Is there a wage growth puzzle? Justin Wolfers says there is, and uses this picture:
to claim:
This puzzle isn’t entirely new, as the usual link between unemployment and the rate of wage growth has totally broken down over recent years.
The recent data have made a sharp departure from the usual textbook analysis in which a tighter labor market leads to faster wage growth, and subsequent cost pressures feed through to higher inflation.
But has the link between wage growth and unemployment "totally broken down"? Eyeball econometrics alone suggests reason to be cautious with this claim as the only deviation from the typical unemployment/wage growth relationship is the "swirlogram" of fairly high wage growth relative to unemployment through the end of 2011 or so. But is this a breakdown or a typical pattern of a fairly severe recession? While, it might seem unusual if you begin the sample at 1985 as Wolfers did, so let's see what the 1980-85 episode looks like:
Same swirlogram. Compare the two recessions:
Fairly similar patterns, although in the 80-85 episode there was more room to push down the inflation expectations component of wage growth. It would appear that in the face of severe contractions, wage adjustment is slow. Now consider the 1985-1990 period:
Notice that wage growth is stagnant until unemployment moves below 6% - past experience thus suggests that we should not expect significant wage growth until we move well below 6% (you could argue the response actually began at 6.5%). Thus, it is premature to believe that there has been a breakdown in this relationship. So far, the response of wages is exactly what you should have expected in light of the 1980's dynamics. Which leads to two points:
- I am no fan of Dallas Federal Reserve President Richard Fisher. That said, he did not pick 6.1% out of a hat when he said that was the point at which wage growth has tended to accelerate in the past. That number fell out of his staff's research for a reason and surprises me not one bit.
- There is a reason the Fed picked 6.5% unemployment for the Evan's rule. There was absolutely no chance that that would be a meaningful number as far as labor market healing is concerned.
Consider now the sample since 1990:
Note four points:
- Notice the minor "swirlogram" associated with the early-90's recession. Again, not a breakdown.
- After 1992, wage growth tends to move sideways until unemployment sinks below 6%.
- Since 2012, the relationship is as traditional theory would suggest, a point that is actually evident on Wolfer's chart as well. The R-squared on the regression line is 0.75. Although notice that again, as wage growth moves into that 2.5% range, it appears to once again move mostly sideways. No mystery - nothing we haven't seen before.
- Clearly, there is some noise in the relationship. You should be able to extract away from the noise and recognize that there is no sudden acceleration in wage growth.
Now let's take another step and consider the relationship between unemployment and real wages (note that the series ends in 2014:8 - we don't have the September PCE price data yet):
The period of the Great Disinflation was generally associated with negative real wage growth. The period of the mid-90s to the Great Recession was generally associated with positive real wage growth. The swirlogram of the Great Recession is again evident, but notice that as unemployment approached the bottom end of the black regression line (R-squared = 0.65), real wage growth actually accelerated before returning to trend. I now have additional sympathy for firms that have complained in the past two years that they could not push wage growth through to higher prices. It does appear that real wage growth was faster than might be expected given the pace of economic activity and, by extension, the level of unemployment.
Oh - and real wage growth has reverted to the pre-Great Recession trend - pretty much exactly where you would expect it to be given the level of unemployment. Honestly, this one surprised me.
Which suggests that labor market healing has progressed much further than many progressives would like to admit. Many conservatives as well.
Which also means a lot of people are not going to like this chart.
And before you complain that the all-employee average wage data holds some great secret that is not in the production and nonsupervisory wage series (I have trouble taking seriously any sweeping generalizations of the business cycle dynamics of a series we only have through one business cycle), here is that version:
Same swirlogram. Pretty much the same idea with wage growth heading right back to where you would expect prior to the great recession.
Bottom Line: Be cautious in assuming that this time is different. The unemployment and wage growth dynamics to date are actually very similar to what we have seen in the past. Low wage growth to date is not the "smoking gun" of proof of the importance of underemployment measures. There very well may have been much more labor market healing that many are willing to accept, even many FOMC members. The implications for monetary policy are straightforward - it suggests the risk leans toward tighter than anticipated policy.