Last week, I took an epistemological look at what it means to say that the ’96 welfare reform was a success. In this essay, I want to lay out a specific empirical case affirming the success of welfare reform — 20 years old as of last week. As I discussed in the previous article, no one can claim to have a rock-solid case that welfare reform did more good than harm, and no one can claim an equivalently strong case for the reverse. With that concession, why might someone think that welfare reform made children better off?
First, child poverty declined, and this decline occurred concurrently with dramatically falling welfare rolls and increasing work among single mothers. That doesn’t prove causality by any stretch, but my conclusions about the Personal Responsibility and Work Opportunity Reconciliation Act (PRWORA) would be very different if all three of these trends hadn’t coincided.
As I discuss in my new paper, “Poverty after Welfare Reform,” child-poverty rates are not only unambiguously lower than in 1996, they are at an all-time low. More controversial is the question of whether “deep poverty” (being under half the poverty line) or “extreme poverty” (living under $2 a day per person) has increased. I argue that the available data don’t allow for strong conclusions, but it is safe to say that neither increased by much. One or both may have declined, and in fact, I found that both could be at or near historic lows.
Even the Center on Budget and Policy Priorities (CBPP) — one of the groups more critical of welfare reform — finds that, at a minimum, the top 80 percent of children in single-mother families saw an increase in family income between 1995 and 2012. In fact, the second-poorest tenth of these children saw their income rise, and the poorest tenth experienced a decline of 6 percent (about $675) when the best cost-of-living adjustment is used. For all but the bottom 20 percent of children in single-mother families, private income rose by more than welfare benefits fell between 1995 and 2005 (the last year shown). CBPP concludes that deep poverty among children with unmarried parents rose from 2.8 percent in 1995 to 4.6 percent in 2010. My own figures, which I argue treat health benefits and inflation more appropriately, find a rise from 0.9 percent to 2.0 percent, with the rate falling to 1.7 percent by 2012 (lower than in 1997).
Both CBPP’s and my figures rely on a variety of imputations and ignore the problem of underreported earnings. We both attempt to do the best with the same data. But strong conclusions from the data are not merited, given the low levels and small changes involved. It does look like deep poverty at the time of welfare reform and today were and are lower than the rates that prevailed in the 1980s. For what it’s worth, my study also found that children of single mothers were about as likely to have a range of food problems at the start of welfare reform as they are today.
Some critics of reform have suggested that child poverty fell only because the rest of the safety net countered the harmful impact of welfare reform. As I discuss in my previous article, in so far as the expanding safety net did help — and it did — we cannot simply assume that that expansion would have happened without welfare reform, and we need to take seriously that earnings might have been lower absent the work incentives in welfare reform.
But in fact, adding food stamps, housing benefits, subsidized school meals, and energy subsidies to income does not affect the poverty trend among the children of single mothers after 1996. (It makes the deep-poverty trend look worse.) In contrast, simply adding refundable tax credits to income (the Earned Income Tax Credit and the Child Tax Credit) and combining the incomes of cohabiting families leads to the conclusion that poverty among the children of single mothers was lower in 2014 than ever before.
What about welfare receipt? No ambiguity there. The number of families receiving cash welfare through the old program, Aid to Families with Dependent Children (AFDC), or its replacement, Temporary Assistance for Needy Families (TANF), fell from 58 per 100 single-mother families in 1994 to 26 in 2000. In 2014, it was 16 per 100 single-mother families. In contrast, the ratio had risen between 1989 and 1994, from 51 to 58 per 100 single-mother families. (See my recent presentation at the Cato Institute.)
Finally, employment among single mothers without a high-school diploma rose from a little over 40 percent in 1992 to 65 percent in 1999. It fell thereafter, but it remains around 55 percent today. Among never-married mothers, employment rose from less than 45 percent in 1992 to about 65 percent in 2000. It was still 60 percent in 2013. During the period in which single mothers’ employment jumped, it rose modestly among married mothers and not at all among single childless women.
Employment among single mothers without a high-school diploma rose from a little over 40 percent in 1992 to 65 percent in 1999.
The rise in employment and fall in welfare receipt began before 1996, but so did welfare reform. By 1996, the administrations of Presidents George H. W. Bush and Bill Clinton had approved waivers allowing the vast majority of states to experiment with work requirements, time limits, and other provisions. And the writing was on the wall before 1996, too. The 1992 presidential campaign was notable for Clinton’s pledge to “end welfare as we know it.” He introduced his own proposal in 1994, and the Republican Congress passed two reform bills (both vetoed by Clinton) in 1995.
Correlation isn’t causation — the coincidence of these trends is only a starting point for asserting that welfare reform caused them. But the research on the factors behind declining welfare rolls and increasing employment among single mothers over the 1990s and early 2000s consistently found that welfare reform was important. The handful of papers that adjudicate between the effect of reform (and the state waivers before 1996) and other factors find that welfare reform was at least as important as the 1993 expansion of the Earned Income Tax Credit (or EITC, a subsidy for low-income workers) in reducing welfare receipt, both of which were larger factors than the improving labor market. The EITC was the most important factor behind rising employment among single mothers, but welfare reform was roughly as important as the economy.
The EITC expansion and welfare reforms were complementary: Child poverty would not have fallen if only one or the other had occurred. Welfare reform made not working untenable for many single mothers, and the EITC expansion made work more attractive.
Some criticisms of PRWORA have used poverty statistics that convey too dour — sometimes, as with the extreme-poverty estimates, far too dour — a view of how children have fared. But, the main criticisms of PRWORA that have emerged come from Peter Germanis, a senior policy advisor in the federal office that administers TANF who worked on welfare reform in the Reagan administration and who now writes essays independently as “Peter the Citizen” decrying what a disaster TANF is.
