A Lot of Noise Where Noise Really Shouldn’t Be

From Alhambra Investment Partners, by Jeffrey P. Snider

Since there isn’t any detectable acceleration in wages or earnings, the plateau across the JOLTS data dating back to various points in 2015 is therefore not likely to be related to the presumed end of labor market slack. Even if the unemployment rate were a valid and relevant interpretation of “full employment”, there would be no reason why businesses might not keep on hiring and advertising for hire anyway. Instead, the slowdown in JOLTS mirrors the wider slowdown across the economy.

This additional BLS data is statistically related by an internal benchmark to the CES, where the “implied” monthly employment change (Hires minus Separations) of JOLTS is fitted to some degree with the Establishment Survey estimate. Even so, there is variation in this data that may not, and often does not, appear in the CES survey.

What that means for any given monthly figure is that it does not necessarily match the mainstream payroll numbers. For February, the headline Establishment Survey estimate was relatively robust (revised down subsequently), a meaningless number on its own, while the JOLTS series particularly in terms of Hires was not. Though the Establishment Survey was reported as weak in March, that doesn’t immediately propose JOLTS estimates produced next month will show the same thing.

Instead, we can rely on averages for both which do show some small level of harmony, at least as far as the same direction. Since the two series are benchmarked to each other, it doesn’t necessarily amount to corroboration given that degree of self-reference. Still, the same shortcomings as seen by the payroll data are found in JOLTS, too.

Year-over-year, the change in monthly HIRES was negative again for the fourth time in the past five months. The 6-month average is now for the first time since 2010 also negative. Thus, while we don’t know the full extent to which the labor market has slowed down there is (still) every indication that it has to some considerable degree. This is a troubling indication because for JOLTS a year has passed (and for the Establishment Survey, a year plus a month) since the bottom of “global turmoil.” That should have been more than enough time for a meaningful turnaround to develop, even considering how the labor market often lags inflections.

Even though there is fuzziness to the interpretation of either, it is the same sort of fuzziness where in an actually improving economy there would not be perfect clarity but a whole lot more of it.

The rate of Job Openings portends something very similar. Going back to 2015, Job Openings (after the last benchmark revisions) have actually stalled out to a greater degree than the rate of Hires has. That leaves only the difference in 2014 for Job Openings to consider by how much the labor market has been affected.

As noted last month, that surge in purported “help wanted” activity calculated by the BLS is unmatched by the Conference Board’s Help-Wanted On-Line Index (HWOL). But that data series has been hit with what is surely some level of methodology breakdown since 2015, obscuring what might have happened in the jobs market after it.

The net result of all this data is that the surge in employment especially starting in 2014, the so-called best jobs market in decades, is questionable, and the degree to which the labor market has slowed since then is nearly impossible to truly unpack.

A full part of the reason for these difficulties springs from the same statistical problems that are in parallel to the real economic issues – the participation problem. The rate of Hires, for example, appears to have rebounded in full to match the prior cycle peak. But that view corresponds to an incomplete estimate of who might be reasonably labeled as unemployed. If we instead measure the level of unemployed by the Civilian Non-Institutional Population rather than the official (stunted) Labor Force, we find that the rate of Hires under that situation should be far greater than 5.3 million, perhaps as much as 1 million greater.

If the “recovery” in jobs is much weaker than it appears in the absolute, non-correlated figures, then what does that mean in terms of the implied improvement in the labor market in 2014? What does it mean about the slowdown after it? It might propose that the former was less than it may seem in these and other BLS estimates, more like HWOL, while the latter might be worse, though not likely as bad as the HWOL series might have it after 2015.

It is this great difficulty in being able to put together a cohesive and comprehensive view of the labor market that is truly important in all this. In many ways, what I am describing is too much noise – and that is a huge problem particularly when the payroll reports are taken as the definitive view of the US economy, especially the unemployment rate that is now completely isolated in its optimistic take. It was easy to take the labor stats at face value before the Great “Recession” because they were constructed for that economic condition, not the one we have now.

The closest thing to a conclusion that we might draw from all this is that the labor market is weaker over the past few years than in the years before them. If it was otherwise healthy prior to 2015, then that might not make much difference. If it was not, then noisy data is where the bond market or eurodollar futures have been for several years now (and likely one big reason why those markets translated only pessimism in 2014) and where they still, largely, remain.

Categories: Economy

Tags: , , , ,

%d bloggers like this: