
Quick Answer
Work from home adoption, more than AI exposure, is the factor that tracks most closely with the fall in employment for recent US college graduates, according to an analysis of Current Population Survey data published on Stripe Economics. Between 2023 and 2026, the employment-population ratio for 22- to 26-year-olds with a bachelor’s degree fell 1.23 percentage points, and telework adoption in an occupation accounts for between 56% and 88% of that decline, while occupational AI exposure shows a much smaller and statistically insignificant relationship, according to the Stripe Economics analysis.
In This Article
How Weak Is the Labor Market for New Graduates?
According to Stripe Economics, among workers age 22 to 26 holding a bachelor’s degree, the employment-population ratio stood at 88% in September 2026, a full percentage point below where it was three years earlier and nearly a point and a half below its immediate pre-pandemic level. Stripe Economics also reported that unemployment for that group was close to 6% at the end of 2025, against around 4% before the pandemic, though it appears to be cooling now. The typical unemployed young graduate had also spent 12 weeks looking for work, a third more than a few years ago, according to Stripe Economics.
The weakening is not confined to graduates. As Ozimek and Goldschlag (2026) have shown, youth employment is also down since 2023 for young non-college-graduates, who are less likely to work in industries that use AI or in jobs exposed to it. Stripe Economics reported that young non-graduate employment rates have fallen slightly more since December 2019 than young graduate rates, while mid-career workers age 30 to 49 have fully recovered to pre-pandemic levels with little difference between those with and without a bachelor’s degree.
Class of 2023 Graduate Outcomes (2023)
Class of 2023 graduate outcomes showed that nearly 85% of bachelor’s degree graduates were employed or engaged in further education within six months of graduation. NACE reported that full-time employment for bachelor’s graduates was 57.2% within six months, down from 59.1% for Class of 2022. The bachelor’s outcome rate declined 1.2% from the prior class but exceeded rates recorded during the pandemic, according to NACE.
What Does the Evidence Show?
The analysis compares payroll employment-population ratios for recent graduates against those of middle-career workers, measuring the difference between the two groups for sets of broad and detailed occupations in each of the nine major Census Bureau divisions, then comparing that spread to work-from-home adoption rates in the same occupational categories as measured in the CPS. Both metrics cover the January to December 2023 period against the 12 months ending September 2026, with October excluded from both periods because the CPS was not collected in October 2025 due to the US federal government shutdown, leaving 11 months in each sample. October was excluded from both sample periods, and each sample period has 11 months.
The work-from-home relationship is negative and statistically significant. According to the Stripe Economics analysis, between 2023 and the last 12 months, for every extra 10 percentage points of work-from-home adoption, employment rates for young college grads grew between 3.9 and 6.6 percentage points more slowly than for middle-aged workers. As an accounting exercise, rather than a causal one, the authors estimate that work from home explains between 0.69 and 1.08 percentage points of the 1.23-point decline in the young-graduate full-time-equivalent employment-population ratio, or 56% to 88% of the total, according to Stripe Economics. Looking only at extensive-margin employment, the rate of young graduates employed at all regardless of hours, work from home explains 97% to 153% of the decline, according to the Stripe Economics analysis.
When occupational AI exposure replaces work from home in the model, the effect on the new-grad employment spread is also negative but much shallower and statistically indistinguishable from zero. With both included, the work-from-home effect stays statistically significant while AI exposure remains insignificant, and industry AI adoption tells the same story: industries self-reporting a higher likelihood of using AI in the first half of 2026 saw, if anything, slightly faster young-grad employment growth relative to mid-career, with that effect also statistically insignificant, according to the Stripe Economics analysis. AI exposure in the model comes from the composite occupational measure in Gimbel, Kendall, and Kulsakdinun (2026) and from AI adoption by industry in the Census Bureau Business Trends and Outlook Survey.
