ObservedWork editorial team ·

West housing construction report: September 2026

West housing permits and starts declined in September’s Census estimates. Compare the figures, sampling uncertainty and revisions to August before using them in a territory review.

Archive edition · September 2026 survey month · October 20, 2026 release (CB26-162) · Retrieved and checked October 22, 2026. September is preliminary; August is revised. All comparisons use this release’s vintage. The August report retains the earlier September 17 vintage.

The regional comparison

The West’s September permit rate was 283,000 housing units, 10.2% below revised August and 11.3% below September 2025. Total starts were estimated at 235,000, down 23.2% from revised August and 25.4% from a year earlier.

Values below are thousands of housing units at a seasonally adjusted annual rate (SAAR), not monthly permit documents or project counts. Census calculates percentage changes from unrounded estimates. Single-family permits were unchanged from revised August at the published precision.

The regional comparison
Measure (thousands, SAAR)Sep 2025Aug 2026 (r)Sep 2026 (p)Month changeYear change
Permits, total319315283-10.2%-11.3%
Permits, single-family1841701700.0%-7.6%
Starts, total315306235-23.2%-25.4%
Starts, single-family202191184-3.7%-8.9%

What the uncertainty allows us to say

The total-starts month change of −23.2% has a 90% confidence interval of ±22.9 percentage points: −46.1% to −0.3%. The year change of −25.4% ±14.3 points spans −39.7% to −11.1%. Both intervals exclude zero, supporting a decrease at Census’s 90% confidence level in this release.

Single-family starts are different: −3.7% ±21.3 points month to month (−25.0% to +17.6%) and −8.9% ±20.0 points year to year (−28.9% to +11.1%). Both include zero, so neither establishes a direction of change. These intervals cover sampling variability, not every possible source of error. Permit estimates have nonsampling error; do not apply the starts intervals to permits.

A statistically significant regional monthly change still does not establish a sustained trend, a local contractor’s workload or future material orders. Preliminary estimates can be revised.

Reconcile August before comparing editions

The earlier August report used the September 17 release. The October 20 release revises August’s West figures as shown below. September’s month comparisons use the revised August column, not the older August values. These are revisions to the same survey month, not new September activity.

July’s overlapping West SAAR figures remain 308 total permits, 178 single-family permits, 244 total starts and 154 single-family starts in both reviewed releases. The prior dated report and its downloads remain intact.

Reconcile August before comparing editions
August measure (thousands, SAAR)September 17 vintageOctober 20 vintageRevision
Permits, total313315+2
Permits, single-family169170+1
Starts, total323306-17
Starts, single-family198191-7

Housing units, annual rates and regional coverage

Permits describe new privately owned housing units authorized in permit-issuing places. Starts are a separate survey measure and generally begin with excavation for footings or foundations. They do not follow the same month’s permits through a project pipeline. An apartment building can represent many housing units.

SAAR adjusts for estimated seasonal patterns and expresses the monthly pace as an annual rate. It is neither the number built so far this year nor a forecast. These measures do not cover all commercial construction, remodeling orders or contractor identities.

The West includes Alaska, Arizona, California, Colorado, Hawaii, Idaho, Montana, Nevada, New Mexico, Oregon, Utah, Washington and Wyoming. No city, county or individual distributor territory is estimated here.

Source vintage and next release

We retrieved the October 20 PDF on October 22 and checked West columns in tables 1a and 3a, including preliminary/revised labels and the confidence-interval rows. The JSON records the source hash, retrieval time, values and revision ledger.

The official calendar checked October 22 schedules September’s revised BPS release for October 27 and the next NRC release for November 18. October 27 data were not yet due at this check and are not included here. A later BPS review must retain its own vintage and explain any differences; it must not silently replace this October 20 edition.

Change log: initial September report, October 22, 2026. If source access fails, retain this dated edition and mark the new review unresolved. No latest-data dashboard or live territory collection is established by this report.

Use the report in a supplier territory review

Use the regional decline as context for questions about the counties you serve, product categories and customer segments. Gather local evidence before changing an account priority or inventory plan. An issued permit does not establish an open bid, an unawarded job or a purchase order.

Business teams can request a source and coverage review for a specific geography and work category. ObservedWork must confirm access and scope; this regional archive does not establish ongoing local collection.

West region: revised August and preliminary September

Thousands of housing units, SAAR; zero baseline and common scale. October 20, 2026 vintage. Starts bars are point estimates; the total-starts monthly decline excludes zero at 90% confidence. Exact values, uncertainty and prior-vintage revisions appear above.

Your checklist

0 of 4 reviewed. Progress stays in this tab; download the worksheet to keep notes.

Sources and method

Sources reviewed October 22, 2026 unless a record carries its own observation date.

Continue with ObservedWork