On jobs numbers, BLS vows there's no conspiracy
Updated: 4:24 p.m. ET
On the heels of this morning's positive September jobs report, which showed a 0.3 percentage point drop in the national unemployment rate, prominent conservative Jack Welch and others on the right are publicly questioning the numbers, finding them so impossibly good as to assume they must have been tampered with.
"Unbelievable jobs numbers..these Chicago guys will do anything..can't debate so change numbers," wrote Welch, the former chairman and CEO of General Electric, on Twitter, shortly after the September jobs report was released.
Conservative radio host Laura Ingraham piled on in a Twitter message of her own: "Jobs #s from Labor Secretary Hilda Solis are total pro-Obama propaganda--labor force participation rate at 30-yr low. Abysmal!"
Welch was not available to comment on the nature of his Tweet or to offer up an explanation for its basis -- according to someone in his office, he posted his message before walking into a meeting without his phone -- but the Bureau of Labor Statistics and others immediately knocked down his and other similar comments as factually unsound. (Welch later doubled down on the sentiment in a Fox News appearance, noting that "these numbers don't smell right.")
Romney, GOP downplay positive jobs reportJobs report: Unemployment rate drops to 7.8 percent
In an appearance on CNBC, Solis called Welch's comments "ludicrous."
"BLS is not manipulating data. Evidence of such would be a scandal of enormous proportions & loss of credibility," Tony Fratto, former deputy press secretary to President George W. Bush, wrote on Twitter.
Steve Haugen, an economist at the BLS who has been involved in the process of analyzing jobs data for nearly 30 years, flatly dismissed the idea that there was any way the White House or Obama campaign could have had a hand in how the numbers turned out.
"The data are not manipulated for political reasons. I've been involved in the process myself for almost three decades. There's never been any political manipulation of the data, period," Haugen told CBSNews.com.
According to Haugen, the BLS has been getting its data for the household survey -- the one on which the unemployment figure is based -- from the Census Bureau since 1942. The BLS took over the responsibility for analyzing the employment and unemployment data in 1959. And Haugen says that survey used in the data collection process has been largely the same since 1994. The BLS says the White House is not involved in the process of gathering or analyzing data, and does not give directives on the collection, production, or dissemination of data.
Asked directly if the Obama administration or the White House had directed the BLS to change its methodology in some way to make the numbers more favorable to the president, Haugen said "no."
In fact, the BLS says it does not at the moment have a single political appointee working in the entire agency.
Haugen explained how the data is collected for the household survey.
"Each month there's a survey of about 50,000 households, or addresses, actually, that are drawn randomly -- a random sample designed to represent the population -- and those data are collected both in person and by phone by Census Bureau interviewers," Haugen said. "The data are collected usually during the calendar week including the 19th of the month. The reference period for the survey -- the questions we're asking about what people are doing -- is related to the calendar week that's the 12th of the month, typically. So there's the interview week, the week of the 19th, then there's the week to process the data and then there's the week we release the data."
A March report from the Washington Post also details how the jobs reports are made, describing the analysis process as "an eight-day security lockdown" during which confidentiality agreements are signed each morning and computers are encrypted and data locked into a safe every time one of the analysts goes even to the bathroom.
"The institutions that do this in government are exceedingly professional and hermetically sealed from political influence and manipulation," said James Thurber, a distinguished professor of government at American University and the founder and director of AU's Center for Congressional and Presidential Studies.
Haugen explained that due to the particular data set in question -- that governing the unemployment statistics -- it wouldn't take a multi-bureau government conspiracy in order to produce a 0.3 percentage point shift.
He said that part of the reason for the relatively large drop in unemployment this month has to do with the sample size of the current population survey, also known as the household survey. Because the sample size is smaller than the payroll survey, for example, there's more variability in the numbers.
"On a month-to-month basis, the household survey employment measure is more variable than the payroll employment measure due to the smaller household survey sample," the BLS said in a press release Friday attached to the jobs report. "Over longer periods, the changes in household and payroll survey employment tend to track more closely."
"The smaller the sample size, the higher the variants," Haugen said. "If you're trying to really track the monthly developments in employment you need to focus mostly on data from the payroll survey."
Haugen explained that one factor that likely led to the 0.3 percentage point drop in this month's unemployment is the fact that people in the 20-24 age group (including college students and people who are often working temporary summer jobs) left the job market this summer earlier than expected.
"In August you had an unusually large decline in employment" among that demographic, he said. Because the BLS does seasonal adjustments for its data, it was primed for a big decline in September, when young people have traditionally left the work force. "What happened was there was a big decline in August and not a decline in September. Because there was no decline [when it was expected] there's a big increase after seasonal adjustment."
"In the [monthly household survey] you occasionally get these large movements because it's based on a relatively small survey," he added. "What we tell people is that you need to wait for additional months of data in order to see what the pattern is."
