Wednesday, November 17, 2010

Puzzling Labor Flow Data

I had hoped to have something intelligent to say about the above data, but instead, after an hour or two of staring at it and fooling around in Excel, I find myself still seriously puzzled. So I'm throwing the graph up in hopes someone else is smarter or better informed than me.

The data are from the BLS, and represent the seasonally adjusted monthly flows between three groups of men (age 16 and over): the employed, the officially unemployed, and the not-in-labor-force (eg retired, students, drug-dealers, disabled, house-husbands, artists-and-writers-in-attics, etc).

The gray bars are recessions.

If it helps any, here are the differences between each of the color groups above (ie the net flows between each of the three groups).  I've taken a seven month centered moving average to mitigate the worst of the noise.

Some things I especially don't understand:
  • Why does the flow from NIL->Unemployed go up in recessions (and the following jobless recoveries), almost as much as the flow from Unemployed->NIL?  I can understand the second going up (discouraged workers), but there's actually little change in the net because the other way goes up as much.
  • Why do movements of NIL->Employed and Employed->NIL track each other so closely?  It would seem like the reasons why people would move in one direction (eg retiring) would have nothing to do with the reasons they move in the other (eg finished school).  Why do the two move together so much?


Philip Brewer said...

I can think of one partial explanation for the odd numbers, which is that temporary sojourns out of the labor force are only partially discretionary.

If there are a relatively fixed number of viable spots outside the labor force (post-doc positions, for example), then it may be that each person who moves moves E->NIL displaces someone who was outside the labor force, forcing them back into the labor market. (Post-doc funding not renewed, so they get a job at Walmart.)

I can easily imagine that even the most informal NIL positions might be relatively fixed. If the #1 son has been living in his parents' basement for a couple of years when the #2 son loses his job and has to move back home, the #1 son might go ahead and accept a low-paying job to avoid doubling the burden on the parents (and avoid having to share a room with his brother).

Gary said...

Indeed puzzling. I sort of think the "Not In Labor Force" label doesn't really exist. I wonder what the data would look like if you just had "employed" and "Unemployed" without the confusing in-between category.

Stuart Staniford said...

That's an interesting theory Philip. I'd always imagined that the size of NIL was more flexible, with the size of employment being fixed by the number of jobs, and unemployment being rather constrained by the duration of benefits. So I'm a little sceptical that folks in NIL can readily transition back to employment to make room for others. Still, I don't have a better story at the moment.

Michael Cain said...

Anecdotes are not data, but I know several households that contributed to the trend. All were basically in the same situation: one spouse was working, the other was NIL staying home to deal with children, an elderly parent, writing a book, working on a degree, whatever. When the working spouse lost their job, both started looking for work.

Alexander Ac said...

Hi Stuart, i know it is again OT, but here is another story on rare earth metals:

"We have a classic supply and demand crisis. Under normal conditions the global demand exceeds supply in about 2011," Professor Brent McInnes from Curtin University in Western Australia told an online briefing last week.

"In 2016, it's quite evident that the Chinese demand itself will exceed the global supply of rare earth elements."

Greg said...

Borjas doesn't have a lot to say, at least on a quick search. However:

Re question 1, there is a continous flow NIL->L (in labour force) from school leavers. In good times most of these get employed; in recessions many of them become unemployed.

Part of the answer to Q2 might be to do with major lifecyle decisions. When a recession hits, couples that were thinking of starting a family are triggered by job loss or a cut in hours. (Anecdotally, I have seen this myself.)

This idea is supported by the observation that the effect is smaller in the earliest recession, when women were a smaller fraction of the labour force, and by the delay between the onset of recession and the spike in E->NIL transitions.

Anonymous said...

My suspicion would be that the NIL <-> Unemployed movement may reflect single-income families with working wives and/or marginally self-employed men. When the economy gets tough, if a spouse loses a job it becomes necessary for the man to to back into the regular labor market in order to obtain the income and/or benefits for the household.

You might be able to test this by plotting this series against the female employment series to see if perhaps female employment dips prior to an increase in male participation. Unfortunately, I don't have time to investigate it myself right now, but the series are curious.