You are here

Welfare Reform and the Safety Net

Evidence Contradicts Likely Assumptions Behind Forthcoming GOP Poverty Plan

June 6, 2016

In releasing their poverty plan tomorrow, House GOP leaders likely will argue that the 1996 welfare law — principally its creation of the Temporary Assistance for Needy Families (TANF) block grant, with its work requirements, time limits, and other features — has been a major success, while other anti-poverty programs have largely failed and should be reshaped along the lines of TANF.  House GOP leaders likely will also talk about the value of “evidence-based policy.”  So let’s look at the evidence about the effects of TANF and other anti-poverty policies on poor families’ well-being.  Three findings stand out:

  • TANF’s results have been mixed.  The child poverty rate fell in the years immediately following TANF’s creation, when the economy was booming and major Earned Income Tax Credit (EITC) expansions took full effect, and TANF likely played a role here, though it’s not possible to determine exactly how much it contributed to the decline and the evidence now suggests its contribution was likely modest.  At the same time, the data are clear that TANF also has contributed to an enduring increase in “deep poverty” (people living below half of the poverty line) and to a significant loss of income at the very bottom.
  • As more evidence of TANF’s effects has emerged, researchers have adjusted their assessments of welfare reform.  Researchers are increasingly concluding that TANF is not the unvarnished success sometimes claimed.  While TANF — in combination with large EITC increases and other factors — appears to have led to increases in employment and earnings among more-skilled single female parents, especially in its early years, TANF also led to losses of income and deeper poverty among a sizable number of other female-headed families.  Concerns over these downside effects have grown as other research has emerged indicating that deep poverty can have serious, lasting effects on young children, including adverse effects on brain development.
  • Various other anti-poverty programs have proven effective.  The evidence is strong that various other key safety-net programs such as SNAP (formerly food stamps) have reduced poverty markedly and produced other important benefits, including improving children’s chances of success in adulthood.  Moreover, the safety net’s effectiveness in reducing poverty has grown about ten-fold since 1967:  in 2014, safety-net assistance lifted out of poverty 42 percent of those who would otherwise be poor, compared to just 4 percent in 1967.

TANF’s Mixed Results

The common argument that the decline in poverty rates that occurred after TANF took effect proves TANF strongly benefited the poor suffers from two significant flaws.  First, how much of the decline in the child poverty rate since TANF’s inception is due to TANF itself is uncertain, especially beyond TANF’s early years.  Second, the poverty rate is highly inadequate by itself as a way to measure TANF’s effects and misses the rise in deep poverty.

TANF and Child Poverty Rates

Changes in poverty rates reflect myriad factors, including the economy and income from other government programs, not simply TANF and families’ responses to it.  For example, the claim that TANF substantially reduced the poverty rate rests on the argument that TANF led many more single mothers to work and that their earnings, supplemented by the EITC, lifted them above the poverty line.  Leading academic studies have found, however, that the major EITC expansions in the 1990s had a larger effect than TANF in inducing more parents to work and that the loss of benefits often offset the increase in earnings for mothers who left welfare for work due to TANF.[1]

Moreover, while the data from TANF’s early years, when the economy was very robust, suggested strong positive results, data for more recent years have been decidedly mixed.  As Robert Moffitt of Johns Hopkins — widely regarded as one of the nation’s leading experts on these issues — recently explained: 

[W]elfare reform had a positive effect on employment and hours worked of single mothers. . . .  However, while the literature also shows positive effects on earnings, the declines in welfare benefits arising from leaving welfare often cancel out the earnings increases, leaving income relatively unchanged (although the literature also shows considerable heterogeneity, with some families experiencing income increases and other decreases).  Especially in more recent years, in addition, a significant number of single-mother families appear to have been made worse off and to have higher deep poverty rates.[2]

Poverty Rate an Inadequate Measure of TANF’s Effects

The evidence suggests that a combination of the EITC expansions, TANF, and a stronger labor market in the late 1990s lifted a number of female-headed families — particularly the most skilled and employable — above the poverty line, thereby lowering the poverty rate.  But that doesn’t tell us whether a number of other poor single parents and their children were pushed deeper into poverty.  Since these families were poor to start with, their increased destitution would not affect the poverty rate.  As former Bush Administration welfare advisor Peter Germanis has noted:  “The poverty rate is not a particularly useful measure for assessing the impact of welfare reform on those receiving cash assistance because the eligibility levels for cash aid are well below the poverty thresholds.”[3]  If TANF made many already poor children poorer, that wouldn’t show up in the poverty rate.

