4 Second step: the household perspective
4.3. Price indices – temporal comparison
93. To convert nominal measures of income and consumption into real measures, price
indices are required. For comparisons of living standards, price indices of bundles of consumer goods and services are the appropriate choice. It is useful to dwell on the nature of the price index to be employed for the measurement of real income and to briefly discuss the notion of cost-of-living indices.
94. Recently, the Panel on Conceptual, Measurement, and other Statistical Issues in
Developing Cost-of-Living Indexes (Schultze and Mackie eds., 2002) presented a series of
recommendations for the development of the US Consumer Price Index. Many points raised in this report are directly relevant for the questions at hand. In particular, the report refers to the concept of cost-of-living indices and to the distinction between conditional and unconditional indices. A cost-of-living index aims at measuring the relative change in expenditure that a household would have to make in order to maintain a given standard of
living25. Every cost-of-living index has to be defined with respect to a particular domain, or
scope (i.e. the goods and services included in the index), keeping other factors fixed. Thus, changes in the standard of living are measured with conditions outside the scope of the index held constant. What exactly should be within the scope of the index, and on which factors the
index should be conditioned, depends on the analytical question at hand. The Panel, for
example, recommended that for purposes of constructing the U.S. Consumer Price Index, only private (market) goods and services should be inside the scope of the index, i.e. changes in non-market commodities (leisure), environmental conditions (climate) and other societal factors (crime) should not influence the evolution of the index.
95. For other analytical purposes, however, the scope of cost-of-living indexes could be
different and this is the case for the measures discussed in the present document. More specifically, for the purpose of comparing living standards, the domain of a cost-of-living index could be made commensurate with a broader notion of income, encompassing at a minimum, government-provided services such as health and education. A further extension (see below) would be to include non-market services that households produce themselves and, possibly, leisure. Accounting for goods and services provided for free or at subsidized prices by government is not without problems, however. The inclusion of such services introduces uncertainty about data quality as, by definition, there are no prices for non-market
products. At the same time, non-market goods and services are important elements for the
living conditions of citizens and their provision potentially affects international comparisons
of living standards26.
25. One particular feature of a cost-of-living index is that it takes into account the effects of substitution by consumers when the relative prices of goods and services change. This is mainly achieved by the choice of superlative index number formulae (Diewert 1976). Diewert (2001) and Triplett (2001) provide in-depth discussions of the consumer price index as a cost-of-living index.
96. Table 4: Price changes and real income of households and total economy 1995-20064 provides estimates of real income trends based of alternative choices for the relevant deflators. Several patterns stand out.
• First, the nominal change in disposable household income is close that for adjusted disposable income for France and the United States, but less so for Finland. . Similarly, the price indices of final consumption expenditure and actual final consumption are quite similar in France and in the United States but not so in Finland. Whether higher inflation in government-provided services than in market-provided services in Finland reflects economic reality or the statistical method employed to measure price (or unit cost) non-market services remains an open issue. But the results show that moving from one price index to another is not a trivial matter, and that getting price and volume measures for publicly-provided services right makes a difference for the assessment of households’ living standards. More detail on this issue in conjunction with government provided health and education services is provided below.
• Second, (and confirming the result already highlighted in Figure 7: Real household disposable income and GDP Percentage growth at annual rate, 1996-20067) differences in the growth rates of household real income and economy-wide real income can be significant, as shown by the Finnish case (where real household income has grown by more than one percentage point less per year over the past decade than economy-wide
26. The cost-of-living index proposed here would have the following components: (i) prices of goods and services of private final consumption, i.e. market prices directly available from national accounts; and (ii) prices of individual services provided by government (such as health and education). For this second group of items, the lack of market prices implies that unit costs should enter the cost-of-living index. Whether these unit costs should refer to the costs per unit of output or the costs per unit of input is an important question and country practice varies on this point. When income or consumption measures are further extended to include the services that households produce for themselves, the scope of the cost-of-living index should be equally extended to cover the unit costs of producing these services.
