Retail Sales +1.2% vs +0.8% Est, Core +1.4% vs +0.5% — Strength Survives the Fuel Strip
Fundamentals · 2026-09-16
Retail and food services sales rose 1.2% m/m in August to $773.9 billion, against a 0.8% consensus, and core sales excluding autos rose 1.4% against 0.5% expected — nearly triple the forecast. July was revised up to -0.5% from -0.6%. Sales are +6.0% y/y, core +6.9%. 12 of 13 categories rose; only building materials fell, at -0.2%. Nonstore retailers +2.6% and gasoline stations +3.1% together supplied 57% of the entire dollar gain from 26% of the base. Strip fuel and it is still +1.1%; strip fuel and autos, +1.2%. The control group that feeds GDP rose 1.36%. Against CPI, real spending was +0.8% m/m and +2.5% y/y. Advance estimate from roughly 4,800 firms, nominal, not adjusted for prices. Next release 15 October; benchmark revisions 28 September.
What It Changes
- It lands five and a half hours before the FOMC decides, and it removes the last soft spot in the case for hiking. A hike to 3.75%-4.00% was already 92% to 93% priced. A consumer spending 1.2% more in a month does not argue for patience.
- The core beat is the number, not the headline. Ex-autos came in at 1.4% against 0.5% expected — a miss by the street of nearly a full point on the measure that is supposed to be the less volatile one.
- A fifth of the gain was people paying more for fuel, and it still does not explain the print. Strip gasoline entirely and sales rose 1.1%; strip gasoline and autos and they rose 1.2%. The strength survives the strip, which is the opposite of what a pure price illusion looks like.
- It is real, not just nominal. Against August CPI of +0.4% m/m and +3.4% y/y, real spending rose roughly 0.8% on the month and 2.5% on the year. That is a genuinely expanding consumer, not one running to stand still.
- It hardens the tension the whole week has been building. Households told Michigan they feel the worst in seventy-four years bar one month, and then went out and spent 1.2% more. Sentiment and spending have now disagreed for long enough that one of them is not measuring what it claims to.
Impact
- USD — Bullish — a large upside surprise on the consumer, on decision day, with the 10-year already at 2007 levels. It supports both the hike and the higher-for-longer path behind it.
- This morning's Open had DXY at 99.66, up 0.04%, extending its climb on Fed repricing, with the 10-year closing Tuesday at 4.996% after touching 5.04% intraday. This print gives that yield level a growth justification it did not have yesterday.
- The dot plot still outranks it. The Open framed today correctly: the hike is priced, so the end-2026 median and the vote count are what move the dollar at 14:00. Retail sales raises the odds of the hawkish version of both.
- US Indices (ES / NQ / YM / RTY) — Mixed, lean bearish — genuinely good news for earnings, arriving on the one day when good news most reliably pushes yields up into a decision.
- Futures came into the print +0.2% to +0.4% across Dow, S&P and Nasdaq-100, a relief bounce after the S&P's fifth decline in six sessions. A hot consumer print is a thin foundation for that relief to stand on with a 5% 10-year.
- The one clean equity positive is discretionary breadth. 12 of 13 categories rose, food services rose 1.2% and nonstore rose 2.6% — that is discretionary spending accelerating, not defensive rotation into groceries, where sales rose just 0.5%.
- Gold (GC) — Bearish — gold is trading as a rates instrument, and this pushes rates the wrong way for it.
- The Open flagged gold's overnight reversal to $4,388 (+1.28%) off Tuesday's six-week low as positioning rather than a change in mechanism. A 1.2% retail print is the kind of data that tests that reading before 14:00 rather than after.
Inside The Number
Advance retail and food services sales came in at $773.9 billion in August, up 1.2% (±0.4%) from July and up 6.0% (±0.5%) from August 2025. Consensus was 0.8%. The margin of sampling error excludes zero, so the increase is statistically significant on the Census Bureau's own test.
The core number is where the surprise actually sits. Sales excluding motor vehicles and parts rose 1.4% against a 0.5% consensus — the measure came in at nearly three times its forecast. On the year, core is +6.9%, running faster than the headline's 6.0%, which tells you autos are the drag rather than the driver: motor vehicle and parts dealers rose just 0.6% on the month and 2.1% on the year.
