Why the Jobs Market May Look Stronger Than It Really Is

Discover why demographics and labor force participation are changing how investors should interpret unemployment.

Key Takeaways

• The unemployment rate has become a less reliable indicator because labor force participation is shrinking. Demographic shifts and immigration policy are changing the denominator of the calculation.

• JOLTS data suggests hiring has essentially stalled despite headlines describing the labor market as resilient. Two consecutive months of net hiring arithmetic have pointed toward virtually no employment growth.

• Healthcare has been carrying much of the labor market's hiring momentum. Signs that job openings are slowing in that sector deserve close attention.

• Leisure and hospitality continue showing weakening hiring trends. Those industries often provide an early read on consumer demand and lower-wage employment.

• Investors should focus on revisions, labor force participation, sector composition, and wage growth rather than relying solely on the headline payroll or unemployment figures.

My Hot Takes

• The unemployment rate has become a backward-looking headline rather than a complete measure of labor market health. Structural demographic changes have fundamentally altered how the number should be interpreted.

• Labor market composition now matters more than aggregate job creation. Concentration of hiring in a single sector can mask broader economic weakness.

• The traditional "bad news is good news" market reaction to weak employment reports is breaking down. Sticky inflation has changed how the Federal Reserve responds to slowing growth.

• Markets remain overly focused on headline statistics instead of understanding the mechanics underneath them. Investors who examine the underlying data gain a better perspective on economic conditions.

• Economic indicators should be viewed as components of a larger dashboard rather than isolated signals. When multiple measures disagree, the divergence itself often contains the most valuable information.

• You can quote me: "The smartest investors won't watch the headline Friday–they'll watch everything underneath it."

Running on fumes. I drove an old Jeep for years with a fuel gauge that quit somewhere around year twelve. It read half a tank whether the thing was full or the engine was coughing on fumes. I never fixed it–I just learned to count miles instead. It was cheaper than a mechanic, and more reliable. Don’t get me wrong, I am not THAT cheap, but the ROI on any additional repairs on that FULLY depreciated asset was just not there–so said the finance expert in me. 😉 I think about that gauge every single time somebody quotes me the Unemployment Rate. We got JOLTS Job Openings yesterday, ADP this morning, Challenger tomorrow, Friday the big one. And the gauge is going to read simply–half a tank.

Here is what the Bureau of Labor Statistics actually put in front of us yesterday morning. Job openings in June came in at 7.359 million, down 178,000 from May's downwardly revised 7.537 million. The openings rate slipped by -0.1 to 4.4%. Hires ticked UP to 5.348 million. Quits rose to 3.232 million, which was the biggest month in a year and comfortably better than the street was looking for. Layoffs sat at 1.766 million, a rate of 1.1%, which is precisely where that rate sat back before any of us had ever heard the phrase "supply chain." The mainstream media narrative basically wrote itself. Steady. Balanced. One economist called the labor market “a duck on a pond.” Another called it “resilient.”

You know what? It IS resilient–in exactly the way my old Jeep’s fuel gauge stuck at half a tank is resilient. It doesn't move.

But there is a number in that release that didn't make a single headline I read, and it is sitting right there in the tables where the BLS puts the stuff it doesn't put in the summary paragraph. Hires in June were 5.348 million. Total separations–the folks who quit, got laid off, retired, or otherwise walked out the door–were 5.351 million. Subtract one from the other and you get -3000. Not negative three thousand as a rounding artifact of some obscure sub-series. Negative three thousand as the survey's own implied read on net employment change, using the exact arithmetic BLS describes in its own technical note as comparable to the payroll survey. May was -8000. That is two months in a row where the churn data says the American economy added nobody.

Now, I want to be careful here, because I have been doing this long enough to know how a good number gets abused. The JOLTS series is statistically aligned to the payroll survey, so that difference is a directional signal and not a competing forecast. It is not a prediction that Friday prints a negative. It is not proof of anything. What it IS, is a quiet disagreement between two government surveys about whether the hiring engine is turning over–and the payroll survey said June was plus fifty-seven thousand. When two instruments on the same dashboard disagree, you don't get to pick the one you like better. If you're smart and you don’t feel like walking to the nearest gas station in a suit–you, er…count miles.

