Your grocery inflation is not the national figure

When food inflation is reported as a single percentage, it describes a basket assembled by a statistics agency to represent average consumption. It is an honest number and a useful one. It is also describing somebody who does not exist.

Why the national figure misses your household

An official index weights each category by its share of average spending. Your own weights are wherever your money actually goes, and they are rarely close to the average.

A household that eats a lot of beef, butter, cheese and coffee has lived through a very different few years from one built on rice, lentils, seasonal vegetables and own-brand staples. During a period when dairy and coffee moved sharply and dry goods did not, those two households could differ by several percentage points while sharing a country, a city and a supermarket.

Substitution widens the gap further. The index measures the price of a fixed basket. You do not buy a fixed basket. You switch to what is on offer, drop things that got expensive, and buy more of what is in season, and all of that is invisible to a national figure and central to your own bill.

What a personal index actually measures

The question worth answering is narrow: for the things I buy regularly, what has happened to the price?

That is deliberately not the same as “has my grocery spending gone up”. Spending moves for reasons that have nothing to do with prices. A visitor, a party, a month with more meals at home, a new baby. An index has to hold the basket still so that only prices are allowed to move, otherwise it is measuring your life rather than the shops.

Building one from receipts

The method is the same one the statisticians use, applied to a basket of one household.

• Choose the basket. Products you have bought repeatedly across the whole period. Something bought once carries no information about change.
• Use unit prices, not line totals. Otherwise buying two of something looks like a price rise.
• Weight by spending. A product that is a tenth of your grocery bill should move the index ten times as much as one that is a hundredth. Unweighted, a 40% rise in vanilla essence counts as much as a 40% rise in milk.
• Fix a baseline period and express everything relative to it.

What to be careful about

A personal index is small, and small samples are noisy. Three things distort it more than anything else.

Promotions. If your baseline period happened to include a sale on something heavy in the basket, every later period looks inflationary. Compare regular prices to regular prices where the receipt recorded both.

Size changes. A pack that shrank is a price rise the line total will not show. This is exactly why the index must be built on unit prices.

Store switching. Moving your shopping from one chain to a cheaper one lowers your bill without any price falling anywhere. That is a real saving, but it is not deflation, and mixing the two makes both meaningless.

What it is good for

Not for arguing with the national statistics, which are measuring something else and measuring it competently. For two more practical things.

The first is knowing where your own increase came from. A single percentage is not actionable; the same figure decomposed into the six products responsible for most of it is.

The second is calibration. Most people substantially misremember what things used to cost, usually in the direction of thinking everything has risen more than it has, because the items that rose are the memorable ones. A personal index built on receipts is the cure for that, and it occasionally delivers the more uncomfortable finding: that prices moved less than you thought, and the bill went up because the basket did.