How Do You Measure the Price of Everything? A Basket and Its Problems

A statistical agency defines a basket of goods and services intended to represent what households actually buy, assigns each item a weight reflecting its share of total spending, collects prices for those items repeatedly in a wide range of outlets, and combines the price changes using the weights to produce an index. The weights come from surveys of household expenditure. The prices come from field collection in shops, from websites, and increasingly from transaction data supplied directly by retailers. The result is expressed relative to a base period set at one hundred, so a value of one hundred and twenty means prices in the basket have risen by a fifth since the base. The annual change in that index is what is reported as inflation.

Several decisions have to be made and each affects the result:

• •Which items belong in the basket, which is revised annually as spending changes
• •How to weight them, since the average household is a statistical construct
• •How to treat quality change, since this year's version of a good may be better
• •How to handle substitution when people buy less of what has risen in price
• •How to treat owner-occupied housing, which different countries do differently
• •Which outlets to sample, since discount and premium shops move differently

Comparing prices over time requires comparing the same thing, and very few things stay the same. A computer costing the same as one five years ago is far more capable, so the price of computing has fallen even though the price on the shelf has not. Statisticians handle this with adjustment techniques that estimate how much of a price difference reflects improved characteristics rather than inflation, and the estimates make a measurable difference to the headline figure. Critics on one side argue that these adjustments understate inflation by treating unwanted changes as improvements. Critics on the other argue they understate it far less than the failure to capture genuinely new goods, which enter the basket only after they are already established and cheap.

Several things people expect the figure to capture are outside it by design. It measures average change across a representative basket, so a household whose spending differs from that basket experiences something different, and the gap is systematic rather than random, since poorer households spend a larger share on food and energy and therefore face higher inflation whenever those rise faster. It measures prices rather than the cost of maintaining a standard of living, which are not the same once substitution is allowed for. It excludes asset prices, so rising house prices and share prices do not appear even though they dominate many people's sense of what is getting more expensive. And it says nothing about incomes, so a rising index describes what is happening to prices and not whether anybody can afford them.

This is not a technical dispute confined to statisticians, because the index determines large flows of money. Pensions and benefits are uprated by it in many countries, so a small persistent bias transfers substantial sums over decades. Wage negotiations reference it. Index-linked government debt pays according to it. Central banks target it, so the measured figure determines interest rates, which affect everybody. Contracts of many kinds are tied to it. Governments have changed which index is used for uprating on several occasions, choosing a variant that rises more slowly, which saves money and is presented as a technical improvement, and the resulting arguments have been conducted in the language of methodology while being about distribution.

A basket of goods is weighted by spending shares, priced repeatedly and combined into an index whose annual change is reported as inflation. Choosing the basket, weighting it, adjusting for quality and handling substitution all require judgements that move the result. Pensions, benefits, debt payments and interest rates are tied to the figure, so methodological arguments about it are arguments about money.