Monday, September 8, 2014

August 2014 Employment Report

I'm sticking to my guns here.  I still think we are due for a gap down in the unemployment rate.  There is a range of about 0.4% of potential reported unemployment in any given month, and I think the odds are that this month's reading of 6.1% (which was under 6.2% by a hair's breadth) was a product of riding the top of the statistical range of a declining rate, as opposed to a leveling off of the trend.

Here are the durations from the past several months.  All of the short duration levels rose.  Especially considering continued declines in insured unemployment, this movement seems inaccurate.  Generally, these durations level off after labor market recoveries, with a continued slight drift downward, and a lot of month-to-month noise.  That is almost certainly the case here.  The 0-4 week category, by itself, was about 0.1% above the average from the past 6 months.  The 15-26 week category had fallen to 0.90% last month, but moved back up to 0.95% this month.  We could conservatively expect this to decline to 0.80% before it levels off.  This month's reading was probably also nearly 0.1% above the trend for the category, although this depends partly on expectations for upcoming months.

Back in May, after the April report, I outlined a detailed forecast of how we might expect unemployment declines to play out for the rest of 2014.  Here is a graph comparing that forecast to the actual August reading.

The reading for August long term unemployment came in very close to expectations - both among the very long term unemployed, who are declining at a fairly linear rate, and the regular long term unemployed, whose movement should generally mimic the 15-26 week group, with a lag.  The chart shows the rise in the shorter durations.  But, the real wild card here is 15-26 week durations.

I would expect this to quickly settle 0.15% lower than it came in at for August, and I would also expect this to lead to another 0.2% drop in 27+ week durations, as those better performing cohorts move into the longer durations.  So, the difference between this being a trend versus noise could mean a 0.3%-0.4% difference in the rate at year's end.  As in April, I believe this month mostly reflects noise, so over the next 4 months, if this month did reflect noise, trends would lead to these reductions:

0-4 Weeks 0.10%
5-14 Weeks 0.05%
15-26 Weeks 0.15%
27+ Weeks, Reduction from August ST Levels 0.05%
27+ Weeks, Reduction from Forecast ST UE Declines 0.20%
27+ Weeks, Reduction from Very LT UE 0.20%
Total 0.75%

Some of the 27+ reduction from future short term unemployment improvements would probably not develop until 2015.  So, this basically puts us at 5.5% in December.  However, this hinges on seeing a continued downtrend in the 15-26 week duration bin.  If we don't see that downtrend, though, this still points to a 5.7% rate in December.

Here is the rate of exit from 15+ week unemployment.  As a confirmation of the notes above, even if this indicator levels off at 40% as we exit 2014, which would be low compared to past trends, it would point to a decline of about 0.3% in long term unemployment between now and December.  That, plus a pull back of about 0.2% in the shorter durations from the unusual levels this month, puts us at 5.6% in December.

In the meantime, continued claims continue to fall - dropping below 2.5 million this week.  This level of unemployment insurance claims has normally coincided with unemployment in the 4.5%-5% range.  Adding the 0.8% of very long term unemployed to that, again, puts us in the mid 5's.  That is based on today's continued claims rate, which still appears to be falling.  So, I continue to believe that the unemployment rate is an outlier.  In this chart, I have separated my estimate of the unusual long term unemployment from regular unemployment to help see the relative level of unemployment to insured unemployment.

Flows are showing the same odd story.  All the flows are showing the same general trends toward normalcy - except the Employment-to-Unemployment flow, which has been unusually high since April.  The difference between the blue line and the green line is the net flow from unemployment to employment.  This is at odds with the data from unemployment insurance from the past 4 months.

The next graph is a comparison of initial unemployment claims and the flow from Employment to Unemployment.  These series move together pretty reliably, except the Employment Flow data is much noisier.  This makes sense, since the flows data is a survey series and the initial claims data is a count.

Keep in mind that these are flows, so that deviations are additive.  In the last 4 months, the unusual rise in the EtoU flow has accounted for about 0.2% of added unemployment.  This flow is probably due for a sharp decline.

Maybe it's just that I said this before the April report, when unemployment was still at 6.7%, and I'm just running on a boost of testosterone from being able to pretend that I know what I'm talking about:
This month could be a real shocker, IMO.  And, I still think we might be tickling 6.0% or at least very low 6's by summer.
 I think a lot of the same dynamics are in place.  From December to March, we were pegged at 6.7% despite strong signs in other indicators, then it all came out in April, dropping to 6.3%.  Now, 4 months later, we barely scratched at 6.1%, despite strong JOLTS and insurance data.  We'll see how the month goes, but I suspect I'll be looking for a drop to 5.7-5.8% in September, which will set us up for a slower decline going forward from there.

No comments:

Post a Comment