What is a consumer price index?

INFLATION HAS DOGGED societies for centuries. Attempts to measure it properly began in earnest in the 20th century. In 1914 the British government calculated a “working-class cost of living index” to help guide adjustments to the wages of essential workers during the first world war. America’s Bureau of Labour Statistics (BLS) published inflation measures for 32 cities in 1919, and followed up with a national index in 1921. How do such consumer price indices work—and how have they changed over the years?

Early approaches involved establishing the composition of a typical shopping basket, and then surveying shops to see how the prices of the items moved from month to month, or year to year. The change in each item’s price was multiplied by its weight in the basket to create an inflation average for all items. The original aim was to work out how much the cost of living had changed over a given period.

The composition of the statistical shopping basket has changed substantially over time. In Britain the retail-prices index, created in 1947, originally included the cost of a mangle, a mechanical device used to wring water out of clothes. Delicacies like condensed milk and corned beef were also included. Statisticians must account not just for changing tastes but for new technologies (video rentals and MP3 players have come and gone from the basket) and for increased spending on services (hotel stays, video-streaming and so on) rather than goods.

Other factors must be accounted for, too. One is substitution: if one item increases significantly in price, consumers may switch to another (chicken instead of beef, for instance). America’s Federal Reserve uses a personal-consumption-expenditure index (PCE) rather than the consumer price index because, in part, the formula the PCE uses better reflects this substitution effect.

Another factor is improvements in quality: a modern laptop is much more powerful than the earliest devices from the 1980s. Comparing the two is like setting a Ferrari against a bicycle. In 1996 the Boskin commission, appointed by America’s Senate to examine possible biases in the CPI, suggested that a failure to account for such improvements—known as hedonic adjustment—meant the CPI had overstated inflation by 1.1 percentage points per year. The BLS subsequently updated its calculation methods.

These days central banks tend to have an explicit or implicit inflation target (2% for the European Central Bank, for example). This means that inflation indices no longer just measure the cost of living but also guide economic decision-making. This raises new calculation problems. For example, housing is one of the biggest costs for households. If mortgage rates rise, the cost of living goes up for people with variable-rate payments. For that reason, mortgage-interest payments were originally included in Britain’s retail-prices index (RPI). The result was that, when the Bank of England pushed up interest rates to fight inflation, the RPI went up in the short term too. To avoid this self-defeating mechanism, Britain created an inflation measure that excluded mortgage payments. Nowadays, central banks tend to use the market level of rents as their measure of housing costs.

Central banks often focus on the underlying, or “core,” rate of inflation: a signal of where overall inflation is likely to drift. Core measures usually exclude food and energy prices, which are highly volatile and subject to extraneous factors such as extreme weather and geopolitical turmoil. Another approach is to use a “trimmed mean”, in which those items that have risen (or fallen) most are excluded. These days, the consumer price index is just one of a range of inflation measures. ■