The Real September Jobs Report
In just two short months the unemployment rate dropped a remarkable half a percentage point to 7.8%.
September’s miracle drop of .3% caught everyone, including economists, off-guard.
Nobody predicted September’s surprise, yet it happened.
If true, it would be an amazing turnaround in a severely depressed jobs market.
When something sounds to good to be true, it usually is. Sadly, in this case, it isn’t true. The explanation is remarkably simple. The statistical anomaly that created it stands out like a sore thumb.
Yet the anomaly has escaped the radar of all the experts. September’s miracle is but a sadistic statistical trick of providence played against unemployed American workers.
It translates into this election year’s October surprise.
You need only understand how the unemployment rate is determined to see the glaring error.
Why is September Different?
Unemployment dropped .2% in August. That is a big drop even though the job market had not improved. It was obvious because only +96,000 new jobs in August were reported. That is less than accountable through population growth alone.
There had to be another explanation for the drop. There was. Its a sad but perfectly reasonable one immediately pounced on by experts. The Civilian Labor Force dropped by -368,000. That’s nearly four times more workers simply gave up and dropped out of the workforce than new jobs were created.
There is a similar situation for September. There were only +114,000 jobs created. Though higher than August, that is still barely enough to account for population growth.
There has to be a reasonable explanation. You can’t have so few jobs created and get a real, very large .3% drop in unemployment. That’s not possible.
How is the Unemployment Rate Calculated?
This is the key question. Knowing the answer unlocks the secret of September’s .3% drop in unemployment.
At its heart The Monthly Jobs Report (the .pdf version) contains two summary tables of data. The first is titled Household Data. The second is called Establishment Data.
They are two entirely distinct and separate databases. Both are monthly surveys. However, they are gathered by two completely different methods from two completely different sources.
The unemployment rate reported for September (and every month for that matter) is calculated from the Household survey data like in this month’s jobs report:
Unemployment is calculated from a simple formula:
Unemployment Rate = 100 – ((Employed / Civilian Labor Force) * 100)
Plug in Employed and Civilian Labor Force as highlight in the graphic above into the formula and round it to the nearest 1/10th of a point and – Voila! – you get 7.8%.
Why Suffer a Math Lesson?
Simple. In the table above the number just to the right of the 142,974,000 Employed number for the month of September is +873,000.
That says +873,000 jobs were created in the month of September!! That raised the Civilian Labor Force by +873,000 new jobs. That is what lowered the unemployment rate to 7.8%.
That is a far cry higher than the +114,000 new jobs reported in September’s Monthly Jobs Report. Why?
Therein, young grasshopper, lies the error in the unemployment rate!
The Household Survey
It helps to see data displayed in chart form.
Above, the Household survey, also known as the A-1 or Current Population Survey report, provides the monthly totals for everyone employed in the United States. It excludes military personnel which are not reported by the Department of Labor.
As can be easily seen in the far right of the chart above, +873,000 new jobs in one month is an unbelievable increase. There was another equally unbelievable new jobs one last January, but that is partially explained since the Household survey is re-calibrated to match the far more reliable Establishment survey every January. The most besides those two anomalies since 2008 is a little over 500K in January of 2010, also subject to adjustment recalibration.
This month’s report is not subject to normal recalibration changes.
The Household survey gets its data by asking questions of 60,000 selected citizens out of a total adult population of 155 million civilians in the workforce. 2,200 government-paid surveyors gather that data.
That sounds impressive and is much larger than most standard 2,000 or so people surveyed in most nationally reported surveys.
However, it differs dramatically from the more reliable Establishment survey that collects federally mandated data electronically from employers for a full 1/3rd of the ENTIRE workforce!
According to the BLS, the ‘Establishment’ survey is accurate to +/- 100,000 whereas the ‘Household’ survey is accurate to a far less +/- 400,000.
The ‘Establishment’ survey is a more reliable measure of employment than the ‘Household’ survey ever will be. That is why job creation stats are quoted from the ‘Establishment’ survey instead of the ‘Household’ one.
A funny thing happened on the way to the ballot box. It was Christmas come early for President Obama. It was this election cycle’s October surprise.
Even CNN’s Ali Velshi, who is a savvy economics guy by TV standards, was fooled into thinking the drop in unemployment rate to 7.8% is real.
As proved above, it isn’t. It is an artifact of the far less reliable household survey.
In the heady days of the tech boom of the 1990s during the Clinton Administration there were only 230,000 jobs a month created. Nobody in their right mind would believe there were +873,000 jobs created in September alone under these economic conditions.
More reasonably, if we assume +114,000 jobs created in September from the more reliable Establishment survey and plug it into the formula above, then the unemployment rate would calculate to a more believable 8.3%… which is more in line with CBO and economist expectations.
Politics works like cause and effect; within it there is no such thing as coincidence!