Why is housing so expensive?
A diagram says “build more.” A ballot in Berlin says “expropriate the landlords.” They are not arguing about the same thing.
Auckland and the supply hypothesis
“Rents in Auckland were 28 percent lower than they would have been without upzoning … the upzoning generated an approximate 24 percent increase in long-run dwelling supply.”
— Greenaway-McGrevy, Pacheco & Sorensen, Journal of Urban Economics, 2024
In 2016 Auckland legalized denser housing on roughly three-quarters of its residential land. Six years later, the rent line had bent. This is the cleanest natural experiment the supply argument has ever had — and it points in exactly the direction the diagram says it should.
Start with the most boring possible model and watch how far it gets you. Housing is a good with a downward-sloping demand curve and an upward-sloping supply curve. Demand in a productive city rises — more jobs, higher wages, more people who want to live near both. In a normal market, supply expands to meet it and the price climbs only modestly. Zoning breaks that. A height limit, a minimum-lot-size rule, a single-family-only designation: each bolts a near-vertical cap onto the supply curve. When demand rises against a fixed quantity, the only thing that can move is price.
Edward Glaeser calls restrictive zoning “the cardinal sin of successful cities”: the places people most want to move to are precisely the ones that have made it hardest to build. The result is not a market failure in the textbook externality sense. It is a quantity restriction, and quantity restrictions have a signature.
Let demand be $D(q)$ and let zoning fix the buildable quantity at $\bar{q}$ below the free-market level. The market-clearing price is then read straight off the demand curve at the cap:
$$p_{\text{cap}} = D^{-1}(\bar{q})$$The gap between $p_{\text{cap}}$ and the cost of supplying that last unit is a regulatory rent — an income transfer from would-be renters to incumbent property holders. It is created by the rule, not by scarcity of land or labor.
Demand kept rising and the city refused to let anyone build. So the price did the rising instead. Auckland lifted the cap. The price stopped climbing. That is the whole mechanism, and the Auckland data is what it looks like when you finally test it at city scale.
There is a famous macro version of this claim, and it is worth getting exactly right because it has been mangled in both directions. Chang-Tai Hsieh and Enrico Moretti argued in 2019 that if a handful of high-productivity cities — New York, San Francisco, San Jose — had loosened their housing constraints to the level of the median US city, workers would have reallocated toward where they are most productive, and the gains would have been enormous. Their headline: housing constraints “lowered aggregate US growth by 36 percent from 1964 to 2009.” That number became the supply argument’s heavy artillery.
It did not survive contact with replication. A 2024 comment by Brian Greaney documented coding errors and a population-unit-dependence problem and found the corrected effect “two orders of magnitude smaller than what they report” — a fraction of a percent, not tens of percent. Hsieh and Moretti contest parts of the correction; the dispute is genuinely unsettled. The honest move is to let the walkthrough’s verdict not depend on the magnitude at all. The directional claim — zoning binds, and where it binds it costs real output — survives whether the number is 36 percent or a rounding error. The formal home of the price-cap diagram is Economics Ch.2 §2.5 (Price Ceilings and Floors); the government-failure framing — a rule that benefits a concentrated incumbent group at diffuse expense — sits in Ch.4 §4.1 (Externalities and government failure).
“Using a spatial equilibrium model and data from 220 metropolitan areas we find that these constraints lowered aggregate US growth by 36 percent from 1964 to 2009.”
— Chang-Tai Hsieh & Enrico Moretti, American Economic Journal: Macroeconomics, 2019
The famous misallocation number was wrong. The argument survives anyway.
For five years, “zoning costs the US economy 36 percent of growth” was the supply movement’s killer statistic. Then someone re-ran the code. The right lesson is not that the supply case collapsed — it is that the case never needed the number.
Take the supply diagnosis at its full strength, because it deserves it. When more housing is legal to build, more housing gets built, and the price stops climbing — this is not ideology, it is the most replicated finding in urban economics. Auckland upzoned and rents bent. Tokyo never zoned the way Western cities did and held real rents roughly flat across a generation of rising demand. Minneapolis abolished single-family-only zoning in 2018 and permits rose. The cities that built stayed affordable; the cities that refused to build did not. If you want to know why San Francisco costs four times what Houston does, you do not need a theory of greed or finance or culture. You need a permit office.
