Income Distribution in the United States

views updated


As the twentieth century drew to a close, the gap between rich and poor in the United States, and between the wealthy and impoverished nations of the world, continued to widen. This entry extends the analysis presented in the 1992 edition of theEncyclopedia (largely based on 1988 data) to include information on income, wealth, and poverty gathered from studies conducted in the midand late 1990s by various government departments, academic economists, and private research institutions.


Data on money income for households, families, and persons in the United States are collected yearly by the Bureau of the Census, based on a national probability sample of approximately 50,000 households. The official definition of "income" is money income before taxes, including some government transfers (Social Security benefits, welfare payments, worker's compensation), returns on investments, and pensions, but excluding capital gains or health insurance supplements paid by employers. In recent years, the Census Bureau has also published computations based on fourteen alternative definitions of income; for example, after deducting a range of taxes or after adding the earned income credit, the value of noncash government benefits such as food stamps, school lunches, and Medicare or Medicaid reimbursements to providers. Some of these alternative measures reduced the level of income inequality, while others increased it, but the general finding is that government transfers have a significantly stronger effect on reducing inequality than does the current tax system. For the sake of simplicity and consistency, this entry will primarily use data based on the official definition.

Household Income. In 1997, the median income for all 102 million U.S. households was $37,005, an increase of almost 2 percent over 1996, continuing an upward trend from a low of $34,700 in 1993. Adjusted for inflation, however, the 1997 median was slightly lower than the previous high of $37,303 in 1989. As Table 1 indicates, there are significant differences among subgroups in median income, and while all but male nonfamily householders and Asian/Pacific Islanders experienced an increase in real income over the previous year, data covering the entire period 1989–1997 present a mixed picture. Gains between 1989 and 1997 were highest for black households, for both married-couple family households and those headed by a female householder, and for persons aged 55–64. Losing ground over the eight years were male family householders without a wife, male nonfamily householders, Hispanic households, and households headed by young adults age 15–24 as well as by those 35–54 years old.

With respect to region and place of residence, households outside metropolitan areas and in the South and Midwest experienced income gains; metropolitan households and those in the Northeast and West suffered declines. In absolute numbers, however, the South continues to lag well behind the Northeast, while at the extremes of the distribution, median household income in Alaska is almost twice that in West Virginia.

Because a "household" can consist of one person, often a young adult at the onset of the work life or an elderly retiree, household medians are typically lower than those for families, officially defined as two or more persons related by blood, marriage, or adoption. The advantages of family households, especially those comprising a married couple, over nonfamily households is shown in Table 1. Clearly, some of the other differences seen in Table 1—by race, ethnicity, region, and residence—can be accounted for variations in the proportion of married-couple family households.

Per Capita Income. Table 1 also reports on per person income, which has risen by close to 7 percent since 1989 in constant dollars, with black gains outstripping those for whites, although white per capita income remains roughly 40 percent higher that that of blacks. In addition, that part of per capita income which is accounted for by earnings continues to show a gender gap in wages, but one that has narrowed from a 31 percent advantage for male workers in 1989 to 25.8 percent in 1997. In part, this narrowing is due to a 3 percent increase in real earnings for women over the eight years, but also an even more significant decline of more than 4 percent for men during the same period. The loss in real earnings for men reflects the general decline in employment in high-paying unionized jobs in the blue-collar sector as well as corporate downsizing at the managerial level.

Although not shown in Table 1, per capita income also varies directly with years of schooling. Since 1980, the adjusted median wage for workers with only a high school education fell by 6 percent, while the median for college graduates rose by 12

