Three point one percent.
That’s where construction unemployment sat in August 2026, an all-time low for the series, and it is the number underneath every wage figure in ENR’s third-quarter cost report. The industry added 120,000 jobs since August 2025, a 1.5% gain, against 0.4% growth for total nonfarm payrolls. Construction is hiring roughly four times as fast as the rest of the economy into a labor pool that has already run dry.
What that bought workers
Union settlements signed in the first half of 2026 carried first-year increases averaging 4.9%, with 4.1% across all contract years. The three years before ran 4.5% to 4.7%, so the first-year number is the one that moved.
By trade, the spread is wide. Carpenters took 7% in first-year settlements. Plumbers 5.9%, sheet metal 5.6%, Teamsters 5.5%, bricklayers 5.1%. At the bottom, roofers at 2.9%, plasterers 4.2%, cement masons 4.3%. A four-point gap between carpenters and roofers in the same bargaining season is a demand signal, not a negotiating-skill signal.
Merit shop averaged 4.1% in 2025 with a 3.9% forecast for 2026, though PAS president Jeff Robinson expects it lands at 4.1% again. The forecast that matters least is the one for electricians: 4.3% on paper, and at contractors above $50 million in revenue an actual average of 6.7% with a median of 8.8%.
The median is the finding
A median above the mean on electrician increases means this isn’t a few outlier contractors dragging an average. Most large electrical employers are paying it. Ken Simonson, chief economist at AGC of America, named the reason without ceremony: “There’s tremendous demand for electricians, HVAC and plumbing folks who can build data centers.”
That’s the mechanism. Data center work concentrates electrical scope into a small number of large contractors who compete for the same journeymen, and the ones who can’t staff a project lose the project.
Two things worth keeping in view
Jerry Crawford, managing director at Turner Construction, pushes back on the broad reading: “Overall, I wouldn’t characterize what we’re seeing as broad-based labor inflation. It’s more about certain markets and trades.” The trade spread supports him. Roofers at 2.9% are not in a wage spiral.
And the provenance is mixed. The 3.1% and the 120,000 jobs are BLS figures. The settlement percentages are Construction Labor Research Council survey data covering signed agreements, not a census of all of them, and CLRC’s own projection is that first-year increases flatten or dip slightly in 2027 and 2028. The merit-shop numbers are PAS survey data, self-reported, with the 2026 figure a forecast whose author is already second-guessing it.
Matt Minarik, a senior data analyst at CLRC, offered the least surprising explanation for the first-year spike: “Some folks said that they were giving higher first-year increases to account for inflation.”
Estimators pricing 2027 labor should probably use the median, not the forecast.
Source: ENR, 2026-09-23.