An illustrative example: a Saturday receipt from a single supermarket, still warm from the bag, set next to a morning headline that says prices rose by a small national percent. The receipt and the headline can both be accurate. They are not the same average. The official all-items inflation rate in U.S. news is usually the 12-month change in the CPI-U, a broad urban basket. A grocery bill is one household, one store or two, one week's food, often one city's prices, and sometimes a change in what was cooked. The gap between them is information about scope. It is not, by itself, evidence that the statistic was built to deceive.
The headline basket is mostly not groceries
BLS classifies consumer spending into more than 200 categories and eight major groups. Food and beverages is one group. Housing, transportation, medical care, and the others are in the all-items index with it. Shelter alone, measured as rent and owners' equivalent rent rather than as grocery prices, is large enough that a year of rent increases can move the headline while a food category does something else. If your eye is on the receipt, you are watching a slice. If the headline is all-items CPI-U, it is watching the whole urban budget, including the shelter concept explained in house prices, rent, and the same number trap.
Food itself is split in the CPI. Food at home and food away from home are different indexes because a bag of rice and a restaurant meal are different products, sold in different kinds of outlets, with different labor and rent inside the price. A household that stopped eating out will see its own spending shift even when both indexes rise. The all-items number will not narrate that choice.
BLS publishes the component indexes. The way to ask "what happened to groceries in the official data" is to read the food-at-home index, or a narrower series such as cereals, meats, or dairy, for a named month and a named geography. The way to ask "what happened to the urban consumption basket" is to read all-items CPI-U. Substituting one answer for the other is the mismatch, and it does not require a motive.
Your cart is not the expenditure survey
CPI weights come from the Consumer Expenditure Surveys, with a lag the BLS FAQ illustrates by noting that CPI data in 2023 reflected expenditures collected in 2021. The average urban household's food share is not your food share. A family that eats most meals at home, or follows a diet the average survey household does not, can see food prices dominate its week while food remains one group inside the index. BLS says the published CPI "does not necessarily measure your own experience with price change," and gives the medical-care example: a heavier-than-average medical budget, in a period when medical prices lead, produces a personal rate above the CPI. The same sentence works for food. A national average summarizes millions of baskets. It seldom reproduces one kitchen.
Geography stacks on top of weights. The headline U.S. city average is not the index for your metropolitan area. Some city indexes are published every month and some every other month. A drought, a port, or a regional retailer can move your receipt before it moves the national food index, or the reverse. Comparing a neighborhood store with a national 12-month percent, on one Saturday, confuses a price level at one outlet with a rate of change across a sample.
Substitution, quality, and the hamburger story
A persistent claim is that when food prices rise, the CPI pretends you traded steak for hamburger and calls that "no inflation." BLS answered this in the misconceptions note. Since January 1999 the agency has used a geometric mean for most basic indexes, and that formula does reflect shoppers shifting toward items that have become relatively cheaper. The shift it allows is inside a basic category, such as among types of ground beef in one city. Hamburger and steak are in different item categories, so the CPI-U does not assume that swap. Inside steaks, the same formula would assume shoppers move toward the cut that became relatively cheaper, which can be the more expensive cut if the cheaper cut's price rose more. The formula is not a nutrition plan and it is not a judgment that households "should" trade down.
BLS has also published its estimate of the size of that basic-index choice: the geometric mean has lowered the annual rate of change of the CPI by less than 0.3 percentage point. That is a statement about a formula's measured effect, from the agency that runs it. It is not a claim that your receipt was edited.
Quality and variety changes are a separate issue. A container that shrinks, a store brand that replaces a national brand, and an organic item that replaces a conventional one are not automatically "the same price." CPI procedures try to price a comparable item in the same outlet and to adjust when quality changes. Your receipt simply totals what you chose this week. If you switched brands to hold the total down, your spending can fall while the index, following a comparable item, rises. If you switched to a higher grade, your spending can rise while the index does not follow you. Both outcomes are compatible with an honest index and an honest receipt.
Average prices are a third object
Alongside the index, BLS publishes average retail prices for a list of specific foods, utility fuels, and automotive fuels. The average price fact sheet draws the line: the CPI measures price change, while average prices estimate the price level all-items indexes cannot show by themselves. At the U.S. level BLS publishes on the order of 70 food average prices, plus a handful of utility and fuel series, and many of those foods are also published for the Northeast, Midwest, South, and West when the sample is large enough. Food average prices are published without sales tax. They are narrowly defined — a dozen large grade A eggs, not every carton in the case — so that the level means something.
BLS warns that the sample is designed to support indexes, not average prices, and that the percent change in an average price can depart from the percent change in the matching index. About a quarter of the item and outlet sample rotates each year. When a dearer outlet rotates out and a cheaper one rotates in, the average price can step down. The index does not treat that rotation as a price cut, because it does not compare different outlets as if they were the same item over time. Quality changes can open the same wedge: an organic peach replacing a conventional peach can enter an average price that allows both, while the index refuses to call the difference inflation. Seasonal foods are not seasonally adjusted in the average-price series.
So there are three grocery-related numbers in the public data, and they are allowed to disagree: your receipt, the food index, and the average price of a tightly specified item. Using the average price of one item as "inflation" inherits the rotation and quality issues BLS lists. Using it as a level — what a specified item tended to cost in a month — is what the series is for. A household log of the same specification, kept the way a personal what-was-this-worth log describes, is the right object to set beside that series.
A calmer reading of a shocking receipt
You might be thinking that a gap this visible has to be somebody's strategy. Design choices do change the number: urban coverage, expenditure weights, rental equivalence for shelter, the geometric mean inside categories. BLS publishes those choices, and its misconceptions note argues, with its own estimates, against the idea that the post-1970s methods were selected to force the index down. You can disagree with a method. The disagreement gets sharper, not vaguer, when it names the method. "My receipt is higher than the headline" usually names a basket difference, which is not yet a method dispute.
A workable habit when the receipt stings:
- Identify the official series that matches the bag: food at home, or a component, for the closest geography BLS publishes.
- Compare percent changes over the same months, not one trip against a 12-month national all-items figure.
- If you care about the level of a specific food, look at the average-price series for that specification, and remember the rotation caveat.
- Write down any change in what you bought. A different brand, a different store, or a guest at the table is a real change in the receipt and not a revision to the CPI.
- Keep all-items CPI-U for the question it answers, which includes shelter, transport, and medical care. The reason two official all-items measures can also disagree with each other is covered in CPI, PCE, and why headlines disagree.
Parents who want a version of this conversation without turning dinner into a crisis can use the sequence in how to explain what money is worth to kids. The currency calculator on this site is for converting currencies and for rough U.S. price context. It is not the CPI food index, and a converted currency amount will not explain a domestic grocery receipt.
The glossary entry for the consumer price index is the short version of the average-basket idea. Inflation, in that same list, means a rate of change in prices, not a synonym for "my total at the register."
Not a verdict on your store
A gap between a household's food spending and the all-items CPI is expected whenever the baskets differ. This page does not say your store is fairly priced, and it does not say the CPI matches your year. Component indexes and average prices are revised in their methods from time to time; the fact sheet is the place to check a definition before you lean on a level. The material here is educational, not a budget plan and not a claim about any retailer.
When your receipt and the headline diverge, which basket did the headline actually price?