The most popular measure of employment growth is the monthly report on nonfarm payrolls. However, the most comprehensive data source on the number of people working is the quarterly census of employment and wages (QCEW). In an attempt to keep the more timely series (nonfarm payrolls) in line with the better data source (QCEW) on a continuous basis, the Labor Department includes a birth/death adjustment in its monthly report.

The birth/death adjustment is intended to account for the employment impacts of both new businesses starting up and existing businesses shutting down. The QCEW points to strong employment growth among startups that the monthly employment report methodology can not detect.

Without that monthly adjustment, the annual benchmark revision to nonfarm payrolls would be upward by a much larger amount. BLS has always benchmarked payrolls to the QCEW, but that annual benchmark revision has been much smaller since the addition of the monthly birth/death adjustment.

The adjustment is reported on a not-seasonally-adjusted basis only, but has a well defined seasonal pattern, generally adding the most to payrolls in April, and subtracting the most in January.

In April 2008, the birth/death adjustment was +267,000. That is a not-seasonally-adjusted number, and it is incorrect to say that the April headline -20,000 seasonally adjusted decline “would have been” -287,000 (-20,000 minus 267,000) excluding the birth/death adjustment. Subtracting 267,000 from the not-seasonally-adjusted total and then applying the seasonal factor is also incorrect, most notably because the overall seasonal factor is influenced by the seasonal pattern of the birth/death adjustment.

Trying to take out the birth/death adjustment is a bad idea, because payrolls will eventually be benchmarked to the series that includes the birth/death concept (the QCEW). Nevertheless, those insisting on trying would find more success in seasonally adjusting the birth/death impact itself, and then subtracting that from the official BLS headline. However, even that method is technically incorrect.

One criticism of the birth/death adjustment is that it does not reflect changes in the business cycle. However, making such an adjustment would be very difficult. Specifically, it is unclear whether new small business start up because conditions are good (demand is stong) or because economic conditions are weak (laid off from a corporate job during a recession so you start your own company).

Being aware of the birth/death seasonal tendencies can be of use in trying to forecast seasonally adjusted payrolls, particularly in January and February, and to a lesser extent in April, October, and December. Those are the months when the birth/death adjustment is the largest and/or most closely correlated with the m/m change in seasonally adjusted nonfarm payrolls.

To further illustrate my point, I've compiled a number of charts and tables in this PDF document. Enjoy!

Mark Serlin is the Chief Economist at Economic Strategies Inc. Prior to joining Economic Strategies, Mr. Serlin authored the widely distributed SERLINON economic indicator analysis service for Bridge Information Services. Mr. Serlin has also worked in the research departments of Bear Stearns, Ried Thunberg, and Manufacturers Hanover.