WASHINGTON -- Manipulating the government's monthly unemployment report is impossible because of the large number of people -- mostly civil servants and not political appointees -- involved in compiling the data, said the former commissioner of the Bureau of Labor Statistics.
"I think it would be impossible to really manipulate the numbers," said Keith Hall, who served from 2008 to 2012 as commissioner of the independent statistical agency, which produces the report. "People shouldn't think at all there is any bias in the numbers. If you wanted to try to mess with these numbers, you are talking a very difficult thing. It almost certainly would ... be next to impossible."
A drop of three-tenths of a percentage point in the jobless rate wasn't unusual, Hall said, and any smaller change isn't considered statistically significant.
On the face of it, there was nothing untoward in the numbers. The number of people who reported that they were employed -- 873,000 as measured in the household survey -- rose by a large number over the number of people who reported that they were unemployed during the month, 456,000. In September, 12.1 million Americans reported that they were jobless.
Hall, who was appointed by former President George W. Bush and served through much of the Obama administration, said BLS commissioner is a nonpolitical position. The commissioner serves a four-year term and is not replaced by an incoming president, as the heads of Cabinet departments and other agencies are.
"I feel like I'm a certified economic geek rather than a political person," said Hall, who is a senior research fellow at the Mercatus Center at George Mason University.
During his four years as commissioner of the BLS, which is part of the Labor Department, Hall said he was never asked by the Bush or Obama White House to change any data.
The unemployment rate is calculated differently than the monthly job-growth figure. To determine the rate, Census Bureau employees survey about 50,000 people each month -- mostly over the phone but sometimes in person -- to determine if they are employed, Hall said.
Dozens, if not hundreds, of people are involved in collecting the data and compiling it, he said.
The household survey data are more volatile than the monthly payroll figures on job growth, which are compiled from about 400,000 businesses, he said.
But the household survey can be an early indicator of changes in the jobs market because it can take a while for new businesses to be included in the payroll survey.
"At turning points, sometimes the household survey turns a little quicker than the payroll survey does," Hall said. "It doesn't mean it doesn't give out false signals."