An official website of the United States government

The .gov means it's official.
Federal government websites often end in .gov or .mil. Before sharing sensitive information, make sure you're on a federal government site.

The site is secure.
The https:// ensures that you are connecting to the official website and that any information you provide is encrypted and transmitted securely.

Bureau of Labor Statistics > Current Employment Statistics > Home

CES introduces use of TRAMO for seasonal adjustment model selection

The Current Employment Statistics (CES) survey uses a program called X-13ARIMA-SEATS to seasonally adjust its data. This program, developed by the U.S. Census Bureau and Statistics Canada, uses a procedure to automatically select from a variety of ARIMA models during the annual seasonal adjustment process. Effective with the release of January 2017 data on February 3, 2017, CES began using an alternative method of automatic model selection known as Time Series Regression with ARIMA Noise, Missing Observations, and Outliers (TRAMO), developed by Victor Gómez and Agustin Maravall of the Bank of Spain.

What has changed?

As part of the annual seasonal adjustment update process, the CES program implemented the TRAMO procedure, available in the current X-13ARIMA-SEATS software, to select the ARIMA models used for forecasting and backcasting.
TRAMO offers a wider selection of ARIMA models than the current method.

What has not changed?

The current method of applying regARIMA models for forecasting and backcasting, outlier detection, and adjustments for calendar-related effects has not changed for monthly production. Only the automatic ARIMA selection method used during annual review has changed.
The current method of seasonal adjustment using the X11 filtering procedure has not changed. All CES series will continue to be seasonally adjusted using the current X11 filtering procedure regardless of the ARIMA model selection.

Research suggested that using the TRAMO procedure to seasonally adjust CES national estimates yielded improvements to forecasts and overall model fit, while showing little-to-negligible change across other statistical measures such as data revisions and model selection errors.

For more comprehensive and technical information about differences between the two methodologies, see Comparing Automatic Modeling Procedures of TRAMO and X-12-ARIMA, an Update (2007) (PDF).

Last Modified Date: February 3, 2017