Data Watch: Employment & personal Income, 11/17/17
With this report, the Demography Sector is initiating a new feature: Data Watch. Our goal is to summarize breaking data trends by showing the range of relevant indicators, by offering a simple visual way to scan the main results, and by identifying what’s important and why. We start with Employment and Personal Income.
OVERALL EMPLOYMENT INDICATORS
The hurricane is over. The CES (NFP) employment change for October (+261K) was the best monthly read in over a year (Jul 2016). Combined with the upward revision in September (to +18K), after the first negative read since 2010, October comes as a relief.
Still, the longer-term trend remains deceleration. YoY, September was 2σ+ below the five-year (z score) trend; even Oct was 1σ+ below. On monthly moving average crossovers (MMAvg; see second chart below), the last 12 months are definitely trending below the last 24, and the last 3 a bit below the last 12.
Other observations:
- The core of the slowdown is the service-providing sector, which was doing so well earlier in the recovery. In services, we’re seeing 3σ’s on the deceleration in the NFP and ADP data—in MoM, in YoY, and in crossovers. Because services comprise 86% of all employment, they steer the ship.
- But give goods-producers their due: They’re not only growing faster than services, they’re also often besting the five-year trend and beating many of the crossovers. The employment component of the Fed manufacturing diffusion metrics (like Empire and Philadelphia) are just killing it. Ditto for employment in the manf ISM, which is an amazing 2 or 3σ over many of its crossovers. Say what you want about the Trump economy: It’s delivering on its promise to tip the workforce toward those who build stuff. (Just pray that the dollar stays low and oil stays high!)
- Average weekly hours remain below their 2014-15 peak and have recently shown little trend higher or lower. Thus, aggregate hours are pretty much following the overall employment trend.
- Average earnings (both nominal and real) are showing a modest rising trend. The Fed of course is watching this closely.
LABOR DEMOGRAPHICS AND INFLATION
In theory, the CES (establishment survey) and CPS (household survey) data are mostly tracking the same jobs. In practice, the two series often show a perverse see-saw relationship: When one goes up, the other goes down. That certainly happened over the last two months. In September, when the CES tanked, the CPS surged—and revealed a robust +0.3% rise in the employment/population ratio (with prime-age adults up by 2.5σ+). In October, when the CES broke out, the CPS took back most of its prior-month gains and the emp/pop ratio sank by -0.2%.
Looking ahead, we clearly see the emp/pop ratio reaching a ceiling, and that ceiling could come sooner than many realize. Over the recovery, CES and CPS employment reached their YoY rate peaks in 2014 and 2015 (at around 1.7 to 2.0%), coinciding with peak GDP growth quarters. This zip came mostly from shrinking unemployment and from a migration of “inactive” adults back into the labor force. Little came from growth in the working-age population (see chart below), which is now growing at a mere 0.3%... and shrinking.
Yet employment can grow faster than the working-age population only so long as the emp/pop ratio in each age bracket keeps rising. How long can this continue? Clearly, with the official unemployment rate (at 4.1%) at a boom level, well below its 2006 minimum, not much more can be expected from that quarter. More help from inactives? These too (as we shall see) are beginning to run scarce. And when they hit bottom, wages will climb the wall and real GDP will hit the wall.
What will force the Fed’s hand of course is inflation—and here we’re beginning to see some acceleration. The CPI is hitting YoY > 2.0% more often this year than over any of the past five years. In October, core CPI jumped to 1.77% YoY—and higher, to 1.83%, on 3MMA YoY. Other indicators are following suit. The Atlanta Fed's "Sticky CPI" (a more sophisticated "core CPI"), now at 2.2%, has been higher this year than at any time since 1H 2014. The New York Fed's "Underlying Inflation Gauge" (UIG) is higher this month than at any time since April of 2012. And the NY Fed's most elaborate version of the UIG, derived from a dynamic factor model across all types of economic data, is up to a stunning 2.96% in October, the highest month since August of 2006.
There's a reason Janet Yellen is growing talons and a sharp beak. The Fed is starting to worry about this. The rest of us should too.
