The Odd Conservative Argument That Food Stamps and Medicaid Saved the Poor From Welfare Reform

This is not the defense Bill Clinton is looking for.

Win McNamee/Getty Images

This week marked the 20th anniversary of welfare reform, the landmark legislation signed into law by Bill Clinton that many on the left, myself included, have accused of eviscerating a key part of the safety net that protected the poorest of America’s poor—namely, penniless single mothers. In honor of the occasion, Scott Winship of the conservative Manhattan Institute has come out with a new report defending reform’s legacy, while attempting to debunk many of the statistics liberals typically wield against it (Republicans have long argued that the bill was a rousing success). It’s a lengthy, careful piece of work, but I doubt it will change many minds on this issue. It also implicitly makes the argument that welfare reform was harmless in part because Medicaid and food stamps saved the most vulnerable from deep deprivation—which is a tad awkward to hear coming from the right.

Clinton made good on his promise to “end welfare as we know it” by dismantling Aid to Families With Dependent Children, the old entitlement program that had sent cash to needy households largely headed by single mothers. He replaced it with a new system called Temporary Assistance to Needy Families, or TANF, which among other things attached work requirements and time limits to cash benefits, and turned welfare’s funding into a fixed block grant that states had the de facto power to spend however they pleased. The idea was to put more emphasis on efforts that would move jobless mothers into work. To simplify a bit, the rap against TANF is that it turned welfare into a slush fund that states have used to plug random budget holes, and that while the reforms may have actually helped reduce the poverty rate overall by convincing some women to find work rather than apply for monthly benefits, it exacerbated the worst kinds of poverty by cutting off cash aid to the truly desperate.

This is, admittedly, a somewhat tricky argument to illustrate, because no single data source does a perfect job tracking the incomes of the very, very poor. When they’re asked to fill out government surveys, Americans notoriously under-report their earnings as well as the benefits they receive from programs like food stamps. Poor families with unstable housing situations can also be hard to track down, especially if they’re living in homeless shelters. So rather than tout a single big number, experts in this field have tended to rely on a constellation of data points that suggest Americans at the absolute bottom of the income distribution suffered in the post–welfare reform era.

Early research, for instance, found that the number of so-called disconnected mothers who were neither working nor in school nor receiving welfare benefits rose after 1996. More recently, Johns Hopkins University’s Kathryn Edin and the University of Michigan’s Luke Shaefer made serious waves when they reported that “the number of households living on $2 or less in cash income per person per day in a given month increased from about 636,000 in 1996 to about 1.65 million in mid-2011, a growth of 159.1 percent.” In the same vein, the Center on Budget and Policy Priorities found that, after including safety net benefits like food stamps and adjusting for under-reporting, the percentage of children living in so-called deep poverty, meaning their family income amounted to less than half the poverty line, rose from 2.1 percent in 1995 to 3 percent in 2005. Using a different data set, a group of researchers including Johns Hopkins’ Robert Moffitt found that, after taxes and transfers, the percentage of families in deep poverty rose from 4.5 percent in 1993 to 6.6 percent in 2004.* The number of food stamp recipients reporting no income has also gone up.

Winship attempts to pick these findings apart one by one. But while he does an excellent job raising questions and demonstrating why it’s dangerous to rest your argument on a single stat, I don’t think he succeeds in his main goal of casting doubt on the idea that, more likely than not, severe poverty has been rising.

Winship’s report starts off by making a basically uncontroversial but important point: Child poverty is lower today than it was in 1996, including among families headed by single mothers. The trend looks even better once you start adding the value of various safety net benefits like food stamps, housing assistance, and Medicaid into the mix. The important thing to keep in mind here, however, is that the 1996 welfare reform law is at best responsible for only some of these improvements. Prior to TANF, states had already begun reforming their welfare programs through a federal waiver program that required them to test their approaches and demonstrate positive outcomes (TANF imposed no such demands). And in 1993, Clinton expanded the earned income tax credit, which boosts after-tax incomes for low-wage workers and drew an enormous number of single mothers into the labor force. “Indeed, the sizable expansions of this particular anti-poverty tool appear to be the most important single factor in explaining why female family heads increased their employment over the 1993-1999 period,” University of Chicago economist Jeffrey Grogger wrote in one often-cited study. As for TANF itself? Economists Robert Schoeni and Rebecca Blank estimated that its creation led to a roughly 2 percentage point decline in the poverty rate for women with less than a high school degree (similar to the effect of the waivers).

Point being: TANF may have had some positive effects on the headline poverty rate, but it didn’t single-handedly work miracles.

Winship’s arguments about the poorest of the poor are interesting but less sturdy. Take his analysis of deep poverty, which, again, is the fraction of children living below half of the poverty threshold. Just like the Center on Budget and Policy Priorities, his results show that once you’ve accounted for the way households tend to under-report the benefits they receive from the government, deep poverty among children rose after welfare reform. He finds the rate is lower overall, because, among other things, he adjusts his numbers using a nonstandard measure of inflation called the Personal Consumption Expenditure Deflator, and includes Medicaid benefits as income. (The latter move is a debatable, but not entirely ridiculous, choice. The value of health insurance partly just reflects the absurd escalation of medical costs in this country, but it’s also obviously valuable). But the same basic pattern remains, as you can see from the green and purple lines below.

