Beyond “Rich” and “Poor”
A meticulous analysis reveals the full depth and breadth of U.S. economic inequality.
Published June 1, 2003 | June 2003 issue
Simple labels and quick dichotomies are usually the easiest way to describe any given state of affairs, but genuine understanding suffers if reality is subjected to substantial simplification. When that reality is as complex as the economic status of U.S. households, generalizations are bound to lead to distortion and confusion. Media reports and even academic work too often reflect such misrepresentation.
In a study of economic inequality published in the Summer 2002 Quarterly Review, four economists go well beyond such superficial treatment, and their analysis reveals a subtle and multilayered interplay of income, earnings, wealth and demographic status among U.S. citizens in the 1990s. They show a nation of wide financial diversity and mobility, substantial growth in various measures of economic well-being and yet a surprising lack of change in the overall picture of economic inequality during the decade
The analysis, "Updated Facts on the U.S. Distributions of Earnings, Income, and Wealth", by Santiago Budría Rodríguez, Javier Díaz-Giménez, Vincenzo Quadrini and José-Víctor Ríos-Rull, is, as its title implies, an update of a 1997 report, (also published in the Quarterly Review). Both analyses rely on the Survey of Consumer Finances (SCF), a national survey of about 4,000 households sponsored by the Federal Reserve to obtain a comprehensive financial profile of U.S. citizens.
Tools of the tradeThere are many ways to measure economic status. The authors use three:
Earningswages and salaries, plus a large portion of business and farm income; earnings, therefore, are the portion of a household's income that comes from its labor.
Incomerevenue from all sources before taxes but after transfers (such as welfare and Social Security); defined thus, income includes earnings but exceeds it by revenue from capital (rent or dividends, for example) as well as government payments.
Wealththe net worth of households (assets minus debts); wealth, therefore, is the stock held by a household of all its unspent income (in other words, its accumulated savings).
There are also many ways of quantifying inequality; the study looks primarily at percentage distributions (often by grouping into quintiles) and Gini indices. A quintile is a fifth, of course, and the study categorizes various economic status measures into five groups: the top and bottom fifths and those in between, with the first quintile being the lowest or having the least.
A Gini index summarizes inequality in a single number, with 1 as complete inequalityone household holding all and the others having nothingand 0 as total equalityall households owning exactly the same amount. The Gini index has the obvious advantage of simplicity, like an average number. But just like an average number, it can also obscure details that a percentage distribution more fully describes.
The economists use these tools to describe the United States as of the 1998 SCF and the changes that have taken place since the 1992 SCF.
The data show clearly that U.S. earnings, income and wealth were all very unequally distributed in 1998. Wealth was by far the most concentrated measure of economic status, with a Gini index of .803, and income was the least unequally distributed, with a Gini of .553. In more intuitive terms, the wealthiest 1 percent of households held 1,335 times the wealth of the bottom 40 percent, while the top 1 percent of households in terms of income received 73 times the income of the bottom 40 percent. Earnings distribution fell in between the other two measures, with a Gini index of .611 and a top 1/bottom 40 ratio of 158.
The same three measures viewed by quintiles: The fifth wealth quintile had an average net worth of $1.2 million; the first quintile's average wealth was actually negative, debts exceeding assets by $4,100. For income, a fifth-quintile household made $159,100 a year, on average, while a first-quintile family brought in $6,400. As for earnings, a top 20 percent household earned $127,500 annually, while a bottom fifth household actually lost $300.
Still one more way to look at economic concentration: Households in the top 1 percent of the 1998 wealth distribution owned 34.7 percent of total wealth. The top 1 percent of income recipients took in 17.5 percent of total income. And the top 1 percent of earners received 15.3 percent of total earnings reported by the survey sample. Comparison of these last two figures points out a curious quirk: While earnings are more unequally distributed than income for the sample as a whole, this is reversed for the richest in both respects, where inequality is greater for income than for earnings. (See chart.)
The Lorenz Curves for the U.S. Distributions
of Earnings, Income and Wealth
What % of All Households Have
What % of All Earnings, Income, or Wealth
% of Households (Ranked by Amount)
Source: 1998 Survey of Consumer Finances
Parsing rich and poor
This wide variation in economic inequality according to what is being measured implies that the terms "rich" and "poor" are themselves ambiguous. Income-poor is by no means the same as wealth-poor. Indeed, the economists point out that while the rich tend to be rich by all three measures, such is not the case for the poor: "The earnings-poor," they write, "are surprisingly wealthy. ... The income-poor own significant amounts of wealth."
For example, almost 23 percent of the households in the SCF sample have zero earningsno wages or salaryand 0.24 percent have negative earnings. Yet the average wealth of the bottom 1 percent of the earnings distribution is about three times the average wealth of the total sample. Two explanations: Many earnings-poor households are retirees who have built up estates, and many households with negative earnings are headed by business owners under financial stress.
At the same time, many of the wealth-poor do fairly well in terms of earnings and income. About 2.5 percent of households have zero wealth and 7.4 percent have negative wealth (a fact that helps to explain the high concentration of the wealth distribution). But the average earnings of the bottom 1 percent in the wealth distribution put them in the fourth quintile (the top 60 percent to 80 percent) of the earnings distribution. The explanation here is demographic: Over 60 percent of the lowest wealth quintile are single and a disproportionate number of them are young. So the young and/or single bring in good money but haven't built up much wealth.
Demographics of inequality
As this last example demonstrates, exploring income, earnings and wealth inequality without looking at age and marital status, or other variables like employment and education, is like driving across country without a map. Demographics help dramatically in understanding the lay of the land.
