I'm trying to resolve a tension between two claims about people receiving various forms of federal assistance in the US. I'll illustrate the claims with a couple of links:

On the one hand, it appears that a lot of people receive some kind of federal assistance. This article at the conservative-leaning CNS.com claims that the number is something like 35% of Americans. Now that claim is somewhat disputed as to its interpretation, but it seems that the overall numbers are legitimate.

On the other hand this article from a California public radio station makes the claim, often repeated on the left, that most people who use welfare programs only use them for a short time.

Now, there seems to be a tension between the idea that 35% of people receive assistance, but most people get off of assistance quickly. I guess it's possible that 70% of the country just goes two-years-on, two-years-off, and that would reconcile these facts, but that's not what happens. Is one of these claims particularly off-base, or what's going on here?

I've seen one person state that the claim from SCPR is simply impossible, in light of the claim from CNS. Is he right?

  • This article from the US Census Bureau has some good information. Their data characterization seems a little sketchy, so some fact-checking might be necessary to make the most of that study.
    – Nat
    Jun 29, 2017 at 14:06
  • I'd say there's a very large discrepancy between a PBS article, and a right-wing pundit's opinion column on a right-wing blog. But as politifact points out, the numbers are OK, it's just that his interpretation is completely false.
    – user1530
    Jul 1, 2017 at 6:58
  • Note: the program that can most accurately be described as "welfare" is the TANF program, but even that isn't "welfare". "Welfare" was essentially eliminated in 1996 with PRWORA. There are still government social programs that could not very properly be called "welfare." The reason I say this is because it's important to distinguish between someone saying "welfare" and meaning TANF vs someone saying "welfare" and meaning any form of government benefit. Related: most people who utilize TANF only use it for short periods of time, which could validate the CPRS claim.
    – John
    Mar 27, 2019 at 19:33
  • "Temporary Assistance for Needy Families is one of the United States of America's federal assistance programs. It began on July 1, 1997, and succeeded the Aid to Families with Dependent Children (AFDC) program, providing cash assistance to indigent American families through the United States Department of Health and Human Services. This cash benefit is often referred to simply as 'welfare.'" en.wikipedia.org/wiki/Temporary_Assistance_for_Needy_Families
    – John
    Mar 27, 2019 at 19:34

2 Answers 2


I suspect that the discrepancy flows from the definition of welfare used. There is probably also a life cycle effect, with some kinds of benefits used only at the end of life and terminating in death, and others used pretty much only when one has young children.

The public radio claim is as follows:

Researchers found that over a four-year period, 79 percent of people who received cash General Assistance - known as "welfare" - left the program within two years. About half left supplemental security income, medicaid, and food stamps in the same amount of time.

The major outlier was housing assistance - including public housing and subsidized rental programs. There, only 40 percent dropped out within two years.

The CNS claim is as follows:

[T]he 109,631,000 living in households taking federal welfare benefits as of the end of 2012, according to the Census Bureau, equaled 35.4 percent of all 309,467,000 people living in the United States at that time.

When those receiving benefits from non-means-tested federal programs — such as Social Security, Medicare, unemployment and veterans benefits — were added to those taking welfare benefits, it turned out that 153,323,000 people were getting federal benefits of some type at the end of 2012. . . .

82,679,000 of the welfare-takers lived in households where people were on Medicaid, said the Census Bureau. 51,471,000 were in households on food stamps. 22,526,000 were in the Women, Infants and Children program. 20,355,000 were in household on Supplemental Security Income. 13,267,000 lived in public housing or got housing subsidies. 5,442,000 got Temporary Assistance to Needy Families. 4,517,000 received other forms of federal cash assistance.

TANF aid, which is the narrowest commonly used definition of welfare, typically is relatively short term and episodic, as is WIC which like TANF has a built in time limit. These programs have the highest turnover in the study cited by the public radio station.

SSI and public housing tend to be much more long term. The public radio story confirms that the housing program is long term, but shows an intermediate amount of turnover in other programs.