Table 1
Summary Measure of Income by Selected Characteristics: 1989, 1996, and 1997
 1997median income (in 1997 dollars)  
characteristicsmedian income (dollars)19961989percent change in real income 1996 to 1997percent change in real income 1989 to 1997
source: U.S. Bureau of the Census (1998d): vii.
all households37,00536,30637,303*1.9–0.8
type of household     
family households45,34744,07144,647*2.9*1.6
married-couple families51,68151,00249,925*1.3*3.5
female householder, no husband present23,04022,05922,315*4.43.3
male householder, no wife present36,63436,47639,1080.4*–6.3
nonfamily households21,70521,45422,2211.2*–2.3
female householder17,61316,77417,865*5.0–1.4
male householder27,59227,89229,036–1.1*–5.0
race and hispanic origin of householder     
white, not hispanic40,57739,67740,166*2.31.0
asian and pacific islander45,24944,26946,6112.2–2.9
hispanic origin126,62825,47728,192*4.5*–5.5
age of householder     
15 to 24 years22,58321,93024,0273.0*–6.0
25 to 34 years38,17436,71138,442*4.0–0.7
35 to 44 years46,35945,43948,554*2.0*–4.5
45 to 54 years51,87551,63053,7380.5*–3.5
55 to 64 years41,35640,72939,9461.5*3.5
65 years and over20,76119,89420,402*4.41.8
earnings of full-time, year-round workers     
per capita income     
all races19,24118,55217,999*3.7*6.9
white, not hispanic21,90520,991(na)*4.4(x)
asian and pacific islander18,22618,332(na)–0.6(x)
hispanic origin110,77310,27910,605*4.81.6

percent. This differential signifies the growing wage gap between unskilled and skilled workers in an economy increasingly geared to processing information rather than to manual labor. For example, according to the Bureau of Labor Statistics, in 1982 workers in the top decile earned roughly four times the wages of workers in the lowest decile; by 1996, this difference had spread to almost five times (Pierce 1998).

Income Dispersion by Households. In 1997, 11 percent of households had income below $9,000, while 9.4 percent reported income of $100,000 or more. Adjusted for inflation, these numbers indicate a decline of 9 percent in households at the lowest level and an increase of 21 percent in the proportion at the top over the decade since 1987. Also showing a slight increase were households with incomes in both the $75,000–$99,999 and $10,000–$14,999 brackets. In contrast, the proportion of households with incomes between $15,000 and $50,000 declined from 45.2 percent of the total to 44.5 percent. It would appear, then, that the "middle" continues to "disappear," but that the overall distribution of household income moved upward, with the greatest gains among those in the highest income categories (U.S. Bureau of the Census 1998d, Table B-2).

Income Dispersion by Families. A similar pattern characterizes the distribution of income by families rather than households: a modest decline in the proportion with incomes below $10,000, from 7.3 percent in 1987 to 6.8 percent in 1997, a 20 percent increase, from 9.3 to 11.8 percent, at the $100,000+ level, and declines in the proportion in the $25,000–74,999 range (U.S. Bureau of the Census 1998d, Table B-4). For both families and households at the lower end of the distribution, much of the gain in income over the decade can be attributed to the longer hours worked by both wives and husbands, as well as the tight labor market that allowed workers to bargain for higher wages (Mishel et al. 1998). Nonetheless, the major gains in family income accrued to the top 10 percent, who realized 85 percent of the increase in the value of stocks.

Income Inequality. Although the Census Bureau's press releases in the late 1990s focused almost entirely on the overall gain in median household income since 1994, especially for black and female-headed households, the data also showed a marked increase in inequality on the two measures tabulated by the Bureau.

  1. Shares of aggregate income received by households. When income groups are divided into quintiles, the share of aggregate income received by the lowest one-fifth of households fell from a high of 4.4 percent in 1977 to a low of 3.6 percent in 1997, while the share going to the highest fifth rose from 43.6 to 49.4 percent. Declines over the past two decades also characterized the three middle quintiles. In other words, the top 20 percent of households accounted for an ever-higher share of the nation's aggregate income, with that of the top 5 percent increasing from 16.1 to 21.7 percent.
  2. Index of income concentration. Another way of measuring inequality is through an index of income concentration called the Gini coefficient or ratio. A Gini score of 0 would indicate complete equality, where there are no differences among units; a score of 1 indicates complete inequality, whereby one unit has all of whatever is being measured and the other units have none. Figure 1 tracks the percentage change in Gini ratios for household income since 1967. As you can see, the period from 1967 to 1980 was one in which income differences among households rose slowly, held in check largely by increases in government transfers to the poor and a relatively progressive tax structure. Inequality rose more sharply between 1978 and 1989, stabilized briefly during the recession of 1990–1992, and then shot upward in the mid- and late 1990s. Some of this increase can be accounted for by a change in Census methodology, but most is very real, and there was no indication that the trend had abated in 1998.

