Without a Net

The welfare rolls have fallen by almost half since 1994. To assess the impact of this dramatic change, both journalists and social scientists have been talking to families that have left the rolls. But these families are only half the story. Even without welfare reform, nearly half the single mothers on the rolls in 1994 would have left by 1999 simply because their children grew older, they found work, or they got married. But in the absence of welfare reform, most of these mothers would have been replaced by other mothers who had just had their first baby, split up with their husband, or lost their job. Because states have made it far more difficult to get on welfare, most of those who left the rolls were not replaced.

The only way to assess the overall impact of welfare reform is to ask how life has changed for all single mothers, including those who no longer even apply for welfare. The most up-to-date information on single mothers comes from the Census Bureau's Current Population Survey (CPS), which collects data every March on about 5,000 unmarried mothers with children under 18. To see how these mothers' lives have changed, we have analyzed the CPS data collected from March 1988 to March 1999, which can help us answer four questions:

  • How many single mothers are working?
  • How have single mothers' incomes changed since welfare reform began?
  • Has declining support from welfare forced more single mothers to double up?
  • Are mothers more likely to marry now than they were before reform began?

The answers to these questions are neither as grim as critics of welfare reform feared nor as encouraging as some advocates of reform promised. More single mothers have jobs. Most also have more money, but a small minority is worse off. The proportion of single mothers doubled up with relatives has not changed. Marriage rates, which stopped falling before welfare reform began, have remained flat.

What the Numbers Tell Us

Welfare reform began in some states in the early 1990s, when Washington began issuing "waivers" allowing states to impose work requirements on certain mothers. But Congress did not abolish Aid to Families with Dependent Children (AFDC) until 1996, and the new system, known as Temporary Assistance for Needy Families (TANF), did not take effect until 1997. The number of households receiving welfare peaked in 1994, two years after the economy began recovering from the recession but three years before TANF took effect [see "Declining Reliance on Welfare," page 39]. By 1996 the rolls had fallen 10 percent, but once TANF took effect, the decline accelerated. Almost everyone agrees that the tightening labor market, the Earned Income Tax Credit (EITC), and welfare reform all played a role in this unprecedented decline.

The labor market has been tighter in the past few years than at any time since the late 1960s, making it far easier for single mothers to find work in convenient places and at reasonable hours. In addition, the EITC can now raise a single mother's earnings by as much as $2 an hour. The EITC reached this level in 1996, but because many mothers were unaware of the change, its impact on employment has been more gradual. An analysis by David Ellwood concludes that changes in the economy and the EITC together account for about half the decline in the rolls since 1994. The rest is welfare reform.

After TANF became law in 1997, states imposed time limits on welfare receipt. Most also imposed much tougher work requirements. Perhaps most important of all, states concluded that the quickest way to reduce their case load was to open fewer new cases. Many states began"diverting" new applicants into employment programs. Some just hand a mother the "Help Wanted" section and tell her to come back after she has made, say, 30 calls. A lot of mothers never come back. As the word spreads, the number of mothers coming into welfare offices begins to fall.

More Single Mothers Are Working

Follow-up studies show that most mothers who leave the welfare rolls find jobs, but a large minority do not. Furthermore, some of those who find jobs soon lose them. The figure "Working Is up among Single Mothers" [page 39] divides single mothers into four groups. The solid red line shows the percentage who reported no welfare income and who also worked at least part of the year. This number has climbed more since 1993 than during any previous recovery. The solid green line shows the fraction of mothers who reported no income from either work or welfare. Between 1987 and 1996, about 10 percent of all single mothers fell in this category. By 1998 the fraction had climbed to 12 percent. This increase is consistent with studies showing that a substantial minority of former recipients are not working.

If a mother has no income for an entire year, she cannot maintain her own household. Such mothers sometimes have to split up their families, but most of them move in with a parent, sibling, or boyfriend. In 1998 such mothers lived in households with incomes averaging about $22,000. Low as this figure is, it is still considerably higher than the household income of the mothers who got welfare. How much of the income was available to support the mother and her children is unclear. Another big worry for these mothers is that they seldom have a legal right to stay in their current households, so if there is a domestic quarrel, they sometimes end up in the streets.

Incomes Are Rising at the Top but Not at the Bottom

Many surveys have found that even when single mothers work, they often remain poor. Census statistics on poverty do not throw much light on this because they ignore the income mothers get from live-in boyfriends and only provide data on single mothers who head their own household. The figure "Single Mothers' Household Incomes" [page 39] shows the total household income of every single mother with children under 18. The income measure includes not only money from jobs, welfare, and the EITC but the estimated value of food stamps, school lunches, subsidized housing, and most other means-tested benefits. The measure does not assign a value to subsidized health insurance because this is worth far more to families with serious health problems than to other people. The income measure also omits the value of child care subsidies because such data do not exist. We convert income in all years to 1997 dollars.