Stripe’s own transaction data points in the same direction. Because Stripe has no direct visibility into whether a person is working from home, the analysis uses proxies such as spending on office supplies and furniture, and notes that office supply, furniture and electronics spending relative to a state’s gross domestic product is moderately correlated with work-from-home adoption rates. According to the Stripe Economics analysis, looking at commuting zones, the clusters of adjacent counties with economic connections, where proxy work-from-home spend grew faster than the regional economy between 2023 and 2024, those areas were likelier to see declines in employment for those ages 22 to 29 from 2023 to 24, with a moderate correlation. Only merchants present in both 2023 and 2024 are included, so Stripe’s mix of business does not become a factor. The 2024 endpoint is the last year available in the American Community Survey, the source of the employment data, and the top 100 commuting zones by spend are the ones compared.
What Does the AI Research Say?
The AI research on young workers is mixed, and several papers point in different directions.
On the negative side, Brynjolfsson, Chandar, and Chen (2026) find that young workers in AI-exposed occupations saw slower employment gains than older workers and workers in less AI-exposed occupations, with findings that hold after controlling for confounding factors including work from home. Tucker (2026) finds a sharp decline in hiring of workers ages 22 to 24 in the most AI-exposed industries after ChatGPT’s release in late 2022. Orr, Tucker, and Warren (2026), using linked US Census Bureau records, find that graduates in the most AI-exposed tenth of college majors saw initial employment fall by about five percentage points and initial earnings by about 13% relative to less-exposed majors after late 2022. Emanuel, Harrington, and Pallais (2026) find that physical proximity to the office increases the job feedback received by software engineers, especially younger and less-experienced workers, and their national analysis finds a relative increase in young graduates’ unemployment in remotable occupations. Lambert and Schindler (2026) find declines in junior hiring shares associated with work-from-home exposure after controlling for AI exposure.
Other studies find little. Fairlie and Wu (2026) find no unusual relative summer 2026 unemployment increase among non-enrolled college graduates age 22 to 25, along with some positive teleworkability interactions. Kauhanen and Rouvinen (2026) find no systematic AI-related displacement or persistent wage differences by AI exposure among young workers in Finland through September 2025. Frank et al. (2026) find the deterioration in US graduates’ entry into AI-exposed jobs begins before ChatGPT, with graduates who have more AI-exposed coursework showing shorter job searches after its release and higher first-job pay when entering highly exposed occupations. Ando et al. (2026) find a negative relationship between AI exposure and labor demand through job openings, but no significant difference between junior and senior roles and even a somewhat smaller negative effect among very junior openings of one year experience.
Brynjolfsson, Chandar and Chen’s working paper, revised August 12, 2026, uses high-frequency administrative payroll data from ADP covering millions of US workers through June 2026 and documents no evidence of widespread, economy-wide job displacement. The authors report that employment of young workers ages 22 to 25 in AI-exposed occupations now stands 19% below where it would be had it kept pace with their less-exposed peers, while experienced workers show no comparable gap. The authors also report that the divergence has widened steadily since they first documented it in August 2025, operates primarily through reduced hiring of young workers rather than increased separations, and is concentrated in occupations where AI usage primarily substitutes for human tasks, with employment flat or rising where usage complements workers. Adjustment occurs through employment rather than base compensation. The authors interpret these as early, descriptive indicators rather than causal estimates, and note the patterns attenuate when controlling for education, show some divergent trends predating generative AI, and are more pronounced in the ADP sample than in national survey benchmarks.
On work from home, the evidence is more consistently favorable to some groups. Bloom, Dahl, and Rooth (2026) find that remote work increased full-time employment among physically disabled Americans. Burn, Gentry, and Lahey (2026) use a related approach for the UK and find positive occupational full-time employment effects for disabled workers and workers ages 50 to 64. According to Stripe Economics, there is increasing consensus that work from home helps the job prospects of disabled and older workers, which the analysis treats as a substantial benefit that should not be dismissed.
