thesisPrototype

Government data

US computer and math employment, September 2026

What will BLS first report for total employed people age 16 and over in Computer and mathematical occupations in September 2026 in CPS Employment Situation Table A-19, not seasonally adjusted, converted from thousands to millions?

Forecast

7M
6.9M–7.2M
80% prediction interval
gpt-5.5Run record ↗
6.56.877.3March 2026August 2026Oct 20267M
historyforecast path80% interval
Probability distribution

Cumulative probability

Chance that the outcome is at or below each value.

0%25%50%75%100%6.6M7M7.5MForecast value

Shaded band: 80% interval (6.9M–7.2M). Dashed line: point estimate (7M).

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

Analysis

§September 2026 CPS computer and mathematical employment forecast

The resolver is the first BLS CPS Employment Situation Table A-19 print for employed people age 16 and over in 'Computer and mathematical occupations,' not seasonally adjusted. Values are in thousands on the BLS table and are transformed to millions by multiplying by 0.001. The live BLS navigation also exposes related occupation rows in the Employment Situation release, but the registered target binds cpseea19.htm Table A-19 and this forecast preserves that exact variant.

▸ Reported tool use: official.lookupmodel report
Open https://www.bls.gov/web/empsit/cpseea19.htm and read the Computer and mathematical occupations row.
↳ BLS CPS Table A-19, Last Modified Date September 04, 2026, reports Computer and mathematical occupations Total 16 years and over: August 2025 = 6,903 thousand and August 2026 = 7,010 thousand, equal to 6.903 and 7.010 million.
▸ Reported tool use: official.lookupmodel report
Fetch Wayback snapshots of BLS cpseea19.htm for spring 2026 first-print history.
↳ Wayback snapshots of the BLS Table A-19 source report Computer and mathematical occupations Total 16 years and over: March 2026 = 6,574 thousand from the April 03, 2026 table; April 2026 = 6,834 thousand from the May 08, 2026 table; May 2026 = 6,903 thousand from the June 05, 2026 table; June 2026 = 6,950 thousand from the July 02, 2026 table.
▸ Reported tool use: public_record.lookupmodel report
Inspect published Thesis A-19 public record generated from the August 07, 2026 BLS Table A-19 page; the record slug is august_2026.first_print because it resolves the August release's July reference-month print.
↳ The public record for bls.cps.employed_people_by_occupation.computer_mathematical.august_2026.first_print records the fetched July 2026 BLS Table A-19 value as 6,924 thousand, equal to 6.924 million, and the same row's July 2025 value as 6,656 thousand.
▸ Reported tool use: official.lookupmodel report
Open official BLS CPS annual Table 9 pages for longer-run occupation context.
↳ BLS CPS annual Table 9 reports Computer and mathematical occupations Total 16 years and over: 2021 = 5,688 thousand, 2022 = 6,171 thousand, 2023 = 6,502 thousand, 2024 = 6,386 thousand, and 2025 = 6,711 thousand; the 2025 annual estimate excludes October because October 2025 data were not collected.
▸ Reported tool use: official.calendarmodel report
Open https://www.bls.gov/schedule/news_release/empsit.htm to verify the Employment Situation release schedule.
↳ BLS schedule rows fetched this run list August 2026 with release date Sep. 04, 2026 at 08:30 AM, September 2026 with release date Oct. 02, 2026 at 08:30 AM, and October 2026 with release date Nov. 06, 2026 at 08:30 AM. This conflicts with the registered resolutionDate 2026-10-08, which I retain because the target contract supplies it.

The outside-view base rate/reference class is latest same-table persistence: August 2026 first print was 7.010 million, after March-August first-print values of 6.574, 6.834, 6.903, 6.950, 6.924, and 7.010 million. Longer-run annual BLS context supports a mild upward drift, rising from 5.688 million in 2021 to 6.711 million in 2025, but the monthly CPS occupation row is noisy.

Level, momentum, one-off, and policy-mechanism effects: the level anchor is 7.010 million; recent monthly momentum is positive but uneven, so I add only +0.025 million rather than the full +0.087 million mean change; I add 0.000 million for one-off effects and 0.000 million for policy or classification mechanisms because I found no release-specific public evidence for a September break.

▸ Reported tool use: model.candidatesmodel report
Compute simple persistence and drift candidates from fetched first-print monthly history.
↳ Using March-August 2026 values 6.574, 6.834, 6.903, 6.950, 6.924, and 7.010 million gives successive changes +0.260, +0.069, +0.047, -0.026, and +0.086 million, mean +0.087 million and sigma = 0.106 million. Persistence candidate point = 7.010 million with mechanical 80% interval 6.874-7.146; selected conservative drift candidate point = 7.035 million with widened 80% interval 6.855-7.215.

