The Statistically Significant Story: Housing Completions Fell 15.3% YoY While Starts Spin Steals the Spotlight

In the official press release dated January 09, 2026 (covering October 2025), the headline narrative leaned on a monthly uptick in single‑family starts and a drop in total starts. But by the release’s own methodology, those monthly and year-over-year moves carry 90% confidence intervals that include zero—i.e., they’re not statistically significant. The number that is? Completions fell 15.3% versus October 2024 with a ±12.1% interval that does not cross zero, signaling a clear, statistically significant weakening in delivered housing supply.

Here’s what the data actually reveals:

• Total starts fell -4.6% m/m and -7.8% y/y, but both changes have 90% confidence intervals that include zero.

• Single‑family (SF) starts rose +5.4% m/m, also not statistically significant.

• Completions rose +1.1% m/m (CI includes zero) but are down -15.3% y/y with a ±12.1% interval—statistically significant deterioration.

• Permits (1,412,000 SAAR) exceed starts (1,246,000 SAAR); the “authorized but not started” backlog rose on an NSA basis: total 251.7k → 261.6k, SF 140.3k → 143.2k, 5+ 107.0k → 113.8k.

• SF permits slipped -0.5% m/m (876k vs 880k) and are -7.0% YTD vs 2024, undercutting the “resilient SF” storyline.

October 2025: Numbers at a Glance (SAAR)

Context matters: permits are near-flat m/m (-0.2%), down -1.1% vs Oct 2024, and well below late‑2024 levels (Nov 1,508k; Dec 1,480k) after a mid‑2025 trough (Aug (r) 1,330k), with only a partial rebound into October (1,412k). Starts are down from September (1,306k → 1,246k; -4.6%, CI includes zero) and below last October (1,352k; -7.8%, CI includes zero).

The Significance Mirage: Directional Language Without Statistical Backing

The release frames October as a month of lower total starts and rising SF starts. True numerically; not true statistically. The 90% confidence intervals for:

• Total starts m/m and y/y changes,

• SF starts m/m change,

• Completions m/m change,

all include zero. By the agency’s own guidance, the direction shouldn’t be over‑interpreted. The statistically sound headline is the -15.3% y/y drop in completions with a ±12.1% interval that excludes zero—fewer homes actually being delivered versus a year ago. That’s the supply story that matters for prices, rents, and shelter inflation.

Permits vs. Starts: The Conversion Squeeze Is Growing

October permits outpaced starts (1,412k vs 1,246k), while the “authorized but not started” backlog rose on a not‑seasonally‑adjusted basis:

• Total: 251.7k → 261.6k

• Single‑family: 140.3k → 143.2k

• 5+: 107.0k → 113.8k

Simultaneously, completions exceeded starts (1,386k > 1,246k), a classic pipeline drawdown. In isolation, that might sound like progress—until you note completions are well below last year (a significant -15.3% y/y). Translation: we’re finishing previously initiated projects at a slower clip than in 2024, and we’re not replacing them fast enough. That’s conversion friction, not momentum.

The multifamily (5+) profile is textbook digestion:

• Permits: 481k vs starts 347k,

• Completions: 367k exceed starts,

• Backlog (NSA) rises (107.0k → 113.8k).

This is not an expansionary set‑up; it’s a slower‑moving pipeline meeting tighter financing and execution constraints.

Regionally, the South holds the largest backlog (NSA “authorized but not started,” preliminary October): total 166.0k; SF 92.6k—a concentration risk the release leaves on the cutting‑room floor.

Single‑Family Momentum? The Pipeline Says No

The single‑family narrative leans on a +5.4% m/m increase in SF starts, but the forward indicator—permits—doesn’t cooperate. SF permits fell -0.5% m/m (876k vs 880k) and are -7.0% YTD versus 2024. SF starts and permits are near parity (874k vs 876k), leaving little immediate headroom. SF completions ticked up to 1,009k m/m (CI includes zero), yet the broader flow of completions remains weak versus last year.

This is consistent with what we’ve seen all year. Back in August 2025, starts were -6.0% y/y; October is -7.8% y/y. The SF permitting path has sagged from late‑2024 highs (Nov 1,508k total permits; Dec 1,480k) to mid‑2025 softness (Aug (r) 1,330k), with only a partial recovery by October (1,412k). If affordability and construction financing pressures aren’t explicitly in the release, the data still screams them. You don’t get weaker SF permits YTD and rising “authorized but not started” without friction in capital, labor, or both.

Multifamily: Big Paper, Slow Shovels

On the multifamily side, October offers a clean trifecta of digestion:

• Permits stay elevated (481k),

• Starts lag (347k),

• Completions exceed starts (367k > 347k),

• Backlog increases (NSA).

Add the YTD permitting mix shift—total permits -3.0% YTD vs 2024; SF -7.0%; 5+ +5.7%—and you get a sector that’s still authorizing projects but struggling to convert them. Financing remains a gating factor. The growing backlog points to extended timelines, not a rapid wave of new units. Meanwhile, with overall completions down sharply y/y, the supply that does hit the market is unlikely to create the kind of relief for rents or prices that many had hoped for in 2025–2026.

What This Means for Markets

• Homebuilders (SF): The “SF resilience” headline is built on non‑significant monthly noise while the SF pipeline softens. Favor builders with strong balance sheets, lower land intensity, and option‑heavy lot strategies. Expect selective pricing power where supply remains tight—particularly in the South, where backlog concentration is highest.

• Apartment REITs and MF developers: Elevated permits with lagging starts and rising backlog imply slower unit delivery in 2026. That supports occupancy and rent stabilization in tight submarkets, but financing friction and construction delays continue to drag. Focus on operators with access to cheaper capital and pre‑funded pipelines.

• Building materials and distributors: The starts‑permits gap and weaker y/y completions argue for choppy demand. Materials suppliers with exposure to renovation/repair may outperform those tied to new‑build volume.

• Rates and the inflation path: A significant drop in completions points to slower shelter supply growth. That’s a headwind to rapid shelter disinflation on a lagged CPI basis, complicating the “quick relief” narrative even if policy rates ease. Duration trades should weigh this against broader macro data; don’t over‑read a single month, but note the directionally tight supply signal.

• Credit and construction lending: The widening authorized‑but‑not‑started backlog and MF conversion friction keep construction lenders in the driver’s seat. Wider spreads and tighter covenants likely persist; nimble borrowers with balance‑sheet flexibility win share.

Positioning and What to Watch

• Track the permits‑to‑starts conversion ratio and the “authorized but not started” stock—especially in 5+. Rising backlogs with flat starts = more friction.

• Watch SF permits for a clean turn; until SF authorizations break decisively higher, SF starts “resilience” is suspect.

• Monitor regional backlog in the South (total 166.0k, SF 92.6k NSA). Absorption there will determine whether supply tightness widens or narrows nationally.

• Scrutinize revisions. The release flags “revised” figures for September and notes that YTD permits reflect revisions not distributed by month—messy for trend analysis and headline spin. Use moving averages and pay attention to revision patterns, not just first prints.

The bottom line: The only statistically clear signal in October is a meaningful drop in completions vs last year. Starts are noisy, permits aren’t inspiring, and the backlog is building. For investors, that tilts the 2026 housing narrative toward continued supply tightness, selective pricing power for quality SF builders, steadier‑than‑feared MF fundamentals in constrained metros, and a slower grind lower for shelter inflation. Don’t chase the starts headline; follow the conversion data—and the backlog—until the pipeline truly clears.