Welfare Redux

When welfare reform passed in 1996, critics (including all of us) feared a substantial increase in material hardship among single mothers and their children. We were wrong. Six years ago, after reviewing dozens of government surveys, two of us wrote in these pages that the record was neither as grim as critics had feared nor as encouraging as advocates had promised. [See Christopher Jencks and Joe Swingle, “Without a Net,” TAP, January 2000.] The welfare rolls had been cut by almost half, material hardship had declined, and the rise in out-of-wedlock childbearing seemed to have slowed.

At that time, however, the economy was still booming and the most stringent requirements of the 1996 legislation had not yet been implemented. Skeptics therefore continued to fear that once the next recession arrived and the new law's requirements were fully implemented a lot more single mothers would start turning up in homeless shelters or would have to send their children to live with grandma.

Now the economy has been through a fairly long (though rather shallow) recession, and the requirements of the 1996 legislation are all in effect. Reviewing the evidence collected since 2000, we have found that single mothers were indeed worse off when unemployment peaked in 2003 than they had been in 2000. (Could it have been otherwise?) But single mothers were no worse off in 2003 than they had been in 1996, before welfare reform.

About 4.5 million families were getting welfare in 1996. Today, as chart 1 shows, that figure has fallen by two-thirds. Because the welfare rolls have dropped so dramatically, welfare is no longer a major political issue. As far as the public is concerned, the problem has gone away. Nonetheless, when the 1996 legislation had to be renewed in 2002, Congress was so polarized that it could not agree on what to do. With Republicans in control of Congress and the White House, the obvious solution might have been to declare welfare reform a success and leave the law unchanged. Had the three of us been in charge, the obvious solution would have been to make work-supports more generous. But conservative Republicans wanted to make it even harder for single mothers to get welfare, and most Democrats objected.

This impasse was finally broken last fall when the Republicans incorporated their welfare proposals into the budget reconciliation agreement for 2006. This maneuver prevented a Democratic filibuster in the Senate, and the changes passed in February. The biggest change is a requirement that the number of people receiving Temporary Assistance to Needy Families (TANF) without engaging in “work-related activities” should be no more than half a state's 2005 caseload. States can meet this requirement either by putting recipients to work or pushing them off the rolls entirely. Since today's TANF recipients are mostly hard to employ, putting them to work will be difficult and expensive. Pushing them off the rolls will be cheaper and easier.

State officials' temptation to choose the cheap alternative will be exacerbated if Congress adopts the administration's recent budget proposals for the next five years. After adjusting for inflation, these proposals would reduce federal grants to the states by 8 to 13 percent over the next five years. If the economy continues to grow -- a big “if” -- state revenues should grow enough to cover the reductions in federal funding. But given all the competing claims on states' resources, the new federal work requirements for single mothers may well bring about the hardships that liberals have been predicting since 1996. Before turning to that possibility, however, we need to say a little more about what has happened over the past decade.

* * *

Declining Poverty

Until 1996, most cash assistance for single mothers came from Aid to Families with Dependent Children (AFDC). In 1996 Congress replaced AFDC with a new program called Temporary Assistance to Needy Families. TANF established federal time limits and work requirements, and it allowed states to set requirements even more stringent than the federal ones. Several million single mothers left the welfare rolls between 1996 and 2000, and many other women who had recently become single mothers or lost their job decided not to apply for benefits or applied but were “diverted” into a job-search program. (Unlike AFDC, TANF does not give indigent mothers a “right” to welfare.)

Chart 2 shows our estimates of how the poverty rate changed for single mothers. Unlike the official poverty count, our estimates subtract taxes from income and add the Earned Income Tax Credit (EITC), food stamps, and the estimated value of school lunches and housing subsidies. Our estimates are also based on household income rather than family income, so they include the income of live-in boyfriends. Finally, our estimates adjust the poverty thresholds for inflation using a slightly better price index than the one used to make the official estimates.

The red line in chart 2 shows that our measure of poverty among single mothers fell from 21 percent in 1996 to 15 percent in 2000. The rate rose again during the recession, but it was still only 17 percent when unemployment peaked in 2003. The Census Bureau has not yet released the data we need to calculate our poverty rate for 2004 but the official rate for single mothers rose slightly between 2003 and 2004, and we predict ours will do the same thing. No poverty data for 2005 will be available until next fall.
The green line in chart 2 tells a different story. It shows the percentage of single mothers with incomes less than 70 percent of the poverty line, a condition we call “severe” poverty. Instead of falling between 1996 and 2003, severe poverty rose slightly. This apparent contradiction reflects a crucial fact about the past decade. While most single mothers have more money today than their counterparts had in 1996, the poorest mothers do not. Tighter work requirements will make this problem even worse if states meet the new federal requirements by pushing families off the welfare rolls entirely.

