Census errs again, makes mistakes measuring poverty nationwide

The U.S. Census Bureau erred in calculating a major national poverty measure — the second mistake reported by the bureau in the last month.

The so-called Supplemental Poverty Measure (SPM) was released with faulty data in mid-September. A corrected version, with a slightly lower poverty rate for the country, is expected to be distributed on Monday or Tuesday, according to a bureau official.

Last week, census officials admitted that they had identified widespread inaccuracies in another set of data specifically for Philadelphia, describing the city’s populace in 2017 in terms of poverty, income, employment, health insurance, rent, and more. However, that material will be revised in a release of data in December known as 5-year estimates, which include information collected on poverty and other topics from 2014 to 2018.

The supplemental measure is considered more comprehensive than the traditional method of gauging poverty because it totes up numerous factors about a person’s life to determine need.

The traditional measure takes into account only family size and income and is “wildly out of date,” said Kate Scully, director of government affairs at Philabundance, the region’s leading antihunger agency.

The supplemental measure looks at all the benefits received by those in poverty, such as food stamps, earned income tax credits (EITC), special aid to young mothers, and school meals.

These programs make a significant difference in people’s lives. For example, according to the census report, the measure showed that the number of Americans who moved above the poverty line thanks to EITC totaled 7.9 million in 2018; food stamps lifted an additional 3.1 million out of poverty, and rental assistance 3 million more.

On the other side of the ledger, the supplemental measure also calculates how much a family has to pay out to survive, including health care, taxes, child care, and housing.

“The supplemental measure gives us a better idea of who’s struggling,” said sociologist Joan Maya Mazelis, a poverty expert at Rutgers University-Camden. “It’s disappointing the Census Bureau had to throw out the numbers.”

Census experts said the mistakes were connected to changes in the U.S. tax code instituted by the Trump administration.

Vital for understanding who we are as a nation, the SPM shows us “whether antipoverty efforts are working for different parts of the population,” said Arloc Sherman, senior fellow at the left-leaning Center on Budget and Policy Priorities in Washington.

The SPM is a rarity in the world of poverty in that it is a measure esteemed by both the left and the right, said Joel Berg, CEO of the nonprofit Hunger Free America. “Progressives favor it because it shows a family’s expenditures; the right likes it because it shows the safety-net help people receive.”

Because wages of working-class and low-income workers stagnated or decreased over the last 20 to 30 years, “the only thing holding people up are benefits programs,” said Jim Weill, president of the Food Research & Action Center in Washington, the nation’s largest antihunger lobby.

“The Supplemental Poverty Measure is the better measure for current times by acknowledging family expenses, but also by counting key in-kind programs like [food stamps].”

Unlike the traditional poverty measure, which is used by the federal government to determine benefits allotted to impoverished Americans, the SPM isn’t tied to any program, but is valued by academics and antipoverty advocates as a more nuanced tool to understand what low-income people face.

In the erroneous SPM report, the U.S. poverty rate was reported to be 13.1% in 2018. This was higher than the 11.8% poverty rate registered by the traditional measure.

The corrected SPM number, to be released nationwide at the beginning of the week, is 12.8%, a census official said.

Explaining how the 2017 Philadelphia glitch came to be, a census official said errors were made among census workers collecting data from residents during a survey.

In the case of the SPM, a census official said miscalculations were attributable “to an error in our tax model.”

This is what that means, based on interviews with experts from the Census Bureau, as well as from universities and think tanks that deal with census material:

When the census compiles the SPM, it surveys about 100,000 households, asking questions about income, expenses, and any safety-net help they receive, such as food stamps.

What census pollsters don’t ask are questions about taxes, because it’s been found that people either don’t want to answer or, more likely, don’t know exactly how much they pay.

To compensate, the Census Bureau created a so-called tax model that estimates Americans’ taxes.

But because the Trump administration instituted numerous changes in the U.S. tax system in 2017 — primarily a tax cut for better-off citizens — the census tax model had to be altered.

In making those adjustments, the Census Bureau inserted a bad line of code relating to a child tax credit. An outside observer found the error and the bureau set about fixing it.

“With all the Census Bureau’s computerization, there’s still possibility for human errors,” said Indivar Dutta-Gupta, co-executive director of Georgetown University’s Center on Poverty and Inequality. “But I’m not worried. Literally every year, the bureau improves.”

Some antipoverty advocates wondered whether the Census Bureau, which is gearing up for the 2020 census count of all Americans, has enough people handling data, and whether more are needed to avoid errors such as the ones discovered in the last month.

In response, a census spokesperson said, “We are confident in the resources we have to conduct the census and continue the work on our ongoing surveys.”

The Inquirer is one of 21 news organizations producing Broke in Philly, a collaborative reporting project on solutions to poverty and the city’s push toward economic justice. See all of our reporting at brokeinphilly.org.

This story has been updated to reflect new information from the U.S. Census Bureau.