Friday, April 02, 2010

Diffusion Index and Temporary Help

by Bill McBride on 4/02/2010 05:08:00 PM

First, I asked on the BLS live chat today about adjusting for the temporary 2010 Census hiring. The Census hiring is reported not seasonally adjusted (NSA), and usually we don't want to mix NSA and SA numbers. The BLS experts explained they don't include the 48K Census hiring in the NSA numbers, and they add the Census numbers in after making the seasonal adjustment. So it is correct to report the headline payroll number was 162K, and 114k (ex-Census). This will really be important in May when the 2010 Census is expected to boost employment by as much at 500k! These are real payroll jobs, but all of these are temporary jobs and will be lost over the 2nd half of 2010.

Here are a couple more graphs based on data in the employment report ...

Diffusion Index

Employment Diffusion IndexThe BLS diffusion index for total private employment increased to 60.0 from 50.0 in February. This is the highest level since 2006. For manufacturing, the diffusion index is at 54.9.

Think of this as a measure of how widespread job gains are across industries. The further from 50 (above or below), the more widespread the job losses or gains reported by the BLS. From the BLS:

Figures are the percent of industries with employment increasing plus one-half of the industries with unchanged employment, where 50 percent indicates an equal balance between industries with increasing and decreasing employment.
This fits with the headline payroll report and is a positive.

Temporary Workers

From the BLS report:
Temporary help services added 40,000 jobs in March. Since September 2009, temporary help services employment has risen by 313,000.
Temporary Help This graph is a little complicated (note: explanation same as last month - repeated for clarity).

The red line is the three month average change in temporary help services (left axis). This is shifted four months into the future.

The blue line (right axis) is the three month average change in total employment (excluding temporary help services).

Unfortunately the data on temporary help services only goes back to 1990, but it does appear that temporary help leads employment by about four months. When we discussed this graph last year, temporary help suggested positive job growth in December 2009. But with revisions - the graph has been shifted a few months.

The thinking is that before companies hire permanent employees following a recession, employers will first increase the hours worked of current employees (hours worked declined in March) and also hire temporary employees. Since the number of temporary workers increased sharply recently, some people think this might be signaling the beginning of an employment recovery.

However, there has been some evidence of a shift by employers to more temporary workers, and the saying may become "We are all temporary now!", so use this increase with caution.

Also the temporary hiring for the Census should be excluded from this graph in the future - and I will exclude the Census numbers starting next month.

I doubt we will see the 300K per month net payroll jobs (ex-Census) over the next few months that this graph seems to suggest.

Earlier employment posts today:
  • March Employment Report: 162K Jobs Added, 9.7% Unemployment Rate for graphs of unemployment rate and a comparison to previous recessions.
  • Employment-Population Ratio, Part Time Workers, Unemployed over 26 Weeks
  • Duration of Unemployment
  • Replay of a Q&A with BLS is here.

  • Last 10 Posts