January 21, 2015
I thought that reporters had finally learned that monthly wage data are erratic and best ignored, but noooooooo, they apparently still believe that they give us real information about the rate of growth of wages. The immediate cause for complaint is a Morning Edition State of the Union fact check segment in which Scott Horsley told listeners that wages rose in November, but then fell in December.
As I tried to explain after the big wage jump in November was reported, the monthly changes are dominated by noise in the data. The 0.4 percent nominal wage rise reported in the month followed a month where the wage reportedly rose by just 0.1 percent and a prior month where it did not rise at all. Employer pay patterns in the economy as a whole do not change that much from month to month, it should have been obvious this was just noise in the data.
The wage drop reported in December should have further confirmed this. Horsley tried to explain the drop as a composition story, that we hired more people in lower paying industries. This is hard for two reasons. First, we added 48,000 jobs in the high-paying construction industry in December, compared to just 20,000 in November. We added only 7,700 jobs in the low-paying retail sector in December, compared to 55,700 jobs in November. In other words, the mix story seems to go the wrong way.
The other reason is the mix from month to month can only make a marginal difference in average wages. To see this, let’s take an extreme case. The gap in pay between the construction sector and the overall average is just over $2 an hour. By contrast, pay in the leisure and hospitality sector is about $10 an hour less than the average. Suppose that we saw 100,000 new jobs in construction and no other jobs in any other sector. This is equal to approximately 0.07 percent of total employment. This means this jump in construction employment would raise wages by less that 0.2 cents an hour. By contrast, the surge in restaurant employment would lower the average hourly wage by 1.0 cent.
In other words, even these extraordinary shifts in composition would have no measurable effect on the pace of wage growth. Anyone looking to explain month to month changes in wages by job mix is looking in the wrong place. The only responsible way to report on the wage data is to take averages over longer periods, the monthly changes simply don’t mean anything.
Comments