Statement of Ronald Haskins, Ph.D., Co-Director, Center on Children and Families, Brookings Institution

Testimony Before the House Committee on Ways and Means

July 19, 2006

Chairman Thomas and Members of the Committee:

It has been ten years since the welfare reform law was signed by President Clinton amid predictions of disaster from the left. Thanks to provisions in the legislation itself that provided millions of dollars for research, to an unprecedented level of research sponsored by foundations, to data reported by states to the federal government, and to national data collected and reported on a routine basis by the Census Bureau, a tremendous volume of information bearing on the effects of the legislation has been produced.  In fact, there is probably more information about the effects of the 1996 welfare reform law than any other piece of social legislation enacted in recent decades.

The most important reform was the replacement of the old Aid to Families with Dependent Children (AFDC) program with the Temporary Assistance for Needy Families (TANF) program.[1] The research on TANF yields a coherent picture that will almost certainly stand the test of time. With its emphasis on work, time limits, and sanctions against states that did not place a large fraction of its caseload in work programs and against individuals who refused to meet state work requirements, TANF was a historic reversal of the entitlement welfare represented by AFDC.  If the 1996 reforms had their intended effect of reducing welfare dependency, a leading indicator of success would be a declining welfare caseload. TANF administrative data reported by states to the federal government show that caseloads began declining in the spring of 1994 and fell even more rapidly after the federal legislation was enacted in 1996. Between 1994 and 2005, the caseload declined about 60 percent. The number of families receiving cash welfare is now the lowest it has been since 1969, and the percentage of children on welfare is lower than it has been since 1966.[2] Although it is often reported in the media that cash welfare caseloads increase during economic recessions and decline during recoveries, this claim is mostly false. In the forty-one years between 1953 and 1994, the number of families on AFDC declined in only five.[3] Only once – between 1977 and 1979 – did the caseload decline (by about 2 percent) two years in a row. By contrast, 2005 was the eleventh year in a row that the caseload declined. Clearly, we are in a new era of welfare use.[4]

Although caseload decline is an important outcome measure of the 1996 reforms, how families fare after leaving welfare is of great importance. The next reasonable test of welfare reform, then, is whether mothers leaving welfare are working. Again, there is abundant information to answer this question. In fact, three lines of evidence can be aligned to produce a consistent story. The first set of evidence is dozens of welfare-to-work studies conducted since the 1980s. These gold-standard studies almost uniformly show reductions in caseloads and increases in employment attributable to work requirements, as long as the programs included job search requirements.[5] The second line of evidence comes from more than forty state studies conducted since 1996 of adults who left welfare.[6] On average, these studies show that a little less than 60 percent of the adults leaving welfare were employed at any given moment and that over a period of several months or longer about 70 percent held at least one job (although there is good evidence that the share of leavers who were working declined somewhat since the recession of 2001).[7]

A third line of evidence, and the most definitive, is statistics on female employment for the nation as a whole.[8] Census data shows historic changes in employment (defined as any earnings during the year) by single mothers, especially low-income single mothers (Figure 1). From 1993 to 2000 the portion of single mothers who were employed grew from 58 percent to nearly 75 percent, an increase of almost 30 percent. Even more pertinent to assessing the effects of welfare reform, employment among never-married mothers, most of whom join the welfare ranks within a year or two of giving birth, grew from 44 percent to 66 percent.[9] Before 1996 never-married mothers were the ones most likely to be school dropouts, to go on welfare, and to stay on welfare for a decade or more. Yet their employment over this period grew by 50 percent. Employment changes of this magnitude over such a short period for an entire demographic group are unprecedented in Census Bureau records.

So employment of poor mothers heading families has increased dramatically.  But what about their income?  One of the most frequent criticisms of the 1996 reforms was that mothers and children would be destitute.  Members of the House of Representatives and the editorial boards of many of the nation’s leading newspapers who opposed the welfare reform bill, used exceptionally colorful language to describe the afflictions to which the legislation would subject poor families and children.[10]

Census Bureau data for female-headed families in the bottom 40 percent of the income distribution for female-headed families (those below about $21,000 in 2000) show that their pattern of income shifted dramatically between 1993 and 2000.[11] In 1993 earnings accounted for about 30 percent of the income of low-income, female-headed families, while welfare payments, including cash, food stamps, housing, and school lunch, accounted for nearly 55 percent. By 2000 this pattern had reversed: earnings had leaped by an astounding 136 percent, to constitute nearly 60 percent of income, while welfare income had plummeted by over half, to constitute only about 23 percent of income (Figure 2). As a result of the growth in earnings and legislated expansions of the EITC, income from the EITC more than tripled. Thus with earnings and EITC payments leading the way, the total income of these low-income families increased by about 25 percent over the period (in dollars adjusted for inflation).[12]  Even after the recession of 2001, earnings remained above their 1993 level.  The predictions of doom turned out to be wrong.

The pattern is clear: earnings up, welfare down. This is the very definition of reducing welfare dependency. Most low-income mothers heading families appear to be financially better off, although work expenses and Social Security taxes consume part of their earnings,[13] because the mothers earn more money than they received from welfare. Taxpayers continue making a contribution to the well-being of these families through the EITC and other work support programs, but the families earn a majority of their income. This explosion of employment and earnings constitutes an enormous achievement for the mothers themselves and for the nation’s social policy.

Members of this committee will recall that one of the most frequently used arguments against the welfare reform legislation was that it would throw over a million children into poverty.  This claim was based on a study conducted by the Urban Institute at the request of the President Clinton’s Department of Health and Human Services.[14]  It is doubtful that any measure of the condition of the nation’s children receives more attention than the poverty rate.  Thus, the impact of welfare reform on poverty has great substantive and political importance.

