Skip to main content

Examining the Safety Net

Testimony of Robert Greenstein, President,
Center on Budget and Policy Priorities,
Before the Human Resources Subcommittee
of the House Committee on Ways and Means

Thank you for the invitation to testify today.  I am Robert Greenstein, President of the Center on Budget and Policy Priorities, a policy institute that focuses on fiscal policy and policies and programs to assist people with low or moderate incomes.  I also served many years ago as Administrator of the Food and Nutrition Service at the U.S. Department of Agriculture, the agency that operates the Supplemental Nutrition Assistance Program (SNAP) and other domestic food assistance programs.  In the mid-1990s, I served as a commissioner on the Bipartisan Commission on Entitlement and Tax Reform, chaired by then-Senators John Danforth and Bob Kerrey.

Overview of Testimony

My testimony today makes the following three points.

  1. Safety net programs that assist people with low or moderate incomes can and should be improved.  But they are far more effective than is often understood.

    Data that measure poverty in the way that most analysts across the political spectrum favor —by counting assistance like SNAP (food stamps), the Earned Income Tax Credit (EITC), and rental subsidies rather than ignoring them as though they didn’t exist or had no impact on the resources a low-income family has to live on — show that the safety net cuts poverty roughly in half.  The data also show that, as a result of safety net programs, poverty barely increased in the Great Recession despite the worst economic downturn in decades.  Further, the data show that poverty is substantially lower today than it was a half century ago, largely as a result of safety net programs.

    In addition, recent advances in poverty research have enabled researchers to track children over several decades as they grow into adulthood.  These studies have found striking evidence that providing income-related benefits like SNAP, the EITC, and other income assistance to poor families with children results in increased test scores and educational attainment for poor children — and subsequently in increased employment and earnings in adulthood.  In other words, various programs that help poor families with children meet basic necessities also improve children’s longer-term outcomes.

  2. The cost of the universe of means-tested programs has grown over the past decade.  But this growth has been due overwhelmingly to two factors:  the Great Recession and the ensuing sluggish recovery; and a substantial growth in the cost of health care programs.  The latter reflects the rise in costs throughout the U.S. health care system (including private-sector health care), the aging of the population (since older people have far higher average health care costs than younger people), and the expansion of Medicaid and premium tax credits to cover more of the uninsured.

    Some policymakers, pundits, and others assume that means-tested programs in general are expanding rapidly and exploding in cost.  Budget data from the Congressional Budget Office (CBO) show that this is not the case; a very different picture emerges once one looks at means-tested programs outside health care.  Total federal spending for programs outside health care that are focused on people with low or modest incomes (hereafter referred to as “low-income programs”) peaked at 2.9 percent of gross domestic product (GDP) in 2010 but has declined as a share of GDP since then and continues to do so.  It stood at 2.3 percent of GDP in 2015.  And, of particular note, it is projected to decline to 1.9 percent of GDP in 2020 and 1.7 percent of GDP in 2025 — levels that are significantly below the average for the past 40 years.  From 1975 to 2014, federal spending on low-income programs outside health averaged 2.1 percent of GDP, which was higher than spending on these programs is expected to be in all years after 2017.

    Similarly, if we look at spending on low-income programs as a share of the budget, federal expenditures for low-income programs outside health care averaged 11.2 percent of federal non-interest spending over the four decades from 1975-2014, and climbed to 13 percent of spending in 2010.  But these costs are slated to fall to 10.4 percent of the budget in 2020 and 9.0 percent in 2025 — well below their average share over the past 40 years, and in fact, the lowest share since 1970.

  3. Coordination across various low-income programs should be strengthened.  Various states have made important improvements in this area in recent years (including through the Work Support Strategies demonstration, which uses innovative ways to better integrate procedures across programs such as SNAP, Medicaid, and child care assistance), but more can be done.

    There is a difference, however, between better coordinating and integrating various procedures and other program elements across these programs and effectively ending key programs and merging their funding into large block grants.  The former approach (strengthening coordination) can boost program effectiveness and efficiency.  The latter approach, by contrast, is likely to result in a marked reduction in program effectiveness in reducing poverty and in increases in poverty and hardship, as I explain below.

I also would note that over the past two decades, the nation has moved to much more of what analysts call a “work-based safety net.”  The degree to which government programs provide benefits to people without earnings has declined substantially, contributing to an increase in “deep poverty” (income below half of the poverty line).  In 1996, for every 100 families with children in poverty, 68 received cash assistance through Aid to Families with Dependent Children; today, only 23 of every 100 poor families with children receive cash assistance through Temporary Assistance for Needy Families (TANF).  Meanwhile, we have expanded the EITC and the low-income component of the Child Tax Credit, both of which go only to families that work.  For these and other reasons, it’s best not to use the term “welfare” to characterize the universe of means-tested programs, since welfare conveys or implies benefits for people who aren’t working, while our country’s safety net increasingly has become one that supports families that work but have low incomes and that supplements their low wages or purchasing power.

