Is rent control ever defensible?
Almost every economist says no. Voters keep saying yes. They are both partly right — and which version of rent control you mean decides who wins the argument.
The crisis and the case for control
“Shall the City of Saint Paul adopt the proposed ordinance limiting residential rent increases?” — Approved.
— St. Paul, Minnesota, ballot question, November 2021 (passed ~53%, capping annual rent increases at 3%)
Rent control is the textbook case where economists agree and the public disagrees — and the public keeps winning at the ballot box. The price-ceiling diagram explains what rent control does. It does not explain why a city would vote for it anyway. That question deserves a real answer before the diagram gets one.
Start with what a rent cap actually promises, because the promise is not crazy. In a city where supply is tight, a demand shock — a tech boom, a wave of investor purchases — pushes rents up fast. A binding ceiling on rents protects the surplus that sitting tenants already enjoy: it freezes their housing cost while the market price climbs around them. For the incumbent renter facing a thirty-percent increase after a boom prices their block, that protection is the entire point. The policy promises to absorb a shock the tenant did nothing to cause.
And there is a real failure underneath the demand for it. In a supply-constrained metro, new housing cannot be built fast enough to answer a demand shock in time. Zoning, permitting, and construction lags mean the supply curve is steep and slow exactly when it most needs to be elastic. So a demand shock does not get absorbed by new building; it gets absorbed by price — and the price falls on incumbent renters as displacement. The market has failed renters in a precise sense: it cannot protect them from a shock it cannot answer quickly. That is the gap a rent cap is reaching into.
A binding ceiling $\bar{P} < P^*$ holds the sitting tenant’s rent below the market-clearing level. In the short run, before the supply response materializes, the tenant captures a transfer roughly equal to:
$$\text{Transfer to incumbent} \approx (P^* - \bar{P}) \cdot Q_{\text{occupied}}$$This is the protection the policy advertises. The cost of that protection — what happens to $Q$ once landlords respond — is Stage 2’s subject. Stage 1’s point is only that the short-run promise is real.
A renter has lived in the same apartment for nine years. A tech campus opens two miles away, and within eighteen months the going rent on her unit has risen by a third. A cap says: not to her, not this fast. Whatever else it does, that is what it is promising her — a wall between her and a shock the city built no housing to absorb.
That is the apparatus you need to evaluate the public’s case — just enough to see what the policy is for. The full price-ceiling diagram, the part that shows what the protection costs, is Stage 2. The short-run-protection reading lives in the intro chapter on price controls.
“The struggle over housing is the struggle over the city itself… To say that housing is a human right is to say that no one should be displaced from their home because they cannot pay an ever-rising rent.”
— David Madden & Peter Marcuse, In Defense of Housing, 2016
Is rent control a necessity, not a mistake?
The tenant-protection case says rent control is the only thing standing between working renters and displacement. Economists call it a mistake. But the displacement it names is real — so what exactly is the necessity claim getting right?
The case from inside the displacement
“The home is the bedrock of personal stability. Yet for tenants, that stability is conditional — it can be revoked the moment the rent rises beyond reach. Housing precarity is not a market signal. It is a form of dispossession.”
— David Madden & Peter Marcuse, In Defense of Housing, 2016
This is the housing-as-a-right argument made by the people who watch neighborhoods empty out, and it deserves to be heard in its own voice rather than translated into “critics of the market say.” The claim is not that prices should never move. It is that a renter’s shelter — the address that anchors her job, her children’s school, her community — should not be revocable at the speed of a demand shock she had no part in. When a tech campus or an investor-purchase wave lands on a supply-constrained city, the rent on an occupied unit can rise faster than any wage, and the tenant’s choice collapses to: pay the impossible, or leave the life you built. The Berlin referendum that voted 57.6% to expropriate corporate landlords was not a misunderstanding of supply and demand. It was a major Western capital deciding that the cost of displacement had crossed a line, and that protecting people from it was worth doing even if the textbook frowned. The displacement is the fact. The demand for protection follows from the fact, not from confusion.
Where this leaves us
The public is not voting irrationally. The displacement crisis is real, the costs fall on people who did nothing to deserve them, and in a supply-constrained metro the market genuinely cannot protect incumbent renters from a demand shock in time. The case for doing something is strong. The question Stage 2 asks is not whether the crisis is real — it is whether rent control is the something that helps. For the larger question of why housing got so expensive in the first place — the supply-and-zoning story this sub-debate sits inside — see Why is housing so expensive?, which owns that controversy at full depth.
