Welfare Reform and Poverty

Ron Haskins and
Ron Haskins Senior Fellow Emeritus - Economic Studies

July 1, 2001

Child poverty will be an important issue as the welfare reform reauthorization debate heats up in the 107th Congress. Since the federal government adopted a standard definition of poverty in the 1960s, scholars, politicians, the media, and the public have used poverty as a measure of the strength and success of government policy and the economy. Having a standard definition of poverty, however, does not prevent disputes on many issues related to poverty. Thus, it comes as no surprise that reauthorization will spark debates about whether the federal government is primarily responsible for poverty reduction, about whether welfare reform has led to adequate reductions in poverty, and about the specific actions government should take to further reduce poverty. We begin this brief with a rendition of the facts about poverty on which there is broad agreement. After that, we address a number of policies aimed at reducing child poverty that we expect to dominate the reauthorization debate.

Evidence on Children’s Poverty

Several key pieces of information provide what we believe is a solid base of knowledge about child poverty for the reauthorization debate. The first piece of information is how the federal government measures poverty. The Census Bureau, the official scorekeeper for federal poverty measures, computes poverty by comparing individual or family income from earnings, cash government payments, and a few other sources to a criterion income level, adjusted for family size and inflation, that represents a minimally adequate standard of living. In 2001, the criterion income, or poverty line, was about $14,600 for a family of three.

Both the definition of income and the poverty line criterion are controversial. One criticism of the Census Bureau’s official measure of poverty is that not all sources of family income are included. Non-cash government benefits, such as food stamps and housing, as well as cash income provided to families through the tax code, are omitted. A mother with two children earning $10,000 is eligible for over $2,000 in food stamps and nearly $4,000 from the Earned Income Tax Credit (EITC). Both of these sources of income are, however, ignored by the official definition. The official definition is also criticized because it ignores work expenses such as transportation and child-care. In addition, many observers believe the poverty line is too low.

The official approach to defining income is especially unfortunate in the new world of welfare reform because more mothers are now working and eligible for food stamps, the EITC, and other benefits that are not counted as income. In response to this concern, the Census Bureau in recent years began computing a broader poverty measure that includes food stamps, the EITC, and several other government benefits in the calculation of income.

The second set of information that provides essential background for the upcoming debate over child poverty is Census Bureau data on the course of child poverty in recent decades. Figure 1 shows changes in the Census Bureau’s official measure of child poverty from 1959 to 1999 and the broader measure from 1979 (the earliest year for which it was computed) to 1999. The top line in figure 1 shows that progress against child poverty under the official definition was rapid during the 1960s. Then, after a brief period of stability in the early 1970s, child poverty drifted generally upwards until the early 1990s, as shown by the fact that as compared with an average of about 15 percent in the early 1970s, child poverty seldom fell below 20 percent in the years after 1980. Although poverty did decline during the economic expansion of the 1980s, it moved back up again and reached the 20 percent level in the late 1980s and early 1990s. Thus, in the quarter century before the welfare reform movement that began in the states in the early 1990s and culminated in the federal welfare reform legislation of 1996, child poverty was rising modestly but noticeably.

Recent trends, however, are more encouraging. As shown in an earlier policy brief in this series (see Welfare Reform & Beyond Policy Brief #1, January 2001), welfare rolls, overall child poverty, and poverty among black children have all been declining rapidly since 1993. At that time, states began aggressively implementing welfare-to-work programs under waivers from federal law, the EITC and other “make work pay” policies were expanded, and the economy started growing rapidly. By 1999, the child poverty rate had fallen to its lowest level since 1979 and poverty among black children to its lowest level ever, although their rate was still well above the rate for white children. The rate for Hispanic children was also declining rapidly and by 1999 had reached its lowest level since 1979.

A third type of background information illuminates what is surely the most important demographic correlate of poverty; namely, the proportion of children in female-headed families. This proportion is important in accounting for poverty because children in these families are much more likely to be poor than children in two-parent families. Thus, it is unfortunate that the percentage of children in female-headed families increased steadily from 8 percent in 1960 to over 20 percent in 1983 (figure 2). It has remained above 20 percent ever since. The figures for black children are even more dramatic. As compared with about 20 percent in the mid-1960s, by the mid-1980s about 50 percent of black children were in female-headed families and the percentage has remained above 50, although in recent years there has been a slight upturn in the percentage of black children living in married-couple families.

