Head Start. Title I of the Elementary and Secondary Education Act. Welfare reform.

(January 2001) With funding for all these programs linked to estimates of poverty published by the U.S. Census Bureau, the agency is drawing on several types of data to refine and update its numbers between censuses.

In November, the Census Bureau released 1997 estimates of income and poverty for every state and all U.S. counties, along with estimates of poverty and population for the country's 15,000 school districts. These estimates, produced by the Small Area Income and Poverty Estimates (SAIPE) program, include:

  • the number and percentage of poor people (for states and counties)
  • the number and percentage of poor people under age 18 (for states and counties)
  • the number and percentage of poor children under age 5 (for states)
  • the number and percentage of children ages 5 to 17 living in poverty in a household headed by a relative (for states, counties, and school districts)
  • median household income (for states and counties)
  • the number of children ages 5 to 17 in poverty (for school districts).

The publication of the SAIPE estimates means that there are now three "official" sources of poverty data: the decennial census, the Current Population Survey (CPS), and SAIPE. But the SAIPE numbers are the official ones for counties and smaller jurisdictions. In fact, they are the only reliable local area poverty figures available between censuses. The Census Bureau produces state-level SAIPE estimates for every year, while county and school district estimates are done for odd-numbered years. At the state level, SAIPE estimates have another advantage: They are more accurate than state estimates based on multiyear CPS averages.


Percent of People in Poverty by State, 1997

Percent of People in Poverty by

Source: U.S. Census Bureau.


SAIPE statisticians produced their estimates for 1997 by combining results from the Current Population Survey with data from federal individual income tax returns, administrative records on food stamps, and 1990 decennial census figures. They derived school district estimates by combining 1990 census data for school districts with 1997 county poverty and population estimates.

To reduce sampling variability for the CPS estimates of income and poverty for counties, SAIPE staff compute three-year weighted averages centered on the nominal estimation year. For example, to estimate income and poverty for 1997, the SAIPE program uses weighted averages of income data from the March 1997, March 1998, and March 1999 CPS surveys. That is why the estimates for 1997 could not be derived until data from the March 1999 CPS survey became available.

The latest results, along with 90 percent confidence intervals showing their variability, appear in the figure above.

For more information on SAIPE, visit the Census Bureau's website: www.census.gov/hhes/www/saipe.html.


Kerri Rivers is a PRB research associate, and Kelvin Pollard is a PRB demographer.