Model-Based Seasonal Adjustment of Survey Series Subject to Benchmark Constraints with a State-Space Smoothing Algorithm

• Activity Number: 339
• Type: Invited
• Date/Time: Wednesday, August 10, 2005 : 8:30 AM to 10:20 AM
• Sponsor: Section on Survey Research Methods
• Author(s): Richard Tiller*+ and Daniel Pfeffermann
• Companies: Bureau of Labor Statistics and Hebrew University & University of Southampton
• Keywords: Small Area Estimation ; Correlated Measurement Error ; Components model

Abstract

The Bureau of Labor Statistics uses state-space models to seasonally adjust small-area labor force estimates from the Current Population Survey (CPS). The models combine classical component models assumed for the population time series with a model for the sampling errors. In order to protect against possible model breakdowns and for consistency in publication, the model dependent estimates at the small-area level are benchmarked to the direct survey estimate in a group of areas for which the survey estimate is sufficiently accurate. In a previous paper, Pfeffermann and Tiller (2004) developed and illustrated a new filtering algorithm that produces real-time estimates satisfying stochastic benchmark constraints along with their variances. This paper extends that work by developing a smoothing algorithm to produce historical seasonally adjusted and trend estimates. As with filtering, variances are developed that account for errors in the benchmark as well as any reduction in uncertainty that may occur from borrowing strength across areas. In addition, one can examine how the smoothness of the original trend model is affected by the benchmarking.