But there is a reason almost every economist who has looked at hard rent control — across schools that agree on almost nothing else — comes back with the same answer. And the cleanest test of that answer happened in San Francisco.
The textbook case against
“In many cases rent control appears to be the most efficient technique presently known to destroy a city — except for bombing.”
— Assar Lindbeck, The Political Economy of the New Left, 1971
The line is hyperbolic, and Lindbeck — a Swedish social democrat, not a free-market ideologue — knew it. But it stands in for something rare: a near-consensus that crosses schools which agree on almost nothing else. And unlike most economist consensus, this one has a clean test.
Here is what the price-ceiling apparatus predicts once you let it run past Stage 1’s short-run promise. A ceiling held below the market-clearing rent does protect sitting tenants — and it sets four mechanisms in motion that, together, make the housing worse for everyone who is not already inside a controlled unit.
Supply withdrawal. When a landlord cannot charge the market rent, the rental unit becomes the least profitable use of the building. So units leave: converted to condominiums, sold to owner-occupants, turned into tenant-in-common arrangements, listed on short-term-rental platforms, or simply not built in the first place. The stock available to the next renter shrinks.
Quality degradation. If the unit rents at the cap no matter its condition, the landlord’s incentive to maintain it disappears. Deferred repairs, deferred upgrades, slow decay. The ceiling caps the rent; it cannot compel the upkeep.
Misallocation and lock-in. A protected tenant does not move even when the housing no longer fits — the empty-nester in the family-sized controlled unit stays, because leaving means re-entering the market at the uncontrolled rent. Turnover freezes; people and homes stop matching.
Reduced mobility. The same lock-in chains workers to cities. A tenant who would take a better job in another metro turns it down rather than surrender a below-market rent. The controlled unit becomes a tether.
The welfare arithmetic: a binding ceiling transfers surplus to incumbents in the short run, but the deadweight loss grows as the supply response materializes. With ceiling $\bar{P} < P^*$, the shortage is $Q_D(\bar{P}) - Q_S(\bar{P})$, and the deadweight loss is the triangle of mutually beneficial rentals that no longer occur:
$$DWL = \tfrac{1}{2}\,(P^* - \bar{P}) \cdot \big(Q^* - Q_S(\bar{P})\big)$$How large that loss grows is governed by the elasticity of housing supply: the more responsive supply is to price over time, the more units the ceiling withdraws, and the bigger the loss. In a tight metro, supply is inelastic in the short run and more elastic in the long run — which is exactly why the damage compounds with time rather than fading.
A landlord with a triple-decker does the math: at the capped rent, the rental units lose money against what the building is worth as condos. He converts. Three apartments come off the rental market and never go back. The tenant who was protected keeps her unit; the family that would have rented the unit upstairs never gets the chance, because upstairs is now a condo. That is the whole problem in one building — protection for the inside, scarcity for everyone else.
This welfare apparatus — consumer and producer surplus, deadweight loss, the way a price control trades a visible transfer for an invisible loss — is the marginalist tradition’s signature contribution to price theory, the formal machinery that lets you put a number on what a ceiling costs. The within-market supply mechanism this shares with the broader affordability question is developed in Why is housing so expensive? Stage 1; here we keep the focus on the ceiling specifically rather than re-deriving the supply-and-zoning case.
“In many cases rent control appears to be the most efficient technique presently known to destroy a city — except for bombing.”
— Assar Lindbeck, The Political Economy of the New Left, 1971
Are the economists right about rent control?
The economist consensus against rent control is unusually broad and unusually confirmed. So is it settled? Yes — about the version Lindbeck and San Francisco were looking at. The trap is assuming that verdict covers every version.
The cleanest test we have
“Landlords treated by rent control reduced rental housing supply by 15%… This led to a city-wide rent increase of 5.1%… rent control has actually fueled the gentrification of San Francisco, the exact opposite of the policy’s intended goal.”
