Frequency Domain Analysis of Seasonal Adjustment Filters Applied to Periodic Labor Force Survey Series

• Activity Number: 549
• Type: Invited
• Date/Time: Wednesday, August 1, 2012 : 2:00 PM to 3:50 PM
• Sponsor: Section on Government Statistics
• Author(s): Richard B. Tiller*+
• Companies: Bureau of Labor Statistics
• Keywords: Trend ; Survey Error ; Real time filtering

Abstract

Statistical agencies routinely seasonally adjust large numbers of time series generated from periodic surveys that often use a rotating panel design in which a portion of the sample is retained each period. While these surveys produce highly reliable estimates for national aggregates, demographic and sub-national series are based on much smaller samples. As a result, survey error is a major source of variation in the observed series which is ignored when using conventional time series decomposition filters. Using frequency domain diagnostics, this paper examines the effects of sampling error on the X-11/X12-ARIMA and SEATS filters and the gains from combining these conventional filters with a survey error filter developed from a model based approach that explicitly accounts for survey error. Of special interest is how the outputs of the conventional filters compare to the optimal filter in real time. Data from the Current Population Series are used as examples.