Unemployment fell sharply in the last two months, down to 9.0%. Much of that is due to a January reshuffling of the aggregate statistics. Little of that is due to the anemic job growth during that period. A good part, undiscussed, is due to people simply dropping out of the labor force.
U3, the rate we refer to as unemployment, the rate of people currently searching for a job, dropped 0.4% to 9.0%. If you look at alternative measures of labor underutilization then U5, which is U3 plus marginally attached and discouraged workers, workers who have looked for work in the past year, want to work but have given up looking for a job, dropped 0.2%. Since that includes the 0.4% drop of U3, that means discouraged and marginally attached workers increased ~0.2%.
However we can’t simply subtract them, since they have different denominators – U3’s denominator is the labor force, U5 is the labor force plus marginally attached and discouraged workers. In order to give a sense of the movement over the past two months, here’s a graph of both. Forgive me for dual y-axes:
[Data is from here and elsewhere on that page. This is a weird calculation, since marginally attached workers aren’t seasonally adjusted (though since it has people in it over the course of a year, it’s a relatively smooth data set) and I use adjusted for U3 since that’s the one everyone uses (same effect either way), and I arbitrarily set the marginally attached’s denominator to the labor force so to anchor it against the 9.0% (it’s the same dynamic if the denominator is labor force plus marginally attached).]
As you can see, in the past two months people are dropping out of the labor force as unemployment is decreasing. I believe this trend will continue, leaving people with an overly optimistic sense of how the labor force is doing. Annie Lowrey and others have covered the long-term hysteresis effects of unemployment, and as people exist unemployment this way it is difficult for them to get reacclimatized to the labor force if the economy ever picks up. I’m going to think more about the right way to represent this data and think through the consequences of it, but this is the worst way possible for us to get headline unemployment numbers down.