Consumer price indices

In principle, the consumer price index measures the price of goods consumed by a typical consumer in a given period of time, and changes in it measure consumer price inflation. Such indices are required for the comparison of standards of living at different times or in different places, and hence for the measurement of real economic growth. Changes in the index have since the 1980s also increasingly been a specific target of central banks. Frequently, they are also used to index-link benefits, allowances and the like (sometimes including wages), often with the objective of "depoliticizing" decisions about them. Stapleford (2009), however, analysed a variety of ways in which this objective is incom- pletely met. One significant source of debate in this regard arises from difficulties about measuring consumer prices, with the suggestion that the index tends to overstate infla- tion, promoting the idea that index-linking should be adjusted accordingly.

One difficulty arises simply from the problem of collecting information on actual prices. The index should be a weighted average of prices paid by consumers, with weights determined by expenditure shares. Expenditure surveys, however, are inevitably inac- curate. As there may be variations in the price of the same good in different areas, for different consumers, or even for different times of purchase, the problem is enormous.

A further difficulty arises from the problems of creating an accurate index when the weights change from period to period. Although such problems were well appreciated by Fisher (1922), most price indexes are calculated by the Laspeyres method: weights are determined by expenditures in an initial period, and the index calculated for a later period. To the extent that differential price increases of different goods lead consumers to substitute away from those goods that increase most in price, the typical consumer will be better off with unchanged "real income" as measured by the consumer price index. This difficulty can be mitigated by more frequent rebasing, or by using more sophisticated measures of the kind surveyed by Diewert (1981), but at greater cost.

A further problem that raises both practical and conceptual issues concerns improve- ments in the quality of goods and the introduction of new goods. Wholly new goods obviously have no comparator in earlier periods. However, following Hicks (1940), one might assume that an earlier "price" of these goods was just high enough to make demand for them zero (when in fact the goods did not exist). If so, then the price at which these goods are actually sold must be lower than this earlier "price", so that the introduc- tion of the good amounts to a fall in the price index.

Changes in quality invite the response that a "quality-adjusted" price for each good might be estimated. It is usually supposed that the norm is for the quality to improve, and this was influentially emphasized by Stigler et al. (1961) and Boskin et al. (1998). Quality deterioration, however, is also possible, and Clague (1962) considered it seriously. In either case, measurement difficulties are apparent, and as services feature more and more prominently in consumer expenditures, the measurement of their quality becomes an important issue. The possibility of there being not only a trend, but also a cyclical aspect, with customer service perhaps deteriorating in boom periods, also becomes of increasing importance.

An important conceptual difficulty, however, arises from the problem of differentiating a "genuine" quality improvement from a change of taste (possibly induced by advertis- ing). If expenditure shifts from a cheaper to a more expensive version of a good, there is no secure way to determine whether the second is "better" or simply fashionable. To treat being fashionable as a characteristic of quality leads to the question of whether the price index is intended to measure the cost of buying a specified basket of goods, or of achieving a specified level of utility. Opting for the latter would suggest, as Deaton (1998) observed, that increasing life expectancy, the spread of asceticism, or the spontaneous development of greater powers of enjoyment would then have a claim to being constitu- tive of a fall in prices. Similar and related points were made by Gilbert (1961). Another point, though, would be that if the price index is not meant to explain the income required to achieve a certain level of welfare, the idea that the Laspeyres index is flawed because consumers can protect their standard of living by changing their consumption pattern is also called into question. Consumers may be able to make advantageous sub- stitutions, but that does not mean that this index overstates the increase in the price of goods they were consuming.

See also:

Core inflation; Inflation; Inflation measurement; Inflation targeting.