US construction spending fell 1.9% in September: what to read before you react

If a single number could build a house, every contractor would own a calculator and nothing else. On Monday, November 2, the Census Bureau published one of those numbers: United States construction spending for September. It fell. Headlines followed. Somewhere, a project manager looked at the headline, looked at the site, and wondered whether to panic or to finish the coffee first.

Finish the coffee. Then read the number properly. This post summarizes what Census reported, where the drop came from and what a project manager or a business owner can take from it. I am an operations person, not an economist, so this is a reading and not a forecast.

What the Census Bureau reported

The Census Bureau's monthly construction spending release for September 2026 came out on Monday, November 2. It estimates the value of construction work put in place, at a seasonally adjusted annual rate. That phrase matters: it means the monthly figure is adjusted for the usual seasonal swings and then expressed as if the month's pace lasted a full year.

According to the release, total construction spending was $2,186.1 billion in September. That is 1.9% below the revised August estimate of $2,227.4 billion, and 2.7% below September 2025, which was $2,245.8 billion. Reuters, as carried by GV Wire, reported that economists polled by Reuters had expected a 0.3% gain, so the number came in well below the forecast.

Where the drop came from

The total is a sum, and the parts did not move together. Census splits it into three groups:

• Private residential: $848.8 billion, down 5.5% from a revised $898.0 billion in August.
• Private nonresidential: $790.6 billion, up 1.1% from a revised $782.3 billion.
• Public: $546.8 billion, down 0.1% from a revised $547.1 billion.

Reuters, via GV Wire, adds some detail. Within residential, spending on single-family homes was unchanged from August and down 2.4% from a year earlier, multifamily rose 0.1%, and the reporter attributes the weak spot to renovation projects, possibly because of high materials costs. Within private nonresidential, spending on power plants rose 1.2%. In the public sector, state and local spending fell 0.3% while federal spending rose 2.5%.

Put those together and the story is less "construction fell" and more "home projects pulled back while commercial and public work held." For a hospital builder, a road contractor and a kitchen remodeler, the same headline means three different things. If you want a refresher on why the parts matter more than the total, the dashboards guide explains how a single big number can hide the one that matters.

The revision: why August is part of this story

Here is the detail that makes the headline less dramatic. August's gain was first reported as 0.9% and has now been revised to 2.0%. A higher August makes September's fall look steeper, because the drop is measured from a higher base.

Revisions are normal in this kind of estimate. Census uses survey responses and other sources, and it updates earlier months as more data arrives. The lesson is not that the number is wrong. The lesson is that the latest month is the least settled month, and a good reader looks at two or three months together before drawing a line.

The margin of error

Every Census estimate comes with a margin of error. In the September release, the monthly change in total spending is shown as -1.9% with a margin of plus or minus 0.7 points, and the change from a year earlier as -2.7% with a margin of plus or minus 1.5 points. I read these as the confidence intervals Census normally publishes, and I have not read the full technical notes, so check them if you plan to cite the numbers.

In plain English: the total very likely fell, because even the best case in that range is still a drop. But the exact size is a range, not a point. This is the same habit I wrote about in the post on two jobs reports: before acting on a number, know how firm it is.

What it means for a project manager

National construction spending is a thermometer for the whole country, not for your building. It does not tell you the price of your concrete next month, how many bidders will answer your invitation or how long your electrician is booked. It can, however, nudge your questions in the right direction. Five habits:

1. Split the total. Whenever you see a big number, ask which part moved. Here it was residential. If you do commercial or public work, the headline is about your neighbor, not you.
2. Read the revision. Look at what the previous month was before it was changed. A fall from a revised high is not the same as a fall from a plateau.
3. Mind the margin. A -1.9% with a margin of 0.7 is a signal. A +0.2% with a margin of 0.7 is noise. Treat them differently.
4. Use three months, not one. Put the last three readings in a line and look at the direction. This is the same logic as the scoreboard in Project Focus: a trend beats a point.
5. Translate it into your project. Ask three local questions: have supplier quotes changed since the last time I priced this, how many bids did my last invitation get and has the lead time on the long-lead items moved? Those answers live in your project, not in a national release. The schedule delays guide and the cost management guide show where to look.

If you build or renovate in Utah

Nothing in this release is about Utah, and I will not pretend otherwise. If you are planning a project here, local conditions matter more: the state's road program, discussed in UDOT's 2026 season and I-215 West, the build-out around the 2034 Games in what project managers can learn from it, and your own contractor's backlog. The best use of a national number is as a reason to ask your contractor and your suppliers one more question, not as a substitute for the answer.

A five-minute exercise

Pick the three numbers you watch most closely on your current project: perhaps the latest cost forecast, the days of float on the critical path and the number of open change requests. For each, write down three things: what the number was last month, what it is now and how firm it is. If you cannot say how firm it is, you have found the real risk. The earned value guide is a good next read for the cost side.

What I do not know

I have read the Census release page and the Reuters report carried by GV Wire, not the full technical notes. I cannot tell you how much of the residential decline is noise and how much is a real pullback, and I cannot predict whether the next release will bring another revision. Census typically publishes the October figures in early December. I will not guess what they will say.

Related reading

For how two sources can describe the same sector in opposite ways, see Two jobs reports, two stories. For a plain method to keep a project honest when the news is noisy, see the 90-day sprint and the Project Focus series.

This post summarizes public data as of November 2, 2026 and reflects my own reading, not an official position. It is not investment, financial or legal advice. Check the original sources for the full figures.

Sources

• U.S. Census Bureau: Construction Spending, September 2026 (Nov 2, 2026)
• GV Wire (Reuters): US construction spending tumbles in September (Nov 2, 2026)

Facts come from the sources above. The reading of those facts is mine.