Did Welfare Reform Increase Extreme Poverty in the United States?

Report Welfare

Did Welfare Reform Increase Extreme Poverty in the United States?

August 21, 2016 11 min read Download Report

Authors: Robert Rector and Jamie Hall

Two decades ago, on August 22, 1996, President Bill Clinton signed the Personal Responsibility and Work Opportunity Act, popularly known as welfare reform, into law. At the time, liberals proclaimed that the bill would slash the incomes of one in five families with children and push 2.6 million people into poverty.[1] Senator Daniel Patrick Moynihan famously predicted that the bill would leave children scavenging in the streets, “sleeping on grates, picked up in the morning frozen.”[2]

In fact, reform cut welfare caseloads by over 50 percent, employment of the least-skilled single mothers surged, and the poverty rates of black children and single-parent families dropped rapidly to historic lows.[3] Doomsday prophets were utterly discredited. Reform was very popular with the public.

Remarkably, 20 years later, Moynihan’s alarm about “children sleeping on grates” has been revived. The left now contends that welfare reform has thrown 3.5 million children into “extreme poverty” of the kind seen in the developing world, living in destitution on less than $2.00 per day. For example, Bloomberg News reports that millions of Americans now have incomes lower than the “disabled beggars of Addis Ababa in Ethiopia.” [4]

These claims of extreme poverty in the U.S. are based on radically defective data. In reality, poverty among single parents, the main group affected by welfare reform, has fallen substantially over the past two decades while it remained constant or rose among groups unaffected by reform.

What Welfare Reform Did

The 1996 reform law replaced the old Aid to Families with Dependent Children (AFDC) welfare program with a new program called Temporary Assistance for Needy Families (TANF). For the first time, a portion of recipients were required to work or prepare for work as a condition of receiving aid. This work requirement led to the historically unprecedented drop in welfare caseloads shown in Chart 1.


Reform was based on the premise that prolonged welfare dependence was harmful to recipients and society. Reformers believed that imposing work requirements on benefits would cause families to rely less on traditional welfare and more on formal employment, informal employment, and support from relatives. It was also anticipated that reform would lead to decreased non-marital childbearing and greater support from non-married fathers and boyfriends, including increased cohabitation and marriage. In general, the reform met these expectations.[5]

Welfare Reform and Income-Based Poverty

Aid to Families with Dependent Children was a cash welfare program primarily for single parents and their children. Typically, 90 percent of AFDC families were single parents and 10 percent were married couples. The impact of welfare reform on poverty can therefore be traced through changes in the poverty rate of single-parent families. The most accurate name for this group would be “non-married families with children” because these families often contain other adults such as grandparents, relatives, boyfriends, and non-married fathers in addition to the single mothers.[6]

Table 1 shows the poverty rate for unmarried families with children based on pre-welfare and post-welfare income.[7] Pre-welfare income includes a broad range of income from earnings, self-employment income, interest, dividends, Social Security benefits, pensions, and unemployment insurance. Post-welfare income includes income from all of these sources plus benefits from six major means-tested welfare programs: AFDC, TANF, food stamps, Supplemental Security Income (SSI), and the refundable components of the Earned Income Tax Credit (EITC) and Additional Child Tax Credit (ACTC).

The table shows changes in the pre-welfare and post-welfare poverty rates from 1995 (the last year before welfare reform) to 2014 (the most recent year for which data are available). For unmarried families with children, the pre-welfare poverty rate (or poverty rate without counting welfare benefits) fell from 38 percent in 1995 to 32 percent in 2005 and then, during the great recession and weak economic recovery, climbed back to 35.1 percent by 2014. The post-welfare poverty rate (or poverty rate after welfare benefits are counted) fell from 27.5 percent in 1995 to 23.2 percent in 2005 and remained at 22.9 percent in 2014. Overall, the pre- and post-welfare poverty rates for non-married parents and their families are substantially lower today than they were before welfare reform began 20 years ago.

The decline in poverty among single parents and their families stands in contrast to the poverty trends of groups unaffected by welfare reform. The poverty rates for married couples with children (a group largely unaffected by AFDC and its reform) fell modestly between 1995 and 2005 but by 2014 had risen above the 1995 levels. Strikingly, the poverty rates of households without children (a group wholly unaffected by welfare reform) have risen sharply over the past 20 years; the post-welfare poverty rate for this group, for example, rose from 13 percent in 1995 to 16.7 percent in 2014.

The overall lesson is clear. Welfare reform was and remains effective in reducing poverty among the group directly affected by the reform: single parents. In distinct contrast, other groups unaffected by reform have experienced increases in poverty over the past two decades.[8]

Spreading Misinformation About “Extreme Poverty”

Remarkably, two decades after the enactment of reform, the left has revived Moynihan’s original claims about “children sleeping on grates.” In their recent book $2.00 a Day: Living on Almost Nothing in America, Kathryn Edin and Luke Shaefer argue that welfare reform has caused a surge in “extreme poverty” in the U.S.[9]

Edin and Shaefer claim that welfare reform has led 3.55 million children (or 4 percent of all families with children) to subsist on less than $2.00 per person per day, which they identify as “one of the World Bank’s measures of global poverty.”[10] According to Edin and Shaefer these families live in “extreme destitution,” regularly selling their blood and collecting aluminum cans to survive.[11]

“Extreme Poverty” and Living Conditions

Does this “extreme destitution” or third world poverty really exist in the U.S.? Edin and Shaefer’s charge is based on the government’s Survey of Income and Program Participation (SIPP), but examination of the SIPP data reveals that families allegedly living in extreme poverty don’t actually appear to be poor at all. According to the SIPP data, some 86.5 percent of families with children apparently living in extreme poverty in the U.S. have air conditioning in their homes or apartments; 89 percent have cell phones; 88 percent have a DVD player, digital video recorder, VCR, or similar device, and 67 percent have a computer. [12] (See Chart 2.)

Only 1 percent of families supposedly in extreme poverty report that they “often” did not have “enough food to eat” over the previous four months; another 8 percent said that they “sometimes” did not have “enough to eat.” The remaining 91 percent report that they “always” had enough food to eat. Despite having alleged incomes of less than $2.00 per day, only 1 percent of these families were evicted during the prior 12 months; 4 percent had their oil, gas, or electricity cut off.


Claims of Widespread “Extreme Poverty” the Result of Defective Data

The “extreme poverty” or “extreme destitution” reported by Edin and Shaefer is almost nonexistent in the United States. It is a statistical fiction based on severe undercounting of earnings and benefits among lower-income Americans.

Edin and Shaefer obtain their sensational claim that 3.55 million children in 1.6 million families live in “extreme destitution” on $2.00 per day by the simple tactic of excluding nearly all welfare benefits from their income count.[13] While they acknowledge that counting the EITC and food stamps drops the number by two-thirds, down to 1.17 million children and 612,000 families,[14] even these lower figures still reflect a massive miscount of income. Edin and Shaefer count a family in “extreme poverty” or destitution if its apparent income in any month falls below $2.00 per person per day, but each month, the SIPP survey (used by Edin and Schaefer) undercounts some 22 million beneficiaries in programs targeted at low-income persons:

  • The average monthly participation in the Supplemental Nutrition Assistance Program (SNAP) was 44.7 million,[15] but SIPP shows only 31.2 million,[16] a deficit of 13.5 million recipients per month.
  • The average monthly participation in the Special Supplemental Nutrition Program for Women, Infants, and Children (WIC) was 8.9 million.[17] SIPP historically has undercounted WIC participation by a third,[18] which implies that 3 million recipients were not counted in SIPP each month.
  • The average monthly participation in Temporary Assistance to Needy Families was 4.6 million,[19] but SIPP shows only 2.6 million,[20] a deficit of 2 million recipients per month.
  • An average of 7.7 million individuals received Unemployment Insurance (UI) benefits each week.[21] SIPP has historically undercounted UI benefits by about 25 percent,[22] which implies that roughly 2 million recipients were not counted in SIPP each week.
  • Monthly participation in the Supplemental Security Income (SSI) program in 2011 was 8.1 million, but SIPP shows only 7.1 million,[23] a deficit of 1 million recipients per month.

Remarkably, Edin and Schaefer have concluded that 1.17 million children live in families that lack both earnings and welfare each month, using survey data that omit benefits distributed to some 20 million recipients each month. Their claim of “extreme poverty” reflects huge holes in their survey instrument rather than conditions in the real world. They are like an individual looking at the world through eyeglasses with cracked lenses and proclaiming that the sky is full of dangerous cracks and fissures and is about to fall.

Disparities Between Expenditures and Apparent Income

Poverty experts recognize that government income surveys, such as the SIPP, substantially underreport income.[24] This underreporting is especially severe at the tails of the income distribution. Underreporting is particularly common for off-the-books earnings and welfare benefits.

Because of income underreporting, government data consistently show that the self-reported expenditures among the poor exceed their apparent incomes. For example, the U.S. Department of Labor’s Consumer Expenditure (CE) survey has shown for decades that the poor households routinely report spending roughly $2.40 for every dollar of apparent income. For families in “extreme poverty,” the expenditure-to-income ratio in the CE rises to around $25.00 to $1.00.[25]

“Extreme Poverty” Nearly Nonexistent for Three Decades

What does the government’s expenditure survey tell us about “extreme poverty”? Based on self-reports of consumer spending, “extreme poverty” has been almost totally nonexistent for three decades.

Adjusting for inflation, the CE shows only 61 instances from 1984 through 2015 in which a family reported spending less than $2 per person per day out of a total of 272,597 quarterly family records.[26] Two thirds of the 61 underspending families lived in public housing.[27] According to expenditure data reported by the families themselves, the number of families with children living on $2.00 per person per day is not one in 25, as Edin and Shaefer contend, but one in 4,469.

Trends in Poverty Measured by Expenditures

Self-reported expenditure data can also be used to assess poverty trends. Chart 3 shows the expenditure-based poverty rates from 1985 through 2014.[28] The chart measures the percentage of households that reported expenditures below 100 percent of the federal poverty line (FPL). The FPL for a family of three in 2015 was $19,096, or roughly $17.43 per person per day.

The poverty trends based on quarterly expenditures in Chart 3 are very similar to the poverty trends based on post-welfare income shown in Table 1. The expenditure-based poverty rate (at 100 percent of FPL) for non-married families with children hovered at slightly less than 30 percent in the decade before welfare reform. Following welfare reform, the rate dropped steadily, reaching around 17 percent in 2008. With the onset of the great recession in that year and during the subsequent weak recovery, the poverty rate rose, but it still remains far below the pre-welfare reform levels.


The chart also provides the expenditure-based poverty rates for two groups unaffected by welfare reform: married families with children and households without children. The poverty rates for these groups remain relatively flat and unchanging throughout the period. The fact that poverty fell rapidly for unmarried families with children between 1995 and 2008 while remaining unchanged for other groups indicates that the poverty decline was caused neither by survey methods nor by general economic factors. Something unique drove down the poverty rate of single parents during this period. The obvious answer is welfare reform.

Trends in Deep Poverty Measured by Expenditures

While consumer expenditure data show that “extreme poverty” or $2.00 per person per day is all but nonexistent, the CE data can also be used to measure the less severe condition of “deep poverty.” Deep poverty may be defined as a family having expenditures less than 50 percent of the federal poverty threshold (50 percent of FPL). In 2015, 50 percent of the FPL for a family of three was $9,548 per year. This translates into $8.72 per person per day in expenditures.

As Chart 4 shows, the CE data show that around 3 percent of non-married families with children, or 400,000 families, currently have quarterly spending that is less than half of the FPL,[29] but expenditure data also show that the percentage of non-married families with children in “deep poverty” is far lower today than before welfare reform. The deep poverty rate for this group was around 5.5 percent before welfare reform; it then dropped steadily to 2.5 percent by 2007. Since then, it has risen slightly to 3 percent, but this reflects the current torpid state of the economy rather than reforms enacted two decades ago.

Finally, contrary to the charges from the left, Chart 4 shows that the greatest drop in deep poverty has occurred among unmarried single parents with children—the group directly affected by welfare reform. This undermines any claim that welfare reform somehow made things worse for this population.



The Future of Reform

The core premise of welfare reform in 1996 was that welfare should not be a one-way handout; beneficiaries would be expected to contribute back in exchange for aid received. This idea remains extremely popular today. Nearly 90 percent of the public agrees that “able-bodied adults who receive cash, food, housing and medical assistance should be required to work or prepare for work as a condition of receiving those government benefits.”[30]

Unfortunately, welfare reform affected only one of over 80 means-tested welfare programs. Reform should be broadened. We should build on the lessons of welfare reform to build a pro-work, pro-marriage welfare system. Able-bodied recipients in all means-tested welfare programs should be required to work or prepare for work in exchange for aid. Penalties against marriage that are prevalent throughout the welfare system should be reduced. Welfare reformed in this manner would be far more efficient than the status quo in reducing poverty and promoting human dignity.

—Robert Rector is a Senior Research Fellow in Domestic Policy Studies, of the Institute for Family, Community, and Opportunity, at The Heritage Foundation. Jamie Bryan Hall is a Senior Policy Analyst in the Center for Data Analysis, of the Institute for Economic Freedom and Opportunity, at The Heritage Foundation.

[1] Center on Budget and Policy Priorities, “Urban Institute Study Confirms that Welfare Bills Would Increase Child Poverty,” July 26, 1996, http://www.cbpp.org/archives/URBAN726.HTM (accessed August 16, 2016).

[2] Robin Toner, “New Senate Push on Welfare Revives Tensions in Both Parties,” The New York Times, September 9, 1995, http://www.nytimes.com/1995/09/09/us/new-senate-push-on-welfare-revives-tensions-in-both-parties.html (accessed August 16, 2016).

[3] Robert Rector and Patrick F. Fagan, “The Continuing Good News About Welfare Reform,” Heritage Foundation Backgrounder No. 1620, February, 6, 2003, http://thf_media.s3.amazonaws.com/2003/pdf/bg_1620.pdf.

[4] Charles Kenny, “Taking Aim at the GOP’s War on the Poor,” Bloomberg, September 30, 2013, http://www.bloomberg.com/news/articles/2013-09-30/taking-aim-at-the-gops-war-on-the-poor (accessed August 16, 2016).

[5] Welfare reform and strong work policies led to a significant increase in marriage among less-skilled mothers. See Robert A. Moffitt, Brian J. Phelan, and Anne E. Winkler, “Welfare Rules, Incentives, and Family Structure,” National Bureau of Economic Research Working Paper No. 21257, Abstract, June 1, 2015, http://papers.ssrn.com/sol3/papers.cfm?abstract_id=2618645# (accessed August 16, 2016).

[6] In some non-married families with children, both parents may be absent, and the child resides with a grandparent or adult relative.

[7] Income-based poverty figures are calculated from U.S. Department of Commerce, U.S. Census Bureau, Current Population Survey, public-use microdata retrieved from http://dataferrett.census.gov (accessed August 16, 2016).

[8] While the poverty rate based on reported income increased for these groups from 1995 to 2014, the poverty rate based on reported expenditures declined slightly but less than the decline for non-married families with children. (See Chart 3.)

[9] Kathryn J. Edin and H. Luke Shaefer, $2.00 a Day: Living on Almost Nothing in America (New York: Houghton Mifflin Harcourt, 2015).

[10] H. Luke Shaefer and Kathryn Edin, “Rising Extreme Poverty in the United States and the Response of Federal Means-Tested Transfer Programs,” Social Service Review, Vol. 87, No. 2 (June 2013), p. 254, http://www.jstor.org/stable/10.1086/671012?seq=1#fndtn-page_scan_tab_contents (accessed August 16, 2016).

[11] Ibid.

[12] Data on living conditions calculated from SIPP 2008 panel wave 9 at U.S. Department of Commerce, U.S. Census Bureau, Survey of Income and Program Participation, public-use microdata retrieved from http://dataferrett.census.gov (accessed March 8, 2016). In the analysis, families headed by a working-age adult with children were defined as being in “extreme poverty” if their cash income was non-negative but below $2.00 per person per day during the final month covered by the interview. This definition closely matches Edin and Shaefer’s extreme poverty population.

[13] Edin and Shaefer’s argument is also misleading because they estimate the percentage of families with children each month that have cash incomes below $2.00 per person per day during that month. This procedure distorts findings because it assumes that families do not shift income resources between months.

[14] Shaefer and Edin, “Rising Extreme Poverty in the United States and the Response of Federal Means-Tested Transfer Programs,” p. 259.

[15] U.S. Department of Agriculture, Food and Nutrition Service, Supplemental Nutrition Assistance Program, “Participation and Costs, 1969–2015,” http://www.fns.usda.gov/pd/supplemental-nutrition-assistance-program-snap (accessed August 17, 2016).

[16] Shelley K. Irving and Tracy A. Loveless, “Dynamics of Economic Well-Being: Participation in Government Programs, 2009–2012: Who Gets Assistance?” U.S. Department of Commerce, U.S. Census Bureau Household Economic Studies, Report No. P70-141, May 2015, p. 20, https://www.census.gov/content/dam/Census/library/publications/2015/demo/p70-141.pdf (accessed August 16, 2016).

[17] U.S. Department of Agriculture, “WIC Program Participation and Costs (Data as of August 5, 2016),” http://www.fns.usda.gov/sites/default/files/pd/wisummary.pdf (accessed August 16, 2016).

[18] Bruce D. Meyer, Wallace K. C. Mok, and James X. Sullivan, “The Under-Reporting of Transfers in Household Surveys: Its Nature and Consequences,” National Bureau of Economic Research Working Paper No. 15181, July 2009, p. 83, http://www.nber.org/papers/w15181 (accessed August 16, 2016).

[19] U.S. Department of Health and Human Services, Administration for Children and Families, Office of Family Assistance, “Fiscal and Calendar Year 2011–Combined TANF AND SSP-MOE: Total Number of Recipients,” April 3, 2012, http://www.acf.hhs.gov/programs/ofa/resource/2011-recipient-tanssp (accessed August 16, 2016).

[20] Irving and Loveless, “Dynamics of Economic Well-Being,” p. 26.

[21] U.S. Department of Labor, “Extended and Emergency Programs (5),” http://www.oui.doleta.gov/unemploy/docs/persons.xls (accessed August 16, 2016).

[22] Meyer, Mok, and Sullivan, “The Under-Reporting of Transfers in Household Surveys,” p. 87.

[23] Irving and Loveless, “Dynamics of Economic Well-Being,” p. 24.

[24] Meyer, Mok, and Sullivan, “The Under-Reeporting of Transfers in Household Surveys.”

[25] The Consumer Expenditure Survey collects income only on an annual basis. Expenditures are collected on a monthly basis. Families with children who report less than $2.00 per person per day in annual income in the CE on average report spending about $25 over the year for every dollar of reported income.

[26] There were 61 instances in which a family with children reported spending on average less than $2.00 per person per day during a quarter out of 272,597 observations. Calculations based on U.S. Department of Labor, Bureau of Labor Statistics, Consumer Expenditure Survey, public-used microdata retrieved from http://www.bls.gov/cex/pumd.htm (1996–present) and on a purchased USB flash drive for years 1984–1995.

[27] Of the 61 families, 41 report living in subsidized housing. Out of the total group, 14 reported annual incomes above the federal poverty level.

[28] The chart presents quarterly poverty rates based on expenditures presented as the centered rolling average of four quarters. Calculations based on U.S. Department of Labor, Bureau of Labor Statistics, Consumer Expenditure Survey, public-used microdata retrieved from http://www.bls.gov/cex/pumd.htm (1996–present) and on a purchased USB flash drive for years 1984–1995.

[29] The chart presents the quarterly poverty rates expressed as the centered rolling average of four quarters. Ibid.

[30] American Perceptions Initiative survey conducted November 13, 2015, http://thf-reports.s3.amazonaws.com/2016/AmericanPerspectivesonWelfareandPoverty.pdf (accessed August 16, 2016).


Robert Rector
Robert Rector

Senior Research Fellow, Center for Health and Welfare Policy

Jamie Hall
Jamie Hall

Research Fellow, Center for Data Analysis