Prior/update/interval: model prior = latest first-print persistence at 7.010 million; historical sample = March-August 2026 first-print levels 6.574, 6.834, 6.903, 6.950, 6.924, and 7.010 million; adjustment components = +0.025 million for mild recent/annual momentum, +0.000 million one-off, +0.000 million policy mechanism, so point = 7.010 + 0.025 = 7.035 million. Successive changes are +0.260, +0.069, +0.047, -0.026, and +0.086 million, giving sigma = 0.106 million. The mechanical 80% half-width is roughly 1.28*sigma = 1.28*0.106 = 0.136 million; I widen to 0.180 million, about 1.32x the mechanical half-width, because the exact first-print monthly history is short and CPS occupation subgroup allocation is noisy. Final bounds are 7.035 - 0.180 = 6.855 and 7.035 + 0.180 = 7.215 million.

Upside risk: a favorable September CPS sample, continued labor-force entry into tech roles, or a rebound from professional services would land above the interval if the Table A-19 row exceeds 7.215 million. Downside risk: adverse CPS sampling, weaker technology hiring, or reversal of the August gain would land below the interval if the row is under 6.855 million. Either would be outside the interval.

Review disposition: accepted both optional reviewer suggestions by tightening the resolutionRule discrepancy note and clarifying that the august_2026.first_print public-record label refers to the August release's July reference-month first print; no required fixes were raised.

Key drivers

  • August 2026 Table A-19 level of 7.010 million
  • recent positive first-print monthly drift
  • CPS occupation-row sampling volatility
  • no discrete policy or classification mechanism identified

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
U.S. Bureau of Labor Statistics Employment Situation, CPS Table A-19
Resolution date
October 8, 2026· outcome not recorded
Resolution rule
Resolve to the first BLS-published September 2026 value in CPS Employment Situation Table A-19, row 'Computer and mathematical occupations,' Total age 16 years and over, not seasonally adjusted, reported in thousands and multiplied by 0.001 to millions. Retain three decimals and ignore all later revisions or corrections unless the registered target is formally amended before release. The official BLS Employment Situation schedule fetched this run lists September 2026 for October 2, 2026 at 08:30 AM ET; the canonical ledger target fixes resolutionDate as 2026-10-08, so this cell remains tied to that registered date.
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 publishable: it preserves the registered target contract, states a persistence prior, uses recent first-print history for a modest update, and gives coherent point, interval, tails, and JSON fields.

  • info optional_suggestion: Consider shortening the resolutionRule discrepancy note so the operative resolver remains visually dominant.
  • info optional_suggestion: Clarify whether the July 2026 public-record lookup label is intentionally named august_2026.first_print to avoid reviewer confusion.

disposition not applicable: Review disposition: accepted both optional reviewer suggestions by tightening the resolutionRule discrepancy note and clarifying that the august_2026.first_print public-record label refers to the August release's July reference-month first print; no required fixes were raised.

disposition not applicable: Review disposition: accepted both optional reviewer suggestions by tightening the resolutionRule discrepancy note and clarifying that the august_2026.first_print public-record label refers to the August release's July reference-month first print; no required fixes were raised.

Activity artifacts

Complete original trace

§September 2026 CPS computer and mathematical employment forecast

The resolver is the first BLS CPS Employment Situation Table A-19 print for employed people age 16 and over in 'Computer and mathematical occupations,' not seasonally adjusted. Values are in thousands on the BLS table and are transformed to millions by multiplying by 0.001. The live BLS navigation also exposes related occupation rows in the Employment Situation release, but the registered target binds cpseea19.htm Table A-19 and this forecast preserves that exact variant.

