Use of ACS Data to Produce SAIPE Model-Based Estimates of Poverty for Counties

摘要

The U.S. Census Bureau's Small Area Income and Poverty Estimates (SAIPE) program has produced median household income estimates and poverty by age group estimates for U.S. states and counties since 1993 (estimates released in 1997), and population and poverty estimates of school-aged (ages 5 - 17) children for school districts since 1995 (estimates released in 1999). The estimates are now produced annually. The state and county estimates have been produced using Fay-Herriott (1979) models fitted to direct income and poverty estimates from the Annual Social and Economic Supplement (ASEC) of the Current Population Survey (CPS). Due to a lack of data for modeling at the school district level, the school district estimates have been produced using a simple updating scheme that takes the estimated shares of the number of school-aged children in poverty among school districts within each county from the previous decennial census and multiplies these by the current years county level model-based estimates of the number of 5 - 17 year old children in poverty. The school district poverty estimates are particularly important because the U.S. Department of Education uses them to allocate Title I funds (about $13 billion dollars in 2006). The state and county estimates are of independent interest, but are also important because these estimates are constrained to be consistent across the various geographic levels. The share approach used for the school district estimates forces them to aggregate to the county estimates, but in addition, the county model-based estimates of number in poverty are raked so they add to their corresponding state model-based estimates. The latter are raked so they add to the CPS ASEC direct national estimate.