To examine whether TANF led a significant share of poor female-headed families and their children to become still poorer, we need to use other measures, such as the deep poverty rate.  The most complete measures show that TANF led to a significant increase in deep poverty.  Between 1995 (before the welfare law was enacted) and 2005 (a year with similar unemployment), the share of children in deep poverty rose from 2.1 percent to 3.0 percent, and the number of such children jumped from 1.5 million to 2.2 million.[4]  Similarly, the deep poverty rate among people in female-headed families jumped from 2.9 percent in 1995 to 6.0 percent in 2005.  (These figures use a broad measure of poverty, similar to the Supplemental Poverty Measure, that analysts favor because it counts benefits such as SNAP and the EITC; these figures also correct for the underreporting of key government benefits in the survey data.)[5]

The data also demonstrate that the rise in deep poverty was due largely to TANF’s sharp reduction in cash assistance for poor families.  As one indicator of this, in 1995, for every 100 poor families with children, 76 received cash assistance through the Aid to Families with Dependent Children (AFDC) program, TANF’s predecessor.  In 2014, for every 100 poor families with children, only 23 received cash assistance through TANF.[6] 

Our analysis shows that if TANF and other safety-net programs had kept the same percentage of otherwise deeply poor children out of deep poverty in 2005 that AFDC and other safety net programs did in 1995, 1.2 million fewer children would have been in deep poverty in 2005.  Stated another way, 1.2 million more children lived below half the poverty line in 2005 due to the weakening of the safety net for these children and their families, chiefly the dramatic decline in TANF cash assistance.[7]

As further evidence, consider what happened to the average income (after the receipt of benefits and payment of taxes) of the poorest one-tenth of female-headed families between 1995 and 2005.  The figures here are sobering.  These families’ average income fell $3,400 (in inflation-adjusted terms), driven by a loss of $3,300 in TANF assistance.[8]

These findings are of particular concern given the growing body of research indicating that deep poverty can be especially harmful to children — especially young children, whose life prospects and chances for upward mobility can be compromised.  Child poverty’s effects, experts note, are “particularly pronounced … for children who live in extreme poverty (that is, 50% or less of the poverty threshold).”[9]  Scientists have warned of the toll of poverty on children’s brain development; as the American Academy of Pediatrics recently warned, “Poverty has direct negative effects on early brain development through the mechanism of toxic stress.”[10]

One study found that every $1,000 change in family income during early childhood caused by anti-poverty and welfare-to-work programs in the 1990s was associated with a significant change in children’s school achievement.[11]   

More Recent Findings Have Altered Experts’ Assessments of TANF

As noted, the research finds that some female-headed families fared better under TANF while others fared worse.  For example, one study found that TANF raised disposable incomes an average of 8 percent among higher-skilled mothers but led to a net loss of income among low-skilled mothers.[12]

Increasingly, researchers examining the full body of research on TANF have concluded that its results are much less positive than the results from TANF’s early years had suggested; this may be because the early years coincided with an unusually strong economy and robust EITC expansions.  Robert Moffitt, for example, notes TANF’s mixed effects on income — with a significant number of families becoming worse off — and on savings and consumption.  He also notes “some negative effects of the 1996 reform on health insurance coverage because of the loss of Medicaid [among some of those losing TANF], no consistent evidence of the effect of reform on family structure and fertility, and mixed results of the effect of reform on children.”[13]

Similarly, Ron Haskins, who led much of the drafting of the 1996 welfare law as a senior Republican congressional aide, testified last year that deep child poverty has increased since the law’s passage and that “some poor mothers and their children have neither cash welfare benefits nor earnings.” “We should acknowledge this problem and try to figure out ways to help these disconnected mothers and their children,” he urged policymakers.[14]

Of particular note, some leading researchers who had termed the welfare law a success have significantly changed their assessment due to the more recent evidence.  For example, Harvard’s renowned Christopher Jencks, one of the country’s most rigorous and highly regarded analysts, noted earlier this year:

People who were able to find work, either because they live in places where work was available, or because they were better qualified than the average welfare recipient, have done pretty well.  People who can’t find work are where they were before they had welfare at all.  That’s a big problem.  People have no means of support for themselves or their children.[15] 

Jencks recently provided data showing that the poorest of the poor have become considerably worse off.[16]

In 2004, looking at the research available at that time, Jencks co-authored an article terming welfare reform a major success.  Now, Jencks says, “I was wrong.”[17]

Evidence Shows Poverty Programs Have Had Important Positive Impacts

As recently as late May at a House Ways and Means Committee hearing, some members declared federal anti-poverty programs a failure even while calling for use of evidence-based policy.  The evidence, however, doesn’t support the claims of failure.

