Contrary to "Entitlement Society" Rhetoric, Over Nine-Tenths of Entitlement Benefits Go to Elderly, Disabled, or Working Households
February 10, 2012
Some conservative critics of federal social programs, including leading presidential candidates, are sounding an alarm that the United States is rapidly becoming an “entitlement society” in which social programs are undermining the work ethic and creating a large class of Americans who prefer to depend on government benefits rather than work. A new CBPP analysis of budget and Census data, however, shows that more than 90 percent of the benefit dollars that entitlement and other mandatory programs spend go to assist people who are elderly, seriously disabled, or members of working households — not to able-bodied, working-age Americans who choose not to work. (See Figure 1.) This figure has changed little in the past few years.
In a December 2011 op-ed, former Massachusetts Governor Mitt Romney warned ominously of the dangers that the nation faces from the encroachment of the “Entitlement Society,” predicting that in a few years, “we will have created a society that contains a sizable contingent of long-term jobless, dependent on government benefits for survival.” “Government dependency,” he wrote, “can only foster passivity and sloth.” Similarly, former Senator Rick Santorum said that recent expansions in the “reach of government” and the spending behind them are “systematically destroying the work ethic.”
The claim behind these critiques is clear: federal spending on entitlements and other mandatory programs through which individuals receive benefits is promoting laziness, creating a dependent class of Americans who are losing the desire to work and would rather collect government benefits than find a job.
Such beliefs are starkly at odds with the basic facts regarding social programs, the analysis finds. Federal budget and Census data show that, in 2010, 91 percentof the benefit dollars from entitlement and other mandatory programs went to the elderly (people 65 and over), the seriously disabled, and members of working households. People who are neither elderly nor disabled — and do not live in a working household — received only 9 percent of the benefits.
Moreover, the vast bulk of that 9 percent goes for medical care, unemployment insurance benefits (which individuals must have a significant work history to receive), Social Security survivor benefits for the children and spouses of deceased workers, and Social Security benefits for retirees between ages 62 and 64. Seven out of the 9 percentage points go for one of these four purposes.
A small number of discretionary (i.e., non-entitlement) programs also provide substantial benefits to individuals, but the lack of full funding for some of these programs means they do not reach all eligible recipients. Indeed, in some cases — such as in low-income rental assistance programs — the vast majority of people who are eligible receive nobenefits because of program funding limits. If we broaden the universe of programs examined to include the principal discretionary programs that provide benefits — low-income housing programs, the WIC nutrition program for low-income women and young children, and low-income energy assistance — the result is essentially unchanged. Some 90 percent of the benefit dollars still go to the elderly, the disabled, and working households.
This figure also changes little if we tweak the definition of a “working household” or of who is “disabled.” This analysis defines a working household as one in which an individual works at least 1,000 hours in a year; raising the threshold to 1,500 hours makes little difference. This analysis defines a disabled person as one who receives Social Security disability benefits or the disability component of the Supplemental Security Income program (SSI) or who qualifies for Medicare on the basis of disability; modifying the definition to include disabled people who are not in one of these categories also makes little difference.
Distribution of Tax Expenditure Benefits Differs Greatly, and Is Much Less Favorable to the Middle Class and Low-Incomes Families
The distribution of entitlement benefits stands in sharp contrast to the distribution of benefits for tax expenditures, which former Federal Reserve Chairman Alan Greenspan has called “tax entitlements.” The Tax Policy Center finds that under current tax policy:a
- The top fifth of the population receives 66 percent of tax-expenditure benefits (compared to 10 percent of entitlement benefits).
- The middle 60 percent of the population receives a little over 31 percent of tax-expenditure benefits (compared to 58 percent of entitlement benefits).
