Our Poverty Data May Be Completely Wrong

The United States federal government spends about $236 billion per year on programs for low-income populations, not counting health insurance. That’s roughly $2,000 per American household. Yet some criticize the American safety net as inadequate, citing increases in deep poverty (defined as less than 50 percent of the official Federal Poverty Level) or those "disconnected" from government assistance. But a new study shows that some statistics on poverty are dramatically overstated because of flawed survey data. This should change how we view the poor in the United States and the programs that serve them.

According to the study, by University of Chicago researchers Bruce Meyer and Nikolas Mittag, survey respondents in deep poverty reported only 52 percent of government resources they received. Considering all people in poverty, survey data captured only 46 percent of government assistance. The implication was that poverty researchers — who often recalculate the poverty rate to include income from programs that the official metric ignores — have been reporting bad numbers. Using this type of measure, with the thresholds from the official metric but more safety-net income included, single-mother households had a poverty rate of 30.2 percent using the survey data, but only 19.2 percent once underreporting was accounted for.

A much smaller percentage of single mothers were also found to "fall through the cracks" completely when factoring in underreporting. Past research suggests that roughly 20 percent of single mothers had no earnings and no income from cash welfare, suggesting that government programs were failing them. But Meyer and Mittag found that relying on survey data overstated the percentage of these "disconnected" single mothers by one-third.

The researchers discovered this phenomenon by looking at data from New York State, comparing what individuals told the Census's Current Population Survey with data from state agencies indicating what they actually received in cash, food, housing, and other benefits. Based on their analysis, the authors believe that the results are applicable outside of New York State, and that the data flaws are likely consistent across most household surveys.

These results require that we reassess our view of the poor and the programs designed to serve them. An analysis in the recent book $2 a Day: Living on Almost Nothing In America, by Kathryn Edin and Luke Shaefer, found that 1.5 million households were living on less than $2 in cash a day in 2011 — a number that almost doubled since welfare reform in 1996. They argued that TANF, the government’s safety net of last resort, was dead. But is that number real? The work by Meyer and Mittag suggests it is likely not.

What does this mean for policy? First, reexamining the reach of government safety-net programs with misreported survey data in mind is needed. Some research does this, but few studies link survey and administrative data at the individual level as Meyer and Mittag did. Doing this at a national level will likely show that the safety net is much stronger than previously thought. Policymakers might have very different conversations about safety-net programs if they are shown to have larger poverty-reducing effects.

Second, researchers should reconsider conducting analyses that rely on flawed Census data. Unfortunately, this means that our understanding of program participation and recipient characteristics might be severely limited. But limited knowledge may be better than flawed knowledge, especially when billions of federal dollars are at stake.

Finally, access to better data is needed. This means more efforts to improve the quality of survey data, as well as increased access to linked survey and administrative data. Interest among academics and policymakers in alleviating poverty is strong. But continuing to rely on flawed data won’t help people in poverty, and might even harm them if it takes us in the wrong direction.

Angela Rachidi is a research fellow in poverty studies at AEI, where she studies the effects of public policy and existing support programs on low-income families.