Ministry of AI · Dispatch from 2047
Does UBI Cause Inflation? The Objection, Priced
Editorial note. The Ministry of AI is a work of disciplined foresight: it describes the year 2047 in the present tense, and treats our own era as history. The institutions are imagined. The economics, the evidence and the historical parallels are real and sourced.
The first serious question anyone asked about the dividend was not moral. It was arithmetic, and it was asked by a woman in the fourth row of a hearing room in the early 2030s, and it was this: if you hand everyone the same money, doesn’t the shopkeeper just raise his prices?
I have spent fifteen years watching people try to wave that question away, and I want to say plainly that they were wrong to. It is the best objection there is. It cannot be answered with sentiment, it cannot be answered by calling the questioner cruel, and for most of my working life the people making my argument answered it badly — which is a large part of why the argument took two decades to win.
So: does a universal payment cause inflation? The honest answer is that the question contains two different claims wearing the same coat, and they have different answers.
The two objections
The first claim is about money. If a state pays every household a sum it did not first collect from somewhere, it has added purchasing power to an economy that has not added goods. More claims chasing the same output raises prices. This is not ideology; it is the least controversial proposition in macroeconomics, and the evidence for it is recent and painful. San Francisco Fed researchers, comparing the United States with a sample of OECD economies, estimated that pandemic fiscal support contributed roughly three percentage points to US inflation by the end of 2021 — with, as they carefully said, considerable uncertainty. Anyone who argues for a universal payment while pretending that episode did not happen is not arguing; they are performing.
The second claim is about supply. Even if not a single new unit of currency is created, a transfer raises demand for particular things, and where the supply of those things cannot respond, the price rises instead of the quantity. This one does not care how the payment is financed. It is the objection that survived into our own era, and it is the one that actually bit.
The mistake of the 2020s was to answer the second objection with evidence about the first, or the reverse. Both camps had a favourite study and neither had a theory of when their study applied.
What the evidence actually shows
Set the two best experiments side by side and the picture stops being contradictory.
In rural Kenya, researchers delivered one-time transfers of about a thousand dollars to more than ten thousand poor households across 653 randomised villages — a fiscal shock exceeding fifteen percent of local GDP. They found large consumption gains, large positive spillovers onto households that received nothing, a local transfer multiplier of about 2.4, and minimal price inflation. An enormous transfer, and prices barely moved.
In the United States in 2021, a comparable-in-spirit transfer coincided with the sharpest inflation in forty years.
The difference is not that one population was more deserving. It is that idle capacity absorbs demand and constrained capacity converts it into price. Kenyan village economies in that study had unused labour and unused inventory; a shock arriving there produced output. The pandemic economy had shuttered factories, blocked ports and a labour force that could not physically show up; a shock arriving there produced numbers on shelf tags.
That is the whole of the theory, and it is why the Ministry’s rules are written the way they are.
| Money-financed transfer | Yield-financed dividend | |
|---|---|---|
| Source of funds | New issuance or borrowing against future output | Income already measured in the Machine Yield Account for a closed period |
| Aggregate demand | Rises above income | Unchanged in total; redistributed between holders |
| Behaviour in a supply shock | Amplifies it | Neutral in aggregate; still shifts composition |
| Effect on inelastic goods | Prices rise | Prices rise |
| What limits the payout | Political appetite | Measured yield, published with its error band |
| Failure mode | Currency debasement | Payout falls in the year households need it most |
The funding-source rule
The Ministry may only distribute yield it has already measured. Not forecast yield, not borrowed against yield, not supplemented from issuance. The Dividend Schedule for a quarter is a division of a number that is already closed.
This makes the dividend structurally different from the transfers that frightened the woman in the fourth row. It is not new money meeting old goods. It is the return on an asset being paid to the asset’s owners instead of accumulating entirely to the parties who happen to operate it. Value that used to leave companies as wages and now leaves as margin is, in part, routed home by a different pipe. The pipe changes who holds the claim. It does not manufacture claims.
And the rule has a price, which we pay openly. In a bad year — a yield contraction, a sector-wide model failure, the 2041 revision — the dividend falls. Households discover that their floor is indexed to the machines’ productivity rather than to their own need, and that is a genuinely worse property than a legislated benefit would have. We chose it anyway, because a payment that can be topped up by issuance is a payment that will be, and the first government to do it would have proved the objection for us.
