Previewing tomorrow’s jobs numbers

Politics,Beltway Confidential,Conn Carroll

Last month, President Obama received a huge boost to his reelection campaign when the Bureau of Labor Statistics (BLS) reported that the unemployment rate had fallen below 8 percent for the first time since he was inaugurated in January 2009. The unemployment had hovered around 8.2 percent for the first eight months of this year.

The quick drop in September’s unemployment rate was driven by a historic 873,000 jump in the number of Americans telling BLS they were employed. Since 1950, the BLS had reported only one other month with a higher job gain: June of 1983, when 991,000 Americans got new jobs and the U.S. economy was growing 9.3 percent. Today, GDP is growing at just 1.3 percent.

September’s unemployment drop looks even more anomalous when you compare it to the jobs numbers reported over the same time frame by U.S. employers. The BLS actually produces two surveys every month: 1) a survey of 60,000 households by Census workers (the household survey that produces the unemployment number); and 2) a survey of approximately 440,000 individual worksites (the employer survey that produces the “jobs created” number). The household survey has always reported more total jobs in the U.S. than the employer survey, but month to month the two surveys mostly move in tandem with strong job gains and losses usually showing up in both surveys.

But not in September’s report. Last month, while the household survey showed a 873,00 job gain, the employer survey reported much more modest 114,000 job growth, a 759,000 job gap. By contrast, in June 1983 when the household survey reported record 991,000 new jobs, the employer survey reported 378,000 jobs created, a gap of just 613,000.

So just how anomalous was the 759,000 job gap between the September household and employer surveys? It has only happened two other times since 1950.

Now the gaps between the household and employer surveys are supposed to be completely random statistical occurrences. But see if you can guess which months they occurred in:
1) An already weakening economy is hit with a spectacular terror attack. In response, the President of the United States stresses how important it is for Americans to keep the economy going. The employer survey reports a loss of 243,000 jobs. But Americans tell the BLS that 605,000 jobs were created, a gap of 848,000.
2) In the midst of a weakening economy, the President of the United States gives a State of the Union where he says America is engaged in a “war against terrorism” and that, “When America works, America prospers, so my economic security plan can be summed up in one word: jobs.” The month after that speech U.S. employers reported a loss of 146,000 jobs while Americans tell the BLS that 737,000 jobs were created, for a gap of 883,000.

The two months in question? September 2001 and February 2002.

So looking ahead to tomorrow’s BLS report, what is the track record for the household and employer surveys the month after anomalously high discrepancies? Here is a chart of the household and employer survey from the months before and after September 2001:

And here is a chart of the household and employer survey from the months before and after February 2002:

In both cases you can see that the employer survey showed steady job losses over all three months while the household survey yo-yoed back and forth between job losses, a job gain, and then losses again. Now here are the last three jobs reports:

Again the employer survey is steady, this time showing weak job growth, while the household survey has gone from small job losses to a huge job gain. If history is any indicator, tomorrows jobs report will show the household survey returning to reality.

In 2001, unemployment ticked up from 5.0 percent in the September report to 5.3 percent in the October report. But in 2002 unemployment stayed flat at 5.7 between February and March 2002 before rising by .2 to 5.9 in April.

View article comments Leave a comment