Monday, October 6, 2014

September Employment Review

Well, I juuuust barely got my gap down - 5.94%, as reported - almost dashed by rounding.

Here are the durations.  Mostly what we saw this month, I think, was a correction in the noise among short term unemployment to finally reflect what we're seeing in the unemployment claims data.  Long term unemployment stalled.  So, it seems that we have seen most of the decline in unemployment that will have come from faster exit rates associated with the end of EUI.  Further declines will come from the 0.8-0.9% of the labor force that has been unemployed for more than 2 years and the slow continued improvements typical of maturing recoveries.  The very long term unemployed had been linearly declining for two or three years at a rate of about 0.05% per month, but this decline has stagnated for 3 or 4 months.  Here is a graph of my estimate.  This could be typical noise.  If it is noise, we might hit 5.5% by December.  If there is some sort of deceleration here, then we might hit 5.7% in December (though, of course, the reading of any given month can range over several tenths).

Here is a comparison of the September reading and my running December forecast levels:

The pessimistic scenario assumes that all the durations under 27 weeks will stabilize and that among the 27+ week durations, there will just be 0.1% of a decline among each of the more recently unemployed as the post EUI cohorts continue to fill this category and as the VLT unemployed level just reverts back to the 0.8% level that it has been hovering at since June.

The optimistic scenario reflects a slight decline in regular 15+ week durations and a reversion to trend among the VLT unemployed.

Here is my graph of unemployment claims and total unemployment.  It tells the same story.  This month, unemployment declined in line with the decline in unemployment claims, but there was no reversion back toward the long term norms, since the very long term unemployed have held steady.

The main question, going forward, is whether the VLT unemployed continue to exit steadily, pulling the unemployment rate slowly down similar to the previous 25 years' experience, or does unemployment stagnate as it did in the 1980s at a slightly higher level.  I'm not sure that anyone has any idea, since this cohort of unemployed workers doesn't have a precedent in post-WW II US.

Looking at flows, I'd say the trends still look good.  The new decline in the unemployment rate comes mainly from an increase in net UtoE (green line in Flow1), which I would characterize as mean reversion after a couple of months where this flow was weak.  Weakness in labor force came from flow from employment.  Flows from unemployment out of the labor force were right on trend.

The unemployment rate comes from the difference between the net UtoE flow and the net NtoU flow.  While my analysis above seems to suggest an unemployment rate that will begin to stall, when we look at flows in this way, with such strong trends in these net flows, it seems more likely that unemployment will continue to fall.

Looking at smoothed net flows over a longer time frame (Flow2 - note the color legend is different than in Flow1) net flows also continue to look strong.  Flows between unemployment and employment (red) still look good.

Changes in labor force participation are reflected in the difference between net NtoE (the green line in Flow2) and net UtoN (the blue line in Flow2).  The incredibly strong labor market in 2005-2007 is clear here.  While net UtoN is still a little high and probably will be for some time while VLT unemployment remains elevated, net NtoE has been very strong this year, hitting levels similar to the 2005-2007 period.  This is another signal of our bifurcated labor market.  While there are a large number of marginally attached workers and very long term unemployed, the rest of the labor market has signs of a mature and bullish business cycle.  Or, maybe as with so many issues, the demographics make this more benign than we imagine.  Possibly, with more older working-age people, there is simply more opportunistic, temporary employment and more marginal employment behavior.  There are clearly cyclical movements, but maybe the demographics inflate both the secular and cyclical trends.

Looking at the individual flows (Flow3), all trends are good.  The flows between E and U finally reflected a fall in EtoU, which has been signaled by unemployment claims for several months.  This decline should be sustained, and not just a single monthly deviation.  The U and N flows continue to decline to recovery levels.  And, the net flow from EtoN appears to mostly have come from and increase in the EtoN flow.  In other words, the decline in labor force participation this month was not a product of marginalized workers leaving the labor force.  It was from employed workers leaving the work force.  Workers coming straight from Not in the Labor Force to Employment remains fairly strong.

Finally, regarding wages, I disagree with the negative reactions that I'm seeing.  Clearly, inflation needs to be accounted for.  And, I think it is more accurate to think of wage levels and employment levels as both being affected by the broader economic context.  Real wage growth is right where we would expect it to be in this context.  Even if the unemployment rate is overstated, which I think it is, and the current labor market is similar to what would normally be about 5% unemployment, 1% YOY real wage growth would be fairly typical.  If inflation is going to keep reverting to around 1.5%, there is little reason to expect nominal wage growth of 3-4% in any labor market.  It could happen, but I don't see why the current wage trajectory would be disappointing.  I suspect the complaints about wage growth reflect social desirability bias, as much finance reporting does.


  1. Excellent blogging. Yes the SSDI disability, but also the VA disability, programs need to be considered. We are talking about 12 million people.

    The VA disability program is also disturbing, as you are talking mostly about working-age males. 3.7 million VA disability recipients---more than on the much-discussed unemployment rolls.

    My worry is that we are creating a cynical society in regards to work and public monies. Why not file for disability if you can get a check?

    As I always say, the federal government and the Fed need to reward the productive and tax the unproductive.

    I say QE-financed cuts in FICA taxes do the trick.

    1. Benjamin, here is a link I posted a while back to some great labor force charts from the Atlanta Fed. It would be great to see the one I copied into my post as quantities instead of proportions. The distortion in this graph understates the level of disabilities because they occur most in the age groups that have the smallest labor force rates.

      The long term decline in 45-54 year old male LFP, compared to the younger age groups, only amounts to about 600,000 workers. It wouldn't take a very large reduction of the SSDI and VA programs to make a pretty big change in LFP rates.

      Of course, the problem is that these programs set up self-destructive incentives. But the state of politics in the US today (maybe always. I don't know) means that the people who put the most effort into appearing altruistic would rather paint you as blaming-the-victim than actually change the structure of the programs, because it's so much more satisfying and so much easier. In a way, the fact that national programs like this are so difficult to run well (understandably - it's a difficult problem to address) is probably the reason why they get so bloated. It's so much easier to throw some money at them, blame the other half of team purple, and ignore them. The national government grows because it is ineffective, not in spite of it.