Then there is my favorite piece of arithmetic in the whole release, and it is a beauty. The ratio of job openings to unemployed workers, which is the single number the Fed has been staring at for four years as its measure of labor market balance, rose in June to roughly 1.04. It was the best reading since January of last year. Improvement! Except job openings FELL by 178,000. Labor demand went down and the balance got better. That happens exactly one way: the denominator shrank faster than the numerator. It’s just math…silly! There were fewer unemployed people to divide by. MATH!!

And this, right here, is the disconnected gauge. It’s not a metaphor, my dear friends, it’s the mechanism.

The Unemployment Rate is a fraction. We all know this and almost nobody trades on it. For most of my career the denominator was boring, because it basically grew a little every year, immigration and demographics did their thing in the background, and you could treat the labor force as a fixed backdrop against which hiring rose and fell. That is no longer true. Between the immigration crackdown and the retirement of the baby boom, the labor force is not growing. Some very nerdy economists now put the break-even rate of job creation–the number of jobs the economy has to produce each month just to hold the unemployment rate flat–at something very close to ZERO. Read that again. 👈 Zero.

Which means you can have an economy that generates essentially nothing and a headline number that reads like full employment. And we already ran that experiment. It was called June. New Nonfarm Payrolls came in at 57k against expectations of a 115k–a solidmiss–and the unemployment rate went DOWN, from 4.3% to 4.2%. The reason it went down is that the Labor Force Participation Rate dropped three tenths of a point to 61.5%, the lowest since March of 2021, and Household Employment (still serious and accurate–just a different methodology) fell by 507,000. Half a million people simply stopped being counted. The gauge read half a tank because the fuel sensor retired after 13 years–it literally fell out of the tank.

So when Friday's number lands and somebody on television tells you the Unemployment Rate held at 4.2% (economists’ estimates), I want you to hear that for what it is: a fraction whose bottom half is being drained by policy and by demographics, reported as if it were a thermometer.

Underneath that broken gauge, the composition tells you something the headline can't. The single largest sector decline in openings last month was health care and social assistance, down 147,000. Health care has been the load-bearing wall of this entire labor market. In June it added twenty-two thousand jobs against a twelve-month average of thirty-eight thousand. In the ADP report, education and health services accounted for 48k of the 98k jobs created–half the private economy's hiring came out of just one bucket. When the sector carrying the whole thing starts reporting fewer forward orders, that matters more than a 100k headline.

Meanwhile, look at the other end of the conveyor belt. Leisure and hospitality openings fell another 86,000 and hires fell 87,000, to 858,000, the third straight monthly decline and the weakest reading since early 2025. Year over year, hires in that sector are down 174,000. That is restaurants, hotels, bars, and arenas. That is where the hourly work lives, where the second job lives, where the kid home from college works the summer. And it has been shrinking for three consecutive months while everybody argued about whether the World Cup was going to save the seasonal adjustment.

Which brings me to the Millers, because none of this is abstract at their kitchen table. Average hourly earnings are up 3.5% over the past year. Headline CPI in June was up 3.5%. That's it. That's the whole raise! It went to the grocery store and the gas pump and never came home. Gasoline alone is up some 27% year over year, which is what happens when your energy complex runs through a strait that somebody keeps shooting at. 🤷 And the traditional escape hatch, the one every one of us used at least once in our careers–quit, take the offer across the street, get the raise the boss wouldn't give you–is a narrow door when the quits rate is sitting at 2.0%. Mr. Miller isn't getting a raise. He's getting more hours.

Want to know whether that's true? McDonald's reported yesterday that U.S. sales growth stalled to 0.8%, and management pointed straight at gas prices. That's not a survey. That's the Millers voting with their wallet, in real time, at the drive-thru. It appears that only TikTokers are driving through these days.