I turn now to a more detailed examination of the three main points.

I. The Safety Net’s Impact on Poverty

The U.S. safety net cuts poverty roughly in half.  Highly regarded research by a team of Columbia University researchers, using a broad measure of poverty that most analysts favor over the official poverty measure because it counts rather than ignores non-cash benefits and tax credits, shows that safety net programs reduced the number of otherwise-poor people by 44 percent in 2012.  Without the income from these programs, the poverty rate in 2012 would have been 29 percent; instead, it was 16 percent.

The figures for 2014 are similar.  Before government benefits and taxes, the poverty rate stood at 27.3 percent.  After benefits and taxes, the poverty rate was 15.3 percent, according to the Census Bureau.  (See Figure 1.)  The safety net lifted 38 million people out of poverty last year.

Moreover, these figures — from the Supplemental Poverty Measure (or SPM)[1]understate the safety net’s effectiveness, because they rely on Census data that substantially undercount the number of households that receive certain benefits like SNAP.  Census and other surveys rely on participants’ recollections months later and typically fail to capture some income and benefits. 

The Urban Institute has developed a highly regarded model that analysts across the political spectrum use to correct for the underreporting of government benefits.  In 2012, the most recent year for which these corrections are available, the safety net lowered the SPM poverty rate by an additional two percentage points once the missing SNAP and other benefits are accounted for; the SPM poverty rate for that year was 29.1 percent before government benefits and taxes are taken into account, but 13.8 percent when benefits and taxes are fully accounted for.  (See Figure 2.  For data showing the impact of benefits and taxes on poverty in each state, see the Appendix.)  This suggests that the safety net actually lowers poverty by slightly more than half.  New research led by Bruce Meyer of the University of Chicago that adjusts for the undercounting of benefits in the Census data produces similar results.

A comparison with the 1960s is also telling.  In 1967, the safety net lifted only 4 percent of otherwise poor people out of poverty, as compared with nearly half today.  This is the main reason that the Columbia researchers found that the poverty rate fell from about 26 percent in the late 1960s to about 16 percent in 2012, using the “anchored” SPM.  (These figures do not adjust for underreporting of benefits.[2])  The researchers noted that if the poverty rate is measured by counting only market-based income (i.e., income before government benefits and taxes), it would have been about the same in 2012 as in 1967 — 27 percent in 1967 and 29 percent in 2012.  The fact that the poverty rate before benefits and taxes was no lower in 2012 than in 1967, but the poverty rate after benefits and taxes was sharply lower in 2012 than in 1967, indicates that virtually all of the progress in reducing poverty since the late 1960s has been due to the development of a broader safety net.

(Note:  Even with these impacts, the U.S. safety net does significantly less to reduce poverty than the tax and transfer policies of most other developed countries.  A recent comparison shows that, in 2011, the U.S. poverty rate before counting benefits and taxes was about at the average pre-tax, pre-transfer poverty rate for the 33 OECD countries.  But when poverty is measured after counting benefits and taxes, the U.S. poverty rate ranked seventh highest among these nations.[3])

Of course, these figures do not reflect the likelihood that employment would have been somewhat higher, and market-income poverty somewhat lower, in the absence of the safety net.  The effect of the safety net on work has been well studied, however, and the leading academic research on this matter finds that the safety net’s overall effects in reducing employment and thereby keeping poverty higher are very small,[4] especially since some safety net programs significantly increase work effort.

Research shows, for example, that the EITC and the low-income component of the Child Tax Credit draw people into the labor market and increase work.  Research has found the EITC to be particularly effective at increasing work effort among single mothers;[5] it is widely considered to be one of the most effective policies for increasing work and earnings among female-headed families and has been a key factor behind the substantial increases in work among single mothers since in the early 1990s.  (In fact, University of Chicago economist Jeffrey Grogger found that the EITC expansions of the 1990s “appear to be the most important single factor in explaining why female family heads increased their employment over 1993-1999,” with an even larger impact on employment than changes in welfare policies.[6])

The Impact of the Safety Net in the Recent Recession

Poverty trends during the recent recession underscore the safety net’s importance.  Various safety net programs expanded automatically (on a temporary basis) in the recession, as they are designed to do when the economy turns down and joblessness rises.  In addition, policymakers enacted temporary expansions of some safety net programs in the 2008 stimulus bill and the 2009 Recovery Act.  The combination of the ongoing counter-cyclical features of key safety net programs and these temporary measures averted a steep increase in poverty despite the severe economic downturn.  Counting only market income, the poverty rate climbed from 23.6 percent in 2007 to 28.1 percent in 2010, according to the Columbia research team.  But counting both market income and government income-support programs and taxes, the anchored SPM rose by less than one percentage point, from 14.7 percent to 15.3 percent — a very modest increase given the depth of the downturn.

