Retail sales vs consumer spending: why can US figures differ?

On this page

Start with what each measure covers
Retail sales help estimate PCE without being the whole of PCE
A fictional example: retail rises while broader spending falls
Keep nominal, real and seasonally adjusted separate
Match the period, units and release version
What the comparison can tell you
Frequently asked questions
Sources and scope

US retail sales and consumer spending are related, but they measure different things. The Census Bureau’s retail and food services report measures sales by particular kinds of businesses. The Bureau of Economic Analysis’s personal consumption expenditures, or PCE, measures a broader range of goods and services purchased by, or on behalf of, US residents.

That difference matters when a strong shopping headline appears beside a weaker consumer-spending figure. Before treating them as contradictory, match their coverage, prices, reference period and release version. This guide compares US statistics; it reports no current sales, spending or inflation reading.

Start with what each measure covers

The Census Bureau’s Advance Monthly Retail Trade Survey, known as MARTS, provides an early estimate of sales at retail and food service businesses. Its coverage includes store and nonstore retailers, as well as food services and drinking places. Online shopping is not automatically outside the retail figures. Census explains the survey’s coverage and confirms that monthly retail estimates include e-commerce sales.

Read the full headline. “Retail trade” and “retail and food services” do not have identical coverage: the second includes food services. An exclusion such as “excluding motor vehicles” changes the measure again.

PCE reaches beyond purchases from retailers. It includes services such as cellphone service and healthcare, including spending paid on a person’s behalf. An insurer paying part of a hospital bill is one example in BEA’s consumer-spending explanation. PCE is therefore not simply a total of purchases made with households’ own cash or cards.

Two US measures with different starting points. Food service appears in the broader retail headline, so the distinction is not simply “goods versus services.”

Question	Retail and food services sales	Personal consumption expenditures
Starting point	Sales at covered retail and food service businesses, classified by kind of business.	Goods and services purchased by, or on behalf of, US residents.
Examples	Purchases at car dealers, grocery stores, online retailers and restaurants.	Consumer goods, meals and a wider range of services, including healthcare and cellphone service.
Price basis	Published sales values are in current dollars, without removing price changes.	BEA publishes current-dollar spending and real, inflation-adjusted measures.
Important boundary	A retailer’s receipts need not all be household purchases.	Spending on a person’s behalf need not appear in that person’s own bank statement.

The boundary in the last row prevents a misleading shortcut. Census says that sales to businesses are not excluded simply because the reporting establishment also sells to households. Its sales definition also excludes sales taxes collected directly from customers. You cannot obtain PCE by taking the retail total and adding a few service bills. Census’s general FAQ explains those reporting boundaries.

Retail sales help estimate PCE without being the whole of PCE

The relationship is closer than two unrelated surveys. BEA uses Census retail data in estimating consumer spending. Its retail control method links the growth of a defined group of PCE goods to a corresponding retail control group. That is a method for part of PCE, rather than an identity between the headline retail total and all consumer spending. BEA describes the retail control method.

This also explains why the term “control group” needs its own definition. A news report’s selected retail categories should not silently replace the full retail headline, total PCE or real PCE. Keep the named series attached to its number.

A fictional example: retail rises while broader spending falls

Imagine two consecutive months in a small, invented economy. All amounts below are US dollars, chosen only for teaching. The table contains every purchase in this example, with no overlap between rows. All purchases are final consumption by the same resident households; there are no business purchases, third-party payments, taxes, cross-border purchases or imputed services. Prices and product quality are unchanged, and we assume no seasonal or calendar effects.

Those assumptions deliberately remove several differences between the real US statistics. The example isolates coverage; it does not reproduce Census estimation or BEA national accounting.

Fictional monthly spending ledger. The retail-style subtotal contains rows 1 and 2; the broader total contains all three rows. The subtotal must not be added to the total again.

Spending category	Month A	Month B	Dollar change
1. Goods bought from retailers	$100	$110	+$10
2. Meals bought from food service businesses	$20	$20	$0
3. Other consumer services	$180	$166	−$14
Retail-style subtotal: rows 1 + 2	$120	$130	+$10
Broader consumption total: rows 1 + 2 + 3	$300	$296	−$4

The retail-style subtotal grows by:

(130 ÷ 120 − 1) × 100 = 8.333…%, or about 8.33%.

The broader consumption total changes by:

(296 ÷ 300 − 1) × 100 = −1.333…%, or about −1.33%.

The $14 fall in other services more than offsets the $10 increase in goods. Both readings can be correct because they cover different purchases. No average of the categories’ percentage changes is needed: calculate each total first, then divide by its own starting value.

The example shows a possibility, not a claim about an actual US month. Real PCE contains additional accounting features and source adjustments. It cannot be reconstructed by copying this three-row ledger into a release spreadsheet.

