In 2014, critics of welfare reform—signed into law 20 years ago this week—received a big boost from a passionate book about the extremely poor. While official statistics indicated poverty among the children of single mothers was lower than in 1996, there were some indications that a sizable minority was worse off.
But in “$2 a Day: Living on Almost Nothing in America,” Kathryn Edin and Luke Shaefer claimed 1.5 million households were getting by on no more than $2 a day per person in 2011—double the 1996 number. They explicitly blamed welfare reform for this increase. The book was a sensation among critics of reform.
The picture they painted, however, bears little resemblance to reality, as I show in my new study, “Poverty After Welfare Reform.” In truth, practically no one in the United States gets by on $2 a day in any meaningful sense for any meaningful length of time. That doesn’t mean we should be cavalier or complacent about the amount of child poverty in America, but it does mean that calls to weaken the work-incentivizing structure of cash welfare—technically, Temporary Assistance for Needy Families, or TANF—are misguided.
Counting the Cost—Accurately
For starters, Edin and Shaefer’s own numbers indicated many fewer children were living on $2 a day per person if they counted federal food stamp and housing benefits and refundable tax credits as income. More to the point, 80 percent of the increase in $2-a-day poverty disappeared. Rather than rising from 1.7 percent of children getting by on $2 a day per person in a typical month in 1996 to 4.3 percent in 2011, their more comprehensive measure rose only from 1.1 percent to 1.6 percent.
My own study found that on the basis of cash income, 1.2 percent of children were getting by on $2 a day per person over the course of 1996, compared with 1.7 percent in 2014. However, when I included food stamps, housing benefits, subsidized school meals, and energy subsidies in income, the percentage increased from just 0.6 to 0.7. The conclusion was the same after counting refundable tax credits in income and using a better cost-of-living adjustment to increase the poverty line over time.
Counting health-care benefits as income resulted in a flat trend, with 0.5 percent of children in $2-a-day poverty in 1996 and in 2014—one in 200. Further, a partial correction of the data to account for the well-known underreporting of government benefits left the rate of $2-a-day poverty at 0.1 percent in both years—one in 1,000 children. The rate was the same for the children of single mothers.
These results are consistent with other analyses that look comprehensively at the resources available to poor families with children. One study looking at spending found that one in 1,000 households consumed no more than $2 a day over the course of 2012. Another study using New York state administrative data found that one in 500 children of lower-income single mothers were in families with no earnings or federal benefits in the form of cash or food stamps.
Poor People Underestimate Their Earnings
Edin and Shaefer maintain the rise in the number of children with no more than $2 a day per person in cash is worrisome and reflects the harm welfare reform has done. Yet these estimates are also deeply suspect. Why? Because we know from comparisons of what people say they earn in household surveys to administrative estimates that collect earnings data from employers that surveys tend to substantially understate earnings among low-income families. In one study, the number of workers among the poorest earners was larger by two-thirds than in the survey that I used.
In fact, Edin’s own research in the late 1980s and early 1990s interviewing low-income welfare recipients found their true incomes were roughly 40 percent higher than the sum of private income that would typically be reported on surveys and government benefits. Nearly all of them had earnings or other sources of income that they concealed from their welfare caseworkers. They remained very poor—just not as poor as one would find absent the careful trust-building work in which Edin engaged.
If earnings underreporting increased after 1996—or was unchanged but well-reported sources of income fell—then it is possible that $2-a-day poverty may have fallen over time, like the official poverty estimates. At any rate, with earnings fully reported, the number of children living in such hardship would be small enough as to be practically nonexistent.
There are other signs that the cash-income $2-a-day figures should not be trusted. Edin and Shaefer found that only three in five of the households living under this threshold either had experienced physical housing problems, food insecurity, medical hardship, or a residential move—not much different from children who were below one-and-a-half times the poverty line but not under $2 a day.
Similarly, in new research, Robert Rector and Jamie Bryan Hall find almost no households supposedly living on $2 a day or less “often” report having insufficient food to eat, having been evicted in the past year, or having had their utilities cut off. The vast majority of them have air conditioning, a DVD player, and a cell phone. Two thirds have a computer.
Inflating the Truth Helps Nobody
Even if we put aside these problems, if the data are taken seriously, they fail to implicate welfare reform in the “rise” in $2-a-day poverty. That’s because the supposed rise began in the 1970s, according to my study—well before welfare reform.
Furthermore, $2-a-day poverty based on cash income rose among groups that were unaffected by welfare reform: childless households, the elderly, the children of married parents, and even married college graduates. It is possible that extreme poverty has been quietly increasing among all these groups—driven by welfare reform among the children of single mothers and other things among the other groups. But it is much more likely that the supposed increase is simply the result of measurement problems.
Advocates for the poor—among whom I count myself—are often tempted to paint the worst-possible picture of how the poor are doing, to rally public support for assisting them. Pointing out when claims are overblown or wrong makes for unpopularity and suspicion of motives. But we do the poor no good if we base policy on an inaccurate depiction of their situation. “$2 a Day” combined a vivid description of the lives of eight extremely poor families with an analysis of national data that incorrectly implied severe hardship was widespread and growing worse. It was effective at convincing many people primed to dislike welfare reform that it had had devastating consequences.
In fact, evidence suggests welfare reform was the most important antipoverty policy since the national expansion of food stamps in 1974. I believe that to be true, yet I favor greater federal spending (spent differently) on the poor. Wanting to help poor children does not require one to believe that poverty is worse than it is or worse than in the past. But whether one finds the case persuasive or not, critics of welfare reform will have to find a different reason to oppose it than the supposed rise in extreme poverty.