As striking as are the data on household income inequality shown in Figure 1, the story for family income is even more dramatic. After declining during the 1960s, the percent change in Gini coefficients for family income has risen sharply higher than for households, reaching the 20–25 percent range by the mid-1990s (U.S. Bureau of the Census 1996, 1998a).

Other Benefits. In addition to money income and government transfers, the well-being of households and families is also affected by employee benefits such as health insurance coverage, vacation pay, sick leave, and pensions. Although it was once thought that fringe benefits would tend to reduce inequality within the labor force, since such compensation represents a larger proportion of lower-wage workers' income than of high-income earners, quite the opposite appears to be the case.

While government-sponsored programs such as Social Security and workers' compensation insurance do indeed have an equalizing effect throughout the labor force, employer-sponsored benefits, in contrast, have "substantially increased compensation dispersion," especially at the lower end of the wage scale, according to data from the Bureau of Labor Statistics (Pierce 1998). This is so because less skilled workers are most likely to be employed in jobs that do not provide benefits, particularly in the small businesses providing most of the new jobs and in workplaces where employers can find a number of ways to limit coverage and/or exclude dependents. Benefits can be denied to entry-level employees and to those classified as "contingent." This latter category—covering all types of part-time employment—also accounts for growing numbers of white-collar workers without employer-sponsored health insurance and pension plans.

As Table 2 indicates, even the proportion of workers at the top of the wage scale with health insurance coverage has declined since 1982, while that of low-wage employees has shrunk from half to one-fourth.

Taxes and Transfers. Thus, although government transfers, most notably Social Security and the earned income credit, have had a moderating effect on income dispersion, the impact of the tax system has been less so, as rates have become less progressive than in the 1960s. At the federal level, the Taxpayer Relief Act of 1997 might benefit

Table 2
Percentage of Employees with Employer-Provided Health Insurance
source: Pierce (1998); see also U.S. Bureau of the Census (1998b).
bottom 10% of wage earners4926
middle 50% of wage earners9084
top 10% of wage earners9890

lower-income taxpayers through a $500 per child tax credit, but the major new programs—education and retirement savings accounts—will lighten the tax burden only of those who can afford to put money aside for the future. The big winners are the top 1 percent of wealthholders, who will receive, on average, more than $7,300 in tax relief from a cut in the capital gains rate, compared with savings of about $7 for the lower 60 percent of taxpayers (Congressional Budget Office 1998).

In addition, the FICA payroll tax, which takes a bigger bite out of the incomes of a majority of American families than does the federal income tax, is regressive in that most workers will pay the tax on 100 percent of income, while higher-income earners will be taxed on only the first $72,600 of wages. Also regressive in their impact are the types of levies currently favored by state governments—sales and sin taxes. As a consequence, the current tax structure has done more to reinforce than to moderate the trend toward income inequality in the United States.

Explanatory Variables. A number of other secular trends have been advanced to explain the long-term increase in income inequality in family and household income.

  1. Labor-market factors. One set of explanatory variables focuses on changes in the economy: (a) the shift in employment from manufacturing to service-related jobs, and the further division among service jobs between high- and low-skilled; (b) a concomitant decline in organized labor and the power of unions to negotiate favorable wage and benefit packages; (c ) global competition and immigration patterns that depress wages of low-skill workers; and (d) growth of the contingent labor force, such as temporary, part-time, and contract workers, who are typically ineligible for fringe benefits.
    All these trends have contributed to an extremely skewed wage structure in which a few at the top crowd out the rest of the field, a "winner take all" situation that characterizes all occupations but is most notable in professional sports and in the compensation package of chief executive officers (Frank and Cook 1995). With the exception of a handful of millionaire athletes, almost all the benefits of economic growth and changes in tax system since the mid-1970s have been reaped by a small stratum of Americans already enriched by education, opportunity, and social capital.
  2. Lifestyle factors. A second set of variables concerns long-term changes in patterns of marriage and living arrangements: (a) an increase in nonfamily and single-parent households due to later age at first marriage, high rates of divorce and separation, nonmarital births, and increased life expectancy, especially for widows in single-person households; and (b) the tendency toward endogamy among high-earning men and women, thus concentrating incomes and widening the split between the few at the top and the rest of the population.

