Assessing the Usefulness of Census Bureau Multi‐Establishment Data to Facilitate Linking Firms with Establishments in BLS Microdata

Elizabeth Weber Handwerker and Lowell G. Mason


For researchers interested in linking firm-level datasets with establishment-level Bureau of Labor Statistics (BLS) microdata, the fundamental challenge is to find all the EINs that these firms use in their reports to the Unemployment Insurance programs of the 50 states. In this paper, we assess the extent to which data compiled by the Census Bureau for multiple-establishment firms can aid this task. We find that for a convenience sample of firms, the Census Bureau data contain most of the Employer Identification Numbers (EINs)—and add additional EINs—to those found through labor-intensive searches of the BLS Longitudinal Database (LDB). However, not every apparently valid EIN for these firms in the BLS LDB appears in the Census Bureau’s lists of EINs for these firms. Furthermore, some of the EINs that appear in the Census Bureau’s lists do not appear to be valid EINs for these firms in the BLS LDB in the relevant year or quarter. We conclude that using Census Bureau data on multi-establishment firms can reduce (but not replace) the labor-intensive work of finding all the establishments for particular firms in BLS microdata.