Where it genuinely failed: the inelastic list
Now the second objection, the one financing discipline does not touch.
Some goods cannot be produced faster because someone has more money. Land in a desirable city is the purest case. Childcare hours, tuition at institutions that select on scarcity, elder care, medical procedures with fixed rota capacity: the Ministry maintains this as the inelastic list, and it exists because of what the pass-through register showed us in the first decade.
The pass-through register is a quarterly table reporting, category by category, how much of a dividend increase surfaced as higher prices rather than higher consumption. We can measure it because the dividend arrived at different times and different rates across regions, which gives us something close to a natural experiment and something considerably better than an opinion.
The register’s verdict on housing was ugly. In supply-constrained metropolitan areas, a substantial minority of the payment was capitalised into rents within three years — the money moved, arrived, and continued moving, into land. Nobody should have been surprised. The pattern was documented long before us: New York Fed researchers studying the expansion of federal student aid found a pass-through to tuition of about sixty cents on the dollar for increases in subsidised loan caps. Subsidise the buyer of a good in fixed supply and you have, to a first approximation, subsidised the seller.
We answered this late and partially: dividend indexation excludes local rent movements, so a city cannot inflate its own claim on the Schedule; and the place share was made conditional on housing supply performance, which is the least popular clause in the entire statute. It is still true that in the tightest markets a portion of every payment lands in a landowner’s account. This is the design’s clearest unresolved failure, and I would rather write it down than let a critic discover it for me.
There was always another route: give people the goods instead of the money. It is not a foolish idea, and it has evidence behind it — a Mexican programme that randomly assigned villages to food boxes or equivalently valued cash found in-kind transfers lowered local prices, a benefit to consumers worth about eleven percent of the transfer, precisely because trucking in goods adds supply as well as demand. We declined. In-kind arrives with an assessor, a queue and a form, and an assessor is the thing an owner does not have. Capacity is not contribution, and work that must be done to survive is not chosen work — the same sentence rules out being handed a box and asked to be grateful for it.
My mother’s grocery bill
She is seventy-three now, and she was the woman’s question in miniature. When her first dividend arrived at fifty-nine, three months after nineteen years of reading claim files was absorbed by a system trained on her own decisions, she did not spend it. She waited. She told me she wanted to see whether the shops would take it back before she got used to it.
They did not, mostly. Food did not run away from her; there was capacity, and competition, and the register shows food pass-through in the low single digits. Her rent did run. Between 2038 and 2043 the landlord’s portion took something close to a fifth of what the Ministry sent her, and she moved once because of it — to the town with the library, as it happens, which is how these dispatches keep circling back to the same street.
So the answer I would give the woman in the fourth row, if I could go back, is not no. It is: not in aggregate, if you refuse to print it — and yes, in housing, unless you build. The dividend was never a substitute for supply. It was a correction to authorship. If we let it become an excuse not to build, we will have taken an inheritance and handed a fifth of it to whoever already owned the ground.
FAQ
Does a universal basic income cause inflation?
It depends on the source of the money and the elasticity of what it buys. A payment financed by new issuance or borrowing adds claims to an economy that has not added output, and prices respond — the pandemic-era estimates are clear enough about that. A payment financed out of income already measured transfers claims rather than creating them, and is not inflationary in aggregate. Both versions raise prices in markets that cannot expand supply.
What is the funding-source rule?
The Ministry distributes only yield already measured for a closed period, with no borrowing against forecast yield and no monetary top-up. The dividend is a division of an existing number. The cost is procyclicality: in a bad yield year the payment falls.
What is the pass-through register?
A published quarterly table showing how much of a dividend increase became higher prices instead of higher consumption, category by category, exploiting differences in dividend timing across regions. It is designed to expose the programme’s leakage rather than defend against the accusation of it.
Did the dividend raise rents?
Yes, in supply-constrained cities, by a substantial minority of the payment within roughly three years. The precedent was visible decades earlier in subsidised student lending, where increases in loan caps passed through to tuition at around sixty cents on the dollar. Indexation rules and the housing-supply condition on the place share reduced it; nothing has eliminated it.
Why not distribute goods instead of cash?
Because in-kind transfers require an assessor, and unconditionality is the point. The price argument for in-kind is real — adding supply alongside demand lowers local prices — and we accepted the loss knowingly.