Who is building. Construction jobs are 10 % or more above September 2019 in 9 of 15 big states, and no more than 2 % above in the other six.

Nine of 15 states have construction jobs 10 % or more above September 2019, five of them 15 % or more; six are within 2 % of it or below.

The Bureau of Labor Statistics counts the jobs in construction in every state each month, and the Federal Reserve Bank of St. Louis publishes the series. We read the construction job count for 15 large states and compare September 2026 with September 2019, the last September before the pandemic.

In September 2026 five states have 15 % to 32 % more construction jobs than in September 2019: Arizona +32.2 %, North Carolina +26.5 %, Ohio +20.7 %, Texas +17.6 % and Florida +15.2 %. Four are 10 % to 15 % above, from Illinois +10.7 % to Michigan +14.5 %. Six are within 2 % of 2019 or below it: Massachusetts +1.8 %, New Jersey +0.3 %, Pennsylvania −0.3 %, California −1.5 %, Washington −1.6 % and New York −5.9 %.

The nation sits in the middle. Construction jobs in the United States are 8,370,000 in September 2026 against 7,547,000 in September 2019, +10.9 %. The 15 states together went from 4.92 million to 5.37 million, +9.2 %. The 15 grew less than the nation, so the other states grew more: together they went from 2.63 million to 3.00 million, +14.2 %.

Construction also took a larger share of work. In 10 of the 15 states the share of all jobs that are in construction is higher than in September 2019; in New Jersey, Pennsylvania, California, Washington and New York it is lower. Arizona has the highest share, 6.9 %, up from 5.8 %; Washington had the highest in 2019, 6.4 %, and is at 6.0 %.

A job count is not a count of crews on housing. Construction jobs include non-residential building, roads and utilities, and specialty trades, and the series counts jobs by where the employer’s establishment is, not what is being built. It says nothing about wages, hours or how hard it is to hire. How we read construction jobs

The bands are a way to count, not a forecast. Illinois at +10.7 % and Michigan at +14.5 % are in the same band here; the next part asks what the permits say about the same states.

Permits per thousand construction workers fell in six states and rose in eight; Texas added 138,200 construction jobs for 0.4 % more permits.

A permit is the start of a building and a crew is what finishes it, so we set one beside the other. The Census Bureau counts the new private housing units authorized by building permits in each state; we take the average of the twelve monthly values to September, in units a year, and divide by the thousands of construction jobs in the same month. It is a count of permits per crew, not a measure of what a crew can build.

In the 12 months to September 2026 the 15 states permitted 77,054 housing units a year, +7.1 % on the 12 months to September 2019. The units per thousand construction workers went from 14.6 to 14.3. The 15 gave a little more to permit and had a little more to build with, and the two nearly cancelled.

Inside the 15 the paths part. North Carolina has the most permits per thousand construction workers, 24.4, then Georgia 21.8, Florida 20.2, Texas 17.5 and Arizona 16.8. Illinois has the fewest, 6.1, then Massachusetts 6.6, Pennsylvania 8.2, Michigan 9.1 and Ohio 9.4. The ratio fell in six states, most in Arizona (20.5 to 16.8), Texas (20.5 to 17.5) and Washington (17.4 to 14.4), and rose in eight, most in New York (8.1 to 10.6) and Ohio (8.0 to 9.4); it is unchanged in Massachusetts.

Texas is the sharp case. It added 138,200 construction jobs, from 783,400 to 921,600, while the permits went from 16,024 to 16,089 a year. Arizona added 55,400 jobs for +7.7 % more permits. North Carolina kept pace: +26.5 % jobs and +30.2 % permits. Switch the views to see it state by state.

What it means for an underwriter. A ground-up schedule or cost assumption is set in a state, and the state’s crew count is one input to it. In Arizona and Texas the crews grew much faster than housing permits, which is consistent with the jobs going to other work or with fewer starts per permit; the series cannot say which, and we did not measure it. We do not claim that a high ratio means cost or delay, because the review has no wage or schedule series.

The permit counts move with the month. Ohio’s +41.8 % is from a base of 1,836 units a year, and Washington’s −18.2 % and Illinois’s −12.8 % are the two falls. A single month in a small state can move the twelve-month figure.

Fourteen states fell to a low in April 2020; six set their high in 2026, and New York has not regained its 2019 level.

The chart shows each state’s construction jobs from June 2016 to September 2026 (124 months), with September 2019 as 100. The lines share a drop: in April 2020 14 of the 15 states hit their lowest month since March 2020, and Texas hit its low later, in February 2021 at 91.0. The drop was deepest in Michigan (59.8), New York (60.9) and Pennsylvania (61.5) and shallowest in Arizona (99.4).

By February 2020, just before the drop, all 15 stood within 3 % of September 2019, between 99.1 and 102.5. Since then the lines part. Arizona is now at 132.2 and North Carolina at 126.5; each is plotted against its own September 2019. Six states set their highest month in 2026: North Carolina and Ohio in August, Texas in June, Illinois in September, New Jersey and Pennsylvania in January.

Pick a state to bring its line forward against the grey of the other 14. California reached 103.6 in January 2024 and is at 98.5, 4.9 % below its high. Washington reached 105.9 in September 2022 and is 7.1 % below it. New York’s highest month since June 2016 was January 2019, 100.2, and it is at 94.1 now. In the last 12 months the count rose in 8 states and fell in 7.

For a lender the date is the point. A loan sized on a ground-up schedule in a state where the crews were at 59.8 % of their September 2019 count in April 2020 carried one labor market, and the same loan today carries another with 200,200 construction jobs in Michigan. The state’s own line, not the national count, is the one to read at the date of the loan.

What to do with this

If you underwrite, lend on or develop ground-up property: read the state’s crew trend and its permits at the date of the decision, and not the national count. A national figure of +10.9 % would hide both Arizona at +32.2 % and New York at −5.9 %.

Titleman reads the offering memorandum, budget and schedule of a property deal and returns the issues and the questions for the sponsor. The review can set a deal’s construction timeline against the construction jobs and permits of its state, so the check is a number in the file, not a feeling about the labor market.