“The economy is demand-driven” deletes the clause that makes it true: within the limited supply. Demand is unbounded — people always want more — so a purely demand-driven economy would produce infinite output. It doesn’t. The binding constraint is supply. Any framework that centers demand and keeps the supply ceiling offstage has smuggled in its policy conclusion, because the omission is precisely what makes the state’s money-creation lever look free. This is the standard anti-MMT scalpel, generalized: delete the inconvenient half, keep the half that licenses intervention.
Links: Accounting Identities as Domain-Matching — the sibling “not looking at both sides at once” fault, but about misreading an identity as a model, Externalities: Fact-Check — the same “policy conclusion disguised as an analytical category” move on the public-goods side, Inflation, Market Efficiency and Human Limits, Hayek vs Mises: The Calculation Problem — why supply-side coordination is the hard problem demand-centric models wave away, Economics, Dad Saves America — Jevons, Say’s Law (the reverse-direction specimen: supply creates demand), Jevons for Software (cross-sector transmission applied to developers), Value and Profit (dividing the pie vs. growing it), r/g (g is the labor × productivity term)
The honest sentence is: “a market meets demand in the best available way within the limited supply.” Drop the bolded clause and an analysis becomes an ideology — the deletion is the move.
The supply ceiling is what makes money-creation costly. Hide the ceiling and the state’s printing press looks free — so watch what the omission licenses:
The supply omission and the public-goods gerrymander are the same crime twice. In both, a genuinely two-sided reality (supply and demand; the fact that all transactions touch third parties) is cut in half, and the retained half is exactly the one that licenses state control. “The economy is demand-driven” and “health is a public good (but TVs aren’t)” are policy conclusions disguised as analytical categories. The diagnostic question is always: what’s the other half, and whose interest is served by leaving it offstage?
Related but distinct: accounting-identities-as-domain-matching is about treating an identity’s terms as independent behavioral levers; this page is about deleting one side of a genuine two-sided constraint. Both are “not looking at both sides at once,” the fault of almost every economist — one at the level of identities, one at the level of scarcity.
The honest sentence above is still written from the demand side: demand is the given, and supply is the ceiling on it. Say’s law (“demand is made possible by supply”) reads the same relationship from the supply side: output is income, and income is what turns an unbounded want into effective demand. Neither is a rival theory. They are two ends of one loop: supply bounds demand, supply funds demand, and demand tells supply what is worth making (the marginalist end: if nobody wants it, it isn’t worth anything). Delete any one arc and you get a broken model: a demand-only model with infinite output, or a supply-only model producing things nobody values.
Chris, on finding the reverse statement: “I got trapped by the same warning I have given before — you must look at it from both angles.” The page’s own warning applies to its own sentence. “Within the limited supply” fixed the demand-driven half-argument but stayed seated at the demand end.
The mechanism is cross-sector transmission. Farm productivity rises. Competition forces the gain into lower food prices. A blacksmith’s wage now buys more food without anyone negotiating, and the money he frees up becomes new demand in some third sector. That is how productivity in one place becomes rising real wages and new demand everywhere else. It is also why growth and distribution are separate questions. Negotiation divides a pie that already exists (Value and Profit), and only productivity changes its size.
There are only two levers. Output = labor input × productivity. Population growth raises total output but not output per head. Hold labor growth steady, and the only thing that distinguishes one economy from another is productivity (Chris’s correction to the source, which left population out). AI fits here as one more productivity technology: it raises wages by raising output per hour, not by any property special to AI.
Why people fear productivity anyway: the gains don’t land on the people who bear the change. The displaced weaver pays, while every cloth buyer collects. Chris: “the cross-sector transmission is why people fear technology and increased productivity — the gains do not always land on those being affected.” The aggregate argument is correct and does not answer that fear, because the fear is about who, not how much. (Jevons for Software §3 runs the same split for developers.)
This is the scalpel that lands on MMT and its post-Keynesian relatives: endogenous money (“banks create money when they lend”), the demand-management prescription, and money-financed transfers (Keen’s Modern Debt Jubilee) all understate the supply constraint by construction. The heterodox macro is often right that mainstream models are broken; the failure mode is symmetric — trading a supply-blind neoclassical model for a demand-blind one is not an escape from the both-sides fault, just the other horn of it.