|
Seasonal Adjustment in Current Employment Statistics (CES) | ||
|
NOTES ON SEASONAL ADJUSTMENT IN COLORADO CES DATA There have been time series breaks of all industry sectors because of the implementation of the NAICS code structure. This has affected the method of seasonal adjustment used in CES estimates. In order to develop seasonal adjustment factors for the new NAICS Supersectors, the BLS has used a reconstructed time series. There are limitations to this series that preclude the calculation of statistically significant seasonal adjustment factors for the Information and Other Services Supersectors. As a result, seasonally adjusted employment totals for the Information and Other Services Supersectors will not be available. These Supersectors account for about 180,000 in employment and contain approximately 8% of the statewide employment total. The Seasonally Adjusted Total Non-Farm employment estimate will also be affected. Even though these two Supersectors are not seasonally adjusted, their employment is still included in the Seasonally Adjusted Total Non-Farm estimate. The Seasonally Adjusted Total Non-Farm estimate is an aggregation of the Supersectors. With these two sectors not being adjusted but still in the aggregate, there may be greater volatility in the Seasonally Adjusted Total Non-Farm employment estimates. | ||
|
Seasonal Adjustment in Colorado’s Current Employment Statistics (CES) Program Why we seasonally adjust Colorado CES data. When one looks at an economic data series - such as employment - over time, many industries show seasonal ups and downs which occur annually. Seasonal changes in weather, school openings and closings, agricultural harvesting and processing, and major holiday seasons are all good examples of this. The General Building Construction Industry follows a highly seasonal employment pattern in Colorado. Colorado Statewide General Building Construction: 1991 - 2001 Not Seasonally Adjusted
Retail trade employment surges up in November and December for the Holiday season, and then drops dramatically in January and February. Since Christmas occurs every year on December 25th, this represents the ultimate example of stable seasonality. Consumers will shop for Christmas on December 25th generally without weather being the major factor. For the most part employers add their seasonal employees at about the same time each Christmas season. Another type of seasonality is called moving seasonality. That is seasonality in which month to month employment changes are not the same from year to year. Here in Colorado, our downhill winter Ski Industry is a good example of this. One year we may have an earlier winter with deep snow and the seasonal employment at Ski Resorts may jump up in November or December. A year with inadequate early snow may not see the employment spike up until January. The lack of stability in moving seasonality, while recognized, cannot be completely compensated for because it will not be the same in future periods either. Of course, the Construction Industry exhibits seasonality to a high degree in most regions, with employment falling as winter sets in and then rising again as spring and summer arrive. Since seasonal trends do generally recur annually, seasonally adjusting a data series allows a more accurate identification and assessment of actual economic changes that exceed the normal seasonal variation. Then, after seasonal adjustment, noticeable movements up or down in the data series will normally be irregular non-seasonal change, or trend cycle movement. Colorado Statewide General Building Construction: 1994 - 2001 - Seasonally Adjusted
The All Employment component of CES Data is seasonally adjusted but Hours and Earnings data are not. One important factor to remember is that some industries are not seasonal in nature. In a case where there is no identifiable seasonality, the series is typically left unadjusted. In addition, since other economic data such as Local Area Unemployment Statistics (LAUS) and Gross Domestic Product are also seasonally adjusted, CES Employment data will be more comparable with them. Overview of seasonal adjustment methodology. The statistical theory behind the X-12 ARIMA (Auto-Regressive Integrated Moving Average) seasonal adjustment process, and its implementation each month in the CES program is a fairly complex subject. A thorough discussion of that topic is beyond the scope of this article, but an overview of the process is warranted for our purposes. The Current Employment Statistics program uses public domain software called X-12 ARIMA, which was developed by the Bureau of the Census, to seasonally adjust the CES Employment series. In Colorado the data are published by the Labor Market Information Unit of the Colorado Department of Labor and Employment, in conjunction with the Bureau of Labor Statistics (BLS) of the U.S. Department of Labor. The X-12 ARIMA software normally uses a data series' last 10 years of actual history to predict the normal seasonal behavior that should occur, given no unforeseen events. Each year numerical seasonal adjustment factors are computed for the most recent five years. However, ten years of data are used in the methodology, if available. These adjustment factors are month specific, based on actual past history of that Industry and geographical region. BLS methodology utilizes prior adjustments to the not seasonally adjusted data that do not recur annually, and which would therefore distort the seasonal adjustment process. Economic analysts evaluate the historical not adjusted data to remove non-seasonal events such as a prolonged strike or shutdown of a major employer. The methodology also takes into account whether a given month has 4 or 5 workweeks in it. The Colorado CES program seasonally adjusts its data annually with a March release date. New: Seasonal Adjustment at the Two Digit SIC level. Prior to 1999 CES data were seasonally adjusted at only the Major Industrial Divisions (MID) level. But beginning in 1999 seasonal adjustments were made at the 2 Digit Industry level. In those MIDs where all 2 Digit Industries meet BLS guidelines, the 2 Digit industries will be aggregated up to the MID level. The guidelines are that there be complete coverage of all 2 Digit Industries and that each of the 2 Digit Industries passes all adjustment tests. If all of the component 2 Digit Industries do not successfully seasonally adjust, then aggregation will not occur for 2 Digit Industry components within that MID. In that event, independent seasonal adjustment will occur at the MID level only. Access detailed NAICS current & historical CES data from the BLS public site.
Colorado Department of Labor & Employment:
Jeffrey M. Wells, Executive
Director CES Home Page: http://www.coworkforce.com/lmi/CES/ceshome.htm Colorado Current Employment Statistics (CES) Estimates are produced from a survey of Colorado employers. CES Estimates are made in cooperation with the U.S. Department of Labor, Bureau of Labor Statistics. Current Employment Statistics (CES) data and any analysis are in the public domain and, with appropriate credit, may be reproduced without permission. Please reference Source: “Colorado Department of Labor and Employment, Labor Market Information”. Email: lmi@state.co.us Last Updated: 04/02/2003
|