- The bottom fifth receives just 2.8 percent of tax-expenditure benefits (compared to 32 percent of entitlement benefits).b
- The top 1 percent of the population receives 23.9 percent of tax-expenditure benefits — more than eight times as much as the bottom fifth of the population, and nearly as much as the middle 60 percent of the population.
a These TPC figures refer to 2011 tax policy. The figures are for individual tax expenditures and do not include corporate tax expenditures. If corporate tax expenditures were included, the results would be skewed even more heavily to the top of the income spectrum.
b The TPC figures are based on tax filing units, rather than entire households (thus, the “bottom fifth” means the 20 percent of tax units with the lowest incomes). If our analysis had been based on smaller units similar to tax units rather than on households, it would have shown that the bottom fifth of the population received a somewhat smaller share of the entitlement benefits (about 25 percent rather than 32 percent) and the middle slightly more (64 percent rather than 58 percent).
Moreover, if we look only at entitlement programs that are targeted to people with low incomes, the percentage of benefit dollars going to people who are elderly or disabled or members of working households remains high. Five of every six benefit dollars in these programs — 83 percent — go to such people.
If anything, these figures understate the percentage of the benefits that generally go to people who are elderly, disabled, or members of working households. As noted, these data are for fiscal year 2010, a year when the unemployment rate averaged 9.6 percent and an unusually large number of Americans were in economic distress. In fiscal year 2007, the share of entitlement benefits going to people who are elderly or disabled or members of working households was a bit higher.
In short, both the current reality and the trends of recent decades contrast sharply with the critics’ assumption that social programs increasingly are supporting people who can work but choose not to do so. In the 1980s and 1990s, the United States substantially reduced assistance to the jobless poor (through legislation such as the 1996 welfare law) while increasing assistance to low-income working families (such as through expansions of the Earned Income Tax Credit). The safety net became much more “work-based.” In addition, the U.S. population is aging, which raises the share of benefits going to seniors and people with disabilities.
The data in this analysis also dispel other common misperceptions, such as a belief (sometimes fanned by political figures) that entitlement programs shift substantial resources from the middle class to the poor. The data show that the middle class receives approximately its proportionate share of benefits: in 2010, the middle 60 percent of the population received 58 percent of the entitlement benefits. (The top 20 percent of the population received 10 percent of the benefits; the bottom 20 percent received 32 percent of the benefits. See Figure 2)
These figures contrast sharply with the distribution of the extensive deductions, credits, and other write-offs in the federal tax code, known as tax expenditures (former Federal Reserve Chair Alan Greenspan has called them “tax entitlements”). The Urban Institute-Brookings Institution Tax Policy Center estimates that for tax year 2011, the top fifth of the population will receive 66 percent of the $1.1 trillion in individual tax-expenditure benefits (the top 1 percent alone will receive 23.9 percent of the benefits), the middle 60 percent of the population will receive a little over 31 percent of the benefits, and the bottom 20 percent of the population will receive only 2.8 percent of the benefits. (See Figure 3 and box.)
Also, contrary to what a substantial share of Americans may assume, non-Hispanic whites receive slightly more than their proportionate share of entitlement benefits. Non-Hispanic whites accounted for 64 percent of the population in 2010 and received 69 percent of the entitlement benefits. In contrast, Hispanics made up 16 percent of the population but received 12 percent of the benefits, less than their proportionate share — likely because they are a younger population and also because immigrants, including many legal immigrants, are ineligible for various benefits. Non-Hispanic African Americans account for 12 percent of the population and received 14 percent of the benefits.
The Programs We Examined and the Data We Used
This analysis uses federal budget data on the benefit costs of various programs and Census data showing the percentage of the benefits in each program that go to different groups by age, employment, race, and other such factors. (See the appendix for more detail.) It covers all major entitlement programs except veterans’ programs and military and civil service retirement, which critics presumably do not have in mind when they warn of the dangers of the “entitlement society” in fomenting dependency and sloth. But if we include veterans’ and federal retirement programs, the share of benefits going to the elderly, the disabled, and working households remains unchanged at 91 percent.
Findings of Analysis Consistent With Leading Academic Research
In a major report on U.S. social programs last year,* leading researchers concluded "First, the U.S. system favors groups with special needs, such as the disabled and the elderly. Groups like these which are perceived as especially deserving receive disproportionate transfers and those transfers have been increasing over time. Second, the system favors workers over non-workers and has increasingly done so over time. The rise of the EITC and the decline of AFDC/TANF is most illustrative of this trend." The study also found that "the demographic group which is most underserved by the system are non-elderly non-disabled families with no continuously-employed members."
