Business Register Quality Metrics

Sheryl Konigsberg, David M. Talan, and Richard L. Clayton


One of the most important concerns involving large datasets is how the integrity of the data is maintained. In 1999, the U.S. Bureau of Labor Statistics (BLS) launched the Longitudinal Database (LDB)—-the Bureau's establishment-based business register. The LDB is a relational database of 8.4 million business establishments linked longitudinally and based on the microdata submitted quarterly by States from Unemployment Insurance (UI) tax files. Data elements on these files include information on monthly employment, quarterly wages, business name and addresses, industry classification, geocodes, and other administrative data. Every business establishment contains a unique identifier that allows for tracking of individual establishments at the microlevel across quarters for the United States. The LDB has three critical functions: producing longitudinal Business Employment Dynamics (BED) statistics, serving as a sampling frame for establishment-based surveys for BLS, and serving as an important resource for labor market research.