Trying to Get a Handle on the Unemployment Figures
There is nothing wrong with healthy skepticism (which is different than outright denialism).
Let me start with what I hope is a clear position: I do not have a problem with people asking questions about numbers (or, really, about anything at all) whether we are talking about poll numbers or the unemployment figures, or whatever the case may be. What I have a profound problem with, however, are people who question numbers not because they have a reason to do so, but because they simply don’t like the result.
An apropos (although hardly perfect, analogy-wise) example from last night: Braves fans have a reason to question their loss in their playoff game last night because of a controversial application of the in-field fly rule by an umpire. Braves fans can argue, with some legitimacy, that the umpire in question contributed to their loss. Now, Texas Rangers fans can try and cook up a conspiracy about how they lost because of the umps, but they would be full of baloney to make such claims.
My problem, as I have tried to explain numerous times, with what I call poll denialism is that the claims made regarding the polls are predicated not on sound arguments, but exclusively on the fact that some people do not like the results. Indeed, some have gone so far as to re-calculate the numbers to make them fit a preferred outcome. Such moves are not science, analysis, nor argument. They are fiction.
The issue to me in all of this is not whether the numbers reflect my personal preferences, but rather whether or not we are having a rational conservation (both individually and nationally) on these issues.
Turning to the unemployment numbers, I have two reactions.
First, the notion that the Bureau of Labor statistics, a group of career civil servants who study stats (i.e., are numbers geeks) are going to cook the books to help an incumbent president are absurd. As such, I think that the 7.8% number has to be accepted within the confines of the process that has produced unemployment rates for decades. Indeed, one of the ironies of the denialist position is that they seem to accept older unemployment figures as true (since they make explicit comparisons to the past), but they just don’t like this one. This is a self-negating position because if the most recent numbers can be as easily manipulated as they are claiming, then there is absolutely no reason to trust the fidelity of past numbers. Once we hit that territory it is impossible to have a conversation about these figures because they all become utterly suspect.
This leads to my second observation: clearly the unemployment calculation is a bit less solid than basic election polling. It is very much a less exact process. As such, it is fair to ask about the degree to which a given number is accurate. For example we can note a key difference between the two processes: it is normal for the unemployment report to be revised upward or downward over time. Further, the margin for error in election polling and determining the unemployment rate are quite different.
Ultimately there is, therefore, a profoundly important difference between questioning numbers because of methodology or because they seem out of sync with previously results and rejecting them as part of a conspiracy.
So, what should we say about the numbers yesterday?
The number that caused, understandably, was the household survey. Catherine Rampell at the NYT‘s Economix blog notes:
the household survey — the survey that the unemployment rate comes from — showed that the number of people with jobs rose 873,000 in September, though the gain had averaged 164,000 each month earlier this year.
These numbers are always tremendously volatile, but the reasons are statistical, not political. The numbers come from a tiny survey with a margin of error of 400,000. Every month there are wild swings, and no one takes them at face value. The swings usually attract less attention, though, because the political stakes are usually lower.
I am going to agree that that 873,00 number is improbable and that we may find out that it was in error in the next survey (or we may find out that they were correct). Regardless, that is one heckuva an MOE.
It should noted that a lot the people in that number (582,000) are part-timers.
It should also be noted that the 873,000 number is supposed to reflect the number of people who found work, while the 114,000 number (the other key number released yesterday and derived from a different survey) is supposed to be the number of new jobs added to the economy. I do agree that there is a reason to ask how the two numbers work together. There are also a variety of variables here, including question of long-term unemployed, people who are, or are not looking for jobs, etc. I do not, for the moment, pretend to be able to succinctly explain (or even claim to fully understand) all of these figures. Indeed, one of the problems here is that we are comparing numbers from two different surveys and which do not avail themselves to quick arithmetic (since part of the issue is not just jobs v. no jobs, but also whether people are actively looking for work or not).
In regards to the private sector job creation number we have the following in terms of history and trends:
Here’s a longer-term view:
Ultimately, however, it isn’t as if 7.8% is a fantastic number. And, for that matter, a movement of .3 points is hardly Olympian (and certainly not unprecedented).
It can also be helpful, when evaluating a given set of numbers, to look for evidence that either corroborates (or negates) the numbers in question. We can do this by turning to Gallup’s unemployment survey (which uses a different methodology):
Now, to me the biggest take-away from the graph above is not the actual number (although the seasonally adjusted number is quite close to the BLS figure), but rather the trend. The trend comports with the general trend of the official unemployment figures. This gives us a reason to think that the numbers from yesterday are legitimate.
In terms of overall numbers, Tom Foreman notes:
The Bureau of Labor Statistics confirms that a lot of jobs have been created under Obama’s leadership — 4.4 million by the bureau’s latest count. What Obama did not say, however, was that the nation shed 4.3 million jobs during the early days of his term, and that the net gain since he took the oath of office in January 2009 is just 125,000 jobs.
A such, we are just slightly better off now than we were back at the end of the Bush administration. And, of course, the unemployment rate at the time was 7.8% (which may sound familiar).
Another measure that strikes me as of interest is the recent increase in the President’s approval rating to 54% in the Gallup survey, which would seem to indicate a general increase of satisfaction in the population. One of the reasons that number gets better, especially at levels not seen in years, is economic improvement.
The general trend has been, in fact, a weak recovery. We should be happy about the recovery part, and not so pleased by the weakness part.
See also, Joe Nocera: Jobs Report: Cooked or Correct?
In conclusion, I would state that I can see why there is debate over the 873,000 figures (a debate that will be settled through the next several jobs reports because, unfortunately, actual conclusions require time and data). However, I would also note that the evidence does suggest a recovery and so, to turn this back to electoral politics, it is not surprising that the incumbent finds himself in a somewhat better position at the moment than does the challenger.