Several UBI pilots are wrapping up multi-year studies — what do the results actually show?
GiveDirectly's long-term Kenya cohort, the OpenResearch US study, the Stockton SEED follow-up, Welsh basic income for care leavers, Sam Altman's expanded cohort — most of these are now 24-36 months in and publishing data.
Headlines vary wildly because the studies are different:
- Different income amounts ($500/month vs $1,000/month vs $4,500 lump sum)
- Different durations (12 months vs 36 months vs ongoing)
- Different control groups, populations, baseline labor markets
What's consistent across most:
- Employment hours: roughly flat or modest decrease (5-8 hours/week). Not a labor-market collapse.
- Subjective well-being: meaningful improvements, consistently
- Financial volatility (overdrafts, payday-loan reliance): consistent reduction
- "Big purchases" (cars, education, housing deposits): modest but real
What's inconsistent:
- Long-term employment trajectory after the pilot ends
- Impact on mental health vs just "feeling better when you have money"
- Whether the intervention itself or the unconditionality is what works
The thing nobody seems to have a clean answer to: what happens at scale. Pilots involve maybe 1,000 people in a labor market of millions. Scale changes the labor-market response in ways the pilots can't capture.
What I'd want to read more about: any honest write-ups of the labor-market scaling question.
3 replies
Labor-market scaling question is the right one. Classic theory says scaling shifts marginal labor supply enough that wages adjust and the income gain partially evaporates. Pilots can't show this. Stage-2 studies need synthetic-control regional rollouts.
The 5-8 hour/week labor reduction is the single most consistent finding across pilots and it's the data point that gets cherry-picked by both sides. Pro-UBI: 'see, no labor collapse.' Anti-UBI: 'see, work disincentive.' Both are right and both miss the point.
Mental-health outcomes are confounded with the 'just having money' effect everywhere I've seen. Real test would be UBI vs equivalent unconditional means-tested transfer. None of the published pilots have that arm.