Did someone rig the unemployment figures last Friday? Former GE head Jack Welch thinks they did, and so does Donald Trump.
A lot of other people probably share that opinion, since the big drop in the jobless rate — from 8.1 percent to 7.8 percent — this close to the election seemed awfully convenient for President Obama.
By the end of this column, you’ll have your own opinion that is based on facts rather than just gut instinct and biases. Let’s take this step by step to the thrilling end.
Each month the Labor Department does two employment reports: One surveys businesses to see how many new people they’ve hired, the other is based on contact with 60,000 households.
The households survey asks people if they are still employed, have found work in the past month or have given up looking. It assumes people are being honest in their answers — so I will, too. The surprise 7.8 percent unemployment rate is derived from that survey of 60,000 households, and that’s where the controversy erupted last week.
There are some things you need to understand up front: The Census Bureau, and not the Labor Department, conducts the household survey. So if you want to level accusations, aim them at the right federal department.
Surveys with sample sizes of 60,000 are very big and are considered very accurate. Political polling, for instance, is done with scientifically picked samples of just a couple of thousand respondents.
But even when 60,000 households are questioned, a relatively small number of responses can affect the statistics. For instance, each response in the Labor/Census jobless-rate survey represents about 2,500 Americans.
So if only an additional 60 people who were surveyed in September said they had found work, that would have caused a 0.1 percent drop in the jobless rate.
The 0.3 percent drop in the September unemployment rate that was reported Friday meant that 180 additional people said they had found some kind of paid employment.
In other words, small numbers of people can change the perception of the job market. Some think they can even change the outcome of a presidential election.
Where did the Census Bureau get these extra 180 people who found work? Probably among 20- to 24-year-olds. In fact, the acting head of the Bureau of Labor Statistics found this so odd that he put a note in his report saying — without explanation — that an additional 368,000 people between those ages became employed in September.
Gee, weren’t they lucky!
All told, there was an increase of 873,000 jobs in the household report, while the survey of employers found only a 114,000 job gain. So, the contribution of people ages 20 to 24 years was significant.
The conspiracy theorists were out in force, saying that the Obama administration had changed the numbers. That’s pretty simplistic, although you’ll still be wondering about that when I’m finished.
Remember, the 368,000 gain in jobs among 20- to 24-year-olds as well as the 873,000 gain are projections based on a much smaller number of actual responses. Still, something clearly happened in the 20 to 24 age group that was extraordinary.
What do college students — usually between the ages of 20 and 24 — do at the end of the summer?
Well, this year a lot of them may have joined political campaigns — for Obama, Mitt Romney and the thousands of other candidates around the country. And if those young people were paid to work the campaign, they could have been counted as newly employed by the unemployment survey.
Remember, Census would just have to reach 180 newly employed college student/campaign workers to make up the entire 0.3 percent drop in the jobless rate. If you think the number was rigged, you might theorize that Census made more than the normal number of survey calls near college campuses.
Theory No. 2, and the one subscribed to by sources inside the Labor Department, is that there was some sort of calendar shift at the beginning of the new college semester that threw off the seasonal adjustments in the unemployment report.
The Labor Department’s computers expect certain things to happen at particular points during the year. In the case of students, the computers expect them to quit their summer jobs in August to get back to school.
If, for some reason, the students remain on their “summer” jobs into September, the seasonal adjustment programs will count them as newly employed because they aren’t expected to still be working.
There is evidence that the seasonal adjustments indeed went crazy this September. Before these adjustments, the job gain for 20- to 24-year-olds in the household survey was around 100,000 — not 368,000.
Here’s where the controversy starts: Why was there a sudden shift in the pattern of students returning to campus? Students this year should have kept their summer employment through August and quit well before the September unemployment data was compiled. The seasonal adjustment programs, in other words, shouldn’t have gone haywire.
Did someone fool with these adjustments?
I’ll leave it there. At least now, if you discuss this at the next cocktail party, you’ll have some facts to back up whichever side you take. Me? I’m not saying what I think.