Interestingly, the widening gap between top and bottom in income shares is not fully reflected in data on the distribution of wealth.


The methodological difficulties in measuring income are minor compared to those encountered in the study of net worth: the total value of all assets owned by a household, family, or person, less the debt owed by that unit. Such assets include investment portfolios, bank accounts, trusts, businesses, real estate, homes and their furnishings, insurance policies, annuities, pension equity, vehicles, works of art, jewelry, and other contents of safe deposit boxes. Because public records of such assets are minimal and/or difficult to trace, researchers are largely dependent on self-reports. In addition, because very few extremely wealthy units would appear in a national random sample, special frames must be constructed. As a result, data reported by the Bureau of the Census, the Federal Reserve Board, and the Internal Revenue Service are not always comparable because of differences in sampling, the type of assets being counted, and the way in which they are measured and weighted. In this section, we will briefly review the history of studying wealth in America, current data, and the generalizations that can be most confidently drawn.

The first systematic study of wealthholding was conducted in 1963 by the Federal Reserve Board (FRB), which found that the wealthiest one-half of 1 percent of households ("superrich") owned 25 percent of the aggregate net assets of the nation. The next one-half of 1 percent ("very rich") accounted for an additional 7 percent of net worth, and the next 9 percent (plain "rich"), accounted for one-third of the total, leaving 35 percent of the total net worth in the hands of the remaining 90 percent of households.

Although comparable data were not collected until 1983, evidence from the Internal Revenue Service (IRS) estate tax records suggests that the share of assets owned by the superrich declined between 1965 and 1976 to a low of 14.4 percent. This drop was due in part to an extended stock market slump and in part to changes in tax policy, as well as the growth of social welfare programs such as Aid to Families with Dependent Children, Medicare and Medicaid, and the liberalization of Social Security benefits, all of which shifted resources from the affluent to the more needy.

Interest in research on wealth picked up again in the mid-1980s, when both the Bureau of the Census and the FRB conducted studies designed to yield data comparable to the 1963 survey, although with different sampling frames and asset measures. The common finding, however, was that between 1976 and 1983 the downward trend of asset ownership by the superrich was dramatically reversed: from owning less than 15 percent of aggregate wealth to accounting for more than 30 percent just six years later. This doubling of asset ownership reflected an upward swing in the value of stocks, reinforced by Reagan administration policies on taxes and welfare favoring the more affluent. It must be noted, however, that these numbers are subject to considerable error due to sampling, nonresponse, and missing data.

Recent Studies of Household Wealth. In a major effort to standardize research findings, the FRB adopted a consistent weighting formula for adjusting data on household wealth from the Board's Survey of Consumer Finances for 1989, 1992, and 1995 (Kennickell and Woodburn 1997). According to these calculations, shown in Table 3, the share of net worth held by the superrich remained constant between 1989 and 1992 at about 23 percent, then rose to 27.5 percent between 1992 and 1995. At the other end of the distribution, the share owned by the 90 percent less well-off households also remained stable—at about 32 percent—over the entire 1989–1995 period. The 1992–1995 increase in the share of net worth of the top one-half of one percent, then, has come largely from the share of the 90–99 percentile "plain rich," which declined from 37 to 33 percent. In other words, there is little evidence that the rich have become richer at the expense of nonaffluent households.

There are marked contrasts between the few at the top and the rest of American households in the types of assets held. Wealth for the bottom 90 percent consists primarily of a principal residence, vehicles, and cash-value life insurance, and has been considerably diluted by rising debt (mortgage and credit card) in the 1990s. In contrast, stocks and bonds, trusts, and equity in businesses account for the bulk of the accumulated wealth of the top decile. By 1997, when stock ownership replaced the value of real estate as the leading component of aggregate wealth, the top 1 percent held more than half of all such investment instruments, with the bottom 80 percent holding 3 percent of the total value (Wyatt 1998). Yet even at this relatively low level of stock ownership, primarily through pension plans, low-wealth households will be especially vulnerable to sudden downturns in the market value of their investments.

