The Wall Street Journal provides a master class this morning in how to mislead with statistics. The subject is unemployment, and every statement here is 100% true:
On average, it now takes people about six months to find a job, roughly a month longer than it did during the postpandemic hiring boom in early 2023.
....More people getting unemployment benefits are drawing on public aid longer. New data released last week from the Labor Department show that 1.8 million people continued to file for previously granted unemployment benefits as of late December, near the postpandemic high.
Year-over-year wage growth has fallen to 4%, down from about 6% at the height of the early 2020s hiring spree. That’s a sign that many employers don’t have to jostle so hard to attract workers.
If this is all true, then what's the problem? First up, here's how long it's taking people to find a job:
The mean length of unemployment is indeed about six months, but that's a lousy measure because it's skewed by the small number of people who have been job hunting for years. That's why the BLS reports median right alongside. It's currently ten weeks, up about 7% over the past year.
Second, there's that stat about "continuing" unemployment claims. This is not a frequently used measure. The one that's reported every week and far more commonly used is new unemployment claims:
New claims have been declining for the past six months and are currently right around their pre-pandemic average.
Finally, there's the claim that wage growth has come down. But that doesn't account for inflation, which has also come down:
Adjusted for inflation, pay growth is higher than it was in 2022 and has been steady for the past two years.
In sum: The median length of unemployment is ten weeks and has barely grown over the past year. New unemployment claims have been dropping and are currently about average. Real wage growth is higher than it was two years ago.
It's true that headline unemployment is up from 3.4% to 4.2% over the past two years. If you leave it at that, you're being honest but, I suppose, a bit dull. That's still a pretty low unemployment rate, after all. But while cherry picking weird statistics might allow you to spice things up, it misleads your readers pretty badly. The Journal makes a habit of this and they should knock it off.
Yeah, they should knock it off. But they're a fully-dedicated agitprop operation, so they won't.
In times of low unemployment, the mean and median would tend to get further apart. As the only people having trouble finding a job will be the unemployable.
The thing I love about these WSJ articles is that they include within them the very things that refute their main contention. Often it's with a helpful graph which tells an entirely different story from what the author is trying to tell. This time we start with the first sentence of the article:
"The U.S. economy has added more than two million jobs over the past year."
That's a remarkable achievement. How can you start with that and spin a story of economic catastrophe? Why, you compare hiring to the peak of the post-pandemic boom, of course:
"The unemployment rate, at 4.2%, remains well below the average during the decade before the pandemic. But there is now just about one job posting per unemployed worker, down from two in early 2022."
"In a survey of consumers by the Conference Board in December, 37% said that jobs were plentiful, down from 57% in the middle of 2022." (emphasis added)
Follow with a bunch of anecdotes of people struggling to find jobs, as if this is a new phenomenon. Then you show the hiring rates for "selected industries", leading one to wonder (not really) on what basis these were selected.
SSDD
It must be incredibly difficult to find 3 people who have difficulty finding a job in a country of 321 million. The WSJ really had to dig for this.
I blame Trumponomics.
Statistical literacy is still both a great educational and infrastructure project.