College Graduate Underemployment in the US in 2026: What Colleges Can Fix

College graduate underemployment in the US edged up to 42% in the second quarter of 2026, according to the New York Fed's recent-graduate tracker, updated August 6, 2026. Put plainly, about four in ten graduates aged 22 to 27 are working in jobs that don't need their degree, and their unemployment rate (about 5.6%) sits above the roughly 4.1% for all workers. Anyone who has walked a first-destination report into a cabinet meeting knows why this one stings. It shows up after commencement, which is exactly when a college has the least influence left.

So what can colleges fix? Most of it happens before graduation. Talent Disrupted, the 2024 study by Strada and Burning Glass Institute, shows that a graduate's first job tends to set the pattern for the next decade, and NACE's 2026 Internship & Co-op Survey shows employers converting interns into full-time hires at the highest rate in five years. Put those together and you get five fixes: count college-level placement instead of any job, get every talent through at least one internship (paid, ideally), add statistics or data coursework to the less quantitative majors, point employer relations at the sectors still hiring, and start coaching in the first year rather than the last.

Related reading: First-Destination Survey Knowledge Rate: How US Colleges Clear 65% in 2026 · Gainful Employment and the Earnings Test: What US Colleges Face in 2026 · Skills-Based Hiring in the US in 2026: What Employers Actually Changed

The 2026 numbers, before anyone spins them

Start with the source. The New York Fed's The Labor Market for Recent College Graduates tracker published its 2026:Q2 update on August 6, 2026. Recent graduates (ages 22 to 27, with at least a bachelor's degree) had an unemployment rate of about 5.6%, and their underemployment rate edged up to 42%. Over the same quarter, the Fed's data has unemployment averaging roughly 4.1% for all workers and about 2.9% for all college graduates.

The Fed counts a graduate as underemployed when they work in an occupation where fewer than half of the people doing it say a bachelor's degree is needed, using the Labor Department's O*NET survey data. A barista with a marketing degree counts. So does a bank teller with a biology degree.

Two details matter more than the headline. First, on the Fed's smoothed series, recent graduates have had higher unemployment than the workforce as a whole in every month since early 2021. For most of the years since the series starts in 1990 it ran the other way, and a fresh degree acted as a shield.

Second, the degree still pays when it lands. The same New York Fed tracker puts the 2025 median full-time wage for 22-to-27-year-olds with only a bachelor's degree at $60,000, against $40,000 for peers with just a high school diploma, and young workers without a degree had higher unemployment too (around 7.2% in the second quarter of 2026). That's the honest version to give a skeptical parent on a campus tour. The degree works. The step from campus into a college-level first job has narrowed, though, and missing it is expensive for years.

Why does the first job matter so much?

The strongest evidence comes from Talent Disrupted, a February 2024 study by the Strada Institute for the Future of Work and the Burning Glass Institute. It followed the career histories of graduates with a terminal bachelor's degree, and the findings read like a warning label:

• 52% of graduates were underemployed one year after graduating, and 45% still were ten years later.
• 73% of those who started out underemployed were still underemployed a decade on, making them about 3.5 times as likely to be underemployed as peers who started in a college-level job.
• Of graduates who started in a college-level job, 79% were still in one five years later.
• A recent graduate in a college-level job typically earned about 88% more than a high school diploma holder. An underemployed graduate earned only about 25% more.

Burning Glass Institute's follow-up, No Country for Young Grads (July 2025), found the same 52% for the Class of 2023 one year after graduation. It also found that unemployment among 20-to-24-year-olds with at least a bachelor's degree rose from 5.2% in 2018-19 to 6.2% in the two years through June 2025, and that in 2023 and 2024 the unemployment gap between young graduates and their less-educated peers was the narrowest in three decades.

Read that as an administrator and it's a timing problem. The window that decides most of this is the first six to twelve months after commencement, and nearly everything that shapes it (the major, the internships, the quality of the first applications) is settled long before anyone orders a cap and gown.

Is AI closing the entry-level door?

The best-known evidence that it is comes from Stanford's Digital Economy Lab. In an August 12, 2026 revision of Canaries in the Coal Mine?, Erik Brynjolfsson, Bharat Chandar and Ruyu Chen use ADP payroll data through June 2026. They find employment for workers aged 22 to 25 in highly AI-exposed occupations sits about 19% below where it would be had it kept pace with less-exposed peers. At the July 2025 data vintage that gap was 15%. Experienced workers show no comparable gap, and the adjustment runs mostly through fewer young hires rather than layoffs.

The New York Fed isn't convinced AI is the main story. In a May 14, 2026 Liberty Street Economics post, Richard Audoly, Miles Guerin and Giorgio Topa report that fewer than 10% of workers and vacancies sit in highly AI-exposed occupations, and that postings for junior and senior roles in those jobs have moved broadly in parallel. Their conclusion: AI may be contributing, but it is not the main driver of the hiring slowdown.

A June 1, 2026 post by Natalia Emanuel, Emma Harrington and Amanda Pallais points somewhere else entirely. They estimate remote work can explain 64% of the recent rise in unemployment among young college graduates, because new hires get far less feedback and mentoring when a team is spread out. One Fortune 500 company in their data went back to hiring inexperienced people once its offices reopened, except on distributed teams, where it kept hiring experienced workers.

