NIMBYism: What the Record Shows About the Veto Over Housing, Energy, and Data Centers

Support for housing, clean energy, data centers, and homeless services runs high in the abstract and collapses at specific addresses, in every ideology and tenure, for reasons the research literature has now measured: losses weigh roughly twice gains, a home is an undiversifiable asset, developer profit alone raises opposition by twenty points, and the hearings that decide hear almost only from opponents.

By Eunice, the desk · Collected and analyzed at 1945 ET, Friday, November 13, 2026.

The opposition has a consistent structure across fifty years of American land-use fights, and the research literature now describes it with unusual precision. Residents support housing, clean energy, data centers, and homeless services in the abstract at high rates. They oppose the same facilities at specific addresses near their homes at high rates. The gap between the two positions is not explained by ideology: renters in expensive cities register it at nearly the same level as homeowners, and liberal homeowners register it alongside conservatives. Three mechanisms account for most of the measured effect. Losses register roughly twice as heavily as equivalent gains. A home is most owners' largest asset and cannot be diversified, so its protection dominates local voting. And opposition rises measurably when a developer's profit is made salient, by twenty percentage points in the best-known experiment, an effect its authors attribute to resentment of the developer's gain rather than fear of the resident's loss. The venues that decide these questions hear almost exclusively from opponents: in the largest study of hearing participation, about two-thirds of commenters opposed the housing before them and about fifteen percent supported it. The outcomes are legible in the record: a 495-home project on a San Francisco parking lot reversed on environmental grounds in 2021 and still unbuilt in 2026; a 25,000-job Amazon campus withdrawn from Queens in 2019 while polls showed most New Yorkers approved of it; data-center projects worth $130 billion blocked or delayed in the first quarter of 2026 alone.

The gap, measured

Michael Hankinson's study of San Francisco's 2015 election and a national survey, published in the American Political Science Review in 2018, found that renters in high-rent cities express opposition to market-rate housing near them at a level similar to homeowners, while simultaneously supporting large increases in their city's overall housing supply. Hankinson calls the preferences scale-dependent: support at the scale of the city, opposition at the scale of the street. William Marble and Clayton Nall, in the Journal of Politics, found the same split running through ideology. Liberal homeowners, in their title's phrase, are where "self-interest trumps ideology": they endorse egalitarian and redistributive housing policies in principle and oppose dense housing development in their own communities, with homeownership predicting local opposition more reliably than party or professed belief. New York City's 2023 Housing and Vacancy Survey supplies the context in which these preferences operate: a net rental vacancy rate of 1.41 percent, the lowest since the survey began in 1968, with roughly 33,000 units available for rent in a city of 8.3 million.

The mechanisms in the literature

Prospect theory, the framework Daniel Kahneman and Amos Tversky published in 1979 and extended in 1992, estimates that losses weigh roughly twice as heavily as equivalent gains; their 1992 estimate of the loss-aversion coefficient was 2.25, and later meta-analyses center somewhat lower without disturbing the asymmetry. Applied to a development proposal, the accounting is structural: the costs, construction, traffic, shadow, change itself, are immediate, visible, and assigned to identifiable households, while the benefits, marginally lower regional rents, a larger tax base, grid capacity, are delayed and divided among millions. William Fischel's homevoter hypothesis, set out in his 2001 book of that name, adds the balance sheet. For most owners the home is by far their largest financial asset; unlike stockholders they cannot diversify it across communities, and Fischel argues that fear of a capital loss turns owners into systematic voters for whatever protects the asset, including restrictions on nearby density. The third mechanism is the least comfortable. Paavo Monkkonen and Michael Manville of UCLA ran a survey experiment with more than 1,300 Los Angeles County respondents and found that telling respondents a developer would earn a large profit raised opposition to a housing project by twenty percentage points, double the increase produced by a traffic warning and comparable to the neighborhood-character warning. Their published conclusion: "some opposition to housing is motivated not by residents' fears of their own losses, but resentment of others' gains." Opposition, in the experiment, could be purchased by framing alone, without changing anything about the project.

The venue

Katherine Levine Einstein, David Glick, and Maxwell Palmer coded thousands of comments from planning and zoning hearings across 97 Massachusetts communities for their 2019 book Neighborhood Defenders. The people who speak are not a cross-section of the public. They are disproportionately older, whiter, more male, and longer-tenured than their communities, and they are far more likely to own their homes; about 41 percent lived on the same block as the proposed project. Roughly two-thirds of commenters opposed the housing under consideration; about fifteen percent spoke in favor. The people who would occupy unbuilt housing have no standing in this venue: they do not yet live in the jurisdiction, and many do not know the hearing exists. In California, the environmental-review statute supplies a second venue. A September 2026 Holland & Knight study of 498 appellate decisions under the California Environmental Quality Act found that housing was the largest single category of appealed project, at 25.5 percent of decisions; about 90 percent of the challenged housing projects were infill; project approvals ultimately prevailed in 75 percent of the decisions, after the delay the litigation imposed; and earlier analyses by the same firm found that suits filed in 2020 alone challenged roughly 48,000 approved housing units, and that only 13 percent of CEQA lawsuits were brought by anyone with a prior record of environmental advocacy. In 2025 the California legislature exempted most urban infill housing from CEQA review, a statutory concession that the channel had become a general-purpose instrument of delay.