Chart 3 helps explain why these changes occurred. The red line shows a dramatic increase in the fraction of single mothers who reported that they had worked during the previous year and that no one in their household had received money from public assistance. The poverty rate for such mothers was 7.6 percent in 1996 and 7.4 percent in 2003, so as more single mothers moved into this category the poverty rate for all single mothers fell. (Readers should bear in mind that our poverty counts include live-in boyfriends' earnings. The poverty rate for single mothers who had to rely entirely on their own earnings would be more than 7 percent.)

The green line in chart 3 shows the fraction of single mothers who reported no income during the previous year from either work or welfare. This group also grew. Fortunately, most of these mothers lived with someone else who helped pay their bills, so “only” 44 percent of them had household incomes below the poverty line. But some of these mothers were very poor indeed. In the years between 1996 and 2003, roughly one in 10 reported that her total household income was less than $2,500. The growth of this group helps explain the rise in severe poverty.

Of course, $2,500 worth of goods and services is not enough to keep a mother and child alive for a year, so we cannot take these reports literally. Some of these mothers had other income that they did not report, either because it was illegal, because they worked off the books, because they had a boyfriend whose presence they did not mention, because they got money from a parent about whom the Census Bureau does not ask, or because the household's principal breadwinner had recently moved out. (The Census does not ask about the income of people who lived in the household last year but no longer live there at the time of its March survey.) That said, we still need to know far more than we now do about who these mothers are, why they are not getting TANF benefits, how they survive, and how their children are faring.

* * *

Not Moving in Tandem

Although chart 2 suggests that most single mothers had slightly more resources in 2003 than in 1996, those who went to work also had more expenses. Working mothers usually have to pay for transportation from home to work, and most of them also have to spend more on clothing. Furthermore, while child care subsidies rose between 1995 and 2003, subsidized child care is not always available at the time and place where a working mother needs it, so out-of-pocket spending on child care often rose. If single mothers' work-related expenses rose as much as their incomes, they would not have had any more money left for food, housing, or children's sneakers than they did before. Indeed, some of them may have ended up with less discretionary income than before.

Health-insurance coverage is another important determinant of a mother's expenses. If going to work meant that fewer single mothers were covered, as it did right after welfare reform, medical bills could have eaten up working mothers' extra income. Indeed, some mothers might have gone without the medical care they previously would have received. We measure trends in health-insurance coverage by calculating the proportion of the people with insurance in each single mother's household. This proportion hardly changed: It averaged 79.7 percent in 1996, 80.2 percent in 2000, and 80 percent in 2004. The proportion of single mothers with Medicaid coverage fell from 1996 to 2000, but by 2004 it had climbed back to its 1996 level. These statistics do not suggest that welfare reform cut single mothers' insurance coverage. But insurance does not cover everything. Because the cost of medical care rose faster than most other prices during these years, single mothers' out-of-pocket costs probably rose too.

One way to see if single mothers felt richer in 2004 than in 1996 is to see whether more of them thought they could afford a place of their own. This is not a measure of material well-being. A single mother who lives in someone else's home often has more creature comforts than a mother who has a place of her own. Single mothers want their own place because they want control. If a mother has a place of her own with her name on the lease, she can evict a boyfriend who misbehaves or stops helping pay the rent. If his name is on the lease, her material standard of living may be higher, but he has the whip hand in domestic negotiations. The same logic applies when a single mother lives with her mother. If a single mother lives in her mother's home, her mother is in charge and she may often be treated as if she were still a child. If she has her own place and lets her mother live with her, she feels like a grown-up.

The proportion of single mothers who headed their own households showed no clear trend between 1995 and 2005. About a sixth of all single mothers lived with relatives (usually one or both parents). Another 4 or 5 percent lived in a non-relative's household. The rest headed their own household. The fact that declining poverty was not accompanied by an increase in the fraction of mothers with their own household suggests that discretionary income did not rise much in the bottom part of the distribution.

