Democrats Confront Reality

Various pundits and commentators in the past week have attempted to legitimize the incredible claim that the economy gained 873,000 jobs last month, given the convenient boost it provides to the President's election prospects.   In the same breath, they are quick to attack anybody who questions the validity of the numbers.  In his article "In conservatives' war on reality, the facts are optional", Leonard Pitts not only elaborates on these themes, but attempts to discredit the very notion that any claim of the Obama Administration is incredible.   Exhibit A: his thesis that Republican skepticism about the numbers constitutes a partisan "war on reality".
 

Last week, the Bureau of Labor Statistics reported that the unemployment rate as reflected in their Household Survey dropped from 8.1 to 7.8 percent, an improvement of 873,000 jobs. For the same time period, the BLS also reported a net job gain of only 114,000 jobs based on their Employer Payroll Survey. It's instructive that in Pitts' world, it is paranoid to note the gross discrepancy in these two numbers, much less suggest that somebody might have cooked the books.
 

The credibility of Pitts and numerous Obama defenders in the Media rests on what they report, and how they report it: The BLS offers two Household Survey employment numbers, the raw number of 873,000 and the "Adjusted" number of 294,000. The raw number includes, among other things, various categories of folks who do not draw a paycheck, who are "off-the-books", or whose existence is otherwise difficult to verify, including "nonagricultural self employed, unpaid family workers, private household workers, and workers absent without pay from their jobs". The BLS itself places little faith in the raw number, and forthrightly states that the "Adjusted" number better reflects reality, since it tracks so much more consistently with the Payroll Survey.
 

The question of the BLS's credibility rests on three things:
 

1) Has the BLS ever reported such a singular discrepancy in the past? For example, has there ever been a period when the raw Household Survey showed 7.5 times as much of an employment gain as the Payroll Survey? The answer is that it is extremely rare, and invariably reflects an over or under-estimation of actual jobs gains/losses which are smoothed out in subsequent surveys. In fact, the BLS's historical figures show that over the entirety of the Obama Administration, the two surveys reflect almost exactly the same number of job gains and losses. So at the very least, the timing of these survey results is extremely suspect, based on the BLS's own reports.
 

2) Does the BLS place much significance in a one-month result? To their discredit, to a certain extent they do, even though we've established that a single monthly number has no meaning. In their press release, they matter-of-factly state that "the unemployment rate declined by 0.3 percentage point to 7.8 percent in September (873,000 jobs)". Clearly, what they can credibly say is that these results are based on a survey and constitute their best guess. The fact that they don't qualify the results does not reflect well on their credibility, because it provides valuable cover to Obama supporters who can then claim that they are only reciting BLS statistics.
 

3) How susceptible is the BLS survey to false results? Per their website, the BLS admits that they survey a far smaller number of Households (60,000) than Employers (486,000), and that as a result, the margin for error on the raw Household number is "four times that of the payroll survey on a monthly basis".  Ironically, the average of reported jobs by combining the Adjusted Household Survey and the Payroll Survey very much validates that margin of error.
 

Another contributing factor is the BLS practice of changing their survey sample by 25% every period, or 15,000 households. This is a sensible practice over the long term, but increases the likelihood that the results might be skewed month to month. Why is this important? Because in a nation of 135 million jobs, in such a small sample size - 60,000 - it would take only 400 extra Households answering in the affirmative to show an "improvement" of 873,000 jobs. So, if the new sample were to over-represent booming areas like Houston to even a very small degree, you would get a false result. 
 

Case in point: the BLS showed 879,000 jobs gained on an Adjusted basis in February of this year. In March and April, those jobs gains were entirely wiped out, and the remainder of 2012 was a see-saw with no net gain in jobs until this month. In fact, a review of available BLS stats going back to 1990 shows the month-over-month job gains/losses to be all over the map. Bottom line, this shows that a single month's results are meaningless, and that only the overall trends are relevant.
 

So, is Leonard Pitts Jr. right or wrong? In my judgment, he is not only wrong, but irresponsible. Pitts insists that, in questioning the numbers, Republicans are challenging the credibility of the BLS statisticians. While it is unclear what the number crunchers think about how their data is used, it is clear that the BLS had an obligation to put a single month's numbers in perspective. They don't. And nowhere does Pitts point out that the White House and most of the Media cherry-picked the numbers and stripped them of context, so either he hasn't done his homework, in which case he's intellectually lazy, or he knew the numbers were questionable but chose to attack Republicans anyway, in which case - in this instance at least - he's a Demagogue.

Given his resume, that's not a hard charge to substantiate.
 

And the larger issue begs the question of what incredible statistic that redounded to the benefit of the Obama campaign might cause Pitts or others amongst the Faithful to be skeptical: 2 million jobs in one month? 5 million?  At what point would they cease to suspend disbelief regarding the things they read?  Nobody in the Media ever asks them that question, and Republican politicians are too lame to make the case that they should.

Regardless, in this instance at least, Stephen Colbert's claim that "Reality has a well-known liberal bias" appears to be incorrect, and Leonard Pitts Jr. is long overdue for a Reality Check of his own.

Read all about it on the BLS website: http://www.bls.gov/web/empsit/ces_cps_trends.pdf and http://www.bls.gov/news.release/empsit.nr0.htm.