— Rebecca Diamond, Tim McQuade & Franklin Qian, American Economic Review, 2019
This is the study the verdict rests on, and it is unusually clean. In 1994 a San Francisco ballot measure extended rent control to small multi-family buildings constructed before 1980 — which created a natural experiment, because buildings just over the cutoff stayed uncontrolled and could serve as a comparison group. Diamond, McQuade, and Qian tracked covered versus comparison buildings over two decades. The covered landlords did exactly what the apparatus predicts: they pulled units out of the rental market — redeveloping, converting to condos, moving to tenant-in-common ownership — cutting rental supply by about 15%. The tenants who stayed got real protection: they were significantly more likely to remain at their address. But the supply they removed drove citywide rents up about 5%, and the policy designed to slow displacement helped fuel the gentrification it was meant to fight. Critically, this is not one ideologue’s reading — the IGM Forum survey shows the profession agreeing across its usual fault lines that ceilings shrink the available stock. The consensus and the cleanest test point the same way, and both are looking at hard rent control.
Where this leaves us
On hard rent control — nominal ceilings below market, applied to the existing stock — the mainstream is right, and not because of ideology. The price-ceiling apparatus predicts supply withdrawal, quality degradation, misallocation, and reduced mobility, and the cleanest natural experiment we have, in San Francisco, found exactly that: a roughly 15% supply reduction, a within-renter-class transfer, and a citywide rent increase that hurt the next generation of renters to protect the current one. The IGM consensus is earned. Hard rent control is indefensible — not as a moral judgment on the tenants who need protection, but as an empirical judgment on whether this tool delivers it. That the transfer runs within the renter class rather than from rich to poor is its own indictment; the distributional-incidence apparatus that makes the point general lives in Is inequality a problem economics can solve?
And if that were the whole story, this walkthrough would be over. But the rent control that keeps passing at the ballot box in 2021 is not the rent control San Francisco tested in 1994. The modern designs changed one thing — and that one thing changes the verdict.
The second-generation nuance
Oregon Senate Bill 608 (2019): annual rent increases capped at 7% plus inflation — with a blanket exemption for any building less than 15 years old.
— Oregon SB 608, the first US statewide rent-stabilization law, 2019
The rent control San Francisco tested in 1994 capped nominal rents on the existing stock. The rent stabilization Oregon passed in 2019 caps how fast rent can rise on a sitting tenant and exempts new construction entirely. That exemption is not a footnote. It is the whole game.
Hold the two policies next to the same diagram and the difference is not one of degree. A hard ceiling tells the landlord what he may charge on a unit, full stop — which is exactly the signal that makes pulling the unit off the market the profitable move. An increase-cap with a new-construction exemption tells him something narrower: he may not raise this sitting tenant’s rent by more than the cap each year, but he may charge whatever the market bears on a vacant unit, and the cap does not touch new buildings at all. The supply curve is left free at the margin that actually determines whether housing gets built. The policy bites on the within-tenancy rent path, not on the build-or-don’t-build decision.
That is why second-generation stabilization is a genuinely different intervention, not a milder dose of the same one. The mechanism that drove the San Francisco result — landlords withdrawing units rather than renting below cost — is precisely the mechanism the new-construction exemption defuses. Builders still face market rents on what they build; the developer’s calculation is untouched. What the cap does instead is limit how fast an existing tenant’s rent can ratchet during a demand shock. It is closer to insurance than to a price control: it does not try to make housing cheap, it tries to make the rent path predictable for the person already living there.
What it costs, honestly. The insurance is not free. An increase-cap still transfers — from landlords and from future renters to incumbent sitting tenants — and the transfer is not perfectly targeted to need. The protected sitting tenant might be a long-tenured professional, not the poorest renter in the city; the cap protects whoever is already inside, regardless of whether they are the person the policy was sold to help. So the welfare case is real but bounded: it is a defensible answer to displacement volatility, not a free lunch and not affordability policy. Reading housing through the lens of what kind of good it is — market good or welfare good — is the general apparatus-flip developed in Is housing a market good or a welfare good?; the anti-displacement-insurance framing here is one instance of that welfare-good apparatus applied to the rent-regulation lever. That the transfer runs to incumbents rather than to the neediest is the same incidence point that Is inequality a problem economics can solve? makes at general scope.