* * *

Getting Enough to Eat

The U.S. Department of Agriculture (USDA) has paid the Census Bureau to conduct a “food security” survey every year since 1995. The survey asks more than 40 questions about whether households had trouble getting enough to eat during the previous 12 months. The April 1997 survey covers the period just before and after August 1996, when Congress replaced AFDC with TANF.

The problem that single mothers reported most frequently was running short of money and having to “stretch” food. Except for an upward blip in the December 2001 survey, this problem declined steadily, affecting only 44 percent of single mothers in 2003 compared to 57 percent in 1995. In 1995 even households near the middle of the income distribution for single mothers often reported this problem. Since these mothers' real income rose 15 percent between 1995 and 2003, the fact that they were less likely to report such problems in 2003 is no surprise.

Other problems the USDA asks about range from worrying that food will run out before the end of the month to reporting that a child went hungry for an entire day. Almost all these problems declined between April 1995 and April 2001. Almost all of them also rose during the recession. They were about as common in December 2003 as they had been in April 1997, but less common than they had been in April 1995. In 1997, for example, 26.9 percent of single mothers said they had worried that they would run out of food before the end of the month. In 2003 the figure was 27.6 percent -- a tiny increase. At the other end of the severity scale, 2.4 percent of mothers said in April 1997 that a child in their household had gone hungry for an entire day during the past 12 months. In December 2003 the figure was 2 percent -- a tiny decrease. The stability of these numbers is consistent with what we saw when we looked at health insurance and at having a place of one's own in which to live. Taken together, these facts suggest that living standards did not change much between 1996 and 2003 for single mothers in the bottom part of the income distribution.

If living standards have not changed for low-income single mothers, the apparent fall in the poverty rate shown in chart 2 is obviously misleading. The same must be true of the Census Bureau's poverty estimates for single mothers, which declined even more than ours and have been widely cited by defenders of welfare reform. The most obvious way of reconciling a declining poverty rate with stable levels of material hardship is to recognize that work-related expenses, which the Census Bureau does not measure, rose after 1996.

But work-related expenses may not be the only reason why our measures of living conditions failed to improve. Because of the EITC, most single mothers with jobs get a substantial check soon after they file their tax return. (Recipients are allowed to take the credit monthly rather than annually, but few do.) Recipients often use their EITC check to make a down payment on a car they can use to get to a better job, on a new living room set, or on other “big-ticket” items. As a result, the monthly income they use for food and rent may then be no higher than it was when they were on welfare.

One final caveat. Our findings do not show that welfare reform per se lowered poverty or left living conditions largely unchanged. Welfare reform was not a laboratory experiment in which everything else remained constant. It was accompanied by a lot of other legislation aimed at helping low-wage workers and their children. Congress had raised the Earned Income Tax Credit in 1993, and single mothers got more EITC money as more of them went to work. Medicaid coverage of low-income children expanded steadily because of changes enacted before 1996. Congress also raised the minimum wage within days of passing welfare reform. Low-wage workers' access to Medicaid also improved after 1996. Without all these changes, welfare reform might well have lowered living standards, just as liberals feared.

* * *

Defending the Two-Parent Family

For many conservatives the primary goal of welfare reform was not to improve single mothers' standard of living but to make them work for what they already got. This idea had overwhelming public support, and the United States has clearly moved in that direction since 1996. The late 1990s were the first economic boom in decades when employment increased faster among single mothers than married mothers. After the boom ended, some single mothers lost their jobs and had a hard time finding new ones. But work remains more common than it has ever been before.

But some conservatives also hoped that forcing single mothers to work would discourage out-of-wedlock childbearing and encourage marriage. For these traditionalists, welfare-reform legislation that raised single mothers' material standard of living would have seemed counterproductive, because it would have made single motherhood even more economically attractive than AFDC had.

According to the National Center for Health Statistics, the birth rate among unmarried women of childbearing age rose almost continuously from 1940 until 1994. It declined briefly between 1994 and 1996, before TANF replaced AFDC, and was basically flat from 1996 to 2002. It jumped in both 2003 and 2004, however, and was higher in 2004 than in 1992. The timing of these changes has no obvious relationship to the timing of welfare reform, but that may be because lots of other things also changed during these years.

