Examining the Impact of Updating More Months of Concurrent Seasonally Adjusted Industry Estimates in the Current Employment Statistics

Brenda Loya

Abstract

With each monthly release of not seasonally adjusted employment estimates, the Current Employment Statistics (CES) program updates the two previous month’s preliminary estimates to reflect additional data receipts. The updated estimates are then used to produce concurrent seasonal adjustment factors and these are applied only to the revised estimate levels. As a result, part of the seasonally adjusted over-the-month change for two months prior to the newly released month is attributed to levels produced from different concurrent runs, creating what we call a seam effect. This paper uses simulated data, created by using the original seasonal adjustment specifications, to examine the impact of changing from the current monthly seasonal adjustment process to updating up to 61 months of seasonally adjusted data with each monthly release, a process similar to what is done during the annual benchmark process. Updating 61 months of data moves the seam effect from two months prior to 61 months back to provide a five year history of seam effect free estimates. The change also results in a higher probability of the peak and trough dates shifting and higher exposure of variability.