Other observations:
- The employment-population ratio, which is still rising for prime-age adults (age 25-54), is starting to experience serious drag from the older 55+ population. As the big Boomer bulge ages, and as its center of gravity begins to move more into its late 60s, this ratio is getting pulled down by a rising relative number of retirees. Sure, Boomers are working longer than earlier generations. But they aren’t working forever. This drag will almost certainly grow heavier with time.
- Most of the survey-based indicators show a job market that is gradually heating up. Conference Board, U. Michigan, and NFIB series all show jobs getting easier to find (for seekers) and harder to fill (for employers). All of these are rising faster than the five-year trend, and most of these show positive crossovers. A warming job market alongside slower overall job growth is a classic hallmark of a late-expansion economy. Note, in the NFIB survey, that the most negative response was to the question about plans to increase future employment.
- Other signs of heating up include fewer workers working part-time for economic reasons and rising temp employment. Online ads, after a long decline since mid-2016, are starting to rise again.
EMPLOYMENT/POPULATION BY AGE AND GENDER
In October, the overall emp/pop ratio declined from 60.4% to 60.2%, essentially bringing it back to where it was in April. Relative to its 2010 bottom, the overall ratio is less than halfway back to its 2007 top. This leaves the impression that we have a long way to go. But this impression is misleading, because the overall rate has been suppressed since 2010 by the aging of very large Boomer (and early-wave Xer) cohorts into age brackets characterized by a much lower propensity to work.
The easiest way to view this so-called “composition effect” is to look at emp/pop ratios by age and gender. Here we see that, when each group is viewed separately, most of these groups are near, at, or above their 2007 levels. This is true for every age bracket of women, with the partial exception of 45-54 year-olds (a bit more than one point down). And it’s true for both genders over age 55. Among prime-age men, yes, there is a shortfall—especially among Millennial males (under age 35) and to some extent among Xer males (age 35-54).
Is it likely that a lot of these men will yet return to active employment? We argue that it is not likely. (See: “The Missing Male Worker.”) In which case, as noted above, we may be closer to the employment ceiling than many suppose.
PERSONAL INCOME, JOLTS, & WAGE TRACKER
The salient feature of the most recent monthly NIPA personal income numbers is their extraordinary stability. Note how many white cells there are in the table below—showing that everything here is pretty much humming along its five-year trend track. Total wages may be slowing a bit, but that’s been made up by accelerating capital income. PCE popped in September along with retail sales as the personal savings rate plunged (to its lowest since 2007). But all this extravagance may pull back a bit in October—again, along with retail.
The BLS JOLTS data show a pretty hot labor market with lots of job openings, few layoffs, and relatively few job seekers or hires relative to the number of openings out there. Remarkably, all four indicators related to openings are at least 1+ σ above trend during each of the past four months. The hurricanes in September didn't cause these indicators to skip a beat. We’ll have to wait until next month (JOLTS data are lagged a month) to see if October brings a fair-weather JOLTS acceleration.
The Atlanta wage tracker, which uses monthly CPS data to follow already-employed workers, is better at providing demographic context than measuring month-to-month rate shifts (since its calculations are all YoY three-month moving averages). As such, it is hard to know how to interpret the deceleration in nominal wage growth shown by the tracker over the past year. It may be influenced by the higher rate of churn, which “resets” each worker’s wage growth rate when they changed jobs and are rehired.
Overall, when looking at the breakdowns, the portrait is of a job market that transitioned most rapidly into higher gear back in 2014 and 2015—and that is still doing very well, but less so on a rate-of-change basis, today.
Job switchers ripped ahead of job stayers in 2016, but the edge has now dropped back a bit. The college-high school gap along with the high skill-low skill gap sank to its low point in 2015. (That helped give a powerful boost to the median-income family, along with plunging energy and food prices.) These gaps have since widened a bit, though they're still narrower than they were back in the dark days of the early recovery. Ditto with white-nonwhite and older workers versus youth. Nonwhite and young employers achieved their greatest relative gains in 2015 and 2016. Today, we’re seeing a bit of a relapse.