Not all female-headed families are created equal. The type of female-headed family most likely to be poor and stay poor for long periods is never-married female-headed families. The percentage of single-parent families headed by never-married mothers increased from less than 1 percent in 1960 to about 31 percent in 1995, an increase of immense proportions and the single most important factor contributing to high and growing child poverty rates.

Given the unfortunate correlation between families headed by women and poverty, and the increases in the percentage of children living in female-headed families-especially those headed by a never-married mother-progress against poverty depends to a substantial degree on improving the financial status of female-headed families. Until recently, the primary means by which federal and state policy tried to achieve this goal was providing welfare benefits.

Beginning in the mid-1960s, federal and state governments developed new programs and spent more money to assist poor, especially female-headed, families. The 1960s saw a substantial decline in poverty among female-headed families from over 50 percent to 38 percent in 1970, perhaps because of both new government welfare programs and a strong economy (figure 3). Progress for black female-headed families was especially notable, with the poverty rate falling from about 70 percent in 1960 to under 60 percent in 1970. In the quarter century between 1970 and 1994, however, little progress was made and the poverty rate for female-headed families was virtually stagnant, averaging nearly 39 percent. But then the rate began to fall again in 1995 and reached 30.4 percent in 1999, the lowest level ever. To a substantial degree, the recent progress against child poverty is associated with the rapid reductions in poverty among female-headed families.

A fourth background fact about poverty uses the broader measure of poverty discussed above to compare relative progress against poverty during the economic recovery of the 1980s with that during the economic recovery of the 1990s. As shown by the bottom line in figure 1, this broader measure consistently shows a much lower level of child poverty than the level shown by the official measure. But an even more interesting insight from the broader measure is that poverty declined by only 15.5 percent during the expansion of the 1980s as compared with 35.5 percent-more than twice as much-during the expansion of the 1990s. The gains against child poverty during the 1990s under the broader measure are comparable to the impressive gains of the 1960s.

There is wide agreement in the research community both on the accuracy of this basic picture of steep declines in child poverty in recent years and greater declines in the 1990s than at any other time since the 1960s. There is also agreement on the major causes of these trends. Employment among mothers who head families, and especially those who have never married, has increased greatly since the early 1990s and today has reached its highest level ever. Although there are undoubtedly many factors related to this remarkable increase in employment by single mothers, welfare reform, recent increases in the EITC, and a booming economy are the most important factors. Second, real wages of low-income workers increased between 1994 and 1999. Third, the often-modest earnings of these female heads are augmented by public benefits enacted by Congress to help working families. These benefits include the EITC, food stamps, Medicaid, the child tax credit, child care, and other benefits. Many of these programs, along with the minimum wage, were created by Congress or have been expanded by Congress since roughly the mid-1980s. It is the combination of increased earnings from work and the system of public subsidies for low-wage workers, both occurring in a rapidly expanding economy, which constitutes the most parsimonious explanation of the substantial progress against poverty in recent years. Working families do not always receive all the benefits for which they are qualified under these programs, but the programs still play a major role in reducing child poverty.

Not All the Poverty Data Are Favorable

Although there is broad agreement that these facts about poverty reduction are correct, many analysts think the reduction in child poverty should have been much greater given the remarkable increases in employment and earnings by low-income mothers. They emphasize data on the poverty gap (the amount of money required to bring every poor family to the poverty line) to argue that the true picture of poverty is not as bright as the one painted above. Data on the poverty gap are especially important because they take into account not only the frequency of poverty, but also its depth. Figures in the first row of table 1 (see page 6) show that the market economy has become much more effective in reducing the poverty gap in recent years. Thus, based only on earnings and other non-governmental income, the poverty gap reached a whopping $51.7 billion in 1993 and $46.1 billion in 1995, but then fell to $34.8 billion in 1999, lower than in any year since 1979. This positive development is consistent with the employment and poverty data reviewed above. The substantial increase in work by mothers has greatly reduced the poverty gap before any government transfers.

However, given the relatively low level of poverty after market income, one might expect that government programs would have a major impact in reducing the poverty gap even more. Unfortunately, this is clearly not the case. Consider the data for 1993 and 1999: based just on market income (before any government transfers), the poverty gap was $51.7 billion in 1993 but only $34.8 billion in 1999. Adding income from social insurance programs (such as Social Security) reduced the poverty gap in both years, but by $7 billion in 1993 as compared with only $5.3 billion in 1999. Even more revealing, means-tested cash and non-cash programs (government programs that provide benefits to families whose income is less than a given amount) reduced the poverty gap by $23.4 billion in 1993, but only $11.9 billion in 1999. Despite improvement in the role of the EITC in reducing the poverty gap (see fourth row of table 1), on the whole, government means-tested programs were much less effective in reducing poverty after 1993.

