(August 2005) Between 1981 and 2001, economic inequality increased substantially in the United States. This increase is apparent regardless of which data or formulas are used to measure it—and regardless of whether the resource measured is income or consumption.

Using income to gauge levels and trends of well-being and inequality, however, has its limitations. First, people get a more tangible benefit from consuming goods and services than from the mere receipt of income. Second, income is often underreported. And finally, changes in income are often transitory. For these reasons, some analysts have argued that tracking consumption levels might be a more appropriate way to measure economic well-being as well as permanent changes in economic inequality. 1

A new study that I coauthored uses the Bureau of Labor Statistics’ Consumer Expenditure Survey to see if the size or trend in inequality among U.S. children, adults, and the elderly between 1981 and 2001 would be altered if consumption measures were used rather than income measures.2 Indeed, the consumption trends of the elderly (especially the single elderly) indicate this group is better off than has been assumed under income measures.

But the study also finds that the consumption levels of children declined over the 20-year period relative to the general population, particularly those of children who are living with single mothers or who are part of less traditional families. And households with children were the only group whose distribution of consumption was relatively more unequal than their distribution of disposable income during the period studied.

A Zero-Sum Game Among Three Groups

The study measures the relative consumption distribution of children, adults, and the elderly as a means of measuring trends in inequality. In other words, for the 20-year period, we measured at six different points the percentage of adults, children, and elderly represented in each quintile of consumption for the total population.

If income and consumption were perfectly distributed, 20 percent of each group would be represented in each quintile. But measuring relative distribution implies a zero-sum game within the total U.S. population: When one group improves its share of consumption under this scheme, it does so “at the expense” of another group in the population. As we shall see, tracking relative distribution shows that the elderly actually improved their share of consumption between 1981 and 2001, while some groups of adults and children fell back.


Adults (defined here as people ages 18-64) have historically been relatively better off than the general population in terms of disposable income and consumption. However, adults had lost some of their relative advantage over the elderly and children in terms of consumption between 1981 and 2001, even though they still enjoyed a relative income advantage.

For instance, the percentage of adults who were in the bottom two consumption quintiles increased from 34 percent in 1981 to 36 percent in 2001. And the percentage of adults without children and who were in the top quintile of the total population regarding consumption dropped from 36 percent to 28 percent over the period.


Because many elderly are retired, they were overrepresented in the lowest disposable income quintiles and underrepresented in the top income quintiles during the period studied. (In 2001, for instance, 26 percent of U.S. elderly were in the bottom income quintile, while 31 percent were in the next-to-bottom quintile.)

But when we switch to consumption, the elderly in 2001 are underrepresented in the lower two quintiles (at 36 percent) and modestly overrepresented in the upper quintiles (at 42 percent). And since 1981, the elderly’s relative distribution of consumption has improved much more than their relative distribution of income.

The increase in elderly home ownership (along with the general increase in value of home ownership in the United States) is the largest contributor to this relative improvement. (Housing consumption is estimated to be the monthly rental equivalence of one’s home.) In 2001, 82 percent of U.S. elderly people lived in their own home, up from 76 percent in 1981, and more than 80 percent owned their homes without mortgages.

Married elder couples—who are more likely to be the “younger” elderly—did even better in terms of consumption over the two decades than did the elderly as a whole. And while analyses of income poverty suggest that single elderly women are among the poorest people in society, even this group improved in terms of consumption distribution since 1981. Much of this added consumption takes the form of medical spending and housing, expenditures that are likely to improve overall well-being.


The improvements in the relative advantages of adults in their income distribution and the relative advantage the elderly have in consumption have been at the expense of the third group, children. The relative distribution of children’s consumption as compared with the overall population was unequal in 1981 and grew more unequal over the next two decades.

About 24 percent of U.S. children were in the bottom income quintile in both 1981 and 2001. But with respect to consumption, children’s overrepresentation in the bottom quintile grew (especially between 1981 and 1994) from 26 percent to 28 percent, while the share of children in the top quintile also grew, from 12 percent to 14 percent. Unlike any other group in the population, the relative deterioration of children’s consumption distribution is larger than the shift in their income distribution over the 20-year period.

Disaggregating the relative distribution of consumption by type of family with children makes it clear why children’s consumption has deteriorated (see figure). The increasing number of children in single-mother households and nonmarried-couple families accounts for the increase in consumption inequality in children from 1981 to 2001. (The situation for children living in married two-parent families has remained relative unchanged over the two decades.) No matter how consumption is defined—as consumption expenditures or as including service flows—the relative consumption distribution of households with children is worse than any other group in the country.

Distribution of Children (Under Age 18) by Family Type Relative to the General Population, Using Equivalent Consumption

Living in married-couple families Living in single-mother families
1981 17.82 25.09 23.00 21.19 12.90 1984 58.24 18.32 13.12 6.14 4.48
1986 20.31 23.35 23.31 18.48 14.55 1986 52.95 18.23 11.52 8.86 8.43
1990 20.08 22.72 22.20 19.56 15.45 1990 65.60 16.49 8.91 4.63 4.37
1994 18.88 23.07 23.76 19.50 14.79 1994 61.76 21.04 5.54 7.68 3.98
1999 20.12 23.12 22.62 19.44 14.70 1999 60.74 19.09 7.85 7.33 5.99
2001 19.08 22.07 21.09 21.00 16.76 2001 54.22 18.18 14.11 6.90 6.59

Living in other non-married families
1981 52.54 13.66 14.74 6.97 12.08
1986 49.70 25.35 12.89 6.18 5.87
1990 47.73 26.05 12.55 9.39 4.27
1994 52.60 23.10 10.28 9.25 4.77
1999 45.24 25.33 17.19 8.33 3.91
2001 48.62 25.52 12.65 7.75 5.47

Source: David S. Johnson, Timothy Smeeding, and Barbara Boyle Torrey, “Economic Inequality Through the Prisms of Income and Consumption,”  Monthly Labor Review (2005).

The Consequences for Policy of Using Either Prism

Children are generally living in young families with both lower income and lower consumption than they themselves may have later in life. Therefore, children may not necessarily be relatively disadvantaged for their entire lifetime. In addition, housing creates a large adjustment in consumption, and children are more often found in families that are in less expensive homes—which lowers their consumption in general. The trends over the 1981-2001 period, however, suggest that successive cohorts of children are moving down the relative consumption distribution of the general population.

Income and consumption are clearly different prisms with which to view well-being. In many cases, it does make a difference which prism is used. The prism of income makes adults look relatively more advantaged than the general population. The prism of consumption makes the elderly appear more advantaged than the general population.

Therefore, the selection of a measure of well-being may have real consequences for how government policies are evaluated, especially for the elderly. But both prisms reveal that the well-being of households with children are at an increasing disadvantage relative to the general population.

Barbara Boyle Torrey is a visiting scholar at PRB.


  1. Using consumption (defined as a person’s expenditures plus the amortized value of the major durables that person purchases) as a measure of well-being also has limitations, because some of the most important factors of well-being are not consumed. But consumption overall is a better measure of what people consider their normal or “permanent” income, which is closer to the concept of well-being than actual (and possibly fluctuating) income.
  2. David S. Johnson, Timothy M. Smeeding, and Barbara Boyle Torrey, “Economic Inequality Through the Prisms of Income and Consumption,” Monthly Labor Review 128, no. 4 (2005): 11-24.