Seasonal Adjustments: Fun With Numbers, Redux

There are three types of people in the world: Those who understand Statistics, and everyone else.


Why do so many fund managers underperform? Our host and overlord frequently references cognitive issues, but I am beginning to suspect the problem is simpler: An glaring lack of fundamental math skills.

Those of us who stress data points in our analysis are bumping up against a legion of people — including strategists, commentators and economists — who seem unable to perform simple statistical analysis. Indeed, the willingness to misuse and abuse numbers makes me wonder whether they are polticially motivated. It goes far beyond the “Lies, damned lies, and statistics” to a point that is, to be blunt, embarrassing.

Last month, we watched Zero Hedge and Rick Santelli demonstrate their seeming innumeracy. The Census adjustment is painfully easy to understand, unless you are a) rooting for the end of the world; or 2) a partisan hack.  The combination of armageddon/hackdom sent the “1.2-million-people-dropped-out-of-the-labor-force” story viral.  Of course, that was embarrassingly wrong, but made its way thru the media, including to a White House press briefing. (BTW, have the sources of that misinformation ever posted a correction or retraction? Or are they standing pat with “Duh“?)

The latest innumerate silliness comes to us via an abuse of seasonal adjustment. This time, it is “on a Not Seasonally Adjusted basis, we’ve lost 1.8 million jobs since December!

This story has been floated around in many guises, but I first saw it via the estimable Art Cashin, who channeled The King Report, quoted here at Business Insider (emphasis in original):

Even more disturbing to us was word from a friend who subscribes to The King Report.  He says the newsletter says that without the filter of seasonal adjustment, so far in 2012 nearly 1.8 million jobs have been LOST.  Wow!  We’ll try to check that out.

I submit that this is, in fact, a perfect example of why we have seasonal adjustments in the first place.  People get hired going into the holiday season and are subsequently laid off or quit after the holidays.

By how much? On an Non Seasonally Adjusted (NSA) basis, from August through November 2011, the economy added…wait for it…1.89 million jobs. Retail and Shipping account for the lion’s share of these, with Bar & Restaurants also significant. After the holidays, 95% of these disappear in January and February.  That is why they call these “seasonal jobs.”

But more to the point, let’s take a look at the stunning NSA loss of 1.8 million jobs since December in the context of the previous decade, shall we?  Here’s the February print minus the previous December’s print for the past 10 years:

2003 -2273
2004 -2050
2005 -1880
2006 -1728
2007 -2105
2008 -2519
2009 -3939
2010 -2432
2011 -2037
2012 -1801

Which one of these is not like the others?  To my eyes, it’s the 3.9  million that printed for the similar period in 2009.  But that might tell a different story.  I suspect, perhaps, that this is not about innumeracy, but rather about folks who should know better yet choose to use numbers to mislead and deceive.  It is, frankly, unconscionable.

Folks, this is Statistics 101. There are plenty of things to criticize about various BLS models — this isn’t one of them . . .

What's been said:

Discussions found on the web:

Posted Under