A James Alexander post
Scott Sumner, like many, was very taken with the latest average hourly earnings figures. The tick-up seemed to be breaking a very dull trend and taking the growth rate back up to the heady heights of 2007.
Unfortunately, the series he used – average hourly earnings for all private sector employees, seasonally-adjusted – only goes back to 2007. The much more traditional series for private sector non-supervisory and production employees goes back to 1965. This chart is perhaps a better guide to the long-run trends. It is far harder to spot any significant trend breakout.
Also, the bigger data set includes many more service sector employees where hourly earnings aren’t a particularly relevant measure of pay. There are 120 million total private sector employees, but only 20 million of them who are outside (above?) the non-supervisory and production category.
It should be a simple matter to check the average hourly earnings of the 20 million supervisory and non-production workers but it isn’t. There doesn’t seem to be a dataset released for that, indicating to me at least that it isn’t a robust data set. It’s not on FRED and it’s not even on the Bureau of Labour Statistics website. There are some (very old) discussions about how the hours for this segment are calculated, but not a lot of recent statistical material, if any. I think this makes it highly dangerous to rely on the all private workers average hourly earnings growth figures.
The seasonal adjustments may also be harder to do and often lead to large revisions. The non-seasonally adjusted version of Scott’s chart looks a lot less compelling.
In the absence of any guidance from the BLS that I can find I did a some simple calculations to find out the average weekly pay of the “white collar” employees. It is more than twice that of the “blue collar” staff. And the gap must be growing given the recent trends in the two indices in the first chart. If I was to derive a longer term chart for the growth in “white collar” average earnings per hour then I bet I would find a very volatile series indeed. The question is then: should this highly volatile, not particularly robust, sub-series drive monetary policy? Of course, not.
All this discussion is a bit beside the point given the horribly low and lower overall inflation figures and the negative growth in US$ base money. The Fed is still wedded to the output gap/Philips theory of inflation and it has been proven wrong – no respectable mainstream economist who believes in basic macro would have predicted a halving of the US unemployment number without a major rise in wages. They (and the central banks) need to rethink their macro, it is time for them to ‘fess up and stop worrying about such ad hoc concepts as “secular stagnation”. Just as the inflationistas have had to rethink their understanding of macro – I should know, I used to be one. Money drives nominal growth and wages, and good nominal growth allows strong real growth.
Historinhas 
Great blogging. The only thing I could add is that Sumner’s post on the matter appeared to me to not voice an opinion about what the Fed is contemplating, but rather to explain why. He has stated on many more occasions than I can count that the more reliable way to determine the stance of monetary policy is to look at inflation and NGDP growth, and in that regard, comparing changes in both over the course of 2015 thus far, it would be difficult to argue that a rise in the FF rate is either warranted or justified.
Great stuff. Unit labor costs have been drunk too. Also auto prices are unchanged over the last 18 years, even in nominal terms.
This is not the 1970s economy.
For what it’s worth I did create an implied index of average hourly earnings for the “elite” employees, going back to 2007. It is volatile, though broadly following the other index. The October 2015 figure looks like an anomaly. I expect it to be revised down substantially.