Analysis by age is a case in point. Inequality in earnings and income tends to be higher among older households than among young households, but wealth inequality decreases steadily from a very extreme level at younger ages until age 40, at which point it remains quite steady, at a Gini index between .700 and .800.
Looking at employment status, the study confirms what one might expect: Workers, who make up almost 60 percent of the sample, are significantly wealth-poorer than the sample average. Retirees, almost 20 percent of the sample, tend to be earnings- and income-poor, but wealth-rich. The self-employed, 11 percent of the total, are very well-off by all three economic measures, with income over 2 times the average and wealth about 3.3 times the average. And nonworkers are poor by all measures. Marital status also has the impact one expects: Married people are better off in terms of income, earnings and wealth than households headed by single people.
In terms of education, those with more of it tend to be better off than those with less, as anticipated. But somewhat surprisingly, inequality as measured by the Gini index tends to be the same among those with no high school, those with a high school education and those with a college degree. In other words, all of the education groupings show very similar levels of inequality: Income inequality among the college-educated, for example, is as high as it is among those without high school diplomas. (This is less true for earnings distribution among no-high school households.)
The American dream
The ability to pull oneself up by one's bootstraps is American legend, but households move in the other direction as well. Aging, the success or failure of business ventures, good luck and bad health all play a part in economic mobility, which, the authors note, "makes inequality an essentially dynamic phenomenon."
To measure mobility, the economists rely on the Panel Study of Income Dynamics, funded by the National Science Foundation, since the PSID (unlike the SCF) follows the same set of households over time. By comparing the earnings, income and wealth status of the same households in the early and mid-1990s, they build a picture of changes in mobility and inequality.
They find, for example, that households in the bottom earnings quintile are by far the least mobile: 90 percent of the families in that quintile in 1989 remained there in 1994. But only 34 percent of households in the second-to-bottom earnings quintile in 1989 were still there five years later: Some moved up and others down.
Income mobility is far greater than earnings mobility, according to the data, and wealth mobility is somewhat lower than that of income. (Age and retirement clearly play a big role in inhibiting earnings mobility. Mobility is considerably higher among households whose heads were between 35 and 45 years old in 1989 than among all households.)
Still, even for incomethe economic measure with greatest mobilitythe chance of moving from the poorest to richest quintile in just five years was low: Just 2 percent of households made that bootstrap leap.
Changes during the 1990s
A comparison of Gini indices between the 1992 and 1998 SCF data shows that there were just small changes in overall inequality during the decade. The Gini index of income inequality, for example, decreased from .574 to .553a minor decline. Other measures of inequality confirm the stability of that overall picture.
But a number of details did change. The correlation between earnings, income and wealth, for example, changed significantly from one period to the next. Earnings and income were not so tightly linked in the last period as in the first. By contrast, the correlation between income and wealthmarginal in 1992was substantial in 1998. This change, suggest the economists, may be connected to the emergence of the "new economy," since the correlation between wealth and business income was much higher at the end of the decade than at the beginning.
The relative economic conditions of the earnings- and wealth-poor showed little change over the decade. Their shares of total wealth and earnings remained largely the same. Among the income-poor, however, relative conditions declined: The share of households with zero or negative income doubled during the decade and their share of total SCF sample wealth in 1998 was half what it was in 1992.
The rich fared better. The earnings-rich, income-rich and wealth-rich households all became relatively wealth-richer. The share of total wealth owned by the top earnings quintile, for example, was 49 percent in 1992 and 55 percent later in the decade. The top 1 percent of earners increased their wealth share from 15.7 percent to 18.3 percent.
These changes were even larger for the income-rich. And the top 1 percent of wealth holders increased their share of the wealth pie from 31.4 percent to 34.7 percent. In other respects the wealth-rich remained little changed: At the end of the decade as at the beginning, the wealthiest obtained most of their income from businesses and capital; they were still, on average, married and over 45 years old.
The demographics of economic inequality changed to some degree during the 1990s. For instance, although the share of the total sample represented by workers increased by 4.6 percentage points, their relative income, earnings and wealth all decreased somewhat. The relative income of retirees declined, as well; in 1992, the income of the average retiree was 78 percent of the total sample, but by 1998, it had dropped to 64 percent.
Changes were also seen in status of different education groups. One surprising example: The relative average earnings of households headed by college-educated individuals was 5.8 times greater than that of no-high school households in 1992, but that edge decreased to 4.7 times larger by 1998. But wealth moved in the opposite direction during that time span; college households became wealth-richer relative to no-high school households.
The economic conditions of singles with dependents improved significantly during the decade, relative both to singles without dependents and to married households. In 1992, the average earnings of singles with dependents were 88 percent of singles without dependents; by 1998, singles with dependents had turned the tables, with earnings 106 percent that of singles without. But single females remained poor compared to their male counterparts throughout the decade.
As these changes imply, there were also some differences in economic mobility during the decade. Earnings mobility decreased slightly, but income and wealth mobility both increased. The most striking mobility difference, according to the economists, was a significant jump in mobility of households in the bottom wealth quintiles.
The overwhelming impression left by the economists' analysis is that of complexity. The data provide a robust demonstration of the fact that inequality cannot be summed up in a simple word or phrase: Different measures of economic welfare, various gauges of inequality, diverse sources of economic well-being and dissimilar types of people all interact over time in ways that defy easy description. And a true understanding of this dynamic phenomenon demands both careful analysis and close attention. As the authors conclude, "Inequality is a complex and multidimensional subject."