But the price you pay for a house is not only a supply story, and the historical layer matters for the stages ahead. Starting in the 1970s, the deregulation and securitization of mortgage finance turned the median home from a place to live into the median household’s primary financial asset — the largest single bet most families ever make. That transformation is the regime shift narrated in Economic History Ch.16 (Stagflation and the neoliberal turn); its hyperglobalized expansion runs through Ch.18 (Globalization and the great moderation), and its detonation is the spine of Ch.19 (The 2008 crisis and after) — the walkthrough Did economics cause the 2008 crisis? takes that detonation as its whole subject. Once a home is an asset, the people who own one have a financial reason to keep new supply scarce. Hold that thought; it is the engine of Stage 3.
So the diagnosis is clean and the cure is proven. Which raises the question the diagram cannot answer and that Stage 3 takes seriously: if loosening the cap works this reliably, why do so few cities do it?
Where this leaves us
Supply binds at the metro level, and the evidence for it does not depend on any contested number. Auckland is the cleanest of several natural experiments, all pointing the same way: legalize building and price restraint follows. The macro misallocation literature is in dispute and the verdict here does not lean on it. What stands is modest and solid — zoning is a quantity restriction, restrictions raise price, and where the restriction is loosened the market clears closer to cost. That is the floor of this debate. The disagreements that follow are all about what sits on top of it.
If the cure is this well-understood, the interesting question is no longer why housing is expensive in cities that refuse to build. It is why those cities refuse. But first, a detour through the version of the supply argument that promises far more than Auckland delivered.
Build, baby, build
“If we deregulated housing supply, the average price of housing would be cut in half … the building boom unleashed by deregulation would simultaneously reduce inequality, increase social mobility, promote economic growth, reduce homelessness, increase birth rates, help the environment, cut crime, and more.”
— Bryan Caplan, Build, Baby, Build, Cato Institute, 2024
This is the supply argument at maximum confidence. Caplan takes the Auckland mechanism and runs it to its limit: deregulation does not merely restrain prices, it halves them, and the spillovers solve half the problems in social policy along the way. If the Stage 1 evidence carries, why not believe him?
The maximalist case rests on a second mechanism stacked on top of supply: agglomeration. Density compounds. Productive cities exist because the marginal idea is cheaper to produce when you can walk to lunch with five other engineers; knowledge spills over, suppliers cluster, specialized labor markets thicken. Output per worker in the densest US metros runs 30 to 50 percent above the national average, and very little of that gap is composition. The same non-rivalry that drives endogenous-growth models — ideas can be used by everyone at once, so concentration multiplies them — is what makes a crowded city more than the sum of its residents.
Read that way, housing policy is industrial policy by another name. Every household priced out of a productive metro is a worker shunted to a place where they produce less; every unit of housing a city refuses to build is foregone output for the whole economy. The formal substrate — increasing returns, non-rival ideas, knowledge spillovers — lives in Economics Ch.13 §13.4 (Romer’s endogenous-growth model), where the same love-of-variety and spillover machinery that powers growth theory underwrites the case that letting people crowd into productive cities is one of the cheapest growth policies available.
The libertarian case, at full strength
“The cardinal sin of successful cities is using zoning and historical-district regulations to artificially limit supply … When the demand for a city rises, prices will rise unless more homes are built.”
— Edward Glaeser, Triumph of the City, 2011
Glaeser is the supply case made by someone who studies cities for a living, and his version is hard to beat on its own terms. The empirical record he points to is consistent across continents. Tokyo’s permissive national zoning code absorbed decades of demand into new construction and kept real rents roughly flat while every other global financial capital saw prices explode. Houston, which alone among major US cities never adopted conventional use-based zoning, sits at the bottom of the price-to-income tables for big metros. The accessory-dwelling-unit reforms that swept California and other states in the 2010s — letting homeowners add a second small unit by right — produced new housing the moment the legal barrier fell. The pattern is monotone: where building is legal, building happens, and where building happens, prices behave.
Add agglomeration and the case sharpens into something close to a moral claim. If concentrating people in productive places makes everyone richer, then a zoning board that blocks an apartment building is not merely inconveniencing a developer. It is taxing the national economy, redistributing toward incumbent homeowners, and locking would-be residents out of the highest-wage labor markets in the country — all at once, and all invisibly. Caplan’s claim that deregulation could halve prices and ripple outward into mobility, fertility, and crime is the logical endpoint of taking this seriously: if the cap is the problem, removing the cap is the solution, and the solution is enormous.
Right diagnosis, overshot dose
The diagnosis is broadly right and the dose is aspirational. No metro has actually halved its housing prices through deregulation; the Auckland and Tokyo evidence is consistent with substantial price restraint, not with cutting prices in half. Caplan’s number assumes supply elasticities no real city has demonstrated, and it quietly imports a counterfactual in which the housing stock can adjust as fast as the rules — but you cannot tear down half of San Francisco, and a city built out at low density stays mostly built for decades. It also assumes the political resistance simply melts, which is precisely the assumption Stage 3 is about to detonate. Keep the direction; deflate the magnitude. Supply restraint is real and large; supply miracles are not.
