Combining Time Series and Cross‐Sectional Data for the Current Employment Statistics Estimates

Julie B. Gershunskaya

Abstract

Estimates from the Current Employment Statistics (CES) Survey are produced based on the data collected each month from the sample of businesses that is updated once a year. In some estimation cells, where the sample is not large enough, the Fay-Herriot model is used to improve the estimates. Under the current approach, the model combines information from a set of areas and is estimated independently every month. Given the design of the survey, it may be beneficial to borrow information not only cross-sectionally but also over time. This paper explores the feasibility of applying such a model. The results are evaluated based on historical "true" employment data available on a lagged basis.