What the Jobs Report Won't Tell You
What the Jobs Report Won’t Tell You
By The Unfiltered Wire | April 3, 2026
Friday morning, the Bureau of Labor Statistics releases the March employment situation summary. The headline number — nonfarm payrolls added or lost — will dominate financial media for roughly four hours before being replaced by something else. Traders will react, pundits will opine, and the Fed will be described as either “under pressure to cut” or “justified in holding.”
Most of that reaction will be based on incomplete data.
The headline payroll number is a lagging indicator derived from a single survey, filtered through statistical models that systematically smooth out turning points. It tells you where employment was, not where it’s going. If you’re making decisions based on the headline alone, you’re working with the least informative number in the report.
Here’s what actually matters.
The Two Surveys Problem
The BLS runs two separate surveys for the jobs report, and they frequently disagree.
The establishment survey (also called the Current Employment Statistics survey) samples roughly 119,000 businesses and government agencies. It counts jobs — specifically, payroll jobs. One person with two jobs shows up twice. This is the number that generates the headline.
The household survey (the Current Population Survey) samples about 60,000 households. It counts employed people. One person with two jobs shows up once. This is where the unemployment rate comes from.
When these two surveys diverge, it’s a signal worth paying attention to.
In 2023 and 2024, the establishment survey consistently reported job gains while the household survey showed employment essentially flat. Over that stretch, the establishment survey counted several hundred thousand more jobs than actually existed at the person level. The divergence was eventually confirmed by the annual benchmark revision in early 2025, which revised nonfarm payrolls down by over 800,000 jobs — the largest downward revision in fifteen years.
The mechanism: establishment surveys overcount during economic slowdowns because they catch businesses adding payroll for workers who are later reduced to part-time or let go. The household survey captures that reality faster.
Watch both numbers Friday. If they diverge again — establishment shows gains while household shows flat or negative — the headline is likely telling a different story than the underlying employment picture.
The Birth-Death Model Adjustment
The most important statistical adjustment most people don’t know exists is called the birth-death model.
Every month, the BLS estimates job creation from businesses that don’t yet appear in their survey frame — primarily new small businesses that haven’t been captured. They also estimate job losses from businesses that closed without notifying anyone. The net estimate gets added directly to the payroll number before publication.
In March of recent years, the birth-death adjustment has added between 100,000 and 200,000 jobs to the headline figure. This is not measured data. It’s a model estimate based on historical business formation and failure patterns.
The structural problem: the birth-death model assumes that current business formation and failure rates resemble historical averages. During economic turning points — when conditions are changing faster than the historical average would predict — the model systematically overstates job creation on the way down and understates it on the way up.
The 2024 benchmark revision confirmed this pattern. The excess estimated from the birth-death model in 2022-2023 accounted for a significant portion of the eventual downward revision.
You won’t see the birth-death adjustment prominently featured in any news report Friday. The BLS does publish it in the supplementary tables. The number to watch: if the month’s total payroll gain is close to the birth-death adjustment, you’re not really looking at measured job creation. You’re looking at a model estimate.
Full-Time vs. Part-Time Composition
The headline counts a full-time job and a part-time job equally. The composition tells a different story.
The BLS breaks down the household survey data into full-time and part-time employment. It also tracks multiple jobholders — people working more than one job simultaneously.
When full-time employment is declining and part-time employment is rising, it can produce flat-to-positive headline numbers that mask underlying labor market stress. A person who lost a full-time position and replaced it with two part-time jobs appears in the establishment survey as three jobs rather than one.
The multiple jobholders figure is particularly informative. When economic conditions are stable and income is sufficient, people don’t typically work multiple jobs by choice. Rising multiple jobholders, especially in the 35-54 age bracket, is a stress signal — it suggests that single-income households are having difficulty meeting expenses.
In 2024, multiple jobholders reached historic highs while full-time employment grew more slowly than the headline number suggested. That’s not a conspiracy — it’s how the measurement works. The question is whether the composition this month tells the same story.
Nominal Wages vs. Real Wages
The report includes average hourly earnings data, and this is where the framing most often misleads.
The financial press typically reports nominal wage growth as a sign of labor market strength. “Workers are seeing raises.” Technically accurate. Financially incomplete.
What matters is real wage growth — nominal earnings adjusted for price level changes. If wages grew 3.8% year-over-year while CPI ran at 3.2%, purchasing power increased by roughly 0.6%. If wages grew 3.8% while CPI ran at 4.5%, workers actually lost ground despite “earning more.”
The BLS publishes nominal earnings data in the jobs report. Real earnings data — using CPI deflators — gets released separately in the real earnings press release. It takes a few extra clicks to get there. Most media coverage doesn’t take those clicks.
There’s also a composition effect worth understanding. If higher-wage workers are laid off and lower-wage workers are hired — which is a common pattern in early-stage slowdowns — average hourly earnings can rise even if no individual worker got a raise. The average goes up because the sample shifted, not because wages went up.
Watch nominal earnings growth, then cross-reference against whatever CPI print is most recent. The spread tells you whether people are actually better off.
What the Fed Is Actually Watching
The Fed’s stated mandate is maximum employment and price stability. In practice, they publish guidance on which specific metrics they weight most heavily.
Prime-age labor force participation rate (ages 25-54) is the cleanest signal in the employment data. This metric strips out the distortions from an aging population — older workers retiring and younger workers extending education. If prime-age participation is rising, the labor market is tightening from genuine demand. If it’s falling, people are leaving the workforce — which is structurally different from being employed.
U-6 unemployment (the “broad unemployment” rate) includes people who want to work but have given up looking, plus people in part-time positions who would prefer full-time work. It runs several percentage points higher than the headline U-3 rate. When U-6 and U-3 diverge significantly, it suggests the headline rate is masking underemployment.
Wage growth in services — specifically services excluding housing — is the inflation component the Fed has had the most difficulty controlling. Services inflation is largely driven by labor costs. If wage growth in services remains elevated, the Fed’s inflation fight is not finished regardless of what’s happening in goods prices.
The Fed explicitly does not set policy based on a single month’s payroll number. What they’re watching is the trend: is participation recovering, is underemployment declining, is services wage growth cooling? One report doesn’t move the needle, but it updates the running tally.
What to Watch Friday
When the report drops at 8:30 AM ET, here’s the sequence:
- Headline number — context only, not a decision input by itself
- Household survey employment — does it confirm or contradict the establishment survey?
- Birth-death adjustment — is the gain mostly model-generated or measured?
- Full-time vs. part-time split — what’s driving any headline gain?
- Multiple jobholders — rising or falling?
- Average hourly earnings — nominal only; apply your own CPI adjustment
- Prime-age participation rate — the cleanest single number in the whole report
- U-6 — gap with U-3 widening or narrowing?
The headline number is the starting point, not the endpoint. It’s the number that moves markets in the first fifteen minutes. Whether that move is based on accurate information depends on whether traders did any of the above.
Most don’t. That creates a brief window where the price discovery is based on incomplete data, and a more complete read of the full release shows something different.
The jobs report is released monthly, revised twice (preliminary and final), and then subject to annual benchmark revisions. By the time the data is accurate, it’s often a year old. The lagging nature of the indicator isn’t an accident — it’s a consequence of how complex, slow-moving systems get measured.
Understanding the structure doesn’t make the number irrelevant. It makes the number legible.
The Unfiltered Wire publishes structural economic analysis through an incentives-first lens. Follow for post-release analysis Friday morning.
Write a comment