The jobs report and its internals
Lesson 10 · about 11 min
The Employment Situation report, universally called "the jobs report" or "NFP" after its headline number, is published by the BLS on the first Friday of the month at 8:30 a.m. Eastern (occasionally the second Friday when the first falls very early in the month, and shifted by holidays). It is the most-watched growth release in the world because it is timely, it covers the whole economy, and the Fed's employment mandate makes it a direct input to the reaction function.
Two surveys, one report
The report bundles two separate surveys that regularly disagree.
| Survey | Also called | What it asks | Headline output |
|---|---|---|---|
| Establishment survey | Payroll survey | Businesses: how many people are on your payroll? | Nonfarm payrolls (NFP), average hourly earnings, average weekly hours |
| Household survey | CPS | Households: are you employed, unemployed, or not looking? | Unemployment rate, participation rate, employment-population ratio |
The establishment survey has a much larger sample and is the source of the headline NFP number. The household survey is noisier month to month but is the only source of the unemployment rate. When the two diverge for several months (payrolls strong, household employment flat), economists argue about which is right, and the argument is usually settled by revisions a year later.
The numbers, in the order the desk reads them
- Nonfarm payrolls change (headline). Consensus is typically forecast with a standard error in the tens of thousands, and the BLS's own stated confidence interval on the monthly change is roughly plus or minus 130,000. Yes, that is larger than most surprises. A 30k miss is noise; a 150k miss is a signal.
- Revisions to the prior two months. These are published in the same release and often matter as much as the headline. A 200k print with a net minus 100k revision is a 100k print in reality.
- Unemployment rate. Reported to one decimal, forecast to one decimal. A 0.1 move is a real surprise; a 0.2 move is a big one.
- Average hourly earnings, m/m and y/y. The inflation component of the jobs report. In an inflation-fighting regime this can dominate the reaction.
- Participation rate. Rising participation can push unemployment up for benign reasons (more people looking), which softens a bad-looking unemployment print.
- Average weekly hours. A small, quiet number that leads layoffs; firms cut hours before they cut heads.
- Sector detail. Health care and government have accounted for a large share of gains in some periods; a print carried by those two sectors is read as weaker than one led by construction and manufacturing.
Reading a mixed print
Most jobs reports are mixed, and the reaction depends on which component the market is most sensitive to that month.
| Headline NFP | Unemployment | Earnings | Likely read in an inflation-fighting regime | Likely read in a growth-scare regime |
|---|---|---|---|---|
| Beat | Down | Hot | Hawkish: yields up, ES down | Relief: yields up, ES up |
| Beat | Up | Cool | Mixed to dovish; earnings and unemployment outweigh headline | Mixed |
| Miss | Up 0.2+ | Cool | Dovish for rates; growth-scare risk if the miss is large | Bearish: growth fear dominates |
| Miss | Flat | Hot | Confusing; often reverses within the hour | Slightly bearish |
There is no rule that resolves the mixed cases. What works is knowing the current reaction function (Module 1) and checking the 2-year yield's first move to see which component the market chose.
Key idea: NFP is two surveys and seven numbers, and the BLS's own error band on the headline is wider than most surprises. Read revisions, unemployment and earnings before you decide what the report said.
The Sahm rule and the level
The Sahm rule is a simple recession indicator built from the unemployment rate: when the three-month average unemployment rate rises 0.5 percentage points or more above its low of the prior twelve months, the economy has historically been in or entering a recession. It has a good historical record in the US but it is a description of past patterns, not a mechanism, and it gave a signal in 2024 that was followed by continued growth. Use it as one input to the level (Module 1), not as a trade trigger.
Data before the data
Several releases arrive in the days before NFP and shift the whisper number:
- ADP private payrolls (Wednesday before, 8:15 a.m. ET): a private estimate with a poor month-to-month correlation to NFP but a strong pull on expectations.
- Weekly initial and continuing jobless claims (every Thursday, 8:30 a.m. ET): the most timely labour-market data available.
- JOLTS job openings (usually a few days before NFP, 10:00 a.m. ET): openings, hires and quits, lagged by a month.
- ISM employment sub-indices.
When ADP and claims have both been weak, the market leans toward a weak NFP, and a print merely on consensus can rally.
Annual benchmark revisions
Once a year the BLS benchmarks the payroll survey to a more complete count from unemployment insurance records. The preliminary benchmark is announced in late summer and the final revision is applied with the January report in February. Preliminary benchmarks have occasionally been large (hundreds of thousands of jobs over a twelve-month period) and can change the narrative of the prior year without changing any single monthly print.
Try it: For the most recent jobs report, build a seven-row table: headline, revisions, unemployment, earnings m/m, participation, hours, and the sector that contributed most. Beside each write consensus, actual, and a one-word read (hawkish, dovish, neutral). Compare your net read to what the 2-year yield did in the first fifteen minutes.
Recap
- The jobs report is published by BLS on the first Friday at 8:30 ET and combines the establishment (payrolls) and household (unemployment) surveys.
- Read in order: headline, revisions, unemployment rate, hourly earnings, participation, hours, sector mix.
- The BLS error band on the headline is roughly plus or minus 130k; small misses are noise.
- Mixed prints are resolved by the current reaction function and by watching the 2-year's first move.
- ADP, claims and JOLTS set the whisper before Friday; annual benchmarks can rewrite the year.