If we compare blacks to non-Hispanic whites, we see a more striking pattern. In 1994, unmarried black mothers were considerably more likely than unmarried white mothers to receive AFDC. Thus if welfare reform made single motherhood less attractive, blacks should have been more affected than whites. And that may be what we see. The birth rate among unmarried white women was flat between 1994 and 2003. The birth rate among unmarried black women declined 18 percent. As a result, nonmarital births have accounted for a declining fraction of all black births since 1994. These racial differences suggest that welfare reform may have had an impact on nonmarital childbearing after all -- a small impact among whites that is offset by other factors, but a larger impact among blacks that is not fully offset by those factors. Still, the evidence is hardly conclusive.

* * *

What Next?

Welfare reform may not have had a big impact on single mothers' typical living conditions, but it transformed the political landscape. For the past 10 years the political action has almost all been in state capitals, not Washington, D.C. Once TANF gave states more control over welfare, most governors concluded that the best way to look good to the electorate was to cut the welfare rolls as rapidly as possible, and that was what they did. Unlike AFDC, which gave a state more federal money when it spent more of its own money, TANF gave states a fixed amount of federal money based on what they had received in the last years of AFDC. Under TANF, therefore, cutting the welfare rolls left states with more federal money for child care, job training, and other programs that help single mothers hold jobs.

In theory, TANF block grants were always contingent on a state's meeting federal work requirements and enforcing federal time limits. Since 2002 states have had to ensure that the number of families getting TANF and not participating in “work-related activities” was less than half the number getting AFDC in 1995. Now they will have to ensure that the number of people getting TANF and not participating in work-related activities is less than half the number getting TANF in 2005. The Congressional Research Service estimates that the new rules will raise the number of families facing work requirements by 70 percent. Because today's TANF recipients tend to have more problems than the typical AFDC recipient had in 1995, moving today's recipients into the labor force will be harder. It is unclear how many states will resist the temptation to just push these families off the rolls entirely.

For welfare reform's critics, time limits have traditionally been an even bigger worry than work requirements. States can exempt 20 percent of their caseload from the five-year lifetime limit on TANF benefits. But the 20 percent figure was not based on much evidence about how many recipients could really be expected to work. Nobody knew. Critics of welfare reform therefore worried about what would happen to mothers who reached their lifetime limit but were caring for a sick child or a frail parent; were sick themselves; or were too disturbed, too addicted to alcohol or drugs, or too slow-witted to hold a job.

Mothers began hitting the TANF time limit in 2001. Thus far, states appear to have managed this problem pretty successfully, using a combination of the 20 percent exemption, transfers to disability programs, and transfers to state-funded programs for the hard-to-employ. Recent reports by the Department of Health and Human Services indicate that very few TANF cases have been closed due to time limits, and some of those cases were transferred to state-funded programs.

The main reason welfare reform has hurt so few families is that the combination of rising wages and work supports like the EITC and child-care subsidies made work an economically viable option for single mothers who could hold a job. But the damage was also limited by the fact that states had enough flexibility to shelter mothers they judged incapable of working. That flexibility is now being reduced dramatically. The economic fate of single mothers is now tied to the business cycle in the same way as that of other working-age parents. Welfare is no longer the poverty trap that it was, but it is also less of a safety net. As other federal funds aimed at the poor are cut, the safety net will become even more threadbare.

By the time Congress abolished AFDC in 1996, everyone hated it. Liberals hated it for being too stingy. Conservatives hated it for supporting the undeserving. The public hated it for promoting idleness instead of work. Recipients hated it for forcing them to cheat in order to survive and then treating them like dirt because it suspected them of cheating. Nobody wept when it died.

But TANF was also a compromise that pleased neither liberals nor conservatives. Liberals thought that its time limits and work requirements would, as Senator Daniel Patrick Moynihan famously said, leave “tens on tens of thousands” of children sleeping on grates. Gingrich Republicans thought it was nowhere near tough enough. President Bill Clinton, who vetoed the Republicans' initial proposals, eventually signed the bill even though he thought it was a bad piece of legislation -- because he thought it was still better than AFDC. For the past decade that has probably been true. Whether it will remain true is an open question.

Christopher Jencks teaches social policy at Harvard University's Kennedy School of Government. He is working on a book about the social consequences of economic inequality. Joseph Swingle is Visiting Assistant Professor of Sociology at Wellesley College. Scott Winship is a doctoral candidate in social policy at Harvard's Kennedy School, and a co-founder of New Vision, a new student-led think tank on social policy (see www.newvisioninstitute.org).