When the data disagrees (as it does not on whether nominal wage growth is rising or falling), we all try to find reasons. This was an interesting hypothesis from @paulkrugman but I'm pretty sure it's not right. Which leaves me looking for more ideas to explain the discrepancy.
One technical issue. Krugman's description is a little off. Atlanta is the median of the growth rates not the growth rate of the medians. Those are the same only under very special conditions (growth rates monotonic by income). In practice could be higher or lower than the mean.
More importantly, the Atlanta tracker shows wage growth rising rapidly for *every* sub-group including industry, occupation, education and wage group.
So this is not simply about low-wage growth rising, growth is rising for every group.
So this is not simply about low-wage growth rising, growth is rising for every group.
Often I've spent lots of time trying to reconcile divergent numbers and usually give up. In this case average hourly earnings come from an employer survey, Atlanta is based on an employee survey. They're somewhat different concepts (composition adjustment, median of growth rate).
So I'm left thinking that one really should be updating one's views based on Atlanta (something I probably should have done a bit more of before but the June jump in Atlanta was so large that it is even clearer now).
P.S. I now see the technical point (about the median of growth rates not the growth rate of medians) was previously pointed out by @ADetmeister (who really needs an avatar picture) and @RiccardoTrezzi.
Loading suggestions...