Friday, September 3, 2010

Visual interpretation of the unemployment number

If there is any relation between mean duration of unemployment and the subsequent drop in the unemployment rate, we are in for a very slow decline (if any) in the unemployment rate:


The following charts show the relationship between mean duration of unemployment and unemployment rate a year later, showing that the relationship has changed in this recession:


Using the fitted equation for the period 2007:1 to August 2010, we have a forecasted unemployment rate of 7.5 percent in August 2011.  This depends on a big ceteris paribus (other things equal) assumption about civilian labor force participation rate and other factors behaving the same in the recovery as in the recession.  If we use the line fitted to the whole sample, the forecasted unemployment rate is 10.2 percent, with a looser fit.

So let us assume unemployment remains unchanged for the next year.  What does that do to “animal spirits”?  We are interested because the distinguishing characteristic of an NBER-defined slump or a business slump as understood by any practical person is the associated failure of confidence; which, as Keynes pointed out, is a palpable entity to the practical person that is adequately proxied by survey measures.  And visual examination of the record (below) shows that every recession has been accompanied one-to-one by a failure of confidence as measured by the venerable Michigan sentiment series, and our measure, A.  I have assumed

Regular readers will recall that our proxy for the determinants of confidence is the relation of the unemployment rate to an adaptation level:

A = - (U – UMEAN)/sigma(U)

where A is “animal spirits.”

We are still not close to a failure of confidence; which is not to say the economy isn’t in an over-indebtedness-induced “underemployment equilibrium” exacerbated by a highly unequal distribution of income and wealth.  The next failure of confidence will probably not occur until about 2013.  For this forecast I assume the unemployment rate will fall for a year, stabilize, and begin to rise again:


If we assume the unemployment rate remains constant for two years and then begins to rise after the next Presidential election, we have this picture:


In the case above confidence never recovers to “positive” levels, but adaptation level effects are evident in that, even as unemployment rises to new highs, the plunge in confidence is not worse than before.  Plus ça change, plus c'est la même chose….

With regard to Hussman’s views on the yield curve noted recently, the 1/10 yield curve is still 2.5+ percent positively sloped, so any logistic yield curve model will still show very low recession probability (ours, of course, also utilizes the A variable; and in any event the effect of ZIRP will not be a distortive on the 1/10 curve as on the 3-month; and changes in the slope are still highly indicative):


The shape of the current cycle resembles the 1970-1974 abortive recovery, only worse. 

At some point I will write up some insights on “America on the hedonic treadmill” and the adaptive necessity of resetting adaptation level “set points” to survive what’s coming, but everybody already knows that we’re going to have to get by with less.  The primary contribution today is to put the current cycle in context along the most important dimension, that of “animal spirits” or confidence.  The next collapse is still a ways off.

No comments:

Post a Comment