Calculating Poverty

To preview how the Census Bureau’s new Supplemental Poverty Measure may change the portrait of poverty, The New York Times consulted multiple alternate measures that researchers have quietly published in recent years.

They included a 2009 version of the Supplemental Poverty Measure, as well as four of eight alternate measures the Census has published, incorporating recommendations from the National Academy of Sciences.

They also included studies from the University of Wisconsin and from officials in New York City, along with three state analyses by Sheila Zedlewski and colleagues at the Urban Institute.

While there are differences in the calculations, all go beyond the official measure by counting benefits like food stamps, work expenses, taxes, and the cost of living.

In addition, The Times did an in-depth analysis of one experimental data set (known as MSI-GA-CPI). It is the Census measure that comes closest to the new Supplemental Measure, while also providing a methodologically consistent look back at prerecession years.

That produced a number of interesting figures that our article length could not accommodate.

Among children, it showed poverty rates falling to 15 percent, from 22 percent, in the official count. That removes about 5.2 million children from poverty. That drop is broadly consistent with what the Urban Institute researchers found in Massachusetts, Illinois and Georgia — an average decline in child poverty of about 24 percent.

It also falls sharply for women ages 25 to 39 — many of them mothers. Many safety net programs, particularly the Earned Income Tax Credit, focus on families with children.

Among elderly, The Times’s examination of the Census data showed poverty rates rising to 13.8 percent, from 9 percent, by the official count. The 2009 Supplemental Poverty Measure and the Urban Institute studies found broadly similar climbs.

Some experts found this possibly portentous — a complication in the generally accepted story that poverty among the elderly has been on a long-term decline. Irwin Garfinkel of Columbia University said his own studies of alternate poverty measures showed that the figure had been climbing for a decade. “We’ve been slipping backwards and we didn’t know that before,’’ he said.

But others, including Bruce Meyer of the University of Chicago, cautioned that it could be a statistical anomaly. Many elderly people use retirement savings but do not report that money as income. And measures of consumption among the elderly, these experts say, have not shown a rising level of hardship.

In The Times analysis, poverty rates in rural America drop sharply, to 10.9 percent from 16.4 percent in the official count, most likely reflecting the lower cost of living. But they rise in metropolitan areas, to 14.9 percent from 13.9 percent. While poverty rises in the Northeast and West, it falls in the South and Midwest, by a total of about six million people.

The Times study found poverty dropping sharply among blacks, falling by 6.7 percentage points to 20.6 percent. Most other studies have also showed significant declines among blacks, which could reflect either that they receive relatively more benefits or live in relatively low-cost locations.

The Times study found poverty growing among Asians, as did all the others. It found poverty declining among Latinos, though by a much smaller amount than it declined among blacks. Some other studies have found Latino poverty rates rising. One possibility is that Asian and Latino immigrants have a harder time getting access to benefits. Another is that differences between the official and alternate studies reflect where people happen to live, perhaps disproportionately in high-cost areas.

What elements of the new measure lift people from poverty? The Times examined the characteristics of those Americans considered poor by the official count, but not poor by the alternate. About 70 percent lived in households that received food stamps a year. A third lived in households with housing subsidies and 19 percent lived in households that received the earned income tax credit.

A caveat: this data is not the same as what the Census Bureau will release on Monday, which among other things uses a different cost-of-living adjustment. But while the numbers may vary, the general trends are likely to remain the same.