Fake Good News of a Phony Jobs Recovery

Nothing reeks of suspicion during a tight presidential race like extraordinary coincidence. After getting pummeled on his jobs record during a listless debate performance, President Obama “unexpectedly” received a much needed lift from good news: the Bureau of Labor Statistics (BLS) reported that 873,000 jobs were created in September, knocking the unemployment rate down significantly to 7.8%. If true, this would be the best overall month since 1983.

However, these numbers are at best implausible. How implausible? Well, over the last 29 years the US has enjoyed 22 quarters of economic growth in excess of 5% and, in none of these, did the overall job creation rise to the level of September’s vertiginous heights. And the last time these lofty numbers were recorded was in 1983, when GDP growth was a dizzying 9.3%.

Today’s growth is at 1.3%, if dizzying certainly for different reasons.

So what gives? One overwhelmingly generous interpretation is that this is just the result of an “implausible statistical quirk” created by a skewed sample used to generate the percentage. It is, of course, statistically possible to have a sample so unrepresentative that it produces results well outside the standard margin of error. But given the timing of the news, and the breathtakingly dishonest way Secretary of Labor Hilda Sols has been explaining the numbers, this all appears more snow job than sudden job bonanza.

It’s instructive to delve a little deeper and parse the numbers presented. Each month there are two entirely separate job surveys conducted. The payroll survey counts businesses (410,000 businesses self-report to calculate this number) and the household survey conducts a phone poll of 60,000 homes. There is no comprehensive survey generated that gives us a simple count of all the jobs added and lost. The percentages ultimately produced are statistical extrapolations based on data collecting models, and those models have all kinds of underlying presuppositions.

For September, the household survey, which is generally the more volatile of the two, recorded the stupefying increase of 873,000 jobs. Meanwhile, the payroll survey recorded a much less impressive 114,000. You don’t have to be a master logician to understand that both of these cannot be right. And while it is not unusual for these numbers to be separated by a significant spread from time to time, it is very unusual for the spread to be this enormous. One survey essentially cheers “We’re on fire!” while the other one laments “We’re drowning!”. For these numbers to be taken at all seriously some account has to be given of the gaping discrepancy between them.

Experts everywhere have greeted the numbers with raised eyebrows. Douglas Holtz-Eakin, a former economics adviser to John McCain and the former head the Congressional Budget Office, calls the numbers “implausible.” CNBC’s Jeff Cox said: “The report presented a slew of contradictory data points, with the total employment level soaring despite the low net number”. “No way in the world these numbers are accurate. Somebody needs to do an investigation. … Investigate these numbers”, said Ed Butoesky of Chapwood Capital Investment. Neil Irwin, writing for the Washington Post opined: ”Weird that payrolls are exactly on forecast but household survey is far better.”

Weird, indeed. Even weirder was the brazenly dishonest attempt by Secretary of Labor Solis, speaking on CNBC to spin the numbers in away more politically helpful for Obama’s reelection efforts. The numbers for July and August were upwardly revised by 86,000 jobs, all of which came from the public sector, but she tried to maintain this was an uptick in private sector employment. Her mendacity, especially given all the statistical red flags, is something south of confidence inspiring.

Mitt Romney responded, that even if accurate, the picture painted is still despairingly bleak:

“This is not what a real recovery looks like. We created fewer jobs in September than in August, and fewer jobs in August than in July, and we’ve lost over 600,000 manufacturing jobs since President Obama took office.

Romney went on to explain that, at the very least, the new report doesn’t adequately account for the increasing number of those simply leaving the work force, basically giving up:

“If not for all the people who have simply dropped out of the labor force, the real unemployment rate would be closer to 11 percent,” Mr. Romney said in his statement. “The results of President Obama’s failed policies are staggering — 23 million Americans struggling for work, nearly one in six living in poverty, and 47 million people dependent on food stamps to feed themselves and their families. The choice in this election is clear. Under President Obama, we’ll get another four years like the last four years. If I’m elected, we will have a real recovery with pro-growth policies that will create 12 million new jobs and rising incomes for everyone.”

So if we take our bearing by the payroll survey, 114,000 jobs were added but there are 456,000 fewer unemployed. This means that 342,000 people must have either dropped out of the work force entirely or left the country. One strategy for reducing unemployment, of course, would be simply exiling underemployed citizens.

It’s not at at clear how the BLS came up with these absurd numbers, but the incontrovertible conclusion is that whether the result of some statistical aberration or manipulated methodology, there are not to be taken seriously. There is simply no way 873,000 jobs were created in September. According to Gallup’s own survey, there was no change at all in the unemployment percentage. If we only use the payroll survey the BLS provides, unemployment actually went up to 8.2%. The statistics put out by the BLS is almost surely due to a new methodology, likely the way they’re counting part-time and underemployed workers, without announcing that they’ve essentially changed the rules of the game.

It’s now commonplace to talk about numbers as the last refuge of rational certitude, as if the plain force of arithmetic is insulated from the craven rhetorical alchemy of political posturing. But numbers themselves reside in a context of analysis that itself is fashioned by necessarily political human judgement-what gets counted and how, when the counting starts and stops, what the overall significance is of whatever computation. In fact, there is nothing easier to game than numbers: their contextualized meaning conceals itself behind the misleading appearance of precision. Numbers don’t lie but the people who concoct surely do, often and well.

My hypothesis is this: the BLS tweaked the existing methodology, the statistical model that drives the final extrapolation, in order to sex up the numbers. This was done for political reasons, likely at the behest of someone in Obama’s administration. However, there was already a small bump registered as a consequence of Fed’s most recent stimulus effort, QE3. A sloppily contrived new methodology and an already existing bump produced the monster number that no sane human being could consider accurate. Even when using faux arithmetic, lying well takes some skill.

Here is an interesting number to consider: according to the Daily Presidential Tracking Poll issued by Rasmussen, Romney is winning, edging out Obama 49%-47%. If Obama is going to overcome that math, he’s going to have to get a whole lot better at lying, or saving that, a whole lot better at governing.

Category: Featured, National