FRBSF Economic Letter
1997-03 | January 31, 1997
In recent months, there has been much public discussion of the continuing increase in income inequality in the U.S. (Weinberg 1996). Some commentators suggest that the rich are getting richer, the poor are getting poorer, and those in the middle are getting nowhere. Others argue that data on household incomes in a single year are a misleading indicator of inequality and that there has been no worsening of long-run opportunities (Cox and Alm 1995). This Economic Letter tries to sort out some of these conflicting claims.
The data underlying the statistics of the distribution of income come from an annual survey of households conducted by the Census Bureau. The degree of income inequality often is measured by the Gini index. This index summarizes the extent to which the income distribution departs from perfect equality. It is constructed by looking at the proportions of the population that receive various proportions of total income. If all families were to receive equal incomes — so that, for example, 20 percent of the population receives 20 percent of the income — the value of the Gini index would be 0. Conversely, if one family were to receive all the income — complete inequality — the Gini index would be 1. Because it is a summary statistic, the Gini index cannot capture all aspects of the income distribution that might be of interest. For example, a disparity between poor and middle-income families might yield the same Gini index as one between middle-income and rich families, but might raise different concerns.
In terms of the Gini index, the distribution of income among American families tended to become more equal in the two decades after World War II. Between 1947 and 1968, the Gini index declined about 7 percent. However, inequality has been rising since 1968, and the Gini index surpassed its 1947 level in the early 1980s (see Figure 1). Since the 1960s, a decreasing proportion of persons has lived in “families” (two or more persons living together and related by blood or marriage), so the focus has shifted to “households” (one or more persons sharing a housing unit). From 1968 to 1989, the Gini index for households increased 11 percent with most of this rise occurring in the 1980s. Inequality has continued to increase in the current decade, but it does not appear to have accelerated.
Two features of these data might cause the numbers to give an incomplete impression of the income distribution and of how it has changed over time. First, the numbers refer only to money income and exclude noncash benefits provided by employers or governments. Second, the data provide a series of snapshots of the income distribution rather than following the fortunes of the same group of households over time.
In recent decades; a rising proportion of employee compensation has come in the form of benefits, such as health insurance and retirement contributions, rather than cash. Between 1968 and 1995, this proportion rose from about 5 percent to almost 12 percent. Many low-income households do not receive these benefits, either because they are not employed or because their employers do not provide them. Hence, the upward trend in these benefits probably has raised the compensation of non-poor households more than that of poor households.
There are no comprehensive data showing the distribution of employer-provided benefits among households. The data that are available suggest that the effect of benefits on the distribution of income goes roughly in the direction suggested above, but has been relatively small. The Census Bureau (Census, April 1996, Table 1) tabulates the distribution of income and calculates the Gini index both including and excluding health insurance supplements to wages and salaries, which represent about two-thirds of employer-provided benefits. Including these benefits increases the share of total income received by middle-income households, has almost no effect on the share of low-income households, and reduces that of upper-income households. Thus, it slightly widens the disparity between middle- and low-income households and narrows that between the middle and the rich. Its net effect on overall inequality measured by the Gini index is very small. There are no data on the distribution of other employee benefits, but it seems likely that these income supplements also disproportionately benefit middle-income households.
The income-distribution data include cash benefits received by households from governments and these transfers reduce overall inequality significantly, since many go to lower income households. In 1994, these benefits reduced the Gini measure of inequality by about 18 percent and increased the share of total income going to the bottom quintile of households by 3.8 percentage points. However, judging by their impact on Gini ratios, changes in these programs appear not to have had much effect on the trend of inequality in the last fifteen years (Census, April 1996, Table 1). As with employer-provided benefits, the inclusion of noncash government benefits — such as Medicare and food stamps — has only a small effect on overall inequality. These benefits-in-kind primarily benefit the poor, but increase the income-share of the bottom quintile of households by less than 1 percentage point.
Commentators who argue that income inequality is not as serious a problem in the U.S. as the Gini measures might suggest often stress that the distribution of income in a single year will be less important if households have opportunities to move up or down the income distribution. If today’s poor have a good chance of being tomorrow’s rich (and conversely), today’s inequality may be less cause for concern.
Cox and Alm, for example, argued recently (1995) that the degree of household mobility is so great that the data for a single year “tell us virtually nothing — particularly about opportunity.” Critics of this optimistic assessment argue either that the degree of movement up and down the income ladder is actually quite modest (so that many households are “stuck” in their current situation) or that the amount of movement up and down the income scale has not changed much, so that changes in the degree of inequality do provide useful information.