July was revised up, not down. The June-to-July change moved from -0.6% to -0.5%, and the confidence interval tightened from ±0.4% to ±0.2%. So the sequence is June +1.2%, July -0.5%, August +1.2% — a soft month between two strong ones, with the soft month now marginally less soft. The three-month annualized run rate on the total is +2.8%.
Breadth was close to complete. Twelve of the thirteen published categories rose. The single exception was building materials and garden equipment at -0.2%, and inside general merchandise, department stores fell 0.8% while the warehouse clubs and supercenters that dominate that line rose. There is no sector here that looks like a consumer pulling back.
Now the part the headline hides. Two categories did most of the work. Nonstore retailers rose 2.6%, adding $3.57 billion, which is 37.6% of the entire $9.49 billion increase. Gasoline stations rose 3.1%, adding $1.85 billion, or another 19.5%. Together those two contributed 57.1% of the month's gain while accounting for 25.9% of the sales base. Everything else — autos, groceries, restaurants, clothing, general merchandise, health, furniture, electronics, sporting goods — split the remaining 43%.
The gasoline line is not a spending story, it is a price story, and the arithmetic is unusually clean. August CPI put the gasoline index up 3.9% on the month. Gas station dollar sales rose 3.1%. Prices rose faster than receipts, which implies the volume of fuel sold fell roughly 0.8%. Americans spent 3.1% more at the pump to take home slightly less gasoline. That is a transfer out of discretionary spending dressed up as retail strength, and it is the single most misleading line in the release. Treat the estimate as approximate — gas stations also sell food and merchandise, which the CPI gasoline index does not cover — but the direction is not in doubt.
And yet the print does not collapse when you take fuel out. Sales excluding gasoline stations rose 1.1%. Sales excluding both gasoline and autos rose 1.2%. The control group — total sales less autos, gasoline, building materials and food services, the aggregate that feeds the consumption line of GDP and which Census does not publish directly — rose 1.36% on the month and 5.58% on the year by our calculation from Table 1. Fuel flattered this report. It did not manufacture it.
In real terms it still stands up. August CPI rose 0.4% m/m and 3.4% y/y. Against that, headline retail sales grew about 0.8% in real terms on the month and 2.5% on the year, with core running about 3.4% real. The consumer is not merely keeping pace with inflation; they are outrunning it.
Against the running narrative, this is the sharpest contradiction the feed has carried in weeks. Michigan sentiment printed 47.8 on Friday, the second-lowest reading since 1952, with expectations down 11.1%. Yesterday's Empire survey showed New York factory orders stalling at 2.0 and shipments turning negative. Households say conditions are dire and manufacturers say demand has stopped, while the actual till receipts accelerated to their fastest month since March. One of those three is measuring something the other two are not, and the till receipts are the only one of them that records a transaction rather than an opinion.
The Internals
The headline aggregates, all seasonally adjusted, from Table 2 of the release:
Measure · Aug m/m · Consensus · Jul m/m revised · y/y
Retail and food services, total · +1.2% · +0.8% · -0.5% · +6.0%
Excluding motor vehicle and parts (core) · +1.4% · +0.5% · -0.2% · +6.9%
Excluding gasoline stations · +1.1% · n/a · -0.6% · +4.9%
Excluding motor vehicle, parts and gasoline · +1.2% · n/a · -0.3% · +5.6%
Retail only, excluding food services · +1.2% · n/a · -0.7% · +6.0%
Control group, our calculation from Table 1 · +1.36% · n/a · n/a · +5.58%
Every category, ranked by how much it contributed to the month:
Category · Aug $m · Jul $m · m/m · Share of the gain · pp of the 1.2%
Nonstore retailers · 141,339 · 137,772 · +2.6% · 37.6% · +0.47
Gasoline stations · 62,303 · 60,455 · +3.1% · 19.5% · +0.24
Food services and drinking places · 105,070 · 103,824 · +1.2% · 13.1% · +0.16
Motor vehicle and parts dealers · 142,379 · 141,565 · +0.6% · 8.6% · +0.11
General merchandise stores · 80,388 · 79,854 · +0.7% · 5.6% · +0.07
Health and personal care stores · 40,721 · 40,355 · +0.9% · 3.9% · +0.05
