2026 Housing Completions by Region: West’s Outsized Drop

The West supplied 24.5% of U.S. housing completions in January–September 2025 but accounts for 38.3% of the decline through the same months of 2026. Comparing the decline by region shows the South lost more units, with a contribution close to its construction share. These are our descriptive calculations from Census/HUD survey estimates; they do not establish statistically significant differences between regions.

The September report, released October 20, 2026, estimates 969,600 privately owned housing units completed nationally during January–September, versus 1,108,700 in the same months of 2025—a reduction of 139,100 units. These cumulative, unadjusted estimates are separate from the annualized monthly rate in the release’s headlines.

The South leads in lost units; the West exceeds its baseline share

The South’s 75,000-unit reduction accounts for 53.9% of the national decline, nearly matching its 53.3% share of prior-year completions. The West’s smaller, 53,300-unit loss represents a larger departure from its construction share.

Totals and reductions below are thousands of housing units. Shares are our calculations from the displayed January–September estimates in Table 5b.

Reductions equal 2025 totals minus 2026 totals. Each region’s baseline share is its 2025 total ÷ 1,108.7 × 100; its contribution is its reduction ÷ 139.1 × 100. For the West: 271.2 ÷ 1,108.7 × 100 = 24.5%, and 53.3 ÷ 139.1 × 100 = 38.3%, rounded.

Because the source figures are rounded, regional reductions total 139.0 thousand, 100 units below the national reduction. We retain the published national denominator.

Single-family homes account for most of the national reduction

Single-family homes contributed 57.0% of the national decline; units in buildings with five or more homes contributed 42.3%. This national breakdown does not identify which building sizes drove each region’s result.

The same Table 5b supplies these January–September estimates, in thousands of units. Contributions equal each reduction divided by 139.1, multiplied by 100.

The displayed reductions total 139.2 thousand, 100 units above the national reduction. Rounding also prevents the shares from totaling exactly 100%.

Sampling supports some declines, not a regional ranking

Census publishes these January–September changes with 90% confidence margins in percentage points in Table 5b:

• National total: −12.5% ±5.2; South: −12.7% ±5.2; West: −19.7% ±11.6. Each interval excludes zero.
• Northeast: −4.3% ±12.2; Midwest: −4.3% ±11.2. Both intervals include zero.
• National single-family: −10.8% ±4.6; five-plus units: −16.3% ±11.1. Both exclude zero. Two-to-four-unit buildings: −8.8% ±22.7 includes zero.

Where an interval includes zero, the direction remains uncertain. Census calculates these changes from unrounded data, so they can differ from percentages recomputed from displayed totals.

These intervals test each category’s change against zero. They do not test our contribution shares, their gaps from baseline shares or differences between regions. We have not calculated intervals for those comparisons. Census’s intervals cover sampling variability; reporting and coverage errors can also affect estimates. Explanatory notes

A completed unit need not add an available listing

Census generally counts a house as completed when finished flooring is installed, or upon earlier occupancy. In buildings with multiple units, all units count as completed once at least half are occupied or available for occupancy. Each apartment counts as a housing unit. The series excludes HUD-code manufactured homes. Census definitions

Completed units can be rentals or homes already sold. The single-family category also includes qualifying attached homes; building size alone does not identify ownership or tenure. Our completed new-home inventory analysis examines the narrower for-sale measure.

Fewer completions alone cannot establish delays, stronger absorption or a local shortage. The table reports neither construction durations nor local demand relative to available stock—distinctions explored in our analysis of what completion figures can establish about delivery and delays.

Use the regional split to choose local checks

Census publishes these estimates nationally and for four regions; its sample does not support state or local estimates. Regional findings can therefore prioritize investigation, but cannot resolve a particular market’s supply balance. Survey methodology

Builders and investors with western exposure can check whether local deliveries show a similar contraction using permit records, occupancy approvals and project schedules. Compare deliveries with unsold homes or vacant rentals and recent sales or leasing. The South’s large absolute reduction also warrants attention, even though its contribution closely tracks its construction share.

Agents and brokers need subdivision-level available inventory, contracts and incentives to assess competition. Mortgage teams need builder closing schedules and their own application pipelines to plan purchase business. Regional shares do not forecast prices, rents, closings or investment returns.

Revisions can change the comparison

The October 20 release marks September estimates preliminary and July and August revised. Census revises the preceding two months with each monthly release; unadjusted monthly estimates normally stop changing after their second revision. August and September revisions can therefore change the 2026 cumulative totals and our contribution shares. Revision policy

Census’s posted calendar schedules the October 2026 New Residential Construction report for November 18, 2026. Refresh both the totals and the share calculations when that release arrives.