Even at its modest version, the supply diagnosis leaves one question standing, and it is the question the diagram can never answer on its own. If building works, why doesn’t the building happen?
The local-democracy trap
“Local democracies have become co-conspirators in the anti-development aspirations of the very few, at the hefty expense of the many … Moving decision-making from the hyperlocal level to the state level is the first step to fixing the broken development process.”
— Jerusalem Demsas, The Atlantic, 2021–2024
Demsas accepts everything in Stages 1 and 2. Supply is the lever; building works. She just asks the question the diagram leaves open: if the cure is this clear, who is keeping their hand on the cap — and why does pushing harder on the lever never seem to move it?
The answer is not a market failure. It is a political-economy failure, and it has a model. Zoning is not handed down by nature; it is the output of a local political process, and that process is structurally rigged toward restriction. A city’s homeowners hold a concentrated dollar stake — for most households the home is 60 to 80 percent of net worth — and they vote in local elections at high rates. The people who would benefit from new housing are future residents who do not live there yet, cannot vote there, and do not know to show up. William Fischel called the homeowner who optimizes municipal policy around their property value the “homevoter,” and the homevoter is not irrational. They are protecting the largest asset they will ever own. That housing-wealth concentration is also the heart of the inequality story — the walkthrough Is inequality a problem economics can solve? picks up the same 60-to-80-percent fact from the distributional side.
This is the logic of collective action in its purest municipal form: concentrated beneficiaries of restriction organize, diffuse losers do not. That logic has a named lineage — Mancur Olson’s account of why small, concentrated groups out-organize large, diffuse ones, the engine of the public-choice tradition surveyed in History of Economic Thought Ch.14 §14.2 (The logic of collective action); the homevoter is that logic applied to a city council. Zoning that began in 1926 as a public-health innovation — keeping the tannery away from the schoolhouse — had become, by 2026, a wealth-preservation mechanism, and the median voter at the municipal level has every incentive to keep it that way. The institutional-economics framing of the same point — rules as the output of who holds power and what they want — lives in Economics Ch.18 §18.6 (Political economy), where the general principle that concentrated losers organize more effectively than diffuse beneficiaries is exactly the homevoter dynamic at city scale.
Let $h$ be the homeowner share of the local electorate, $\bar{v}$ the per-homeowner gain from blocking new supply (preserved scarcity value), and $\bar{c}$ the per-future-resident cost of that block. Restriction wins the local vote whenever
$$h \cdot \bar{v} > (1 - h) \cdot \bar{c}$$and because future residents are not yet in the electorate, the right-hand side is systematically under-counted at the ballot box. State-level preemption changes the inequality by raising the denominator of the political unit — diluting the concentrated homevoter bloc inside a wider electorate that contains the would-be residents.
Picture the town meeting. The people in the room own homes and want their property values protected. The people who would move in if the town built more housing are not in the room — they live somewhere else, for now. The room votes to block. Every room votes to block. The only way to change the answer is to make the decision in a bigger room — the state capitol — where the future residents finally outnumber the incumbents.
This reframes the whole debate. The supply economists were right that building is the cure, and they spent a decade losing anyway, because they were arguing with the wrong people. You cannot persuade a homevoter to vote against their own balance sheet; the homevoter is not confused. So the abundance movement’s real insight is procedural, not economic: stop trying to win the argument at the city council and move the decision to a level of government where the math runs the other way. California’s 2025 zoning overhaul, which stripped much local discretion to block housing, and Minneapolis’s 2018 abolition of single-family-only zoning are not new economic theories. They are the same supply theory routed around the political body that was blocking it.
The honest version concedes what preemption costs. Local control is not only a NIMBY weapon; it is also how communities express genuine preferences about the places they live, and a state government that overrides it is making a real trade — cheaper housing for the region against self-determination for the neighborhood. The abundance frame answers that the trade is worth it, because the losers from local control are silent and numerous and the winners are loud and few. That is a defensible position. It is also a reminder that the supply cure is not free; it is paid for in political authority, taken from one level of government and handed to another.
The other heterodoxy: tax the ground-rent, don’t preempt the locality
“This association of poverty with progress is the great enigma of our times. It is the central fact from which spring industrial, social, and political difficulties that perplex the world, and with which statesmanship and philanthropy and education grapple in vain.”
