Retail Sales and Consumer Spending Data
Retail sales is a fast, noisy monthly read on consumer spending, and how markets trade it depends heavily on which sub-line — headline, ex-autos, or the "control group" — actually feeds GDP.
Prerequisites: GDP and the National Accounts
Consumer spending is about two-thirds of US GDP, which makes any timely read on it market-moving. Retail sales, released monthly by the Census Bureau roughly two weeks after month-end, is the fastest broad signal available — well ahead of the quarterly GDP consumption figure it feeds into. That speed comes at a cost: it's a survey of a sample of retailers, not a full census, and the first print is routinely revised in the following two releases as more complete data comes in.
Retail sales covers goods spending — stores, restaurants, gas stations, online retailers — but not services, which is now the larger and faster-growing share of what US consumers actually spend on (healthcare, travel, rent, streaming subscriptions). That's the single most important thing to remember about the number: a strong or weak retail sales print says something about goods consumption specifically, not the full consumer.
The sub-lines that matter
Markets rarely trade the raw headline number. Three sub-lines get watched separately because each strips out a different source of noise:
| Line | What it removes | Why |
|---|---|---|
| Headline retail sales | nothing | includes autos and gas, both volatile and price-driven |
| Ex-autos | vehicle sales | auto sales swing with financing conditions and supply, not just demand |
| Control group | autos, gas, building materials, food services | the cleanest proxy for the consumption line that actually flows into GDP |
The "control group" is the one GDP modelers watch most closely, because it excludes the categories most prone to one-off swings (a spike in gas prices mechanically inflates headline retail sales in dollar terms without any change in real spending) and most closely tracks the goods-consumption component of the national accounts.
Worked example
Suppose headline retail sales rises 0.9% month-over-month, driven mostly by a 4% jump in gasoline station sales after a spike in pump prices. Ex-autos still shows a solid 0.6% gain. But the control group — stripping out gas along with autos and building materials — comes in flat at 0.0%. A trader reading only the headline would conclude consumer spending is accelerating; reading the control group instead shows the "strong" print was almost entirely a price effect at the pump, not more real goods being bought.
What this means in practice
Because retail sales is reported in nominal dollars, not inflation-adjusted, a hot print during a period of rising goods prices can overstate real spending growth — the same trap as the gasoline example above, generalized to any category with volatile prices. Economists building nowcasts for GDP typically deflate the control group by a relevant price index before feeding it into a consumption estimate, rather than taking the nominal print at face value.
Retail sales is fast but narrow — goods only, nominal dollars, and heavily revised — so the market-relevant signal usually comes from the control group, not the headline, and even that needs a price adjustment before it says anything about real consumption.
The most common mistake is reacting to the headline retail sales beat or miss without checking whether autos or gasoline drove it. Both categories can move for reasons that have nothing to do with the health of the consumer — a chip shortage limiting auto supply, or an oil spike changing prices, not volumes.
Related concepts
Practice in interviews
Further reading
- US Census Bureau, Advance Monthly Retail Trade Survey methodology