sge

New Quarterly Data on Business Employment Dynamics by Size of Firm from BLS

January 2006 Luncheon

Speaker:  Jack Galvin
Affiliation:  Associate Commissioner for the Office of Employment and Unemployment Statistics, Bureau of Labor Statistics

Rapporteur: Gregory Niemesh

John Galvin addressed the following questions: What are the business employment dynamics (BED) data? What is the appropriate unit of analysis for the new BED data by employer size class: firm or establishment? How should businesses be classified into size classes? What does the BED data show about the underlying dynamics of net job change?

The Bureau of Labor Statistics (BLS) began publishing BED data in September 2003; the data series are quarterly and run from the third quarter in 1992 to the first quarter 2005. The BED data includes gross job gains, gross job losses, and employment and establishment counts, levels and rates. It becomes available about seven months after the end of the quarter. In addition to the national time series, the Bureau of Labor Statistics (BLS) started publishing BED data categorized by major sector industry in May 2004, and by firm size class in December 2005.

The original BED data track changes in employment at the establishment level and provide a picture of the dynamics underlying aggregate net employment growth statistics. The change in the number of jobs over time is the net result of gross job gains and losses that occur at all firms in the economy. Gross job gains are the sum of all jobs added at firms that have expanded their payroll, plus the payroll of new firms that have opened. Gross job losses are the sum of all jobs lost at either firms that have closed or firms that contracted their payroll.

Looking at the underlying gross flows of a net change in employment adds important information that can be used to assess the business cycle, the level of labor market volatility, and the effect of establishment employment changes on aggregate employment. For example, over the period Q3 1992 to Q1 2005, the new BED data by employer size class show that firms with less than 500 employees accounted for 65 percent of net job growth, although their share of total employment in March 2005 was 55.8 percent.

Firm size class data can also be used to analyze patterns of employment changes across firms of varying sizes during the business cycle. The federal government does not officially define a small business. However, the Small Business Administration classifies firms with less than 500 employees as small, and the BLS uses this definition in its analysis.

Job loss associated with the 2001 recession was driven mainly by large firms: Firms with 500 or more employees accounted for 60 percent of job losses and only 35 percent of net gains over the new job decline that took place between March 2001 and June 2003. Since the beginning of the recovery, firms with less than 500 employees have accounted for 64.1 percent of new jobs, while firms with greater than 500 employees have accounted for 35.9 percent.

The BLS uses the firm, defined as a legal business, instead of the establishment as the unit of analysis for the new BED data. An establishment is defined as an economic unit that produces goods or services, usually at a single physical location, and engages in one or primarily one activity. A firm may consist of several establishments. While establishment data make use of applicable single industry codes, firm data is most consistent with similar data from other national and international sources.

Firms are grouped into size classes from Office of Management and Budget based on the level of employment. The BLS uses dynamic sizing of firms to allocate gross job gains and losses. Dynamic sizing initially classifies the firm based on the previous quarter’s employment. If gross job losses or gains cause a firm to cross a size class threshold, then the remaining gross job losses or gains are allocated to the new size class. An important reason for choosing dynamic sizing, rather than the other three sizing alternatives evaluated, is its symmetric treatment of seasonal gains and losses.

The BED data is based on data from the BLS’s own Quarterly Census of Employment and Wages program, and does not increase the burden on respondents. The data cover 97 percent of private wage and salary jobs and come from quarterly unemployment insurance microdata, to which BLS adds industry and geographical coding. Longitudinal linking of microdata across quarters allows the BLS to follow the gross job losses and gains posted by individual firms. For more detailed information, Galvin referred researchers to the following websites: