Updated: Changes to procedures for producing Current Employment Statistics (CES) State estimates

CES monthly employment data are widely recognized by policymakers, business analysts, and the media as a critical and sensitive gauge of the health of the U.S. and State labor markets.

CES National and State estimates, while using the same underlying survey reports and statistical methodology, are developed independently by BLS and the individual State Labor Market Information offices, respectively.  The independent development of estimates can lead to discrepancies between the employment trend shown by the National CES estimate and those obtained by summing the State CES estimates.  In the past, the discrepancy was generally greatest at labor market turning points, and resulted in the sum of the States estimates recording less employment change than was picked up by the National estimates.  This relationship was seen both when National estimates were rising sharply or falling sharply.

Since 2006, BLS and its State partners have been developing and testing alternative approaches for substantially eliminating the potential for discrepancy between the National and sum of States estimates.

Initial Changes

Beginning with the development of November 2008 final and December 2008 preliminary State CES estimates (which were released by BLS on January 27, 2009), BLS implemented improved estimate review procedures. The improved review process required States to provide fact-based justification for estimates that diverge significantly from the estimates derived solely from standard BLS methods.  An example of an acceptable fact-based justification that a State might provide is documented information about a large strike or layoff that is not captured in a CES sample report.  The changes that BLS implemented to the State estimate review resulted in a process that corresponds more closely to the National estimate review process, and strengthened the consistency and transparency of the CES State estimation process.  Further, the changes improve the collective accuracy of the CES State estimates, bringing them into closer alignment with the National estimates.

Further refinement

Beginning with the production of October 2009 final and November 2009 preliminary State CES estimates (which are to be released by BLS on December 18, 2009), BLS implemented a further refinement to the State estimation process; the selection of outliers, or atypical reports, has been automated through implementation of a standard statistical method for the statewide industry super-sector estimates. Businesses identified with atypical over-the-month employment changes account only for themselves in the estimation process and are not used to represent other firms in the population. Hence the designation of a sample unit as an outlier downweights its influence in the published series and helps to control statistical error in the estimates. The new atypical designation procedure ensures that this process is consistent across states and industry super-sector estimates. This change further improves the collective transparency of the CES State industry estimates.

 

Last Modified Date: October 21, 2009