This is a completely reasonable question. Let me try to answer it.
The unemployment rate and related measures are based on a monthly survey of households. But they don't just call you up and say "Are you unemployed?" They ask a series of questions to decide how you should be characterized.
For most people, this is pretty straightforward. If you have a job and went to work last week, you're "employed." If you got laid off, you're "unemployed." If you're retired, you're "not in the labor force."
But there are a ton of gray areas and this is where it gets tricky.
But there are a ton of gray areas and this is where it gets tricky.
Let's say you take a week off work, unpaid. BLS says you should be counted as "employed, not at work."
On the other hand, if your boss tells you not to come in next week because of a furlough, you should count as "unemployed, on temporary layoff."
On the other hand, if your boss tells you not to come in next week because of a furlough, you should count as "unemployed, on temporary layoff."
When the pandemic hit, BLS told the people who conduct the surveys (who, confusingly, work for the Census Bureau) to count anyone who was on an unpaid absence from work due to the virus as "unemployed."
Apparently, though, this didn't happen in all cases.
Apparently, though, this didn't happen in all cases.
It's not hard to understand how this could happen. Someone who's been told, "We're closed down because of COVID so I don't have any hours for you" might not think of themselves as "laid off," even if, as far as BLS is concerned, they are.
It's not clear why, 2+ months into this crisis, this hasn't been fixed. In today's press release, BLS said: "BLS and the Census Bureau are investigating why this misclassification error continues to occur and are taking additional steps to address the issue."
It's reasonable to ask, "If they know about the problem, why don't they just adjust the data?" The answer is that BLS is very reluctant to make these kind of one-off edits. From the release: "To maintain data integrity, no ad hoc actions are taken to reclassify survey responses."
The reaction to today's surprising numbers gives a pretty good sense of why ad hoc editing might be a bad idea. The more subjective judgment you introduce into the process, the more opportunity there is for bias (intentional or otherwise) to slip into the numbers.
What BLS does, instead, is call out the issue so that we can all make our own adjustments. The issue is noted prominently in our story, and in many other outlets' stories today. It certainly hasn't been buried.
See this from @aliciaparlap for example:
nytimes.com/interactive/20…
See this from @aliciaparlap for example:
nytimes.com/interactive/20…
Should the Times highlight the (adjusted) 16% number instead of the (official) 13% number? I'd argue no, for reasons similar to the BLS's: It's dangerous to start making our own one-off adjustments. But it's not cut and dry, and I think reasonable people could disagree. <fin>
One addendum: It's important to note that the unemployment rate fell *regardless* of what choice you make here, because the same issues affected April's data. In April, the adjusted rate (accounting for this issue) was just under 20%, vs just over 16% in May.