Food and beverage stores · 85,580 · 85,214 · +0.4% · 3.9% · +0.05
Miscellaneous store retailers · 16,821 · 16,515 · +1.9% · 3.2% · +0.04
Clothing and accessories stores · 28,353 · 28,163 · +0.7% · 2.0% · +0.03
Electronics and appliance stores · 8,307 · 8,180 · +1.6% · 1.3% · +0.02
Sporting goods, hobby and book stores · 9,061 · 8,951 · +1.2% · 1.2% · +0.01
Furniture and home furnishings stores · 11,398 · 11,296 · +0.9% · 1.1% · +0.01
Building material and garden equipment · 42,227 · 42,318 · -0.2% · -1.0% · -0.01
Year-over-year, which separates the price story from the volume story:
Category · y/y · Read
Gasoline stations · +21.0% · Almost entirely price; CPI gasoline is up sharply on the year
Miscellaneous store retailers · +14.0% · Small base, noisiest line in the report
Sporting goods, hobby and book stores · +10.7% · Genuine discretionary strength
Nonstore retailers · +9.9% · The structural share gainer, still compounding
Electronics and appliance stores · +7.8% · Strong, and consistent with a replacement cycle
Food services and drinking places · +5.8% · Discretionary, and the cleanest read on consumer willingness
Building material and garden equipment · +5.1% · Positive on the year despite falling on the month
General merchandise stores · +4.5% · In line with the broad retail trend
Clothing and accessories stores · +4.3% · Steady
Motor vehicle and parts dealers · +2.1% · The laggard, and why core beats headline
Health and personal care stores · +1.9% · Barely above inflation
Furniture and home furnishings stores · +1.9% · Still housing-constrained
Food and beverage stores · +0.5% · Below inflation, so grocery volumes are falling
Nominal against real, using August CPI of +0.4% m/m and +3.4% y/y:
Measure · Nominal · Real · Read
Headline m/m · +1.2% · +0.8% · Real spending clearly expanded
Headline y/y · +6.0% · +2.5% · A real gain, not an inflation artifact
Core y/y, excluding autos · +6.9% · +3.4% · Stronger still once autos are removed
Food and beverage stores y/y · +0.5% · -2.8% · Grocery volumes are shrinking
Gasoline stations m/m · +3.1% · -0.8% · More money for less fuel
How much to trust an advance estimate, from Table 3 of the release:
Measure · Median CV · Standard error on the m/m change · Median absolute revision
Retail and food services, total · 1.1% · 0.2pp · 0.1pp
Excluding motor vehicle and parts · 1.3% · 0.2pp · 0.1pp
Nonstore retailers · 2.8% · 0.3pp · 0.4pp
Miscellaneous store retailers · 4.9% · 2.2pp · 0.9pp
Food services and drinking places · 3.2% · 0.9pp · 0.1pp
Building material and garden equipment · 3.7% · 0.8pp · 0.3pp
Where The Spending Actually Went
The concentration is the finding, so it is worth stating plainly: $5.42 billion of the $9.49 billion increase came from two categories.
Nonstore retailers did the heavy lifting. Up 2.6% on the month and 9.9% on the year, adding more than a third of the entire gain. This is the e-commerce line, and it has been taking share from physical retail for two decades, so a strong month here is partly structural rather than cyclical. It also carries the second-largest revision risk in the report — a median absolute revision of 0.4pp against 0.1pp for the total — so of all the numbers in this release, this is the one most likely to look different in three weeks.
Gasoline stations were the second-largest contributor and contributed nothing real. Up 3.1% in dollars while pump prices rose 3.9%, implying volumes fell about 0.8%. On the year the line is up 21.0% — a number that will be quoted as consumer strength and is very nearly pure price.
Restaurants are the honest discretionary signal, and they accelerated. Food services and drinking places rose 1.2% on the month and 5.8% on the year, contributing 13.1% of the gain. Eating out is the first thing a stressed household cuts. It is not being cut.
The grocery line is the one that supports the gloom. Food and beverage stores rose 0.4% on the month and just 0.5% on the year — well below 3.4% inflation, which means real grocery volumes are down roughly 2.8% year over year. Households are buying less food while spending more on restaurants and online goods. That is not a picture of broad hardship; it is a picture of substitution, and possibly of a consumer base splitting in two.
The only decline was building materials, at -0.2%, which fits a housing market that has been frozen by rates all year, and which is still up 5.1% on the year.