For example, one often-heard claim is that the poverty rate is about the same as it was 50 years ago and this proves federal anti-poverty programs have failed.  House Speaker Paul Ryan repeated a version of this claim on June 3, asserting that “Washington has spent trillions of dollars on dozens of programs to fight poverty.  But we have barely moved the needle.”[18]  Yet such claims are bereft of merit, as they rest on serious misuse of data.

It’s true that the “official” poverty rate today is about what it was 50 years ago.  But analysts across the political spectrum agree that one cannot compare poverty then and now using the official poverty measure.  The official measure counts as income only those government benefits that are provided through spending programs which pay benefits in cash.  It ignores SNAP, rental assistance, the EITC, the refundable part of the Child Tax Credit, and more.  This means it ignores virtually all anti-poverty assistance created or expanded over the past half century, while counting the main form of assistance cut sharply over this period — cash assistance for poor families with children.  You can’t learn anything about the efficacy of most anti-poverty programs by using a measure that doesn’t count them.

Most analysts favor a broader measure of poverty that counts, rather than ignores, those benefits.  And Columbia University researchers have found, using a version of the Supplemental Poverty Measure (the principal such broader measure), that the poverty rate has shrunk by two-fifths since the late 1960s, with nearly all of the progress due to the creation and expansion of various anti-poverty programs.[19] 

The data also show that while in 1967, federal programs lifted above the poverty line only about 4 percent of Americans who would otherwise be poor, the programs now lift out 42 percent of those who would otherwise be poor.[20]  A separate study found that SNAP lifted 10.3 million people (including 4.9 million children) out of poverty in 2012 and cut the number of children in deep poverty by 1.2 million, more than any other program.[21]  

Moreover, extensive academic research has found other important benefits from key anti-poverty programs — from reduced child hunger and malnutrition and better health to various indicators of upward mobility.  A strong body of research links programs such as SNAP and the EITC to increased educational attainment among children and to stronger employment and earnings in adulthood.  Charges that these programs may reduce material hardship but do nothing to improve mobility are inconsistent with the evidence.

These and other findings on the effects of various safety-net programs can be found in an array of important papers and book-length studies, such as the recent NBER volume on the safety net edited by Moffitt, papers issued last month by the Hamilton Project, a recent book synthesizing the research on SNAP, and much other work.[22]

It’s encouraging to hear policymakers call for evidence-based policymaking.  For such calls to be more than sound bites promoting one’s ideological preferences, however, policymakers also need to rely on the actual research and data in the field.

End Notes

[1] Jeffrey Grogger, “The Effects of Time Limits, the EITC, and Other Policy Changes on Welfare Use, Work, and Income among Female-Head Families,” Review of Economics and Statistics, May 2003; James P. Ziliak, “Temporary Assistance for Needy Families,” in Robert Moffitt, ed., Economics of Means-Tested Transfer Programs in the United States, National Bureau of Economic Research, November 2015,

[2] Robert Moffitt, “Introduction to Volume I,” in Robert Moffitt, ed., Economics of Means-Tested Transfer Programs in the United States, National Bureau of Economic Research, November 2015,

[3] Peter Germanis, “Will Conservative Reforms to the Safety Net Reduce Poverty? A Skeptical Conservative Responds to Scott Winship,” January 17, 2016,

[4] Arloc Sherman and Danilo Trisi, “Safety Net for Poorest Weakened After Welfare Law But Regained Strength in Great Recession, at Least Temporarily,” Center on Budget and Policy Priorities, May 11, 2015,

    Other studies have found similar results, including a rise in deep poverty for the U.S. population as a whole from 4.5 percent in 1993 to 6.6 percent in 2004 (Yonatan Ben-Shalom, Robert Moffitt, and John Karl Scholz, “An Assessment of the Effectiveness of Anti-Poverty Programs in the United States,” Institute for Research on Poverty, Discussion Paper no. 1392.11, revised June 2011) and a near-doubling between 1996 and 2011 in the number of families with children living on monthly cash income of less than $2 per person per day ( H. Luke Shaefer and Kathryn Edin, “The Rise of Extreme Poverty in the United States,” Pathways, Summer 2014,; see also Kathryn J. Edin and H. Luke Shaefer, $2 a Day: Living on Almost Nothing in America, 2015).

[5] Data for the last few years are not yet available, and data for the years of the Great Recession don’t provide an appropriate comparison, both because of the increase in unemployment and because of significant temporary increases in safety-net income as a result of stimulus measures enacted in 2008, 2009, and 2010.

[6] Ife Floyd, LaDonna Pavetti, and Liz Schott, “TANF Continues to Weaken as a Safety Net,” Center on Budget and Policy Priorities, revised October 27, 2015,

[7] Sherman and Trisi.  The analysis shows that the safety net lifted fewer children above half the poverty line in 2005 (7.6 million) than in 1995 (10.7 million).  The majority of that decline was driven by TANF, which protected 0.6 million children from deep poverty in 2005, down from 2.4 million in 1995.

