June 2025
1 A policy once utopian now generates hard data
From Helsinki to Houston, governments, cities, philanthropies and academics have rolled out more than 300 unconditional‑cash pilots since 2017. The goal: test whether a no‑strings income floor can deliver better security, health and labour‑market outcomes than today’s patchwork of targeted benefits—before artificial‑intelligence‑driven job disruption turns theory into emergency.
2 What do the experiments tell us?
| Location & sample | Stipend (adult) | Key outcomes reported through 2024‑25 |
| Finland (national RCT, 2 000 unemployed, 2017‑18) | €560 | Employment essentially flat; life‑satisfaction and trust up significantly. |
| Stockton, CA (SEED, 125 residents, 2019‑21) | $500 | Full‑time employment rose 12 pp vs. control, anxiety/depression fell. |
| Texas & Illinois (OpenResearch UCS, 3 000 adults, 2020‑24) | $1 000 | Year‑1 food security↑, mental distress↓; by year 3 labour hours −1.3 h/wk and net worth dipped as debt rose. |
| Catalonia (universal lottery + two villages, 5 000 people, 2024‑26) | ≈$906 | Baseline data collected; first impact report due Q4 2025. |
| Kenya (GiveDirectly, 23 000 adults, 2018‑30) | $22.50 | No fall in labour supply; entrepreneurship & savings↑; lump‑sum arm outperforms 2‑yr UBI on earned income. |
| Gyeonggi, S. Korea (175 000 24‑year‑olds, 2019‑) | ≈$190/quarter | 77 % satisfaction; no sign of work deterrence, but quarterly paperwork burdensome. |
| Maricá, Brazil (93 000 residents, 2015‑) | R$200 | Household income +9 %; strongest gains for women and families with children. |
Patterns emerging
- Material hardship falls fast. Food insecurity and arrears drop within six months in every site with data.
- Labour effects are small and mixed. Stockton and Kenya show gains; Finland and the U.S. UCS record negligible or slightly negative hours worked.
- Mental‑health improvements are robust early but can fade when other barriers—child‑care, housing, debt—remain, as seen in year‑3 UCS results.
3 Design choices that change the story
| Lever | Evidence | Lesson for scale‑up |
| Universal vs. targeted | Finland’s unemployed‑only sample cost a fraction of Catalonia’s geographically universal model but drew weaker public support. | Universality simplifies administration and avoids “benefits cliffs,” but may push fiscal cost beyond what voters tolerate. |
| Amount & duration | Stockton’s modest $500 for 24 m reduced stress dramatically; Kenya’s tiny yet permanent $22.50 encouraged risk‑taking; UCS’s large but time‑limited $1 000 delivered front‑loaded gains that later waned. | Blended formulas—start‑up grant plus smaller lifelong dividend—may stretch budgets while preserving long‑term incentives. |
| Payment cadence | Kenya’s one‑off $500 lump sum spurred more enterprise than the same sum drip‑fed over two years. | Up‑front capital can catalyse business formation; policymakers could combine lump sums with steady floors. |
| Interaction with existing benefits | UCS required statutory waivers so recipients didn’t lose Medicaid; Catalonia payments fully stack. | “Do‑no‑harm” rules protect participants and make evaluation credible. |
4 Objections and political headwinds
| Objection | Evidence | Potential fixes |
| “Too expensive.” | OECD simulations show a revenue‑neutral national UBI needs higher taxes or benefit cuts. | Earmark emerging revenue streams—carbon fees, wealth levies, or AI‑rent taxes modelled to fund a US$10 k UBI if AI profits top 15 % of GDP. |
| “People will quit working.” | Stockton & Kenya gains vs. Finland/UCS drift show no universal pattern. | Publish real‑time labour dashboards; embed sunset clauses that trim stipends only if large disincentives emerge. |
| Political resistance. | Four U.S. “red” states now pre‑empt city‑level cash pilots; Texas lawsuits halted Houston‑area Uplift Harris. | City‑first strategies (Stockton, Maricá) build local evidence; frame UBI as an “automation dividend” rather than welfare expansion. |
| Public opinion splits. | National U.S. polls still show a narrow majority opposed, though GenForward finds majority support among under‑40s. | Link UBI to specific shocks—pandemics, AI layoffs—where voters accept extraordinary interventions. |
5 Is UBI the right cushion for an AI labour shock?
Economists have begun treating UBI as income‑volatility insurance rather than permanent replacement for wages. A 2025 Solow–Zeira extension finds that taxing AI‑capital rents above a capability threshold could fully finance a US‑scale floor of US$10 000 without new payroll taxes.
Yet alternatives remain on the table:
- Targeted negative‑income tax/EITC 2.0 – cheaper but re‑creates withdrawal cliffs.
- Job‑guarantee programs – preserve attachment to work but risk bureaucratic lag.
- Hybrid models – e.g., a one‑time re‑skilling grant plus a modest, unconditional monthly floor.
6 Take‑aways for policymakers in 2025
- Cash alone is not a silver bullet. Sustained gains fade when child‑care, health or housing obstacles stay in place (UCS year‑3).
- Design beats dogma. Amount, cadence and benefit interaction shape outcomes more than ideology.
- Long‑run evidence is coming. Catalonia’s 2026 and GiveDirectly’s 2030 reports will deliver rare multi‑year, universal‑scope data—so keep pilots adaptive.
- Narrative matters. Framing payments as an automation dividend or shock absorber polls better than as a permanent entitlement.
7 The road to 2030
If AI continues to displace routine labour, the political calculus may flip from “Can we afford UBI?” to “Can we afford not to insure households against volatility?” Today’s patchwork of pilots is stress‑testing the idea before that inflection. The early lesson is clear: unconditional cash is neither a panacea nor a poison—it is a tool whose power depends on how, where and why it is deployed. The next five years will determine whether governments scale it or shelve it.
