thesisPrototype

Government data

US Nonfarm Payrolls Change, September 2026

What will the BLS first-print seasonally adjusted change in total nonfarm payroll employment be for September 2026?

Forecast

75k
-52k–202k
80% prediction interval
gpt-5.5Run record ↗
-139-9.7120249January 2026August 2026Oct 202675k
historyforecast path80% interval
Probability distribution

Cumulative probability

Chance that the outcome is at or below each value.

0%25%50%75%100%-243k75k393kForecast value

Shaded band: 80% interval (-52k–202k). Dashed line: point estimate (75k).

Derived from the point estimate and 80% interval; the agent did not report a full distribution.

Analysis

§September 2026 first-print nonfarm payrolls forecast

The target is the BLS Current Employment Statistics total nonfarm payroll employment over-the-month change, seasonally adjusted, first print for September 2026, in thousands. The resolver is the BLS Employment Situation release page and table B-1 / summary table B; the series variant is the seasonally adjusted total nonfarm establishment payroll change, not household employment, not not-seasonally-adjusted employment, and not a later revised CES level.

▸ Reported tool use: official.lookupmodel report
Checked BLS Employment Situation schedule and current release page for the September 2026 publication date and latest labor-market print.
↳ BLS schedule lists September 2026 Employment Situation release on Oct. 02, 2026 at 08:30 AM; the Aug. 2026 release reports total nonfarm payroll employment +162 thousand, unemployment rate 4.1 percent, prior-12-month payroll average +31 thousand, June revised from +20 to +31 thousand, and July revised from -23 to +21 thousand.
▸ Reported tool use: official.lookupmodel report
Read BLS archived Employment Situation releases for early-2026 first-print total nonfarm payroll changes.
↳ First-print seasonally adjusted total nonfarm payroll changes fetched from BLS archives: January 2026 +130 thousand, February 2026 -92 thousand, March 2026 +178 thousand, April 2026 +115 thousand; associated unemployment rates were 4.3 percent, 4.4 percent, 4.3 percent, and 4.3 percent.
▸ Reported tool use: official.lookupmodel report
Read BLS archived Employment Situation releases for recent first-print total nonfarm payroll changes.
↳ First-print seasonally adjusted total nonfarm payroll changes fetched from BLS archives: May 2026 +172 thousand, June 2026 +57 thousand, July 2026 -23 thousand, August 2026 +162 thousand; August details included food services +59 thousand, local government education +42 thousand, manufacturing +16 thousand, health care +13 thousand, and information -23 thousand.

Reference class/base rate: the same-variant 2026 first prints from January through August are 130, -92, 178, 115, 172, 57, -23, and 162 thousand. Their mean is 87.4 thousand, while the last three first prints average 65.3 thousand. I put more weight on the recent slower hiring regime than on the full-year mean.

Level, momentum, one-off, and policy-mechanism effects are mixed. Level: unemployment at 4.1 percent is not recessionary. Momentum: June and July were weak on first print, and their later upward revisions are current public evidence available before the September target resolves, not outcomes for the target. One-off: August's +162 thousand had a visible local-government-education rebound and food-services surge that I do not fully extrapolate. Policy-mechanism: no resolver-relevant policy level changes the first-print rule; federal and information-sector weakness keep the forecast below the full 2026 mean.

Prior/update/interval: persistence prior is the 2026 first-print reference class mean of 87.4 thousand from the 8 official BLS prints listed above; adjustment components are -15 thousand for recent three-month softness versus the full sample, -10 thousand for likely August one-off education/food-services payback, and +12 thousand for unemployment still near 4.1 percent and June/July upward revisions, giving about 74.4 thousand rounded to 75. For interval sizing, using the values themselves for this change/flow series, sample sigma = 99.1 thousand from only eight 2026 first prints; 1.28*sigma = 126.9 thousand, so the 80% interval is 75 +/- 127 = [-52, 202] thousand, with high sampling uncertainty.

Upside risk is a second month of education normalization plus broad private services hiring that would land above the interval if September prints above +202 thousand. Downside risk is renewed information/professional-services contraction or a weak seasonal-adjustment print that would land below the interval if payrolls fall by more than 52 thousand. Outside the interval would likely require a broad shock rather than ordinary month-to-month noise.

Review disposition: accepted the blocking resolver fix by setting the cell resolutionDate to the ledger-bound 2026-10-08 while retaining the official BLS scheduled first-print date of 2026-10-02 in the resolver evidence; accepted the interval caveat about eight-print sampling uncertainty and clarified that June/July revisions are pre-resolution evidence, not target outcomes.

Key drivers

  • 2026 first-print payroll base rate averages about 87 thousand
  • last three first prints average 65 thousand despite August rebound
  • August strength was partly concentrated in food services and local government education
  • health care trend remains positive but slower than its prior 12-month pace
  • July and June revisions are current pre-resolution evidence but not target outcomes

Tool evidence

This run has no captured tool responses. “Reported tool use” in the analysis is the model’s account, not an archived tool response.

Sources and resolution

Official source
Official agency release
Resolution date
October 8, 2026· outcome not recorded
Resolution rule
Resolve to the first-print over-the-month change, in thousands, for seasonally adjusted total nonfarm payroll employment in the BLS Employment Situation release for September 2026, table B-1 / summary table B. Use the value as first published by BLS, without later revisions, corrections, or release-day grace exceptions; the ledger-bound resolution date is the registered expected-release-window end, while the official BLS calendar cited in this run scheduled the September 2026 release for October 2, 2026.
Run details

The analysis is the model’s written report. Tool-use descriptions in that report are model claims; the activity artifacts contain the execution record.

thesis.analyst · gpt-5.5 · fast · v2.5.11

pre-submit review · completed

Draft is mostly publication-ready, but its resolutionDate conflicts with the canonical ledger target contract.

