One million children pushed into poverty: That was the prediction of a widely cited study on the likely effect of welfare reform, released just before Congress passed the landmark legislation in August 1996. The "Personal Responsibility and Work Opportunity Reconciliation Act" gave states unprecedented discretion in setting eligibility standards, established more stringent work requirements for those receiving federally funded benefits, and imposed a five-year lifetime limit on federal benefits for most recipients. When President Clinton signed the bill into law, in the middle of the presidential campaign, several members of his administration resigned in protest. Liberals, advocates for the poor, and poverty researchers were nearly unanimous in their opposition. Even most conservatives, with their talk of group homes and private charities, implicitly conceded that the benefits of welfare reform lay in the long-run behavioral changes that they expected it to produce. In the short run, all agreed, things would have to get worse before they got better.
Fast-forward to 2002, when the welfare legislation was set to expire. That year the welfare rolls were less than half their size in 1996. Female-headed families with children were less likely to receive welfare benefits than at any point in at least 40 years. The magnitude of the change surpassed everyone's predictions. Even more remarkably, however, the official poverty rate among female-headed families with children based on $14,500 for a woman with two children in 2002 had fallen from 42 percent to 34 percent during this period. At no time between 1959 (when the Census Bureau first began tabulating such data) and 1996 had this figure dropped below 40 percent. Welfare reform is now widely viewed as one of the greatest successes of contemporary social policy.
Nonetheless, social scientists who study anti-poverty policy disagree about whether welfare reform really improved living standards among female-headed families with children. Analyses that we recently conducted shed new light on this question. Our research leads us to conclude that welfare reform did not increase material hardship among single mothers and their children and may well have helped reduce it. That was because the reform was part of a larger package of policy changes including a more generous Earned Income Tax Credit (EITC), a higher minimum wage, and expanded childcare subsidies. These policy changes were politically dependent on one another. Congress expanded the EITC in order to help single mothers make ends meet when they took low-wage jobs, and it raised the minimum wage within days of passing welfare reform for the same reason. Likewise, the money that states saved by reducing their welfare caseloads was often used to expand childcare subsidies. Taken together, these changes constitute what we will call the "welfare-reform package." This was a major policy shift, which simultaneously imposed more stringent work requirements on single mothers and provided more government assistance to those who found low-wage jobs.
Even so, the experience of the 1990s allows policymakers to draw only limited inferences about future reforms. Although welfare reform has succeeded in its current form, in our view legislators should now leave it alone, rather than trying to fix what is not broken.
Did Hardship among Single Mothers and Their Children Decline?
If official poverty rates among female-headed families with children declined in the years following welfare reform, why might social scientists question whether hardship declined? One reason is that official poverty rates ignore noncash benefits such as food stamps and Medicaid, the major healthcare program for low-income families. In fact, unmarried female heads were less likely to get food stamps and Medicaid in the late 1990s than in either the 1980s or the early 1990s, even though they often remained eligible. (As we shall see, however, other federal and state programs to help these families became more generous during the late 1990s.)
Another reason for not taking changes in the official poverty rate at face value is that poverty estimates are based solely on income and family composition, and are not adjusted for work-related expenses. The social-policy reforms of the 1990s dramatically increased the proportion of single mothers who worked. Working usually increased these mothers' income, but in most cases it also increased their expenses for childcare, transportation, and clothes. Greater participation in the formal labor market is also likely to have reduced welfare recipients' earnings from off-the-books jobs, as well as the amount of financial help they got from family members and boyfriends. One might therefore expect the shift from welfare to regular employment to increase unmarried mothers' reported income more than it increased their standard of living.
To get around the limitations of income and poverty statistics, we examined changes in material hardship among all families headed by a single mother, using the Food Security Survey, which the Census Bureau conducts every year for the Department of Agriculture. This survey measures families' ability to feed themselves adequately. It asks about problems that range from relatively common (having to "stretch" the food supply) to the very rare (a child went hungry for an entire day).
Our analysis of nearly 50 such measures revealed that food problems among single mothers and their children declined consistently between 1995 and 2000, when the economy was expanding. In April 1995, for instance, 57 percent of single mothers reported having to stretch their food supply at some point during the previous year because their monthly budget came up short. By April 2001, this figure had fallen to 46 percent. The share of single mothers reporting that a child was not eating enough fell from 11 to 8 percent. In short, while insecurity among mother-only families remained remarkably high in April 2001 given that the national unemployment rate was only 4.5 percent that month the improvement since 1995 was sizable.
