Solid headline, uneven details | Oct 15, 2026 / Census / Retail Sales (Sep)

📺 Watch the Full Video Analysis

📊 The U.S. Census Bureau’s advance estimate for September retail and food services sales was $745.0 billion, up 0.8% m/m and 6.3% y/y.
📈 August was revised up to +1.3% m/m, and July-September sales rose 5.5% from a year earlier.
⚠️ Excluding autos and gasoline, sales rose only 0.2%, which is not statistically significant, while gasoline stations jumped 6.4% m/m. All figures are nominal.
💡 We unpack industry strengths and weaknesses, survey limits, and the level shift from the September 28 annual revision.
🗓 Next release (October data): November 17.

The Ultimate Summary: 総額は堅調、中身には濃淡

What this release is

The Advance Monthly Retail Trade Survey is the U.S. Census Bureau’s first estimate of monthly sales at retail stores, restaurants and bars. It samples roughly 4,800 firms and weights the results to represent over three million businesses. Figures are seasonally adjusted but not adjusted for inflation, so they mix price and volume effects.

September at a glance

• Total sales: $745.0 billion, +0.8% m/m (margin of error +/-0.4 pts), +6.3% y/y
• August revised to +1.3% m/m (from +1.1%)
• July-September: +5.5% from a year earlier
• Excluding autos and gasoline: +0.2%, statistically indistinguishable from zero

The Census Bureau states: ‘If the interval contains 0, then one does not have sufficient evidence to conclude at the 90 percent confidence level that the change is different from zero.’

How to read it

The headline is a statistically significant gain, and the three-month year-over-year rate is solid. The narrower measure excluding autos and gasoline is much softer for the month, so the strength is not evenly spread. Retail sales covers goods and food service; it is one input among many on consumer activity, and this release contains no information about Fed policy, the dollar or equities. The next report (October data) arrives on November 17, 2026.

月次の推移と前年比: 2か月連続の増加

Why the revision history matters

Advance estimates rest on a small sample and are revised in later releases, and once a year the whole series is re-benchmarked. On September 28, 2026 the Bureau published an annual revision using results from the 2023 and 2024 Annual Integrated Economic Survey.

The roughly $34.6 billion gap in the August level most likely reflects the re-benchmarking rather than a change in spending, though the release does not itemize the cause.

Month-to-month path in the latest release

July -0.4%, August +1.3%, September +0.8%. The direction of the July and August moves matches the earlier release (-0.5% and +1.2%), so the revision changed magnitude, not sign.

What we can and cannot compare

Because the two releases sit on different benchmarks, treating +6.0% versus +5.5% as a slowdown would be unsafe. Within the October release, the like-for-like comparison is 5.6% (August) versus 6.3% (September) y/y, both nominal. As a general matter, nominal series can rise with prices alone; this report contains no price data to test that.

ガソリンが押し上げ、除くと小幅増

Gasoline stations: the outlier line

Gasoline stations posted +6.4% m/m and +28.1% y/y in September, following +4.0% m/m in August. Sales were $68.02 billion versus $63.95 billion, and the line is about 9% of total sales. Of the roughly $5.7 billion total increase, about $4.1 billion came from this line (simple arithmetic on Table 1).

Two ways to read it

• Constructive: even without gasoline, sales are up 4.5% y/y, and ex autos and gasoline up 5.0% y/y, so the year-over-year picture is not narrowly based.
• Cautious: the ex autos and gasoline m/m series ran -0.1%, +1.1%, +0.2% for July-September, and the July and September readings are not statistically different from zero. Two consecutive significant moves in one direction are absent, so the trend call stays open.

Scope note

As a general matter, fuel station revenue moves with pump prices as well as gallons sold, and this nominal release cannot split the two. Nothing in the report speaks to Fed policy, FX or equity markets; those questions need other data.

業種別の強弱と統計の限界

Industry detail

The Census Bureau notes that comparisons across industries have not been tested for significance.

Reading the mix

The y/y column shows breadth in restaurants, online and catalog sellers and electronics, with grocery-type stores and pharmacies lagging. Miscellaneous retailers show a month-to-month swing (-3.0%) alongside a +10.9% annual rate, a reminder that one month rarely defines a category. Some detailed lines are not published in the advance report (marked with an asterisk) or fail publication standards (marked S).

Reliability

The advance estimate uses a link-relative method and does not impute most nonrespondents. Over the past 12 months, revisions to the total month-to-month change averaged 0.0 points, with a median absolute revision of 0.1 points; August itself moved from +1.1% to +1.3% in this release.

What to watch

The October report is due November 17, 2026. The series to follow is growth excluding autos and gasoline, where September’s +0.2% was not statistically significant.