Here's our view, and it's a slightly unfashionable one. For a provost, the cause debate matters less than it seems. Generative AI, distributed teams and the lean staffing habits Burning Glass Institute describes all push in the same direction: employers are spending less on training beginners, so they want graduates who arrive already half-trained. You can't fix an employer's org chart from campus. You can change how ready your graduates are when they walk into it, whichever economist turns out to be right.

Outcomes by major, straight from the Fed's table

Major still explains more of the gap than almost anything else. The New York Fed's outcomes-by-major table (2024 data, refreshed in February 2026) shows how wide the spread is for graduates aged 22 to 27. A selection:

Look at computer science. It has one of the higher unemployment rates on the list and one of the lowest underemployment rates. CS graduates tend to land a college-level job or sit out for a while; few of them drift into retail. Biology runs the other way, with modest unemployment but more than half working below degree level, partly because many biology graduates head on to medical school or graduate study (the same Fed table shows 64% of working-age biology graduates holding a graduate degree). Talent Disrupted makes the same point with different data: 53% of biology graduates held a college-level job five years out, compared with 74% of engineering graduates.

Talent Disrupted's authors suggest a fix that is modest and practical. Graduates in less quantitative fields should add statistics, data analysis or computer science courses alongside the major. That's a decision for the curriculum committee, not the career center, and it tends to stall in the space between the two.

Internships still do the heavy lifting

If you only fund one thing, fund this. Talent Disrupted (2024) found that, after controlling for gender, race and ethnicity, and institution characteristics, graduates with at least one internship had 48.5% lower odds of being underemployed. Five years after graduating, 54% of graduates without an internship were underemployed, against 41% of those who had one. In computer and information sciences the gap was 38% versus 20%.

The employer side lines up. NACE's 2026 Internship & Co-op Survey found employers converted 63.1% of their 2024-25 interns into full-time hires, the highest rate in five years. NACE's Job Outlook 2026 Spring Update (April 2026) has employers planning to hire 5.6% more Class of 2026 graduates, with organizations above 5,000 employees up 8.7%, after a fall survey that had pointed to roughly flat hiring. Hiring plans are recovering, and a good share of that hiring will flow through intern pipelines that were filled a year earlier.

The uncomfortable number is access. Talent Disrupted cites 2022 Strada research showing only 29% of college graduates had completed a paid internship. Think about who misses out. A talent who works part-time to cover rent, or who cares for family, often can't take an unpaid summer role at all, and on Talent Disrupted's numbers that pushes their odds of a college-level start the wrong way.

Picture two biology majors at the same regional public university, both with a 3.4 GPA. One spends the summer after junior year in a paid lab internship that a faculty member flagged in March. The other works retail to cover rent, because nobody mentioned the lab role until May. On the Fed's table they face the same field, where about half of recent graduates work below degree level. On Talent Disrupted's numbers, the first one has much better odds of landing on the right side of that line, and the difference started with an email in March.

What universities can change before graduation

None of what follows is new. What's new is the price of skipping it, which went up as employers cut back on training beginners.

1. Measure college-level placement, not just employment. A first-destination survey that counts a barista job as a positive outcome hides the exact problem described above. Code first destinations by the education level the occupation typically requires (the O*NET method the Fed uses) and report underemployment by program, every year, to the people who set budgets.
2. Set an internship target by program, and pay for the gap. Aim for every talent to finish at least one internship or paid work-based project before the final year. Where the good roles are unpaid, stipends from institutional funds or employer partners cost less than a cohort that starts out underemployed.
3. Put quantitative coursework into the less quantitative majors, through required statistics or data courses, or a short applied-data minor that history and communications majors can actually fit into a schedule.
4. Point employer relations at the industries that are hiring. NACE's spring 2026 update named information, engineering services, wholesale trade, construction and miscellaneous professional services among those adding hires, and the 8.7% planned rise at organizations above 5,000 employees is a good reason to court the big recruiters early.
5. Start coaching in the first year, and scale it. Talent Disrupted, citing NACE's 2022 Career Services Benchmarks Survey, puts the ratio at one career services staff member for every 2,263 enrolled learners (NACE's newer benchmarks are in our staffing-ratio breakdown). No hiring plan closes a gap like that on its own. Practice tools can take the repetitive work, mostly mock interviews and portfolio feedback, so staff time goes to the conversations that need a person. Talenlio is one practical option here: four AI agents (Portfolio AI Agent, Interview Coach AI, Challenges AI Agent and Job Hunter AI Agent) alongside a readiness dashboard with weekly reports and exportable data, live in about two weeks. The exportable data is the useful part for point one, since it gives you readiness numbers to set beside the placement figures.

A forecast for the Class of 2027

Our guess, and it is a guess: recent-graduate underemployment stays at or above 40% through 2027, even if headline unemployment eases, because employers' preference for people who can contribute on day one looks structural rather than cyclical. If that's right, the universities that pull ahead won't be the ones with the busiest career fairs. They'll be the ones that can tell their board, program by program, what share of last year's graduates landed a college-level job within six months, and what they changed because of it.

Could your institution produce that number by the end of this semester?

Book a walkthrough to see how the four agents and a readiness dashboard would work for one of your cohorts, or read how Talenlio works with universities.