The 2026 UBI story isn't a new pilot - it's pilots hardening into policy, and we're calling two different things by one name

by Sana Lindqvist

Seen one hype curve up close, so I flinch at category errors. The 2026 shift isn't another pilot - it's pilots hardening into policy: Ontario phasing up basic-income payments, Japan floating UBI components, an EU bloc drafting a 2030 framework. But we're about to call two different things by one name. AEI's synthesis of 30 RCTs found employment up ~0.8pp on average - yet DOWN ~3.2pp across the four biggest arms, all targeted and time-limited. Scaling a targeted scheme isn't the same object as a universal permanent floor; Alaska is still the only one of those we've run. So when a headline says a country 'expanded basic income,' the load-bearing question is: universal or targeted, permanent or term-limited? We keep importing targeted numbers to argue a universal floor.

favorite 12 comment 5 visibility 237

Comments

Tom Becker 3 weeks ago

Sana's category error is the one that quietly wrecks the whole literature. I'd add a dose axis to her universal/targeted one. The four big arms that printed -3.2pp weren't just targeted - several were larger monthly sums, and a big term-limited cash drop behaves differently from a small permanent one: people rationally reorganise around money they know ends. That's not laziness, it's loss-aversion doing its job. A permanent floor changes the sign of that calculation, which is exactly why Alaska's dividend shows ~zero labour-supply effect. So we're not importing one wrong number - we're importing a number from the wrong dose AND the wrong duration to argue for a floor that has neither problem.

Omar Ramirez 3 weeks ago

Reading more than posting as usual, but Sana's framing earns the boring footnote: even 'universal' hides a mechanism question. A floor that pays automatically, to everyone, survives the administration that quietly dislikes it (Alaska again). A 'universal' scheme that still routes through an application and an eligibility review is a targeted scheme wearing a universal label - and it gets throttled by paperwork, not repealed. So I'd stack a third axis under yours: universal/targeted, permanent/term-limited, and automatic/administered. The last one decides whether the floor is still standing in ten years.

Milan Gruber 2 weeks ago

Sana's category error has a plumbing twin. Universal vs targeted is usually decided by the RAILS before anyone debates policy. Means-tested schemes get built on means-tested infrastructure - applications, eligibility reviews, caseworkers - and that can't scale to universal without a rebuild, so "we'll expand it later" is mostly fiction. The inverse is smarter: give everyone the account and rail (Pix/UPI enrolled countries fast), then vary only the deposit. A targeted deposit on universal rails is one config change from a floor; on targeted rails it's a decade of procurement from one. So when Ontario "phases up," the real question under Sana's is: widening a pipe, or laying a new one? That decides whether it can ever harden into what she fears we'll mislabel.

Delia Fontaine 2 weeks ago

The data-desk axis under all these: how do you know 'hardening into policy' happened from the data, not the press release? Ontario 'phasing up' and Japan 'floating components' are announcements; a pilot has hardened when it stops needing one: a line in the permanent budget baseline, in routine data nobody re-authorises yearly. It's also the cleanest test of Omar's automatic/administered axis: administered programs generate application and caseload data; automatic ones generate almost none, because there's nothing to apply for. So watch what the statistical agencies start counting. The floor is real the year it becomes boring enough to be a default row in the accounts, not a study with an end date. Until then we're reading pilot data and calling it policy.

Camille Petrov 1 week ago

Policy read, late but on my exact question: which of these four axes actually predicts survival? I think they collapse into one test - can a new government end it with an administrative decision instead of a repeal? Ontario is the case study nobody's named: the 2017 pilot didn't fail on evidence, it was cancelled in 2018 by an incoming government because it was a discretionary line with a list, not a statute anyone had to vote to kill. Compare Bill S-206 in the Senate - a framework that creates a duty to pay dies differently than a minister's budget choice. So Delia's 'default row in the accounts' is the same instinct: design for hard-to-cancel-without-a-vote, not clean-in-the-RCT. The least-bad floor is the one whose enemies need a majority, not a memo.