Eventually, if I keep trying hard enough, I will find a way to make these two lines converge:
In a comment to my last chart comparing job losses now to those in the early 1930s, plukasiak wrote:
the problem I have with this chart is that it conflates “job peak” with the beginning of comparable periods — but the 2008 recession was created (in large part) by the spike in oil prices, and employment was recovering when the financial/banking crisis hit, and the economy (and employment) went into free-fall just like it did as a result of the 1929 market crash…
You can pick and choose starting dates for a chart like this to make almost any point you want—the reason to go with peak employment as the starting point of the chart is it at least enforces a little consistency. Still, plukasiak is right that the current downturn acquired a very different character in September/October of last year. So I’ve charted job losses from September of both 1929 and 2008—in both cases the month before financial markets really tanked. An added advantage to doing it this way is that the lack of seasonally adjusted data becomes less of an issue, because now the chart is comparing October to October, January to January, etc.
And what do we learn? That the employment trajectory of this recession is looking more and more comparable to that of the early Depression—although mainly because January 2009 was a truly, spectacularly bad month for non-farm payrolls. It was bad enough in seasonally adjusted terms: 741,000 jobs lost. But when you look at the raw data it is truly staggering: 3.6 million jobs gone. There’s always a big decline in employment in January as all those Christmas-season jobs disappear—that’s why the Bureau of Labor Statistics adjusts the data for seasonal factors. But in normal times it’s a decline of 1 million jobs or so. This January accounts for about three-quarters of the total job loss since the fall financial panic. Since then, in February and March, the unadjusted job losses have been modest (172,000 in February, 58,000 in March), even while the adjusted job losses have stayed above 600,000 a month. Unadjusted employment usually goes up in February and March even in a recession, so there’s good reason for doing seasonal adjustments. But given the many ways in which this recession differs from the others of the post World War II era, I do wonder if the seasonally adjusted numbers that get all the play in the media come anywhere close to representing the true job-loss trajectory this time around.