Also consistent with official poverty rates, most measures of food problems increased as unemployment rose between 2000 and 2002. But these increases were almost always much smaller than the declines between 1995 and 2000, so mother-only families still reported significantly fewer problems at the end of 2002 than they had in 1995. Official poverty statistics, then, appear to predict recent trends in hardship among female-headed families with children quite accurately.
Did the Welfare-Reform Package Reduce Hardship among Female-Headed Families?
The fact that hardship declined during the period in which the federal welfare-reform package was being implemented does not mean that the reforms were responsible for the improvement. These were also years of sustained economic growth. Unemployment fell from 1995 to 2000, and real wages among the worst-paid workers rose. Some scholars think that these changes fully account for the gains that mother-only families experienced. Indeed, it is even conceivable that single mothers and their children might have experienced even bigger declines in poverty and hardship had welfare reform not pushed so many unskilled recipients off the rolls.
To distinguish the effects of social-policy changes from the effects of the economic boom of the 1990s, we first examined whether falling unemployment had had comparable effects on poverty rates among single mothers during earlier business cycles. From the early 1960s to the mid 1990s, economic expansions reduced poverty more among two-parent families than among single-mother families, and recessions harmed two-parent families more. This pattern was widely cited both as evidence that welfare was a poverty trap and as evidence that it was a safety net. Both claims may be correct. It is conceivable that during booms, single mothers did not benefit from economic growth as much as they might have, but during busts they were shielded from rising unemployment. At any rate, this pattern is exactly the opposite of the one we observe during the most recent business cycle, when single mothers gained more than married couples during the boom and lost more during the bust. The implication of this change is that, in the absence of the welfare reform package, falling unemployment would have had less impact on poverty among single mothers and their children in the late 1990s.
The economic boom of the late 1990s was unusual, however, because the wages of America's worst-paid workers rose faster than prices for the first time in a generation. Real hourly wage rates among the bottom 20 percent of workers rose about 11 percent between 1995 and 2000. Mean family income among single mothers in the bottom half of the earnings distribution for all single mothers grew 16 percent during this period. Because single mothers' earnings are only one component of their family incomes, and because not all single mothers work, we estimate that the growth in wages for the worst-paid single mothers can account for only one-fourth of the income gains their families experienced. The difference was largely attributable to increases in employment and hours worked. Social-policy reforms, in conjunction with rising wages paid to low-skilled workers, strengthened single mothers' attachment to the labor force to a greater extent than in previous economic booms.
Why Did the Sky Not Fall?
Predictions of widespread destitution turned out to be wrong for three reasons. First, some of the law's provisions were not as severe as critics assumed. States were supposed to require a rising fraction of their caseloads to participate in work-oriented activities, but any reduction in a state's welfare rolls below the 1995 level counted toward the required target. In 2002, 50 percent of welfare recipients were to be engaged in work activities, but if a state's caseload had fallen by 50 percent since 1995 as was commonly the case then the work requirement was fulfilled. The dramatic decline in welfare receipt greatly eased pressure on states to force the least-able women toward work. Another example is what happened with the seemingly draconian five-year lifetime limit on welfare receipt, which states were allowed to shorten even further. Many did so. But states have considerable flexibility in determining who is subject to time limits: they may exempt 20 percent of their caseload from time limits on federal funding. Furthermore, time limits may be waived in practice for the vast majority of recipients assisted by state funds. As a result, states can substitute federal and state funds as needed to retain longer-term recipients. (Of course, not all states choose to take advantage of this option.)
Those who predicted disaster may also have underestimated the magnitude of the increase in government support for low-income workers. President Clinton's talk of "ending welfare as we know it" referred not just to negative incentives for single mothers to avoid or exit welfare, but also to positive incentives to join the workforce. The welfare-reform legislation included some of these incentives, such as greater childcare spending and stricter enforcement of child-support responsibilities. Expansion of the EITC, the State Children's Health Insurance Program (SCHIP), and a minimum-wage increase were enacted separately, but they were still linked to welfare reform politically.