Another set of numbers comes from the Bureau of the Census, which has, since 1990, conducted a panel study of households—the Survey of Income and Program Participation (SIPP)—that permits following the economic status of units over time. These data differ from those collected by the FRB in three crucial respects: the sample

Table 3
Percentage of Total Net Worth Held by Different Percentiles: 1989, 1992, and 1995
  percentile of net worth 
source: Kennickell and Woodburn (1997): 22.

frame does not yield many very high income households; a different set of assets are counted; and the distribution of wealth is measured differently. In the SIPP study, the distribution of "asset ownership" is computed on the basis of the median net worth of households at each quintile of monthly income. For both 1991 and 1993, the one-fifth of households with the lowest monthly income owned about 7 percent of the total net worth of all households, while those in the top fifth owned 44 percent (U.S. Bureau of the Census 1995). Unfortunately, comparable data for the next wave (1995–1996) have not yet been published, nor can the SIPP numbers be used for historical comparisons.

Personal Wealth. In addition to studies of household wealth by the Census and Federal Reserve Board, the Internal Revenue Service periodically publishes reports on individual wealth-holding, based on estate tax returns. These data are also subject to error—from sampling, from calculating mortality rates, and from underreporting, since not all assets can be tracked and since high-income earners have ways of dispersing and hiding assets prior to death. The most recent data come from surveys of estate tax returns of the very wealthy carried out by the IRS's Statistics of Income Division (SID) in 1992 and 1995 (Johnson 1997).

In 1992, it was estimated that 3.7 million adults, or 2 percent of Americans aged 21 or older, had gross assets of at least $600,000, which accounted for 28 percent of the aggregate personal wealth of the nation. In 1995, the number of wealthy persons had increased (to 4.1 million), as had their total net worth, but once adjusted for inflation, these differences were minimal. Indeed, looking at the very wealthiest—persons with a net worth of at least $1 million—in terms of numbers, total assets, and net worth, there was a sharp decline between 1989 and 1992, due primarily to the recession of 1990–1992. Between 1992 and 1995, however, the number of millionaires increased slightly (to 1.32 million), as did their total assets and net worth, but both still remained below the levels of 1989.

The SID also computed the percent of total U.S. net worth held by the top 1 percent and one-half of 1 percent of individual wealthholders. As shown in Figure 2, the share of total wealth owned by the wealthiest individuals rose from 1989 to 1992 before declining to 1989 levels. Clearly, according to these data, there was no dramatic shift in net worth from the less affluent to the top, although the dollar amount of their assets did appreciate. Between 1989 and 1995, then, the pattern for individual wealthholders was similar to that for household net worth in showing minimal increases in the concentration of wealth.

At the time of publication, detailed reports from the IRS and FRB on trends between 1995 and the end of the decade were unavailable, but material from Forbes magazine's yearly compilation of the 400 wealthiest individuals suggests that the fortunes of the very affluent have increased significantly (Forbes, October 12, 1998). In 1998, the minimum needed to appear on the list of the 400 wealthiest individuals in the United States was $500 million, up 5 percent from 1997 and more than double that of a decade earlier. Most dramatic was the increase in the proportion with a net worth of more than $1 billion, up from 170 persons in 1997 to 186 in 1998, almost half the total, compared to only 23 in 1986. Ten of the billionaires had fortunes in excess of $10 billion, including the five children of the founder of Walmart Stores, and Bill Gates, the founder of Microsoft, whose net worth almost equals the gross domestic product of New Zealand. Thus, while the very rich may not be getting richer at the expense of the less affluent, they received the lion's share of wealth created in the economic boom years of the mid-1990s.