Some have noted, referring to the official poverty rate (which doesn’t count SNAP, the EITC, rental vouchers or the like), that the number of people in poverty was considerably higher in 2014 than in 2007 — and have concluded that this means the safety net hasn’t been working effectively.  Such a conclusion is unwarranted.  It is not appropriate to use the official poverty rate to assess the safety net’s performance over this period since virtually all of the expansions in safety net programs after 2007 occurred in programs whose benefits aren’t counted in the official poverty rate.  When the SPM is used, the increase in the poverty rate between 2007 and 2014 is only about one percentage point.

And such an increase in the poverty rate as compared to 2007 would be expected, because the economy remained much weaker in 2014 than in 2007, the year in which the prior economic recovery reached its peak.  For example, median family income was $1,100 lower in 2014 than in 2007, hardly a factor attributable to the safety net.  Similarly, median hourly wages were lower in 2014 than in 2007, and the unemployment rate was higher.  In fact, the number of long-term unemployed workers was nearly double in 2014 what it was in 2007.

Research on the Impact of Income Support on Poor Children

Traditionally, safety net programs have been regarded largely as measures to alleviate current poverty and hardship.  Over the past few years, however, evidence has mounted from an array of important academic studies showing that basic income support for families with children — including tax credits and vouchers to help purchase food or pay rental costs — has significant long-term positive effects on poor children in areas such as school performance, high school completion, and labor-market outcomes in adulthood.

To note just a few such studies, economists Gordon Dahl and Lance Lochner have found that children affected by a major EITC expansion in the 1990s exhibited improved math and reading test scores.[7]  Harvard economist Raj Chetty and his colleagues similarly found that additional EITC and Child Tax Credit income is linked to increases in students’ test scores; they concluded, based on those data and the strong links between educational attainment and future earnings, that EITC receipt in childhood leads to significantly higher earnings and employment rates when the children become adults.[8]  They estimated that each dollar of income provided through these credits increases a child’s future earnings by more than one dollar.

Gordon Berlin, the president of MDRC, one of the nation’s premier evaluation and research organizations, has noted that based on a review of rigorous random-assignment evaluations and other research, “We have reliable evidence involving thousands of families in multiple studies demonstrating that ‘making work pay’ [through cash supplements to low earners] causes improvements in young children’s school performance.”[9] 

A landmark study of the impact of food stamp benefits on children provides further evidence of the long-term impacts of income support and similar assistance.  The researchers, led by economist Hilary Hoynes of the University of California at Berkeley, were able to use the uneven roll-out of the Food Stamp Program in the early 1970s to compare poor children who had access to food stamps in the early 1970s with comparable poor children from counties that hadn’t yet instituted the program.  They examined records of these children as the children grew up and into adulthood, and found that access to food stamps in early childhood (and the prenatal year) was associated with an 18 percentage-point increase in high school completion rates, lower rates of metabolic syndrome (obesity, high blood pressure, heart disease, and diabetes) in adulthood, and among girls, increases in “self-sufficiency” (using an index of education, earnings, income, and decreases in welfare receipt in adulthood).[10]

Consistent with these and other studies, economists Greg Duncan and Katherine Magnuson have estimated that a sustained $3,000 increase in the income of low-income families with young children results in an average of 17 percent higher earnings and 135 more hours of work per year when the children become adults.[11]  An additional 135 hours of work represents nearly a third of the gap in adult work hours between children raised in poor families and children raised in families whose incomes exceeded twice the poverty line.

Exactly why income-support and related programs yield these benefits is not yet clear.  Duncan, Magnuson, and others are launching a demonstration project to examine these issues.  A number of researchers believe, however, that part of the answer may lie in another important set of research findings that have emerged in recent years — those relating to “toxic stress” in poor families and brain development in children.

A growing body of evidence indicates that children living in unusually stressful situations may experience chronic stress levels severe enough to damage the developing neural connections in their brains and to impede their ability to succeed in school and develop the social and emotional skills needed to function well as adults.  These stressful situations include living in dangerous neighborhoods, in families with difficulty putting food on the table or paying the rent, or with parents who cannot cope with their daily lives.  One study documented that a 17-year-old’s working memory “deteriorated in direct relation to the number of years the children lived in poverty (from birth through age 13).”[12]  Another study found slower brain growth in MRI scans among poor children between the ages of five months and four years.[13]  Still another study found that temporary spells of low income during pregnancy tend to come with a rise in the mother’s stress hormone cortisol to levels associated with negative child outcomes, including “a year less schooling, a verbal IQ score that is five points lower and a 48 percent increase in the number of chronic conditions” for the exposed children, as compared to siblings born at times when the family had lower stress and, usually, higher income.

