On the first Friday of most months, at 8:30 a.m. New York time, the Bureau of Labor Statistics publishes the Employment Situation report. Within a second, currency pairs move, index futures gap, and gold repositions. The headline that causes it is non-farm payrolls: the net change in US employment over the previous month.
Traders who only read that headline are reading about a third of the report, and often the least informative third. NFP is three data sets released simultaneously, and the market can decide to trade any one of them.
What non-farm actually excludes
The "non-farm" label exists because agricultural hiring is violently seasonal and would swamp the signal. But the exclusion list is commonly misstated, so it is worth being precise.
Payrolls exclude farm workers, the self-employed and unincorporated proprietors, private household employees, unpaid family workers and active-duty military. They include government employees at all levels and employees of nonprofit organizations, both of which are frequently and wrongly listed as excluded.
That matters in practice. A large government hiring or firing program shows up directly in the headline, and a monthly print can be flattered or dragged by public sector payrolls with no private sector signal at all. Reading the private payrolls line separately is worth the extra ten seconds.
Two surveys, one report
The report is built from two independent surveys, and this is the single most useful structural fact about it.
The establishment survey asks roughly 120,000 businesses and government agencies how many people were on their payrolls. It produces the headline payrolls number and average hourly earnings. Because it counts jobs, a person holding two jobs is counted twice.
The household survey asks about 60,000 households about their employment status. It produces the unemployment rate and the participation rate. Because it counts people, that same person is counted once, and it captures the self-employed, whom the establishment survey misses entirely.
Two different samples measuring two different things will disagree, sometimes sharply. A month showing solid payroll growth alongside a rising unemployment rate is not a contradiction or an error; it is two instruments reading different parts of the same economy. The household survey is smaller and therefore noisier month to month, but it picks up turning points earlier because it sees self-employment and marginal attachment.
The three numbers that matter
Payrolls. The net change in jobs. As a rough frame, prints above 150,000 to 200,000 describe an economy expanding at trend, below 100,000 a slowdown, and negative prints a contraction. Those thresholds drift with population growth and immigration, so they are a guide, not a rule.
Unemployment rate. The share of the labor force without a job and actively looking. Recent estimates of full employment in the United States sit somewhere in the mid-3s to low 4s, but the level consistent with stable prices is unobservable and gets revised. The trap here is direction: the rate can fall for a bad reason, when discouraged workers stop searching and leave the labor force entirely. Always read it alongside the participation rate.
Average hourly earnings. Wage growth, monthly and annual. This is the number that has done the most damage to positioning over the last few years, because it is the bridge between the labor market and inflation. Firms facing rising labor costs raise prices, which shows up in services inflation. Wage growth around 3% to 3.5% is broadly consistent with 2% inflation given normal productivity; sustained readings above 4% are not.
Combinations tell the story
Individually these numbers are ambiguous. Together they describe a state of the world.
Solid payrolls, stable unemployment, moderate wages. Growth without price pressure. The most comfortable outcome for a central bank with a dual mandate, and generally the friendliest for risk assets.
Strong payrolls and accelerating wages. An overheating labor market. Implies policy stays tight for longer, which typically supports the dollar and pressures equities.
Weak payrolls but wages still rising. The worst combination: slowing activity with persistent inflation. Cutting rates feeds prices, holding them deepens the slowdown, and markets price the uncertainty by demanding a discount from everything.
Weak payrolls, rising unemployment, decelerating wages. A clear slowdown, which markets sometimes greet warmly because it brings rate cuts forward. Whether that holds depends entirely on whether the slowdown stops at "slowdown."
Good news is bad news, and when it flips
Newcomers are regularly confused by strong employment data sending equities down. The mechanism is straightforward once you separate the economy from the discount rate.
When inflation is the binding constraint, a hot labor market means the central bank cannot ease. Higher-for-longer rates lower the present value of future earnings, and equities fall even though the economy is doing well. When growth is the binding constraint, the same strong number is read as earnings support and equities rise.
The regime determines the sign. Before every release it is worth asking one question: what is the market currently afraid of? The answer tells you how a beat will be interpreted, which matters more than whether there is a beat.
Revisions: the part nobody reads
Every NFP report revises the previous two months. Those revisions routinely run into the tens of thousands of jobs, and they are published in the same release that everyone is reading for the current month.
The effect can invert the message. A headline of 180,000 looks solid until you notice the prior two months were cut by 50,000 each — at which point the three-month average is materially weaker than it appeared last month, and the labor market has been cooling for a quarter without anyone noticing.
Persistent one-directional revisions are a signal in themselves. A run of downward revisions means the initial samples are systematically overstating hiring, which historically happens near turning points, when business births and deaths are hardest to model. The annual benchmark revision, which reconciles the survey against actual tax records, occasionally moves the level by hundreds of thousands.
The practical habit: read the three-month average, not the print, and look at what the revisions did to it.
The rest of the labor calendar
NFP is the loudest labor release, not the only one, and the others often frame it.
ADP National Employment Report, two days earlier, estimates private payrolls from actual payroll processing data. Its correlation with the official number is weaker than its reputation suggests, and it should be treated as a different measurement rather than a preview.
Initial jobless claims, every Thursday, are the highest-frequency labor data available and the closest thing to real-time. Layoffs show up here first. A sustained climb in the four-week average has historically led weakening payrolls.
JOLTS measures job openings, hires and quits. The quits rate is the underrated series: people resign when they are confident of finding something better, so it tracks labor market tightness and, with a lag, wage pressure.
All of them sit on our economic calendar with consensus and previous readings, which is what you need to judge a surprise rather than a level.
Trading the release
In the minutes after 8:30, EUR/USD can move fifty to a hundred pips, gold twenty to forty dollars, index futures hundreds of points. Spreads that are normally one pip can widen to five or ten. Slippage is severe, and a stop is filled at whatever price exists, not the one you chose.
For most retail accounts the honest answer is not to hold exposure through it. Closing positions half an hour before and waiting fifteen to thirty minutes afterwards costs nothing except the trade you would probably have lost.
Traders who do work the release generally wait out the first two to five minutes. The initial move is machine-driven off the headline; the second move comes when humans have read the wage line and the revisions, and it frequently goes the other way. That reversal is where the informed flow is. Chasing the first candle means providing liquidity to it.
The deeper point is the same one that applies to the CPI release: markets do not trade the number, they trade the distance between the number and the forecast. And on NFP there are three forecasts to miss, which is why the direction of the reaction is so hard to call even when you correctly guess the headline.
What the report is really for
Payrolls matter because employment is half of the Federal Reserve's mandate, and the Fed sets the price of dollar money for the world. Every asset that discounts a future cash flow in dollars is affected by what this report does to rate expectations.
Reading it properly means reading four things together — jobs, unemployment, wages, revisions — and placing them against what the market already assumed. One month is noise. A quarter of consistent data is a trend, and trends are what move policy, which is what moves everything else.