Autos are the quiet drag. Motor vehicle and parts dealers rose 0.6% and are up only 2.1% on the year — below inflation, so real auto spending is falling. That single line is the entire reason the core measure beats the headline, and it is why the 1.4% core figure is the more representative one.
On the survey itself: the advance estimate draws on roughly 4,800 firms subsampled from the larger Monthly Retail Trade Survey, and Census imputes almost nothing for non-respondents, using a link-relative estimator instead. Over the last twelve months the average revision from advance to final on the total has been 0.0pp, with a median absolute revision of 0.1pp. The headline is usually right to a tenth. The category detail is not — miscellaneous store retailers carries a median absolute revision of 0.9pp and a coefficient of variation of 4.9%.
Against This Morning's Open
- This release is missing from The Open's calendar, and that is now two days running — The Open lists NAHB at 10:00, the FOMC at 14:00 and the press conference at 14:30, with nothing at 08:30.
- Yesterday it was the ADP weekly print; today it is retail sales, which is a top-tier, market-moving release that beat consensus by 0.4pp on the headline and 0.9pp on the core. That is a calendar gap worth fixing rather than noting.
- On everything else The Open read the day correctly. It argued the hike is close to fully priced so the dot plot and the vote count are the fulcrum, not the decision itself. This print does not change that; it raises the odds of the hawkish version of both.
- It also called the pre-decision setup accurately — the 10-year at 4.996% after touching 5.04%, futures up 0.2% to 0.4% in relief after five declines in six sessions, gold reversing to $4,388 on positioning. A 1.2% consumer print is the sort of data that tests a relief bounce rather than supports it.
What This Sets Up
- Next advance report — Thursday 15 October, covering September, which is the same morning the next Empire State survey lands.
- The bigger date is Monday 28 September at 10:00 ET, when Census publishes revised estimates incorporating historical corrections and the 2023 and 2024 Annual Integrated Economic Survey. Benchmark revisions can move the level and the recent trend at once, and this one has been flagged in the release itself.
- Whether nonstore retailers holds. It supplied 37.6% of this month's gain and carries a 0.4pp median absolute revision — four times the total's. If August gets marked down, this is where it happens.
- Whether the fuel distortion widens or unwinds. Pump prices are still climbing with crude above $100 and diesel at a record. Another month like this one and the gasoline line will flatter the nominal headline again while doing nothing for real spending.
- Whether grocery volumes keep falling. Food and beverage stores at +0.5% y/y against 3.4% inflation is the one line in this report that agrees with consumer sentiment. If it keeps deteriorating while restaurants accelerate, the split-consumer reading gets much harder to dismiss.
What Is This?
- What it is: The Advance Monthly Retail Trade and Food Services Survey, known as MARTS, is the Census Bureau's first estimate of what Americans spent at retailers and restaurants in the month just ended, published around the middle of the following month. It is built from a subsample of roughly 4,800 employer firms drawn from the larger Monthly Retail Trade Survey and weighted to represent a universe of more than three million firms. Because imputation is not performed for most non-respondents, Census uses a link relative estimator — the ratio of current-to-previous-month sales among firms that reported in both months, applied to the prior month's fuller estimate. Figures are adjusted for seasonal variation and for holiday and trading-day differences, but not for price changes.
- Why it matters: Consumer spending is roughly 70% of the US economy, and this is the earliest hard read on it — not a survey of how people feel, but a count of what they actually paid. Traders watch three cuts of it: the headline, the core measure excluding autos, which strips out the most volatile line, and the control group excluding autos, gasoline, building materials and food services, which is the aggregate that feeds the consumption component of GDP. Census publishes the first two and not the third, so the control group has to be calculated from Table 1.
- How to read it: Four cautions. It is nominal, so in a month when fuel prices jump, dollar sales can rise while the volume of goods sold falls — always check it against CPI before calling it real growth. It is an advance estimate, accurate to about a tenth on the total but much looser by category, with revision risk concentrated in the smaller lines. It measures goods and restaurants, not services, so it misses most of the spending that has been driving this cycle. And the margin of sampling error matters: Census publishes a 90% confidence interval with every figure, and when that interval contains zero, the release says explicitly that there is not enough evidence to conclude the change differs from zero.
Want the full explainer? Economic News Events, Explained breaks down this release and every other one we cover, in plain English.
_For informational purposes only. Not investment advice._