A Hierarchical Bayesian Approach to Estimation for the Annual Survey of Public Employment and Payroll

• Activity Number: 163
• Type: Topic Contributed
• Date/Time: Monday, August 10, 2015 : 10:30 AM to 12:20 PM
• Sponsor: Government Statistics Section
• Author(s): Brian Dumbacher* and Michael Larsen
• Companies: U.S. Census Bureau and The George Washington University
• Keywords: Small area estimation ; Composite estimator ; Hierarchical Bayes ; Model validation ; Government units

Abstract

The Annual Survey of Public Employment and Payroll (ASPEP) is conducted by the U.S. Census Bureau to collect data on federal, state, and local government civilian employees. Estimates of local government totals are calculated for domains created by crossing state and government function, where functions range from air transportation to water supply. To calculate estimates at such a detailed level, the Census Bureau uses small area methods that borrow strength from other domains through auxiliary information from the most recent Census of Governments. In this paper, we study the properties of the composite estimator used during ASPEP's 2009 sample design and explore a new hierarchical Bayesian approach to small area estimation. We consider various models and investigate model diagnostics.