The Stock Market Just Hit an All-Time High. The People Who Built It Cannot Find Jobs.

Photo by Melanie Rosillo Galvan on Unsplash

The S&P 500 sits near 7,500, up over 25 percent in a year. The NASDAQ has gained more than 36 percent, driven by an AI infrastructure boom that technology companies are funding with hundreds of billions of dollars in capital spending.

At the same time, the unemployment rate for recent college graduates sits at approximately 5.7 percent, well above the national rate of 4.3 percent. Underemployment, meaning graduates working in jobs that do not require their degree, affects 41.5 percent of them, according to the Federal Reserve Bank of New York.

Recent graduates have a higher unemployment rate than the overall population. That sentence would have sounded strange for most of the past three decades. Between 1990 and 2018, college graduates almost always had lower unemployment than the general workforce. The degree was the safety net. It is no longer functioning as one, and the gap has now persisted for five consecutive years.

The most striking part of this story is not that it is happening. It is that nobody is entirely certain why.

The explanation everyone assumes turns out to be only part of the story

The instinctive explanation, the one repeated constantly in headlines and dinner table conversations, is that artificial intelligence is quietly replacing the entry-level white-collar jobs that graduates have traditionally relied on to start their careers.

Goldman Sachs Research examined this directly and found only a modestly positive correlation between AI adoption and unemployment among young workers aged 20 to 30 across industries. The information sector, the part of the economy most exposed to generative AI tools, has experienced an 8.5 percent employment decline since 2023. But that sector employs only 2.3 percent of young college graduates. A decline that severe, in a sector that small, cannot explain a national pattern affecting graduates across every field of study.

Researchers at the Economic Policy Institute found that the increase in the unemployment rate for young college graduates over the last two years appears to be driven primarily by an increase in labour force participation rather than a declining probability of having a job once you are in the labour force. The employment to population ratio for young graduates has held essentially steady over the past two years even as the unemployment rate climbed. More graduates are looking for work. The share of them actually finding work has not collapsed in the way the headline number suggests on its own.

This is an important and frequently missed distinction. A rising unemployment rate can mean fewer jobs are available. It can also mean more people are actively searching, which mechanically raises the rate even if the underlying job market has not deteriorated as dramatically as the headline implies.

The culprit that researchers increasingly point to instead

A study from the Federal Reserve Bank of New York, alongside independent research from the London School of Economics, identified a different and less politically interesting cause: the disappearance of in-person entry-level work.

Remote work can explain 64 percent of the increase in unemployment for all young college graduates between 2017 to 2019 and 2022 to 2024. The logic is straightforward once explained. Junior employees historically learned their professions largely through informal channels, watching how senior colleagues handled difficult conversations, absorbing unwritten norms, building relationships through proximity rather than scheduled meetings. Employers have grown reluctant to hire inexperienced workers into roles where that informal mentorship and absorption cannot happen, because a remote junior employee with no professional experience cannot benefit from osmosis the way an in-person one historically could.

This produces a specific and somewhat counterintuitive outcome: experienced workers who already know how to operate independently have benefited from remote work flexibility, while inexperienced graduates entering the workforce for the first time have been disproportionately locked out of exactly the kind of roles where they would have learned the most.

A separate and complementary explanation comes from Oxford Economics, which found that recent graduates, who make up only 5 percent of the total workforce, have contributed 12 percent of the overall rise in the national unemployment rate since mid-2023, more than double their proportional share. The Economic Policy Institute’s research points to a depressed hiring rate as the central mechanism: employers across the economy have simply slowed the pace at which they bring in any new workers, and the demographic that depends most heavily on being hired fresh into the labour force, recent graduates with no existing professional network or track record, absorbs that slowdown disproportionately.

Why this matters even if AI is not the primary cause

It would be reasonable to read all of this and conclude that artificial intelligence is therefore not a meaningful part of the story. That conclusion would be premature.

There is a slight positive correlation between industry-level AI adoption and unemployment among college graduates, even after accounting for remote work and hiring rate effects. The labour market turnover rate, meaning how frequently workers change jobs or get hired into new ones, has decreased sharply across the entire economy over the past few years, and the hardest hit by that broader freeze are workers entering the labour force for the first time, regardless of the specific cause of the freeze.

What the research collectively suggests is not that AI displacement fears are baseless, but that they are one strand within a more complicated convergence of forces: a structural shift toward remote work that has quietly eliminated the apprenticeship function entry-level roles used to serve, a broader economy-wide slowdown in hiring that hits new entrants hardest by definition, and a technology shift that is beginning to affect specific roles and industries even if its current measured impact remains smaller than headlines suggest.

These forces compound rather than cancel each other out. A graduate today is navigating an economy where employers are hiring less overall, where the remote roles that are available teach them less than in-person roles once did, and where the specific skills employers want are shifting under their feet due to AI tools reshaping what entry-level competence even means in many fields.

The market versus the workforce that built it

This is what makes the simultaneous record stock market and weak graduate labour market more than just an interesting coincidence.

The stock market’s gains are heavily concentrated in a small number of companies whose AI infrastructure investment is driving most of the index’s return. These companies are spending capital on the computing infrastructure and software systems that, whatever the precise mechanism, are part of a broader economic shift that is making certain kinds of entry-level human labour less necessary or less valuable, at exactly the same moment that those companies’ stock valuations are rewarding investors handsomely.

The investors benefiting from that NASDAQ outperformance are, overwhelmingly, people who already own significant equity holdings: established professionals, retirees, institutional investors, the wealthiest households who hold a disproportionate share of all stock market wealth. The graduates struggling to find their first professional role are, by definition, people who have not yet had the chance to build any equity holdings of their own.

The market is not lying about the strength of corporate earnings or AI-driven productivity gains in specific sectors. It is also not measuring whether the economic structure generating those gains is creating proportional opportunity for the people entering the workforce underneath it. Those are two different questions, and the current moment is a clear illustration of how far apart their answers can be.

What this means if you are navigating it directly

For graduates and their families currently living through this market, three things follow from the research rather than from the headlines.

First, the underemployment statistic matters as much as the unemployment statistic. At 41.5 percent, the more common outcome for a struggling graduate is not joblessness but taking a role below their qualification level. This is painful but it is a different problem with different solutions than outright unemployment, and it typically resolves over a longer horizon as the graduate gains any work experience at all, which remains the strongest credential in a frozen hiring market.

Second, in-person and hybrid roles appear to offer a genuine structural advantage for early-career workers specifically, separate from any consideration of company prestige or starting salary. The research on remote work’s effect on junior employee development suggests that proximity to experienced colleagues during the first two to three years of a career has real, measurable value that is easy to underweight when comparing job offers primarily on salary or flexibility.

Third, the skills that reduce exposure to whatever combination of AI and hiring freeze is operating in a given field are not generic. They are specific to demonstrating competence that is difficult to delegate to either a remote arrangement or an AI tool: direct client or customer interaction, physical or hands-on technical work, and roles where judgement under ambiguous, non-standardised conditions is the core function rather than a secondary one.

The stock market’s record highs are real. The difficulty facing this graduating class is also real. Understanding that these two facts are connected through a more complicated chain of causes than a single AI headline can capture is the difference between navigating this labour market with a clear strategy and navigating it with a diagnosis that may be only partially correct.