Recent-graduate unemployment reaches 10%
Will the New York Fed report an unemployment rate for recent college graduates of 10% or more for any period?
Chance by Q4 2031
30%
Cumulative probability
- Latest forecast
| Year | Q1 | Q2 | Q3 | Q4 |
|---|---|---|---|---|
| 2026 | <1% | |||
| 2027 | <1% | <1% | <1% | 1% |
| 2028 | 3% | 4% | 6% | 8% |
| 2029 | 10% | 12% | 14% | 16% |
| 2030 | 18% | 20% | 21% | 23% |
| 2031 | 25% | 26% | 28% | 30% |
Model reasoning
The New York Fed's unemployment rate for recent college graduates is about 5.7% and has topped 10% only once, during the 2020 pandemic. Because young graduates now fare worse relative to other workers than in the past, a moderately deep recession lifting overall unemployment to about 7.5% would likely push it past 10%. I estimate about a 30% chance by late 2031. A recession beginning in 2027 would most raise the odds.
Where things stand
- Latest release. The New York Fed's last update came out on August 6, 2026 and covers data through 2026:Q2. It says recent-graduate unemployment "stayed elevated at about 5.6 percent."
- The series itself. I downloaded the underlying CSV (
college-labor-unemployment-data.csv, column "Recent graduates"). The data are monthly, seasonally adjusted, and smoothed with a three-month moving average; the chart notes incollege-labor-chart-meta.jsonconfirm this. The June 2026 value is 5.698. For comparison, "All workers" is 4.061. - History, 1990–2026. The series reached 10.0 or more only in May–August 2020, peaking at 13.411 in June 2020. Other peaks:
- Great Recession: 7.876 (December 2010)
- Early 1990s: 5.13
- Early 2000s: 5.32
- Next releases. The November 2026 release covers July–September. Summer 2026 data showed no spike (Fairlie & Wu, UCLA Anderson Review, October 2026), so the 2026 and early-2027 quarter-ends are close to zero.
- Macro backdrop.
- Headline unemployment was 4.2% in September 2026 (BLS series LNS14000000; CNN, October 2, 2026).
- Payrolls rose only 29,000. Jobless claims are near a 57-year low (Reuters, October 8).
- The Fed raised rates on September 16 amid an oil shock tied to the Iran war. The 10-year Treasury yield is about 5.2%.
- Kalshi priced a 2027 recession at about 20% (October 6). Polymarket puts a recession by end-2026 at about 8%.
Key structural fact
Recent graduates' unemployment divided by the all-workers rate was 0.67–0.98 from 1990 to 2018. It is now about 1.35 (1.09 in 2019, 1.23 in 2023, 1.33 in 2025).
Within recessions this ratio has stayed roughly stable, so the graduate rate rises in proportion:
| Recession | All-workers rate rose by a factor of | Graduate rate (peak) rose by a factor of |
|---|---|---|
| 1990–92 | 1.39 | 1.42 |
| 2001 | 1.59 | 1.92 |
| 2007–09 | 2.22 | 2.30 |
| 2020 | 3.58 | 3.42 |
From today's level, the graduate rate needs a factor of about 1.77 to reach 10%. That roughly corresponds to headline unemployment of about 7.3–7.6%, a rise of about 3–3.4 points, once you allow for sampling noise. I measured that noise against a 13-month trend: SD about 0.3 points, with a typical 12-month maximum about 0.43 above trend. If instead the gap is additive (about +1.5 points), headline unemployment would need to reach about 8.4%.
Base rate and model
- Historical windows. Using BLS unemployment from 1948 to 2026, I looked at every 60-month window starting near a cyclical low. The 3-month-average unemployment rate rose 3 points or more in 35% of them (38% after 1970, 32% after 1985). I also ran counterfactual graduate paths in those same windows:
- Proportional model: 33% reach 10%.
- Additive model: 21% reach 10%.
- Excluding windows that contain the pandemic: 24% (proportional) and 11% (additive).
- Monte Carlo simulation. Assumptions: recession risk of about 20% in 2027 and about 15% a year afterwards; recession sizes drawn from the historical rises in unemployment; a 70/30 weighting of the proportional and additive models; and a 15% chance of an AI-driven collapse in entry-level hiring. Result: 34% by September 2031. Variants:
- No AI path: 29%
- Additive model only: 20%
- Ratio drifting back down: 26%
- Ratio trend speeding up: 40%
Strongest counter-considerations
- For NO. Recessions have been less frequent since 1985, and layoffs are currently very low. Struggling graduates often go to graduate school (and drop out of the sample) or take non-college jobs rather than stay counted as unemployed. Remote-work effects may fade. Even the Great Recession only reached 7.9% on this series.
- For YES. The Metaculus "New grad unemployment" community medians for annual averages are 7.3% (2027), 10.1% (2030) and 12.8% (2035). A related Metaculus question gives about 19% to graduate unemployment of 20% or more before 2030. I judge these skewed by AI-disruption priors, since current data show stabilization, but they argue against going lower.
Forecast
I put about 30% on YES by late 2031. The hazard is near zero until 2028. It then runs about 2.2% per quarter, reflecting the lag between a recession starting and the graduate rate peaking, plus the publication lag from the February/May/August/November schedule.
Resolution criteria
Question. By the last day of each quarter, has the New York Fed reported an unemployment rate for recent college graduates of 10% or more for any period?
Definitions.
- New York Fed: the Federal Reserve Bank of New York.
- Unemployment rate for recent college graduates: the national aggregate “Recent college graduates” series in the Unemployment chart of The Labor Market for Recent College Graduates, including its downloadable underlying data. Recent graduates are people aged 22–27 with a bachelor's degree or higher, excluding those currently enrolled in school; unemployment is measured as a share of the group's labor force, not its total population.
- Any period: any observation published in that series, using the New York Fed's own period labels, smoothing, and seasonal-adjustment methodology. The quarterly update schedule does not restrict eligibility to quarterly observations. The Observation period convention applies: the observation period must end after 2026-09-16.
- 10% or more: a published numerical value of at least 10.0 percent. Use the downloadable series value in preference to rounded chart labels or narrative summaries when they differ within the same release; do not round a lower value up to 10%.
Resolves YES if a qualifying New York Fed publication within the Resolution window reports at least one eligible observation meeting the threshold. Initial releases and subsequently revised observations both count. A later downward revision does not undo a YES. No attribution to AI is required.
Resolves NO if no qualifying observation has been published by the horizon. Underemployment, rates for individual majors or other subgroups, forecasts, scenarios, rates for all college graduates, and independently calculated rates do not count. Publications repeating or revising only periods ending on or before the opening date do not count. A renamed or relocated successor series counts only if it continues to measure the same national recent-graduate population and unemployment concept.
Resolving source. New York Fed publications only: The Labor Market for Recent College Graduates, its Unemployment chart and linked downloadable data, or a New York Fed-authored release or Liberty Street Economics article explicitly reporting that series. This is a narrower source rule than Public confirmation (A2). Apply Resolution date (A3), the Resolution window, and the Observation period convention. The series is normally updated in February, May, August, and November; resolution uses the publication date, not the observation date.