Saturday, August 8, 2009

The Song of the Dead Black Swan

Historically initial jobless claims (i.e., applications to begin payment of unemployment benefits) have been a leading indicator of recessions when they go up, and recoveries when they go down. After peaking on a smoothed 4 week average basis at 659,000 in early April of this year, they delined slowly through June and then rapidly in July. The current reading is 559,000, or a decline of 15% from the peak.

But the BLS keeps statistics not just of initial claims, but also unemployment of less than 5 weeks duration, 5-14 weeks, 15-26 weeks, and finally, 27 weeks duration and up. And those statistics show a consistent pattern as recessions end: the initial claims peak first and are most leading; as you go longer out into the series, the data tends to get less "noisy" and smooths out, but it also becomes more and more lagging, until the 27+ weeks data always peaks well after recessions are over. In the graphs below, in which initial claims are in blue, unemployment for 5-14 weeks old are in red, an 27+ are in green, notice that blue peaks first, then very quickly afterward red, and with a significant lag, green.

The first graph is for the 1974-75, 1980 and 1981=82 recessions:


This second graph is for the 1991 and 2001 recessions and the "jobless recoveries" thereafter:


And this final graph shows our own recession:


During July, it was contended by some pessimists, including on a Receommended diary here, that the event was a "Black Swan"; that the decline only was because of early shutdowns of auto plants, so laid off autoworkers weren't putting in for benefits in July. It was further contended that once that seasonal effect wore off by about the third week of July, new jobless claims would rapidly rise again through their April highs. Unadjusted jobless claims were increasing, we were told, and were the "real" trend for those who wanted to tell the truth.

But instead of rising, non-seasonally adjusted claims (shown in red in the graph below) have declined by 200,000. Seasonally adjusted claims have also continued to decline on a smoothed 4 week basis (in blue):



Now let's go back to the previous graph. Note in particular the sharp decline in the red line in the graph of our current recession: that is claims that were between 5-14 weeks old in July. In other words, this is unemployment that began no later than June, and therefore cannot include the supposedly downward distortion by auto layoffs that didn't happen in July, and yet still steeply declined.. This data confirms that July's steep decline in new unemployment claims was the "real thing" and not just a statistical blip. In other words, the Black Swan is well and truly dead. It is not just pining for the Autralian outback. It has passed on. It is no more. It has ceased to be. It has expired and gone to meet its maker. It's a stiff. Bereft of life, it rests in peace. It's metabolic processes are now history. It's kicked the bucket, it's shuffled off this mortal coil, run down the curtain and joined the choir invisible. In short, THIS IS AN EX-BLACK SWAN.



But in its dying swan song, it has provided us with further information about what might be ahead. Because as I previously pointed out, unemployment data for the period of 5-14 weeks duration is still slightly leading, and also has less noise.

First of all, in the past, the most the 5-14 week average has ever declined without heralding the imminent end of a recession is 7.5% in 1970. It is now down 17.5%.

Secondly, go back and look at the graph covering the period 1974-82. Recoveries from those recessions, two of which were quite deep, were "V: shaped and featured strong job growth. Notice how the red line declined steeply and consistently.
Next, look at the graph of the 1991 and 2001 "jobless recoveries". Notice that the red line does not decline very steeply at all. In fact, the maximum it has turned down during the entire duration of those "jobless recoveries" was -11% and -14.5%, respectively.
Finally, look at the graph of our own recession. Note that the red line is again declining steeply, just as it did after the 1974-1982 recessions, already off 17.5%, just through the end of June. And because of those aforementioned auto plant closures and start-ups, it is almost certainly going to reported as having declined considerably further when the data is next released in a month.

In summary, the Song of the Dead Black Swan tells us not just that the recession is nearly over, but suggests that at least in the next few months, there could be actual job growth, more robustly than almost anyone suspects.