That’s the short answer, but the longer answer is an examination of something I talk about a lot on Twitter, which is that, by and large, journalists are very bad at maths and numbers.
(A note: Muphry’s Law dictates that I will make at least one number mistake in this post, so let me apologise in advance for that.)
If you want to just see the working, skip to the heading “Did 140,000 women lose their jobs?” at the end.
I noticed this striking claim today in a tweet about a New York Times story. The story itself is well worth your time. It’s concerning but very well reported.
Jobless, Selling Nudes Online and Still Struggling
OnlyFans, a social media platform that allows people to sell explicit photos of themselves, has boomed during the…
The claim is that there were 140,000 jobs lost in the USA in December 2020, and they were all jobs that were held by women. That is to say, 140,000 American women lost their jobs in that month, and no men did.
As they say, extraordinary claims — of which this seems to be one — require extraordinary evidence. So I did as the tweeter — a New York Times editor — requested, and looked it up. Here’s what you see if you do that:
Not all of these sources look entirely reputable on the subject, but even with modern journalism’s tendency to what Nick Davies called ‘churnalism’ — repetition, without checking, of press releases and other people’s reporting, by writers pressed to squeeze those valuable clicks out of anything and everything — there seemed like a big mass of reporting to suggest the claim had been checked.
As you can see from this second search page screenshot (the one above is from Google News and the one below is from Google web), it’s been reported by CNN and the American financial TV channel CNBC. So maybe — I thought — it’s true and my sceptical instinct was wrong.
Sadly not. It is wrong. I found the actual data release from the US Bureau of Labor Statistics (pdf link) which does show a headline figure in the very first sentence of an employment decline of 140,000 in December.
Let’s delve further. CNBC even helpfully links to the data source, which does indeed show a net drop of 156,000 in women’s employment in December (row 1 of the table drops from 71,038m in November to 70,882m in December). When we add the net gain of 16,000 in men’s employment in the same month to the net loss of 156,000 in women’s employment we get 140,000.
What’s going on? The third page of the BLS news release explains. The net loss of 140,000 is the across all (non-farm) sectors of the American economy. But on that page the BLS breaks out the various sectors.
Leisure and hospitality: -498,000
Private education: -63,000
Other services: -22,000
That’s a total loss in those sectors of 628,000 jobs. Then it goes on to discuss sectors that gained jobs in December.
Professional and business services: 161,000
Retail trade: 121,000
Transportation and warehousing: 47,000
Health care: 39,000
Wholesale trade: 25,000
It then describes a few other sectors in which the December change was minimal. So clearly the above tallies don’t show the whole picture, but they do make it clear that “140,000 women lost their jobs, and no men” cannot possibly be true.
We see from the above gains that there were at least 482,000 new jobs in December. Take away the losses from the gains and we’re left with a net loss of 146,000 jobs. Let’s assume that the discrepancy of 6,000 is down to rounding and to sectors not broken out by the BLS press release. Or I’ve done the maths slightly wrong somewhere.
So there can’t have been “140,000 jobs lost by women, and none by men” because the job losses were nearly five times that: in fact, 682,000 people lost their jobs in December. The 140,000 is the net loss. It can’t be translated neatly into 140,000 of anything, which I think is one thing that has got the journalists confused.
Secondly, does it make any sense that 140,000 women lost their jobs, and no men did? We should exercise our scepticism when we see claims such as this. How likely is it that no men lost their jobs in a given month? It’s a striking claim, and it’s easy to go and post something like that without checking, just because it’s so surprising. We should turn that on its head and say that surprising claims should be checked before we share them.
It’s also easy to get lost in the numbers. It took me about half an hour to pull the data, understand it, and work out where people had gone wrong. And I’m still not totally sure I haven’t made a mistake somewhere. So I can understand how, if you’re being pressed to file a story — any story — and this one comes across your desk, it’s tempting to just print the legend.
I can understand it more when such reporting comes from “content mills” and poorly paid small publishers, but it’s very disappointing for CNN not to have bothered to check its claim, or for CNBC — supposedly a financial publisher — to have been similarly careless. I should praise the one outlet I could find, Crain’s Detroit Business (left), that correctly reported the figures.
Did 140,000 women lose their jobs?
The data above doesn’t give us the men-women breakdown, but it’s pretty clear that, almost certainly, a lot more than 140,000 women lost their jobs.
Did every man employed on November 30 still have his job on December 31?
Clearly not. I suspect the BLS tables break that out clearly but I wanted this to be a quick look, and I don’t have the time to check them out. But it is clear from the numbers that such a claim does not — and cannot — hold water.
Does this matter? It seems highly likely, given the sectors (leisure, hospitality, education and government) that a lot of those 682,000 people who lost their jobs are women, and so the sentiment behind the claim is true: December job losses probably did hit women hard. But I think it’s important that journalists in particular take care to make sure that things they report are true.