July 2026 Residential Construction Spending: Dollars vs. Output

Census’s initial estimate of July 2026 residential construction spending put private residential work at an $859.0 billion annual rate, down 1.3% from revised June. Single-family spending led the estimated decline, but the aggregate change was not statistically significant: its 90% confidence interval included zero. The September 1 release does not establish that building costs fell or that fewer homes were delivered.

What the September 1 release measures

Released September 1, 2026, at 10 a.m. EDT, report CB26-140 measures July activity. Its headline figures are seasonally adjusted annual rates in current dollars: seasonal adjustment removes recurring calendar patterns, while annualization expresses a month’s adjusted pace over a year. Neither operation removes inflation. Table 1 explicitly says the figures are not adjusted for price changes.

The $859.0 billion estimate is therefore neither an amount spent during July nor a forecast for the full year. Construction spending measures the dollar value of work done on new structures and improvements to existing ones, as Census explains in its construction-spending overview.

A project can contribute spending while it remains unfinished. To assess delivery, pair this report with new-construction indicators that track units and project stages. Dollars alone cannot answer how many homes became available.

Single-family spending drove the estimated decline

The table uses the September 1 release throughout, including its revised June estimates. Dollar levels are billions at seasonally adjusted annual rates; percentage changes are point estimates.

Source: Census Table 1. Improvements are calculated by subtracting new single-family and multifamily spending from private residential spending. They are included in the total but not shown separately in Census Table 1; their percentage changes are our calculations.

Single-family spending fell by $12.931 billion at an annual rate. Multifamily added $0.274 billion and the improvements residual added $1.085 billion, partially offsetting that decline. These figures describe each component’s contribution to the estimated movement, not its cause or statistical significance.

The broader residential total, including public construction, was $871.205 billion at an annual rate. Keep that separate from the private-sector total when comparing reports. Both figures come from the same Census table.

For contractors, the component split matters. A business concentrated in new houses faces a different reported spending pattern from one concentrated in improvements. The combined total can conceal that difference, and the improvements residual does not count renovation jobs.

Lower spending does not establish lower construction costs

Spending combines the amount of work, its price and the mix of projects. Lower expenditure could reflect less work at unchanged prices, cheaper work at unchanged volume, or a shift toward projects with lower costs. This release does not identify which mechanism produced July’s estimated change.

Consider a hypothetical contractor performing comparable work packages. Completing 100 packages at $10,000 each produces $1 million of expenditure. Completing 90 at $10,500 produces $945,000: spending falls 5.5% even though the cost per package rises 5%. This illustrates the distinction; it is not an estimate of July construction.

Project mix adds another complication. A change in the share of larger homes, apartments or improvements can change dollar totals without a matching change in unit counts. Nor does a national spending estimate establish what a local subcontractor will charge.

For decisions about construction costs and financing, compare bids for the same scope separately from the spending trend. Check quantities, specifications and schedule before interpreting a lower bid as evidence of lower unit costs.

Revisions and uncertainty affect the comparison

The August 3 release initially put June private residential spending at $877.118 billion. The September report lowered it to $870.574 billion. Comparing July with the original June figure would produce a roughly 2.1% decline instead of the reported 1.3%. Use both months from the same release.

Census reports July’s private residential monthly change as −1.3%, with a margin of ±1.3 percentage points, and flags it as statistically indistinguishable from zero. That does not prove spending was unchanged. It means the estimate does not establish the direction of the aggregate movement at the stated 90% confidence level. See the release and explanatory notes.

Those notes also caution that confidence intervals cover sampling variability, not every source of error. Late responses, missing reports and undercoverage can affect estimates, and revisions can alter the picture. Census says identifying an underlying trend can take two months for total construction and as long as eight for particular categories.

For a development budget or loan draw review, use the report to identify questions for the project records: how does spending compare with verified work completed, remaining quantities and current quotes? Keep a release date beside each benchmark drawn from housing data and reports, so the basis of the decision remains clear when estimates are revised.