A misleading chart on ‘welfare’ spending
By Glenn Kessler,
Source: budget.senate.gov/republican “Converted to cash, we spend enough on federal welfare to mail every household living beneath the poverty line a check for $60,000 each year. Can anyone honestly say this huge sum of money is being wisely and effectively spent, that no improvements are needed?”
— Sen. Jeff Sessions (R-Ala.), Budget Committee hearing, Feb. 13, 2013
The chart above, as well as Sen. Sessions’s statement, is derived from his calculation that what he describes as some 80 “welfare programs” make up the single-largest item in the federal budget — larger than Social Security, Medicare or Defense Department spending. On Wednesday, we examined how Sessions counted these programs, using a 2012 report from the Congressional Research Service.
As we noted, Sessions is trying to shake up traditional notions of what constitutes welfare. Now, let’s look at how he puts these data to use in a chart — and whether it illuminates or obscures the issue.
The chart concludes that welfare spending “equates” to $168 in cash per day for each household in poverty, which it says exceeds the median income by 20 percent. Alternatively, as Sessions put it at the hearing, this amounts to $60,000 per year, compared to a median income of $50,000 in 2011.
We had long discussions with Sessions’s staff about this figure, which they contend is mathematically correct and intended to illuminate the money now spent by the federal government on low-income people. But we have some serious problems with both the numerator and denominator in this calculation.
First, health-care spending, especially Medicaid, makes up nearly 50 percent of the total figure. But a majority of Medicaid spending goes to the elderly and disabled, not families with children.
Moreover, health-care spending is different from food stamps or the earned income tax credit in that such aid generally does not add to a family’s income level; instead, such assistance helps pays for bills that are the direct result of how sick or disabled a patient is. (That’s why so much of Medicaid spending is directed to the elderly in the last years of life.)
“Medicaid is a federal program that is intended to provide health-care services to people who are poor or near-poor,” responded a Sessions aide. “It also provides health benefits to sick people, but those people must first meet certain income criteria (and in some cases an asset test) in order to qualify for the benefit. In other words, being sick alone doesn’t qualify one for Medicaid in the same way as being hungry doesn’t qualify one for food stamps.”
Still, while the chart compares what Sessions terms welfare spending to median income, the Census Bureau does not include health benefits (such as employer-provided health care) in that calculation, even though such benefits account for half of the welfare side of the ledger. (See Page 29.) So, he’s really comparing apples and oranges.
Finally, Sessions adds up many means-tested programs, which are aimed at people with low incomes, but then divides the figure by the number of people under the poverty level — even though millions of people above the poverty level receive these benefits. That also significantly gooses up the figure for spending per household.
At first glance, many might assume that Sessions is saying this is how much money is given to each household under the poverty line. Sessions’s staff fiercely disputed that, noting that the chart says “equates,” which they say indicates it is not claiming that this money is spent only on people below the poverty line.
But that impression is certainly left, particularly given the way Sessions discussed the figure at the hearing: “We spend enough on federal welfare to mail every household living beneath the poverty line a check for $60,000 each year.”
In testimony before the House Budget Committee in 2012, Robert Rector of the Heritage Foundation said that simply dividing the means-tested spending by the number of the poor “can be misleading because many persons with incomes above the official poverty levels also receive means-tested aid.” He recommended dividing the figure by the bottom third of the income distribution, which yielded a figure of $36,000 for a family of four.
The Congressional Budget Office, in a report this month, had an even more nuanced approach, estimating the average federal spending per household in 2006 for the 10 largest means-tested programs (worth about 75 percent of Sessions’s total) by different income quintiles (See Box 1.) For the lowest quintile, the figure is nearly $9,000, after adjusting to 2012 dollars.
In both cases, when a more nuanced approach was taken, the headline number shrinks.
“This calculation was not intended to trick people into thinking that poor households receive $60,000 in benefits. It was also not intended to suggest that these programs are available only to those in poverty,” the Sessions aide explained. “It was intended to start a discussion about how much we spend on programs that most people believe are intended to support poor people, what the benefits include (e.g., we’ve talked about showing how small an amount relative to the total is directed at jobs programs), and who the benefits are for…. We intended this statistic to open up a dialogue about how we (as a society) define the welfare state and who should be covered by that welfare state.”
Those are certainly worthy questions. One could also explore whether some benefits add to a disincentive to work, because workers face a “cliff” in which they rapidly lose more benefits for each extra dollar they earn, as the chart below from Gene Steuerle of the Urban Institute shows.
Gene Steuerle, blog.governmentwedeserve.org
The Pinocchio Test
Sessions — who says he went to college on Pell Grants, which he lists as a welfare program — clearly wants to jar some of the conventional wisdom about anti-poverty programs. Calling attention to their cost and questioning their effectiveness are certainly worthy endeavors for a lawmaker.
But he runs into trouble with his chart and the shorthand description he gave at the hearing. The comparison to medium income is specious, given that it does not include health-care benefits, while the cost per household appears inflated, given that the programs he lists cover a whole range of programs that assist more than just those in poverty. As we noted on Wednesday, it is a matter of opinion what to consider “welfare.” But the presentation in the chart is simply misleading.
We wavered back and forth between Two and Three Pinocchios, but the apples-and-oranges comparison to median income tipped it to Three. This chart needs to go back to the drawing board.
UPDATE: Stephen Miller, spokesman for Sessions, asked to post a response to this column:
Who watches the Post's watchman? Your piece is disappointingly anti-intellectual, appealing to sentiment over reason. No reasonable analysis can exclude the cost of providing low-income families with subsidized health care when calculating the total financial aid that they receive.
You claim our chart is inaccurate and yet you fail to provide your readers with the math. So here it is: according to CRS’ report (which you do not contest), spending on 80-plus federal poverty programs equals $746 billion per year. Include state contributions to those programs and it grows to over $1 trillion. (That number would be even larger if one added spending on state- and local-run programs, which we did not.) By comparison, we spend $480 billion on Medicare, $540 billion on defense, and $725 billion on Social Security.
Although you did not acknowledge it, the Committee released the charts with a detailed and extensive explanation of the underlying methodology. As our release stated, “almost 110 million Americans received some form of means-tested welfare in 2011.” We then went on to explain that if “the $1 trillion spent on federal welfare programs [were] converted into cash and divided exclusively among the 16.8 million households who lived beneath the federal poverty line last year, the government would be able to mail each of those households an annual check for $60,000.” The release concluding by saying that “this figure underscores the fragmented, inefficient nature of welfare in this country.”
We can have a political debate about what to call these poverty programs, how to most effectively allocate our resources, and what kind of reforms would best alleviate the chronic poverty created by the Left’s policies. What is not debatable is the math. Unlike your post, our analysis is honest, accurate and, most importantly, a constructive step towards helping those in need.
Check out our candidate Pinocchio Tracker