How bad is it?

  We really need a labor economist.  Anyone?  Anyone?  Kaplan?

Calculated risk produced a chart based on a chart published by Barry Ritholtz.   It shows the peak-to-recovery job losses for the recessions back to 1948.  So far, it's been about as bad as 1981-82.  I thought it would be interesting to add 1929 and 1937.  How does the situation look compared to the opening months of the Great Depression and the so-called “Roosevelt recession” of 1937? 

To do that, I had to use the “total employment” series from the BLS Current Population Survey in order to make the post-1939 numbers comparable to the Depression-era data.  What I got looked somewhat different from the charts currently running around the internets. 

The big red line is the current downturn.

Qualitatively, the above picture gives the same picture as the other charts.  The current downturn is more-or-less tracking 1981, 1974, 1957, and 1948, all pretty bad recessions.  And since past performance does not predict future results, we are not promised the same sort of rapid recovery we saw from those four.

But two things surprised me, and are very different from the other data. 

First, most of the 20th century recessions turn out to be double-dips in employment terms.  Even 1960 has that pattern, although the second dip is not on the graph because employment rose above the pre-recession level before the second dip.  Only 1937 and 1980, which I didn't put on the graph, don't show a double-dip ... and in the case of 1980, it seems perfectly correct to lump it in with 1981 as one long triple-dip recession. 

The second is the length of time before employment levels recovered.  I had thought that 1990 and 2001 were unusually long in that respect, but once again, the Total Employment series doesn't support that.  If these data are right, then the only reason the 2001 recovery seemed “jobless” was that the downturn wasn't that severe ... and 1990 looks indistinguishable from other recessions.  (To be fair, these graphs only go out 24 months, and job growth in the 1990s remained sluggish well after total employment recovered.  But the point holds.)   

Are these stylized facts correct?  Were 20th-century recessions characterized by double-dips and long periods of low employment?  Or have I used the wrong data in the wrong way?  An inquiring mind would like to know.

Of course, I don't want to know that badly.  It is advantageous to have an obscure place on the web, even if it means that inquiries aren't always answered.

Comments

[Andrew R.]

I have a question on such a graph. Do we have similar data for systems that got sand in the gearbox of the financial sector? Or are the economies where that happened to different from the U.S. to be able to draw any comparisons?

[Noel Maurer]

I don't have time to pull up the data, but it exists. I believe that Mexico recovered rapidly from the 1994-95 collapse, at least in terms of total employment.

One difference that I can imagine (Dave?) is that many of them were small economies in a world economy that continued to tick along nicely. A large devaluation, frex, drove export booms in Mexico and Sweden.

Another, of course, is that the U.S. has famously flexible labor markets. Hire-and-fire at will, no required severance payments.

But those are hypotheses, based on nothing. It would be nice to see the employment time series before making pointless guesses to explain patterns that may not exist.