Beware the Job Report

TW: The jobs numbers today actually spurred some relief as expectations were so low that they were exceeded. But the challenge with the data is that like most government reports it is very preliminary subject to significant revision. A problem with all such reports is that when the data is moving in a volatile manner the preliminary reports become more and more unreliable usually understating the move either up or down depending upon the underlying direction. Obviously these days down would be the likely direction. This is not subjective bias on the part of the government merely stats. They use formulas and estimates based on past trends as those trends themselves evolve they skew the initial estimates.

From Floyd Norris:
"Throughout much of 2008 — until the fourth quarter’s obvious economic collapse — there were a lot of economists denying a recession had begun, and asserting that the employment numbers were a prime piece of evidence.

They seemed to have a point. Job losses from February through August averaged 83,000 a month. That was not good, but it was not nearly the number you would expect as a recession was getting going.

Today the numbers were revised, and the recession appeared. The average over that span is now 147,000 jobs lost per month.

Why the change? The figures for March 2008 and before reflect the use of unemployment tax data. After that there are revisions stemming from the benchmark established by that data, as well as changes in seasonal adjustments and changes in the birth-death model the government uses to estimate jobs created by new companies.

Those economists who believed the numbers were fooled. Those who reasoned that a recession was obvious, given the troubles of the home and auto industries, were vindicated.

And now? Over the five months from September through January, the number of private-sector jobs is down 2.1 percent — the highest for any comparable period since 1975.

Among the unemployed, an unusually high percentage got that way by losing their jobs — not because they quit them, or were laid off with the possibility of being recalled, or by not having found work since joining the labor force.

Finally, the January figure may be worse, in reality, than the loss of 598,000 reported by the Labor Department. That is a seasonally adjusted figure. Before adjustment the figure was a loss of 3.6 million jobs.

That big adjustment reflects the fact that temporary holiday season workers see their jobs end. But this season there were fewer such workers hired, and that means that more of those who lost their jobs were probably permanent workers."
http://norris.blogs.nytimes.com/2009/02/06/lost-a-lot-of-jobs-that-never-were/