Aspects of Survey and Model-based Postcensal Estimation of Income and Poverty Characteristics for States and Counties

Congress has charged the Census Bureau with the responsibility to develop postcensal estimates of basic income and poverty statistics by state, county, and subcounty areas. The March Supplement to the Current Population Survey (CPS), based on a sample size of about 65,000 households, provides national estimates for these characteristics annually, but the sample size is insufficient to provide reliable direct estimates at all or even most levels of detail requested by Congress. The paper reports on two aspects of the estimation problem: 1) variance estimation for the CPS sample estimates, and 2) features of some candidate small area models that might be used to produce the required estimates at the state and county level. The two problems are interrelated: many small area models employ estimated variances for the survey estimates both in the estimation of the small area models and in their evaluation. The paper reports progress in developing a first set of estimates to meet the Congressional request.