–2Y Treasury spread, the employment data now add another early signal that the economy’s internal structure is weakening.
This does not guarantee a recession. But it does suggest that the U.S. labor market is much weaker than it looks. The warning is not in the collapse of the headline number. The warning is in the narrowing of the base, the dependence on health care, the contraction across major sectors, and the emergence of a broader early-warning cluster.
Publications
GITT: Week 10 – Cooling Prices, Persistent Constraints (May 8)
The Gulf Shock: How One Corridor Can Reprice the World (May 5)
The U.S. LNG Moment: How the Qatar Shock Rewired Global Gas (April 26)
GITT: The Framework and Week 8 (April 24)
Gasoline: The Pump Shock Nobody Is Ready For (April 23)
The Architecture of a Global Economic Crisis:
Part 2: The Hidden Layer: Petrochemicals
Part 3: When It Reaches the Real Economy
Part 5: Financial System Impact
Part 6: Early Signals: Stress Already Visible
March 15: Energy Crises – Historical Scale (open article)
March 18: Strait of Hormuz Risk: How a Middle East War Could Trigger a Global Supply Shock
March 19: RAS LAFFAN: GLOBAL ENERGY SHOCK: Part 1
March 19: Dutch TTF – Technical Forecast
March 25: Who Blinks First? The Energy War Reshaping Markets
April 3: ABU DHABI: SYSTEM STRESS EXTENDS: Part 2
Tags:
#StraitOfHormuz #MiddleEastCrisis #GlobalEconomy #SupplyChain #EnergyCrisis #OilPrices #NaturalGas #LNG #Petrochemicals #Aluminum #Fertilizers #FoodSecurity #Inflation #Stagflation #Semiconductors #Helium #Commodities #GlobalMarkets #SP500 #MarketCrash #EconomicOutlook #Geopolitics #InvestingAngles