Interesting changes are also taking place at the lower end of the income distribution, among the persons, families, and households officially designated as living below the poverty threshold. In the early 1930s, Franklin Delano Roosevelt could speak movingly of one-third of the nation being ill-housed, ill-clad, and ill-nourished in a society without an extensive social welfare system. Yet three decades later, in the early 1960s, despite the introduction of Social Security and other programs designed to minimize the effects of unemployment, more than 20 percent of Americans could still be defined as lacking a minimally adequate standard of living.

Defining Poverty. By 1964, as the nation geared up for Lyndon Johnson's War on Poverty, the Social Security Administration (SSA) was pressed to construct clear parameters for measuring impoverishment. The SSA turned to research conducted by the by the Department of Agriculture (DoA) in 1955, which found that families typically spent one-third of their income on food and which also computed the cost of a least expensive nutritionally adequate food plan. The SSA simply multiplied the cost of the DoA's food basket by three and, with some corrections for family size, age and sex of householder, and rural/urban residence, arrived at a dollar figure for yearly income—the poverty level—that neatly demarcated the poor from the nonpoor.

The value of the minimal food plan is adjusted each year to the cost-of-living index, but it no longer distinguishes rural from urban residence (the rural threshold had been higher, since it was thought that country folk could grow some of their own food) or female from male heads of household (on the grounds that women ate less than men). Only number of children and age of householder have been retained in the calculations. Nor has there been an adjustment for the fact that the cost of housing today typically exceeds that for food.

By the official yardstick, the poverty threshold for a single person in 1997 was $8,183, slightly higher for those under age 64 and slightly lower for someone aged 65 or older on the assumption that an older person eats precisely $276 worth less food per year than does a younger person. The poverty line was $12,802 for a family of three and $16,400 for a family of four. These are the dollar amounts considered adequate to house, feed, and clothe household members. Income in excess of the threshold officially lifts the unit out of poverty and therefore makes the unit ineligible for additional benefits, including both income supports (the former Aid to Families with Dependent Children and its successors, and Supplemental Security Income) and in-kind programs (food stamps, Medicaid, school lunch assistance, and housing subsidies). At the urging of conservative critics, the Bureau of the Census also publishes computations that include the cash value of these in-kind benefits in the definition of income, thus automatically reducing the poverty rate by about 25 percent. Nonetheless, at this writing, the official poverty level is still calculated on the basis of money income earned or received.

As Figure 3 indicates, the number of people below the poverty level was cut in half between 1959 and 1973—from 22.4 to 11.1 percent of the population—as a result of federal programs designed to assist the elderly, low-income families, and single-parent households. As the domestic War on Poverty fell hostage to the war in Vietnam, poverty rates began to rise, reaching a high of 15.2 percent during the recession of the early 1980s, and again in 1993, before declining to the current 13.3 percent, or 35.5 million persons.

In its 1998 report, the Census Bureau drew special attention to the fact that poverty declines in 1995–1997 have been much steeper for black and Hispanic persons than for Asian/Pacific Islanders and whites, but this effect is partly due to the fact that blacks and Hispanics were much further behind to begin with, so that any income increase will translate into a higher percentage compared to those initially less disadvantaged. Nonetheless, poverty among African Americans is at an all-time low, though the rate remains more than twice that for whites. The three factors most responsible for declining poverty rates today are: (1) The rise in the minimum wage that took effect in 1996, which accounts for the increased incomes of black and Hispanic single mothers, most of whom are hourly workers; (2) the earned income tax credit (EITC) now available to low-income workers, which has been particularly helpful to low-wage married-couple families; and (3) the economic boom that has generated a large number of jobs at the lower-skill level of the service sector, thus reducing unemployment and allowing low-wage workers to bargain for higher wages and benefits. In 1998, however, Congress voted down further raises in the minimum wage and failed to expand the EITC, thus leaving low-skill workers increasingly dependent on market forces.

The Census also computes the "ratio of income to poverty level"; that is, the number of Americans in families whose income is under half the poverty level, the "severely poor," as well as those with incomes 25 percent above the threshold, the "near poor." In 1997, 14.6 million persons, or 41 percent of the poor, were "severely poor." Another 12.3 million were "near poor," including many full-time workers, since the earnings of a year-round, full-time minimum-wageworker would still fall below the poverty threshold for a couple with one child.