If we turn from figures on reducing the poverty gap to simple measures of the income of mother-headed families issued by the Census Bureau, we find this same pattern of increased earnings before taxes and transfers but big offsetting losses of welfare income. The poorest 40 percent of single-mother families increased their earnings by about $2,300 per family on average between 1995 and 1999. But their disposable income increased on average only $292. Once again, increased earnings are not translating into greater disposable income because these families are losing most of their cash welfare income and much of their food stamp income as well. There is some evidence that underreporting of welfare income increased over this period, but more families are also encountering substantial work expenses.

Thus, although there is broad agreement that child poverty has fallen substantially under the 1996 welfare reforms, many liberals are dissatisfied with the degree of poverty reduction. Based on data from the Current Population Survey, Wendell Primus has estimated that in 1999 there were 700,000 families that were worse off than their counterparts before welfare reform. The primary reason is that the substantial earnings increases by mothers have not been matched by the ability of government programs to remove them from poverty as effectively as in the past. In short, liberals believe that government programs are letting these families down.

These findings about family income are especially troubling in light of recent evidence from the Manpower Development Research Corporation (MDRC) that increased income-and not just employment-has a positive impact on children. The MDRC study examined five experiments covering 11 different welfare reform programs and found that increased employment among the parents in a family did not by itself significantly improve their children’s lives. It was only in programs in which increased employment was accompanied by increased income that there were positive effects, such as increased school achievement for elementary-aged children. These results lead to the conclusion that boosting income, especially by ensuring that working families actually receive the work support benefits for which they are qualified, is a major issue and should receive careful attention from Congress during reauthorization.

Policies to Reduce Child Poverty

As might be expected, there is little consensus on steps Congress should take during reauthorization to address child poverty. Since welfare reform began in the states, the general pattern of income by poor and low-income mothers has consisted of two very clear trends: increased income from earnings and the EITC, and decreased income from cash welfare and food stamps. Conservatives generally believe this is a desirable pattern because it shows declining dependency by welfare mothers. Most liberals agree that the increases in work and earnings are desirable, but also believe that the decreased income from cash assistance and food stamps is a troubling indicator because those decreases greatly exceed the declines in poverty.

Both liberals and conservatives are likely to agree that it is necessary to maintain substantial work incentive by ensuring financial rewards for work. Most liberals and at least some conservatives are also sympathetic to the view that providing work subsidies to help working families escape poverty is a worthy goal of welfare reform. However, conservatives emphasize the importance of time limits and work requirements. If recipients do not meet work requirements, they must be subject to strong penalties, including benefit termination. Thus, the new welfare system has both increased financial incentives and powerful new limits on welfare. Most conservatives see these limits as the heart of the new reforms and a major reason for declining welfare rolls, increasing work, and declining poverty among female-headed families. Even when the limits on welfare increase the odds of poverty for some individual families, many conservatives are willing to accept this unfortunate result as a cost of maintaining a system of authoritative emphasis on work, especially because most adults who lose benefits and increase their risk of poverty are doing so on the basis of their own choices. Nonetheless, most conservatives are willing to make investments to reduce poverty-as long as the benefits are tied to work and the discipline of work is maintained.

Food Stamp Reforms

Because fewer than half the mothers leaving welfare continue to receive the food stamps to which they are entitled, Congress should carefully examine food stamps to ensure that eligible children and parents have every opportunity to receive these entitlement benefits. Several changes in food stamp policy could boost participation among the working poor. The federal quality control system, designed to control fraud and abuse, should be modified because it has led many states to take administrative actions that reduce participation among working families. The food stamp program itself should be simplified. Asset tests, in particular, should be revisited, because current tests make the program hard to administer and make too many low-income working families ineligible. The adequacy of the food stamp benefit structure also warrants reexamination, particularly in light of the cuts in benefits enacted in 1996. Another reform supported by many conservatives would be to provide states with a food stamp block grant and then allow them to solve coordination issues without interference from the federal government.

One of the major reasons low-income working families have not improved their financial situation more than they have is that many of them do not receive the food stamp benefits to which they are entitled. Fixing this food stamp problem would have a major impact on the poverty gap figures reviewed above and would substantially reduce child poverty.

