By Dave Altig, executive vice president and research director, and Pat Higgins, a senior economist, both of the Atlanta Fed's research department
Just before Wednesday's confirmation from Fed Chairwoman Janet Yellen that the Federal Open Market Committee (FOMC) does indeed still see slack in the labor market, Jon Hilsenrath and Victoria McGrane posted a Wall Street Journal article calling notice to the state of debate:
Nearly four-fifths of those who became long-term unemployed during the worst period of the downturn have since migrated to the fringes of the job market, a recent study shows, rarely seeking work, taking part-time posts or bouncing between unsteady jobs. Only one in five, according to the study, has returned to lasting full-time work since 2008.
Deliberations over the nature of the long-term unemployed are particularly lively within the Federal Reserve.... Fed officials face a conundrum: Should they keep trying to spur economic growth and hiring by holding short-term interest rates near zero, or will those low rates eventually spark inflation without helping those long out of work?
The article goes on to provide a nice summary of the ongoing back-and-forth among economists on whether the key determinant of slack in the labor market is the long-term unemployed or the short-term unemployed. Included in that summary, checking in on the side of "both," is research by Chris Smith at the Federal Reserve Board of Governors.
We are fans of Smith's work, but think that the Wall Street Journal summary buries its own lede by focusing on the long-term/short-term unemployment distinction rather than on what we think is the more important part of the story: In Hilsenrath and McGrane's words, those "taking part-time posts."
We are specifically talking about the group officially designated as part-time for economic reasons (PTER). This is the group of people in the U.S. Bureau of Labor Statistics' Household Survey who report they worked less than 35 hours in the reference week due to an economic reason such as slack work or business conditions.
We have previously noted that the long-term unemployed have been disproportionately landing in PTER jobs. We have also previously argued that PTER emerges as a key negative influence on earnings over the course of the recovery, and remains so (at least as of the end of 2013). For reference, here is a chart describing the decomposition from our previous post (which corrects a small error in the data definitions):
Our conclusion, clearly identified in the chart, was that short-term unemployment and PTER have been statistically responsible for the tepid growth in wages over the course of the recovery. What's more, as short-term unemployment has effectively returned to prerecession levels, PTER has increasingly become the dominant negative influence.
Our analysis was methodologically similar to Smith's—his work and the work represented in our previous post were both based on annual state-level microdata from the Current Population Survey, for example. They were not exactly comparable, however, because of different wage variables—Smith used the median wage while we use a composition-adjusted weighted average—and different regression controls.
Here is what we get when we impose the coefficient estimates from Smith's work into our attempt to replicate his wage definition:
Some results change. The unemployment variables, short-term or long-term, no longer show up as a drag in wage growth. The group of workers designated as "discouraged" do appear to be pulling down wage growth and in ways that are distinct from the larger group of marginally attached. (That is in contrast to arguments some of us have previously made in macroblog that looked at the propensity of the marginally attached to find employment.)
It is not unusual to see results flip around a bit in statistical work as this or that variable is changed, or as the structure of the empirical specifications is tweaked. It is a robustness issue that should always be acknowledged. But what does appear to emerge as a consistent negative influence on wage growth? PTER.
None of this means that the short-term/long-term unemployment debate is unimportant. The statistics are not strong enough for us to be ruling things out categorically. Furthermore, that debate has raised some really interesting questions, such as Glenn Rudebusch and John Williams's recent suggestion that the definition of economic slack relevant for the FOMC's employment mandate may be different from the definition appropriate to the FOMC's price stability mandate.
Our message is pretty simple and modest, but we think important. Whatever your definition of slack, it really ought to include PTER. If not, you are probably asking the wrong question.