[An updated version of this article can be found at Poverty in America in the 2nd edition.]
Poverty is one of America's most persistent and serious problems. The United States produces more per capita than any other industrialized country, and in recent years has devoted more than $500 billion per year, or about 12 percent of its gross national product, to public assistance and social insurance programs like Social Security, Medicare, Aid to Families with Dependent Children (AFDC), food stamps, and Medicaid. Despite our wealth and these efforts to reduce income inequality, poverty is more prevalent in the United States than in most of the rest of the industrialized world. It is also more prevalent now than it was in the early seventies, when the incidence of poverty in America reached a post-war low. According to the Census Bureau, 33.6 million Americans were poor in 1990, almost 14 percent of the population.
These official figures represent the number of people whose annual family income is less than an absolute "poverty line" developed by the federal government in the midsixties. The poverty line equals roughly three times the annual cost of a nutritionally adequate diet. It varies by family size and is updated every year to reflect changes in the consumer price index. In 1990 the poverty line for a family of four was $13,359.
Many researchers believe that the official method of measuring poverty is flawed. Some argue that poverty is a state of relative economic deprivation, that it depends not on whether income is lower than some arbitrary level, but whether it falls far below the incomes of others in the same society. But if we define poverty to mean relative economic deprivation, then no matter how wealthy everyone is, there will always be poverty. Others believe the official method is conceptually correct but errs by omission. For example, official poverty figures take no account of the value of noncash government transfers like food stamps and housing vouchers, which serve as income for certain purchases. The Census Bureau estimates that the inclusion of the market value of these benefits in family income would have reduced the measured poverty rate by 1.4 percentage points (or by approximately 10 percent) in 1990.
The official definition also ignores the value of assets like owner-occupied housing and consumer durables that do not generate money income but increase household resources nonetheless. According to one study based on data from the early eighties, 31 percent of the poor owned their own homes and 48 percent owned a motor vehicle, and the average net worth of poor families was thirty thousand dollars. The Census Bureau estimates that if the net imputed return on equity in owner-occupied housing were included in income, the poverty rate would have been 1.2 percentage points lower in 1990.
Another problem with the poverty measure arises from the dramatic shift in household composition since World War II. Smaller, more fragmented households are more common today than ever before. This suggests that some poor households were formed voluntarily for the sake of privacy and autonomy for their members. To the extent that some people have willingly sacrificed their access to the economic resources of parents, spouses, or adult children, some of the increase in poverty may actually represent an improvement in well-being. Such inaccuracy is inherent in a poverty measure based solely on a household's money income.
Whatever their flaws, the official figures are widely used as a simple gauge of the trends in poverty. According to the official Census Bureau figures, the poverty rate declined from 22.2 percent in 1960 to 12.8 percent in 1989. Most of this decline occurred in the sixties. By 1969 the poverty rate had fallen to 12.1 percent. It then hovered between 11.1 and 12.6 percent in the seventies, increased to a recent peak of 15.2 percent in 1983, and then decreased to 12.8 percent in 1989. Although the lack of rapid progress in recent years is discouraging, a longer-term perspective leaves a net positive impression. For example, according to one estimate by Christine Ross, Sheldon Danziger, and Eugene Smolensky, more than two-thirds of the population in 1939 was poor by today's standards.
The trend in poverty masks the divergent incidence of poverty among various demographic groups. The poverty rate among the elderly, for example, after declining dramatically from 35.2 percent in 1959 to 12.2 percent in 1990, is now lower than for the rest of the population. The poverty rate among children also declined after 1959, but only through the early seventies. It has swung up sharply since that time, and at 20.6 percent in 1990 remains higher than poverty rates among other age groups. The poverty rate among black households has also declined over the last thirty years, but at 31.9 percent in 1990 remains three times as high as the rate among white households.
The incidence of poverty also is higher among households headed by women. Although the poverty rate among these households declined from 49.4 percent in 1959 to 37.2 percent in 1990, they remain far more likely to be poor than other types of households. This higher incidence of poverty, together with the rising share of households headed by women, has led to what researchers call the "feminization of poverty," with an increasing fraction of the poor in female-headed households. Between 1959 and 1990 this fraction rose from 17.8 percent to 37.5 percent.
