Ever since the JOLTS (Job Openings and Labor Turnover Survey) came into existence, labor market analysts have looked at the ratio of unemployed to job openings. In fact, the BLS publishes that very metric monthly (see chart 1 here). My old EPI colleague Jeff Wenger used to call it the “musical chairs” number, as when it goes up (high levels of job seekers per job), you can envision a bunch of folks trying to get the one seat/job when the music stops.
The figure below replicates the BLS line—it’s the non-blue one (sorry, color blindness strikes again)—and it shows that while the number of job openings per worker climbed to almost seven during the downturn, it’s now pretty close to its pre-recession level.
But that’s partly because our measure of unemployment is biased down right now, for reasons I articulate here. And if that’s the case, then a measure like this will look better, where “better” means a tighter match between labor supply and labor demand, than it really is.
We’ve (CBPPs Bryann DaSilva and I) developed our own version of a more complete measure of the job market slack based on the work of IMF economist Andrew Levin. It’s built off of three inputs: the ‘unemployment gap’ (the unemployment rate minus the so-called “natural rate of unemployment”—the lowest jobless rates associated with stable inflation according to CBO), and gaps in labor force participation and involuntary part time work.
This result is a measure which arguably adjusts for the downward bias in the current unemployment rate and therefore paints a more realistic picture of the amount of slack in the labor market.
Turning back to the figure, unlike the standard measure, the more comprehensive alternative has not reverted to its prerecession levels. It has fallen considerably, but its current level is just about equal to its peak in the previous downturn.
In other words, when we correct for current biases in the number of unemployed, the jobless are still engaged in high-stakes musical chairs, with too few chairs when the music stops.
Source: BLS, our version of Levin’s slack measure (see text)