Gallup Had Something Wrong and I Caught Them!

Too bad I didn't catch this four weeks ago.  

Gallup posted a short article comparing their unemployment rate with the standard BLS unemployment rate.  Their seasonally unadjusted rate is higher than the BLS' unadjusted rate for April and the Gallup economist was explaining and discussing this.  (Notice in the graph below from the Gallup site that the BLS and Gallup numbers tend to rise and fall in tandem, though BLS may be higher than Gallup for any one month or vice verse.)  

(What does any of this mean for the big Friday jobs report--- or for any one of us out here in jobs land?  Though estimates are all over the place, and they are generally pretty glum, we'll just have to see.  By "pretty glum", I mean that most are suggesting that we won't add enough jobs; not that we are going to be losing jobs.)  

In any case, this was the paragraph that caught my attention:

The slow economy suggests April new jobs created were likely a lot closer to the BLS payroll increase of 115,000 than the household survey's nearly 600,000. It also suggests that the unadjusted unemployment rate is probably closer to Gallup's 8.3% than the government's 7.7%. In turn, the foundation supporting the government's decline in its seasonally adjusted unemployment rate to 8.1% seems a lot weaker than that supporting Gallup's survey results showing an increase in the adjusted unemployment rate in April.     

Except that Gallup was referring to seasonally adjusted jobs numbers when they referred to the 115,000 job increase and seasonally UNadjusted numbers when they mention the household survey's 600,000 extra workers.

So I took keyboard in fingers and explained the problem.  We'll see if my comment makes it through the moderation process.

Gallup, you are mixing apples and oranges on this one.  You are using seasonally UNadjusted numbers on the household survey.. which makes sense as the Gallup rate you are comparing it to is also seasonally unadjusted.  As you mention, the BLS shows that 600,000 more people reported themselves as employed in  seasonally unadjusted numbers in April.

But then you mention that the BLS Establishment survey's (jobs) numbers seem to support Gallup's numbers vs. the BLS numbers.  Not so.  The 115,000 jobs that you mention is the BLS estimate of jobs added in April in seasonally ADJusted numbers.  BLS actually estimates (in UNadjusted numbers) that 900,000 jobs were added in March. 

March to April is always a hiring time in unadjusted numbers.  Over the past 20 years, an average of 875,000 "real" jobs in UNadjusted numbers have been added between March and April.  Those 875,000 "real" jobs have resulted in an average of 118,000 jobs in ADJusted jobs numbers between March and April over those last 20 years.

If you don't understand this seasonal adjustments stuff, please read HERE.

Note:  As of 4 p.m. CDT 5/31/2012, my comment has still not been posted.  I doubt that it will ever see the light of day.

May Monthly Jobs Numbers

 Friday, June 1st. 

 Check back then!