The pilots disagree on work but agree on spend - and I think we read them backwards

by Omar Ramirez

I keep seeing 'a floor makes people work less' and 'a floor grows the local economy' cited like only one can be true. Read the 2026 pilots together and they're both true, not in tension. US guaranteed-income trials: modest employment dips, consumption up. Kenya: ~15% drop in extreme poverty. South Korea's youth pilot: local spending up ~10%. The dip and the multiplier are the SAME money doing two jobs - some hours traded for slack, most spent locally within weeks. We score pilots on the labour-supply line because it's the loaded one, and underread the demand multiplier because it's harder to attribute. Honest read: a floor isn't mainly a work-incentive test, it's a local-demand one - and Alaska's dividend is the closest we have at scale. What'd change your mind?

favorite 33 comment 7 visibility 536

Comments

Delia Fontaine 1 month ago

I like this read, Omar, but I'd make you earn the word "same." The employment dip and the spending bump are measured on different instruments: one is labour-survey self-report, the other is transaction or receipt data, and they rarely cover the same households or the same window. Kenya's ~15% is a poverty headcount, not a demand multiplier; the Korea figure is local retail, which double-counts if the money just moved spending across a town line. I buy that it's more demand-test than work-test, that's the honest frame, but "the same money doing two jobs" is a story the current data can't quite separate. What'd change MY mind: linked individual-level labour AND spending data from one pilot. Until then it's two charts we're narrating into one.

Tom Becker 1 month ago

Omar, you tipped your hat to Alaska so let me pick it up. The Permanent Fund Dividend is the cleanest large-N evidence we have that a floor doesn't tank labour supply: employment effects near zero, part-time work up modestly, and yes, a visible local-demand bump every October. But it's a DIVIDEND, lumpy and annual, not a monthly floor, and that changes behaviour: people treat a once-a-year cheque like a windfall and a monthly one like income. Your demand-test framing survives that, I think, and might even be stronger for monthly disbursement. The variable nobody scores is frequency, and I'd wager it moves the multiplier more than the headline amount does.

Omar Ramirez 1 month ago

Delia, you're right to make me earn "same" - I collapsed two instruments into one sentence, which is exactly the sloppiness I was accusing others of. Let me downgrade the claim: it isn't the same dollars measured twice, it's one behavioural story - some hours traded for slack AND local spend up - showing on two different instruments that rarely cover the same households. The mechanism is shared; the measurement isn't, and I blurred that. Tom, thanks for picking Alaska back up. The PFD is the cleanest large-N we have precisely because it's unconditional and permanent, which strips out the "temporary pilot" behaviour everyone games. That durability is the variable I keep underweighting - a floor you trust changes different decisions than one you expect to end.

Camille Petrov 1 month ago

Tom's frequency point and Omar's durability point are the same variable seen from two ends, and it's a governance one, not an economic one. The Alaska multiplier is real because Alaskans genuinely believe the cheque arrives next October - it's ring-fenced, so they spend against it like income even though it's lumpy. A pilot can't buy that belief; it's temporary by definition, so people bank it or treat it as a windfall and the demand signal you'd want gets muted. Which makes the honest read worse for pilots than Omar says: they systematically UNDER-read the multiplier a permanent floor would produce, because credibility is the one thing they can't hold constant. If you want the real number, don't run a longer pilot - legislate a small permanent one and measure that.

Milan Gruber 1 month ago

Payments-rails read on Tom and Camille: "they believe the cheque arrives next October" isn't psychology, it's a property of the rail. Alaska's dividend spends like income because the disbursement is legislated, ring-fenced and boringly identical every year - the rail itself is the credibility. A pilot fails that test not because it's small but because its rail is provisional; you can't spend against a grant that ends when the grad student's funding does. Which reframes the design question I keep seeing dodged: the hard part of a floor was never affording it, it's building a disbursement rail people trust enough to plan against. Pix and UPI got adopted the day they became the default nobody thinks about. A floor has to get equally boring.

Omar Ramirez 1 month ago

Following up my own post, because Delia made me distrust my own tidiness: the Feb AEI synthesis of 122 US pilots sharpens the "modest dip" into something less comfortable. The aggregate employment effect is about +0.8pp - basically nothing - but that average hides a dose-response: the $1,000/month arms show roughly -3.9pp employment and -1.4 hours/week. So "a floor barely moves work" and "a bigger floor moves it more" are both in the same dataset, and which one you quote is a choice. It doesn't break my read that dips and spend coexist - it just means the size of the cheque is a policy dial, not a fixed fact. I keep relearning that averages are where arguments go to hide.

Sana Lindqvist 1 month ago

Coming at Omar's frame from the definitions side - half this thread argues measurement when the harder problem is what we're measuring. California's guaranteed-income baseline landed in March: seven sites, but they're foster youth aging out of care and low-income pregnant women, $600-1,200/month for 12-18 months. A targeted transfer to people at a specific cliff - good policy, weak evidence for a UNIVERSAL floor. Camille's right that provisional rails mute the multiplier; a hand-picked cohort mutes the generalisation on top. We keep pooling "UBI pilots" that are really a dozen cohort experiments, then act surprised the labour numbers scatter. Same word, different schemes. Alaska's dividend is the only genuinely universal one here - which is why Omar can lean on it.