THE STASH NOVEMBER 6, 2009
You've probably already heard the grim news: The economy shed another 190,000 jobs last month, driving the unemployment rate up to 10.2 percent (though the job-loss total wasn't so far out of line with what economists expected). But here's the number I'm seizing on: 5.594 million. That's the number of people who've been unemployed for 27 weeks or more, a horrifically large number of people to be struggling through such oppressive circumstances. That's up from last month's 27-weeks-and-over figure of 5.438 million.
So how can this possibly be good news? It has to do with a point I blogged about earlier this week, which comes from a recent issue of Grant's Interest Rate Observer (not online). In the issue, Jim Grant and Dan Gertner argue that one big indication of whether the recovery will be "jobless" or "jobful" is when different durations of unemployment peak relative to the bottom of the business cycle. For example, as I explained in the earlier item:
[I]n the 1981-2 recession--which saw a rapid return to job growth--the number of people out of work less than 5 weeks peaked two months before the bottom of the business cycle (i.e., the end of the recession); the number of people out of work 5-14 weeks peaked exactly at the bottom, as did the number of people out of work 15-26 weeks; and the number of people out of work 27 weeks and over peaked 7 months after the bottom. It turns out that most postwar recessions/recoveries prior to 1990 followed this pattern: the various durations of unemployment tend to peak not too long after the bottom of the recession itself, and there's a fairly quick overall return to job growth.
On the other hand, the 1990-1 and 2001 recessions/recoveries looked very different. In the first case, the number of people out of work 5-14 weeks peaked 11 months after the end of the recession; the number of people out of work 15-26 weeks peaked 15 months after the end of the recession; and the number of people out of work 27 weeks and over peaked 19 months after the recession. (The corresponding peaks for 2001 came 5, 20, and 22 months after the end of the recession.) And, of course, these recessions corresponded with an agonizingly slow labor-market recovery.
Grant and Gertner argue that the dynamics of the current recovery, in terms of those unemployment peaks, look a lot more the pre-1990 business cycle than the post-1990 business cycle. Which is to say, they resemble the era of "jobful" recoveries rather than the era of jobless recoveries. During the current recession/recovery, the number of people out of work 5-14 weeks peaked the same month as the presumed end of the recession (May, according to Bloomberg), and the number of people out of work 15-26 weeks peaked a month later.
The only real hitch in Grant/Gertner analysis was that final category: people out of work at least 27 weeks. As Grant and Gertner conceded, the number of people out of work that long hadn't yet peaked. In fact, in the last set of data prior to today, the number actually jumped pretty sharply, from 4.988 million to 5.438 million.
Which is why the latest 27-weeks-and-over number is so encouraging. After jumping by 450,000 in September, it only jumped about 150,000 in October, suggesting that the number of very-long-term unemployed could be about to level off. If it does peak in the next month or two, that would really strengthen the Grant/Gertner case that employment growth could come a lot more quickly than we think, since it resembles what we observed before 1990, when "jobless" recovery wasn't yet part of the lexicon.
P.S. In fairness, the 27-week-and-over number actually leveled off in August, too, before shooting up again in September. So we won't have a great sense of whether it's stabilizing until we get another month or two worth of data. But the early indication is certainly encouraging.