Guest Post: How You Too Can Be As Good (Or Bad) As Goldman In Predicting Non-Farm Payrolls

Submitted by James Freeze
It doesn’t take a fancy European name like Jan or a PhD to take the guesswork out of the
monthly Non-Farm Payroll data.
A simple 3-factor model can predict payrolls within spittin’ distance of the actual post-revision result. The monthly average of the 4-week moving average of initial unemployment claims (sa), the monthly average of the insured unemployment rate (sa), and the %Y/Y change in employee withholding taxes.
Perhaps the recent trend of improvement in payrolls is stimulus related and will reverse course, perhaps not, but the relevant data is provided each week for the world to make that
determination.
And with the average range of bond contracts and spoos expanding by 34% and 30%,
respectively, with each release, it can be very profitable to get it right… that’s all I got to say about that.
Relevant data set support:
- Login or register to post comments
- Printer-friendly version
- Send to friend



on Fri, 12/04/2009 - 14:12
#152735
Goldman knows the numbers because it gets advanced information from the Gov't (I think).
on Fri, 12/04/2009 - 14:17
#152748
So its not Dr. Freeze just Mr. Freeze LOL.
Great post thanks.
on Fri, 12/04/2009 - 14:36
#152779
Steve Liesman CNBS " I've been looking this over and these numbers are real"
Whoa there partner, real? You mean the crap you normally spout is not?
on Fri, 12/04/2009 - 14:39
#152783
Gimme an R - Gimme an E - Gimme an A.....
on Fri, 12/04/2009 - 15:02
#152830
Gimme an C - Gimme an H (space) Gimme an A - Gimme an R - Gimme an O - Gimme an U - Gimme an N - Gimme an D... What's that spell - REACH AROUND! Yeah, what I give Carl Quintanilla!
on Fri, 12/04/2009 - 14:42
#152790
Monkeys can recognize photographs of other monkeys they know, proving that they can both detect differences in faces and figure out if they've seen them before, researchers report.
on Fri, 12/04/2009 - 15:30
#152813
It's the Fam-French unemployment model! Genius! However, the r-sqaure (.75) seems a little low. The model only seems somewhat useful. Still, the Obama-bernanke models of employment and economic recovery seem like complete bullshit to me.
on Fri, 12/04/2009 - 15:28
#152907
Where is the employee withholding tax data coming from?
on Fri, 12/04/2009 - 15:53
#152974
What model of employmetn and economic recovery?there is only one model available now,and that is the recovery of the stock market. Speaking about fancy names,just look at what a fancy name did(Von Braun) under whose supervision Saturn 5 was designed. And look at the current fancy name(Vinnair,or Van Praag)and tell me what they are designing. Using Rocket science to design debt wrapped in many layers of debt,that if you dare open it in a bath tub,you will find bubbles coming to the surface because there was only air inside. Is there any wonder that there is no actual growth in the economy?. Forget about GDP numbers.All you have to do and look at P/E multiples,and they will tell you a different story. Any company of a small hint of growth and good balance sheet,is trading by 50-100x earnings.So what does that tell you?nothing else to buy.
on Fri, 12/04/2009 - 16:01
#152988
I don't believe anything from GoniffSachs.
Here is an interactive unemployment map. Very interesting, not sure how accurate but you will get the idea. Just click play.
http://cohort11.americanobserver.net/latoyaegwuekwe/multimediafinal.html
on Fri, 12/04/2009 - 18:56
#153349
Yes, but the insured unemployment rate you include in the model is misleading since it is censored at 6 months unemployment. The true insured unemployment rate is 7.44% once one includes Currently Ins. Unemployment (SA)= 5,465,000 + Extended Benefits= 597,688 (compared to 22,125 one year prior) and EUC 2008 = 3,859,553 (compared to 777,393 one year prior).
For comparison Extended Benefits were only 22,125 one year prior) and EUC was 777,393 one year prior. In other words, these elements would have minimal/negligible impact a year ago.
It would be interesting to recalibrate the model with the complete insured unemployment rate.