Jobs revision show that Obama has created jobs overall

New jobs numbers show jobs figures under-reported by 380,000. Matthew Zeitlin checks the data.

John Moore / Getty Images

News flash! In 2012, there were 386,000 more people at work on payroll jobs in March 2012 than originally reported, according to a new report from the Bureau of Labor Statistics. That’s an increase of a mere 0.3 percent of the previously reported total. But the annual revision shows BLS understated the number of new payroll jobs created between March 2011 and March 2012 by about a fifth. Although this is just a preliminary estimate of the new employment level, it does have some political significance. Job growth has been slightly stronger in the last year. And now the Obama administration can say that there has been overall job growth since the president’s inauguration in January 2009.

Should this revision hold up, it will mean that 125,000 net new jobs had been created since Obama’s inauguration in January 2009—rather than a net job loss. The economy lost around 4.3 million jobs in the first year he was in office. With the new revision, roughly 4.4 million jobs have been created since the job market started consistently adding new jobs in March 2010.

But how could the numbers change so much? When the Bureau of Labor Statistics reports its jobs figures every month they are estimates. BLS polls a sample of firms and then extrapolates those figures to reflect the experience of all firms. The benchmark revision, on the other hand, is based on extensive data from unemployment insurance records that nearly all firms are required to file.

So, instead of surveying a small portion of employers and crossing their fingers that the sample is representative, BLS actually counts the number of jobs lost, maintained, or created across all firms. In the initial employer survey, the BLS could have been slightly off in its estimation of how many firms actually exist. So the sample could have been bigger than they thought it was compared to all companies.

That’s because companies are created and destroyed all the time, and the BLS estimate cannot always keep track of these changes in the total number of employers. The unemployment insurance records, however, must be filed by virtually every company in the state in which it is located from the day it starts hiring to the day it goes out of business. So data based on these records are the gold standard of employment measures, according to University of Michigan economist Justin Wolfers. The final revision will be issued in February.

Although the BLS has only issued a revision to the total level of employment for March of this year and hasn’t revised to the month-to-month data, the increase can be averaged out across the 12 months back to March of last year. So, instead of creating 162,000 jobs a month over that period, it now appears the economy was creating 194,000 a month, a nearly 20 percent increase over the previous estimate.

“The broader context is still disappointing, but job growth was faster than we thought,” Wolfers said. Wolfers added that the size of the benchmark revision puts into context the focus on the monthly job report, which is oftentimes significantly revised up or down. This past month, the number missed economists’ expectations by about 30,000. “We missed the August number by 30,000, this re-benchmarking is 12 times larger, it really is important,” says Wolfers.