  • blocking resolver: The draft sets resolutionDate to 2026-10-02, while the canonical target context binds the target to the ledger registration and expectedReleaseWindow ending 2026-10-08.
  • info optional_suggestion: In the interval sentence, note that sigma is estimated from only eight 2026 first prints, so the 80% interval reflects high sampling uncertainty.
  • info optional_suggestion: Clarify that June and July revisions are current evidence available before the September target resolves, not target outcomes.

disposition accepted: Review disposition: accepted the blocking resolver fix by setting the cell resolutionDate to the ledger-bound 2026-10-08 while retaining the official BLS scheduled first-print date of 2026-10-02 in the resolver evidence; accepted the interval caveat about eight-print sampling uncertainty and clarified that June/July revisions are pre-resolution evidence, not target outcomes.

disposition not applicable: Review disposition: accepted the blocking resolver fix by setting the cell resolutionDate to the ledger-bound 2026-10-08 while retaining the official BLS scheduled first-print date of 2026-10-02 in the resolver evidence; accepted the interval caveat about eight-print sampling uncertainty and clarified that June/July revisions are pre-resolution evidence, not target outcomes.

disposition not applicable: Review disposition: accepted the blocking resolver fix by setting the cell resolutionDate to the ledger-bound 2026-10-08 while retaining the official BLS scheduled first-print date of 2026-10-02 in the resolver evidence; accepted the interval caveat about eight-print sampling uncertainty and clarified that June/July revisions are pre-resolution evidence, not target outcomes.

Activity artifacts

Complete original trace

§September 2026 first-print nonfarm payrolls forecast

The target is the BLS Current Employment Statistics total nonfarm payroll employment over-the-month change, seasonally adjusted, first print for September 2026, in thousands. The resolver is the BLS Employment Situation release page and table B-1 / summary table B; the series variant is the seasonally adjusted total nonfarm establishment payroll change, not household employment, not not-seasonally-adjusted employment, and not a later revised CES level.

▸ Reported tool use: official.lookupmodel report
Checked BLS Employment Situation schedule and current release page for the September 2026 publication date and latest labor-market print.
↳ BLS schedule lists September 2026 Employment Situation release on Oct. 02, 2026 at 08:30 AM; the Aug. 2026 release reports total nonfarm payroll employment +162 thousand, unemployment rate 4.1 percent, prior-12-month payroll average +31 thousand, June revised from +20 to +31 thousand, and July revised from -23 to +21 thousand.
▸ Reported tool use: official.lookupmodel report
Read BLS archived Employment Situation releases for early-2026 first-print total nonfarm payroll changes.
↳ First-print seasonally adjusted total nonfarm payroll changes fetched from BLS archives: January 2026 +130 thousand, February 2026 -92 thousand, March 2026 +178 thousand, April 2026 +115 thousand; associated unemployment rates were 4.3 percent, 4.4 percent, 4.3 percent, and 4.3 percent.
▸ Reported tool use: official.lookupmodel report
Read BLS archived Employment Situation releases for recent first-print total nonfarm payroll changes.
↳ First-print seasonally adjusted total nonfarm payroll changes fetched from BLS archives: May 2026 +172 thousand, June 2026 +57 thousand, July 2026 -23 thousand, August 2026 +162 thousand; August details included food services +59 thousand, local government education +42 thousand, manufacturing +16 thousand, health care +13 thousand, and information -23 thousand.

Reference class/base rate: the same-variant 2026 first prints from January through August are 130, -92, 178, 115, 172, 57, -23, and 162 thousand. Their mean is 87.4 thousand, while the last three first prints average 65.3 thousand. I put more weight on the recent slower hiring regime than on the full-year mean.

Level, momentum, one-off, and policy-mechanism effects are mixed. Level: unemployment at 4.1 percent is not recessionary. Momentum: June and July were weak on first print, and their later upward revisions are current public evidence available before the September target resolves, not outcomes for the target. One-off: August's +162 thousand had a visible local-government-education rebound and food-services surge that I do not fully extrapolate. Policy-mechanism: no resolver-relevant policy level changes the first-print rule; federal and information-sector weakness keep the forecast below the full 2026 mean.

Prior/update/interval: persistence prior is the 2026 first-print reference class mean of 87.4 thousand from the 8 official BLS prints listed above; adjustment components are -15 thousand for recent three-month softness versus the full sample, -10 thousand for likely August one-off education/food-services payback, and +12 thousand for unemployment still near 4.1 percent and June/July upward revisions, giving about 74.4 thousand rounded to 75. For interval sizing, using the values themselves for this change/flow series, sample sigma = 99.1 thousand from only eight 2026 first prints; 1.28*sigma = 126.9 thousand, so the 80% interval is 75 +/- 127 = [-52, 202] thousand, with high sampling uncertainty.

Upside risk is a second month of education normalization plus broad private services hiring that would land above the interval if September prints above +202 thousand. Downside risk is renewed information/professional-services contraction or a weak seasonal-adjustment print that would land below the interval if payrolls fall by more than 52 thousand. Outside the interval would likely require a broad shock rather than ordinary month-to-month noise.

Review disposition: accepted the blocking resolver fix by setting the cell resolutionDate to the ledger-bound 2026-10-08 while retaining the official BLS scheduled first-print date of 2026-10-02 in the resolver evidence; accepted the interval caveat about eight-print sampling uncertainty and clarified that June/July revisions are pre-resolution evidence, not target outcomes.

calibrated forecast · 80% CI
75k[-52k · 202k]
Target metadata

Data point: bls.ces.total_nonfarm.payroll_employment.change.sa.2026-09.first_print

More government data forecasts