Winship argues that these changes are too small, and the data too imprecise, to make any judgments about it. When the CBPP looked at the issue, however, it found the rise in deep poverty after welfare reform to be statistically significant. Winship also suggests that the whole issue could be an artifact of bad data collection. Families receiving fewer welfare cash benefits, he argues, may just be under-reporting their incomes more severely than before:

Imagine a family where earnings are completely unreported—and large enough that, if reported, they would be enough to escape deep poverty. Imagine, too, that welfare income is reported fully and also amounts to over half the poverty line. Now imagine that welfare income disappears over time, so that those earnings eventually become the only income received by the family. The family will, at some point, appear to fall into deep poverty, even though it remains above the deep poverty threshold the whole time, thanks to its (unreported, unchanging) earnings.

This is an elegant theory. But in practical terms, it’s also not very reassuring, at least if you’re concerned about the well-being of America’s children. Winship is essentially saying that while it’s entirely possible  welfare’s demise left a substantial number of needy families a bit poorer, they may nonetheless have enough secret income to keep them above the somewhat arbitrary mark of 50 percent of the poverty threshold. I am not sure how many people are going to comforted by this speculation. Which is, I repeat, just speculation.

Winship’s attempt to debunk Shaefer and Edin’s now-famous $2-a-day statistic is equally shaky. The point of the pair’s research is that a slice of America is living largely without access to ready cash income, which leaves their lives especially prone to hardship and disaster. Food stamps and Medicaid are wonderful, but unless you illegally sell your SNAP benefits (which some people of course do), they don’t let you pay a phone or electricity bill. They don’t let you fill up your car to get to work or fix a leak that’s destroying your house. Even if you do include the value of certain benefits as income, the researchers find that the number of Americans living below the $2-per-person-per-day mark still rose in the wake of welfare reform.

Winship’s response is to mimic Shaefer and Edin’s analysis using a different data set. (They rely on the Survey of Income and Program Participation, while he chooses census figures. You can argue about which is superior for these purposes. I won’t bore you.) His graphs show that, in fact, the number of children living below the $2-per-day mark began its current upward trend prior to welfare reform. This suggests that Clinton’s legislation may have merely exacerbated an old trend rather than singlehandedly caused it. That’s an interesting point, but not necessarily helpful to his cause.

Winship then attempts to make the trend disappear by adding in Medicaid and other noncash benefits and tweaking his inflation measure before finally doing away with the conceit of measuring dollars, per-person. As he puts it:

Rather than dividing income by the number of household members to determine whether income is above $2 a day, Line 7 divides income using an “equivalence scale” that better accounts for the needs of households with different numbers of adults and children. I used an adjustment based on the recommendation of an expert panel convened in the mid-1990s. This adjustment lowers extreme child poverty rates even further.

Suffice to say, Winship is no longer measuring the same thing as Edin and Shaefer at all. Instead he’s gone and created his own measure of extreme hardship that, among other things, illustrates how government benefits—especially Medicaid and food stamps—have helped prevent a good deal of desperate need in this country in the years after cash welfare was eviscerated. This is encouraging, but a bit odd coming from a Republican, given that the party’s congressional delegation has spent the last six years vowing to shred those programs.

Winship’s paper is long and detailed, and I can’t possibly do it justice by addressing all of its arguments, point by point. But his main ones, again, do little in my mind to dispel the idea that something has gone wrong for the very poor in the two decades after welfare reform. He points out that that data on consumption tend to suggest lower levels of poverty than data on income. Of course, part of that may have to do with the ready availability of predatory credit in this country. Moreover, research has shown that the Consumer Expenditure Survey (which Winship’s favored researchers rely on) is a bit of an outlier when it comes to poverty trends, and doesn’t track well with other indicators like unemployment or consumption figures from the Panel Survey of Income Dynamics. He suggests that the rise of food stamp applicants with no income may be a product of the same under-reporting that he thinks influences deep poverty. This is, again, mostly speculative. But he does marshall some evidence. For instance, he cites a government study that interviewed 50 food stamp recipients and found a little more than half made money they didn’t report by doing things like odd jobs for friends and family, like mowing lawns. Of course, were their part-time lawncare businesses properly accounted for, I’m guessing most of those people would still qualify as dirt poor—to use a technical term.

Winship’s study is a valuable demonstration of just how much freedom researchers have when it comes to crafting poverty data. Add in some Medicaid benefits here, impute some food stamps there, and you can totally change a trend line. But while it raises questions about the stats underlying the case against welfare reform, it hardly disproves them.

Still, I do hope people read this paper. It’s not often a conservative mounts this kind of a defense of what remains of our safety net.

*Correction, Aug. 26, 2016: This post originally misspelled Robert Moffitt’s last name.