Early in June 2020, the U.S. Bureau of Labor Statistics released the jobs numbers for May. The media rejoiced as though it were a brisk rain after a drought—which, in a sense, it seemed. Here are some of the headlines:
- NPR: “May Surprise: U.S. Adds 2.5 Million Jobs As Unemployment Dips To 13.3%”
- New York Times: “Unexpected Drop in U.S. Unemployment Helps Markets Rally”
- CNN: “America’s unemployment rate falls to 13.3% as economy posts surprise job gains”
But by the next day, or later, a different set were appearing, like the Philadelphia Inquirer’s “Why the May job numbers were wrong and how to read economic data in the coronavirus era” or the Washington Post’s “A ‘misclassification error’ made the May unemployment rate look better than it is. Here’s what happened.”
Or you could have gone with a write up by Ben Casselman at the New York Times or my own at Forbes.com: “Trying To Figure Out The Crazy ‘Drop’ In Unemployment”.
There were a number of problems that should have been obvious if reporters had read what the BLS published and had understood the implications, meaning had more comfort with numbers.
One issue was a major misclassification problem, with people being put into one category that meant they didn’t seem unemployed. That shouldn’t have been as much a surprise as it was to most. BLS had the same issue in the April jobs report, as I noted when that came out.
As many journalists piled onto this—which should have screamed out at them when they wrote their original reports—they missed one or more of others:
- Standard definitions ignore many who are working. Someone in the survey has to have actively looked for and applied for work in one of the accepted ways within the previous four weeks. If everything is shut down and you just lost your job, where can you look for another? That means you’re technically not looking for work, as so don’t count as unemployed.
- Not only was there misclassification, but the response rate to the survey was 70%, or 13 percentage points lower than usual. The same was true in April. People weren’t responding, and when you have a sudden change in how people react to a survey, you have to wonder whether you’re losing accuracy.
- The “official” unemployment rate, called the U-3, screens out many who might reasonably otherwise be considered unemployed, including those who are discouraged or working only part-time because they can’t find full-time employment. Many of those additional people are collected in the U-6 definition, which was 22.8% in April and 21.2% in May. The difference between the U-3 and U-6 was much larger than it typically runs, which should have caught attention from journalists covering it.
And then, as some noticed that African Americans saw 16.8% unemployment under the U-3 definition, few if any bothered to mention that 17.6% of Hispanics were out of work.
This was one sorry example of coverage. Yes, there’s lots of time pressure and you probably are pushed to get something up quickly. But after filing, immediately go back and look at the details for what you may have missed. And then add it in.
