Ernie Tedeschi: "Jason... 3 points: The 1st is: if... prime-age... EPOP... better... than the headline unemployment rate, then why haven't wages accelerated more, given those measures have improved faster than U3?...
...This framing... starts with a U3 wage Phillips Curve and then assumes that we can then linearly map that into, say, a prime-age EPOP wage Phillips Curve. But U3 hasn't been a decent predictor of wage growth in this cycle....
Jason's second point... is concern over positing a relationship between a stationary variable like wage growth and an apparently-nonstationary variable like prime-age EPOP.... This is an important concern because if prime-age EPOP were truly nonstationary in recent years, then the tight relationship between it and wage growth could be spurious and not hold in future years. One condition under which the level of prime-age EPOP would be stationary would be if its trend / potential level were generally flat. That doesn't strike me as a crazy assumption if we're talking about recent history, say after female LFPR had mostly peaked in the 1990s....
An additional point about this... also speaks to Jason's third concern, which is that a wage Phillips Curve relationship ought to speak to the multitude of labor market phenomena we've seen over the past 25 years... like incarceration, deunionization, part-time work, aging of the population, educational attainment, female LFPR, etc.... Relating prime-age EPOP to ECI post-mid-1990s addresses many of these broad issues. Female LFPR had largely peaked by then, and inflation expectations stabilized around 2-ish after dropping in the 80s. On the wage side, ECI controls for occupation, industry, unionization, and part-time status. It's not a perfect measure by far (for one, its sample size is small), but it addresses many of the big concerns. On the EPOP side, I would in principle prefer a measure that controlled for things like race, education, sex, age, and metro status. But... controlling for all of those factors yields a series that is basically just a static level shift from prime-age EPOP....
My suspicion here is that the "right" way to think about a wage Phillips Curve is by relating compositionally-adjusted wages to compositionallly-adjusted employment, but that compositional effects mostly just happen to wash out in prime-age EPOP.
Last thing I'll say is that I don't think this approach satisfactorially answers all questions. For example, you'd need a more sophisticated approach to go back before ~1990, when female LFPR was still rising. Also, the role, or lack thereof, of productivity needs more rigorous teasing out. Other issues like incarceration and nominal wage rigidities also remain. But in terms of recent wage growth dynamics, the empirics suggest that demographically-adjusted employment explains much more of the mystery than U3 or U6 can.... I don't think it should be surprising that U3 and U6 -- which are ratios against measures of the labor force—have lost explanatory wage power in a cycle notable for large nonparticipation margins of slack...
#shouldread
