Thursday, October 2, 2014

The unemployment rate as a leading indicator for wage growth

 - by New Deal democrat

One of the most pronounced issues in the US economy generally is the stagnation of wages since the turn of the Millennium, and specifically the lack of wage growth since the current economic expansion began over 5 years ago.

Several months ago I undertook to explore whether there were one or more leading indicators for wage growth.  While I have not found any Holy Grail, the exercise has begun to bear fruit.

In this post, I'm going to explore nominal wage growth.  Because most employers probably determine raises without reference to the inflation rate, and partly because raises are usually only given once a year, nominal wage growth is much less subject to noise than real wage growth.  This makes it easier to distill signal from noise.  I'll deal with real wage growth in one or more subsequent posts.

Here's what I'll show in this post:

  • nominal wage growth makes a good "mid-cycle" indicator
  • the unemployment rate typically makes tops and bottoms months before nominal wage growth
  • the exception, following severe recessions, is where the unemployment rate peaks at a number higher than 8%.  In those cases nominal wage growth has bottomed when the unemployment rate has fallen to about 7.5% +/-1%.
This rule allows us to make useful predictions.  I'll test the rule with a forecast at the end of this post.

To begin with, I have been looking for some worthwhile "mid-cycle" indicators, i.e., data series that typically have an inflection point about midway through an economic expansion.  The YoY% change in nominal wages appears to be such an indicator.

Here are nominal wages from 1965-83:

and here they are from 1984 to the present:

In each case, there has been a sudden surge in nominal wage growth of about 2% close to the midpoint of the economic expansions.  In other words, there are no false negatives.  Further, there are only 2 false positives, in 1974 and 1976.  Since we had a turnaround and a 2% gain that began in mid-2012, that suggests that late 2012 was close to the midpoint of this economic cycle.

But more importantly, I want to be able to forecast whether or not we can expect improvement in wages going forward.  Since nominal wage growth appears to have an inflection point near mid-cycle, I wondered if a normally lagging indicator like the unemployment rate might be a worthwhile leading indicator for wage growth.  It turns out that it is.

In the below graphs, the unemployment rate is in blue, and is inverted, so that post-recession peaks show as downward spikes.  Expansion lows in the unemployment rate are broader affairs and show as rounded peaks.  Average hourly earnings are shown in red. I've normed the series for relative ease of viewing, particularly in the case of the inflationary 1970's.

Here is the unemployment rate compared with average hourly earnings for 1965-1983:

and here it is from 1984 to the present:

These show that the unemployment rate peaked before nominal wage growth bottomed in 5 of 7 cases.  Similarly the unemployment rate bottomed before nominal wage growth peaked in 5 of 7 cases. The worst contrary result was only -7 months.

Next, note that the unemployment rate is subtracted from 7.5.  This means that a declining unemployment rate from, e.g., 9% to 6%, shows as a rising line crossing 0 at the point where the unemployment rate is 7.5%.  Of the three cases (1974, 1982, and 2008) where the recession peak in unemployment was worse than 7.5%, nominal wage growth bottomed when the unemployment rate was 8.4% (1975), 7.0% and 6.6% (two months in 1986), and 8.1% and 7.8% (two months in 2012).  This gives us a rule of thumb that all it takes is for the unemployment rate to fall to 7.5% +/-1% to generate sufficient tightness for nominal wage growth to begin to rise.

As I said at the beginning of this post, the relationship is consistent enough to be able to generate useful forecasts.  Since we know that (1) initial jobless claims lead the unemployment rate, and these have still been improving over the last 6 months; and (2) nominal wage growth generally does not peak until after the unemployment rate has made its expansion bottom; then (3) over the next 6 months or so, we should expect nominal wage growth to continue to rise, and make a new YoY% high for this economic expansion.

That's my prediction.  We'll see if it pans out or not.