by Calculated Risk on 5/06/2005 07:16:00 PM

In the previous post I cautioned about the difference between seasonally adjusted data and not seasonally adjusted data. The reason was some commentators were misusing the birth / death model that is not seasonally adjusted.


Click on graph for larger image.

This illustrates why the BLS reports seasonally adjust job growth. There is a very distinct pattern to hiring / firing. As an example, every January all of the temporary retail help is let go.

For the current month, 1.179 million new "not seaonally adjusted" jobs were created according to the payroll report. The BLS reported 274 thousand new jobs. They seasonally adjust each category, so it is very difficult to determine the actual factors.

For those that want to check these numbers, go here. Just check the boxes for both "Not seasonally adjusted" and "Seasonally Adjusted" Total nonfarm employment. Then click on Retrieve Data.

Two charts will appear. For seasonally adjusted: If you subtract the March number from April you will get the headline number of 274K. For not seasonally adjusted: If you do the same subtraction (March from April) you will get the 1.179 Million number.

CAUTION: Don't use the "Not Seasonally Adjusted" for anything except these exercises. Take a look at January - we lost 2.692 Million jobs! (but seasonally adjusted we gained 124K). We lose a ton of jobs every January and July for seasonal reasons ... this is a series that really needs seasonal adjustment.