Other researchers have concluded that early-onset arthritis, hypertension, and other conditions previously linked with early childhood stressors may explain a not-insignificant fraction of the lower work hours and earnings of adults who were raised in poverty.[14]

In short, programs that help families with children afford basic necessities appear to improve children’s longer-term outcomes, in part by reducing the added stress that parents or children may experience if they can’t pay their bills or put food on the table — and thereby easing the negative effects that poverty-related stress can have on children’s brain development.  More research in this area is vital and may provide important policy-relevant findings.

II. Low-Income Programs and the Nation’s Fiscal Challenges

As we all know, the nation faces long-term fiscal challenges.  These challenges stem primarily from projected increases in the cost of health and retirement programs, an insufficient revenue base, and the effect of this mismatch on interest costs in future decades.  Yet low-income programs outside of health care are not driving or contributing to the nation’s long-term fiscal problems, as a review of the budget data shows.

To be sure, low-income program spending grew substantially between 2007 and 2010 in response to the severe economic downturn, helping to mitigate its worst effects.  But since peaking in 2010 and 2011, federal spending on low-income programs outside health care has fallen considerably as a share of GDP and is projected to continue falling as the economy recovers more fully.  Based largely on CBO estimates, by 2018 it will fall below its average over the past 40 years (from 1975 to 2014) and continue declining as a share of GDP after that.

Specifically, federal spending for low-income programs outside health care (including expenditures for refundable tax credits such as the EITC) climbed from 1.9 percent of GDP in 2007 to a peak of 2.9 percent of GDP in fiscal years 2010 and 2011.  But since then, such spending has dropped to an estimated 2.3 percent of GDP in 2015 and is projected to drop back to its prior 40-year average of 2.1 percent by 2017 — and then to fall further to 1.7 percent of GDP by 2025.  (See Figure 4.)

Low-Income Outlays Outside Health: Percent of GDP
  Avg, 1975-2014 In 2010 In 2015 In 2020
In 2025
Discretionary 0.8% 0.9% 0.7% 0.6% 0.5%
Mandatory 1.3% 1.9% 1.6% 1.4% 1.2%
TOTAL 2.1% 2.9% 2.3% 1.9% 1.7%

Source: CBPP calculations of Congressional Budget Office and Office of Management and Budget data Note: May not add due to rounding

Groups of programs that are declining rather than rising as a share of GDP — and are falling below their average costs as a share of GDP in prior decades — are not contributing to the nation’s long-term fiscal challenges. 

We can also analyze trends in spending for these programs other than by measuring their costs as a share of GDP, such as by looking at spending on means-tested programs outside health care as a share of total federal non-interest spending (i.e., spending for everything except interest payments on the debt).  The same pattern shows up here:  outlays for these programs peaked at 13 percent of federal non-interest spending in 2010 and 2011, but have since declined.  They are projected to fall to their prior 40-year average of 11.2 percent by 2017, and then fall below the 40-year average.  By 2025, these expenditures are projected to drop to 9 percent of the budget — one-fifth lower than their average over the four decades from 1975 to 2014, and their lowest share since 1970.

Low-Income Outlays Outside Health: Percent of Total Program Outlays
  Avg, 1975-2014 In 2010 In 2015 In 2020 (baseline) In 2025 (baseline)
Discretionary 4.3% 4.3% 3.4% 3.0% 2.7%
Mandatory 6.9% 8.7% 8.5% 7.4% 6.4%
TOTAL 11.2% 13.0% 11.9% 10.4% 9.0%

Source: CBPP calculations of Congressional Budget Office and Office of Management and Budget data

Still another way to measure expenditure trends for these programs is to look at their cost over time in dollars, adjusted for inflation and U.S. population growth.  In 2005, expenditures for means-tested programs outside health care stood at $365 billion (measured in 2016 dollars and adjusted for the size of the U.S. population).  In 2025, these expenditures, measured the same way, are projected to be $354 billion, about 3 percent less than in 2005.

What About Health Programs?

Health care programs constitute a very large share of overall low-income spending.  As a result, increases in health care spending as a share of GDP (and as a share of the budget) can mask declines in the rest of the means-tested program universe.  This is one reason that health care should be considered as its own category.  There are other reasons as well.

First, a very large share of means-tested health spending is for seniors and people with disabilities, not children and non-elderly adults.  Although seniors and people with disabilities made up only 22 percent of Medicaid beneficiaries in 2014, 56 percent of Medicaid spending went for care for these groups, a share that will rise as the U.S. population ages.  Providing care to a frail elderly person or a person with a disability generally costs much more than the typical cost of care for a child or a working-age parent.

Moreover, long-term care alone constitutes 25 percent of all Medicaid costs and a larger share of Medicaid costs for seniors and people with disabilities.  Yet a substantial share of Medicaid spending on long-term care is for seniors who had middle-class incomes for much of their working lives but whose long-term care needs now exceed their ability to pay for that care.  In 2015, private nursing home care is projected to average $91,250 per year, assisted living facility costs to average $43,200 per year, and home health aide services to average $20 per hour.  In 2009, the average long-term care cost for a Medicaid beneficiary receiving such care was $34,579, a figure certain to be higher today.

It should be recognized that high and rising health care costs are not caused by Medicaid’s means-tested nature.  To the contrary, health care costs are rising throughout the U.S. health care system, and costs per beneficiary are substantially lower in Medicaid than in Medicare or private-sector health care (largely because Medicaid pays health providers lower rates).

Still another factor that plays a large role in raising costs throughout the health care system is medical advances that prolong life — especially in old age — and improve health, but add considerably to costs.  In addition, since older people incur about five times higher average health care costs than younger people do, the aging of the U.S. population is pushing up costs for federal health care programs such as Medicaid.

The nation will have to do more in the years and decades ahead to slow rising health care costs.  These are fundamentally issues related to health care and health care delivery, however, rather than issues related to the nature of low-income assistance.

One last point on this matter.  As noted, spending on low-income health care programs is projected to rise as a share of GDP.  Yet spending on low-income programs other than health programs is actually expected to decline somewhat faster.  As a result, spending on low-income programs as a whole — including health — is projected to decline somewhat as a share of the economy by 2025, relative to its current level.  After peaking at 4.9 percent of GDP in 2010, total spending on all low-income programs (including health) receded to about 4.5 percent of GDP in 2015 and is projected to equal 4.2 percent of GDP in 2025.  (Note:  These figures include the costs of tax-credit subsidies under the Affordable Care Act to make health care affordable to people with incomes up to 400 percent of the poverty line.)

Spending on Low-Income Programs (percent of GDP)
  1975-2014 average 2010 2015 2025
Low-income programs, excluding health 2.1% 2.9% 2.3% 1.7%
Low-income health programs 1.1% 2.0% 2.2% 2.5%
Total, all low-income programs 3.2% 4.9% 4.5% 4.2%

Source: CBPP calculations of Congressional Budget Office and Office of Management and Budget data

III. Improving Coordination Across Programs

Coordination across various means-tested programs can — and should — be strengthened.  Lack of adequate coordination can have deleterious effects.  It can make it harder for poor families to access benefits or services that they qualify for and could be helped by.  It can be especially hard on working-poor parents if they have to take time off from work to make repeated visits to program offices to provide the same income and other information over and over again for different programs.  And it can reduce efficiency and increase administrative costs for state and local governments.

Fortunately, a number of states have made significant progress in improving program coordination in recent years.  And promising developments — especially advances in information technology — offer the potential for significant further progress.

In particular, a number of states — including the six states that participate in the Work Support Strategies demonstration project (which include a mix of  “red” and “blue” states) — have been taking important strides to better integrate various procedures related to major safety net programs such as Medicaid, SNAP, and child care subsidies.  These states are improving coordination related to intake, verification, periodic redetermination of eligibility, and other matters in order to create a more coherent and easy to navigate package of work supports.  The states believe a more integrated safety net will strengthen families’ financial stability, better support work, improve efficiency and program integrity, and reduce administrative costs.   These states are working to take more of a family-centered approach to service delivery rather than a program-by-program approach, and their efforts have resulted in important progress. 

One factor that has helped make progress possible is that states turn out to have more flexibility to better align and coordinate various programs than most state or federal officials had realized until recently.  Some years ago, a group of state officials working with a National Governors Association project began to compile lists of various things they would like to do to better coordinate programs but that federal rules prohibited.  To the surprise of most state officials involved, it turned out that states could do a large share of the things that states thought federal rules barred them from doing.  But hardly anyone — including federal officials — was aware of that, because securing the needed flexibility for a state required extensive technical expertise in an array of federal rules and guidance governing disparate programs like SNAP and Medicaid, as well as knowledge of how to assemble a coordinated package of state options and waivers covering the different programs.  Not only did few state officials have the necessary expertise, but few federal officials did either, because they tended to know the rules for their own agency’s programs but not the details of various rules for other means-tested programs.

Fortunately, this is now changing.  The Work Support Strategies states, the efforts of a number of other states, and the growing conversation among states on these matters are showing how greater coordination efforts can be undertaken.  In addition, federal officials have become somewhat more knowledgeable about coordination strategies.  To facilitate further progress, a task force of state health and human services officials recently developed a series of proposals for administrative actions that federal agencies can take to eliminate various barriers to increased coordination.  The task force’s recommendations are now under consideration by federal agencies.