A major part of the Germanis critique focuses on the fairly undisputed fact that few states have been especially innovative or attentive in trying to move TANF recipients into jobs (and that is putting it nicely). Because states have fallen down on the job in this regard, many liberals, Peter Germanis, and even a few libertarians have concluded that PRWORA must have failed in encouraging single mothers to work.
But PRWORA might have increased employment in at least three ways. It might have led states to invest in welfare recipients and to actively assist them in finding work. Germanis is correct that states have disappointed by this yardstick. And yet PRWORA might have simply convinced a lot of single mothers to leave the rolls on their own by simply looking for and taking jobs. Moreover, welfare reform might have persuaded many single mothers to stay off the rolls in the first place and to work instead.
In fact, that is what welfare reform did, and because it was so successful in this regard, what states did to help TANF recipients get jobs ended up being relatively unimportant. Employment rose and poverty fell. Millions ended up better off than they would have been, full stop. Germanis’s latest salvo appears to concede (finally) that this story is theoretically plausible, but he rejects it for reasons that are not apparent. He argues that the fall in welfare receipt among single mothers was quite a bit larger than the rise in employment. Elsewhere he argues that the share of families eligible for welfare who are actually enrolled has fallen. But in the end, he has no rejoinder to my evidence that child poverty fell quite a bit and that deeper child poverty is rare and did not rise by much or at all (and may have fallen).
Employment rose and poverty fell. Millions ended up better off than they would have been, full stop.
Poverty can fall for reasons other than moving from nonwork to work. Hours-worked can rise and wages can increase. Families can alter their living arrangements and out-of-wedlock childbearing can fall. Germanis appears to believe that his favored metrics are indicators of rising hardship, but they are not.
Could welfare reform have done more to make states accountable for investing in recipients? Absolutely, though that would have meant spending more toward that end and less elsewhere, and successful work programs might have lured more people back onto the rolls or kept them from leaving. Designing a policy along these lines may or may not have reduced poverty more than PRWORA did. Reforming PRWORA along these lines may or may not reduce poverty in the future more than a continuation of the current program would. But PRWORA critics can’t simply point to the dearth of impressive work programs funded by TANF as evidence that welfare reform failed to put people to work or reduce poverty (or as evidence that some other approach would do better).
More hyperbolically, Germanis and others characterize TANF as a slush fund that states use for spending that doesn’t help poor single mothers and their families. PRWORA did give states much more discretion in how they spent money that, in the past, would had to have gone toward cash benefits, child care, or work programs. But even in 2014, 58 percent of TANF funds were spent toward these ends or refundable tax credits. That was down rather modestly from nearly 80 percent in 1997.
What critics call the “slush fund” is the one-third of TANF expenditures that go toward what CBPP characterizes as “a broad range of uses, including child welfare, parenting training, substance abuse treatment, domestic violence services and early education.” That’s up from 12 percent in 1998, an increase, but not an especially dramatic one, particularly since these functions are at least indirectly related to poverty and upward mobility. An earlier CBPP analysis found that TANF spending in 2011 was essentially at the same inflation-adjusted level as in 1997 outside three categories of spending that — in part, and not necessarily in large part — reflect gaming of the system by states.
Furthermore, because the rolls fell so much, the amount of inflation-adjusted spending on cash benefits per TANF beneficiary actually has held steady. (Compare the number of families receiving cash assistance to the spending in this CBPP report, which I adjust for inflation. CBPP reports that the inflation-adjusted value of monthly TANF benefits has fallen markedly over time. It is unclear how to reconcile these numbers, but the monthly TANF benefits are estimates of maximum amounts for a family of three, potentially failing to capture changes in the average duration of benefit receipt, in family size, and in the share of beneficiaries receiving the maximum amount. CBPP also uses a cost-of-living adjustment that has not been used for research purposes in over 20 years by the major federal statistical agencies.)
As my first article indicated, no conservative believes that PRWORA was the single best way that child poverty might have been lowered. We (most of us) simply believe that it did, in fact, lower child poverty. Viewing it as at least something of a success, then, we are reticent to make major changes to current law that would weaken the work-promoting incentives.
The unjustified hyperbole from the left about welfare reform’s negative effects puts conservatives on the defensive in proportion to how influential it becomes. Critics who seek compromise with conservatives around the design of the future safety net would get much further if they would not insist that welfare reform caused a substantial increase in child poverty — deep, extreme, or otherwise.
If they did that, they would find that many conservatives are open to changes in TANF, from allowing more education and training to count as “work activities,” to requiring states to spend more on the core functions of TANF, to curbing state excesses in discouraging eligible families from getting the benefits they deserve. Some conservatives — more than liberals might guess — are open to returning most decision-making to the federal level. Others of us might be willing to raise spending levels. Some are proponents of expanding the EITC or Child Tax Credit.
The real priorities among conservatives are retaining work requirements and time limits and avoiding entitlements — whether in the form of a guaranteed open-ended cash benefit or a government-subsidized job. We want to experiment with work requirements and time limits in other programs because we believe they were successful — poverty-reducing — in the case of TANF and that their success was partly the result of state experimentation.
I was opposed to welfare reform in 1996 because I was worried that change would be worse for poor children. As political scientist and Daniel Patrick Moynihan–biographer Greg Weiner notes, the concern about PRWORA espoused by Moynihan was not necessarily wrong in 1996, even if it looks misplaced in 2016. Today, though, the same kind of first-do-no-harm instinct calls for retaining the central features of PRWORA and cautiously trying to build on them to improve other programs.