Skip Navigation Links    
DOL Seal - Link to DOL Home Page
Photos representing the workforce - Digital Imagery© copyright 2001 PhotoDisc, Inc.
 www.bls.gov  Search | A-Z Index

Important Information


Benchmarking and the Annual Revision Process

Model-Based Revisions: Benchmark revisions became available on February 27, 2004, for States, the District of Columbia, the Los Angeles-Long Beach metropolitan area, and New York City. Estimates for these areas have been updated with new model inputs, model reestimation, benchmarking to Current Population Survey (CPS) annual averages, and the application of new seasonal adjustment factors. Estimates for Puerto Rico also were re-seasonally adjusted. Census region and division data, which are the sum of statewide estimates, also were revised. Seasonally adjusted data for these areas have been revised on a monthly basis for 1999 forward, while not-seasonally adjusted monthly and annual average data were revised back to 2000.

Substate Area Revisions: Benchmark revisions for substate estimates are released several months after the model-based estimates due to the amount of time it takes to review data for over 7,000 areas. The purpose of benchmarking is to incorporate any changes in the input data, such as revisions to establishment-based employment estimates and unemployment insurance claims counts, as well as updated historical relationships. The revised estimates are then readjusted to sum to the latest statewide estimates of employment and unemployment. For 2004, this procedure is being performed on four years of data, rather than three years, one year at a time, working backwards.

1. Revisions to 2003 all area data were released on May 5

2. Revisions to 2002 all area data were released on June 9

3. Revisions to 2001 all area data were released on June 30

4. Revisions to 2000 all area data were released on July 20

*NOTE: All annual revisions will be annotated in the data base with a "b" for benchmarked.

Revision of Historical Labor Force Data

All monthly and annual average labor force estimates for 1990-99 for regions, divisions, States, the District of Columbia, the Los Angeles-Long Beach metropolitan area, and New York City were revised on February 11. The basis of these revisions was updated CPS population controls that better link the results from the 1990 Census (adjusted for the estimated undercount) with those from the 2000 Census (as enumerated). In addition, the revised 1990-99 labor force estimates reflect subsequent reestimation of models, benchmarking to the new CPS levels, and updated seasonal adjustment factors. The impact of the revisions on labor force estimates vary. For example, States in which the 2000 Census enumeration was close to the projected 1990 Census-based population estimate had smaller annual average labor force revisions than States with large differences between 1990-based and 2000-based population figures. Also, revisions were larger for time periods later in the decade than for periods earlier in the decade. Finally, changes to monthly estimates were generally greater than the revisions to annual averages.

Revised data for all substate series--consistent with the new not seasonally-adjusted statewide controls--also were released on Februay 11. All revised data for 1990-99 are annotated with the letter "c" to indicate that they have been adjusted to incorporate revised intercensal population controls for the 1990s. Data for 1990-99 for Puerto Rico reflecting 2000 Census population controls are expected to be revised in April.

Corrections to the Database

All areas in Oregon were corrected for 1990-1991 monthly and annual average data on October 22.

Monthly and annual average data were corrected for North Chicago and Palatine Village, Illinois for 1990-1996 on October 22.

Data were corrected on September 28, for all substate areas in Wisconsin for July and annual average of 2003.

Monthly data for July 2003 through February 2004, as well as annual averages for 2003, for all areas in Oregon were corrected on August 20. Klamath County was the area most affected. The same correction was made on August 4 for April through June 2004 data.

Corrections were made on August 20 for April through December 2003 to monthly data for all areas in Iowa. The Iowa City, metropolitan area was most affected. Interstate areas shared with Illinois and Nebraska were also corrected.

On August 20, monthly and annual average data for 2003 were corrected for three cities in Lehigh County, PA--Allentown, Bethlehem City (part), and Whitehall.

Benchmarked 2001 data for Utah were loaded to the Labstat data base on July 7. Also, corrections were made to 2001 benchmarked data for three areas shared between Virginia and West Virginia--Frederick-Winchester-Hampshire, VA-WV LMA; Montgomery-Greenbrier-Pulaski, VA-WV LMA; and Rockingham-Harrisonburg-Page, VA-WV LMA.

On June 18, corrections were made for Allentown City, Whitehall Township, and the part of Bethlehem City located in Lehigh County for January-April 2004.

Data were corrected on May 21, for all areas in Mississippi and the Memphis, TN-AR-MS, MSA for January 2004. The correction primarily impacted Itawamba County and the Lee-Prentiss-Pontotoc, MS LMA.

