Given the roller coaster markets have been on, it was only fitting that the most important macro data point on the planet printed a wholly befuddling upside surprise to cap another dizzying week for traders.
The headline nonfarm payrolls number came across as something of a nonsense print. The accompanying annual revisions made parsing the report all but impossible.
Statisticians and those accustomed to dealing with government data will calmly walk you through “the process,” but the wide range of forecasts and a week of preparatory damage control from White House officials plainly suggests nobody knows what to expect anymore.
Read more: Blockbuster Jobs Report Caps Dizzying Week For Traders
The inherent ambiguity should give policymakers pause. Yes, it’s important to be data-driven, but if the data can’t be trusted, what’s a hawk (or a dove) to do?
Obviously, I’m not suggesting the numbers are fabricated. That’s not what I mean by “can’t be trusted.” Rather, the pandemic has laid bare the limits of humanity’s capacity to “snapshot” a fluid situation. On the ground, the labor market is constantly evolving. And our “best” measures of inflation have a long history of being lampooned as woefully inadequate.
All we know with any degree of certainty is that i) the jobs market is hopelessly distorted by an acute supply-demand mismatch, and ii) price pressures are still percolating in the economy and there’s an outside chance inflation becomes embedded.
The pace of wage growth in leisure and hospitality slowed in January, but only slightly. It’s still off the charts and lines up well with job openings in the space (figure on the left, below).
The sector remains more than 10% short of pre-pandemic levels (figure on the right, above).
Bloomberg Economics’ Yelena Shulyatyeva noted that “neither the headline payrolls print nor the unemployment rate captured the full impact of the Omicron wave.” That’s because “workers who were paid for just a few hours during the early part of January would still be counted on payrolls and were also counted as employed for the household survey, from which the unemployment rate is calculated.”
If that’s the case, I’m not sure the rest of the soundbites emanating from various pundits on financial television Friday meant much.
At the risk of adopting an unduly dismissive cadence, this report may be as close to meaningless as NFP reports get with the exception of the wage data, which, even if you do think the jobs numbers “count,” would still be the biggest takeaway.
Average hourly earnings are near $32 now. The 5.7% YoY gain (figure above) would be anomalous outside the pandemic context.
“With companies desperate to hire and the biggest issue being the lack of suitable staff, wages are rising sharply and the Fed will respond,” ING’s James Knightley said, adding that “the narrative of intensifying labor market inflation pressures and strong employment growth when Omicron is supposedly depressing activity only makes it more likely that the Fed will embark on an aggressive series of interest rate increases.”
“The strength, including the revisions, appears to have been due in part to updated seasonal factors, and hourly earnings were probably boosted by the drop in the workweek, but the data are unambiguously strong nonetheless,” TD’s Jim O’Sullivan remarked, weighing in. The rise in the participation rate “suggests the strong labor market is attracting more workers back into the labor force [which] should ultimately bring wage inflation lower and suggests the Fed doesn’t need to panic about a wage-price spiral forming,” he added.
That latter bit introduces a bit of extra ambiguity into the already impossible task of divining what the report entails for the Fed’s reaction function.
Ultimately, the knee-jerk response in rates — and particularly any attempts to push the issue on 50bps for March — will probably be faded in the days ahead. Fed officials made it pretty clear this week that they don’t see much utility in “going big” out of the gate.
At the least, policymakers seem to think the risk/reward of a 50bps move as asymmetric, with the potential for a severe adverse reaction in markets outweighing the assumed signaling effect for the inflation expectations channel of a determined, angry bunch of hawks.
A big move isn’t going to have an immediate impact on expectations, and it won’t have any impact at all on realized inflation in the near-term. It could, however, torpedo equities and balloon credit spreads. The accompanying snap tighter in financial conditions would perhaps be welcomed — right up until it’s not, where that means right up until it’s acute enough to risk limiting the Fed’s capacity to tighten further.
As for any lingering confusion around the January payrolls numbers, don’t expect clarity. “It is what it is, but there will be skepticism [as] the BLS provided no explanation for why it’s so strong,” ING’s Knightley went on to say, calling the 151,000 increase in leisure and hospitality hiring “hard to fathom given restaurant dining is down more than 20%… based on Opentable data.”
My late wife’s first job was as a statistician for the state of Ohio’s labor bureau. She was responsible for producing the monthly employment and other statistics for the non-profit and government sectors. Altogether, six statisticians produced the final monthly employment data for the state, which was then submitted to the BLS. What everyone needs to know about this data is that the people working and non-working aren’t actually all counted. Selected employers are surveyed and the results of this sample are compiled and run through the current estimation equations. To keep data comparable over time the equations remain relatively stable and aren’t modified changed unless periodic tests prove they create significant errors. The final data are affected by the choice of organizations to sample, participation of those organizations, the estimation equations, actual changes in employment and other conditions, of course, and other factors. My wife did this job for four years and during that time none of the process mechanics were changed. Her qualifying exams showed her to be the most qualified statistician at her employment grade in any state agency for that time. Sampling theory alone will guarantee that a certain percentage of the monthly BLS data will be crazy. The fact that national data comes from the data from fifty states is also an issue. The more constituent parts that contribute to the final outcome, the less reliable the final overall data will be. We talked about this topic frequently and it was clear that she and her colleagues knew there were flaws and quirks in the data but had no power to fix parts of the system that were flawed.
Thanks for that post, Lucky One.
For years i had happily assumed that the ADP Payroll numbers were strongly based upon what their clients were doing. I finally learned that much of it is sampling based on that data.
Now you reveal that the Establishment Survey is just that, a survey. So much for “hard” data points.
If you’ll tolerate an old man’s tale, between years at B-school I learned to code in Fortran and built some models to predict cotton future prices. A valuable use of my time on many levels. The best thing I learned was to distrust modelling after my efforts revealed that Latin American soccer scores had a high R Squared with cotton futures.