October 5, 2012

October 5, 2012 | Commentary on Economy

Something 'Really Wrong' in Jobs Numbers

Jack Welch is wrong, but that doesn’t make today’s unemployment numbers right. This morning the former General Electric CEO tweeted: “Unbelievable jobs numbers . . . these Chicago guys will do anything . . . can’t debate so change numbers.”

It would be virtually impossible for the Obama administration to fudge or otherwise tweak the employment numbers. The survey was conducted two weeks ago — well before the debate.

Further, President Obama has no political appointees in the Bureau of Labor Statistics (BLS). Until January the BLS commissioner was a holdover from the Bush administration whose term had not expired. The agency’s acting head is a member of the career civil service. Even if his campaign wanted to, Obama could not tweak the numbers.

That said, it is highly unlikely that today’s unemployment numbers are accurate. BLS conducts two main labor-market surveys: the payroll survey (polling employers) and the household survey (polling individuals). The household survey showed a 0.3 percent drop in unemployment and 873,000 net new jobs — the most since mid-1983. The payroll survey showed one-eighth as many jobs added (+114,000).

Unfortunately, the payroll survey is probably correct. Of the two, the household survey has a much smaller sample size and thus larger margin of error. Although the two surveys tend to show the same results over time, the household survey jumps around more on a month-to-month basis. Even without fudging, the laws of statistics dictate that some polls will produce results outside the margin of error. One out of every 20 polls is somewhat wrong, and one out of 100 polls is really wrong.

Today’s household survey looks a lot like that one poll in 100. It reports stronger job creation than any time since the height of the Reagan economic boom. However, nothing else shows anything similar.

The BLS payroll survey shows continued slow job growth. So does another payroll survey conducted by a private-sector firm. New claims for unemployment insurance remain stuck around 375,000 a week – higher than you see in a strong economy. The government just revised second-quarter economic growth estimates down to an anemic 1.3 percent. No other indicators point to an economic boom.

This looks a lot like an anomalous survey, which is quite possible without any political hijinks. It would be fantastic if today’s unemployment numbers are accurate. However, there is a good chance the next household survey (due November 2, four days before the election) will show unemployment jumping back up to 8 percent.

Analysts of all stripes should hesitate to read too much into today’s numbers.

About the Author

James Sherk Research Fellow, Labor Economics
Center for Data Analysis

First appeared in National Review Online's The Corner.