* 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, May 2011, pp. 36,37.
The entitlement and mandatory programs covered in the analysis are Social Security, Medicare, Medicaid, unemployment insurance, SNAP (formerly known as the Food Stamp Program), SSI, Temporary Assistance for Needy Families (TANF), the school lunch program, the Children’s Health Insurance Program (CHIP), the Earned Income Tax Credit, and the refundable component of the Child Tax Credit. This is the full list of entitlement or mandatory programs (other than the veterans’ and federal retirement programs) for which the Census Bureau collects data on which beneficiaries receive them. It includes every entitlement or mandatory benefit program with annual federal and state expenditures of over $10 billion other than veterans’ and federal retirement programs.
These programs accounted for $1.8 trillion in federal expenditures in 2010 out of a total of $2.1 trillion in entitlement and mandatory program costs; most of the remaining amount goes for the veterans’ and federal retirement programs. We also include in this analysis $130 billion in state funding for benefits in the three programs that operate as federal-state partnerships: Medicaid, TANF, and CHIP.
We examined as well a somewhat broader group of programs that includes the same entitlement and mandatory programs but adds low-income housing programs, WIC, and LIHEAP — the largest discretionary programs that provide benefits, other than Pell Grants (for which reliable Census data on recipients are not available).
- In 2010, 91 percent of the benefits provided through entitlement programs went to people who were elderly (65 or older), disabled (receiving Social Security disability benefits, SSI disability benefits, or Medicare on the basis of a disability — all three programs use essentially the same disability standard, which limits eligibility to people with medically certified disabilities that leave them substantially unable to work), or members of a household in which an individual worked at least 1,000 hours during the year. As noted, the 91 percent figure is unchanged if one includes veterans’ and federal retirement programs.
This analysis defines working households as those in which someone has worked at least 1,000 hours a year. This is a conservative definition. If two household members work more than 1,000 hours between them but no single individual works at least 1,000 hours, we do not classify the unit as a working household.
- We also do not count people receiving unemployment insurance benefits as workers, although such individuals must have amassed a significant work record to qualify for UI. If we include people receiving UI as workers — in other words, if we ask what share of entitlement benefits go to people who are elderly or disabled or receive UI, or are members of households in which an individual works at least 1,000 hours — the share rises from 91 percent to 94 percent. The percentage edges down to 92 percent if we count UI recipients but raise the “hours-of-work threshold” from 1,000 hours of work to 1,500 hours. If we define working households as those in which an individual worked at least 1,500 hours but do not count UI beneficiaries as workers, the percentage declines slightly to 88 percent.
- If we add in the principal discretionary programs that help people meet basic needs (low-income housing, WIC, and LIHEAP) and examine both them and the entitlement and other mandatory programs, the 91 percent figure drops to 90 percent.
- This analysis uses a narrow definition of disability that misses individuals who become disabled so young that they haven’t amassed enough work history to qualify for Social Security Disability Insurance or Medicare, and whose countable household income or assets are over the very low SSI eligibility limits, which are below the poverty line. If we broaden the definition to include other adults who report work-limiting disabilities in the Census survey data, the percentage of benefits going to people who are elderly or disabled or members of working households rises from 91 percent to 92 percent.
- If we look only at means-tested entitlement and other mandatory programs — that is, programs limited to low-income people — the percentage of benefits going to the elderly, the disabled, or working households remains high at 83 percent, a robust percentage for programs that are limited to people with low incomes. This high percentage reflects policy changes in recent decades that have substantially restricted benefits for poor people who lack earnings (other than the elderly and disabled), while increasing assistance for low-income working families with children, especially in the form of tax credits. The 83 percent figure edges down to 82 percent if low-income housing programs, WIC, and LIHEAP are included.
- More than half (53 percent) of entitlement benefits go to seniors. Some 73 percent of the benefits go to people who are elderly or disabled; most of the rest goes to working households.
