
A 1.6 million-establishment gap between BLS and Census counts stems from elderly-care providers and multi-unit firm reporting differences, a new Fed paper finds.
The Bureau of Labor Statistics and the Census Bureau have been reporting sharply different counts of U.S. employer establishments since the 1990s. By 2023, the gap reached roughly 1.6 million, with the BLS showing much faster growth. A new Federal Reserve working paper, using linked microdata from both agencies, identifies the two main drivers.
A large and growing number of employers providing services to the elderly and persons with disabilities fall within the BLS survey frame but not the Census Bureau's. Many of these providers are small, home-based, or operate under licensing structures the Census frame does not capture. The BLS, which draws on state unemployment insurance records, picks them up. As the population aged and demand for these services grew, the BLS establishment count pulled away from the Census count.
The second factor is structural. Many multi-unit firms appear with substantially more establishments in the BLS data than in the Census data. When a company operates multiple locations – retail chains, restaurant groups, logistics hubs – the two agencies sometimes record different numbers of establishments for the same firm. The paper shows this is not a one-time reconciliation issue. It is persistent and growing.
The authors – Dan Cao, Henry Hyatt, Toshihiko Mukoyama, and Erick Sager – did not find a single data error or methodological flaw. The divergence reflects real differences in which businesses are counted and how multi-location firms report their footprints.
The discrepancy substantially affects the measured establishment size distribution. A researcher using Census data will see a smaller, more concentrated business sector than one using BLS data. The choice of source changes the baseline for any quantitative policy analysis of business dynamism, entry rates, or concentration trends. A policymaker relying on Census data to assess whether competition is declining would get a different answer than one using BLS data.
The paper does not declare one frame correct. It documents the divergence and its consequences. For anyone tracking the U.S. business sector, the gap means that any single-agency series should be read with a footnote. The BLS and Census are not measuring the same thing.
Prepared with AlphaScala editorial tooling from the source reporting linked above. Indexable analysis may include a cited Alpha Score value. Publishing checks screen each story before release. Educational coverage, not personalized advice.