In August 2026 construction carried 0.81 unfilled openings for every hire it made, against 0.63 a year earlier. Openings rose 18% over that year while hiring fell 8%. Demand is not collapsing. It is getting harder to convert into people.
Hiring conditions right now
Everything else published about the trades describes the next ten years. This is the only part that describes this month.
- 0.81 construction job openings per hire in August 2026, up from 0.63 a year earlier.
- 2.9% construction job openings rate, below the all-industry rate of 4.3%. The trades do not have unusually many open seats. They have seats that stay open.
- −8% construction hires against the same month a year earlier, while openings over the same period rose 18%.
| Series | 2025/08 | 2026/08 | Change from prior month | Rate |
|---|---|---|---|---|
| Job openings | 213 | 251 | −48 | 2.9% |
| Hires | 336 | 308 | −50 | 3.7% |
| Total separations | 351 | 272 | −43 | 3.2% |
Source: BLS Job Openings and Labor Turnover Survey, August 2026, Table 1, Table 2 and Table 3. August 2026 figures are preliminary. The Bureau of Labor Statistics revises the most recent month in the following release, so these levels will move.
All three series fell month over month - openings by 48, hires by 50 and separations by 43 (thousands). A market cooling on every measure at once is not the same as a market with no demand: it is one where fewer people are moving, which is precisely when an open seat takes longest to fill.
The decade: 1,082,700 hires a year to stand still
Across the 18 federal occupations our 20 trades map to, employers must hire about 1,082,700 people a year against a workforce of roughly 10.5 million - 10.3% turning over annually. Almost none of that is growth. It is replacement.
We have published the full analysis separately, occupation by occupation, with the arithmetic shown: Commercial Trades Replacement Demand →
The age gap is in the supervisor layer, not the field
The industry worries about an ageing trades workforce. The field does not look old. 12 of the 18 trades we publish a median age for sit below the 42.1-year median for all US workers, and the youngest, commercial roofing, is 36.6 - 5.5 years under it.
The layer above them is a different story. Every one of the 4 supervisory occupations sits above the national median, from 45.2 to 47.6. Between the youngest trade and the oldest supervisory layer there are 11.0 years. The scarcity that actually bites is not apprentices. It is the people who run them, and they are the hardest seat in the trade to refill.
The full breakdown by trade, with every median age sourced: Trades Workforce Age Report →
Where the public data runs out
3 of the 20 commercial trades we recruit for cannot be measured from federal statistics at all: Fire Protection, Physical Security, Commercial Irrigation. There is no dedicated occupation code with published employment, wage and projection data for them.
We say so rather than filling the hole. Any national figure offered for those trades is built on a proxy occupation that includes a great deal of work that is not theirs, and the error is not small or knowable. Where we have nothing trustworthy to publish, the honest output is the gap itself.
The same caution applies to the figures we do publish. The two federal surveys disagree with each other: the Occupational Outlook Handbook counts 821,000 electricians where the Current Population Survey counts over a million. They measure different things - one counts jobs, the other counts people who call themselves electricians - and neither is wrong. It is a reason to treat any single national number as a direction rather than a measurement.
Common questions
Is commercial trades hiring getting easier or harder right now?
Harder on the posting side, slower on the filling side. In August 2026 construction carried 251,000 job openings against 308,000 hires, or 0.81 openings for every hire. A year earlier that ratio was 0.63. Openings rose 18% over the year while hires fell 8%. August 2026 figures are preliminary. The Bureau of Labor Statistics revises the most recent month in the following release, so these levels will move.
Does construction have more unfilled jobs than other industries?
Proportionally, no. Construction's job openings rate in August 2026 was 2.9% of employment against 4.3% across all industries. The difficulty in the trades is not an unusual volume of open seats; it is that the seats take longer to fill and the people who fill them are harder to replace.
How many people do the commercial trades need to hire each year?
About 1,082,700 a year across the 18 federal occupations our trades map to, against roughly 10.5 million workers - 10.3% of the workforce annually. Almost all of that is replacement rather than growth. The full analysis is in our replacement demand report.
Which commercial trades cannot be measured from federal data?
3 of the 20 trades we recruit for have no dedicated federal occupation: Fire Protection, Physical Security, Commercial Irrigation. Any figure presented for those trades at national scale is an estimate built on a proxy occupation, and we do not publish one.
Method and sources
- Current conditions - BLS Job Openings and Labor Turnover Survey, August 2026, release USDL-26-1547 (29 September 2026). Levels and rates are seasonally adjusted and read directly from Tables 1, 2 and 3. August 2026 figures are preliminary. The Bureau of Labor Statistics revises the most recent month in the following release, so these levels will move.
- Openings per hire - our arithmetic: construction job openings divided by construction hires in the same month.
- Decade outlook - BLS Occupational Outlook Handbook, US Bureau of Labor Statistics, Occupational Outlook Handbook. Wages are May 2025; employment is the 2025 base year; growth and annual openings are the 2025-35 projections.. Per-occupation sources are listed on the replacement demand report.
- Median age - BLS Current Population Survey, Household Data Annual Averages, Table 11b, 2025 annual averages. Per-trade figures are on the workforce age report.
- Every figure on this page is computed at build time from those datasets. None is typed into the text, so a revision at the source changes the page.