Medicaid Reforms

Changes in Medicaid are also needed to address the problem of eligible children not receiving the benefit. Typically, a parent in a family of three loses Medicaid eligibility when her income surpasses 67 percent of the poverty line (about $9,800 for a family of three in 2001). Research has shown that expanding state Medicaid programs to cover parents also increases the number of low-income children who receive coverage, apparently because parents are more likely to apply if they themselves receive coverage. An additional option is to expand funding for the State Children’s Health Insurance Program (SCHIP) and allow states to use the funds to extend coverage either through their Medicaid or SCHIP programs to low-income working parents and their children. Some liberals would also allow non-custodial parents who are supporting their children to be covered by Medicaid or SCHIP.

Expansion of Tax Code Benefits

There are a host of tax code changes that could help reduce poverty. One change on which there appears to be broad agreement is to reduce the marriage penalty in the EITC by starting the phase-out range (the income at which the benefit begins to fall) at a higher income for married families. Another option is to make the child tax credit partially refundable (meaning that, in a kind of negative income tax, low-income families without tax liability would be given a cash return equal to the amount of the credit). One approach outlined in an earlier brief in this series is to provide a significant credit to parents who work full time by phasing in the credit as earnings increase. The recent tax bill signed into law by President Bush in June 2001 adopted this approach and included modifications of the EITC to reduce the marriage penalty. Another tax policy that favors low-income families would add another tier to the EITC that would provide higher benefits for families with three or more children. All of these policies would significantly reduce child poverty.

Help for Floundering Families

According to one study, 44 percent of TANF recipients reported at least two significant obstacles to work, such as low education, no recent work experience, language barriers, mental or physical health problems and disabilities, and lack of transportation or child care. This study also showed that the share of recipients reporting no work activity increased steadily with the number of significant obstacles. Research also indicates that families that fail to meet work requirements (and have their welfare benefit reduced as a result) have greater obstacles to employment than other families receiving welfare. Providing services and accommodations appropriate to the greater needs of such floundering families, while maintaining the emphasis on work, should enable more of these parents to succeed in the workplace. Some states have been able to help many families that initially failed to comply with work requirements, rather than simply sanctioning them and setting them adrift. All states should be encouraged to follow this policy. Congress should also provide incentive funding to states that are willing to design and conduct demonstration programs (including careful evaluations) aimed at developing new ways to help these families work toward self-sufficiency.

The Five-Year Time Limit

Most liberals and conservatives disagree sharply on the fairness and efficacy of welfare time limits. Most liberals would favor dropping time limits altogether because they believe work requirements and sanctions are sufficient measures to address the issue of dependency. However, recognizing the low political probability of eliminating time limits, they would expand the share of participating families that states may exempt from time limits-now 20 percent of caseload-to at least 30 percent. Today, the exemption covers many fewer people than was envisioned when the law was passed because current caseload size is much smaller than anyone expected (and 20 percent applied to a smaller number yields fewer exemptions). Many conservatives oppose this reform.

By contrast, another modification of time limits appears to enjoy considerable bipartisan support. In many states, work requirements and earnings disregards encourage welfare recipients to make the transition to work by combining earnings with a wage supplement in the form of cash assistance payments. But working adults who receive these wage supplements risk using up their five years of TANF eligibility because federal law requires states to count every month adults receive assistance from the TANF program as a month of welfare use, even if the adult is working full time. Thus, liberals would require states to stop the five-year clock whenever parents are working. Many conservatives would support this policy if it were a state option and applied only to parents working at least 25 or 30 hours per week.

Child Support and Supporting Non-Custodial Parents

Non-custodial parents (most of whom are fathers) should be encouraged to provide for their children both financially and emotionally. Child support, for poor families that receive it, is the second largest source of family income-about $2,000 or more than one-quarter of income on average. Since only about 35 percent of poor families receive child support, child poverty could be greatly reduced if a higher percentage of poor families were to receive $2,000 or so in child support payments.

Another way to increase the anti-poverty impact of child support, even without increasing child support payments by fathers, would be to ensure that all child support payments are given to mothers and children. In the case of families that have been on welfare in the past, the government keeps nearly $1 billion per year of the support that fathers pay as reimbursement for welfare costs. Ending this “welfare cost recovery” focus of the current child support system would substantially reduce child poverty. It would also eliminate a major disincentive for fathers to pay child support because it would result in even more of the fathers’ payments going to their children.

States could go further and build on the proven success and cost-effectiveness of the EITC by creating similar financial incentives for payment of child support. For example, states could be encouraged to use funds from their welfare block grant or state dollars to supplement low-income fathers’ child support payments so that fathers would have additional incentive to pay child support and mothers and children would receive more than the amount paid by fathers. The size of the supplemental payment could vary with the father’s income and gradually phase out as income increases.