New Geographic Definitions

In 2001, Broomfield city, CO, was first designated a county. Prior to the change, the city had been divided among three counties--Adams, Boulder, and Jefferson. Boulder County was coterminous with the Boulder-Longmont, CO PMSA under the old definition, and part of Broomfield that was in Boulder county was within this metropolitan area. With the creation of Broomfield County, BLS decided to move Broomfield County to the Denver, CO PMSA because data from Census 2000 indicated that it would become part of the Denver metropolitan area when the new area definitions were introduced. Broomfield was assigned to the Denver, CO PMSA for LAUS estimation purposes in 2003. As a result, the Boulder-Longmont, CO and Denver, CO PMSAs no longer match the official OMB geographic definitions.

New Population Data

Approximately every 10 years, the LAUS program incorporates the most recent decennial census estimates into its estimation methodology. As part of this updating process, Current Population Survey (CPS) estimates based on the latest decennial census population estimates were incorporated into the LAUS signal-plus-noise model estimates. Unlike the usual revision process for intercensal population controls, where the monthly CPS estimates are ratio-adjusted to reflect independent population estimates, revisions based on the decennial census data involve re-estimating the CPS labor force estimates using the 2000 decennial census population values in the estimation process. While only levels are affected by the usual annual revision for intercensal population controls, the incorporation of the decennial census data means that both levels and ratios, such as employment-population ratios and unemployment rates, can be affected.

Another part of the updating process affects the weights applied to the CPS micro-data used to develop monthly CPS estimates (the second-stage weighting procedure) in the intercensal period. In addition to updating the estimates for sex, age, race, and ethnicity with 2000-based data as part of the incorporation of decennial census results, the CPS estimates will also reflect new multi-race population categories and a new second-stage weighting procedure that will improve the quality of the State CPS estimates.

With the implementation of the 2000 decennial census in the CPS, there will be two “breaks” in the CPS series: one, between December, 1999 and January, 2000, due to the incorporation of 2000 decennial census population values and the use of modified CPS weights in reconstructing the series, and the second, between December, 2002 and January, 2003, when the new CPS second-stage weighting procedures will be used.

From January 2000 to December 2002, the monthly CPS series will be based on (1) the new population controls from the decennial census, and (2) modified weights that estimate the new multi-race classifications. The monthly CPS series has been re-estimated using the 2000-based population controls and weights that adjusted the 1990-based CPS data associated with the old single race definitions to the new 2000-based multi-race population values.

Revisions observed in the re-estimated CPS data are primarily due to two types of differences between the 1990-based and 2000-based estimates. The first type is the “population” difference – the difference between the extrapolated 1990-based CPS population estimates and the CPS population estimates based on the new 2000 decennial census counts. This type of difference impacts levels and rates of growth, and is most noticeable in population and employment estimates. States and areas where these population differences were greatest will see the largest change in their estimates. The second type of difference is the “composition” difference – the difference between the extrapolated 1990-based demographic composition of a State, adjusted annually to independent population estimates, and the demographic composition observed in the 2000 decennial census. This type of difference has the potential of impacting rates and ratios. For example, areas where the composition of the population showed a disproportional increase in demographic groups with high unemployment will see their unemployment rate increase.

Area Titles

A title change was implemented in early 2003 with the labeling of the St. Louis, Missouri-Illinois MSA switching from an LMA in news release tables and the Labor Market Area Directory to an MSA, with references to the unique inclusion of a partial city and the difference from the official OMB designation cited in a footnote.

A new "city" code and label were created in the database for the city of Washington, D.C. Previously this area was treated as only a state and a county (called District of Columbia) in the database, but a new code also provides a city designation.

A new title was created for Lexington-Fayette city, KY. The previous titles were Lexington/Fayette city and Fayette County. The city is coterminous with the county, but the new city title uses the corporate name. The county title remains Fayette County.

The Local Area Unemployment Statistics program reviewed its file of area titles and made a number of changes in early 2003. All of the changes are listed in the file at http://www.bls.gov/lau/latitles.pdf. A small number of areas identified in the file are listed twice. These are counties that meet 3 criteria: (1) each is coterminous with a city; (2) each is identified by a single LAUS code (usually as a county); and (3) each has an alias record with the same title but a different FIPS code (usually for the city record). Juneau Borough/city, AK, is an example of this situation.

Please check this page for further updates.

 

Last Modified Date: October 22, 2004

 

Back to Top Back to Top www.dol.gov
U.S. Bureau of Labor Statistics
Local Area Unemployment Statistics Information and Analysis
Suite 4675
2 Massachusetts Avenue, NE
Washington, DC 20212-0001

URL: http://www.bls.gov/LAU
Phone: (202) 691-6392
Fax: (202) 691-6459
LAUS data questions: lausinfo@bls.gov
Technical (web) questions: webmaster@bls.gov
Other comments: feedback@bls.gov