The failure of the aggregate poverty rate to decline in the seventies, and its subsequent rise in the eighties, suggest to some that the War on Poverty launched by the federal government in the midsixties failed. Indeed, the incidence of poverty was as high in the late eighties as it was in the late sixties, and the average poverty rate for the eighties was 2 percentage points higher than the average for the seventies. Researchers have suggested a number of plausible explanations for these trends, including changes in the composition of households, slower economic growth, the failure of government training programs to increase the skills of the poor, and the rise of a permanently poor urban underclass. Some also argue that the income transfer policies designed to alleviate poverty have themselves helped perpetuate it. Although all of these factors have likely contributed to the problem, the relative importance of each remains somewhat unclear.
The rapid growth of households headed by women and unrelated individuals, who typically cannot earn as much as married-couple families, has left a larger share of the population in poverty. This demographic trend appears to have put especially strong upward pressure on the poverty rate in the seventies, when the share of female-headed households rose most rapidly. Decennial census data indicate that if demographic characteristics such as the age, race, and gender composition of households had not changed between 1950 and 1980, the poverty rate would have been 3 percentage points lower in 1980 than it actually was.
Trends in economic growth also influence the incidence of poverty. Researchers have found that recessions have a disproportionate impact on the poor because they cause rising unemployment, a reduction in work hours, and the stagnation of family incomes. The link between macroeconomic conditions and the incidence of poverty was clearly visible during the 1982 recession, when the poverty rate rose to 15.2 percent, up from 13.0 percent in 1980. It was likewise with structural unemployment—the unemployment that results not from temporary declines in aggregate demand, but from a long-term mismatch between the skills demanded by employers and those supplied by workers. The rising trend in structural unemployment that started in the sixties appears to have contributed to the persistence of poverty. One study by Rebecca Blank and Alan Blinder finds that each 1-point increase in the unemployment rate of males aged twenty-five to sixty-four increases poverty by 0.7 percentage point. It should not be terribly surprising, then, that poverty was almost as high in the second half of the eighties, when unemployment averaged 6.2 percent, as it was in the second half of the sixties, when unemployment averaged only 3.8 percent.
Training and compensatory education programs like the Job Corps and Head Start, designed as part of the War on Poverty to increase the skills of the poor, may also have influenced trends in poverty. One study, by Gary Burtless of the Brookings Institution, estimates that the federal government spent $282 billion in 1986 dollars on these programs between 1963 and 1985. Most of these programs have not been carefully evaluated, but of those that have, some have been successful. For example, some education programs like Head Start have had a positive effect on poor children, and some employment and training programs have raised the earnings of adult women but were generally less helpful to adult men.
Some researchers believe that the growth of an urban underclass locked in a cycle of welfare dependency, joblessness, crime, and out-of-wedlock pregnancy has also contributed to the persistence of poverty. Although researchers define the underclass in numerous ways, one common definition is the number of poor who live in inner-city neighborhoods where poverty rates are 40 percent and above. By this definition the underclass grew by 36 percent between 1970 and 1980 to 1.8 million people but is still only about 7 percent of the poor population nationwide. The fact that the underclass is a relatively small group means that its growth cannot explain much of the trend in aggregate poverty.
Finally, some researchers blame the persistence of poverty on income-transfer policies. These are typically divided into two categories: public assistance programs, like AFDC, food stamps, and Medicaid, which were designed to help people who are already poor; and social insurance programs, like Social Security, unemployment insurance, and Medicare, which were designed to prevent poverty when certain events like retirement or layoff threaten a household's well-being. Expenditures on these programs totaled roughly $570 billion in 1988, up 360 percent in real terms since 1965. In 1988, social insurance expenditures accounted for three-quarters of this total, with Social Security alone accounting for nearly 40 percent. Within the social insurance category, some 80 percent of expenditures were in the form of cash. In contrast, of the $137 billion spent on public assistance programs in 1988, less than one-third was paid out in cash. The rest was distributed through in-kind transfer programs like food stamps, housing vouchers, and Medicaid, which can be used only for buying food, housing, and medical care, respectively.
