Something odd is happening in the U.S. job market.
The rate of unemployment is falling, and doing so pretty quickly (see first figure), to levels approaching the Federal Reserve’s estimate of full employment. That is, the Fed says full employment corresponds to a jobless rate of 5.1 percent and the unemployment rate last month clocked in at 5.3 percent.
Underemployment (aka “U6”), a broader measure of slack which counts involuntary part-timers (who want, but can’t find, more hours of work), has been falling even faster than the topline rate. Job growth is trucking along at a nice clip too, with employers adding about 240,000 jobs per month on net over the past year.
All good, right?
So why is the damn labor force participation rate (LFPR) stuck in the mud? And why aren’t we seeing much in terms of wage growth?
On the first point, the LFPR—the share of the population working or looking for work—topped out at around 66 percent before the recession and was last seen stuck at 62.6 percent, the lowest it’s been since the late 1970s. That’s partly a benign function of the aging workforce, as we baby-boomers age out of our working years.
But there’s also a far less benign cause: persistently weak labor market demand that’s led a bunch of working-age people to sit out the job market. One way to see this is to take the retirees out of the picture, as economist Elise Gould does in this post. The share of employed prime-age workers (those between the ages of 25 and 54) is climbing back to its pre-recession level, but it’s still only about halfway there.
Low labor force participation can make the unemployment rate a less reliable indicator than it otherwise would be. If you’re out of the labor force, you’re not counted as unemployed (because you’re not, by definition, looking for work). But if you’re one of those people who could get pulled back into the job market if you saw some welcoming opportunities, then you ought to be counted as contributing to the slack.
This “shadow slack” that doesn’t show up in the unemployment rate has a lot to do (as work by economists Danny Blanchflower and Adam Posen indicates) with the fact that, despite the trends you see in the figure above, wage growth remains mired at around 2 percent. That is, with the un- and underemployment rates falling so much, you’d be well within your rights to expect to see wages accelerating beyond the 2 percent anchor they’ve been pretty much stuck at for about five years.
To show you what I mean, I’ve employed a very simple model of wage growth based on underemployment (see data note below). As you see, the model tracks wage growth pretty well, generally catching turning points.
However, based on the assumption—one in which I’m pretty confident—that underemployment continues to decline at the rate you see above, this model predicts wages should be growing at a pretty good clip by now, which they ain’t much doing. Chair Yellen considers 3.5% to be the non-inflationary rate of nominal wage growth, which is around where this forecast ends up.
But this out is probably too optimistic re future wage growth because it leaves out the low labor force problem. Adding the labor force to the model (as Posen and Blanchflower found) makes it track recent flat-lining wage growth more closely, and, under the assumption that the labor force rate remains where it is right now, even while underemployment continues to decline, the forecast shows a later and shallower liftoff.
These simple models are all speculative, of course. If the job market heats up and more sideliners find work, the LFPR would improve, as would the chances for faster wage growth. But as Yellen herself said last month, “The lower level of the unemployment rate today probably does not fully capture the extent of slack remaining in the labor market–in other words, how far away we are from a full-employment economy.”
That sort of thinking, along with being correct, suggests that if the Fed must raise interest rates, they should do so slowly and carefully, with a close eye on those who still don’t have much to show from the current recovery.
Data note: The wage variable is year-over-year percent change in a combination measure of four wage/compensation series: average hourly wage of non-supervisory workers; Employer Cost Index, average compensation and average wage; and median weekly earnings for full-time workers. The wage series in the figures uses principal components analysis to weight the underlying component series. Copying Goldman Sachs (GS) analysts, I start with their “non-linear wage Phillips Curve” model, which regresses the wage series on underemployment and underemployment squared. For the projection to 2017, underemployment continues to decline at its recent rate until hitting 9 percent, which GS considers the full employment rate for this variable. For the LFPR simulation, I hold the labor force constant at 62.8 percent, its quarterly value since 2014Q2.