Of particular importance are data sharing and other advances that developments in information technology make possible.  In the old paper-bound systems, it was often difficult or administratively burdensome for state agencies to share household information across programs.  Now, many states are able to do so.  For example, most states now use document imaging systems to save and file household verifications.  Many states also now provide call centers that clients can contact by phone to report a change in their circumstances and need for benefits.  These and other technological improvements can make it easier for participants and state caseworkers to use a single process to update information on a household’s income and circumstances that is used to determine the household’s eligibility or benefit levels across various programs.[15]

When state agencies can spend less staff time processing eligibility as a result of better data-sharing across programs, they can redeploy the saved resources to important tasks like connecting families to work and supporting families who need more intensive supports.  Similarly, many states believe families can better spend time looking for work (or remaining at the job) when they don’t need to spend hours every few months (or more often than that) at local human services offices standing in line or filling out forms with the same information as they recently suppled to another program.

Another promising development is Congress’ reauthorization last year of the Workforce Innovation and Opportunity Act (WIOA).  In doing so, Congress sent a strong signal that people in the greatest need should be served under WIOA to a much greater degree than previously, and that this should be accomplished in ways that are well coordinated with TANF.  For example, WIOA includes new “priority of service” provisions requiring that when state or local agencies use WIOA Title I Adult funds to provide career and training services, they should give priority to serving “public benefits recipients, other low-income individuals, and individuals who are basic skills deficient.”  Prior to enactment of this legislation, the workforce training infrastructure largely bypassed TANF recipients; in 2013, fewer than 4 percent of WIA participants were people receiving TANF cash assistance.  That now should change. 

Certain reforms in TANF could facilitate further progress in WIOA/TANF coordination.  TANF work requirements often aren’t well aligned with WIOA programs, and this can discourage the collaboration and priority of service that Congress intended in passing WIOA.  In particular, states can be discouraged from allowing TANF recipients to participate in WIOA programs because such participation won’t necessarily “count” for TANF Work Participation Rate purposes.  For example, the TANF Work Participation Rate doesn’t recognize participation in many basic education and skills programs as a stand-alone activity (since these aren’t considered “core” activities under TANF).  And the exact hours of the WIOA program may not match the TANF hourly requirements.  We recommend that Congress deem participation in WIOA programs as counting toward the TANF Work Participation Rate, which would facilitate better integration of TANF and WIOA and allow TANF recipients fuller access to WIOA programs and services. 

In short, federal policymakers and program managers can do more to promote coordination.  The federal-state partnership will be further strengthened if more federal agency staff develop more in-depth expertise in other federal assistance programs.  State and local governments ought not be left on their own to disentangle the differing federal rules across programs and figure out how to navigate them to strengthen coordination.  I should note that FNS, CMS, and ACF have recently taken steps in this area, but more can be done.

What Not to Do

What would not be sound is to take a more radical course by eliminating key safety net programs and merging their funding into mega-block grants.  Such a course would likely prove counter-productive.  It likely would result in increased poverty and hardship over time, for several reasons.

First, it would entail converting programs like SNAP that respond automatically and immediately to changes in need, such as during economic downturns, into fixed grants to states that do not rise when the economy falters and poverty increases (and, similarly, do not decline automatically when the economy experiences robust growth and poverty recedes).

Second, overall resources for low-income assistance would likely decline.  History shows that when policymakers consolidate various federal programs into a block grant whose funds can be used for a broad array of purposes, states often substitute some of the federal block-grant dollars for state dollars that they previously were spending on those purposes, thereby shrinking the total pool of federal and state resources devoted to those purposes (and freeing up money for state officials to plug state budget holes).  Although maintenance-of-effort requirements can reduce the scope of this problem, they cannot eliminate it, as such requirements are notoriously difficult to enforce.  This is evidenced by state substitution of federal TANF funding for state funding despite the 1996 welfare law’s maintenance-of-effort requirement, as the Government Accountability Office has documented.[16]  (Georgia is a case in point.  Under the welfare law, it has a $173 million maintenance-of-effort obligation, but it has used maneuvers that aren’t prohibited to count non-government spending toward this obligation — and to withdraw $99 million a year in state funding and shift it to other areas of the state budget not related to helping low-income families become self-sufficient or meet basic needs.) 

Third, in part because funds from broad federal block grants tend to be used in diffuse ways, the effect of such funds is hard to document, and federal funding for such block grants tends to erode markedly over time.  A forthcoming Center on Budget and Policy Priorities analysis examines the 13 major health, housing, and social services block grants created in recent decades.  Funding for the vast majority of these block grants — 11 of the 13 — has shrunk in inflation-adjusted terms since the block grants’ inception, in some cases dramatically.  Funding for the median block grant has declined by 25 percent since its inception, after adjusting for inflation.  Funding for four of the block grants has dropped by more than 60 percent.

The forthcoming analysis also examines changes in funding for these block grants from 2000 to 2015.  Here, as well, 11 of the 13 block grants suffered funding reductions.  Moreover, combined funding for the 13 block grants plunged 28 percent over the 2000 to 2015 period, after adjusting for inflation.

These funding declines occurred even as need increased.  The U.S. population has grown by 14 percent since 2000, and the number of people living in poverty has increased as well, as the economy hit headwinds and wages stagnated.


*   *   *   *   *

This concludes my testimony.  Thank you again for the opportunity to testify here today and to provide my thoughts on these issues.


These state-by-state figures average data for four years (2009-2012) for increased reliability.  They are SPM figures that correct for the underreporting of income.  The data combine the effects of federal and state policies, but the vast majority of the reduction in poverty is due to the effect of federal programs, including refundable tax credits.

Effects of the Tax and Benefit System on Poverty by State
Using the Supplemental Poverty Measure after corrections for underreported income;
average of 2009-2012
  All Ages   Under 18
  Percent in Poverty   Percent in Poverty
State Number Lifted Above SPM Poverty Line Before Counting Taxes and Benefits After Counting Taxes and Benefits   Number Lifted Above SPM Poverty Line Before Counting Taxes and Benefits After Counting Taxes and Benefits
U.S. TOTAL 46,364,000 28.9 13.8   11,924,000 29.8 13.8
Alabama 935,000 31.2 11.5   222,000 31.2 11.7
Alaska 79,000 21.1 9.8   25,000 22.9 9.7
Arizona 998,000 32.4 17.3   290,000 36.0 18.8
Arkansas 601,000 34.8 14.0   138,000 33.0 13.6
California 4,851,000 33.6 20.7   1,443,000 37.3 22.0
Colorado 568,000 23.6 12.3   157,000 24.2 11.7
Connecticut 415,000 22.7 10.9   84,000 20.7 10.3
Delaware 145,000 28.0 11.8   38,000 29.1 11.1
Dist of Columbia 82,000 32.8 19.3   32,000 47.0 18.8
Florida 2,892,000 33.2 17.8   582,000 32.9 18.5
Georgia 1,401,000 30.6 16.2   414,000 32.4 16.3
Hawaii 224,000 30.3 13.1   72,000 33.9 10.3
Idaho 283,000 27.9 9.6   81,000 28.0 9.0
Illinois 1,745,000 27.1 13.5   474,000 29.2 13.9
Indiana 1,060,000 29.0 12.3   260,000 29.5 13.4
Iowa 413,000 20.8 7.0   93,000 18.1 5.2
Kansas 435,000 25.2 9.6   116,000 26.4 10.2
Kentucky 809,000 30.0 11.2   199,000 29.9 10.2
Louisiana 792,000 32.2 14.4   221,000 33.0 13.8
Maine 236,000 26.5 8.6   42,000 23.0 7.3
Maryland 535,000 21.5 12.3   137,000 21.5 11.4
Massachusetts 918,000 26.3 12.3   199,000 25.3 11.4
Michigan 1,756,000 29.2 11.2   371,000 27.4 11.3
Minnesota 670,000 21.5 8.7   160,000 20.4 7.8
Mississippi 593,000 35.2 14.7   163,000 35.0 13.8
Missouri 1,054,000 28.4 10.7   245,000 27.5 10.2
Montana 172,000 27.7 10.2   36,000 25.3 8.8
Nebraska 227,000 20.9 8.4   57,000 19.9 7.5
Nevada 345,000 30.3 17.4   92,000 32.9 19.1
New Hampshire 137,000 19.7 9.3   22,000 15.5 7.5
New Jersey 1,036,000 25.0 13.1   244,000 25.0 13.0
New Mexico 377,000 31.9 13.2   112,000 33.5 11.9
New York 3,062,000 31.0 15.1   839,000 34.4 15.1
North Carolina 1,676,000 30.4 12.6   427,000 30.8 12.5
North Dakota 73,000 18.7 7.7   15,000 15.9 5.8
Ohio 1,969,000 27.9 10.6   461,000 26.6 9.4
Oklahoma 620,000 27.3 10.5   172,000 28.5 10.2
Oregon 641,000 29.2 12.5   156,000 29.1 10.9
Pennsylvania 2,012,000 26.7 10.7   375,000 23.1 9.4
Rhode Island 176,000 28.4 11.5   43,000 30.3 11.1
South Carolina 854,000 31.3 12.7   204,000 30.3 11.5
South Dakota 122,000 23.8 8.7   30,000 22.1 6.9
Tennessee 1,176,000 31.6 13.1   293,000 30.9 11.1
Texas 3,691,000 28.9 14.3   1,246,000 33.0 15.2
Utah 314,000 20.6 9.5   94,000 19.8 9.2
Vermont 96,000 23.7 8.2   18,000 21.6 7.2
Virginia 781,000 21.4 11.5   190,000 21.2 11.2
Washington 1,038,000 25.4 10.1   280,000 26.4 8.9
West Virginia 383,000 31.1 10.0   70,000 26.9 9.1
Wisconsin 833,000 24.0 9.2   180,000 22.4 8.7
Wyoming 61,000 19.3 8.2   13,000 14.9 5.5

Government benefits are Social Security, unemployment insurance benefits, workers compensation, veterans’ benefits, TANF, SSI, SNAP, school lunch, WIC, rent subsidies, higher education grants, general assistance, and home energy assistance. The tax system is federal and state income and payroll tax owed, net of federal or state tax credits received such as the Earned Income Tax Credit (EITC) or Child Tax Credit.