— Henry George, Progress and Poverty, 1879, Introductory
There is a second heterodoxy here, older than the homevoter model and pointing at a different lever. Henry George looked at the same enigma — rising wealth, persistent housing poverty — and located it not in zoning but in land itself. Land value, George argued, is distinct from the value of anything built on it: a vacant lot in Manhattan is worth millions because of what surrounds it, not because of anything the owner did. That unearned increment — the ground-rent — is the wealth-extraction mechanism. His remedy was a land-value tax: tax away the unearned rise in land prices, leave the return to actual building untouched, and the incentive to hoard scarce urban land collapses. Notice the diagnosis is the same one the homevoter model reaches by a different road — the asset-rent component of housing is what extracts wealth from renters to owners. George’s lever is fiscal, not procedural: rather than preempt the locality, tax the ground-rent the locality is protecting. The land-rent tradition George radicalizes runs back through Ricardo’s theory of differential rent, the classical apparatus surveyed in History of Economic Thought Ch.3 §3.2 (Ricardo: trade, distribution, rent).
Supply binds, and so does the politics that produced it
The political economy is the binding constraint on the Stage 1 cure. Supply works, but the body that controls supply is captured by the people who profit from scarcity, and no amount of better economics persuades them out of their own interest. For US conditions the lever the evidence endorses is state-level preemption — route the decision around the captured municipality. But George marks the alternative diagonal: a land-value tax attacks the same wealth-extraction mechanism without requiring any preemption at all; it needs fiscal-political will instead. Both diagnose ground-rent as the extraction engine. They disagree only on whether you defeat it by moving the decision upward or by taxing the rent away. Either way, the diagram was never the problem. The politics was.
There is a move that goes further than preemption and further than a land tax. What if the frame itself — housing as a market — is the thing to reject? Berlin, in 2021, voted to find out.
Berlin, Vienna, and the verdict
“57.6% of Berlin voters approved the proposal to socialize the residential property of large landlords — those with more than 3,000 units — affecting around 240,000 apartments.”
— Deutsche Wohnen & Co. enteignen referendum, Berlin, September 2021
A democratic majority in a major European capital voted to expropriate its corporate landlords. The vote was non-binding and an expert commission later concluded the constitution permitted it; the political fact was unmistakable either way. The voters of Berlin did not have a market-failure theory of housing. They had a property-rights theory — and they were not arguing about the supply curve at all.
Before reaching for expropriation, the textbook offers a milder lever, and it is worth seeing why the textbook itself is skeptical of it. Rent control is the obvious market-side answer to unaffordable rents: cap the price. The cleanest modern study is Diamond, McQuade, and Qian’s 2019 analysis of San Francisco’s 1994 rent-control expansion. Landlords of covered buildings responded by pulling units out of the rental stock — converting to condos and tenancies-in-common, redeveloping, or selling to owner-occupiers — cutting rental supply for future renters by about 15 percent. Protected tenants gained; new arrivals faced a smaller, pricier market that roughly offset those gains. The net effect was long-run redistribution within the renter class rather than an expansion of housing.
That is the price-ceiling diagram doing its standard work: a binding cap below the market price withdraws supply, transfers surplus among consumers, and creates deadweight loss. The welfare arithmetic — consumer and producer surplus, the deadweight triangle a price ceiling carves out — sits in Economics Ch.3 §3.4 (Consumer and producer surplus), with the deadweight-loss treatment in §3.6 and the underlying price-ceiling mechanics back in Ch.2 §2.5. But here is the trap waiting for anyone who files Berlin and Vienna under “rent control on steroids”: they are not on this diagram at all.
“Around 60 percent of Vienna’s residents live in social or limited-profit housing; the city sustains the system at roughly 0.25 percent of GDP a year, and Viennese tenants spend on average about 18 percent of income on rent.”
— Composite from IURC and OECD housing data, 2025
Vienna proves decommodification works. The question is whether it scales.
Vienna is the existence proof the whole decommodification frame points to: a major Western city where most people rent from the public sector and housing costs a fraction of what it does elsewhere. The catch is not whether it works. It is what it took to build.
Here is the move that the whole stage turns on. The textbook treats housing as a market good with a supply curve, and that treatment is a modeling choice, not a law of nature. Vienna’s structurally low median housing cost is not a market clearing at a lower price. It is a policy outcome — the result of deciding that most housing should sit outside the market entirely. Once you see that the market frame is a choice, the Berlin vote stops looking like an outburst and starts looking like a coherent alternative theory: if housing is the median household’s shelter and not its casino chip, why should it be priced like a casino chip at all?
Berlin’s 57.6 percent was not a deviation from European norms but a surfacing of a logic already written into the German constitution. The Grundgesetz, in Article 15, explicitly provides for socialization of land and means of production into common ownership against compensation. The legal scaffolding for treating large-scale landlordism as something a democracy may simply end was already there; the referendum was a society deciding whether to use it. That is not rent control. It is a contest over the property-rights structure of housing itself.