[8] These figures use Census data, adjusted for the underreporting of key safety-net benefits.  The $3,400 decline is in 2015 dollars using the CPI-RS to adjust for inflation.  If the PCE deflator is used instead, the decline is $2,600.  If other technical adjustments are made (such as to adjust for changes in the average number of people in a family, subtract expenses as the Supplemental Poverty Measure does, and the like), the dollar loss shrinks somewhat but substantial dollar losses remain, regardless of the measure used to adjust for inflation.  All figures adjust for changes in family size and are expressed in four-person-family equivalents.

[9] Jeanne Brooks-Gunn and Greg J. Duncan, “The Effects of Poverty on Children,” The Future of Children, Vol. 7, No. 2, Summer/Fall 1997.

[10] John M. Pascoe et al., “Mediators and Adverse Effects of Child Poverty in the United States,” Pediatrics, March 2016,

    The Harvard Center on the Developing Child defines toxic stress as “strong, frequent, or prolonged activation of the body’s stress management system” without adequate support from a caring adult.  Toxic stress in sensitive periods of child growth can adversely affect brain architecture, leading to underproduction of neural connections in areas of the brain dedicated to reasoning.  National Scientific Council on the Developing Child, Excessive Stress Disrupts the Architecture of the Developing Brain: Working Paper 3, updated 2014,

[11]  Greg J. Duncan, Pamela A. Morris, and Chris Rodrigues, “Does Money Really Matter? Estimating Impacts of Family Income on Young Children’s Achievement with Data from Random-Assignment Experiments,” Developmental Psychology, June 2011, pp. 1263–1279.

[12] Christopher Bollinger, Luis Gonzalez, and James P. Ziliak, “Welfare Reform and the Level and Composition of Income,” in Welfare Reform and its Long-Term Consequences for America’s Poor, James P. Ziliak (ed.), Cambridge University Press, 2009.  Another study, of Connecticut’s time-limited welfare pilot project, found that the program raised income inequality among affected low-income families; families with incomes slightly below the median of affected families lost income, while those slightly above the median gained income.  Marianne P. Bitler, Jonah B. Gelbach, and Hilary W. Hoynes, “What Mean Impacts Miss: Distributional Effects of Welfare Reform Experiments,” American Economic Review, 2006, Vol. 96 No. 4: 988-1012; cited in Moffitt, “Introduction to Volume I.”

[13] Moffitt, “Introduction to Volume I.”

[14] Testimony of Ron Haskins of the Brookings Institution and Annie E. Casey Foundation before the House Subcommittee on Human Resources, Committee on Ways and Means, “Hearing on Challenges Facing Low-Income Individuals and Families,” February 11, 2015,

[15] Max Ehrenfreund, “Bernie Sanders is right: Bill Clinton’s welfare law doubled extreme poverty,” Washington Post, February 27, 2016,

[16] Christopher Jencks, “Why the Very Poor Have Become Poorer,” New York Review of Books, June 9, 2016,

[17] Ehrenfreund.

[18] Speaker Paul Ryan, Weekly Republican Address, June 3, 2016.

[19] Christopher Wimer et al., “Trends in Poverty with an Anchored Supplemental Poverty Measure,” Columbia Population Research Center Working Paper No. 13-01, 2013,

[20] Danilo Trisi, “Safety Net’s Anti-Poverty Effectiveness Has Grown Nearly Ten-Fold Since 1967,” Center on Budget and Policy Priorities, January 4, 2016,  The 42 percent figure, which is for 2014, updates an estimate of 44 percent for 2012.  Trisi calculated these figures, and the 4 percent figure for 1967, from unrounded figures provided by Christopher Wimer.

[21] Arloc Sherman and Danilo Trisi, “Safety Net More Effective Against Deep Poverty Than Previously Thought,” Center on Budget and Policy Priorities, May 6, 2015,

[22] Judith Bartfeld et al., eds., SNAP Matters: How Food Stamps Affect Health and Well-Being, Stanford University Press, 2015; Chuck Marr et al., “EITC and Child Tax Credit Promote Work, Reduce Poverty, and Support Children’s Development, Research Finds,” Center on Budget and Policy Priorities, updated October 1, 2015,; Diane Whitmore Schanzenbach, Lauren Bauer, and Greg Nantz, “Twelve Facts about Food Insecurity and SNAP,” Brookings Institution, Hamilton Project Economic Facts, April 21, 2016,; Danilo Trisi, “Chart Book:  Accomplishments of the Safety Net,” Center on Budget and Policy Priorities, updated November 17, 2015,