October jobs, before the Friday print.
Nonfarm payrolls, the unemployment rate and hourly earnings. The payroll call learns directly from first prints (the number markets trade on) and leans on weekly jobless claims from the survey month, which arrive before the report.
What September 2026 told us
The last release against what was expected, what drove it, and what it changes.
The print
| First print | Now | Consensus | Model | Key Data call | Surprise | |
|---|---|---|---|---|---|---|
| Payrolls | +29k | +29k | +89k | +120k | +101k | −60k |
| Unemployment rate | 4.2% | 4.2% | 4.1% | 4.11% | 4.10% | +0.1 pt |
| Hourly earnings | +0.1% | +0.13% | +0.3% | +0.26% | +0.30% | −0.2 pt |
What it showed
- September 2026 payrolls rose 29k as revised (first print 29k) against a consensus of 89k; the three-month average is 51k and the twelve-month average 41k.
- Unemployment 4.2% (−0.2 pt on the year); U-6 7.6%; participation 61.8%.
- Hourly earnings +0.13% on the month, 3.0% on the year.
- Hiring is concentrated: health care & social assistance +23k, government −17k. Job openings were 7.08 million in August 2026.
Where the jobs came from
Change in payrolls by sector, thousands.
Analyst notes
Hiring has slowed to a crawl. Payrolls rose just 29,000 in September against an 89,000 consensus, leaving the three-month average near 51,000 and the twelve-month average near 41,000. That is well below the pace needed to absorb labor-force growth in most estimates.
Narrow gains. Health care and social assistance added 23,000; government shed 17,000. Outside health care, hiring is roughly flat.
The unemployment rate is not confirming the weakness yet. It rose a tenth to 4.2% but is lower than a year ago, helped by a participation rate of 61.8%. Wages rose only 0.1% on the month and 3.0% on the year, consistent with a cooling market.
October. Initial jobless claims in the October survey weeks have not jumped, so the model expects a modest rebound to roughly 85,000–90,000, with unemployment near 4.2%.
Where October 2026 should land
Odds of each outcome, sized by the model's own recent misses with fat tails for surprise months.
Payrolls model +93k
Unemployment rate model 4.15%
Hourly earnings model +0.28%
Track record
Every month the model is refit using only data available at the time, then scored against the number first published, the same yardstick used for the consensus.
| Period | Releases | Model miss | Key Data call miss | Consensus miss | Same 25k bucket: model / call / consensus | Model closer / further | |
|---|---|---|---|---|---|---|---|
| Since 2023 | 44 | 70k | 67k | 70k | 14% / 11% / 7% | 24 / 20 | Call ahead |
| Since 2019 | 76 | 88k | 87k | 90k | 17% / 11% / 5% | 42 / 34 | Call ahead |
| Since 2014 | 123 | 75k | 73k | 74k | 16% / 11% / 8% | 61 / 62 | Call ahead |
| Period | Releases | Model miss | Key Data call miss | Consensus miss | Exact rounded print: model / call / consensus | Model closer / further | |
|---|---|---|---|---|---|---|---|
| Since 2023 | 42 | 0.092 | 0.098 | 0.098 | 31% / 26% / 26% | 22 / 20 | Model ahead |
| Since 2019 | 74 | 0.102 | 0.103 | 0.103 | 24% / 26% / 27% | 37 / 37 | Model closer |
| Since 2014 | 121 | 0.100 | 0.094 | 0.093 | 26% / 31% / 31% | 50 / 71 | Even |
| Period | Releases | Model miss | Key Data call miss | Consensus miss | Exact rounded print: model / call / consensus | Model closer / further | |
|---|---|---|---|---|---|---|---|
| Since 2023 | 44 | 0.090 | 0.086 | 0.086 | 32% / 30% / 30% | 17 / 27 | Even |
| Since 2019 | 76 | 0.105 | 0.107 | 0.107 | 33% / 24% / 24% | 37 / 39 | Model ahead |
| Since 2014 | 123 | 0.104 | 0.102 | 0.102 | 31% / 27% / 27% | 59 / 64 | Even |
How it works
Payrolls: a regression on recent payroll momentum and the year-over-year change in initial and continuing claims (state programs, unadjusted, so seasonality cancels), trained on the first-printed payroll number rather than the revised series.
Unemployment rate: anchored on last month's unrounded rate (unemployed divided by the labor force), which shows how close the published one-decimal figure sat to a rounding boundary, plus a predicted change driven by claims; it learns from first prints. Hourly earnings: a blend of last month, the three-month and twelve-month pace.
Payrolls are where economists have the most private data (ADP, firm surveys, state reports), so the model alone is roughly level with the consensus. The Key Data call, which combines the two, has done better than either in the backtest; the track record below has the numbers.
The Key Data call: once Wall Street's consensus is published, the model's view is combined with it. The combination starts from the consensus and moves toward the model by an amount learned from past releases, also weighing last month's surprise and revision. It is used only while it has beaten the consensus over the previous three years of releases, scored out of sample; otherwise the call is the consensus.