▸ Reported tool use: official.lookupmodel report
Open https://www.bls.gov/web/empsit/cpseea19.htm and read the Computer and mathematical occupations row.
↳ BLS CPS Table A-19, Last Modified Date September 04, 2026, reports Computer and mathematical occupations Total 16 years and over: August 2025 = 6,903 thousand and August 2026 = 7,010 thousand, equal to 6.903 and 7.010 million.
▸ Reported tool use: official.lookupmodel report
Fetch Wayback snapshots of BLS cpseea19.htm for spring 2026 first-print history.
↳ Wayback snapshots of the BLS Table A-19 source report Computer and mathematical occupations Total 16 years and over: March 2026 = 6,574 thousand from the April 03, 2026 table; April 2026 = 6,834 thousand from the May 08, 2026 table; May 2026 = 6,903 thousand from the June 05, 2026 table; June 2026 = 6,950 thousand from the July 02, 2026 table.
▸ Reported tool use: public_record.lookupmodel report
Inspect published Thesis A-19 public record generated from the August 07, 2026 BLS Table A-19 page; the record slug is august_2026.first_print because it resolves the August release's July reference-month print.
↳ The public record for bls.cps.employed_people_by_occupation.computer_mathematical.august_2026.first_print records the fetched July 2026 BLS Table A-19 value as 6,924 thousand, equal to 6.924 million, and the same row's July 2025 value as 6,656 thousand.
▸ Reported tool use: official.lookupmodel report
Open official BLS CPS annual Table 9 pages for longer-run occupation context.
↳ BLS CPS annual Table 9 reports Computer and mathematical occupations Total 16 years and over: 2021 = 5,688 thousand, 2022 = 6,171 thousand, 2023 = 6,502 thousand, 2024 = 6,386 thousand, and 2025 = 6,711 thousand; the 2025 annual estimate excludes October because October 2025 data were not collected.
▸ Reported tool use: official.calendarmodel report
Open https://www.bls.gov/schedule/news_release/empsit.htm to verify the Employment Situation release schedule.
↳ BLS schedule rows fetched this run list August 2026 with release date Sep. 04, 2026 at 08:30 AM, September 2026 with release date Oct. 02, 2026 at 08:30 AM, and October 2026 with release date Nov. 06, 2026 at 08:30 AM. This conflicts with the registered resolutionDate 2026-10-08, which I retain because the target contract supplies it.

The outside-view base rate/reference class is latest same-table persistence: August 2026 first print was 7.010 million, after March-August first-print values of 6.574, 6.834, 6.903, 6.950, 6.924, and 7.010 million. Longer-run annual BLS context supports a mild upward drift, rising from 5.688 million in 2021 to 6.711 million in 2025, but the monthly CPS occupation row is noisy.

Level, momentum, one-off, and policy-mechanism effects: the level anchor is 7.010 million; recent monthly momentum is positive but uneven, so I add only +0.025 million rather than the full +0.087 million mean change; I add 0.000 million for one-off effects and 0.000 million for policy or classification mechanisms because I found no release-specific public evidence for a September break.

▸ Reported tool use: model.candidatesmodel report
Compute simple persistence and drift candidates from fetched first-print monthly history.
↳ Using March-August 2026 values 6.574, 6.834, 6.903, 6.950, 6.924, and 7.010 million gives successive changes +0.260, +0.069, +0.047, -0.026, and +0.086 million, mean +0.087 million and sigma = 0.106 million. Persistence candidate point = 7.010 million with mechanical 80% interval 6.874-7.146; selected conservative drift candidate point = 7.035 million with widened 80% interval 6.855-7.215.

Prior/update/interval: model prior = latest first-print persistence at 7.010 million; historical sample = March-August 2026 first-print levels 6.574, 6.834, 6.903, 6.950, 6.924, and 7.010 million; adjustment components = +0.025 million for mild recent/annual momentum, +0.000 million one-off, +0.000 million policy mechanism, so point = 7.010 + 0.025 = 7.035 million. Successive changes are +0.260, +0.069, +0.047, -0.026, and +0.086 million, giving sigma = 0.106 million. The mechanical 80% half-width is roughly 1.28*sigma = 1.28*0.106 = 0.136 million; I widen to 0.180 million, about 1.32x the mechanical half-width, because the exact first-print monthly history is short and CPS occupation subgroup allocation is noisy. Final bounds are 7.035 - 0.180 = 6.855 and 7.035 + 0.180 = 7.215 million.

Upside risk: a favorable September CPS sample, continued labor-force entry into tech roles, or a rebound from professional services would land above the interval if the Table A-19 row exceeds 7.215 million. Downside risk: adverse CPS sampling, weaker technology hiring, or reversal of the August gain would land below the interval if the row is under 6.855 million. Either would be outside the interval.

Review disposition: accepted both optional reviewer suggestions by tightening the resolutionRule discrepancy note and clarifying that the august_2026.first_print public-record label refers to the August release's July reference-month first print; no required fixes were raised.

calibrated forecast · 80% CI
7M[6.9M · 7.2M]
Target metadata

Data point: bls.cps.employed_people_by_occupation.computer_mathematical.september_2026.first_print

More government data forecasts