At the 90th percentile of the distribution for single mothers, household incomes were about $55,000 in 1998, and rising. But mothers at the 50th percentile were just scraping by, with 1998 incomes near $23,000. Mothers at the 10th percentile reported total household incomes of only $9,400. Furthermore, while most single mothers had more income in 1998 than in 1996, those at the 10th percentile reported no gain and those at the 5th percentile actually lost a little ground. Incomes at the 5th percentile averaged only $6,400 in 1997 and 1998, down from $6,900 in 1995 and 1996. But even at the 5th percentile, single mothers were doing better than they had in the late 1980s, when their incomes averaged only $5,600.

At this point the reader should be wondering whether any mother can really make ends meet on less than $10,000 a year, especially when the income measure includes the value of food stamps and housing subsidies. Such skepticism is reasonable. When the Consumer Expenditure Survey asks single mothers how much they spend, those who report incomes below $10,000 mostly spend well over $10,000. A few of these mothers may be living on credit cards, but most are just underreporting their true income. In Making Ends Meet (1997), Kathryn Edin and Laura Lein found that almost all unskilled single mothers relied on gifts from family members or boyfriends, or on off-the-books employment to balance their budgets. Mothers are unlikely to report such income to the Census Bureau. But while these mothers are not quite as poor as the CPS implies, Edin and Lein's findings suggest that most of them still experience a lot of material hardship.

"Single Mothers' Household Incomes" [page 39] also overstates some mothers' disposable income. The income figures are adjusted to eliminate taxes, but not to eliminate work-related expenses. Now that more single mothers work, they are spending more on child care, transportation, and office clothes. This increase in work-related expenses could mean that some mothers in the lower part of the distribution end up with less money for food, housing, and other necessities. To see how common this is, we need data on single mothers' living conditions. The federal government does collect such data, but not much of the post-1996 data is currently available, so we have to rely on the CPS for clues.

Doubling Up

When single mothers cannot make ends meet on their own, they usually try to move in with relatives. Thus if welfare reform had left single mothers worse off, we would expect more of them to be doubled up. In fact, the percentage of single mothers living with relatives has not increased [see "Single Mothers' Living Arrangements," page 39], and neither has the percentage living in a household headed by a nonrelative (usually a boyfriend). The only change in single mothers' living arrangements is that those who head their own households are more likely to report having a nonrelative in their household. Four-fifths of these nonrelatives are boyfriends.

Although more single mothers report live-in boyfriends, it is not clear what this means. Historically, many mothers have had boyfriends whose presence they did not mention to census takers. ("He doesn't live here. He's just staying here.") As mothers leave welfare, they have less reason to conceal a boyfriend's presence. But single mothers who no longer get welfare and cannot find steady work may also be relying more on problematic men whom they would previously have kept at arm's length. Or couples may just be cohabiting longer before they decide whether to marry.

When mothers are unable to feed their children or keep a roof over their heads, they often send the children to live with relatives. When that is impossible, the kids often end up in foster care, at least for a while. Thus if single mothers were having more trouble making ends meet, we would expect to find more children living apart from both parents. In fact, the number of such children shows no clear trend between 1994 and 1999, fluctuating between 9 and 11 percent among blacks and between 2 and 3 percent among whites.

Less Health Insurance Coverage

Health insurance coverage has fallen for almost all groups, including single mothers, since the early 1990s. Because almost all poor children are now eligible for Medicaid, while many poor adults are not, we assess changes in coverage by looking at every single mother's household and calculating the percentage of all household members with coverage. From 1987 to 1993, coverage fluctuated between 81 and 82 percent. After 1993 it began to fall. In 1997 and 1998, coverage averaged only 79 percent.

The reason for declining coverage is clear. When a single mother goes on welfare, she is automatically enrolled in Medicaid. In 1993, when the welfare rolls peaked, 40 percent of all single mothers said they had Medicaid. As the rolls fell, Medicaid coverage fell too. By 1998 only 33 percent of single mothers said they had Medicaid coverage. Private coverage increased somewhat as mothers entered the labor force, but not enough to offset the decline in Medicaid.


Although most Americans favored welfare reform primarily because they favored work requirements and time limits, many hoped that eliminating AFDC would also discourage single parenthood, which had spread steadily since the 1960s. "Custodial Parents Who Are Unmarried" [page 41] shows that this trend came to a halt in 1994. Single parenthood has not become any less common, but the long-term increase has stopped, at least for the moment. "Children Born out of Wedlock" [page 41] shows changes in the percentage of children born out of wedlock. Here too we see a steady rise from 1970 to 1994. Since 1994 we have been on a plateau.

Both of these figures suggest that the rise of the single-parent family may finally have been arrested. Because this happened well before 1996, welfare reform does not seem a likely explanation. State-by-state analysis reinforces this judgment. Wisconsin began serious welfare reform earlier than almost any other state and has cut its rolls more than any other large state. Yet the proportion of Wisconsin children born to single mothers has not fallen. Indeed, the proportion climbed from 27.1 percent in 1994 to 28.5 percent in 1998—an increase of 1.4 points—at a time when the increase for the nation as a whole was only 0.2 points. David Ellwood has investigated this issue more systematically using data from all 50 states. He has found no evidence that marriage became more popular in states that implemented welfare reform more assiduously.