Table 4: Price changes and real income of households and total economy 1995-2006
Percentage change at annual rate
France
United
States Finland
D is p o s a b le in c o m e * (c u rre n t p ric e s ) 3 .7 % 5 .2 % 4 .4 % P ric e in d e x fin a l c o n s u m p tio n e x p e n d itu re 1 .3 % 2 .0 % 1 .6 %
R e a l d is p o s a b le in c o m e 2 .4 % 3 .1 % 2 .8 %
A d ju s te d d is p o s a b le in c o m e * (c u rre n t p ric e s ) 3 .8 % 5 .2 % 4 .5 % P ric e in d e x a c tu a l fin a l c o n s u m p tio n 1 .6 % 2 .2 % 2 .0 % R e a l a d ju s te d d is p o s a b le in c o m e * 2 .2 % 3 .1 % 2 .5 % N a tio n a l d is p o s a b le in c o m e (c u rre n t p ric e s ) 3 .6 % 5 .6 % 5 .7 % P ric e in d e x d o m e s tic d e m a n d 1 .6 % 2 .1 % 1 .9 % R e a l n a tio n a l d is p o s a b le in c o m e ** 2 .1 % 3 .5 % 3 .8 %
*F o r p riva te h o u s e h o ld s a n d n o n -p ro fit in s titu tio n s s e rvin g h o u s e h o ld s ; n e t o f d e p re c ia tio n .
real income). The gap is smaller for the United States and absent in France but likely to be present in many other countries.
97. While the concepts and problems that are involved in constructing good price indices
are well understood27, the rapid changes in relative prices, economic structures, and product
design and characteristics have meant that conventional methods towards measuring price
change may be inadequate in capturing these changes. For instance, in 1996 the US Advisory
Committee to Study the Consumer Price Index (widely known as the Boskin Commission
after its chair, Michael Boskin) concluded that the CPI overstated the cost of living by about 1.1 percentage points per year, implying that real growth was understated by the same amount. The report called attention to problems associated with adjustments for quality changes, to the appearance of new products (particularly relevant in fast changing sectors like health and IT, but also in retailing), and to problems in data collection of data (e.g. the increasing fraction of sales done over the internet, and at discount stores). Quality change is
also an important topic in the report by the Panel on Conceptual, Measurement, and other
Statistical Issues in Developing Cost-of-Living Indexes (Schultze and Mackie eds., 2002).
Diewert (1998) estimates that for a measured price change of about 2% per year, the quality bias may be as large as 0.4 percentage points in the US CPI. Deaton (1998), on the other hand, questions the practical feasibility as well as some of the conceptual foundations for the need to capture quality change and to target a cost-of-living index. The reader is referred to the extensive literature on price indices that brings out the many conceptual and empirical issues involved in capturing quality change. It is evident that there is no simple or unique methodology.
98. A point of particular relevance from a welfare perspective is the question about ‘whose’
price index is evaluated. Often, conceptual discussions about price indices are conducted as if there were a single representative consumer. Statistical agencies calculate the increase in prices by looking at the costs of an average bundle of goods. However, different people buy different bundles of goods (e.g. poor people spend more on food and less on entertainment) and they may buy their goods and services in different types of stores (which sell “similar” products at very different prices). When all prices move together, having different indices for different people may not make much a difference. But recently, with soaring oil and food prices, these differences may have become more marked and people at the bottom of the income distribution may have seen real incomes fall by much more than those at the top of the income distribution.
99. Deaton (1998) relates the income-specific price indices back to the problem of quality
measurement. He argues that quality effects are income related and that the “benefits of quality upgrading and of new goods will only be distributionally neutral if the affected goods are neither luxuries nor necessities. While it is true that advances in the technology of consumption have benefitted people in all parts of the income distribution, it is hard to believe that many of the relevant goods are not luxuries. When new goods are consumed disproportionately by the rich, whose price indexes are weighted in the CPI according to their incomes, the quality-corrected plutocratic index can provide a very poor measure of prices for the average consumer.”
27. For an overview of price index theory and price measurement see Diewert (1987), IMF et al. (2004) and ILO et al. (2004).
100. A price index for (actual) private consumption28 of major groups in society (by age, income, or place of residence such as rural vs. urban population) is necessary if we are to
appraise their economic situation. One of the recommendations of the Commission mésures
du pouvoir d’achat des ménages (2008) (Commission on the measurement of purchasing
power of households) in France was to develop consumer price indices for owners of dwellings, for renters and for households that are about to purchase dwellings (Ruiz 2009). Other relevant categories potentially exist. However, such prices indices are not available in most countries. While indices with different weights for different groups of the population are easily computed and exist in several countries (such as the United Kingdom, France, Germany), they tend to show relatively similar movements. A full development of price indices differentiated by socio-economic groups would require, however, the collection of different prices for different parts of the population, so that socio-economic aspects are taken into account in data collection design. This is likely to be difficult and costly. This development should constitute a medium-term objective – a recommendation that echoes a
similar conclusion by Schultze and Mackie (2002)29. Such work would not only foster the
quality of deflation procedures, but also make it easier for citizens to assess their personal situation through some of the income and price data released by statistical offices.