Although child poverty dropped during the 1960s, after the early 1970s it gradually drifted upward, primarily because an increasing percentage of American children were being reared in female-headed families, the family type with the lowest work output and the highest poverty rate.[15] However, between 1994 and 2000, child poverty fell every year and reached levels not seen since 1978 (Figure 3). In addition, by 2000 the poverty rate of black children and of children in female-headed families was the lowest it had ever been. The percentage of families in deep poverty, defined as half the poverty level (about $7,000 for a mother and two children in 2000), also declined until 2000, falling about 35 percent during the period.[16] Even after four consecutive years of increasing child poverty between 2001 and 2004, poverty was still 20 percent below it 1993 peak.

A special analysis by the Department of Health and Human Services (HHS) and the Congressional Budget Office provides a clear understanding of the impact of work on poverty rates among families headed by poor mothers. The analysis examined the changing impact of earnings and government taxes and transfer payments on poverty during the 1990s. In 1990 the poverty rate among children in households with an unmarried female head before any taxes or government transfers was 50 percent. But in 1999 this poverty rate (which might be thought of as the market poverty rate, because it is computed without regard to government taxes or benefits) fell by 20 percent, to a little over 39 percent. Virtually all this decline in poverty is attributable to increased employment and earnings by mothers during the 1990s.[17]

The analysis then added various combinations of government transfers and taxes to market income among these unmarried mothers. One of the analyses shows that in 1990, before welfare reform, the combination of all government non-tax transfers such as cash welfare and food stamps reduced poverty by about 12 percentage points, from around 50 percent to a little more than 37 percent. Although the market poverty rate in 1999 was 11 percentage points lower than in 1990, government cash and in-kind transfers in 1999 still reduced poverty by almost an additional 10 percentage points, to a little under 30 percent.

The final step in the analysis was to examine the effect on poverty when income from the EITC was added and federal tax payments were subtracted from income. Not surprisingly, given the relatively low level of work and earnings in 1990, adding the EITC increased income only enough to reduce poverty by less than 1 percentage point. By contrast, in 1999 adding the EITC to income and subtracting federal taxes reduced the poverty rate by 4.50 percentage points. Based on total income, including both market earnings and all government taxes and transfers, poverty among single mothers and children was therefore 36.8 percent in 1990, compared with 25.1 percent in 1999, a decline of nearly one-third. If the 1999 poverty rate had been the same as the 1990 rate, nearly 4.2 million more single mothers and children would have been poor. The prediction that welfare reform would lead to major increases in child poverty was flawed.

Promoting child well-being was a major goal of all participants in the 1995-96 welfare reform debate. Republicans argued that increased work by mothers on welfare would lead to positive impacts on children because mothers would be setting an example of personal responsibility, would impose schedules and order on chaotic households, and would increase family income. By contrast, many Democrats thought that welfare reform would be disastrous for children. They believed that mothers would not be able to find and maintain work, would hit time limits or be hit by sanctions, and would experience serious declines in family income, driving them into destitution. Perhaps the most frequent charge, based on a reputable study by the Urban Institute, was that welfare reform would throw a million children into poverty.[18] There were also predictions that more children would be removed from their parents and placed in the child protection system.

Several types of research evidence are now available to make informed judgments about which predictions have come true. A reasonable place to begin is with broad survey data on the well-being of American children. As we have seen, poverty not only did not increase but actually declined every year between 1994 and 2000, with black child poverty reaching its lowest level ever. Although poverty increased after 2000, it remained well below its 1994 level. So great was the decline in poverty that, as Paul Jargowsky and Isabel Sawhill show, the number of neighborhoods with concentrated poverty fell precipitously, as did the number of neighborhoods classified as underclass because of the concentration of poverty and the high frequency of problems such as school dropout, female-headed families, welfare dependency, and labor force dropout by adult males. The authors conclude that the 1990s were a “remarkable decade in which substantial progress was made.”[19]

Besides measures of poverty and underclass neighborhoods, a host of additional measures of child well-being is available. One of the best collections of national indicators is the Child and Youth Well-Being Index (CWI), published annually by Ken Land of Duke University with support from the Foundation for Child Development. The Land index reports twenty-eight key indicators of child well-being; these indicators are based on nationally-representative surveys, most of which have been administered annually since 1975. The overall index shows a clear pattern of changes over the past three decades. After a few years of modest changes in no clear direction, in 1982 the index showed a decline in well-being that lasted almost continuously until 1995. Since 1995, the index shows an improvement in well-being in almost every year, more than recovering the ground lost in the 1980s and early 1990s. Using 1975 as the base year, the index descended to about 75 percent of its original level by 1995. Since then, it has increased by about 30 percentage points, to about 5 percent above its 1975 level. The CWI is organized into seven domains, each of which measures an important dimension of child well-being such as economic, health, safety, and emotional and spiritual well-being. Most of these domains reflect the overall CWI pattern of continuous increases since 1995. Only the health domain shows a decline, and this only because child obesity increased dramatically.  Other measures of child health showed improvement. As Land concludes, “Children are faring better in recent years.”[20]

A similar conclusion is reached by the Federal Interagency Forum on Child and Family Statistics.[21] The forum presents many of its indicators separately for various income and ethnic groups. In nearly every case in which indicators are presented in this way, low-income and minority children reflect the pattern of general improvement, often showing even greater improvement than white children and children from wealthier families. Similarly, Donald Hernandez of the State University of New York has studied ethnic differences in the Land index. Compared with the huge differences in the early 1990s between white children and both black and Hispanic children, both minority groups closed the gap with whites by about one-third over the last decade, both groups narrowing the gap on six of the seven index domains.[22]

