Friday's Employment Situation news release caught me by surprise. While the establishment survey showed an increase of about one hundred thousand in private sector employment, the household survey showed that 873 thousand more people were employed in September 2012 than in August 2012, representing a 0.61% increase in employment in just one month, corresponding to an annual employment growth rate of 7.57%.
As a result, the official unemployment rate fell to 7.8%, first time below 8% in a long time. President Obama and his friends pounced on this news, proclaiming
It's a reminder that this country has come too far to turn back now.
I did not expect such an improvement in the unemployment rate. After all, GDP growth for the 2nd quarter of this year had just recently been revised down to 1.3% and the Fed had initiated another round of quantitative easing.
I just couldn't reconcile the two signals.
So, I wrote a program to download the entire archive of Employment Situation press releases to see exactly how unusual to have an upswing like this. I wrote another program to extract the unrevised employment figure from the household survey table in the press release for each month. I did this, because the official data available from the BLS includes revisions and corrections, and therefore may end up hiding much larger swings that may have occurred with unrevised, initial release data in the past.
Note that there is a discontinuity between January of each year and December of the previous year due to changing population controls. Therefore, I am going to exclude the changes between those two months from the chart below.
It turns out, September 2012 is the third largest one-month increase ever in the initial employment estimate from the household survey. Only the January 1999 (0.66%) and the February 2002 (0.64%) surveys show higher increases.
The chart below shows month-to-month percentage changes in the initial employment estimate based on the household survey between January 2007 (when the Democrats took over the House and the Senate) and September 2012:
January 2012 (0.60%) was the only other month since January 2007 that showed anything close to what we saw in this month's report. However, when the effect of population controls is removed, the increase is only 0.45%.
That means, the increase in September is 35% larger than the largest increase during President Obama's term. The second largest increase was April 2010 with 0.4%. Guess what happened in spring 2010?
Yup, increased Federal hiring due to the 2010 Census, as you can see in the figure below:
Incidentally, Federal employment, excluding the USPS, is up 14% since the Democrats took over the Congress, and almost 9% since President Obama was inaugurated (with the biggest increase in Federal Government employment happening in the first months of President Obama's administration).
The President's supporters seem to think this one month of unemployment below 8% vindicates the stimulus.
It doesn't. The President and his team claimed they knew the exact buttons to push and the exact levers to pull to prevent the unemployment rate from ever going above 8%. I have shown elsewhere how they fell short.
The argument against the stimulus was not that someone else knew that it would be better to push some other buttons, and pull some other levers.
No, the argument was to keep calm, and weather the storm.
The President and his team decided to use the crisis as an excuse to push every little economic intervention they ever dreamed of. Along the way, a few recovery summers came and went with some fanfare.
Now, it would certainly be cause for a celebration if the U.S. finally gets on a robust growth path.
However, it is curious, how, just when President Obama happened to need some serious good news on the economic front because his administration has been, shall we say, coming up short in protecting U.S. interests in the world, we ended up getting this huge increase in employment.
Until there is more evidence otherwise, I would certainly recommend regarding September's employment number from the household survey as a very exceptional data point.