Kellyanne Conway, counselor to President Donald Trump, drew derision from labor economists Monday after her assertion of "fakery" in reference to jobs data.
Conway's comments came as she defended Trump's position on Friday's jobs report, which he once denounced as "phony" but now accepts as "very real."
"Any allegation that the numbers are deliberately biased are nonsense," said Harry Holzer, professor of public policy at Georgetown University. "The procedures are very well known and highly respected."
No Recent Changes
Those protocols also have changed remarkably little over the decades, said Gary Burtless, a labor economist at the Brookings Institution. "They haven't changed them for quite a long time. Every now and then they tweak the questions a little bit," he said, changes that are documented in the survey methodology and on the BLS website.
"The official unemployment rate is the most comparable historically most comparable internationally," said Erica L. Groshen, former Commissioner of Labor Statistics and head of the Bureau of Labor Statistics. "But we know in this large and complex market that's not the only measure of activity you might find," she said, which is why the BLS draws on different sources for its jobs reports, and produces six different measures of labor utilization.
"Broadly speaking, all six of the measures move the same way over the business cycle," Groshen said.
A Representative Sample
The raw data actually comes from two monthly surveys, a survey of around 60,000 households and another one of 600,000 workplaces, Groshen said. "These companies that report to us have been chosen to be representative geographically and by industry of all the employers in the country," she said.
The household survey, called the Current Population Survey and conducted in conjunction with the U.S. Census Bureau, is similarly exhaustive. Households that are chosen are surveyed over a four-month period, after which they cycle off the survey for eight months, then return for a second four-month participation period. Like the employer survey, Groshen said that the BLS works to get a representative sample of Americans. Participants start off getting a visit from a Census worker for an in-person interview, and surveyors follow up using contact info the respondents provide.
Skeptics argue that the topline unemployment figure doesn't capture a full picture of out-of-work Americans.
Trump Isn't the Only Naysayer
Trump, who at one point claimed the unemployment rate could be as high as 42 percent, is not the only public figure to express skepticism about the official unemployment rate. In a 2015 blog post titled "The Big Lie," Gallup CEO Jim Clifton wrote, "The official unemployment rate, as reported by the U.S. Department of Labor, is extremely misleading."
In 2012, former General Electric CEO Jack Welch ignited a social media firestorm when he tweeted that "these Chicago guys" — a reference to President Obama's appointees — "change numbers" in the BLS reports, an allegation widely panned on Twitter and by Labor Department officials.
Labor market economists say the BLS includes more comprehensive snapshots in its employment releases every month.
"That's based only on people who are actively looking for work," Holzer said of the unemployment figure cited most frequently by politicians and the media. "But we all know that, and BLS knows that. The BLS presents a whole range of unemployment numbers and people are free to look at the whole batch of them."
The widest categorization of unemployment is in the U-6 table, which measures underutilization; it includes people who want to work full-time but can only find a part-time job as well as people aren't working and want to work but haven't looked for a job in the last four weeks, and those who aren't looking for a job because they don't think there are any available for them.
There are also other measures for calculating unemployment, such as state-level data on people filing unemployment claims for the first time, which Burtless said corroborates what the BLS publishes. "These number reflect what Americans are telling the Census Bureau," he said.
Almost Impossible to Massage
The sheer size and scope of the data-gathering operation would thwart any deliberate attempts to manipulate the unemployment rate, Groshen said.
"You should think of statistical agencies as more like a factory than you might realize. There's a huge division of labor in order to get these numbers out on a monthly basis," she said. "There is no way for one bad actor — or two or three — to affect the numbers."