The antipoverty effectiveness of these programs is typically measured by counting the number of people with pretransfer incomes below the poverty line whose incomes are raised above the poverty line by the income transfers. According to government estimates, social insurance and public assistance programs moved over 40 percent of the pretransfer poor above the poverty line in 1989. This implies that the poverty rate is reduced by nearly 9 percentage points by these programs.
Economists realize, however, that by ignoring the incentive effects these programs have on recipients, this method of analysis overstates the success of transfer programs. Some critics of welfare policy argue that means-tested cash-income transfers like AFDC prevent recipients from leaving poverty by reducing their incentives to work and to form stable two-parent families. For example, when a recipient receives an AFDC payment, work becomes less necessary because the payment can be used instead of a regular paycheck to buy necessities like food and housing. In addition, work becomes less attractive because AFDC administrative rules require the reduction of benefits as the recipient's earned income rises. If a woman finds a job paying four dollars an hour but welfare rules require a fifty-cent reduction in her AFDC benefits for each dollar in wages, the woman's effective pay before taxes falls to only two dollars per hour and is even lower after taxes. As a result, she may be less willing to take the job.
Economists have found, however, that these incentive effects do not reduce the work efforts of recipients substantially. Sheldon Danziger, Robert Haveman, and Robert Plotnick estimate that if all income-transfer programs, including Social Security, disability insurance, unemployment insurance, and AFDC, had been eliminated, transfer recipients would have increased their work hours by 4.8 percent during the seventies. Roughly 80 percent of the increase would have been caused by the removal of social insurance programs, and only 20 percent by the removal of means-tested transfers.
Critics of welfare policy argue that because AFDC is more readily available to families headed by women than to married-couple families, it encourages divorce, discourages remarriage, and increases out-of-wedlock childbearing. While their point is well taken, economic research suggests that these effects are small. One study by Mary Jo Bane and David Ellwood concludes that an AFDC benefit increase of a hundred dollars per month (in 1975 dollars) to a family of four (a 38 percent increase over the median state benefit level for that year) would increase the number of female-headed families by 15 percent. The study finds that most of this increase results from the movement of single mothers out of the homes of their parents. There is little or no evidence that welfare encourages out-of-wedlock childbearing or that it has much of an influence on divorce or remarriage rates.
This body of evidence suggests that the persistence of poverty cannot be attributed to income-transfer programs themselves. Although transfer programs surely have not reduced poverty by the full 9 percentage points mentioned earlier, they clearly have reduced poverty significantly. Indeed, one of the greatest success stories is the decline in poverty among the elderly, due in large part to the growth of Social Security and Medicare.
In sum, a variety of factors have influenced the incidence of poverty. Those that have reduced the poverty rate, in rough order of importance, are the growth of cash transfers, the investments in government training and education programs, and the overall growth in the economy since the midsixties. Factors that have increased the poverty rate include, in order of importance, the increase in the unemployment rate, the growth of female-headed families, and (possibly) an increase in dysfunctional behavior associated with the rise of the underclass. All of these factors together have left the incidence of poverty much the same as it was in the late sixties.
Isabel V. Sawhill is a senior fellow at Brookings Institution. She was previously program associate director for human resources, veterans, and labor with the Office of Management and Budget and a senior fellow at the Urban Institute in Washington, D.C. From 1977 to 1979, she was the director of the National Commission for Employment Policy. This article was prepared with the assistance of Mark Condon.
Blinder, Alan. "The Level and Distribution of Economic Well-Being." In The American Economy in Transition, edited by Martin Feldstein. 1980.
Danziger, Sheldon, and Daniel Weinberg. Fighting Poverty: What Works and What Doesn't. 1986.
Ellwood, David. Poor Support: Poverty in the American Family. 1988.
Haveman, Robert. Starting Even: An Equal Opportunity Program to Combat the Nation's New Poverty. 1988.
Sawhill, Isabel V. "Poverty in the U.S.: Why Is It So Persistent?" Journal of Economic Literature 26 (September 1988): 1073-1119.