• The disturbing serial pattern of underestimating job growth continued...
  • ...and now tallies ~2 million missed jobs in 17 months
  • Bears pointing to the rise in the UR have it all wrong
  • The case for a June hike remains solid through careful interpretation of Fed-speak
  • US nonfarm employment m/m 000s / UR %), wage growth m/m %, May, SA:
  • Actual: 339 / 3.7 / 0.3
  • Consensus: 195 / 3.5 / 0.3
  • Scotia: 225 / 3.5 / 0.3
  • Prior: 294 / 3.4 / 0.4 (revised from 253 /3.4 / 0.5)
  • Two-month revisions: +93k

US job growth strongly beat consensus yet again (see table). As chart 1 shows, the pattern of serially underestimating job growth is rather unflattering to the community of economists and market participants that keeps wishing for weak reports but keeps getting routinely beaten on estimates of job growth. Consensus has underestimated payrolls in all but one month since the start of last year. The sum total of this underestimation bias amounts to missing about 1.9 million of the jobs that have been created over this time and so we’re dealing with a sizeable bias. That’s comparing consensus estimates to initial nonfarm payrolls estimates on release day. If instead we compare consensus estimates to current estimates of past job growth including revisions then the sum total forecast error has underestimated job growth by 1.6 million over this same period.

Chart 1: Consensus Has a Nonfarm Bias Problem

As a further sign of this bias, some of the commentary is playing on the fact that the unemployment rate edged up three-tenths to 3.7% in May and therefore somehow that’s what the Fed will focus upon. This is pure rubbish. The unemployment rate is derived from the companion household survey that showed a decline of 310k jobs and a 130k rise in the size of the labour force. Those numbers are pure statistical noise. Don’t take my word for it? Well then, listen to the Bureau of Labour Statistics (here) since they estimate that the 90% confidence interval for the household survey’s reported change in jobs is +/-600k versus +/-130k for nonfarm. Therefore we cannot conclude at this implied level of significance that the 310k drop in the household survey’s estimate of job growth is anything but pure noise whereas the 339k rise in nonfarm payrolls meets the bar for defining it as a statistically significant jump.

Chart 2 is another way of demonstrating this point. It shows that deviations like the 649k spread between this month’s nonfarm and household survey job growth estimates are not at all uncommon in both directions. Recall that earlier in the year the spread was in the opposite direction. The household survey beat nonfarm payrolls by +478k in December, 422k in January, and 360k in March and yet the bears were silent then. Imagine that, data not conforming to the narrative.... I'm sure some folks will spin the household survey in May as a darker signal than the payrolls report but there's simply no credibility to doing so.

Chart 2: Household Survey Deviations from Nonfarm Payrolls

In all, these are major reasons why the Fed pays much more attention to nonfarm payrolls rather than the household survey’s job growth estimates.

Across other details, revisions added 93k more jobs to the prior two months than previously estimated.

Private payrolls were up by 283k with government jobs up by 56k almost entirely at the state and local levels of government.

Goods producing sectors added 26k while services were up 257k in a continuation of the pattern that’s reflective of the modern day economy. There was decent breadth within services. Leisure/hospitality was up 48k, health care and social assistance added 75k, transportation/warehousing hired 24k folks, professional services were up 64k and not due to the temp employment category. Chart 3 provides the breakdown.

Chart 3: May Changes in US Non-Farm Payroll Employment

Wages were up by 0.3% m/m SA for a 4% m/m SAAR gain which is softer than the 4.8% distorted surge in the prior month but up from 3.3% before that. The three-month moving average for this measure is 4.1%. Chart 4.

Chart 4: US Hourly Earnings

US hours worked are not a plus for Q2 GDP. After a 2.3% q/q SAAR gain in Q1, hours are tracking flat (-0.1% q/q SAAR) in Q2 (chart 5). Unless there is a strong productivity rebound from Q1 (and there may well be one) then Q2 GDP faces downside risk.

Chart 5: US Total Hours Worked


I think today’s set of numbers keeps open the door for additional Fed tightening. The camp that wants to believe that Fed rate hikes are over and done with is distorting impressions about what is contained in recent Fed-speak and in a way that may be destabilizing to markets. The only voting FOMC official that I’ve heard clearly spell out support for a June pause has been Philly’s Harker, and even he sounds a tad data dependent. What follows is a review of recent Fed-speak.

For starters, did Chair Powell’s comments on May 19th really set up a pause as some are suggesting? Recall that he said “Having come this far, we can afford to look at the data and the evolving outlook to make careful assessments.” When Powell said he’s going to “look at the data” I heard that he’s going to, well, “look at the data” and then decide what to do on a meeting-by-meeting basis rather than having hard set guidance in advance of future decisions. That’s not necessarily a pause signal. Recall that Powell also said in that same event that they haven’t made any decisions about future moves and emphasized they will follow the data at the margin from here forward. Powell’s point is that since they are in restrictive territory now and facing lagging effects of their decisions they can afford to step back from explicit hike guidance and take the decisions one at a time depending upon the evolution of data and events like the debt ceiling agreement’s passage.

Jefferson’s comments the other day are also being exaggerated by markets and are not necessarily representative of the full range of opinions as noted below in a recap of positions taken by FOMC officials who vote this year. On balance, I think it will come down to the debt ceiling vote’s passage, a strong nonfarm report and the next CPI on day 1 of the FOMC meeting.

  • Governor Waller: depends on data.
  • Governor Jefferson: He said “A decision to hold our policy rate constant at a coming meeting should not be interpreted to mean that we have reached the peak rate for this cycle.” Emphasize “at a coming meeting” and not necessarily June. That’s much more ambiguous on timing a potential pause than the coverage I’ve seen that read what some folks wanted to read as a clear June signal which it was not.
  • Governor Bowman: Pointed to rebounding house prices as a challenge to their inflation goals. No comment on next decisions.
  • Governor Barr: No recent comments.
  • Governor Cook: No recent comments.
  • Philly’s Harker: “I am definitely in the camp of thinking about skipping any increase at this meeting, barring what we see in the next few days.”
  • Dallas’s Logan: Data dependent. “The data in coming weeks could yet show that it is appropriate to skip a meeting. As of today, though, we aren’t there yet.”
  • Chicago’s Goolsbee: Said he doesn’t wish to pre-judge June decision and hasn’t made a decision.
  • Minneapolis Prez Kashkari: “We may have to go higher from here, but we may not raise rates quite as aggressively and as quickly as we have over the course of the past year.” Ambiguous on timing and amounts if any. May pause, may not, rates may have to go higher.

As an add-on point, I don’t like the view that is creeping into markets that the Fed can halt and start and halt and start rate hikes in jerky fashion. First, I don’t have much trust for such guidance from a minority of Fed officials since it may just be about trying to manage markets by halting but holding out the threat of hikes they have no intention of delivering. Second, such an erratic path could damage market confidence and confidence in the economy through its effects upon the ability to plan. Third, I still think there is more merit to front-loading as long as the data supports it rather than a fits and starts path forward. If you’re going to pause, then signal a lengthy one as opposed to this nonsense of how well, maybe we don’t go in June, but July could be a live meeting. That’s just muddled thinking imo.