A collaboration between Lewis McLain & AI
The unemployment rate says the job market is fine. A broader set of numbers — the people working part‑time who want full‑time, the ones who’ve stopped looking, and the millions of graduates in jobs that never needed a degree — tells a more complicated story.
Analysis · July 2026 · Data: U.S. Bureau of Labor Statistics via FRED · Structural estimates: Federal Reserve Bank of New York
Introduction
The United States unemployment rate is 4.2% — a number that, by the standards of the last half‑century, describes a healthy labor market. But there is a second number the government publishes on the same morning, and it is nearly double: 7.9%. That figure, called U‑6, counts the people the headline rate leaves out — and once you add a third, structural layer that no monthly rate captures at all, “underemployment” turns out to be one of the largest and least‑discussed features of the modern economy.
4.2% Headline unemployment (U‑3), Jun 2026
7.9% Broad underemployment (U‑6)
4.7M Working part‑time but want full‑time
~41% Recent grads in jobs not needing a degree
Two very different things called “underemployment”
The word gets used for two distinct problems that are worth separating, because they have different causes, different cures, and very different time signatures.
The first is labor underutilization — people who want more work than they can get. This is what the Bureau of Labor Statistics measures with its broadest official gauge, U‑6. It starts with the officially unemployed (roughly 7.1 million people who have no job and looked for one in the past four weeks), then adds two groups the headline rate ignores: the marginally attached (about 1.5 million who want a job and have looked in the past year but not the past month — including “discouraged” workers who’ve given up because they think nothing is out there), and the involuntary part‑timers (about 4.7 million who are working part‑time only because their hours were cut or they can’t find full‑time work). Add those in, and the share of the workforce that is unemployed or underused roughly doubles.
The second is a skills or credential mismatch — people whose jobs sit well below their qualifications. This is the barista with a bachelor’s degree, the engineer driving a rideshare. It doesn’t show up in U‑6 at all, because these workers are counted as fully, happily employed. The clearest measure comes from the Federal Reserve Bank of New York, which tracks how many recent college graduates work in jobs that don’t require a degree: the answer has hovered around 40% for years. Two out of five.
The first kind of underemployment is cyclical — it explodes in recessions and heals in recoveries. The second is structural — it barely moves with the business cycle, which is exactly what makes it easy to ignore and hard to fix.
A history written in two spikes
The BLS only began publishing U‑6 in 1994, but its short history is vivid. Through the late‑1990s boom it fell to around 7% — the tightest labor market in a generation. It drifted up during the early‑2000s “jobless recovery,” then came the two defining shocks of the modern era, both plainly visible below.
In the Global Financial Crisis, U‑6 reached 17.2% — better than one in six workers unemployed or underused — and, tellingly, it stayed in double digits for six years. The headline rate recovered far faster than the broad one; the 2010s were, beneath the surface, a decade of involuntary part‑time work and stalled careers. Then came the pandemic: in April 2020, U‑6 hit 22.9% — nearly one in four — the highest reading ever recorded, before collapsing at record speed as the economy reopened.
By 2022–23, U‑6 fell to about 6.6% — the lowest in the history of the series, a genuinely extraordinary labor market. The story of the past two years, and the reason this is worth writing about now, is that it has quietly climbed back to 7.9%.
The two moving parts
Two components explain most of that drift, and each carries its own signal.
The first is simply the gap between the broad and headline rates — the extra slack you only see in U‑6. It is the single cleanest picture of “the numbers under the number.”
Notice what this shows: the gap normally sits around 3.5–4 points and blows out in crises (it hit 8.1 points in 2020, when layoffs and hour‑cuts hammered part‑timers hardest). Today it is 3.7 points — squarely in the normal range. On this measure alone, nothing looks alarming. The concern, as we’ll see, is less about the level than the direction.
The second moving part is involuntary part‑time work — the people cut to part‑time hours who want full‑time. This is the most cyclically sensitive corner of the labor market, because employers trim hours before they cut heads. It is, in effect, an early‑warning gauge.
This series surged past 9 million in the financial crisis and stayed elevated for years — the “part‑time economy” that defined the slow 2010s recovery. It spiked to 10.9 million in April 2020, then fell to a record‑low 3.7 million in 2022 as employers scrambled for workers. It is now back near 4.7 million and rising — a soft but real signal that the labor market is cooling from the inside out.