- The data contradict beliefs that entitlements take heavily from the middle class to give to people at the bottom or that they shower benefits on the very wealthy. The middle 60 percent of the population receives close to 60 percent of the benefits. The top 5 percent of the population receives about 3 percent of the benefits.
This analysis describes who receives government benefits. To estimate where government benefit dollars go, we use Census Bureau data to calculate the percentage of these benefits received by particular groups, and apply those percentages to budget totals for each program from administrative data from the Office of Management and Budget or program records.
The analysis is limited to assistance in programs for which data are available from our main Census Bureau source, the Annual Social and Economic Supplement of the Current Population Survey (CPS), commonly called the March CPS. Our 2010 figures use the March 2011 CPS, which asks about calendar year 2010 income.
People who live in institutions present a special case. Some 17.6 percent of Medicaid benefits and about 5.2 percent of Medicare benefits go to people in nursing homes and other institutional settings. The CPS is a household survey and does not include people who live in institutions. However, the Census Bureau’s American Community Survey doescover those individuals. Accordingly, we use ACS data to distribute Medicaid and Medicare benefits that go to people living in institutions. (For other benefits we examine, the share going to people in institutions is generally very small or non-existent, and we do not make a similar adjustment.)
This analysis includes spending for benefits (assistance to individuals); it excludes program expenditures that do not go for benefits — primarily administrative costs. State funding is included in jointly run federal-state programs that require state matching funds — Medicaid, the Children’s Health Insurance Program (CHIP), and Temporary Assistance for Needy Families (TANF). Payments to beneficiaries living in U.S. territories or other countries (such as a Social Security retiree living abroad) are excluded because overseas data are not available from the CPS.
We examine Census data on the age, disability status, and race of the individuals who receive the assistance, and the number of hours worked by the household member with the most work hours. Annual work hours are estimated by multiplying the number of weeks worked by the usual number of work hours per week.
Wherever possible, we distribute dollars, not persons. For example, rather than looking at the percentage of Social Security recipients who are 65 and older, we look at the percentage of Social Security dollars that go to recipients 65 and older.
For some programs, the Census Bureau does not ask survey respondents about the dollar value of benefits received but Census estimates the dollar value based on a combination of other available survey data and program data. For example, while Census asks respondents about whether they receive housing assistance, it does not ask about the value of rental assistance received; however, the Bureau calculates the value of the rent subsidy using other data including annual household cash income, the size and composition of the household, local fair market rent levels, and program rules. We use these Census estimates of the amount of housing assistance that various households receive, as well as similar Census estimates for the value of school lunch and WIC benefits. For the value of tax credits, we use estimated credits available on the CPS file, which Census calculates based on the family’s reported income, earnings, and composition, as well as on national tax statistics from the IRS.
In cases where Census has data on who receives a benefit but has not estimated the dollar value of the benefit — specifically, for Medicare and Medicaid recipients in institutions — we allocate total benefit dollars from program records using the Census data on the distribution of program recipients rather than program dollars. This effectively assumes that benefits have the same average value across the institutionalized population.
In general, we use individual income values in the Census data, assigning all Social Security benefits, for example, to the specific individual or individuals within the household who report receiving that income. In the case of benefits such as housing assistance, SNAP, and home energy assistance that generally go to an entire household rather than to individuals, we assign each household member their per-person share of the benefit. Likewise, in the case of benefits that typically support families (such as TANF), we allocate the per-person value of the benefit equally across all members of the family.
We define old age and disability at the individual level. For example, if a family of two contains a 65-year-old Social Security recipient and his or her 64-year-old spouse who also receives Social Security as an early retiree, we count only the 65-year-old individual’s benefit as going to a person who is elderly. (In our core analysis, disability refers to reported receipt of Social Security disability benefits, SSI, or Medicare among those younger than 65.)
We define work at the household level. Thus, if a household contains a child, an individual who worked substantially through the year, and that individual’s unemployed spouse, the benefits going to all household members are counted as going to a working household.