Congress could increase the capacity of low-income fathers to pay child support by providing employment services and work supports. Many of these fathers do not pay child support regularly because they are unemployed or under-employed and have only a limited income from which to make payments. Welfare reform required and helped more low-income mothers to enter the workplace so they could better support their children. Now states should be encouraged or required to help low-income fathers increase their employment and earnings and to address some of their difficulties with the child support system including child support orders that are high relative to their income and large accumulated child support debts owed to states.

Promote Marriage

As shown above, living in a female-headed family greatly increases a child’s chances of living in poverty, especially if the female-headed family was created by a non-marital birth. Living in a married-couple family, in addition to reducing poverty, confers a host of other benefits on both children and adults. It follows that promoting marriage would be a good way to reduce child poverty and to promote child and adult well-being more generally. Early indications are that the Bush administration and Republicans will emphasize marriage during the reauthorization debate.

According to recent research, about half of the children born outside marriage live with both biological parents at birth. As times goes by, however, these fragile families tend to break apart. Policymakers need to find ways to help these families stay together and move toward marriage. A large-scale study in Minnesota conducted by the Manpower Demonstration Research Corporation produced results that suggest one way to promote marriage. The study found that the combination of work requirements, an income supplement, and work supports such as child care substantially increased the marriage rate. This outcome was caused primarily by the fact that two-parent families that had both the work requirement and work supports were almost 40 percent less likely to separate or divorce than two-parent welfare families that did not have the work requirement and work supports. Because two-parent families participate in food stamps, Medicaid, and cash welfare assistance at a lower rate than single-parent families, even when their incomes are similar, states should eliminate any remaining barriers or eligibility restrictions that apply to two-parent families. The federal government should encourage states to actively seek to serve larger proportions of eligible two-parent families through their welfare programs, thus giving them access to employment services and other work support programs.

Despite the evidence on marriage from the Minnesota experiment and a very small number of similar experiments, little is known about how to promote marriage. Last year, the House (but not the Senate) passed legislation by a vote of 405-18 that would have appropriated about $160 million over five years to fund community-based (including faith-based) projects to help poor and low-income fathers marry the mother of their children, become better parents, and improve their employment and income. Although there are now no programs that have proven they can achieve these goals, the nation is fortunate to have a growing number of programs that are trying. It would be a prudent investment of public funds to stimulate the growth of these programs and to carefully evaluate their effects.

Making Poverty Reduction Official Federal Policy

Prime Minister Tony Blair has recently set the official goal of eliminating child poverty in Great Britain within 20 years. The Blair government plans to achieve this goal, in part, by using the same means the United States is now successfully using to reduce poverty-namely, stressing the importance of work and finding ways to supplement income.

Many liberals in the United States would like to follow Blair’s lead and use the reauthorization debate to prompt a discussion about the federal government’s responsibility for reducing child poverty, both to make the responsibility more explicit and to set a specific target for child poverty. Some liberals have suggested that the federal government should declare its responsibility for poverty reduction and set the goal of reducing child poverty by at least half within 10 years. Most conservatives oppose making poverty reduction an official goal of federal policy both because they reject the federal responsibility for poverty reduction and because they believe that liberals would use the goal as leverage to increase government spending on social programs that may or may not reduce poverty.


Further reductions in child poverty, like those achieved in recent years, can be realized through a mixture of federal and state policies that encourage individual effort and then subsidize work and perhaps marriage. An aggressive pursuit of some of the policies outlined above would constitute an expanded battle against poverty and would in all likelihood have the effect of reducing child poverty even below the remarkably low levels achieved since enactment of the 1996 reforms.

Additional Reading

Bos, Johannes, and others. 1999. New Hope for People with Low Incomes: Two-Year Results of a Program to Reduce Poverty and Reform Welfare. New York: Manpower Demonstration Research Corporation.

Citro, Constance R., and Robert T. Michael, eds. 1995. Measuring Poverty: A New Approach. Washington, D.C.: National Academy Press.

Gennetian, Lisa, and Cynthia Miller. 2000. Reforming Welfare and Rewarding Work: Final Report on the Minnesota Family Investment Program, Volume 2: Effects on Children. New York: Manpower Demonstration Research Corporation.

Jencks, Christopher. 1993. Rethinking Social Policy: Race, Poverty, and the Underclass. HarperTrade, 1993.

U.S. Census Bureau. 2000. Poverty in the United States (Series P60-210). U.S. Government Printing Office.