What non-farm payrolls count, and why revisions follow

What does the non-farm payroll number count?

The number counts employees on payrolls and reports how many were added or lost since the previous month. It leaves out farm workers and private household employees. Wikipedia's summary puts the covered share at about 80 percent of the workforce. It says the figure comes from a monthly survey of about 131,000 businesses and agencies, covering about 670,000 worksites (source: Wikipedia, read 2026-10-03).

Two details of that definition matter for everything below. The headline is a change, so a reading of 29,000 means 29,000 more jobs than the month before and says nothing about how many jobs exist. And it is an estimate from a survey, so it carries the error any survey carries.

Per the same summary, the US Bureau of Labor Statistics publishes it in the monthly jobs report. That is typically the first Friday of the month, at 8:30 a.m. Eastern Time. The same report carries other figures, and traders read several of them at once: the unemployment rate, jobs by sector, average hourly earnings, and revisions to earlier months.

What did the September report show, with the arithmetic?

The September report is a clean case for seeing how the headline and the revisions fit together. As of 2 October 2026, CoinDesk reported these figures from the release:

• September payrolls rose by 29,000, against a consensus forecast of 90,000.
• August's gain was revised down from the originally reported 162,000 to 133,000.
• July's gain of 21,000 was revised to a loss of 10,000.
• The unemployment rate rose to 4.2 percent, against an expected 4.1 percent and August's 4.1 percent.
• Average hourly earnings rose 0.1 percent for the month, against a forecast of 0.3 percent, and 3 percent over the year, against a forecast of 3.2 percent.

Source: CoinDesk, "U.S. added just 29,000 jobs in September, with unemployment rate rising to 4.2%", 2 October 2026, read 2026-10-03.

Now the arithmetic. August lost 162,000 − 133,000 = 29,000 jobs in revision. July moved from +21,000 to −10,000, a swing of 31,000. The two revisions together removed 29,000 + 31,000 = 60,000 jobs from the record, a little more than twice the 29,000 that September added.

Add the three months as first reported: 21,000 + 162,000 + 29,000 = 212,000 jobs. Add them as the September report now states them: −10,000 + 133,000 + 29,000 = 152,000 jobs. The same three months look 60,000 jobs smaller than they did when each was first printed.

The new month takes the headline, and the older months carry the correction.

Why does the first number get restated?

A first estimate comes from a survey, and a survey figure is restated as fuller counts arrive. That is why the release that gives a new month also gives new figures for the two months before it, as the September report did for August and July.

For a reader, the practical rule is to hold the first number loosely and the direction of several months more firmly. A single print of 29,000 and a single print of 133,000 are both estimates that can move. A run of months that keeps coming in below the forecast says more than any one of them, and the revision line is where a run shows up.

Coverage often quotes the headline as if it counted the jobs that exist. The figure is a change, taken from a survey and restated later, and a quote of it reads best with those two facts attached.

Why does a market move on the gap to the forecast?

A market moves on the gap between the number and the forecast because the forecast is already in the price. By the time the report is released, traders have priced in something close to the consensus of 90,000. A reading of 90,000 would add little new information. A reading of 29,000 is 61,000 below it, and that gap is the news.

CoinDesk's report shows the reaction in that form. The 10-year Treasury yield slid 7 basis points to 5.17 percent. The 2-year yield fell by a similar amount to 4.71 percent, and gold rose more than 1 percent (source: CoinDesk, 2 October 2026). A basis point is one hundredth of a percentage point, so 7 basis points is 0.07 percentage points. Wikipedia's summary adds that revisions to earlier months can also move markets, because traders re-price growth expectations when an earlier number changes (source: Wikipedia).

What a given month's number does next, to rates or to any price, is unknown when the report lands.

What does it mean for a holder in rupees?

A US jobs report lands in the US morning, and prices can move within minutes of it. That is a reason to know what moves the screen on the first Friday, while the weights stay where the thesis put them.

Take a holder who puts ₹25,000 a month into a basket. Over twelve months that is 12 × ₹25,000 = ₹3,00,000 invested, across twelve instalment dates. The year also brings twelve jobs reports, each with a headline that can be restated later. None of the twelve reports changes an instalment date or an amount. A holder who knows what the headline counts can read the day's coverage without treating it as an instruction.

The same logic holds at the other end of the scale. A position sized at three percent of a ₹1 crore portfolio is ₹3,00,000. A headline figure from a survey, restated a month later, is a thin basis for changing a position of that size. The sizing decision was made against a thesis and a horizon, and one report does not change either.

What does a rule-based index do on a report day?

A rule-based index does what its rules say, on its own schedule. A Crypto Index is a curated basket of digital assets, weighted on a methodology and rebalanced on a schedule. An investor in one holds the basket and not a set of coins to store. Qatobit offers four QSI Crypto Indices (QSI Core, QSI Growth, QSI VRION and QSI GEQ8), three of which hold crypto, each rebalanced monthly. Rules decide the basket and a schedule carries the rebalance, so the mood of a first Friday has nothing to act on.

Where to go next

The jobs report matters to markets mostly through rates, and two earlier pieces cover that link. What the 10-year bond yield is and what a rise reprices explains the yield that moved 7 basis points on the September report. Fed meeting: hike, hold or cut, and the monthly SIP arithmetic works through what a rate decision does to a monthly instalment. For the schedule side, four crypto indexes, one methodology sets out the rules each QSI index follows.

Frequently asked questions

What does non-farm payroll measure?

Non-farm payroll measures the change in the number of jobs on US business and government payrolls, outside farms, from one month to the next. The September report put the change at 29,000 jobs, against a consensus forecast of 90,000.

Why is non-farm payroll revised?

It is a survey estimate, and a survey figure is restated as fuller counts arrive. The September report revised August from 162,000 to 133,000 and July from a gain of 21,000 to a loss of 10,000, removing 60,000 jobs between them.

What is the difference between the headline and the unemployment rate?

The headline is the change in jobs from the month before. The unemployment rate is a separate figure in the same report, and in September it rose to 4.2 percent against an expected 4.1 percent.

Why do markets react to the gap against the forecast?

The forecast is already in the price before the report lands. In September the headline came in 61,000 jobs below the 90,000 consensus, and the 10-year Treasury yield slid 7 basis points to 5.17 percent in response.

Should a jobs report change a monthly investment plan?

A report is one estimate, restated later, and a monthly plan is built on a horizon that spans many reports. A holder putting ₹25,000 a month into a basket invests ₹3,00,000 a year across twelve instalments, and no single jobs report is part of that rule.

Crypto investments are subject to market risk. Not financial advice.