Who are America's poor? Forty percent are children under age 18, whose poverty rate of about 20 percent is unchanged since 1989. Especially disadvantaged are the 59 percent of children under age six living with a female householder, no husband present. And while the poverty rate for all such households has hovered at slightly under one-third since 1987, there was a significant decline in poverty from 1996 to 1997 for black female householders. Even so, four in ten black female householders and almost half of Hispanic female householders were officially poor in 1997.

The most powerful variable affecting poverty status, however, is not race or ethnicity, but sex. It is women of all ages, especially those without husbands, from teenage mothers to elderly widows, who are most at risk of being poor. If economic well-being depends on working full time, remaining married, and being free of child-care responsibilities, then single mothers with limited education and job skills will be especially disadvantaged (U.S. Bureau of the Census 1998e). Yet it is precisely these women, many of whom were dependent for survival upon the income provided by Aid to Families with Dependent Children (AFDC), who came to be blamed for their own misfortune and to symbolize the failure of public welfare programs.

The Poverty Debate of the Mid-1990s. Although poverty programs absorbed less than 5 percent of the national and state budgets, they became a focal point for political debate in the mid-1990s. The public had come to believe, contrary to empirical studies, that AFDC families remained on the welfare rolls for generations, that poor women had additional children in order to increase their monthly benefit, and that it was characteristics of the poor themselves (laziness, substance abuse, sexual immorality) that accounted for poverty. In fact, fewer than 10 percent of the poor can be considered long-term welfare recipients, primarily women who entered the system as very young unwed mothers and who have been unable to develop the job skills or to find employment near their home that pays enough to lift the family above the poverty level. Nor is there any consistent relationship between benefit levels and the fertility of poor women.

As the SIPP data clearly show, poverty is a transitory state for the great majority who fall below the threshold. For example, although 21.4 percent of Americans were poor at some point in 1994, the proportion who were poor for all of 1993 and 1994 was only 5.3 percent. Almost half of all spells in poverty (47.3 percent) lasted 2–4 months, and 75 percent lasted less than one year (U.S. Bureau of the Census 1988f). People fall into poverty when a marriage ends, when employment stops, or when children become ill; they are lifted out of poverty when they remarry, when they are employed, and when family members are restored to health.

As part of their attack on all government programs to which entire classes of citizens are entitled, conservative critics of the American social welfare system have long maintained that there is a "culture of poverty," whereby maladaptive attitudes toward work and family are transmitted across generations (e.g. Murray 1984). In this view, welfare dependency itself is the problem, and the solution is to remove income supports, so that employment becomes the more attractive alternative. In contrast, most sociologists would focus on the structural conditions that produce unemployment, family dissolution, educational failure, and homelessness, with people's behavior perceived more as a response to than as a cause of their situation.

The debate was won by those seeking radical change in the welfare system, and in 1996 the Personal Responsibility and Work Opportunity Act replaced AFDC and several other federal assistance programs with block grants to the states for Temporary Assistance for Needy Families (TANF). The states were given relative freedom to construct their own programs for moving former welfare recipients into the paid labor force.

It is too early to tell what effect the new systems will have on poor women and their children. Some states will be more effective than others in providing the job training and child care required for full-time labor-force participation; others will be more or less punitive. It will also be very difficult to disentangle the effects of the business cycle from those of public policy, or to determine whether people leaving welfare would have done so anyway given the transitory nature of most poverty spells. At the moment, welfare rolls have dropped dramatically, as a function of both economic growth and tighter eligibility requirements. And although the poverty rate also fell in 1997, the decline was minimal. Much will depend on whether the private sector can continue to generate jobs that former recipients can find and hold—and that will pay a family wage.


Any discussion of wealth, poverty, and income inequality within a society must take into account the vast differences between modern industrial and the less developed nations. Poor people in high-income societies are rarely as deprived of the basic necessities of survival as are most people in the Third World, where income inequality is typically much greater than in the industrialized countries. But Americans do not compare themselves to Sudanese; rather, they compare themselves to other Americans to whom they feel similar. It is a sense of relative rather than absolute deprivation that tends to fuel resentment. At this writing, despite the extent of income inequality in the United States, there are few appreciable signs of anger directed toward the top wealthholders. Rather, whatever ill will has been generated by blocked opportunity appears to be directed toward racial and ethnic minorities at the same or lower social-class level.

