The JOLTS series is, in my view, a second tier data set made so by its computational connections to the CES. The Job Openings portion I believe has proven over the past few years why it is less reliable, surging so far out of character that it is all alone in suggesting huge economic positivity that even the main BLS figures won’t touch. The fact that Janet Yellen and the FOMC purported rely a great deal upon the estimate is importantly disqualifying in that regard.

That doesn’t mean, however, there is no use for the series at all. If the numbers have been literally unbelievable in their increase and acceleration, then it still may be significant where they no longer increase or accelerate. In other words, very much like the Establishment Survey, what might be important is relativity, not the number itself but rather the current month in relation to past months.

This is behavior we have already observed in JOLTS in the Hires data set. In October 2014, Hires hit a seasonally-adjusted level of about 5.1 million. The current estimate for August 2016 is 5.2 million, meaning that apart from the suspicious likely SNAP-related changes toward the end of last year the JOLTS view of hiring has largely flat-lined or plateaued. So even in what was a somewhat optimistic view of the labor market conveys that “something” changed in late 2014 such that what looked like healthier labor fluidity very likely stopped being so much so.

abook-oct-2016-jolts-hi-recent abook-oct-2016-jolts-hi

The upward bend in terms of Job Openings was far more severe, thus less believable, but it, too, seems to have expressed the same “something”, though a few months later than Hires. The seasonally-adjusted rate of Job Openings was about 5.6 million in April 2015 where the current figure for August 2016 is 5.4 million. Like Hires, Job Openings increasingly appear more like a plateau than the 45-degree ascent from before.

abook-oct-2016-jolts-jo-recent abook-oct-2016-jolts-jo abook-oct-2016-jolts-hi-to-jo

As with the Fed’s LMCI perspective of jobs and labor:

To start with, this year was supposed to feature significant acceleration more befitting the “full employment” rhetoric. The labor market potentially being compared, at best, to 1986 or 1995 is the opposite of favorable. During those periods the prior recovery slowed but only temporarily before taking off again (bubble or not). In 2016, there was no recovery from which to slow, only a slowing itself belying these kinds of cyclical interpretations. In other words, what is truly significant is that it does appear as if the economy slowed further after being merely bad to start with.

We have to conclude the JOLTS series largely agrees with the last part even if it, especially in Job Openings, described a different set of circumstances before the “rising dollar” intruded. In other words, the slowing seems to be universal, and thus more likely “real”, whereas these various labor statistics may disagree from what starting point or rate.

The usefulness of Hires and Job Openings is again more second rate as confirmation that the “rising dollar” and its “global turmoil” appear (as we thought) to have been sufficiently disruptive as to impact the US (and global) economy enough to register even across all the BLS statistics. It hasn’t led directly to recession, but that is a debate more so about what happened before this “something” weakness than the validity of the weakness itself (as well as how it doesn’t appear to be “transitory”). Even Job Openings, once way out of line for this economy, seems to have been knocked off that trend so as to be now much less belligerent toward broad corroboration.