Because households do move up and down the income scale, the income distribution in a single year is likely to be a biased indicator of inequality over a longer period. At any time, it is likely that, among low-income households, more are having an unusually bad year than an unusually good one, so their current income under-estimates their longer-run situation. Conversely, current income tends to overstate the longer-run average among high-income households. Thus, focussing on a single year exaggerates the degree of true inequality. It is important to recognize, however, that this factor would mitigate the upward trend in inequality only if this source of bias has been increasing. It seems unlikely that this has occurred because fluctuations in the economy do not appear to have worsened since the 1960s.
The overall distribution of income is affected by demographic changes that cause some classes of households to become a larger share of the population and other classes to decline. Two examples of such demographic changes that may have increased inequality in recent years are the greater number of two-career households, which has increased the number of high-income households, and of single-person households, which has added to the proportion of low-income households. These changes have widened income disparities, but both may be viewed as reflecting rising living standards and the greater opportunities available to many individuals. For example, one interpretation of the increased number of two-income households is that it reflects the improved opportunities available to married women, rather than being forced on couples by worsening opportunities for married men. Similarly, the rising proportion of one-person households may be a symptom of rising incomes that have made living alone — especially for older persons — more feasible. As a final example of a demographic change that may have added to inequality, the increase in immigration in the last two decades probably has tended to widen the gap between rich and poor, because new immigrants generally have lower incomes than native-born Americans and often rise more slowly up the income scale (Samuelson 1996).
In an attempt to get at the degree of income mobility in the U.S., various studies have attempted to track the incomes of a sample of households over a series of years. Cox and Alm cite a study of 14,000 tax returns from 1979 to 1988. Among this sample of taxpayers, 86 percent of those who were in the lowest quintile in 1979 had moved up one or more quintiles nine years later and 15 percent had risen to a real income that would have put them in the top quintile in 1979. Conversely, 35 percent of those who began the period in the top quintile had moved down by one or more quintiles by 1988. Cox and Alm argue that these data imply that only a minority of households remain stuck in one part of the income distribution for long periods.
However, these numbers may exaggerate the degree of true mobility since many of the income changes reflect the normal effects of aging: young persons experience rising incomes as they advance into their prime earning years, while older households move into retirement and lower incomes. And as the whole economy grows, most households move up in absolute (even if not in relative) terms, so it is not surprising that Cox and Alm find that most Americans enjoy rising real incomes as time passes.
In a critique of Cox and Alm, Gottschalk (1996) finds (using a different data set) that of those individuals who were in the lowest quintile of incomes in 1974, more than two-fifths remained in this same relative position in 1991. Similarly, more than half of those who were in the top quintile in 1974 were still there in 1991. These results imply less relative income mobility than Cox and Alm claim and thus more reason to be concerned about rising inequality.
Gottschalk also argues that there is no convincing evidence that mobility has improved over time, so that although many persons at the bottom will move upward over time, their chances of doing so have not improved as their relative position has worsened. While it may be the case that many of those who are now at the bottom of the income distribution will not be there in later years, it remains true that these households are relatively worse off than their predecessors were in the past and there is little evidence that their chances of moving up have improved.
Nonetheless, it is well to remember that significant numbers of persons do improve their relative position as time passes and even more improve their absolute position. Gottschalk’s data show that of those who were in the lowest quintile in 1974, only 30 percent had not moved to higher income levels by 1991.
The gap between the incomes of the rich and the poor has been widening for three decades, though this trend has not accelerated in the 1990s. Income inequality clearly is mitigated by the fact that many persons move up (and down) the income scale over their lifetimes, but there is no clear evidence that mobility has improved as inequality has increased. Nonetheless, many persons do improve their relative position as time passes and a substantial majority experience rising absolute incomes as the economy grows. Measured inequality also may have been increased by social and demographic changes, some of which may be due to these rising absolute living standards.
Cox, W. Michael, and Richard Alm. 1995. Annual Report. Federal Reserve Bank of Dallas.
Gottschalk, Peter. 1996. “Notes on ‘By Our Own Bootstraps: Economic Opportunity and the Dynamics of Income Distribution’.” Boston College. Mimeo.
Samuelson, Robert J. 1996. “Immigration and Poverty.” Newsweek (July 13).
U.S. Bureau of the Census. 1996. “Income, Poverty and Valuation of Noncash Benefits: 1994.” Current Population Reports, P60-189 (April).
Weinberg, Daniel H. 1996. “A Brief Look at Postwar U.S. Income Inequality.” Current Population Reports, P60-191 (June).
Opinions expressed in FRBSF Economic Letter do not necessarily reflect the views of the management of the Federal Reserve Bank of San Francisco or of the Board of Governors of the Federal Reserve System. This publication is edited by Sam Zuckerman and Anita Todd. Permission to reprint must be obtained in writing.
Please send editorial comments and requests for reprint permission to
Attn: Research publications, MS 1140
Federal Reserve Bank of San Francisco
P.O. Box 7702
San Francisco, CA 94120