Another feared effect of welfare reform was an increase in the number of children taken from their destitute families by the foster care system.[23] By the mid-1990s, the national foster care caseload had increased every year for fourteen consecutive years, rising from 262,000 in 1982 to 507,000 in 1996. The caseload then increased over the next three years at approximately the same rate as in previous years. Then in 2000, for the first time in two decades, the foster care caseload began to decline and has declined every year since then, falling from 567,000 in 1999 to 518,000 in 2004, a fall of almost 10 percent.[24] Similarly, the incidence of child maltreatment of all types has declined in most years since 1993, falling by over 20 percent between 1993 and 1999, before rising somewhat beginning in 2000. However, the rate in 2001 was still well below the rate of the early 1990s.[25]

In addition to these broad indicators of child well-being, there is a growing body of scientific research on the direct effects of welfare reform on children, including gold standard studies based on random assignment. Most of these studies were initiated before the 1996 legislation, but nonetheless examined the effects of work programs similar to those mounted by states both before and after the 1996 reforms. Pamela Morris of MDRC and her colleagues have reviewed the impacts on young children of seven random-assignment demonstrations, including thirteen employment programs in the United States and two in Canada, yielding data on 30,000 low-income children.[26] Morris and her colleagues confined their review to children who were between the ages of two and nine when the programs began (between four and fifteen at the point of final data collection). Five results are notable: positive impacts on school achievement were evident among children whose mothers were in certain work programs; impacts were confined to children age five and under at the beginning of the studies; impacts were confined to work programs that increased family income by providing earnings supplements; impacts faded after three years; and positive impacts on school achievement were related to attendance at center-based child care programs during the preschool years. These results are broadly consistent with the large literature on effects of maternal employment, including the finding that when mothers’ work leads to increased family income, young children often show modest improvement on measures of social and intellectual development.[27]

A similar review by Lisa Gennetian of MDRC and her colleagues on the effects of work programs on adolescents complements the Morris review.[28] The Gennetian review is based on pooled data from seventeen random-assignment programs. The nearly 6,600 participating children were between ages ten and sixteen at the beginning of the studies; at the point of final data collection they were between twelve and eighteen. Averaged across all the experiments, mothers participating in work programs, compared with mothers in the control programs, rated their children as performing below average in school. In addition, children in the experimental programs were slightly more likely to repeat a grade and to be enrolled in special education classes. They were not, however, more likely to be expelled from school, to drop out, or to have had (or have fathered) a baby. Data from the individual studies provide some evidence that the negative effects on school performance seemed to be concentrated in adolescents with younger siblings, suggesting that the poor school outcomes might be associated with early assumption of adult responsibilities because working mothers shared child care with their older children. Similar negative effects of maternal employment on adolescents have been noted by several other researchers and reviewers.[29] Although these effects are modest and were not found in all of the individual studies, there is nonetheless reason for concern. Gennetian and her colleagues call for “more investigation rather than . . . an immediate policy response.”[30]

Taken together, the survey and experimental information available on the well-being of poor, low-income, and minority children in the decade following welfare reform does not justify the fears expressed by liberals. With some exceptions, measures of child well-being show that children, and especially poor and minority children, have generally lived under improved conditions and have shown modest gains on indicators of development since 1996. On the other hand, the hopes of conservatives about the impact of welfare reform on children have not been vindicated either. High-quality studies of welfare reform show that preschool children of families participating in welfare-to-work studies may experience modest gains in their development and behavior, but equally good studies show that adolescents experience modest problems in school performance. From the perspective of one decade, it does not seem likely that welfare reform will alleviate the serious lags in development and performance shown by children from poor and minority families. Direct interventions with these children will be necessary if the nation is to close the ability and education gaps between them and more advantaged children.

Although welfare reform is a major cause of the dramatic rise in earnings and the decline in welfare dependency and child poverty, at least two other factors account for the improving financial well-being of female-headed families. First, the economy of the 1990s was exceptionally strong. By 2000 almost 137 million Americans had jobs, up by more than 16 million since 1993. Before the recession hit in 2001, 64.4 percent of all noninstitutionalized adults in the United States were working, the highest share ever. Not surprisingly, the unemployment rate fell from 6.9 percent in 1993 to 4.0 percent in 2000, the lowest in several decades.[31] Sophisticated statistical studies have been conducted by economists to determine the relative contribution of the economy, of welfare reform, and of other factors to the dramatic rise of work and earnings by low-income mothers heading families.[32] These studies all show that both welfare reform and the booming economy are important, but there is little agreement about the relative contributions of each factor. However, previous economic booms did not lead to either the reduction in welfare rolls or the increase in work by low-income mothers heading families that were seen in the 1990s.[33] Without welfare reform cajoling and where necessary pushing mothers into the labor force, a growing economy would have had a more modest effect on the employment and earnings of these mothers, as was in fact the case during all previous economic expansions.

Second, beginning more than a decade before the 1996 reforms, the federal government made existing benefit programs friendlier to low-income working families and created entirely new programs designed to help working families. These actions include expansions of child care, creation of the child tax credit, changes in the standard deduction and the personal exemption in the income tax code, changes in Medicaid, and above all several expansions of the EITC. Two studies by nonpartisan and highly respected congressional agencies – the Congressional Budget Office (CBO) and the Congressional Research Service (CRS) – provide an idea of the magnitude of these changes.

