What to make of the historically low unemployment rate
Jan 10, 2019
One of the amazing economic stories of the moment is the unemployment rate, which at around 4% has returned to the level last reached during the peak of the tech boom in 2000. The story is much more complex than it seems.
I devoted a chapter of Numbersense (link) to explain how the government computes unemployment rates. The most important thing to realize is that an unemployment rate of 4 percent does NOT mean that four out of 100 people in the U.S. are unemployed, and 96 out of 100 are employed.
It doesn't even mean that four out of 100 people of working age are unemployed, and 96 out of 100 of working age are employed.
What it means is of the people that the government decides are "employable", 96 out of 100 are employed. Officially, this employability is known as "in labor force." There are many ways to be disqualified from the labor force; one example is if the government decides that the person is not looking for a job.
On the flip side, who the government counts as "employed" also matters! Part-timers are considered employed. They are counted just like a full-time employee in the unemployment metric. Part-time, according to the government, is one to 34 hours worked during the week the survey is administered.
So two factors can affect the unemployment rate a lot - the proportion of the population considered "not in labor force" (thus not counted at all); and the proportion of those considered employed who are part-timers. (Those are two disjoint groups.)
The following chart then shows that despite the unemployment rate looking great, the U.S. labor market in 2018 looks nothing like what it looked like from 1990 to 2008.
Technical notes: all the data are seasonally adjusted by the Bureau of Labor Statistics. I used a spline to smooth the data first - the top chart shows the smoothed version of the unemployment rates. Smoothing removes month-to-month sharp edges from the second chart. The color scale is based on standardized values of the smoothed data.
P.S. See Part 2 of this series explores the different experiences of male and female workers. Also, the entire collection can be viewed here.
just a quick note: percent translates to one out of hundred (and not one out of ten as mentioned in the article)
Posted by: percent | Jan 10, 2019 at 09:08 AM
You say "4 percent does NOT mean that four out of ten people in the U.S. are unemployed".
I should hope so! I guess it also doesn't mean that four out of 100 people in the US are unemployed either ;)
Posted by: Nick | Jan 10, 2019 at 09:08 AM
percent, Nick: thanks for the catch! Corrected.
Posted by: Kaiser | Jan 10, 2019 at 10:37 AM
Getting closer. I think you want to fix the working case to say 96 out of 100 now. It is an interesting article regardless. Thanks.
Posted by: David | Jan 10, 2019 at 12:30 PM
David: Of course! It's a case of having too many balls in the air. Thanks for the note.
Posted by: Kaiser | Jan 10, 2019 at 12:54 PM
Is the chart yours or is it a Bureau of Labor Statistics production? Is it possible to have a link? Thank you.
Posted by: Antonio Rinaldi | Jan 18, 2019 at 04:22 AM
AR: It's my chart, using data from BLS. I put it up on the JMP Public website here: https://public.jmp.com/packages/Visual-Exploration-of-Unemployment-Data/js-p/5c3a727dda21e00c0c44ac1c
I believe if you create an account, you can access the underlying dataset.
Posted by: Kaiser | Jan 20, 2019 at 01:41 PM