The EITC, a tax credit that goes to working parents with low earnings, is refundable, so it is essentially a cash benefit for those with no income-tax liability. For a minimum-wage worker with two children, the EITC has the same effect as a 40 percent increase in annual earnings. It is a bigger program in real terms than welfare ever was. SCHIP provides health coverage to children from low-income families who are not poor enough to qualify for Medicaid. Because Medicaid eligibility also expanded over the course of the late 1980s and 1990s, all children in families with incomes less than 185 percent of the poverty line roughly $34,500 for a married couple with two children in 2003 are eligible today for health coverage through Medicaid or SCHIP (although many eligible children remain uninsured).
Because federal block grants to states were based on the size of their caseloads prior to welfare reform, the dramatic decline in welfare rolls also freed substantial sums that states could spend on work supports. When combined with time limits and work requirements, these policy changes made working advantageous for more single mothers and enabled them to take advantage of the 1990s boom. Without the work supports, welfare reform almost certainly would have hurt more mother-only families economically than it helped.
Finally, welfare reform occurred in an environment in which demand for low-skilled workers was quite strong. Had welfare reform been implemented during a recession, recipients would have faced pressure to leave the rolls, but few jobs would have been available. Had it been implemented during the late 1980s, when unskilled workers' real wages were falling, minimum-wage jobs would have been somewhat easier to find than they are now, but surviving on what they paid would have been harder. In some sense it was the interaction of welfare reform, expanded work supports, and the economic boom that produced such unexpected outcomes among female-headed families with children.
Where Do We Go from Here?
Despite the bipartisan consensus that welfare reform has been a great success, Congress has yet to reauthorize the legislation, which expired in September 2002. As this is written in mid September, hardly anyone expects congressional reauthorization before the presidential election. Instead, Congress-watchers predict still another temporary extension. The political stalemate has almost nothing to do with disagreement about the impact of the last round of reform. Instead, there is bitter disagreement over the merits of adopting even tougher work requirements.
The leading proposals before Congress would increase the share of welfare recipients expected to work and require them to work more hours. Given the absence of job growth since 2000, this seems like the wrong time to toughen work requirements. Furthermore, the most employable women have already left the welfare rolls. Those who remain on the rolls tend to have low skill levels, poor mental and physical health, sick children, or other barriers to work. Finally, more welfare recipients will be reaching their time limits. In future economic downturns, there may be no safety net to speak of unless states are willing to take on the responsibility.
Tougher work requirements will require greater spending on work supports if policymakers are to avoid exacerbating hardship among single mothers. In particular, legislators will need to ensure that adequate funds are allocated for childcare. Adequate federal funding for work supports is particularly important right now, because states are no longer flush with cash, as they were in the late 1990s. Indeed, a number of states have completely spent down the reserves they built up then.
In the long run, funding for work supports needs to keep growing as it did in the 1990s. After all, even at the peak of the economic boom, one-third of single mothers were still below the official poverty line, even though nearly three out of four single mothers worked at least part time. Expanding the EITC and simplifying the application process for Medicaid and SCHIP would be straightforward steps toward helping single mothers and their children. Policymakers could also expand federal housing subsidies for female-headed families in which the mother works by excluding, say, the first $200 a month of earnings when calculating rents. Of course, making single motherhood more economically attractive increases the risk that it will become more common, but with work requirements and time limits, welfare today is a far less attractive option than it was in earlier decades. If policymakers wish to discourage single motherhood, their best strategy is to extend work supports to poor two-parent families.
The last round of welfare reform shows that, contrary to the fears of liberals, a policy that combines "sticks" with "carrots" can simultaneously promote work and improve the living standards of single mothers and their children. But it did not teach us much about what we should do about single mothers who cannot find work. Nor do we yet know how mother-only families will fare under this new regime if high unemployment persists. The latest estimates show, for example, that the poverty rate for single mothers increased more between 2002 and 2003, when unemployment was high but flat, than between 2000 and 2002, when unemployment was rising. Welfare reform was the product of a compromise between Democrats and Republicans. It has succeeded. This is not the time to unravel the compromise and try an experiment of doubtful wisdom.
Scott Winship is a doctoral student in the sociology and social policy program at the Kennedy School of Government. Christopher Jencks is Wiener professor of social policy at the school. Their study is available at http://ksgnotes1.harvard.edu/research/wpaper.nsf/rwp/RWP04-027?OpenDocument.