At the request of the Ways and Means Committee in 1997, CBO undertook a study to determine whether federal policy changes between 1984 and 1998 had resulted in more support for low-income working families. CBO examined several major entitlement programs that help working families, including child care, the EITC, Medicaid, and the child tax credit. Taken together, we can label these and similar programs the nation’s work support system, because the programs provide financial and in-kind support to poor and low-income working families. CBO calculated the benefits that would have accrued to low-income working families from the work support system under 1984 law and compared that level of support to the level under 1999 law. Because every work support program examined by CBO had been expanded or created since 1984, the analysis was expected to show an increased commitment by federal policymakers to low-income working families. But it is fair to say that even experts were surprised by the finding that if the work support system had remained as it had been in 1984, working families in 1999 would have received only around $6 billion in government work support benefits (Figure 5). By contrast, the 1999 version of the work support system – that is, the one that actually existed in 1999 – provided nearly $52 billion in support to working families. In other words, the expansions in the work support system after 1984 resulted in working families receiving $46 billion more in cash and other benefits than they would have received if Congress and a series of presidents had not expanded the work support programs. It would be difficult to exaggerate the extent to which the nation’s social policy to help low-income families has shifted from one that provided most of its benefits to families dependent on welfare to one that provides enormous benefits to working families.[34]

The second study, based on information computed by CRS, strengthens the CBO conclusion. Whereas the CBO study provides an estimate of changes in aggregate federal spending on work support programs, the CRS data can be used to compare the financial work incentive for a typical mother with two children on welfare in a typical state (Pennsylvania) in 1986 and 1997.[35] For years, a major charge against the welfare system was that it posed a substantial disincentive to work because families that accepted jobs could be worse off working than on welfare. A mother and two children in a typical state in 1986 received about $8,970 in cash welfare and Food Stamps (all figures are in constant 1997 dollars). If the mother worked and earned $8,000, her welfare income would fall drastically, to $1,900. She would also pay nearly $1,200 in federal taxes but would gain about $540 from the EITC. Thus for working full-time she would have net income of about $9,275, or about $350 more than if she had stayed on welfare. In addition, both the mother and children would lose their Medicaid coverage, the insurance value of which would be around $3,000, after nine months, and the mother would get very modest if any government help paying for child care. Clearly, a mother who elected to stay on welfare rather than accept a low-wage job in 1986 would be making a financially rational decision. By contrast, because of the broadening of the work support system and changes in welfare laws, by 1997 this same mother with a $10,000-a-year job (roughly equivalent to $8,000 in 1986) would have net income of around $15,350, or $7,550 more than the $7,800 she would have received if she had stayed on welfare. The EITC alone was worth an additional $3,000 in cash, and changes in federal income tax law had removed the mother entirely from paying income tax. Further, the mother would have Medicaid coverage for one year, and the children would be covered as long as the mother had low income. Finally, there was much more money available for child care in 1997 than in 1986. All in all, the work support system had made work a more attractive option for welfare mothers in 1997 than in 1986.[36]  Given that the EITC is pegged to inflation, that funds for child care have expanded dramatically since 1996,[37] and that the child tax credit was made partially refundable in 2001, it seems likely that the work support system is even more generous today than it was in 1997. In any case, at the time the 1996 reforms were enacted, as well as today, the work support system provides compelling financial incentives for mothers to leave welfare even for low-wage jobs.

The positive impacts of the 1996 reforms on income, earnings, and poverty have been pervasive and, in some cases, profound. However, no policy produces all benefits and no costs. Although the 1996 law did not produce the failures predicted by its critics, it nonetheless has created challenges that states and the federal government should address. In my view, the most important of these challenges is the finding that there is a group of mothers at the bottom of the income distribution who appear to be floundering under the new and more demanding welfare system.  Generally, these are mothers who live without another adult in their household and who do not have income from cash welfare, from employment, or from unemployment insurance. In the past, these troubled parents could stay on welfare for many years. Under the old AFDC program, the average length of spells for adults on the rolls at any given moment was twelve years. It would be naïve to believe that all these welfare-dependent parents were suddenly capable of finding and retaining jobs for $7 or $8 an hour. A demanding welfare system requires at least some minimum level of competence and motivation, and not all parents have these minimum levels.

There are several types of evidence that a number of mothers are in fact floundering. Surveys show that about 60 percent of the mothers who leave welfare are working at any given moment and that around 70 percent have held at least one job since leaving welfare.[38] The 40 percent who do not work regularly raise some concern, but the 30 percent who have not worked at all since leaving welfare raise even more serious concern. States frequently use sanctions and thirty-six states have policy that allows them to completely terminate cash benefits for rule infractions. At least one study found that mothers who were sanctioned off the rolls had characteristics that make it less likely they will be able to get and hold a job. More specifically, they are less likely to have a high school degree or job experience and more likely to have substance addictions, mental health problems, or three or more children than other welfare mothers.[39] Also of concern are poor mothers heading families who are financially worse off since welfare reform passed. Kasia Murray and Wendell Primus have compared Census income data for mothers for the 1993-96 and the 1996-2000 periods and found that that mothers in the bottom 10 percent of single earners actually lost income during the latter period.[40] These findings are placed in a broader context by Rebecca Blank and Robert Schoeni from the National Poverty Center at the University of Michigan. Blank and Schoeni, using data from the Census Bureau’s Current Population Survey, compared the change in income between the 1992-95 period (before TANF) and the 1997-2000 period (after TANF). Controlling for factors such as family size and inflation, they plotted income for two groups: all families with children and families with children without both parents present. Blank and Schoeni find that all but the bottom 2 percent of families with children had improved their income in the late 1990s relative to the mid-1990s. Even in the case of children living outside a two-parent family, 92 percent of families improved their income.  However, the bottom 8 percent declined.

