Using Area Characteristics to Model Nonresponse and Late Reporting in the Current Employment Statistics Survey

John S. Dixon

Abstract

Previous research has found a relationship between characteristics of an establishment, such as firm size, and nonresponse. To build on this, we incorporate information about the establishment's geographic area in models to explore response patterns in the Current Employment Statistics Survey (CES). Area information is known to have an impact on employment levels; for example, education level of the local population may limit employment which may lead to establishments being less likely to respond to the CES. Using information from the Census planning database, we add characteristics about the establishment's area, such as age, education, unemployment rate, to other information known about the establishment from the Quarterly Census of Employment and Wages (QCEW). Building a model with both area and establishment characteristics, we explore patterns in both nonresponse and late responding.