Hi everyone. Welcome to the May Market Snapshot. Given the attention lately on the health of the labor market, that’s the topic for this month’s video; especially in the wake of what was—at least on the surface—a better-than-expected jobs report for April.
Let’s start with an important discussion around buckets? Buckets? What is she talking about, you ask? Pretty much every economic data point can be sorted into one of three broad buckets: leading indicators, coincident indicators and lagging indicators. In addition, there are subsets of leading indicators—in other words, certain data points that lead the common leading indicators. We’ll get to that in more detail shortly.
[Table for Claiming recession? is displayed]
Focusing first on leading indicators, initial unemployment indicators represent one of those “heads up” indicators that move in advance of broader economic trends. This table shows every official recession start point back to the late-1960s; in addition to the dates and levels of troughs in the four-week average of claims; as well as the percentage increase in claims leading into each recession’s start point. This is a perfect example of my favorite adage: better or worse tends to matter more than good or bad when it comes to economic data. Yes, the latest reading of 239k for the four-week average of unemployment claims is still low in level terms, but that’s up more than 25% from the trough last September.
[Table for Claiming recession? with average is displayed]
As you can see, the average increase in claims heading into recessions has historically been only 20%. Again, it’s the rate of change that matters at least as much as the level.
[High/Low Chart for Payrolls often high at start of recessions is displayed]
Moving on to coincident indicators, nonfarm payrolls is one such indicator.
[High/Low Chart for Payrolls often high at start of recessions with recession bars is displayed]
As you can see, payrolls are often still trending higher at the onset of recessions … in part because the NBER, the official arbiters of recessions, back-date recessions starts to at/near the recent peak in the aggregate data they track, including payrolls.
[High/Low Chart for Unemployment rate always low at start of recessions is displayed]
Moving on to one of the most lagging of all economic indicators, here’s a long-term look at the unemployment rate. I can’t tell you how often I hear something to the tune of “there is no way the economy is at risk of a recession with such a low unemployment rate.”
[High/Low Chart for Unemployment rate always low at start of recessions with recession bars is displayed]
As a lagging indicator, the unemployment rate doesn’t foretell recessions; in fact, as you can see, it’s historically been very near its low at the outset of recessions. In other words, a rising unemployment rate doesn’t bring on recessions; recessions ultimately bring on a rising unemployment rate.
[High/Low Chart for Survey divergences persist for Establishment survey is displayed]
An understanding of the relationship between payrolls and the unemployment is also important. The monthly nonfarm payrolls release comes from the Bureau of Labor Statistics’ Establishment Survey, which counts jobs.
[High/Low Chart for Survey divergences persist for Household survey is displayed]
On the other hand, the unemployment rate is calculated from the Household Survey, which counts people. Over the past 12 months, the Establishment Survey suggests that 4.25 million jobs were added; while it’s only 2.7 million jobs per the Household Survey.
The Establishment Survey also includes qualitative assumptions and adjustments; tied to both seasonality as well as what’s called the “birth/death” model, which estimates the birth and death of businesses. For what it’s worth, the birth/death assumptions—in particular—tend to overstate business births, and understate business deaths, at important inflection points down in the economy. In addition, the Household Survey tends to be more accurate around those same inflection points; with the Establishment Survey data ultimately subject to significant revisions to prior releases. Related to that, the April jobs report showed payroll growth that was stronger-than-expected; however there were significant revisions to the prior two months’ data. In fact, the downward revision to March’s data was about the same amount by which the April data “beat” expectations. Keep that in mind.
[High/Low Chart for Rising prime-age participation rate is displayed]
In another sign of at least a loosening up of what has been a very tight labor market, the prime-age labor force participation rate continues to move higher; and has actually finally surpassed the pre-pandemic peak. This, at least, is in keeping with what the Federal Reserve is looking for to help bring inflation down.
[High/Low Chart for Rising prime-age participation rate with recession bars is displayed]
Another component of labor market data tied to the ongoing inflation problem is wage growth, which is not yet approaching what might be considered the Fed’s comfort zone. As of April, average hourly earnings were slightly higher than the prior month. Keep in mind though that this is an average and is likely skewed lower by something called “mix shift,” which I’ll explain in a moment.
[High/Low Chart for Median > average for Atlanta Fed’s Wage Growth Tracker is displayed]
Because of this, we also need to look at median measures of wage growth; not just average measures and we can look at median courtesy of the Atlanta Fed’s Wage Growth Tracker, as it is called. You see a fairly meaningful divergence between these two measures recently. That’s because layoffs to date have been disproportionally-biased toward higher-wage jobs within higher-wage industries. So get back to the average thing and the ‘mid shift’ and what happens when you take a bunch of high numbers out of an average? The average goes down. A median measure is not biased as such; hence the spread between these two.
[High/Low Chart for Median > average for Average hourly earnings is displayed]
We can look at this in a little more detail here. This plots average hourly earnings against payroll growth for each defined sector. The mix-shift effect is in play if you look at the information and education/health services sectors, as two examples. Information as you can see is the highest-paying sector, but job creation was the weakest in April. Conversely, education/health services is in the middle of the pack in terms of hourly earnings, but saw the strongest job growth last month.
[Bubble Chart for Watch shifts in mix-shifts for average hourly earnings vs. monthly nonfarm payrolls change is displayed]
Finally, we can move on to the recent release of data from the JOLTS report—that stands for the Job Opening and Labor Turnover Survey. Job openings fell from nearly 10 million in February to less than 9.6 million in March. by the way the data lags other labor market data by a month that’s why we are talking about March data not April.
[High/Low Chart for Openings down a lot, layoffs up only slightly so far for JOLTS: Job openings rate is displayed]
Put another way, the job openings rate fell to 5.8%, keeping its swift move off the peak in place. The rolling over in job openings has been a key supporting factor for those hoping for a soft landing; but that wasn’t the case in March given that the layoffs and discharge rate rose sharply. If we continue to see this, it would confirm that the reduction in job openings is consistent with a recession, or hard landing not a soft landing.
[High/Low Chart for Openings down a lot, layoffs up only slightly so far for JOLTS: Layoffs and discharge rate is displayed]
So lots of details. Let me wrap up with a quick summary. Unemployment claims lead (with job openings and layoffs leading those), payrolls are coincident indicator (also subject to revisions and adjustment vagaries), and the unemployment rate lags. In keeping with what the Fed wants to see, the labor force participation rate is moving higher, but wage growth may still be a little too “hot.” Unique in this cycle is the “top-down” (or higher wage) nature of layoffs to-date. But of course, so many things are unique in this pandemic-afflicted cycle. My final thought is to remind viewers that it’s often the case that better or worse matters more than good or bad. Keep that in mind as you look at economic data to judge just where we are in this unique cycle.
[List of Takeaways is displayed]
[Disclosures and Definitions are displayed]