Normally, SSI is paid to seniors or disable people who also qualify for Social Security benefits, can't work, and don't have savings, so participation in those programs tends to be long term.

The waiting lists for public housing assistance are typically many years long, so only the long term poor tend to end up getting it.

Eligibility for food stamps (officially known as SNAP) is materially more generous than eligibility for the cash welfare TANF program and so people tend to be recipients of it for a longer period of time.

Medicaid really consists of multiple programs with different eligibility requirements and different benefits.

For example, one very significant component of the Medicaid program which is rarely characterized as "welfare" is its nursing home assistance program which is utilized mostly by the working and middle classes and provides a majority of the funds for nursing home care in the U.S. This program actually is usually for a short period of time because the average nursing home stay is about one year (predominantly ending in death rather than returning home). Obviously, people who leave the Medicaid program in that manner don't return to the general population afterwards.

Another significant component of the Medicaid program involves people who are permanently disabled and usually receiving some form of disability benefits as well. This is also long term in most cases.

A third component of the Medicaid program involves working age, not disabled people and their children who either receive pre-expansion benefits because they are in or very nearly in poverty, or who receive post-expansion benefits because they are working class. (There are actually some additional subsets of this population, CHIP for children, a pregnant women's program and a program for adults, in addition to expansion and buy-in components). Often people are made eligible for Medicaid on the basis of the same application that qualifies them for food stamps and TANF.

But, lots of people who are formally in the Medicaid program don't actually receive any Medicaid benefits. Some are just healthy people who don't need medical care. Others are people who need medical care but can't find a provider who accepts it (because Medicaid reimburses providers at far below the market rate or the rate paid by Medicare, many providers refuse to accept new Medicaid patients). Definitionally, it isn't obvious that someone eligible for the Medicaid program who uses it when they get step throat or a urinary tract infection once every three years, for example, is a continuous or an episodic beneficiary of this means-tested program.

  • This answer only mentions federal programs - could part of the discrepancy also lie in state level welfare programs?
    – user4012
    Jun 29, 2017 at 17:05
  • @user4012 Probably not. States programs make up a quite small proportion of all means tested welfare programs. Even when states fund means tested welfare, most of the money funds part of a federally designed system.
    – ohwilleke
    Jun 29, 2017 at 19:04

First, these examine fairly different economies. The CNS claim is based on data from 2012. The numbers of people receiving assistance obviously goes up in bad times, and 2012 was still a pretty bad time. So, you would expect that number to be higher than in 2015 (the NPR article).

Second, these examine pretty different sets of programs as "welfare". CNS claim merges all forms of means-tested assistance programs together, and calls it welfare. The NPR article focuses on General Assistance as welfare, although also claim that it extends to a lot of other programs.

The data that the CNS study is based on puts unemployment insurance (is that the same as the NPR's General Assistance?) as having ~5 million. Is it the Federal Supplemental Security Income (~20 million).

Certainly, Medicaid, which the CNS survey counted and the NPR study did not, is vital to understand. It's the major driver of CNS data, with ~82 million people on Medicaid in their data set. (note, this does not reduce their example to 25 million; people can be in multiple groups, with ~50 million on foodstamps alone. However, it's the largest driver, with foodstamps a distant second.) However, the majority of people with Medicaid work. (Although the NPR article also says that about 50% of people will leave Medicaid as well, so this may be moot.)

Third, I think it's very reasonable to assume that for a lot of people hovering on the poverty line, they get on and off assistance. A lost job may cycle someone onto assistance and then back off. A child may require assistance and then age out of it (for instance, WIC only covers children up to 5). The number of hours may be reduced for a few weeks. Or it may be possible to pick up a second job/extra hours during the busy period (say Christmas time).

  • "General Assistance" is a term often used to refer to a state level means tested cash benefit program that is distinct from TANF because it has beneficiaries who aren't eligible for federal welfare payments. Usually the payments are small and the number of beneficiaries is a quite small share of all means-tested program benefits.
    – ohwilleke
    Mar 27, 2019 at 18:26

You must log in to answer this question.

Not the answer you're looking for? Browse other questions tagged .