One San Francisco study measured both the cost and the benefit of rent control

A rent cap runs into the building it governs

A rent cap sets a ceiling on what a landlord may charge a sitting tenant, and every argument about it begins inside one building. When the market rent on a unit climbs above the capped rent, the difference is money the owner no longer collects, and the owner has choices: hold the tenant and absorb the gap, defer maintenance, convert the building to condominiums or owner-occupancy, or sell. The single study both camps reach for measured exactly this sequence. In 2019 the economists Rebecca Diamond, Timothy McQuade and Franklin Qian traced what happened after San Francisco's 1994 ballot measure extended rent control to small multifamily buildings, comparing treated buildings to nearly identical ones just past the cutoff.¹ They found that rent control cut the rental supply from treated landlords by 15 percent, and that the same tenants it covered were 10 to 20 percent more likely to still live at their 1994 address years later. Both figures sit in one abstract, and each camp quotes one of them.

The question is not academic in 2026. In July 2024 the Biden administration proposed denying federal depreciation write-offs to corporate landlords holding more than fifty units unless they kept annual rent increases to 5 percent, a cap its own fact sheet said would cover more than 20 million units.² California voters rejected Proposition 33, which would have repealed the state's Costa-Hawkins limits and let cities cap rents on any housing, including new construction; the state's nonpartisan analyst warned it would leave "fewer homes" available to rent,³ and the measure lost by roughly 62 percent, the third straight defeat for a California rent-control initiative after 2018 and 2020.⁴ St. Paul, which passed a strict 3 percent cap by ballot in 2021, has spent every year since softening it.⁵ On one side stand economists who argue the policy backfires on the renters it targets. On the other stand housing scholars who argue it protects them, and that the backfire is overstated.

The tenant in year one against the renter in year ten

The two camps are not mostly disagreeing about what happened. They are counting different people over different spans. The defenders' tenant is the incumbent of 1994, measured in the first years under the cap, while she is still in her apartment and has not been priced out; on that person, over that horizon, the policy plainly worked. The skeptics' renter is the newcomer of 2004 or 2026, measured after a decade of withdrawn supply has pushed market rents up; on that person, over that horizon, the policy plainly hurt. Diamond, McQuade and Qian measured both in one dataset, which is why both camps cite it honestly.¹ Part of the split is a forecast, about whether the long-run supply loss eventually swamps the near-term protection, and part is a value choice, about which of the two renters a city owes more.

A second question is whether "rent control" even names one thing. The results that show large damage measured strong, hard-to-evade regimes: San Francisco's expansion and Cambridge's pre-1995 control with vacancy control. The results that show little measured moderate regimes with vacancy decontrol and new-construction carve-outs. St. Paul is running the conversion in real time: a flat 3 percent cap in 2021, a new-construction exemption and vacancy increases of up to 8 percent plus inflation by 2023, and a permanent exemption for buildings built after 2004 in 2025, leaving perhaps 90 percent of the stock still covered but the bite much reduced.⁵ Whether the design or the label governs the outcome is unsettled, and the same tension runs through the survey evidence: skeptics say the near-unanimous panel reflects theory plus the strong-regime findings, while defenders say it ran ahead of the messier evidence on moderate caps. Both readings fit the record.

The most-cited evidence that caps kill construction is St. Paul's building slump, and it is exactly where a causal claim should slow down. After the 2021 vote, St. Paul housing production fell 81 percent from its own three-year average. But rent-control-free Minneapolis, next door, fell 88 percent over the same window, as interest rates and construction costs hit both cities.¹⁵ The councilmember who led St. Paul's rollback blamed rates and costs and noted the ordinance was "the only thing that we can do anything about."¹⁵ Rent control may well have worsened St. Paul at the margin, but the macro shock did most of the damage in a city that had no cap at all, and the 81 percent figure cannot carry a clean causal story by itself.

The horizon cuts against the defenders too, and their own researchers are the ones who show it. The Furman Center that found no short-run degradation in 2021 reported in 2024 that in older, fully rent-stabilized Bronx buildings "the average unit was operating at an annual shortfall of $1,444," as regulated rents fell behind the cost of running the building.¹⁶ Same institution, opposite-looking findings, because they measure different years: the long-run maintenance decay the skeptics predict is now surfacing in the numbers of a pro-tenant research center rather than a landlord lobby.

What would settle the factual half is a study neither side has run at scale: follow the same population of vulnerable renters through both a capped and an uncapped regime over a full decade, and count both who stays housed and what the market rent does to the newcomer. In the near term the evidence will show up in three places: whether Furman's Bronx shortfalls spread to the rest of New York's stabilized stock, whether St. Paul's softened cap lets its permits recover in a way Minneapolis's do not, and whether any large public system reinstates a strong regime worth measuring. The value question, whose renter a city owes more, no dataset will settle. On it, the economists who built the strongest case against the policy offered an alternative rather than a rebuttal: if a city wants to insure tenants against rent spikes, Diamond and her coauthors wrote, it "may be less distortionary to offer this subsidy in the form of government subsidies or tax credits" than to freeze the price and let supply adjust.¹ That keeps the 1994 tenant housed without asking the 2026 tenant to pay for it, and it is the rare place the mechanism the skeptics fear and the goal the defenders hold do not collide.