There’s a huge debate over how the very poorest are doing. There’s no debate about whether food stamps help them.

There’s a huge debate among scholars who study the lives of the very poorest about just how many very poor people there are in America.

Some researchers, using survey data, estimate there is a large and growing portion of Americans living on less than $2 day — in other words, in truly extreme poverty. Other researchers have made a strong claim that this data is severely exaggerated, arguing that surveys either exclude or just don’t end up counting substantial sources of income and assistance that the poorest get: tax refunds, Social Security, housing benefits, and especially the Supplemental Nutrition Assistance Program (SNAP), otherwise known as food stamps.

But where these analyses agree is that SNAP has become a primary income support to poor individuals and especially families. And a new paper from a group of researchers — Dean Jolliffe, Juan Margitic, and Martin Ravallion — published last week as a working paper through the National Bureau of Economic Research argues that the “floor” of US living standards, the very lowest amount that people in the US live on per day, has been falling, at least according to an analysis of survey data, and how SNAP is raising it higher.

The three economists try to estimate a floor of income using survey responses from the poorest, weighted toward the most poor, and then seeing how that floor has moved over time. They put this figure at about $4.55. With SNAP, however, the floor goes up to $4.93, about a third of the official poverty line.

But SNAP coverage in America is incomplete: Not everyone who’s eligible ends up receiving benefits. (They cite a range of estimates for how many people are left out, ranging from about a third, according to the Congressional Research Service, to 15 percent, according to the Center on Budget and Policy Priorities.) But, the authors argue, if SNAP coverage were more complete — “if all the poorest received the mean SNAP spending per SNAP recipient” — the floor would rise to around $8 per day.

This type of data has been the subject of a massive debate recently, not just because of what it counts and what it doesn’t, but how accurately it reflects what those surveyed say it does.

Bruce Mayer and his fellow researchers have argued that the bottom of the surveys are “likely to yield a group filled with more gross errors than households that are truly impoverished,” pointing to factors like those receiving benefits that are greater than the lowest levels of poverty that show up on the survey. They have heavily criticized other estimates of extreme deprivation, arguing that very few Americans and vanishingly few families with children live at or below $2 or $4 a day. For the specific survey, the Current Population Survey, used in the Jolliffe paper, they argue that the rate of extreme poverty is much lower, thanks to misreported earnings and assets data, as well as including transfer programs like Social Security and tax credits. In Mayer’s work, some of the biggest adjustments to reported poverty come from transfers like SNAP, showing that even if there isn’t a consensus about how poor the poorest are or if that portion of the population is expanding, there is a consensus on the importance of food stamps.

Joliffe and his collaborators have tried to address limitations in the data. They use survey data, and they do adjust it to throw out non-zero incomes, but they assume that the lowest reported positive income “has the highest probability of being the poorest.” They note that doing similar studies measuring what people buy, which some argue is a better way of measuring the living standards of the very poorest, in other countries has resulted in higher estimates of the “floor” but “a very similar evolution over time.”

One of the more striking findings is that while poverty measures of all sorts jumped in response to the financial crisis and subsequent economic downturn, the authors’ income “floor” stayed “relatively stable,” indicating that efforts to expand SNAP as a stimulus and income support measure were effective for the very poorest. But they also argue that the program could be better at reaching the very poor, pointing to how welfare reform in the mid ’90s limited the very poor’s access.

“The program helped assure that the poorest could at least maintain their (low) living standards during a period of inequitable growth,” the authors write. “We find a marked longer-term decline in the extent to which SNAP has raised the floor, associated with declining efficacy of food stamps in reaching America’s poorest since the mid 1990s.”