Five Decades of Poverty

Consumption data show greater improvements in poverty rates and the depth of poverty than income data.

Despite a range of public programs that are designed to ameliorate poverty, the official U.S. poverty rate is virtually the same today as it was in 1970. Rather than being an indictment of the effects of economic growth or anti-poverty programs, this fact reflects important flaws in how official poverty is measured. In Five Decades of Consumption and Income Poverty (NBER Working Paper No. 14827), Bruce Meyer and James Sullivan present alternative measures of poverty that more accurately reflect how well-being has changed for disadvantaged families since the early 1960s. They examine improved measures of income based poverty as well as measures based on consumption. They emphasize that changes in the poverty rate over the past five decades differ considerably depending on how poverty is measured.

Among the biggest concerns with the official U.S. poverty rate is that it is based on pre-tax money income. Consequently, the measure ignores tax programs and in-kind transfers that provide resources to the poor, including the Earned Income Tax Credit, food stamps, housing assistance, Medicaid, and the National School Lunch Program. The Earned Income Tax Credit is one of the largest anti-poverty programs in the United States, but its subsidies show up in after-tax rather than pre-tax income. This study shows that between 1990 and 1996, the expansion of the Earned Income Tax Credit resulted in a decline in poverty of 1.3 percentage points. And between the early 1960s and 2005, a money-income poverty measure that incorporates taxes declined by an additional 3.9 percentage points as compared to a pre-tax measure of income poverty.

Another important concern is that official poverty, "while described as an absolute measure, is far from it." The official poverty measure over-adjusts for inflation because it is based on the Consumer Price Index, which has some well-known biases. The authors make clear that poverty rates vary significantly depending upon how the poverty measure is adjusted for inflation. For example, correcting for known biases in the Consumer Price Index increases the decline in poverty between the early 1960s and 2005 by 14 percentage points.

Another way of gauging poverty is in terms of consumption, rather than income. The authors argue that a consumption based measure of poverty offers advantages over an income based measure, not only because consumption is a more direct measure of well-being but also because evidence suggests that consumption is more accurately measured for those at the bottom. For single mothers, for example, the fifth percentile of expenditures exceeds the fifth percentile of income by 50 percent; the difference is 25 percent at the twentieth percentiles, the authors report. This difference is attributable in large part to underreporting of income, particularly transfer income such as food stamp benefits and welfare payments.

In general, the authors find that consumption data show greater improvement in poverty rates and the depth of poverty than income data. They note that the amount of improvement varies across family types: for example, since 1980 poverty as measured by consumption has fallen faster than income based poverty for single parents and the elderly, but consumption based poverty has fallen more slowly for married couples with children.

-- Linda Gorman

The Digest is not copyrighted and may be reproduced freely with appropriate attribution of source.

National Bureau of Economic Research, 1050 Massachusetts Ave., Cambridge, MA 02138; 617-868-3900; email:

Contact Us