4.14 Structure of consumption in PPP terms
See Table 4.14 hereCommentary
About the data
Definitions
Data sources
International comparisons of consumption
Cross-country comparisons of consumption expenditures are difficult to interpret. When the expenditures are denominated in national currencies or in a single currency using an exchange rate conversion, the comparisons do not account for the sometimes substantial differences in relative prices. As a result, they also do not reflect differences in the real relative quantities of different types of consumption embodied in each country’s expenditure patterns.
This problem has led to the use of purchasing power parities (PPPs) to convert reported values to a common unit of account. PPPs measure the relative purchasing power of different currencies over equivalent goods and services. PPP-based expenditures therefore correct for differences in relative prices and so allow meaningful comparisons of consumption across countries. PPP-based figures also reveal the underlying relationships between the structure of consumption and income, because the quantities of different types of consumption are valued at average international prices. Thus PPPs provide a consistent and meaningful approach to analyzing how the structure of consumption changes with the level of development.
Table 4.14 presents the structure of private consumption for 65 countries with a population of more than 1 million using the most recent PPPs from the International Comparison Programme (ICP). The summary tables 4.14a and 4.14b divide the countries into five income groups based on their GDP per capita in PPP terms, and give the average shares for selected components of consumption based on both PPP and local currency values.
Comparison of the averages based on local currency and PPP values reveals some interesting differences. The most significant relates to the share of food in total consumption, used as a basic indicator in many poverty studies. While the share of food varies inversely with income in both summary tables, the size of the shares is quite different. For example, the average share of food for the lowest-income group is 48 percent of consumption in nominal terms, but only 29 percent in PPP terms. But the difference tends to disappear as income rises, because the relative price of food (measured by the ratio of the PPP for food to the PPP for GDP) is higher for lower-income economies. Because this tendency holds both for countries and for income groups within a country, one conclusion is that the incidence of poverty is likely to be higher if measured in PPP terms than if measured in nominal terms.
Another interesting difference relates to the relative shares of services in consumption. For lower-income countries services generally account for a higher share of consumption in PPP terms than in nominal terms. The conventional view is that the share of income spent on services increases with per capita income. This is true only in nominal terms, however. In real terms the share of services remains more or less constant regardless of income level. But the relative prices of services are generally lower in poorer countries. So the implication is that people in higher-income countries do not buy proportionately more services—they just pay more for the services they buy.
Although PPPs are more useful than official exchange rates in comparing consumption patterns, caution should be used in interpreting PPP results. PPP estimates are based on price comparisons of comparable items, but not all items can be matched perfectly in quality across countries and over time. Services are particularly difficult to compare, in part because of differences in productivity. Many services are not sold on the open market in all countries—for example, government services—so they are compared using input prices (mostly wages). Because this approach ignores productivity differences, it may inflate estimates of real quantities in lower-income countries.
The data here are based on the most recent estimates from the International Comparison Programme (ICP). The ICP database is compiled in two steps. First, regional comparisons are carried out, and aggregate PPPs are computed for developing countries by region and for the OECD countries. Second, the results are linked across regions to establish global consistency. The figures for the OECD countries and for the countries of the former Soviet Union and Eastern Europe are from the 1993 round of the ICP. The rest are from the 1985 round.
Consumption refers to private (that is, household) and nonprofit (nongovernmental) consumption as defined in the U.N. System of National Accounts. Estimates of private consumption of education and health services include government as well as private outlays. The ICP concept of enhanced consumption, or total consumption of the population, focuses on who consumes goods and services rather than on who pays for them. That is, it emphasizes consumption rather than expenditure. This approach improves international comparability because aggregate measures based on consumption are less sensitive to differences in national practices in financing health and education services.
Because countries tend to concentrate on production numbers, however, estimating the detailed structure of consumption is one of the weaker aspects of national accounting in low- and middle-income economies. The composition of consumption is estimated through household expenditure surveys and related survey information and therefore shares any bias inherent in the original sample frames. For example, in some countries surveys are limited to urban areas or even more narrowly to capital cities and so are not representative of national expenditure patterns. Urban surveys show lower than average shares for food and higher than average shares for gross rent, fuel and power, transport and communication, and other consumption. Controlled food prices and incomplete national accounting for subsistence activities may also contribute to low shares for food. Adjustments based on other available indicators may have to be made to improve the cross-country comparability of consumption patterns. See Ahmad (1994) for an extensive discussion of the ICP and its methods.
Definitions
• Private consumption includes the consumption expenditures of individuals, households, and non-profit, nongovernmental organizations. It also includes government expenditures for education and health services.
• Household consumption shows the percentage shares of selected components of consumption computed from details of GDP converted using purchasing power parities.
• Bread and cereals comprise the main staple products—rice, flour, bread, all other cereals, and cereal preparations.
• Gross rent consists of both actual rent and imputed rent (the hypothetical cost of renting the same property in the open market) and repair and maintenance charges.
• Fuel and power exclude energy used for transport (rarely reported to be more than 1 percent of total consumption in low- and middle-income economies).
• Health care and education may include government as well as private expenditure.
• Transport and communication cover all personal costs of transport, telephones, and the like.
• Other consumption covers beverages and tobacco, nondurable household goods, household services, recreational services, and services (including meals) supplied by hotels and restaurants; the purchase of carryout food is also recorded here. This group also covers consumer durables, comprising household appliances, furniture, floor coverings, recreational equipment, and watches and jewelry.
The source of purchasing power parity data is the International Comparison Programme (ICP), coordinated by the United Nations Statistical Division. The World Bank collects detailed ICP benchmark data from regional sources, establishes global consistency across the regional data sets, and computes regression-based estimates for nonbenchmark countries. For detailed information on the regional sources and compilation of benchmark data see the World Bank’s Purchasing Power of Currencies: Comparing National Incomes Using ICP Data (1993b). For information on how regression-based PPP estimates are derived see Ahmad (1992).