The structural difference is which margin the policy constrains. A hard ceiling binds the level of rent on all covered units, including the marginal new unit, so the supply response collapses: $Q_S$ falls as units exit. An increase-cap with a new-construction exemption binds only the growth rate of rent on sitting tenancies and exempts the build margin:
$$\Delta \text{Rent}_{t} \le \text{CPI} + k, \qquad \text{(new construction exempt)}$$Because the build margin is untouched, the supply response that the hard ceiling kills is preserved. The deadweight loss from supply withdrawal — the dominant term in the San Francisco result — is sharply reduced. What remains is the incumbents-versus-future-renters transfer, which is real but is not a destruction of stock.
Two rules. Rule one: “you cannot charge more than $2,000 for this apartment.” That makes the landlord want to stop renting it. Rule two: “you cannot raise this sitting tenant’s rent by more than inflation-plus-seven-percent a year — but you can charge anything you like on a brand-new building.” That leaves the builder’s incentive alone and just smooths the shock for the person already there. Same family name, opposite effect on whether the next apartment gets built.
There is also an older tradition the insurance framing draws on without quite belonging to: treating housing as a good whose provision can sit partly outside the market — the cost-rent municipal-housing model of inter-war “Red Vienna,” where the city built and let housing at the cost of upkeep rather than the market rent. That is a different apparatus from capping rents on a private market, and it is the lineage the housing-as-a-right voice is reaching toward when it calls housing a right rather than a commodity.
“The exemption for new construction is the design feature that makes the difference. It preserves the incentive to build while still protecting existing tenants from sudden, large rent increases.”
— on Oregon SB 608’s 7%+CPI cap with a 15-year new-construction exemption, 2019
Does the second-generation design rescue rent control?
Modern rent stabilization caps increases instead of rents and exempts new buildings. Supporters say that fixes the supply problem that doomed the old version. Is it a real difference, or a relabeling?
Insurance, not affordability
“A well-designed cap on rent increases that exempts new construction functions less like a price control and more like insurance: it protects sitting tenants against the volatility of demand shocks without freezing the supply response that the housing shortage actually requires.”
— the housing-economist case for rent stabilization as displacement insurance, building on Diamond, McQuade & Qian (2019)
Argued at full strength, the insurance case is this. The San Francisco study is a verdict on hard rent control, and the modern designs learned from it. Exempt new construction, so the supply response that the hard ceiling kills is preserved. Cap the rate of increase rather than the level, so a sitting tenant is protected without the market being frozen. Set the cap high enough that it binds only in a genuine shock, not in an ordinary year. Do all three and the policy delivers what it was always actually trying to deliver — not cheap housing for everyone, but a buffer against the displacement that a demand shock imposes on people who cannot move fast enough. Displacement has real, documented costs, and insuring against them is a legitimate goal even in a world where the underlying shortage should also be fixed. The honest version of this case carries its own limit: the protection still transfers to whoever is inside, and the cap is no substitute for building. It is insurance, priced accordingly, and it asks to be judged as insurance — not as the answer to affordability.
Where this leaves us
The modern designs changed the one thing that mattered: they stopped killing the supply response. By exempting new construction and capping increases rather than nominal rents, second-generation stabilization protects sitting tenants from demand-shock displacement without the supply withdrawal that doomed the San Francisco experiment. It is not free — it still transfers to incumbents, and the protected tenant is not always the neediest renter — and it is not affordability policy. But as narrow anti-displacement insurance, the well-designed version is defensible in a way the hard version is not. The textbook verdict on rent control was right about the rent control it tested. It does not automatically extend to the rent control that keeps passing now.
So one version is indefensible and one version is defensible-as-insurance. But both versions are doing the same thing: managing the consequences of a shortage neither of them cures. Which raises the question the whole debate has been circling.
The verdict
The rent is high because supply is short. No rent policy builds an apartment.
— the first-best framing every stage of this debate has circled
Every argument so far — the displacement crisis, the price-ceiling diagram, the second-generation exemption — has orbited the same fact. The rent is high because the city under-built for a generation. Both versions of rent control act on the price. Neither one acts on the shortage.
Put the whole debate on one diagram and the verdict resolves into a comparison of two interventions. Rent control — hard or soft — operates on the price: it fixes, or caps the growth of, what a sitting tenant pays. Supply expansion operates on the quantity: it shifts the supply curve outward, so the market-clearing rent falls for everyone and the regulatory-rent wedge that the shortage created begins to close. Only one of these addresses the underlying scarcity. A rent cap can make the distribution of a shortage more humane; it cannot make the shortage smaller. Building can.
