Data & Accountability • Thursday, November 19, 2026

October Housing Starts Rise in Estimate, but Error Range Includes Decline

Census and HUD report a 1.309 million annual pace while permits fall; the release cautions against calling one month a trend.

Tabitha Tally — approved human correspondent in a representative reporting setting

Illustrative correspondent image; not a photograph of the reported event.

By Tabitha Tally • Thursday, November 19, 2026

The latest Census Bureau and Housing and Urban Development construction report estimates that privately owned housing starts reached a seasonally adjusted annual pace of 1.309 million in October. Released November 18, it puts the increase from revised September at 9.9 percent. But the accompanying margin is plus or minus 16.8 percent, so the reported change does not provide sufficient statistical evidence to conclude that starts actually increased rather than decreased.

Permits tell a different part of the pipeline. October authorizations were at a 1.295 million annual rate, 7.4 percent below revised September. Starts were measured against September’s revised 1.191 million pace. These are October activity estimates published in November, not counts of homes built during November. The release warns that monthly movements can be irregular and that establishing an underlying trend in starts may take six months. That caution belongs beside the headline figure.

Census definitions help explain why permits, starts and completions should be read separately. A permit is a local authorization to proceed. A start occurs when excavation begins for a building’s footings or foundation, with all units in a multifamily building counted as started at that point. Completion follows a different milestone, including finished flooring for a house. Each measure describes a stage, so an increase at one stage need not be matched by an increase at another during the same month.

The scope is also narrower than all housing activity. The construction statistics cover new privately owned residential units and exclude such categories as publicly owned units, manufactured homes and units created within existing structures. New residential sales are another series, limited to qualifying new single-family transactions involving both house and land. Those boundaries matter for local debates: an apartment conversion, a manufactured-home delivery and a newly excavated subdivision are not interchangeable observations in these statistics.

The Survey of Construction methodology defines a seasonally adjusted annual rate as the adjusted monthly value multiplied by twelve. It enables comparison of a month’s pace with an annual total; it is not a statement that 1.309 million homes were started in October alone. The methodology also describes routine revisions to the latest estimates. Readers comparing releases need the current revised September number, rather than an older preliminary value, to reproduce this month’s percentage comparison.

The resulting picture is mixed and uncertain, rather than a clean verdict on a housing recovery. The point estimate for starts rose, permits fell, and the starts confidence interval includes zero. None of those observations establishes the cause of a particular builder’s decision or the direction of a local rental market. The next releases will help show whether the apparent movement persists. For now, the best reading keeps the estimate, its uncertainty and the construction stage together.

Sources

Follow the Entire Crew

Follow @FrontPageCrew for daily reporting and newsroom updates.