Why Work From Home Is Hard to Separate From Other Forces
The modern era of AI began in late 2022, at the same time other major economic shifts were taking shape, which makes it hard to disentangle the impact of any single factor. High inflation and the resulting higher interest rates likely pinched certain industries. According to Stripe Economics, job switching surged during the pandemic “Great Resignation” and then swung to record lows as companies pulled back on both hiring and firing, which could disproportionately hurt new labor force entrants. Immigration flows rose and fell sharply, changing job dynamics in uncertain ways. AI exposure and teleworkability are also highly correlated with each other, making the two difficult to distinguish definitively, and the analysis notes factors it did not test, such as higher interest rates and past hiring surges. The Stripe Economics analysis describes the model as suggestive rather than truly causal and says the full extent of AI effects on the labor market might not appear until the next recession, much as the effects of the internet did for travel agents.
Key Considerations
- The work-from-home results are accounting relationships, not causal estimates, and the authors describe the model as suggestive rather than truly causal.
- AI exposure and teleworkability are highly correlated, so the two factors are difficult to separate definitively even though work from home remains significant when occupational AI exposure is included as a control.
- According to Stripe Economics, work-from-home adoption correlates with higher full-time employment for disabled workers and for workers ages 50 to 64, both in US data and in UK data.
- According to Stripe Economics, the work-from-home finding applies to occupations where work from home is now an expectation, and young workers themselves might demand work-from-home jobs less than older workers.
- Stripe Economics reports that the AI research on young workers is not universal: several studies find no unusual unemployment increase, no systematic displacement, or declines that begin before ChatGPT’s release.
- According to Brynjolfsson, Chandar and Chen, the AI employment gap among young workers, at 19% below less-exposed peers in the ADP payroll data, is a descriptive indicator rather than a causal estimate.
What to Do Now
- Review where your own hiring happens and whether those roles are remote-eligible.
- Check whether roles you post to recent graduates are classified as work-from-home occupations.
- Track the employment-population and job-feedback patterns in your own teams.
- Evaluate roles that depend on in-person learning and feedback.
- Compare your industry AI adoption against AI exposure measures for your occupation.
FAQ
What Share of the Decline in Young Graduate Employment Does Work From Home Explain?
According to the Stripe Economics analysis, between 0.69 and 1.08 percentage points of the 1.23-point decline, equal to 56% to 88% of the total, are explained in an accounting rather than causal exercise. On extensive-margin employment alone, work from home accounts for 97% to 153% of the decline.
Does AI Exposure Explain the Decline in Young Graduate Employment?
In the model used for the analysis, the effect of occupational AI exposure on the new-grad employment spread is negative but much shallower than the work-from-home effect and statistically indistinguishable from zero. It remains insignificant when work from home is included in the same model, as does industry AI adoption from the Census Bureau Business Trends and Outlook Survey.
What Is the Employment-Population Ratio for US College Graduates Age 22 to 26?
According to Stripe Economics, it was 88% in September 2026, a full percentage point below where it was three years earlier and nearly a point and a half below its immediate pre-pandemic level. Unemployment for the group was close to 6% at the end of 2025, against around 4% before the pandemic.
How Long Are Unemployed Young Graduates Searching for Work?
According to Stripe Economics, the typical unemployed young graduate had spent 12 weeks looking for work, a third more than a few years earlier.
Who Benefits From Work-From-Home Jobs?
Remote work increased full-time employment among physically disabled Americans in Bloom, Dahl and Rooth (2026), and Burn, Gentry and Lahey (2026) find positive occupational full-time employment effects for disabled workers and workers ages 50 to 64 in UK data. Work from home also weakens employment for young people, who lose the benefits of in-person learning and agglomeration effects, and employers are likelier to hire older workers for jobs where work from home is now an expectation.
Is There Widespread Job Displacement From AI?
Brynjolfsson, Chandar and Chen (2026) document no evidence of widespread, economy-wide job displacement in ADP payroll data covering millions of US workers through June 2026. They report that young workers ages 22 to 25 in AI-exposed occupations stand 19% below where they would be had they kept pace with less-exposed peers, with the divergence operating primarily through reduced hiring of young workers.
Sources
See our editorial disclaimer for how our articles are produced.
Run a free scan to see your AI Visibility Score, SEO rating, and local citation accuracy.