/ 14 May 1999

40m citizens can’t all be

wrong

The consumer price index plays an important role in the determination of our annual increases. Tony Twine explains what it’s all about

All around the world, John Citizen takes on a slightly glazed expression whenever the term consumer price index (CPI) and its associated inflation rate are reported or brought up at annual increase time.

The main reason for this is that individual John Citizens have difficulty in associating their cost of living increases with the John Citizen represented by the official CPI data. If the given data does not represent our own circumstances, it must surely be wrong. But is it?

The most common mistake made by the uninitiated is to confuse inflation, which is the rate of change of a given price level, with the price level itself. Thus, inflation reducing from, say, 9% to 5% does not mean that prices have fallen – they are merely rising more slowly.

A second reason for common public incredulity comes from the breadth of items covered within the CPI (271 items in 17 groups, in South Africa). Few individuals or households spend equal amounts of money on all these items every month, or even every year, distancing the common experience from the overall movement of prices.

This is demonstrated by the fact that so- called headline inflation is actually media shorthand, borrowed from economists, for “the rate of change of the CPI for all items, all income groups and all areas” as published by Statistics South Africa. Most print or broadcast media do not have the space or time for the title, let alone any meaningful disaggregation of the data, presented in some detail by Statistics South Africa.

In South Africa, a base year for the CPI is researched and constructed once every five years. Countries with sustained lower inflation rates may rebase at longer intervals. The basing exercise consists of a huge survey of more than 15 000 households in 15 metropolitan areas around the country, which sets out to establish, within the bounds of reporting accuracy, how households disposed of their income on consumer goods and services over the previous year. Reported spending is then analysed for the base year, and a table of weights for each individual item, adding up to 100% of consumer spending, is created.

Those weights are frozen for as long as the particular base year is in use. On a monthly basis, usually on the seventh day of each month, the prices of the 271 individual items are measured in the 15 metro areas, except for prices which do not move very often, like telephone and postage rates. Weighting individual items allows price indicators for 17 groups (for example, food, transport, education, clothing) to be calculated, with the 17 groups contributing by weight and price level to the all items index.

Clearly, different income groups will allocate their resources differently over the items and groups of goods and services within the CPI. Because price changes for individual items are not uniform, the overall weighted price changes for each income group will differ from time to time.

Statistics South Africa publishes not only headline CPI and inflation figures on a monthly basis, but also five income groups (reflected by expenditure, rather than claimed income).

Every individual will have their particular set of weights for their own consumption basket. Those weights will change from month to month. Unfortunately, fiscal, monetary and corporate policy cannot be set against a backdrop of price movements for more than 40- million individuals, and measures of central tendency have to be used to form an information base for informed decision- making. The CPI principles described here are used around the world, although they are sometimes known as “retail price” or “cost of living” indices.

Therefore, John Citizen should be more suspicious if the rate of headline inflation matches his own perceptions of his individual inflation status, rather than if there is a difference.

As individuals’ personal circumstances change, they may well move between specific weighting groups, further confusing their own inflation pictures with that of price movements as a whole. CPI-measured inflation is neither flawed nor inaccurate, simply because 40-million John Citizens are not all Mr Weighted Average.

Tony Twine is a director at independent consultancy Econometri