By Jeffrey P. Snider
I think it is entirely fitting, even useful in the long run, that December's payroll report was yet another perfect month; the fourth of 2015 by my unofficial count. There was absolutely nothing wrong with any of the components, at least in the raw job count estimated by various statistical regressions and adjusted with imputations (wages, not so much). The problem with surging labor statistics becomes obvious in the context of everything not labor-calculation related. Does anyone really buy that the economy shifted so much higher, and in December of all months?
The BLS is doing a very good job of leading its own creations to the point of irrelevancy, which is perhaps where they best belong in this "cycle." Everything to like and dislike about the payroll reports remained exactly the same; the labor force has been much better for the past two months, but doesn't come close to confirming what the Establishment Survey suggests; ditto for the Household Survey, which is still showing only about two-thirds the gains even after more than 700k employee growth November and December combined.
In circumstances where even the stock market is gravely concerned about much more than the Chinese financial condition, it just doesn't add up (still). Very much like PMIs, the payroll reports don't really imply what the mainstream seems to infer of them. Setting aside all trend-cycle concerns about whether or not the numbers the BLS presents are real jobs or phantoms (I'm as tired of writing about it month after month as you probably are in reading it; besides, there will be a much more useful opportunity to revisit the legitimacy of the Establishment Survey in about seven weeks when Q4 GDP is updated for productivity, setting up again, given another "perfect" payroll report, for more negative productivity nonsense), in closer examination of the payroll figures as they are show that they are neither extraordinary nor useful for determining future economic direction.
When last year the surplus of payrolls was defined as the "best jobs market in decades," it only counted in the very narrowest of definitions. It is easy to be fixated on the monthly change in raw payrolls or employment, especially since what is really a rather a wide statistical dispersion and confidence interval is taken as an actual count of the numerical creation of jobs, but geometric progression is a much tougher assignment. A monthly increase of 250,000 jobs in 2015 is not the same as 1995; population and economic expansion changes the base comparison. In percentage terms, the Establishment Survey might have been the best in 2014 since 1999, but it wasn't appreciably different than the years before it.
Using total calendar years for comparison, 2014 doesn't truly stand out in any way other than just a little better than the reduced growth we have come to expect. This past year, 2015, is just plain unspectacular by every count as it really wasn't at all different in far more important percentage terms than any of the prior years within the cycle. Growth of 1.9% last year was indistinguishable from the 1.6%, 1.7% and 1.8% in the years 2011-2013, respectively. None of those years would be now, nor were they then, in any way confused with potential "overheating."
By contrast, the cumulative comparison of the labor market using the Household Survey didn't suggest anything remarkable of either 2014 or 2015. Even 2014 failed to match the increase recorded for 2006 while both of the last two years are nearly identical to the gains estimated for 2012 (a year in which revisions belatedly discovered tremendous, sustained and still unappreciated broad weakness in the economy).
As is plain on the charts above, it's as if the payroll expansion from the unexceptional and in many ways woeful recovery from the dot-com recession was transplanted to the recovery portion of the "cycle" after the Great Recession. While it was in many ways painful then when it occurred, it speaks even more ill of the labor gains now given the disparity in the size of the recessionary hole that defined each cycle to begin with. In other words, the very distinct lack of symmetry even in the Establishment Survey during and after the Great Recession only suggests further just how nondescript and even unsatisfactory the current payroll expansion might be at the current level of estimation.
Overheating as a kind of qualitative economic measurement, as is clearly being used as some soft standard in setting monetary policy, would require something like that seen in 1983 and 1984, if not at least 1993 and 1994 (when Alan Greenspan first lurched federal funds "tightening" toward the intended "soft landing"). Those years stand out easily in the charts above for both direction and intensity, a qualitative comparison wholly unlike 2014 and 2015 no matter how many superlatives are attached to the absolute numbers in the monthly series. In 1993, for example, the Establishment Survey gained 2.6% compared to the overly lauded 2.3% in 2014, but that was followed by a 3.4% increase in 1994 compared to 2015's clear setback, just 1.9%. The estimates for 1983 and 1984 make the claims about the current labor expansion seem even sillier.
And those years heading into "overheating," such that that may even exist, were notably at the front ends of each cycle. The payroll and jobs estimates at the back ends of them, especially 1989 and 2000, tell us absolutely nothing about the economy to come. In both the Household and Establishment Surveys, the labor market seems normal one year only to suffer recession the very next. That actually makes a great deal of sense since jobs and labor are lagging economic indications.
That point even applies to full-time job estimates, which is remarkably more consistent in what to expect year to year. By the BLS's estimates, the economy gained 2.2% in full-time jobs last year, which is the same as +2.3% in 2006 and +2.2% in 2000. All of which suggests perhaps more of a warning than anything about reading too much into the averageness or steadiness of assumed labor market conditions. The comparison to 2000 may be especially relevant when payrolls were believed then, as now, to demonstrate an economy still growing and gaining despite financial turmoil in stocks gathering more economic attachments.
That leaves everything that has been hitched to the Establishment Survey and payroll reports in a really precarious state. I still question everything about them, including and especially the asymmetry and trend cycle, but it doesn't actually matter for the interpretations of the economy as it might stand right now. The payroll reports don't suggest what is claimed of them on their face; the numbers as they stand currently still don't add up to a surging jobs market, and even if they did that doesn't in any way imply that labor growth, as overall economic growth, will continue this next year.
If this is all Janet Yellen has left upon which to base her "overheating," with GDP no longer at all comforting, then she truly has nothing. After all, the six-month average monthly increase in the Establishment Survey is 229k, which is slightly less than the average of 233k now estimated (after many revisions) for March 2012. No overheating then as just six months later the Fed was initiating QE3.