Yet, when compared with other Western democracies, the United States has the most unequal distribution of income and highest poverty rate. For example, while the percentage of American children in poverty was among the highest in the developed world, proportionately fewer were lifted out of poverty by government aid (Atkinson et al. 1995; United Nations 1998). This is so because the United States has the least extensive social welfare system of any modern state, and since 1996 even this limited "safety net" has been reduced. Alone among its industrialized peers, the United States is without a comprehensive family policy, lacks a national health insurance system, and provides minimal assistance to the most needy. As a consequence, there are few institutionalized mechanisms other than Social Security—which has also come under attack from those who would turn it into a private rather than public responsibility—for the redistribution of income that might narrow the gap between the very rich and very poor or that might substantially reduce both the likelihood and impact of poverty. In the absence of a revitalization of a sense of collective responsibility, income inequality will continue to characterize the United States. Indeed, all signs point to a continuation of economic, social, and political trends that elevate individual over collective interests and that reinforce the power of the market and weaken that of government.


Atkinson, Anthony B., Lee Rainwater, and Timothy M. Smeeding 1995 Income Distribution in OECD Countries: Evidence from the Luxembourg Income Study. Washington, D.C.: OECD (Organization for Economic Cooperation and Development) Publications and Information Center.

Congressional Budget Office 1988 Data provided to Citizens for Tax Justice, Washington, D.C.

Frank, Robert H., and Philip J. Cook 1995 The WinnerTake All Society: How More and More Americans Competefor Ever Fewer and Bigger Prizes, Encouraging EconomicWaste, Income Inequality, and an Impovershed CulturalLife. New York: Free Press.

Johnson, Barry W. 1997 "Personal Wealth, 1992–1995." Internal Revenue Service Statistics of Income Bulletin 16(3):70–95.

Kennickell, Arthur B., and R. Louise Woodburn 1997 Consistent Weight Design for the 1989, 1992, and 1995SCFs, and the Distribution of Wealth, Revision II. Washington, D.C.: Federal Reserve Board.

Mishel, Lawrence, Jared Berbstein, and John Schmitt 1998 The State of Working America. Washington, D.C.: Economic Policy Instutute.

Murray, Charles 1984 Losing Ground. New York: Basic.

Pierce, Brooks 1998 Compensation Inequality. Washington, D.C.: Bureau of Labor Statistics.

United Nations 1998 Human Development Report 1998. New York: United Nations.

U.S. Bureau of the Census 1995 Asset Ownership ofHouseholds: 1993. Current Population Reports, Series P70, No. 47. Washington, D.C.: U.S. Government Printing Office.

——1996 A Brief Look at Postwar U.S. Income Inequality. Current Population Reports, Series P60, No. 191. Washington, D.C.: U.S. Government Printing Office.

——1998a Changes in Median Household Income: 1969to 1996. Current Population Reports, Special Studies, Series P23, No. 196. Washington, D.C.: U.S. Government Printing Office.

——1998b Health Insurance Coverage: 1997. Current Population Reports, Series P60, No. 202. Washington, D.C.: U.S. Government Printing Office.

——1998c Measuring 50 Years of Economic ChangeUsing the March Current Population Survey. Current Population Reports, Series P60, No. 203. Washington, D.C.: U.S. Government Printing Office.

——1998d Money Income in the United States: 1997. Current Population Reports, Series P60, No. 200. Washington, D.C.: U.S. Government Printing Office.

——1998e Moving Up and Down the Income Ladder. Current Population Reports, Series P70, No. 65. Washington, D.C.: U.S. Government Printing Office.

——1998f Trap Door? Revolving Door? Or Both? Current Population Reports, Series P70, No. 63. Washington, D.C.: U.S. Government Printing Office.

Wyatt, Edward 1988 "Share of Wealth in Stock Holdings Hits 50-year High." New York Times January 11, p. 1ff.

Beth B. Hess