A chart circulating this week claims 158,000 women and just 4,000 men gained jobs in August 2026 — "98 of 100" new jobs going to women. The chart's 162,000 total is real. Its 12-month men/women numbers are real. But its headline monthly split doesn't appear in any BLS table we can find — and the actual household-survey number for August shows men gaining more jobs than women, not less. This report walks through exactly which BLS tables say what, with direct links, so you can check it yourself.
Every figure below comes directly from the Bureau of Labor Statistics' own published Employment Situation release for August 2026 (issued September 4, 2026): the summary narrative at empsit.nr0.htm, Table A-1 ("Employment status of the civilian population by sex and age") at empsit.t01.htm, and Table A-7 ("Employment status of the civilian population by nativity and sex, not seasonally adjusted") pulled via BLS's own historical-data query tool at webapps/legacy/cpsatab7.htm. The viral chart under review was posted by the X account @LayoffAI ("LayoffHedge") and is quoted and screenshotted directly below; nothing in it is misquoted or taken out of context. We checked seasonally adjusted and not-seasonally-adjusted figures, and both the 16-years-and-over and 20-years-and-over age cuts, before concluding the claimed 158,000/4,000 monthly split does not match any published BLS series.
The chart cites "BLS Employment Situation, Aug 2026" as its source for the monthly figures and "BLS CPS" for the 12-month figures — two different attributions for the two halves of the same graphic, which turns out to be the whole story.
Every monthly BLS Employment Situation release is built from two independent surveys that measure different things, run by different methods, and are not designed to be added together or split into each other:
A men/women split of "who got the new jobs" can only come from the household survey — the establishment survey structurally cannot produce one. So when a chart's monthly split adds up to precisely the establishment survey's total (158,000 + 4,000 = 162,000), that is itself a signal something has been merged that shouldn't be: a payroll-count total from one survey, paired with a demographic split that didn't actually come from measuring that total.
Table A-1 reports seasonally adjusted employment levels by sex every month. Comparing July 2026 to August 2026 — the same comparison the viral chart claims to make — gives a result that runs the opposite direction:
| Group (16 yrs+, seasonally adj.) | July 2026 | August 2026 | Change |
|---|---|---|---|
| Men, employed | 84,978,000 | 85,337,000 | +359,000 |
| Women, employed | 77,199,000 | 77,409,000 | +210,000 |
| Total, employed | 162,177,000 | 162,746,000 | +569,000 |
Source: BLS Table A-1, series LNS12000001 (men) and LNS12000002 (women), seasonally adjusted employment level. The men + women rows sum exactly to the total row — a basic internal-consistency check the claimed 158K/4K split does not need to pass because it isn't drawn from this table.
The chart's 12-month figures are a different story — they're real, and they trace cleanly to the same Table A-1, comparing August 2025 to August 2026:
Both round almost exactly to the numbers on the chart. This is the real, well-documented, twelve-month divergence in the household survey between men's and women's employment — a genuine and citable data point. The problem isn't that half the chart is wrong; it's that the two halves come from doing two different kinds of math (one real year-over-year household-survey comparison, one unsourced monthly split grafted onto an unrelated payroll total) and presenting them as if they were measured the same way.
Table A-7 breaks the household survey out by nativity and sex instead of just sex, and it's not seasonally adjusted — a second, independent way to sanity-check any single-month demographic claim. We maintain a local monthly mirror of this table (see our presidential job-creation scorecard) going back to February 2022. Charting native-born and foreign-born month-over-month employment change side by side looks, at a glance, like the two move in lockstep-opposite directions — one up, the other down, almost every month. Actually measuring it tells a more precise story:
If native-born and foreign-born gains were truly trading off one-for-one every month, their combined total would barely move. It doesn't — the combined total is almost as volatile as either piece on its own. That's consistent with what the household survey's demographic subgroup breakdowns are known for: a survey of roughly 60,000 households produces a lot of sampling noise once you slice it down to a subgroup, whether the slice is by sex, by nativity, or any other single-month cut. That's a real reason to treat any single month's demographic split — ours included — with caution, and to check a claimed trend against more than one month before repeating it.
bls.gov/news.release/empsit.nr0.htm. It states the establishment-survey total in the first sentence and separates "Household Survey Data" from "Establishment Survey Data" in its own section headers — that split is the whole ballgame.empsit.t01.htm). For race, Table A-2. For Hispanic origin, Table A-3. For nativity, Table A-7. If a claim breaks out payroll jobs (the establishment-survey total) by any of these, it's not using a table that exists.LNS12000001 for men's seasonally adjusted employment level) that you can look up directly at data.bls.gov/timeseries/<series ID> for the full published history, not just the two most recent months.This investigation is produced independently — one person, 23 years of data work, living on $1,419/month Social Security. If this work matters to you, a small donation makes a real difference.
♥ Donate via Square More Reports