The gross job gains and gross job loss statistics from the BLS Business Employment Dynamics (BED) program measure the large gross job flows that underlie the quarterly net change in employment. In the fourth quarter of 2004, employment grew by 869,000 jobs. This growth is the sum of 8.1 million gross job gains from opening and expanding establishments, and 7.2 million gross job losses from contracting and closing establishments. The new BED data have captured the attention of economists and policymakers across the country, and these data are becoming a major contributor to our understanding of employment growth and business cycles in the U.S. economy.
Following the initial release of the BED data in September 2003, the BED data series expanded in May 2004 with the release of industry statistics. The BLS then began work on tabulations by size class. The production of size-class statistics is a complex task involving several economic and statistical issues. Although it is trivial to classify a business into a size class in any given quarter, it is difficult to classify a business into a size class for a longitudinal analysis of employment growth. Several different classifications exist, and many of these possible classifications have appealing theoretical and statistical properties. Furthermore, these alternative classification methodologies result in sharply different portraits of employment growth by size class.
In this article, we discuss the alternative statistical methodologies that the BLS considered for creating size class tabulations from the Business Employment Dynamics data. Our primary focus is on four methodologies: quarterly base-sizing, annual base-sizing, mean-sizing, and dynamic-sizing. We discuss the evaluation criteria that BLS considered for choosing its official size class methodology.