Article

June 2013

Linking firms with establishments in BLS microdata

(continued...)

The following tabulation, for the fourth quarter of 2009, shows the percentage of establishments and total employment matched when just the firm name was used, for each of the sample firms involved in the four case studies:6

 EstablishmentsEmployment  
Sample firm studiedPercentage correctly matchedPercentage incorrectly matchedPercentage correctly matchedPercentage incorrectly matched  
1. listed in Dow Jones80.5170.095.838.8  
2. listed in S&P 5004.8.02.4.0  
3. listed in Russell 200099.0.097.6.0  
4. privately owned.8.052.8.0  


In only two of the four case studies—the Dow Jones and Russell 2000 listings—was a large percentage of establishments and employment correctly identified. However, the Dow Jones listing also halarge percentage of incorrectly identified establishments and total employment. Reviewing the resulting matches in this case study revealed that a number of establishments were acquired by the firm after the fourth quarter of 2009 and that these establishments were incorrectly matched. (The QCEW name and address files are continuously updated, and versions corresponding to past dates are not available.) The remaining two case studies identified much smaller percentages of establishments and employment. A review of the establishments that were not identified indicated that the unmatched establishments’ names listed in the QCEW were those of the associated firms’ subsidiaries and not the firms themselves.

As the following tabulation shows, better results can be obtained by using the names of firms’ subsidiaries as well as the addresses of the firms’ establishments:

 EstablishmentsEmployment  
Sample firm studiedPercentage correctly matchedPercentage incorrectly matchedPercentage correctly matchedPercentage incorrectly matched  
1. listed in Dow Jones99.70.3100.00.0  
2. listed in S&P 500100.0.0100.0.0  
3. listed in Russell 2000100.0.0100.0.0  
4. privately owned99.8.099.9.0  


These names and addresses are culled from the firms’ websites and from their Form 10-K filings (for each firm that is publicly listed), but this manual process is time consuming. To aid the process, the search may be expanded to include establishments in the QCEW with names that do not exactly match those of the associated firms and their subsidiaries, but rather match only parts of the names. This approach increases the number of possible matches, both correct and incorrect. Thus, the matches are reviewed manually and compared against addresses found on the firm’s Form 10-K listings and websites, and then the incorrect matches are removed. As the following tabulation shows, this additional manual step adds more time to the matching process (but it is much better than simple searches by name or single EIN):

 Minutes spent  
 Sample firm studiedReviewing firms' Form 10-K listing Reviewing firms' websitesSearching the QCEW by subsidiaries' names and addresses Total time taken (minutes)  
1. listed in Dow Jones10212758  
2. listed in S&P 50026392186  
3. listed in Russell 200013101841  
4. privately owned0442569  
   Average minutes spent12292364  


Still, these efforts do not find every correct match or remove every incorrect match. Fortunately, for the four case studies presented, there is additional information about the true matches, and that information can be used to evaluate the matching efforts. (Note, however, that, for most firms, such information is not available.)

In each of the four case studies, all of the firm’s establishments were found. Not every matching attempt, however, is successful. For example, Handwerker, Kim, and Mason attempted to find all of the establishments in the QCEW for the largest 500 multinational manufacturers in the United States.7 Using every resource currently available at the Bureau, they were able to find establishments that matched employment within 20 percent of total employment reported to BEA for only 454 of the firms examined.

RESEARCHERS SOMETIMES NEED TO FIND all of the establishments associated with a single employer in BLS data. With most employers, this task for the researcher is straightforward. As shown in tables 1 through 4 and by Elvery and colleagues,8 the vast majority of employers are small, with EINs in only one state and with a single UI account and a single establishment. However, the large companies that frequently are of interest to researchers often use multiple EINs in reporting their employment to the UI system (the source of QCEW data), and there is no straightforward way to find all of the EINs and establishments associated with a particular firm.

Notes

6 To verify in all four cases that the establishments that were found through the matching process used were the correct establishments, the names of establishments were examined and the EINs that were found by matching against the (highly incomplete) BLS listings of EINs for employers with multiple EINs were checked.

7 Handwerker, Kim, and Mason, “Domestic employment.”

8 Elvery, Foster, Krizan, and Talan, “Preliminary Micro Data Results.”

prev page5next page

View full article
About the Author

Elizabeth Weber Handwerker
handwerker.elizabeth@bls.gov

Elizabeth Weber Handwerker is a research economist in the Office of Employment and Unemployment Statistics, Bureau of Labor Statistics.

Lowell G. Mason
mason.lowell@bls.gov.

Lowell G. Mason is an economist in the Office of Employment and Unemployment Statistics, Bureau of Labor Statistics.