Source: CBPP analysis of U.S. Census Bureau's March 2010-2013 Current Population Survey; Supplemental Poverty Measure public use files; and HHS/Urban Institute TRIM baseline microdata files.

End Notes

[1] Unlike the official poverty measure, which measures only a family’s cash income before taxes, the SPM counts after-tax income plus non-cash benefits less various medical and work expenses.  It thus better captures the large and growing portion of government support that is not in the form of cash and thus is missed by the official poverty measure. 

[2] The “anchored” SPM is a variation of the SPM that compares poverty rates in different years using a 2012 SPM poverty threshold adjusted only for inflation.  Christopher Wimer et al., “Trends in Poverty with a Supplemental Poverty Measure,”

[3] CBPP analysis of OECD data at

[4] Yonatan Ben-Shalom, Robert Moffitt, and John Karl Scholz, “An Assessment of the Effectiveness of Anti-Poverty Programs in the United States,” National Bureau of Economic Research, 2011, (also available in the Oxford Handbook on the Economics of Poverty, Oxford University Press, 2012); see also Robert Moffitt, “The Deserving Poor, the Family, and the U.S. Welfare System,” presidential address to the Population Association of America, May 4, 2014; and Robert Moffitt, “The U.S. Safety Net and Work Incentives:  The Great Recession and Beyond,” in Journal of Policy Analysis and Management, Spring 2015, vol. 34, no. 2.

[5] Chris M. Herbst, “The labor supply effects of child care costs and wages in the presence of subsidies and the earned income tax credit,” 2009,

[6] 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.  Using different data, in another study, Grogger reaches similar conclusions.  Jeffrey Grogger, “Welfare Transitions in the 1990s: the Economy, Welfare Policy, and the EITC,” NBER Working Paper No. 9472, 2003,

[7] Gordon Dahl and Lance Lochner, “The Impact of Family Income on Child Achievement: Evidence from the Earned Income Tax Credit,” American Economic Review 102, no. 5 (2012): 1927–1956.

[8] Raj Chetty, John N. Friedman, and Jonah Rockoff, “New Evidence on the Long-Term Impacts of Tax Credits,” Statistics of Income Paper Series, Internal Revenue Service, November 2011.  For additional studies finding that the EITC increases college attendance, see Dayanand S. Manoli and Nicholas Turner, “Cash-on-Hand & College Enrollment: Evidence from Population Tax Data and Policy Nonlinearities,” NBER Working Paper No. 19836, 2014,; and Michelle Maxfield, “The Effects of the Earned Income Tax Credit on Child Achievement and Long-Term Educational Attainment.” Michigan State University Job Market Paper, November 14, 2013, Maxfield %20EITC%20Child%20Education.pdf. 

[9] Gordon L. Berlin, “Remarks at National Summit on America’s Children,” MDRC, May 22, 2007.

[10] Hilary W. Hoynes, Diane Whitmore Schanzenbach, and Douglas Almond, “Long Run Impacts of Childhood Access to the Safety Net,” NBER Working Paper 18535, 2012,

[11] Greg Duncan and Katherine Magnuson, “The Long Reach of Early Childhood Poverty,” Pathways, Winter 2011, Duncan.pdf.

[12] Gary W. Evans, Jeanne Brooks-Gunn, and Pamela K. Klebanov, “Stressing Out the Poor: Chronic Physiological Stress and the Income-Achievement Gap,” Pathways, Winter 2011, scspi/_media/pdf/pathways/winter_2011/PathwaysWinter11.pdf.

[13] Jamie Hanson et al., “Child Poverty Affects the Rate of Human Infant Brain Growth,” PLoS One (2013),

[14] Anna Aizer, Laura Stroud, and Stephen Buka, “Maternal Stress and Child Outcomes: Evidence from Siblings,” National Bureau of Economic Research Working Paper 18422, 2012,

[15] For more information on how states are leveraging new technology with respect to health and human services programs, including SNAP, see “GAINING GROUND:  A Guide to Facilitating Technology Innovation in Human Services,” Freedman Consulting,

[16] U.S. General Accounting Office, “Welfare Reform:  Challenges to Maintaining a Federal-State Fiscal Partnership,” August 2001,