And these are not failed experiments dressed up as success. Red Vienna built roughly 65,000 municipal apartments — the Gemeindebauten — between 1923 and 1934, financed by a luxury-consumption tax designed explicitly to redistribute, and that stock is the foundation the modern system still runs on, narrated as part of the interwar settlement in Economic History Ch.12 (Interwar monetary collapse). Singapore’s Housing and Development Board now houses around 80 percent of the population in homes most residents own on long leases — a developmental-state model surveyed in Ch.14 (The postwar golden age and decolonization), with the broader East Asian developmental-state context in Ch.17 (China’s reform and the Asian century). These are mature systems that have run for generations. The decommodification frame is not asking whether they could work. It is asking why the rest of the rich world decided housing had to be a market.
The financialization diagnosis
“Financialization refers to the increasing dominance of financial actors, markets, practices, measurements and narratives … and the resulting structural transformation of economies, firms, states and households. Due to the financialization of housing in today’s market, housing risks are increasingly becoming financial risks.”
— Manuel Aalbers, The Financialization of Housing: A Political Economy Approach, Routledge, 2016
There is a frame that explains why the supply restriction of Stage 3 is so durable, and it is not on any of the diagrams so far. Manuel Aalbers’ financialization thesis says the post-1970s transformation of mortgage finance — securitization, housing-as-collateral, the household balance sheet absorbing excess global capital — did not merely make housing expensive. It rewired what housing is for. Once a home is the median household’s primary financial asset and the foundation of the entire credit system, restricting new supply is not a regulatory accident. It is the rational behavior of an asset-holding voter base protecting the value of its holdings, ratified by a financial system that needs house prices to keep rising. This is the diagnosis that names why the homevoter wins: the homevoter is not a quirk of local democracy but the political expression of a financial order that turned shelter into the largest asset class on earth. It connects directly to the 2008 story — when housing-as-financial-asset broke, it broke the world economy — the subject of the walkthrough Did economics cause the 2008 crisis?.
The verdict
Housing is expensive because three layers stack and no single frame captures them all. Zoning binds at the metro level, and supply expansion is the right lever — the Auckland evidence is solid and the maximalist halving claim overshoots it. The political economy that produces zoning is the binding constraint on that cure, so for US conditions state-level preemption beats persuasion. And part of the world rejects the market frame itself: Berlin voted to expropriate and Vienna proves decommodification works — where the century-long political prerequisite has been paid. Around those three sit two heterodox diagonals that sharpen rather than replace them: Aalbers’ financialization names why the supply restriction holds — not as a regulatory mistake but as an asset-holder order defending itself — and Henry George’s land-value tax attacks the same ground-rent extraction without requiring expropriation. The housing crisis is solvable. Supply-and-preemption is the cheapest path; decommodification is the most generous; an LVT is the cleanest theoretical lever where political consensus for it exists. None of them is free, and the cost of each — political authority, fiscal commitment, state capacity — is itself the substantive question, not a market-design footnote. The right policy question is conditional on which of the three layers a polity actually intends to address.
The shape of this dispute — a mainstream apparatus producing a clean diagnosis, a political economy producing the lever, and decommodification advocates reframing the question as one about property rights rather than markets — is not unique to housing. It appears again, almost beat for beat, in healthcare and in climate.
Where this leaves us
We started with a diagram that says “build more” and a ballot in Berlin that says “expropriate the landlords,” and the point of the four stages was that these are not rival answers to one question — they are answers to three different questions stacked on top of each other:
- Supply binds at the metro level. Zoning is a quantity restriction; loosen it and price restraint follows. Auckland is the cleanest proof, and the verdict does not depend on the contested Hsieh-Moretti macro number.
- The maximalist version overshoots. Agglomeration makes the supply case a growth argument, but no city has halved its prices through deregulation. Right diagnosis, aspirational dose.
- The politics is the binding constraint on the cure. Homevoters profit from scarcity and control the cap, so the cure runs through state-level preemption — or, on George’s diagonal, through taxing the ground-rent the scarcity creates.
- Part of the world contests the frame itself. Berlin and Vienna treat housing as a welfare good, not an asset; financialization explains why everyone else treats it as an asset. Decommodification works where the political prerequisite has been paid.
Build more where you can, preempt local restriction where you must, and decide as a polity whether housing is an asset class or a welfare floor — because that last question is the one the textbook cannot answer for you. The supply curve tells you what happens if you lift the cap. It is silent on whether the thing being priced should have been a market in the first place.