Blank and Schoeni explicitly tied their analysis to welfare reform by comparing states with strong cash work incentives (which allowed mothers who went to work to retain relatively more of their welfare benefit) and strong penalty incentives (strict time limits and strong sanctions) and found that both cash and penalty incentives were associated with higher income. The authors conclude that “it is the more lenient states with softer penalties where children’s income seems to have grown the least.”[41] Although the authors interpret their findings as “good news,” their work is similar to Murray and Primus’s in showing that there is a group of mothers at the bottom – in this case about 8 percent of the distribution of female-headed families – that is worse off now than before welfare reform. This finding is reinforced by Census Bureau data analyzed by Richard Bavier of the Office of Management and Budget.  Bavier finds a disconcerting increase in the number of mothers in the bottom fifth of income for female heads of families who report zero earnings and zero income from cash welfare (ignoring SSI). The number of mothers in this category increased in every year between 2000 and 2004, jumping by 60 percent over the period.[42]

Several other researchers, including Robert Moffitt and Katie Winder at Johns Hopkins; Pamela Loprest, Sheila Zedlewski, and others at the Urban Institute; Sandra Danziger and Sheldon Danziger of the University of Michigan; and Robert G. Wood and Anu Rangarajan at Mathematica Policy Research report similar findings on increased hardship among mothers who leave welfare, live in a household without another adult, and do not have earnings.[43] The studies by Wood and Rangarajan and the Danzigers and their colleagues are especially interesting because they both have many years of longitudinal data (data collected on the same subjects over time) on mothers who had been on welfare. Wood and Rangarajan followed a representative group of 2,000 recipients who had received welfare in 1997 or 1998 in New Jersey. Although the group that was off welfare and employed increased from about one-third to one-half over the fifty-four-month follow-up period, the group of greatest concern – those who were off welfare but without a job – was consistently a little more than one quarter of the sample. Of this group, about 60 percent had other sources of income, including SSI, unemployment compensation, a working spouse or partner, or recent employment. Thus the mothers who were the least financially stable constituted about 40 percent of those who were off welfare and unemployed, or around 11 percent of the total sample.

All the evidence reviewed above, showing that mothers and children at the bottom of the distribution experience hardship, is based on income data. Surprisingly, consumption data provide a different picture. In studies using two nationally representative data sets, Bruce Meyer and James Sullivan show that the material conditions of low-income mothers, as measured by their consumption, improved somewhat after welfare reform.[44] On the other hand, a large part of the additional consumption in the late 1990s appears to be related to work. More specifically, the mothers spent more on housing, food away from home, and transportation. Additional housing costs could well be explained by the fact that the federal housing programs in which many of these mothers participate charge families 30 percent of their income, with the remainder of the family’s rent being paid by the government. If mothers earn additional money, they must pay 30 percent of it on housing: in effect, federal housing policy all by itself imposes a 30 percent tax on increased earnings. Additional spending on food away from home and transportation could also be associated with mothers working and needing to use some of their increased earnings to get to work and to eat out because of time pressures.[45]

Evidence on the well-being of mothers and children can also be gleaned from information on food consumption. Christopher Jencks, one of the major critics of the 1996 reforms, and his colleague Scott Winship conducted extensive analyses on the Food Security Supplement to the Current Population Survey for the years 1995-2001. Based on twenty-eight questions related to food security, Jencks and Winship conclude that single mothers had fewer problems related to food in 2001 than in 1995, the last year before welfare reform. Further analyses shows that, although the number of low-income mothers receiving welfare between 1995 and 1999 fell from 58 percent to 29 percent, food-related problems dropped dramatically. The decline in food problems leveled off in the 1999-2001 period, but food problems in 2001 were still substantially below the level of problems reported in 1995. Similarly, based on the Department of Agriculture’s definition of food security, the percentage of food-insecure female-headed families declined from around 31 percent in 1995 to about 27 percent in 1999, as the welfare rolls were declining rapidly. Even during the period following the mild recession of 2001, the percentage of food-insecure families did not increase significantly, remaining below the 1995 level. The authors conclude that “single mothers’ material standard of living probably improved more during [the economic expansion of the 1990s] than during earlier ones.”[46] In an op-ed piece published in the Christian Science Monitor, the authors state flatly that their study of food problems led them to conclude that “welfare reform worked.”[47]

Income data thus suggest that there was a group of single mothers, comprising perhaps 10 percent of all single-mother families that had been on welfare, who were worse off following welfare reform. Data based on consumption and on food insecurity tend to offset this conclusion, although even here there is some evidence of problems at the bottom of the distribution. On balance, it seems prudent to conclude that scholars should examine this problem in much greater detail and search for solutions that will help mothers hold jobs. Given the research reviewed above showing an increase in the number of poor mothers with no obvious sources of income, more public and private funds should be devoted to conducting research and demonstration programs to determine how these floundering mothers can be helped. The trick will be to maintain a demanding welfare system that strongly discourages welfare dependency while simultaneously allowing states, counties, and cities enough flexibility to identify and help these mothers. Some mothers may never be able to achieve steady employment. Welfare programs should figure out how to help them without reducing the pressure on more capable mothers to leave welfare for work or to avoid welfare in the first place.

Compared to any major change in social policy in the last several decades, I think it fair to conclude that welfare reform stands out as federal legislation that actually met its goals.  The entire political spectrum was in agreement that mothers on welfare should become self-sufficient. All but the left-most part of the spectrum agreed to support legislation that terminated entitlement cash welfare and replaced it with a system that required work. The data summarized above show that poor mothers on welfare responded exactly as they were expected to do – they went to work in droves. In addition, as Republicans predicted, they increased their family income and reduced the poverty rate of their children in the process. They left welfare for work, but government continues to support their efforts through child care subsidies, health insurance, food stamps, and above all, the Earned Income Tax Credit. There’s something here for everybody to like:  both more work and lots of government support – except now the bulk of government support is for those working, not those avoiding work. The results of major changes in public policy rarely work out this well.