Case file: a parking lot in San Francisco

At 469 Stevenson Street in San Francisco, a developer proposed a 27-story building with 495 rental homes, 73 of them affordable, on a 28,790-square-foot surface parking lot holding 176 cars. The Planning Commission certified the project's environmental impact report on July 29, 2021. On appeal by a neighborhood consortium, the Board of Supervisors voted 8 to 3 on October 26, 2021 to reverse the certification and require further study, citing, among other things, gentrification and geotechnical questions, on a site that was a parking lot. The state housing department warned the rejection risked violating the Housing Accountability Act. A partially recirculated report followed; the project was re-approved in 2023, into a high-interest-rate construction market. As of 2026, nine years after the project was first proposed, the site remains a surface parking lot and no homes have been built.

Case file: Amazon in Queens

On November 13, 2018, Amazon announced it would split its second headquarters between Arlington, Virginia, and Long Island City, Queens: 25,000 jobs and $2.5 billion of investment at each site. New York's package comprised $1.525 billion in performance-based state incentives, including up to $1.2 billion in job-linked tax credits and a $505 million capital grant; with city tax benefits the total was universally reported as roughly $3 billion, most of it contingent tax relief on activity that had not yet occurred rather than cash in hand. Opposition organized immediately, led by state and city officials and by Representative Alexandria Ocasio-Cortez, whose district borders Long Island City, on grounds that included the subsidies, the secrecy of the negotiation, and Amazon's labor practices. Polling cut against the opposition's premise and for its strongest argument simultaneously: Quinnipiac found New York City voters approved of the deal 57 to 26 percent, while splitting 46 to 44 on the incentive package itself; Siena found 56 percent of state voters approved even with the incentives stated. On February 14, 2019, Amazon withdrew, citing the opposition of state and local politicians, and quoted a 70 percent support figure that matches neither published poll. Ocasio-Cortez, celebrating the withdrawal the same day, said: "if we were willing to give away $3 billion for this deal, we could invest those $3 billion in our district ourselves if we wanted to," naming teachers and subway repairs. The statement treated contingent tax relief as a spendable fund, which, as critics across the spectrum noted at the time, it was not. The aftermath is on the record: the centerpiece city-owned site, a Department of Education warehouse at 44-36 Vernon Boulevard, remained in that use for roughly six years; in February 2025 the city issued a request for expressions of interest to redevelop it. Amazon, for its part, expanded in Manhattan instead, leasing office space at Hudson Yards and hiring thousands of employees in New York without a headquarters campus or the Queens incentives.

The complication: data centers

The data-center fights of the last three years follow the same geometry, concentrated costs against diffuse benefits, but the local benefit side of the ledger is genuinely thinner, and the record supports part of the opposition's case. Virginia's Joint Legislative Audit and Review Commission, in its December 2024 study of the world's largest data-center market, reported industry figures of approximately 50 full-time permanent workers, about half of them contractors, at a typical 250,000-square-foot facility, against roughly 1,500 workers on site at peak construction. The fiscal side runs the other way: in the five Virginia localities with mature markets, data-center revenue supplied from under 1 percent to 31 percent of total local revenue in fiscal 2023, the 31 percent in Loudoun County. The resource side is the opposition's strongest ground: data centers consumed 26 percent of Virginia's electricity sales in 2023, and the commission's independent forecast found that unconstrained demand would double the state's total power demand within about a decade. Nationally, Data Center Watch, the tracking project whose figures Bloomberg and the banks cite, counted projects worth $130 billion blocked or delayed amid local opposition in the first quarter of 2026 alone, with at least $156 billion across 2025; the project's period figures overlap and are not additive. Opposition groups numbered more than 800 across 49 states by early 2026.

The complication: shelters and supportive housing

The facilities most often opposed on predicted-harm grounds have the most direct evidence available, and it runs against the predictions. The NYU Furman Center's study of New York City supportive housing found that property values within 500 feet of the developments dipped modestly during construction and then rose steadily after opening, performing no worse than comparable properties elsewhere. A 2026 working paper by Brown, Amoah, and co-authors found the temporal structure of the opposition itself: voters living 100 to 300 meters from newly opened homeless facilities became more supportive of homelessness spending after the facilities opened. The opposition, in their phrase, is "more prospective than retrospective": it is strongest before the facility exists and weakens on contact with it.

The responses on the record

Three bodies of evidence describe what happens when the veto is overridden or removed. Evan Mast's moving-chain research, published in the Journal of Urban Economics in 2023, traced some 52,000 residents of new market-rate buildings through six rounds of moves: building 100 new market-rate units leads 45 to 70 people to move out of below-median-income neighborhoods within about five years, as each vacated unit opens for the next household. Asquith, Mast, and Reed, in the Review of Economics and Statistics, found that large new apartment buildings in low-income areas lower rents within a few hundred meters by 5 to 7 percent relative to comparable areas farther away. At the scale of a whole city, Auckland's 2016 Unitary Plan upzoned roughly three-quarters of the city's residential land; a synthetic-control study by Greenaway-McGrevy and So found that eight years after the reform, Auckland rents were about 23 percent lower than the no-reform counterfactual, alongside a documented construction boom. New York's City of Yes, adopted December 5, 2024, is the American test now running: the Department of City Planning projects it will enable 82,000 homes over fifteen years, a projection of permitted capacity rather than of completed buildings.

What the record does not show

No study in this literature measures hypocrisy, and none needs to. The findings are consistent with sincere holders of every position involved: the loss-aversion asymmetry is a property of human cognition, the homevoter's incentive is rational given the asset, the hearing skew is a property of who can attend a weeknight meeting, and the spite component, though measured, accounts for a share of opposition rather than its whole. What the record shows is a system in which each of those individually comprehensible forces points the same way, toward refusal, and in which the beneficiaries of building, future residents, future employees, future taxpayers, are structurally absent from every room where the decision is made. The outcomes are therefore not a mystery requiring a theory of bad faith. They are the arithmetic of the rooms.