October has been another month with strong signals from unemployment insurance claims, which continue to push to new lows in both initial and continuing claims. Here is my favorite chart on the topic, with the comparison of total unemployment and unemployment insurance. The forecast is based on a linear decline in very long term unemployment (VLTUE), so the recent divergence of total unemployment and insured unemployment suggests that VLTUE has stagnated. This is corroborated by my analysis of unemployment durations, which also suggests that the decline in VLTUE has stalled. In addition (see the second graph), normal unemployment (after subtracting the VLTUE) also remains elevated compared to insured unemployment.
There is a tendency for unemployment to settle in near 2.8*insured unemployment, except for cyclical shocks when insured unemployment briefly rises. There was one brief period in 1993-94, while emergency unemployment insurance (EUI) was in place, that total employment briefly remained elevated. And, throughout the post-recession 80's total employment deviated from the expected level. Both periods occurred when unemployment rates were higher than they are now.
Here is another way to look at the numbers. Graph 3 is continued claims (which cover up to 26 weeks) as a percentage of unemployment less than 26 weeks. Here we can also see the ratio of claims to total unemployment falling below the typical bottom range, toward the atypical range of the 1980's.
I don't have a theory regarding this deviation from typical trends, and so I don't have any firm clues about the persistence of this deviation. The clear stagnation in residential construction is an obvious source of possible persistent unemployment, so that I wouldn't be surprised if this pattern proceeds along with developments in housing. This speculation comes from other evidence, though, and not directly from the numbers here.
The outline of unemployment last month was:
4.3% Unemployment proportional to insured unemployment
0.7% Normal unemployment above the expected level
5.9% Total Unemployment
Insured unemployment has declined again this month, so that normal unemployment would be expected to decline nearly 0.1% (from 4.34% to 4.26%). I don't know what to expect of the current 0.7% deviation in normal unemployment. And VLTUE had been declining by about 0.05% per month, but has leveled off. As we can see in graph number 4, the question remains whether this is noise or a new trend.
JOLTS are also mysterious. Job openings have been through the roof, but hires and quits have stagnated in recent months (JOLTS have been reported through August). So, are the hires and separations hitting a plateau because we have full employment for 99% of the labor force while 1% of the labor force remains marginalized? Or are hires are separations simply at a temporary low point around their upward trends? The drop in hires and quits in August was sharp, and it didn't seem to be corroborated by other data, so I suspect that we will see a recovery, and possibly even a strong upward revision, especially considering the strong hiring reported in September flows.
I expected to look at the numbers and predict another gap down. But, I think the recent deviations from my forecast may be changes in trends and not just noise. There is a chance we could see another gap down to 5.6-5.7% this month. But, I'm not as confident as I have been in previous months. The unemployment rate may remain in the 5.8-5.9% range.