Now the question is:  What’s next? The obvious part of the answer is that states must continue to aggressively implement the work requirements in the TANF program. This Committee wisely led the way to fixing a glitch in the 1996 legislation that weakened the work requirement in the 1996 law. The first priority of this committee should be to ensure that states aggressively implement the new requirements. A second priority should be for this committee to work with the Agriculture Committee and the Banking Committee to strengthen the work requirements in the Food Stamp program and create strong work requirements for able-bodied adults in housing programs. If work requirements are successful in one program, I can see no reason why they cannot be successful in other programs.

Beyond these obvious next steps, I think this committee has already taken the most important action that holds great promise for further reductions in poverty and improvements in child development and well-being; namely, stimulating a national marriage movement. Years of research on poverty have convinced me that there are only three ways to reliably reduce poverty:  economic growth, increased work, and increased marriage rates.  Unfortunately, primarily because of low and often declining wages at the bottom of the income distribution, economic growth is less effective than in the past at reducing poverty.[48] However, the nation’s experience with increasing work levels following the 1996 welfare reforms shows unequivocally that increased work by mothers heading families drives down the poverty rate.  Now comes marriage.  Work that we have undertaken at Brookings provides solid evidence that increasing marriage rates to the level the nation enjoyed in 1970 would reduce poverty by almost 30 percent.[49] In addition, as shown in a recent volume of the Future of Children, published by Brookings and Princeton University, the academic world is in almost unanimous agreement that increasing marriage rates would be good for children.  This committee should provide strong oversight of its recent legislation that provides $150 million per year to stimulate healthy marriage and responsible fatherhood programs around the nation. In addition, the Committee should keep a close eye on the ground-breaking experimental research the Department of Health and Human Services is funding to test marriage education and other approaches to strengthening marriage.  If some of these programs are successful, the Committee should make funds available to expand them throughout the nation.  I believe the evidence strongly supports the view that if we can increase the nation’s marriage rates, especially among poor and minority parents, the parents themselves, children, and the nation will greatly benefit.  Not least among these benefits will be a declining need for government welfare programs.

1. A remarkable characteristic of the 1996 welfare reform law was its breadth.  Here I examine results only for the new TANF program.  But in a forthcoming book (Ron Haskins, Work Over Welfare: The Inside Story of the 1996 Welfare Reform Law, Washington: Brookings, 2006), I review evidence on the effects of the law’s provisions on family composition, Supplemental Security Income (SSI) for Children, SSI for drug addicts and alcoholics, public benefits for noncitizens, child care, and child support enforcement.  In each of these social policy domains, the 1996 law has had effects that range from substantial to dramatic.  Among the more notable are a substantial reduction in the number of children with mild disabilities receiving SSI; a complete end to SSI benefits for drug addicts and alcoholics; an unprecedented reduction in the number of noncitizens receiving TANF, food stamps, SSI, and Medicaid; a doubling of funds for child care and extensive use of saving from welfare payments to pay for child care; and a doubling of child support payments and a substantial increase in paternity establishment.  The original estimate of savings from the 1996 law by the Congressional Budget Office was $54 billion over seven years (Congressional Budget Office, “Federal Budgetary Implications of the Personal Responsibility and Work Opportunity Reconciliation Act of 1996,” CBO Memorandum, December 1996).  Although there are technical problems with trying to estimate the accuracy of the CBO forecast, I have no doubt that the actual savings were several billion above the CBO estimate.

[2]. Because the population increases every year, data on the number of children or families on AFDC or TANF over time can be misleading. Both the number of children and the percentage of all U.S. children on AFDC or TANF fell dramatically after 1994. The percentage of children on AFDC peaked in 1993 at 14.3 percent. It fell every year thereafter until reaching 5.3 percent in 2002. See House Committee on Ways and Means, 2004 Green Book, pp. 7-31. By way of comparison, in 1970 the percentage of children on AFDC was 8.8 percent, more than 65 percent higher than in 2002 (see Annual Statistical Supplement, Social Security Bulletin 13-11700 (Department of Health and Human Services, 2005), table 9.G1.

[3]. Department of Health and Human Services, Administration for Children and Families, “Cash Assistance for Needy Families, Aid to Families with Dependent Children (AFDC) and Temporary Assistance to Needy Families (TANF), Average Monthly Families and Recipients for Calendar Years 1936--2001,” May 10, 2002, available at www.acf.dhhs.gov/news/stats/3697.htm.

[4]. House Committee on Ways and Means, 2004 Green Book, section 7, pp. 27-37; Department of Health and Human Services, Indicators of Welfare Dependence: Annual Report to Congress, 2003 (Government Printing Office, 2003), appendix A, table TANF 1; for welfare caseloads information, see the Administration for Children and Families, www.acf.hhs.gov/newstat2.shtml.

[5]. Judith M. Gueron and Edward Pauly, From Welfare to Work (New York: Russell Sage, 1991); Jeffrey Grogger, Lynn Karoly, and Jacob Alex Klerman, Consequences of Welfare Reform: A Research Synthesis (Santa Monica, Calif.: Rand, 2002).

[6]. House Committee on Ways and Means, 2000 Green Book, pp. 1471-74, 1500-10. See also Gregory Acs, Pamela Loprest, and Tracy Roberts, Final Synthesis Report of Findings from ASPE’s ‘Leavers’ Grants (Washington: Urban Institute, 2001), pp. 23-47.

