DavidRosenbergIt’s getting hard to keep track of all the different ways people are attacking Friday’s employment number.

You’ve got TrimTabs which keeps banging the drum about the birth-death model.

You’ve got folks talking about how it was all in temp work, or the public.

Then there are those who point out all of the people who have left the workforce.

David Rosenberg chimes in with another: that it just can’t be right, because it doesn’t jibe with the contraction in the services sector

——

A 1-IN-35 EVENT 

It’s remarkable nobody talks about this.  The big surprise in the payroll data was
the service sector component; it rose 58k.  But we know from the ADP report
that service sector employment fell 81k, which was fractionally worse than the
79k decline in October.  Such a discrepancy has occurred less than 3% of the
time in the past, and each time, the following month after the big gap, there was
a convergence … with headline nonfarm payrolls swinging 100k lower on
average, which would imply a 111k decline when December’s figure comes out.
Also take note that the +58k print in the service sector payroll was completely at
odds with the 41.6 reading in the ISM non-manufacturing employment index in
November — a figure that in the past was consistent with a -192k tally in service
sector payrolls and never before aligned with a positive number.  Go back to the
2001 recession, and the worst ISM non-manufacturing jobs subindex was 43.9
(right after 9/11) and here we published a figure that was more than two points
shy of that! 

So as we wonder how the headline number could only be -11k on Friday, there
were some very lumpy increases in some very non-cyclical segments of the
economy: 
• Administration/waste management +87k
• Health/education +40k
• Government +7k

The rest of the economy shed 145 jobs and the declines were spread across
nearly 60% of the industrial base from retail, to transports, to manufacturing, to
construction.  For some reason, we didn’t see this dichotomy mentioned
anywhere in the weekend press.