When we examine benefits by income, we sort the population into income groups based on household cash income adjusted for household size, using the same method that CBO uses to examine income and benefit receipt and tax burdens by income group. Like CBO, when we group individuals into fifths by household income, each fifth contains the same number of persons (not households).
Our Census data have several limitations. For example, the CPS likely misses most benefits paid to individuals who died between the calendar year (2010) and the survey month (generally March of the following year); this omission may bias our results toward younger recipients. Conversely, benefits paid to someone who was 64 years old in 2010 but turned 65 early the next year before the March 2011 CPS may be classified as going to someone age 65. Also, as previously noted, the CPS leaves out the institutionalized population; while our analysis includes the dollars going to people in institutions, we lack complete data on how these dollars are distributed. Total spending levels from program records are for the fiscal year, while the Census data used to distribute them are for the calendar year.
Another limitation is under-reporting. In the CPS and ACS, as in most surveys, respondents may forget or omit some of their income. Moreover, some benefits tend to be under-reported more than others. We deal with this data limitation by using Census data only to estimate the percentage distribution of each program’s benefits; we apply that distribution to the actual spending totals for each program as shown in official budget and program records. This eliminates any bias due to differences in under-reporting between programs. (If we had used pure Census data, our figures would show that 92 percent, rather than 91 percent, of entitlement and other mandatory benefit spending in 2010 went to seniors, people with disabilities, and working households.) It is still possible for some bias to exist due to variations in the amount of under-reporting between different types of recipients within a program.
The appendix table lists the programs included in the analysis. The category of “means-tested” mandatory benefits, which are those targeted to low-income individuals and families, includes SSI, SNAP, housing assistance, Medicaid/CHIP, TANF, EITC, the refundable portion of the Child Tax Credit, and the means-tested portion of school lunch. The means-tested discretionary programs examined are low-income housing programs, the low-income home energy assistance program, and the WIC program.
|Mandatory spending (core analysis)|
|Social Security a||689||689|
|Unemployment compensation b||156||156|
|Supplemental Security Income||44||44|
|Temporary Assistance for Needy Families (basic assistance only) c||7||4||11|
|Supplemental Nutrition Assistance Program d||65||65|
|School Lunches e||10||10|
|Earned Income Tax Credit (refundable share) f||55||55|
|Child Tax Credit (refundable share)||23||23|
|Children's Health Insurance Program i||8||3||11|
|Total, mandatory spending included in core analysis j||1,834||130||1,964|
|Memorandum: All mandatory spending k||2,096||#N/A||#N/A|
|Selected discretionary programs|
|Rental assistance l||34|
|Special Supplemental Nutrition Program for Women, Infants, and Children (WIC) m||6|
|Low Income Home Energy Assistance n||5|
Unless otherwise noted, data are from U.S. Office of Management and Budget (OMB), Historical Table 8.5, "Outlays for Mandatory and Related Programs" (http://www.whitehouse.gov/omb/budget/Historicals) and Congressional Budget Office (CBO), The Economic and Budget Outlook: Fiscal Years 2012 to 2022, Appendix F, “Historical Budget Data” (http://www.cbo.gov/ftpdocs/126xx/doc12699/01-31-2012_Outlook.pdf).
a According to OMB table 8.5, mandatory spending for Social Security was $701 billion in 2010. We have excluded an estimated $12 billion paid to residents of Puerto Rico and other territories and foreign countries; see http://www.ssa.gov/policy/docs/statcomps/supplement/2011/5j.pdf.
 As explained below, this analysis covers Social Security, Medicare, Medicaid and CHIP, unemployment insurance, SNAP, SSI, TANF, the school lunch program, the EITC, and the refundable component of the Child Tax Credit.
 Mitt Romney, “What kind of society does America want?,” USA Today, December 19, 2011.
 CQ Transcriptions, “Former Sen. Rick Santorum, R-PA., Presidential Candidate, Holds A Town Hall Meeting in Brentwood, NH,” January 4, 2012.
 Some mandatory programs, such as TANF and CHIP, also have capped funding and may turn away some eligible applicants due to funding limitations.