The part that’s genuinely huge: the graduate mismatch
All of the above is cyclical — it rises and falls with the economy. The structural story is different, and arguably the bigger one. Even at the tightest labor markets on record — 2019, 2023 — roughly 40% of recent college graduates have worked in jobs that don’t require a degree, and a large share never climb into one. That number has been remarkably stable for decades, through booms and busts alike. It is not a recession artifact; it is a permanent feature of how the American economy matches people to jobs.
Three forces keep it there. Credential inflation means more jobs now ask for a degree than genuinely need one, while the supply of graduates has outrun the supply of graduate‑level work. Scarring means the damage compounds: a worker’s first job predicts their earnings a decade later, so starting out underemployed can depress a lifetime of wages, not just a first paycheck. And now a third force is arriving — automation of entry‑level white‑collar work. The junior, routine, “stepping‑stone” roles that let underemployed graduates climb into degree‑level careers are exactly the roles most exposed to AI. If that ladder loses its bottom rungs, structural underemployment could stop being a young person’s temporary problem and become a permanent one.
Why it matters
Underemployment matters for four reasons that go well beyond any single statistic.
It means the headline rate overstates the economy’s health. A 4.2% unemployment rate implies a tight market with rising wages. But if millions are involuntarily part‑time or have drifted out of the labor force, there is more slack — and less wage pressure — than the headline suggests. This is precisely why the Federal Reserve watches U‑6 and participation, not just U‑3, when it judges how hot the labor market really is.
It wastes human capital. Skills atrophy when they go unused; a chemistry graduate making coffee is not building the expertise the economy paid to create. Multiplied across a generation, that is a real and permanent hit to national productivity.
It is an inequality story. Underemployment does not fall evenly. It lands hardest on the young, the less‑educated, Black and Hispanic workers, and — in the case of involuntary part‑time work — disproportionately on women. A single national average hides very different experiences underneath.
It is a leading indicator. Because hours get cut before jobs are, the U‑6 gap and involuntary part‑time tend to turn up before the headline unemployment rate does. When they drift the wrong way while U‑3 still looks pristine — which is roughly the situation today — it is often the first crack.
So how worried should we be?
Honestly, it cuts both ways, and it’s worth holding both halves at once.
The reassuring read. By historical standards, this is not a crisis. U‑6 at 7.9% is a world away from the 17% of the financial crisis or the 23% of the pandemic, and only about a point above the pre‑COVID best. Involuntary part‑time work, near 4.7 million, sits at the low end of its twenty‑year range. The gap between broad and headline unemployment is normal. If you were handed only these levels, with no trend and no context, you would call the labor market healthy.
The worrying read. The problem is the direction and the composition, not the level. U‑6 has climbed about 1.2 points off its 2022–23 low; involuntary part‑time is rising; and — the subtle part — the broad rate itself is understating the softness, because the labor‑force participation rate is falling. People who give up and leave the workforce entirely stop being counted as either unemployed or underemployed; they simply vanish from the numerator. So U‑6’s rise, real as it is, is happening even as some of the slack hides itself by exiting. Underneath, the structural graduate mismatch grinds on at 40%, and AI is now aimed squarely at the entry‑level jobs that used to be the way out.
The verdict. Cyclically, underemployment is a yellow flag, not a red one — the levels are fine; it’s the trend that deserves watching. Structurally, it is a chronic, under‑appreciated problem that deserves more concern than it gets, especially for workers under 30. The genuinely striking fact isn’t that underemployment is spiking — it isn’t. It’s that in an economy still described as strong, a broad measure of wasted and underused labor sits near 8%, involuntary part‑time work is quietly climbing, and two out of five young graduates are working below the level they trained for. That is the “huge” part: not a crisis in the headlines, but a large, permanent inefficiency hiding in plain sight.
What to watch
Four gauges will tell you which way this is breaking, and none of them is the headline unemployment rate: the U‑6/U‑3 gap (widening = hidden slack building), involuntary part‑time work (the earliest cyclical warning), the labor‑force participation rate (falling participation flatters U‑6 by hiding dropouts), and the recent‑graduate underemployment rate (the structural gauge, and the one most exposed to AI). Together they see the labor market the headline number can’t.
Sources & method. All time series are public U.S. data accessed via FRED (Federal Reserve Bank of St. Louis): U‑6 (U6RATE) and U‑3 (UNRATE) unemployment rates and part‑time for economic reasons (LNS12032194), all from the U.S. Bureau of Labor Statistics. The recent‑graduate underemployment figure (~40%) is from the Federal Reserve Bank of New York’s research on the labor market for recent college graduates. Data through June 2026. This is an analytical essay, not investment advice.