What Next?

These are still early days for welfare reform, which did not begin in earnest until 1997. The rolls have been cut in half, but everyone agrees that this was the easy half. Employment among single mothers has increased more than almost anyone expected, but not everyone is finding work, even in today's booming economy. Most surveys of former recipients suggest that their economic situation has not changed much. If we consider all single mothers instead of just former recipients, most are a little better off today than they were in 1996, but the bottom 5 percent seem to be a little worse off. If we could take account of work-related expenses and out-of-pocket health care spending by those who no longer have insurance, we might find that a somewhat larger fraction of single mothers had lost ground.

Strictly economic considerations aside, two facts stand out. First, almost all mothers who are working tell interviewers that they prefer work to welfare. Second, many working mothers report problems finding satisfactory child care. There is a lot of anecdotal evidence that young children are being left alone for long periods. These reports suggest that welfare reform could end up helping parents but hurting their children. Because we have no reliable system for monitoring children's well-being, we will probably never know how welfare reform affected them.

If the welfare rolls were to keep falling at the current rate, TANF would disappear within a few years, but we think that is unlikely. Some current TANF recipients have serious mental or physical problems that make it hard to see how they could get a job, much less keep it. Other current recipients have children with mental or physical problems that demand constant attention at unpredictable times, making employment difficult and child care almost impossible to find. No knowledgeable welfare official believes that every mother currently on the rolls is capable of holding a steady job. But to the best of our knowledge, no state has a well developed plan for dealing with its "hard core" cases.

No one knows how low the rolls can go without a sharp increase in suffering. When Congress abolished AFDC in 1996, it allowed states to exempt a fifth of their case load from work requirements, but it did not exempt anyone from time limits. Drawing a line between those who can and cannot work is never easy, as we have learned over the years in battles over eligibility for disability benefits. In theory, any mother whom no private employer will hire should presumably be eligible for disability benefits. In practice, that will not always be the case. Some people will be denied benefits because reviewers judge them capable of work, but they will nonetheless fail to find a job. To some, failure to find work will constitute evidence that disability benefits should have been approved. To others, failure to find work will just look like malingering. Whether a state errs more often on the side of excessive generosity or excessive stinginess will ultimately depend on the political climate in which decisions are made.

One big unresolved question is whether popular support for work requirements will lead states to spend more money helping marginally employable single mothers join the labor force. This is happening in a few states, but many states seem to care more about saving money than getting everyone into the labor force. Another unresolved question is how states will handle children who need so much attention from their mother that she cannot hold down a job while responding to their problems.

There is a disturbing parallel between America's recent effort to eliminate welfare and its earlier effort to deinstitutionalize the mentally ill. When deinstitutionalization began in the late 1950s, we had been locking up and sedating far too many people. Many patients were clearly better off once they left the hospital and began living with relatives or in halfway houses. Many also found that they could hold a job most of the time.

But during the 1970s, after hospitals had discharged the patients who were most likely to succeed in the outside world, two problems arose. First, some civil liberties lawyers became convinced that mental hospitals were just jails by another name, and that involuntary commitment should be subject to the same restrictions as imprisonment. Judges responded by making involuntary commitment increasingly difficult. Second, economic growth slowed during the late 1970s, and real tax revenues stopped rising. As a result, governors began looking for programs to cut. Conservative budget cutters soon formed an unholy alliance with radical civil libertarians, in which both groups sought to empty state mental hospitals. This led to a train wreck.

The housing and psychiatric services that were supposed to help the mentally ill cope after they left the hospital system never materialized, so the mentally ill often ended up homeless. Even patients who sought hospitalization were often turned away, either because the shrinking hospital system had no space or because they did not meet the hospitals' ever-more stringent standards for admission. And patients who threatened or attacked their fellow citizens were often released a few days later, on the grounds they were less disturbed than they had been when they were admitted.

Not many single mothers are mentally ill, but both groups are politically weak. This makes it easy to imagine another alliance between budget cutters and ideologues who think society should insist that every single mother work. Work is increasingly essential for self-respect in America, so given the right kinds of support, most single mothers are probably better off working, at least part time. But almost any good idea can lead to disaster if it is pursued too relentlessly, and the idea that single mothers should work is no exception.

Some single mothers can't manage both employment and parenthood simultaneously. Even those who have the energy and skill to juggle work and parenthood often earn so little that they cannot make ends meet without additional help. If such help is not available, the long-term impact of welfare reform on both single mothers and their children could well turn out to be like the long-term impact of deinstitutionalization on the mentally ill: good for some but terrible for others. This is a worst-case scenario. But it is a possibility we should bear in mind as states keep cutting their welfare rolls.