That is the sense in which even the defensible second-generation version is a symptom-treatment. It manages the distributional consequences of a supply constraint it does not relax. The disease is the shortage — the regulatory and physical limits on how much housing a city allows itself to build — and the first-best is to relax it. That is the YIMBY answer, and it is the load Why is housing so expensive? carries at full depth: the Auckland upzoning evidence, the regulatory-rent wedge, the political economy of who blocks building. This walkthrough names that first-best and points to it rather than re-arguing it — because the rent-control question is decided by what we just established, and the supply question is a different walkthrough’s territory.
“Rent control treats the symptom. The disease is that we have not built enough housing — and no cap on rents has ever built a single home.”
— the supply-first framing of the housing debate
Is rent control treating the symptom instead of the disease?
Even the defensible version of rent control manages a shortage it does not cure. So what does “defensible” actually commit you to — and what does it refuse?
Past the false binary
“Rent control: good or bad?” is the wrong question. The honest answer is layered — indefensible in its hard form, defensible as narrow insurance in its soft form, and subordinate in every form to building more housing.
— the layered verdict this walkthrough commits to
The public debate polarizes into a binary — rent control good, rent control bad — and the binary is what makes both sides wrong. The strong-tenant-protection position says rent control is the answer to the affordability crisis; it is not, because affordability requires supply and a cap builds nothing, and even the good version is insurance rather than an affordability fix. The strong-textbook position says all rent regulation is harmful; that is true of the hard version and false of the well-designed soft version, which escapes the supply-withdrawal mechanism through the new-construction exemption. Naming which layer the disagreement sits at is not fence-sitting. It is the answer. The disagreement on hard rent control is frame-level and settled; the disagreement on the soft version is a calibrated conditional on design; and both are subordinate to the supply question. That is a position, held in three parts, not a refusal to take one.
Where this leaves us
Is rent control ever defensible? Hard rent control — nominal ceilings on the existing stock — is not. San Francisco tested it and it reduced supply, degraded quality, misallocated housing, and transferred within the renter class to the next generation’s cost; the economist consensus is earned. Well-designed second-generation rent stabilization — increase-caps that exempt new construction and target displacement risk — is defensible, narrowly, as anti-displacement insurance: it protects sitting tenants from demand-shock displacement without the supply withdrawal that doomed the hard version. But neither cures the shortage. The rent is high because supply is short, and no rent policy builds an apartment. The defensible version of rent control is the narrow one — second-generation stabilization deployed as insurance against displacement while the real work, supply expansion, is done. Rent control is defensible exactly to the extent that it is humble about what it can do: protect people from a shock the market cannot absorb in time, and get out of the way of the building that actually fixes the problem.
One last thing the verdict implies: price controls are not categorically harmful. They are harmful here, on this good, in this market design — a nominal ceiling on an otherwise-free rental market. Move one good over and the answer flips. In Is healthcare a market?, procedure price controls embedded in a well-designed regulated multi-payer system — Germany, Switzerland, the Netherlands — work, because the surrounding market design carries them. The verdict on price intervention is never categorical. It depends on the good and on the design around it.
Where this leaves us
We started with a city voting for rent control against near-universal expert advice, and the question of why a place would do that. The four stages answered it without flattening the disagreement into a slogan:
- The crisis is real. In a supply-constrained metro, a demand shock displaces incumbent renters faster than any supply response can absorb. The case for protecting them is strong, and the public is not voting irrationally.
- Hard rent control fails. Nominal ceilings on the existing stock withdraw supply, degrade quality, and misallocate housing. San Francisco is the cleanest test, and it transferred within the renter class while shrinking the stock. The economist consensus is earned.
- The second-generation design is different in kind. An increase-cap that exempts new construction leaves the build margin free, so the supply-withdrawal mechanism never fires. It is insurance against displacement volatility — defensible, conditional, and not affordability policy.
- Neither cures the shortage. Both versions act on the price; only building acts on the scarcity. The defensible version is the humble one, deployed as insurance while supply expansion does the real work.
So: is rent control ever defensible? Not the hard version — San Francisco settled that. The soft version, narrowly: yes, as insurance against a displacement the market cannot absorb in time, paired with the supply expansion that actually fixes the problem. The next time someone tells you “rent control destroys cities” or “rent control is the answer to the housing crisis,” you have the tools to ask the only question that decides it — which rent control, and paired with what.