Keep nominal, real and seasonally adjusted separate

A higher dollar value does not necessarily mean more goods or services were bought. For one unchanged product, 50 units at $2 each cost $100. If the same 50 units cost $2.20 each next month, sales rise to $110, or 10%, while the quantity stays at 50. These are fictional prices and purchases, not a US series.

Census explicitly states that its adjusted retail sales data account for seasonal, holiday and trading-day differences, but not price changes. Seasonally adjusted sales can therefore still rise because prices rose. Census’s time-series notes keep those adjustments distinct.

BEA separately publishes current-dollar PCE and real PCE. The PCE price index is a price measure; it is not the dollar amount of consumer spending. BEA’s release definitions and statistical conventions distinguish current-dollar values, real values and price indexes.

For actual real PCE, use the published measure and its method. BEA estimates prices and quantities at detailed levels and aggregates them using a Fisher chain-weighted formula. Dividing the entire retail-sales headline by a broad CPI index does not turn it into official real PCE. Nor should you sum published chained-dollar components as though they always add to the published total. BEA’s methods FAQ and statistical conventions explain these limits.

Match the period, units and release version

First compare the reference months. A release published this month can describe last month’s activity. A month-over-month change and a year-over-year change use different starting points, even if both appear beside the same reference month.

Then read the level’s units. BEA presents personal income and outlays levels at seasonally adjusted annual rates. A level stated at an annual rate is not the cash actually spent in one month, and it does not make every reported monthly percentage change an annualized growth rate. Do not compare that level directly with a retail-sales monthly dollar total. BEA’s statistical conventions describe the presentation.

Finally, retain the publication date and estimate version. Census’s advance release provides early estimates and includes preliminary information for the preceding month. Later estimates can differ. Its survey description explains the advance and preliminary stages. BEA estimates can also be revised: its September 2026 annual-update notice is a dated example, not a permanent revision schedule or a current spending figure used here.

Compare two original releases if your question concerns what readers knew at those publication times. Use a consistent later vintage if your question concerns the subsequently estimated spending path. Mixing those tasks can create a gap that reflects the information available rather than only the underlying purchases.

What the comparison can tell you

Retail sales offer an early view of activity at covered businesses. PCE offers a broader consumption measure. Reading them together can reveal why a shopping headline leaves out an important part of household consumption.

Neither aggregate describes every household’s budget, income or financial comfort. Neither fixes the next GDP release, interest-rate decision or asset-price direction. If you are interpreting a market move, our news-reading guide explains why the prior expectation and observation window also matter.

For price-index concepts, the US CPI guide distinguishes index levels, inflation rates and PCE inflation. For the pay side of the comparison, the real-wage guide separates hourly pay, paid hours and purchasing power. Those are related questions, rather than substitutes for measuring consumption.

Frequently asked questions

Can retail sales rise while total consumer spending falls?

Yes, the measures cover different activity. In the fictional ledger, retail-style spending rises from $120 to $130 while broader consumption falls from $300 to $296 because other services fall more. Actual US comparisons also require matching prices, adjustments, periods and release versions; the ledger is not an official PCE calculation.

Is PCE the same thing as the PCE price index?

No. PCE measures consumption expenditure. The PCE price index measures price changes for its covered consumption. Real PCE is an inflation-adjusted measure of consumption. A spending increase, a price increase and a real-consumption increase answer different questions.

Does seasonally adjusted retail sales mean inflation has been removed?

No. Census’s seasonal, holiday and trading-day adjustments do not remove price changes. Keep the adjustment label and the current-dollar price basis together. You need a suitable real measure to study consumption volume rather than treating every dollar increase as extra quantity.

Are online purchases missing from retail sales?

No. Census states that e-commerce sales are included in monthly retail sales estimates. “Retail” does not mean only purchases in physical shops. Coverage still depends on the businesses and activities in the named series; it does not expand to every service purchased online.

Can I use PCE as a record of my own household’s spending?

No. PCE is an aggregate and includes purchases made on people’s behalf, not just their own payments. Your spending mix and circumstances can differ from the aggregate. A personal budget requires your own transactions and liabilities; the national figure cannot establish what you can afford or predict your next month’s spending.

Explore HypaForecast pages

Browse the market list and BTC 5-minute and 15-minute pages. These general publisher links are separate from the comparison of US spending statistics.

View markets View 5-minute BTC View 15-minute BTC

Sources and scope

Census MARTS coverage, reporting boundaries and adjustment notes, and BEA PCE definitions and methods were checked through public web text on 20 October 2026. The monthly dollar ledger and 50-unit price example are original fictional illustrations, not actual US data, official PCE estimates or a reconstruction of national accounts. The BEA September 2026 annual-update notice is used only to illustrate revisions; no release values or future schedule are reported. No current economic forecast or market response is assessed.

HypaForecast publishes educational guides and operates a prediction market. This explanation of US statistics does not establish a trading opportunity, market availability or a reader’s eligibility to use a service.