Census reported on November 2 that private nonresidential construction spending ran at a seasonally adjusted annual rate (SAAR) of $790.6 billion in September 2026, 2.3% above September 2025. That figure counts dollars of work put in place at the prices paid that month. Over the same year, the BLS producer price index for new nonresidential building construction rose 5.4%. Divide one by the other and September’s spending bought about 3.0% less building than a year earlier.
A CFO comparing this year’s revenue or backlog to last year’s faces the same arithmetic. The method below comes from the way the Federal Reserve Bank of Richmond and the Bureau of Economic Analysis convert spending to constant dollars, meaning dollars held at one period’s prices.
What the two numbers measure
Census’s Value of Construction Put in Place survey estimates the dollar value of construction work done each month. The Richmond Fed describes those dollars as covering materials, labor, contractor profit, architectural and engineering work, overhead, and interest and taxes paid during construction. AGC’s weekly Data DIGest notes that the series is not adjusted for inflation.
A producer price index (PPI) tracks what sellers receive for the same output over time. BLS builds the nonresidential building indexes from model buildings broken into assemblies, such as a 5-inch reinforced slab on grade. A cost-estimating firm updates material and installation costs for each assembly every quarter. Each month BLS asks general contractors and trade contractors what overhead and profit they would add if they bid that work now. The index therefore moves with input costs and with contractor markups.
AGC’s guide to construction cost indexes explains that BLS combines its five building indexes, for warehouses, schools, offices, industrial buildings and health care buildings, into the PPI for new nonresidential building construction, series WPU801. It is not seasonally adjusted, and its base is June 2009 = 100.
The nonresidential building price index rose 5.4% in the year to September
BLS producer price index for new nonresidential building construction, index June 2009 = 100, not seasonally adjusted
Most of the index’s movement comes in the first month of a quarter. It rose 1.7% in October 2025 and 2.3% in July 2026, and changed by less than 0.1% in each of the next two months. That pattern fits the quarterly cost update in the BLS method, with monthly markup surveys adding smaller moves in between.
The calculation
- Pick a base month whose prices you want to hold constant. The example below uses September 2026.
- Divide each month’s spending by that month’s price index.
- Multiply by the base month’s index. The result is spending in base-month prices.
- Compare the adjusted figures across months the same way you would compare the raw ones.
For September 2025, that is $773.1 billion times 186.243 divided by 176.749, or $814.7 billion in September 2026 prices. September 2026 spending of $790.6 billion is 3.0% below it.
| Month | Private nonresidential spending ($ billions SAAR) | New nonresidential building PPI (Jun 2009=100) | Spending in September 2026 prices ($ billions) | Change in current dollars | Change in prices | Change in September 2026 prices |
|---|---|---|---|---|---|---|
| September 2024 | 802.3 | 175.48 | 851.5 | |||
| September 2025 | 773.1 | 176.75 | 814.7 | -3.6% | 0.7% | -4.3% |
| September 2026 | 790.6 | 186.24 | 790.6 | 2.3% | 5.4% | -3.0% |
Over two years the difference is larger. Current-dollar spending fell 1.5% from September 2024 to September 2026, and the adjusted figure fell 7.2%.
Matching the index to the work
One index for all private nonresidential spending is a shortcut. The Richmond Fed, following BEA’s method, matches each Census category to its own deflator. Office and health care spending get their own construction PPIs. The warehouse PPI stands in for retail, food service and other commercial buildings. Residential spending is deflated with Census price indexes for new houses under construction, and BEA uses the ENR and Turner cost indexes for some other categories.
That matters for September’s figure. The AGC summary reports that data center spending, which Census counts inside private office, rose 90% from a year earlier, and that private power is the largest private nonresidential segment. The electrical scope of a data hall and the work on a substation differ from the five model buildings that BLS prices for WPU801. Read the 3.0% decline as an estimate for building work in general. A firm that builds mostly power or data center projects would want a deflator closer to that work.
For your own firm, use the index closest to your scope. BLS publishes separate indexes for the five building types and for nonresidential electrical, plumbing and HVAC, roofing and concrete contractors. The steel escalation post covers how to find and pull a single BLS series.
A worked example
This example uses round numbers. A general contractor builds schools and offices, and the price index for that work averaged 5% higher this year than last year.
| Measure | Last year | This year | Change in dollars | Change in the price index | Change in last year's prices |
|---|---|---|---|---|---|
| Revenue | $40,000,000 | $42,000,000 | 5.0% | 5.0% | 0.0% |
| Backlog at year end | $55,000,000 | $60,000,000 | 9.1% | 5.0% | 3.9% |
| Bids submitted | $100,000,000 | $98,000,000 | -2.0% | 5.0% | -6.7% |
Revenue grew 5%, the same as prices, so the firm put in place about the same amount of work as last year. Backlog grew 9.1% in dollars and 3.9% after the price change. Bid volume fell 2.0% in dollars, which is a 6.7% drop in the work the firm chased.
A crew-count check helps here. If field headcount is flat and adjusted revenue is flat, the two agree. If adjusted revenue is flat and headcount rose, productivity or mix changed and the job cost reports should show where.
Limits of the method
BLS revises PPIs for up to four months after first publication. The September spending figure is Census’s first estimate for the month, and Census revised August spending in the same release. Re-run the calculation when either series updates. The next PPI release is November 13, and the next Census spending release is December 1.
The Census figure is seasonally adjusted and WPU801 is not. Over a 12-month comparison the mismatch matters little, because both ends fall in the same calendar month. For month-to-month changes it adds noise.
BLS leaves site preparation, landscaping and design fees out of the building PPIs, while Census counts architectural and engineering costs in spending. A firm with heavy sitework will see its own price change differ from the building index.
How to use it on your jobs
Add one column to the annual review. Next to revenue, backlog and bid volume, show the change in the price index that best fits your work and the change after dividing by it. The backlog in months post covers the companion measure, how many months of revenue the current backlog represents.
To start, download WPU801 from FRED, or the BLS index for your building type or trade. Average it by fiscal year, divide this year’s revenue by the ratio of this year’s average to last year’s, and compare the result with last year’s revenue.