Job seekers wait to enter a job fair at the Alameda County Office of Education on April 24, 2013 in Hayward, Calif.
The U.S. economy added 169,000 jobs in August, according to the Labor Department -- slightly underperforming economists' expectations. And the unemployment rate ticked down a tenth of a percentage point to 7.3 percent.
There was some bad news buried below the decent news. The economy added 74,000 fewer jobs in June and July than originally reported -- pushing the job-creation average during the summer to 148,000 per month. And the decline in the unemployment rate resulted primarily from fewer people in the labor force looking for work -- a decline of 312,000 people. Labor-force participation is now at a low point in the downturn.
And there were contradictory indicators embedded in the report as well.
While employers had 169,000 more jobs on their payrolls, 115,000 fewer Americans reported that they were working.
Doug Handler, chief U.S. economist at IHS Global Insight, calls that a “weird” month. He says it happens from time to time because the two statistics come from different sources.
In what’s called the “household survey,” 60,000 American households are asked: How many adults were working last month? How many were looking for work? How many were doing neither? The sampling of households includes young people, old people; married and single; rich and poor from all over the country. The data crunchers at the Bureau of Labor Statistics then extrapolate from the sample to the whole U.S. population. That’s how they estimate the number of people with jobs.
In what’s called the “establishment survey,” approximately 500,000 business and government workplaces are asked: How many people were on the payroll? Again, the sample is extrapolated to the entire U.S. employer base. That’s how the BLS estimates the number of jobs added to the economy.
So, in the August report on both surveys, the job numbers were down, and up.
“It is just sort of sampling noise,” says Tom Nardone, the labor bureau's associate commissioner for employment and unemployment.
Nardone explains that each survey has a margin of error. “In the case of the household survey, it’s nearly 400,000,” says Nardone. “It might sound odd, but the minus-115,000 [in the household survey] is not statistically different from the [plus-169,000] number from the payroll survey.”
Nardone points out that the margin of error may appear high compared to the number of jobs being reported every month as having been added or lost. However, compared to the entire employment base, the margin is actually very small. In the household survey, the margin of error is 0.3 percent (399,000 distributed among 144 million total U.S. workers). For the establishment survey, the margin of error is 0.07 percent (92,000 distributed among 136 million total U.S. payroll jobs).
Economists typically put more weight on the establishment survey than the household survey in any given month: it is derived from several data sources, and is based on a much larger sample size. But over time, the surveys tend to even out and agree. Nardone says both show total job growth of approximately 2 million over the past year.
And Nardone also points out that the surveys don’t define "job" or "working" precisely the same. The household survey counts a person as “employed” if they have multiple jobs; in the establishment survey, each job would be counted separately. The household survey includes the self-employed and those working on farms. Nardone says that when these technical differences are accounted for, the surveys converge even more.
This was the last monthly jobs report before the next meeting of Federal Reserve policymakers. Economists today were quick to speculate on whether the Fed would consider the labor market strong enough to weather a tapering of monetary stimulus beginning in September.
John Canally, economist at LPL Financial, says Fed policymakers won’t be thrown by the discrepancy in the August jobs report between the household and establishment surveys.
“The Fed’s going to look not just at today’s jobs report, but at 5 or 10 or 15 other labor-market indicators,” says Canally. These include weekly jobless claims, private payrolls, job openings, layoffs, manufacturing and non-manufacturing payrolls. “If you add it all together, it skews a little closer to what the establishment survey said this month, which is that the economy is adding jobs at a relatively modest pace.”