[7]. Acs, Loprest, and Roberts, Final Synthesis Report, chap. 3.

[8]. See the Bureau of Labor Statistics, http://stats.bls.gov/cps/home.htm, for labor force statistics from the Current Population Survey. All the measures discussed in this section are point-in-time measures.

[9]. Gary Burtless of the Brookings Institution, unpublished calculations using U.S. Bureau of Labor Statistics data.

[10].  Democrats on the floor of the House referred to the Republican bill as “extreme,” “harsh,” “cruel,” “abusive,” and “mean-spirited;” they also said the bill “attacked children,” “lashed out at children,” and “punished children;” Bob Herbert of the New York Times said the bill conducted a “jihad” against the poor, that it “makes war on the kids of this country,” and that it would “deliberately inflict harm” on children and the poor.  See Ron Haskins, Work Over Welfare, Chapter 8.

[11]. These figures are from unpublished tables of demographic and economic characteristics of female family heads, distributed by income quintile, that are prepared by Richard Bavier of the Office of Management and Budget using the Census Bureau’s Current Population Survey. Bavier makes these tables available to anyone who requests them. All figures are given in 2004 dollars.

[12]. A study by two other researchers, using longitudinal data from the Survey of Income and Program Participation, shows the same pattern of increased work and earnings and declining poverty among families leaving welfare. See June O’Neill and M. Anne Hill, “Gaining Ground, Moving Up: The Change in the Economic Status of Single Mothers under Welfare Reform,” Civic Report 35 (Manhattan Institute Center for Civic Innovation, 2003).

[13]. Jeffrey Grogger and Lynn A. Karoly, Welfare Reform: Effects of a Decade of Change (Harvard University Press, 2005), p. 156.

[14]. Sheila Zedlewski and others, “The Potential Effects of the Budget Reconciliation Bill on Family Incomes” (Washington:  Urban Institute, 1995).

[15]. David J. Eggebeen and Daniel T. Lichter, “Race, Family Structure, and Changing Poverty among American Children,” American Sociological Review, 56 (1991): 801-17.

[16]. Bernadette D. Proctor and Joseph Dalaker, Poverty in the United States: 2002 (Census Bureau, 2002), table 5; Douglas J. Besharov, ed., Family and Child Well-Being after Welfare Reform (New Brunswick, N.J.: Transaction, 2003), figure 4; Bureau of the Census, Poverty and Health Statistics Branch, Current Population Survey, Annual Social and Economic Supplements, December 14, 2005, table 3, available at www.census.gov/hhes/www/poverty/histpov/hstpov3.html.

[17]. This analysis retains the same thresholds as the official poverty index (about $13,300 for a family of three in 1990) but uses a broader definition of income than the official measure. House Committee on Ways and Means, 2004 Green Book, table H-21.

[18]. Zedlewski and others, “The Potential Effects of the Budget Reconciliation Bill on Family Incomes.”

[19]. Paul A. Jargowsky and Isabel V. Sawhill, The Decline of the Underclass, policy brief, Center on Children and Families (Brookings, 2006), p. 6.

[20]. Kenneth Land, Child and Youth Well-Being Index (CWI), 1975-2004 with Projections for 2005, available at www.soc.duke.edu/~cwi/.

[21]. Federal Interagency Forum on Child and Family Statistics, America’s Children:  Key National Indicators of Well-Being 2005 (Government Printing Office, 2005).

[22]. Donald J. Hernandez and Suzanne E. Macartney, “Measuring Social Disparities: A Modified Approach to the Index of Child Well-Being (CWI) for Race-Ethnic, Immigrant-Generation, and Socioeconomic Groups with New Results for Whites, Blacks, and Hispanics,” paper prepared for the forum, “Review of the Child Well-Being Index,” Brookings Institution, May 10, 2006.

[23]. House Committee on Ways and Means, Contract with America: Overview, Serial 104-20, January 5, 10, 11, and 12, 1995, pp. 58-93.

[24]. House Committee on Ways and Means, 2004 Green Book, section 11, table 23; Department of Health and Human Services, Administration for Children and Families, Children’s Bureau, “Trends in Foster Care and Adoption, FY2000-FY2004,” September 2005, available at www.acf.hhs.gov/programs/cb/stats_research/afcars/trends.htm.

[25]. House Committee on Ways and Means, 2004 Green Book, section 11, figure 1.

[26]. Pamela Morris, Lisa A. Gennetian, and Greg J. Duncan, “Effects of Welfare and Employment Policies on Young Children: New Findings on Policy Experiments Conducted in the Early 1990s,” Social Policy Report 19 (2005): 3-22.

[27]. For reviews of the literature on the effects on children of working mothers, see E. Harvey, “Short-Term and Long-Term Effects of Early Parental Employment on Children of the National Longitudinal Survey of Youth,” Developmental Psychology 35 (1999): 445-59; Martha J. Zaslow and Carol A. Emig, “When Low-Income Mothers Go to Work: Implications for Children,” Future of Children 7 (1997): 110-15.

[28]. Lisa A. Gennetian and others. How Welfare and Work Policies for Parents Affect Adolescents: A Synthesis of Research (New York: MDRC, 2002).

[29]. Greg J. Duncan and P. Lindsay Chase-Lansdale, “Welfare Reform and Children’s Well-Being,” in The New World of Welfare, edited by Rebecca M. Blank and Ron Haskins (Brookings, 2001); Jennifer L. Brooks, Elisabeth C. Hair, and Martha J. Zaslow, “Welfare Reform’s Impact on Adolescents: Early Warning Signs” (Washington: Child Trends, 2001).

