There does not seem to be much incremental in today’s JOLTS report. The vacancy rate ticked up, but the quit rate ticked down. The latter may be a more objective metric, being based on actual employment decisions, rather than mere postings. To quit a job is a big deal. To post — or to forget to remove — a posting is not. In the old days, this distinction did not matter, as evidenced by the lower right panel in the chart below. But recently it has. Still, I am out of consensus on that point, so there is no value in pressing it on the day the JOLTS report prints.
Instead, I would like to offer a wild, fact-free speculation on how some dovish analysts seem to process the openings data. If I may be so bold, they seem to assume that the JOLTS data are a fundamental signal and noise at the same time.
Data are actual to April.
Here is why that seems to be the case. The dovish take is that wage growth has been temporarily elevated by the influence of job vacancies. The idea that the effect is temporary is novel, but the influence itself is based on conventional analysis. To wit, when the vacancy rate is high relative to (what is implied by) the unemployment rate, then the short term natural rate of unemployment is meant to be higher, because of matching frictions. So far, so coherent. Vacancies signal a real thing. (I don’t believe that for a reason I will allude to below, but it is coherent.) And closely related, the ratio of v/u is a better measure of labor market tightness than u alone. See the top right panel of the chart above for the v/u ratio.
But then the doves go on to say that the job openings figures will just decline. Why? Because the vacancies are probably just fake and need to correct to reality, I assert they believe, for your consideration. I admit it is totally unfair to read minds like that. But full disclosure: that is how I view it.
If the vacancy data were valid and did measure actual matching frictions, then the dovish analyst would have to explain why matching frictions are reliably going to decline, a couple years after the Covid disruption. But they don’t really make the argument. Right? They just assert it, knowing that the most directly testable part of the claim, the trend in vacancies, is easy because vacancies are obviously out of line with reality, not anchored in anything likely durable, and fated to decline regardless.
So, why does the Fed also pay so much attention to this? I have an opinion on that but will not share it today for fear of you really thinking I have the tinfoil hat on. Instead, I will make a less subjective point. It is possible that the trend in vacancies will end up just being high beta to labor demand, which is now in a down cycle as a function of the Fed’s fight against inflation, which is a serious and non-transitory issue. Vacancies, then, may eventually decline precipitously without telling us much about matching frictions. They will just be a noisy way of seeing that employment growth has slowed. Possibly. My own view is that it does not matter much.
Let me conclude by addressing what may be a possible source of confusion. There is a debate running now between the likes of Olivier Blanchard and Larry Summers on the one hand and some Fed associated doves on the other hand. The Blanchard-Summers take, to use shorthand, is that the rise of vacancies may signal a durable rise of matching frictions, in which case success in the inflation fight may require a prolonged significantly higher unemployment. Feel free to see some irony in that. The opposing view is that the vacancy rise is not really that interesting, even though there is no reason to challenge the validity of the vacancy data, in part because matching always slows when the labor market is tight and there is nothing unique in this episode. Effectively they are saying that u does a good enough job of picking up the tightness of the labor market and there is no need to pile on by getting distracted by v/u, which is too reductionist. These people are not making the case for the immaculate disinflation. They are just saying it is not so bad as Blanchard and Summers have suggested. Recently, Blanchard has conceded that the data have been on their side. But it is worth keeping in mind what this alternative debate is actually about. Separately, to the credit of both sides of this debate, neither is assuming contradictory things about how vacancies are determined.
