September 1, 2004

The jobs never existed...

An economist friend wrote this, and I'm putting it in as a guest-post...

A note on Payrolls

The BLS has just finished a study (link to PDF) of the differences between the payroll survey (CES) of employment and the household survey (CPS). Chart 1 below shows the results in brief. The blue (top) line is the household survey with the population controls smoothed so that their impact is spread throughout the year instead of hitting in one shot in January. This answers a major complaint by Alan Greenspan.

The yellow line is the blue line adjusted for so that jobs are defined in the household survey to be the same as in the payroll survey (this has to do with excluding agricultural workers, the self-employed, those holding two jobs, etc.). The red line is just the payroll survey as it currently exists. Note that the yellow line now agrees with the red line prior to 1998 and after 2002. The gap yet to be explained is the bulge in payroll employment over adjusted household in the 1998-2002 period.

employment chart one

My contention is that this bulge in payroll jobs probably never occurred in the first place and hence those jobs were not "lost" in any meaningful sense. As the BLS notes on pages 18-22 in attached study the reference periods in the two surveys are substantially different. The household survey is limited a one week each month whereas the payroll survey uses full pay periods for its reference and these are typically two weeks or a month. The point made by the BLS is that a worker who quit one job and took another within the same pay period could be counted twice. The BLS never pursues this reasoning but notes that the double counting problem in the payroll survey is potentially significant most likely to occur during economic expansions.

I don't think this is quite right. It may take more than a mere expansion. It may take a "red hot" labor market. As shown in the chart below bulge in payrolls (CES) over household (CPS) employment is associated with unemployment rates below 5 percent.

Employment chart two

This is bad news for the Bush administration. No one will ever believe that those payroll jobs did not exist. The Democrats have won that argument and Bush will be the "first president since Herbert Hoover to jobs decline during his administration." Furthermore, the labor market is unlikely to get "red hot" anytime soon. This means that reported payroll expansion will get no boost from a renewed bulge in payrolls over household employment growth. Instead, payrolls will have to proceed at the same pace as adjusted household employment. In other words, the red and yellow lines in Chart 1 will continue to stay together.

I hope I am wrong.

* My guess is that the Democrats have pushed so many "worst economy since Herbert Hoover" lies that no one will believe them even if they are telling the truth for once. And most Americans don't even know who Hoover was...--JW

Posted by John Weidner at September 1, 2004 7:45 PM
Weblog by John Weidner