Latest Pajamas Media Column (‘Revisionomics’) Went Up Yesterday

It’s here.

It will go up here at BizzyBlog tomorrow (link won’t work until then) after the blackout expires.

_____________________

Left on the cutting room floor: For the first 15 weeks listed, initially released unemployment claims were revised upward the following week, until the most recently reported week broke the string, as seen below:

The large majority (almost all, per Zero Hedge) of the reports issued during the past year or so contained upward revisions of the prior week’s original initial claims numbers:

Since the first week is an estimate and not a count of documents on hand, and is supposed to be an educated stab at what the right number really is, it is hard to handle the idea that revisions can occur in the same direction for so many weeks in a row without at least a tiny bit of manipulation being involved. To be clear, I’m not engaging in similar speculation about GDP growth or job additions/losses covered in the column, which is why I left unemployment claims out of it.

A flashback: I haven’t looked at an extended time period during the Reagan Era (hopefully, I will get to that in the next few days), but in the one month I did review in a previous post, I found that the government’s initial estimate of jobs added in April 1984 was 269,000 (readers who go to the link will note that the New York Times “cleverly” rounded the number down to 250,000). The final number listed at the Bureau of Labor Statistics web site is 363,000. That’s a 94,ooo-job, +35% difference from initial to final in just one month, and is consistent with what occurred during 2003-2006 as reported in the column. 2007-2010 went the other way, as final numbers badly trailed initially reported figures — and they’re not done revising those figures yet.

Go to the column to see why I believe that the upward revisions to jobs added from 2003-2006 occurred, and why 2007-2010 saw the opposite.