[30]. Gennetian and others, How Welfare and Work Policies for Parents Affect Adolescents, p. 22.

[31]. Census Bureau, Statistical Abstract of the United States, 2004--2005, p. 371.

[32]. Rebecca M. Blank, “Declining Caseloads/Increased Work: What Can We Conclude about the Effects of Welfare Reform?” Economic Policy Review 7 (2001): 25-36; James P. Ziliak and others, “Accounting for the Decline in AFDC Caseloads: Welfare Reform or the Economy?” Journal of Human Resources 35 (2000): 570-86.

[33]. It is difficult to discern a clear effect of the 1991 recession on the welfare caseload. When the recession hit, the caseload had already grown in five of the previous six years, rising from 3.69 million families in 1985 to 4.38 million families in 1991, during the healthy economy of the 1980s. The caseload continued to increase over the next three years, rising from 4.38 million families in 1991 to 5.05 million in 1994, before its prolonged decline during the welfare reform period. By contrast, as we have seen, the welfare caseload continued to decline throughout and following the recession of 2001. Between 1983 and the peak employment year of 1991, the employment-to-population ratio for never-married mothers rose from 34.5 to 44.0, or by 9.5 percentage points. During the expansion of the 1990s, the comparable ratio rose from 43.4 in 1992 to 65.8 in 2000, an increase of 22.4 percentage points. Clearly a lot more poor mothers went to work during the expansion of the 1990s than the expansion of the 1980s. Employment data is from unpublished calculations by Gary Burtless using Census Bureau data. Caseload information is from the Administration for Children and Families, available at www.acf.hhs.gov/news/stats/newstats2.shtml.

[34]. Congressional Budget Office, Policy Changes Affecting Mandatory Spending for Low-Income Families Not Receiving Welfare (1998).

[35]. House Committee on Ways and Means 1986 Green Book, p. 369; House Committee on Ways and Means 1998 Green Book, p. 408.

[36]. House Committee on Ways and Means, 1996 Green Book, section 8, p. 399, table 3.

[37]. Douglas J. Besharov and Caeli A. Higney, Federal and Sate Child Care Expenditures (1997-2003) (College Park, MD:  Welfare Reform Academy, University of Maryland, May 2006).

[38]. House Committee on Ways and Means, 2004 Green Book, pp. L24-L31.

[39]. Andrew J. Cherlin and others. “Operating within the Rules: Welfare Recipients’ Experiences with Sanctions and Case Closings,” Social Services Review 76 (2002): 387-405.Thirty-six states have a policy that allows them to completely terminate cash benefits for rule infractions. See Department of Health and Human Services, Temporary Assistance to Needy Families Program (TANF): Fourth Annual Report to Congress, May 2002, p. 346. But see Jeffrey Grogger and Lynn A. Karoly, Welfare Reform: Effects of a Decade of Change (Harvard University Press, 2005), pp. 235-36, who conclude that there are too few studies to draw firm conclusions about the effect of sanctions.

[40]. Kasia O’Neill Murray and Wendell E. Primus, “Recent Data Trends Show Welfare Reform to Be a Mixed Success: Significant Policy Changes Should Accompany Reauthorization,” Review of Policy Research 22 (2005): 301-24.

[41]. Ibid., pp. 307, 308.

[42]. These figures are from unpublished tables of demographic and economic characteristics of female family heads, distributed by income quintile, that are prepared by Richard Bavier of the Office of Management and Budget using the Census Bureau’s Current Population Survey. All figures are given in 2004 dollars.

[43]. Robert Moffitt and Katie Winder, “Does It Pay to Move from Welfare to Work? A Comment on Danziger, Heflin, Corcoran, Oltmans, and Wang,” Journal of Policy Analysis and Management 24 (2005): 399--409; Pamela Loprest, “Disconnected Welfare Leavers Face Serious Risks,” Report 7, Snapshots of America’s Families III (Washington: Urban Institute, 2003); Sandra Danziger, Elizabeth Oltmans Ananat, and Kimberly G. Browning, “Childcare Subsidies and the Transition from Welfare to Work,” Family Relations 52 (2004): 219--28; Robert G. Wood and Anu Rangarajan, “What’s Happening to TANF Leavers Who Are Not Employed?” Issue Brief 6 (Princeton, N.J.: Mathematica Policy Research, 2003).

[44]. Bruce D. Meyer and James X. Sullivan, The Well-Being of Single-Mother Families after Welfare Reform, Policy Brief 33, Welfare Reform & Beyond (Brookings, 2005).

[45]. Christopher Jencks, Scott Winship, and Joseph Swingle, “Welfare Redux,” American Prospect 17 (2006): 36-40.

[46]. Scott Winship and Christopher Jencks, “How Did the Social Policy Changes of the 1990s Affect Material Hardship among Single Mothers? Evidence from the CPS Food Security Supplement,” Faculty Research Paper RWP04-027 (Cambridge, Mass.: John F. Kennedy School of Government, 2004), p. 2.

[47]. Scott Winship and Christopher Jencks, “Welfare Reform Worked: Don’t Fix It,” Christian Science Monitor, July 21, 2004, p. 9.

[48] Sheldon Danziger and Peter Gottschalk, Diverging Fortunes:  Trends in Poverty and Inequality (New York:  Russell Sage Foundation, 2006).

[49] Ron Haskins and Isabel Sawhill, “Work and Marriage:  The Way to End Poverty and Welfare” (Brief #28), Washington, DC, Brookings Institution, September 2003; and Adam Thomas and Isabel Sawhill, “For Richer or for Poorer: Marriage as an Antipoverty Strategy,” Journal of Policy Analysis and Management, 2002, 21(4), 587-599.