My first stab at comparing non-farm payroll employment declines in the current recession with those from the Great Depression was beset by a couple of flaws. One was that I was using seasonally adjusted numbers from the current recession while the figures from the 1930s were unadjusted. That was easy enough to fix. The other was that agricultural pursuits occupied a much greater share of the workforce in the 1930s than they do today, so nonfarm employment wasn’t quite as important an indicator as it is today. Sebastian Dartevelle, a scientist at Los Alamos National Laboratory and an occasional reader of this blog, devised a simple way to correct that: He came up with rough estimates of the total number of people who could work in 1929 (those 14 and older) and in 2007 (those 16 and over), and divided the change in nonfarm employment in each downturn by the appropriate labor-force number. Here’s what he came up with (updated by me to include the latest BLS data):
There’s still a pretty big gap between the Great Depression line and the Great Recession line, but it’s not nearly as big as on my earlier chart that didn’t adjust for the size of the overall labor force. The pace of job losses in the first 16 months of the Depression appears to be about twice as fast as in the first 16 months of this recession; in the earlier chart it was four times as fast. This isn’t the final word, mind you: Farmers and farm workers weren’t exactly fully employed in the 1930s (remember the Joads?). So while the previous chart exaggerated the Depression job disaster, this one underplays it. But I do think this version is probably closer to reality.
Losing the seasonal adjustment delivers some interesting results. The most obvious is that it makes the current recession’s job losses look a lot choppier, and thus harder to identify clear trends in. It also moves the employment peak that preceded this downturn back to November (it’s December when you use the seasonally adjusted numbers). January 2009 (month 14) now appears to have seen a sharper job loss than any of the first 16 months of the Great Depression. Not good. But what really matters is what comes next: In the Depression, the job losses continued for two more years after this chart ends. Let’s hope that doesn’t happen again.
Finally, just to make one thing crystal clear (because several commenters didn’t seem to understand it last time around): This chart shows the decline in nonfarm payroll employment. It has nothing to do with the unemployment rate.