Final thoughts on the April NFP report, via Barron’s Alan Abelson. He cites Merrill Lynch economist Daid Rosenberg, who like us, feels that "strange things are afoot at the Circle-K:"
"Alas, we came across David Rosenberg’s take on April jobs, and it was pretty gloomy. Right off the bat, David, who’s Merrill Lynch’s top economist, casts doubt on the seemingly glowing numbers, which were, as it happens, much higher than the Street was forecasting (there’s always a first for everything).
"Do you believe the economy is growing at a 6% annual rate right now?" he asks rhetorically. "Because if you don’t, in my view, there is no way you can believe the strength in today’s nonfarm payroll report."April’s jobs data, he feels, "vastly overstated economic conditions last month." The gains that supposedly occurred in retail, construction and telecom just don’t square with what has been happening in those sectors. He notes that the economy is undergoing a classic inventory correction and the trend in the leading economic indicator is down.
All of which means, he contends, job additions in the months ahead closer to 130,000 than 170,000. Dementia, shmentia, we’ll just have to postpone being bullish."
I am trying not to be guilty of my own biases in selective perception and interpretation. Still Rosenberg’s take is very consistent with my own. Food for thought.
Source:
Up and Down Wall Street: Car Crazy
Alan Abelson
Barron’s Monday, May 9, 2005
http://online.barrons.com/article/SB111542268857427309.html
If you look back over the last few years you see several examples of a big one month jump in employment that just seemed to be outliers
and that is what this one looks like. In particular,
it is in sharp contrast to both the ISM manufacturing and services surveys that showed employment growth weakening last month
Interestingly, even with the surge in employment average weekly earnings growth is still only 2.6% on
a year over year basis versus its peak of 3% last November. In other words real wage growth probably is still negative.
% error, % error, significant numbers, significant numbers. The number is around 140,000,000 non – farm employees. Neither the BLS or anyone can calculate the accuracy of the number. BLS says it varies by 100,000 after the third revision!!! This number isn’t even a revision let!!! People are slicing and dicing numbers that are not significant. Come back in 3 months after all the revisions. And the number they will report will still have an unknown error. The BLS will say its the best they can do. I wouldn’t be surprised that the error after the 3rd revision is 2-4% that 280,000 to 560,000. It’s the government!
The accuracy of all the ecomonic numbers are all that way, GDP is +/- 1.1% at best. That translates into a GDP of 2 to 4 % last qtr.
Is it any surprise that a blowout number comes in April, a month where the CES birth/death model has added an average of over 200,000 jobs historically? While the birth/death model may be consistent on a year-to-year basis it sure skews the results month-to-month.
http://www.epinet.org/content.cfm/webfeatures_econindicators_jobspict_20050506
While hourly wages grew at a similar pace as in recent months—up 2.7% over the past year—weekly earnings got a 0.2 tenths of an hour bump up to 33.9 hours per week, the highest level of weekly hours since September 2002. This pushed weekly earnings up 3.3% over the past year, slightly above the latest inflation readings, which have been around 3% per year. These represent the first real weekly earnings gains for production, non-managerial workers in recent months.
The Birth/Death (B/D) model isn’t a hedonic adjustment, it’s just another time series added on to the main data. The model also cuts both ways: without the January adjustment of -280k, NFP would have been pushing half a million! I don’t know what kind of GDP growth that would imply, but whatever the number, I’m prettty sure we didn’t see it. From that perspective, the B/D model is not completely out to lunch. The 2005 pattern is also consistent with that seen in 2004, when April had a +225k B/D adjustment.
It’s also good to keep in mind that in a nation of 300M people, a 250k adjustment is one-quarter of one-third of one percent – ie, it’s just noise. Even taken over a complete 12 month cycle, the B/D model only added 600k in total to the number – still nothing but noise.
Given what the model is supposed to represent, it might prove fruitful to track the year-over-year changes in the adjustment; April was the first month for which that is possible.