So here is what I'll actually be watching for the rest of this week, and I'd suggest none of it is the headline. This morning's ADP: strip out education and health services and see what's left, because last month there wasn't much. Also watch the Weekly Pulse ADP publishes (my followers know I love this number–probably the most timely read on the “pulse” of the labor force), which showed 16,500 jobs per week in the four weeks through July 4th, down from 19,250–a fourth consecutive deceleration. Tomorrow's Challenger report: everybody will quote the job cut number, and everybody will be looking at the wrong half. Cuts have been cooling with June at 45,849, down 53% from May. The number that tells you something is HIRING PLANS, the announcements of jobs companies intend to add, which have been running at levels we haven't seen since 2010. Low fire is only half a story. Low hire is the other half, and it's the half that traps people.

And finally, Friday. Consensus is 100k and 4.2%. Four things to look at before you look at either one: the revisions to prior months, because the last report knocked 74,000 off April and May combined; payrolls excluding health care and social assistance, because if that's near zero then one sector IS the report; the Participation Rate, because a falling Unemployment Rate on falling participation is the sensor, not the tank; and Average Hourly Earnings against that 3.5% inflation print.

Now here's the part that should genuinely change how you read all of it. For roughly forty years, a weak jobs report was GOOD news for your portfolio. It was the “bad is good” reflex. Bad number, Fed eases, stocks up, bonds up, everybody home by five. That reflex is dead, and it died recently enough that a lot of very smart people haven't updated. The July FOMC held at 3.50% to 3.75% on a 9-3 vote, and all three dissents wanted to HIKE. Markets are pricing something like a 57% probability of a quarter-point increase in September. The thirty-year Treasury Bond closed yesterday at 5.17%, after touching its highest level since July of 2007 last week. A soft payroll number on Friday does not buy the Millers a cheaper mortgage. It does not buy you a rate cut. It just buys you a soft payroll number, on top of 3.5 % inflation, into a Fed that has told you plainly it is watching prices and offering less guidance than any Fed in living memory.

Weak growth plus sticky inflation plus a central bank leaning the other way is not a configuration most portfolios built over the last decade were designed for. That is not a forecast. That's just what's on the board.

Which brings me back to my Jeep. I sold it, finally. Somebody younger and more optimistic than me now owns it, gauge and all, and I am genuinely happy for him–he gets to start depreciating an asset I finished depreciating years ago, and he'll learn what I learned, which is that the gauge is not the tank. Sooner or later he'll be on a two-lane road at night, watching that needle sit calmly at half, doing the arithmetic in his head about how far it is to the next station. That arithmetic is the whole job. The needle isn't lying to him, exactly. It just stopped being connected to anything that matters.

We're going to get four employment numbers this week and about nine hundred headlines. Most of them will read the needle. Count the miles instead. 😉

YESTERDAY’S MARKETS

Stocks rallied broadly yesterday, with the S&P 500 gaining 1.79%, the Dow Jones Industrial Average adding 1.71%, and the Nasdaq Composite climbing 2.59%. Both the S&P 500 and the Dow finished at record closes, the S&P's first in two months and the Dow's second consecutive all-time high. Treasury yields eased modestly across the curve, with the 30-year down roughly 5 basis points to 5.17% and the 10-year off a similar amount to the 4.62% area. Palantir surged 29% following its second quarter results, Caterpillar gained 5.97% on record quarterly revenue, gold rose 1.01%, and crude declined on reports of progress toward reopening the Strait of Hormuz.

NEXT UP

• ADP Employment Change (July) is expected to come in at 65k, down from June’s 98k hires.

• ISM Services PMI (July) may have advanced to 54.5 from 54.0.

• Fed speakers today: Cook and Daly.

• Important earnings today: Phillips 66, Primo Brands, Carlyle Group, Circle Internet Group, CVS Health, BorgWarner, Dynatrace, Freshpet, Disney, Eli Lilly, Uber, Shake Shack, Kraft Heinz, Zimmer Biomet, Owens Corning, Galaxy Digital, Honeywell Aerospace, Block Inc, DoorDash, Dutch Bros, Axon, HubSpot, Motorolla Solutions, elf Beauty, NuScale Power, Realty Income Trust, Figma, Western Digital, Solaris, Sandisk, Etsy, Chime Financial, IonQ, McKesson, Expedia Group, Albemarle, Zillow, MetLife, AppLovin, Warner Music Group, and eBay.