 This analysis divides the population into income fifths, or quintiles, by annual household cash income, adjusting income for household size in accordance with the methods of the Congressional Budget Office.
 Asian Americans and other groups received the other 5 percent of the entitlement benefits.
 Administrative costs — program spending that does not go to individuals or families — are not included. These costs make up a small share of overall program costs; see Robert Greenstein and CBPP Staff, “Romney’s Charge That Most Federal Low-Income Spending Goes for ‘Overhead’ and ‘Bureaucrats’ Is False,” Center on Budget and Policy Priorities, updated January 23, 2012, http://www.cbpp.org/cms/index.cfm?fa=view&id=3655.
 The $2.1 trillion in entitlement and mandatory expenditures also includes programs that are not benefits, such as federal deposit insurance and the Postal Service Fund.
 If we did include Pell Grants using the available data, the 90 percent figure would not change.
 We did not use a 1,500 hours-of-work threshold because work of at least 1,000 hours from an individual shows substantial work effort, especially in a year when the unemployment rate averaged 9.6 percent. In addition, to accept a 1,500 hour threshold while continuing not to count UI beneficiaries would classify as non-workers those people who worked 1,000 to 1,499 hours during the year but were then laid off, had amassed a significant work record in the prior year as well, and consequently qualified for UI. We took the middle ground of including as workers individuals who worked at least 1,000 hours in 2010, while excluding those who received UI benefits but had less than 1,000 hours of employment.
 If one spouse in a family is elderly or disabled but the other spouse is not, we count only the benefits for the elderly or disabled spouse as going to people who are elderly or disabled.
 The Urban Institute estimates that in 2009, institutional care for Medicaid beneficiaries was 17.6 percent of total Medicaid expenditures. (Source: John Holahan et al., “Medicaid Spending Growth Over the Last Decade and the Great Recession, 2000-2009,” The Kaiser Family Foundation and the Urban Institute, February 2011, www.kff.org/medicaid/upload/8152.pdf.) The most recent CBO Medicare baseline shows that in 2010, institutional care spending was 5.2 percent of total program spending (www.cbo.gov/budget/factsheets/2011b/medicare.pdf).
 Roughly 2 percent of Social Security and SSI benefit dollars go to people living in institutions. This percentage is small enough that we do not attempt to analyze these recipients separately; for these individuals, we follow the approach explained in footnote 18. If we were able to analyze these recipients separately, the likely result would be to increase the estimated share of benefits going to seniors and persons with disabilities, but only slightly. Other benefits analyzed here, such as food stamps, are generally not available to persons in institutions.
 The Census housing assistance calculation we used (called fmrcap) is available for researchers in the Experimental Poverty Measures Public-Use Research Files at www.census.gov.
 Like Census, we assume WIC has a fixed value per WIC participant listed in the Census data.
 Although the CPS identifies recipients of Social Security disability benefits, the ACS, which we use for analyzing benefits going to persons in institutions, does not. This does not present a problem for our analysis of Medicare benefits, since all Medicare beneficiaries are age 65 or older or have a serious disability. For our Medicaid analysis, we can identify which Medicaid recipients who live in institutions and are under 65 receive SSI or Medicare (and hence are disabled), but not which Medicaid recipients are in their first two years of receipt of Social Security disability benefits and hence do not yet receive Medicare coverage. (Such individuals have a two-year waiting period for Medicare.) To more completely capture people who are institutionalized, we use self-reported disability status on the CPS. This method results in a modest understatement of the proportion of Medicaid benefits provided to people in institutions that goes to people with a disability. This method shows an estimated 75 percent of Medicaid benefits for people in institutions going to people with disabilities. Data from the 2009 Medicaid Statistical Information System show the actual figure is 80 percent.
 Under the CBO methodology, household income is adjusted for household size by dividing household income by the square root of the number of household members.
 For Medicaid and Medicare spending, our correction using ACS data relies on the assumption that average benefits for those two programs are equal across all types of recipients in institutions. For Social Security and SSI, as noted, the percentage of recipients who live in institutions is very small, and our method effectively assumes that their benefits are distributed similarly to those of recipients not living in institutions.