What is the market impact from the interpretation of data?
It is an open and free debate, but there is a problem. The results follow from advanced statistical methods and processes. Even the smartest hedge fund managers, columnists, and pundits cannot draw independent conclusions. They did not take the right classes. They never did any time series modeling or survey research. Briefly put, they lack the necessary skills to evaluate most data.
The result: Nearly everyone relies upon the analysis of those accepted as experts. What choice is there?
This is a recognized principle in social science, called the two-step flow of communications.
Application to the Monthly Payroll Employment Report
When do we know that data interpretation has a market impact? One test might be the widespread citation of a conclusion.
Our "go to guy" for those on the NYSE floor is Art Cashin. In his daily comment (very valuable and available to UBS customers) he wrote as follows:
Payroll Numbers – Three pros, Dennis Gartman, John Mauldin and Greg Weldon each deconstructed the non-farm payroll data. Their conclusions were that the data was, actually, anything but bullish. Things are not always what they seem at first glance.
So the perception on the floor relies on these influential interpreters of data. What are their sources?
John Mauldin
John Mauldin does a number on the number! Even though this is an extended excerpt, readers need to check out the entire analysis.
Without that addition from the birth/death number, total private employment would have dropped by 296,000. Now, if that had been the headline number, the market would have tanked. Now, I have no doubt that the economy did create a lot of new jobs last month. But when the final revisions are in, we will see that job losses were well south of 100,000. If memory serves me correctly, the BLS had to add about 800,000 jobs that they missed during the recovery in 2003-4. (The birth/death model misses job growth during recoveries, the opposite result of the miss in slowing periods.) They did this just last year, in a major revision of the data. We will see the same type of revisions in 2010, only this time it will be downward.
And even the BLS says that the birth/death numbers have little statistical meaning. The following is from their own website (courtesy of Dennis Gartman) [emphasis obviously mine]:
“Birth/death factors are a component of the not seasonally adjusted estimate and therefore are not directly comparable to the seasonally adjusted monthly changes. Instead, the birth/death factor should be assessed in the context of its effect on the not seasonally adjusted estimate... The components are not seasonally adjusted separately because they do not have particular economic meaning in and of themselves.”
Mauldin also cites Gartman and The Liscio Report.
Barry Ritholtz
Barry Ritholtz did his regular review of the employment numbers, with, as usual, a special focus on the birth/death adjustment. The overall conclusion is that employment growth is overstated and the BLS methods are seriously flawed.
He also quotes at length from Alan Abelson, doing his monthly bearish take on the BLS report.
David Merkel
David Merkel undertakes his typical thoughtful analysis of the problem, well worth reading in the entirety. David looks at the addition of jobs from the B/D adjustment and compares these to the net job change over time. He graciously notes our dissent on some of the key issues.
Importance
The significance of these analyses is demonstrated by Art Cashin's citation. The notion that the BLS methodology is wrong has become accepted as conventional wisdom. Nearly everyone thinks that these are "phantom jobs", added by a flawed methodology, and that "turning points" in the market are missed.
It is not part of the job description for BLS employees to write on blogs or to appear on CNBC. As a result, there is no spokesman for their method. It is a one-sided debate.
All of the sources cited in this article have significant power. Their arguments have influenced active traders to believe that economic data have been artificially inflated.
Our Approach
After many months of attempting to refute specific claims about the BLS approach, we have decided to take a different tack. More to come....
