This month’s forum is in collaboration with Econlib, part of the Liberty Fund network.
In the last thirty years or so, the United States has seen persistent shortages of housing, waste disposal facilities, electricity generation, and oil refineries. More recently, we have begun to see the difficulties in building data centers, portending new shortages. This is confusing in a wealthy nation. A “shortage” occurs when there is too little of a good or service at the prevailing price. Normally, though, when something is too scarce, the price goes up, reducing the amount people want to buy; on the other side, suppliers find ways to produce more of it. Both usually happen at once, and thus the shortage quickly disappears. So how can a shortage be persistent? What is going on?
Freddie Mac has put the most persistent shortage, the housing gap, at roughly 4 million homes. Of course, there have been cost increases in land, labor, and lumber, important inputs into construction. But the shortage is largely due to the accumulated price of permits, hearings, studies, delays, and lawsuits that stand between a willing builder and a willing buyer.
In the most regulated markets, more than 30 percent of the cost of a new unit can be traced directly to regulatory delay and uncertainty, and a developer may need to charge nearly $3,000 a month on a modest apartment just to break even. None of that shows up as a line item on anyone’s ballot, which is exactly the problem: the costs are diffuse, hidden, and borne by people who do not yet live in the building.
The process that generates these costs looks, on its face, like democracy working as intended. That process is the public hearing. The origin of this practice (unknown in most of the world) rests on a plausible-sounding premise, as K. L. Einstein pointed out in an Econtalk episode, based on her book Neighborhood Defenders. The claim is that letting the neighborhood weigh in before anything gets built is the “gold standard” of good planning. (The phrase “gold standard originated in a famous, hugely cited 1969 article by Sherry Arnstein.)
That seems fair. Top-down urban renewal in the early and mid-twentieth century ran roughshod over communities, especially minority neighborhoods, with no public input at all. The siting of hazardous waste facilities and roads was based on the cost of the land, and that meant that poor neighborhoods were always singled out. The shift toward participatory review was in part a corrective to that history. A hearing is also, in principle, a sensible way to handle a real externality: a new building changes its neighbors’ traffic, light, and parking, and a public process gives those neighbors a formal say.
The trouble is that the folks who show up at the hearing are profoundly unrepresentative of the community. The most important excluded group is potential future tenants, who generally don’t yet know they would move to the new housing if it were built. But Einstein shows that the people who attend hearings are not even representative of the neighborhood as it currently exists. They are older, whiter, wealthier, and almost always opposed: only about 10-15 percent of speakers favor the project in front of them.
Planning board members, who are usually unpaid volunteers, face this same small, vocal, repeat-player audience in meeting after meeting. The rational response to being harangued for three hours is compromise by delay—another traffic study, another redesign, another few months of costs for the developer, and ultimately years of delay for people who just want to rent an apartment.
Einstein cites a Cambridge project near Porter Square that entered review as four units with one parking space each and emerged as three units with two spaces apiece. This is the kind of change that any one, single hearing can present as reasonable, but thousands of such hearings, repeated nationwide, turn into a housing shortage. Add the threat of a lawsuit, which developers describe as the real project-killer, and delay compounds into abandonment. The hidden cost, of course, is projects that are never started in the first place.
The Core Problem
Consider housing, hazardous waste, oil refineries, and data centers together, and a common pattern emerges. Everyone benefits, in aggregate, from enough housing being built somewhere, from the country having somewhere to put spent nuclear material, from having stable fuel prices, and from having enough data-center capacity to run the services people now treat as infrastructure. But no resident wants the “somewhere” to be their neighborhood.
Each locality’s individually rational move is to keep it out and let it land elsewhere, but if every locality reasons the same way, the thing nobody wanted to host does not get sited at all, and everyone is worse off than if some jurisdiction, somewhere, had said yes.
Several scholars have made this formal. Yale law professor David Schleicher frames exclusionary zoning itself as a prisoner’s dilemma among neighboring municipalities: each would prefer a regional mix of affordable and market-rate housing to concentrated poverty, but each also fears that if it alone opens itself to lower-income residents while its neighbors stay closed, it absorbs a disproportionate share of the region’s poor. Residents worry about traffic, crime, and declining schools, along with the aesthetic disadvantages of filling the neighborhood with high-rise apartments. In most cases, the tax and other benefits of saying “yes” even benefit the community that accepts the project, adding further to the attractions of mixed housing. But given the potential costs of a massive influx of low-income residents, the dominant strategy for every municipality is to exclude, even though every municipality would do better under mutual inclusion.
A national shortage measured at 4 million units keeps growing every year that construction lags household formation; every additional cycle of three-month continuances is another cohort of renters priced further from opportunity
Economist Sen Eguchi built the same logic into an explicit two-player game for siting an unwanted facility like a landfill or waste site, showing formally that, depending on how costs and benefits are distributed, the interaction between two candidate host communities collapses into either a prisoner’s dilemma or a war of attrition—two different games, but both ending in an unbuilt or badly delayed facility.
This is the reason that the remedies that have actually worked share one feature: they take the “Reject” option off the table for at least one party, rather than hoping better arguments at the microphone will change local minds. Massachusetts’s Chapter 40B “anti-snob zoning” law lets developers override local zoning boards in towns that have not met an affordability threshold, functionally forcing an “Accept” move that no individual town would choose on its own.
The federal Nuclear Waste Policy Act did the same thing for nuclear waste, designating Yucca Mountain as the national repository by statute rather than leaving the siting decision to a hearing that Nevada would rationally have voted down forever. Data-center siting has not yet found its equivalent circuit-breaker, which is one reason it is producing the same fights, town by town, that housing and nuclear waste produced before it. The remaining sections turn to what those circuit-breakers look like in practice, and where a YIMBY movement can—and cannot—build one.
The Price of No Progress
The theory says the outcomes will be bad; the data say how bad. Edward Glaeser coined the term “zoning tax” for the gap between a home’s market price and what it would cost to build absent regulatory obstacles. In their original estimates, that tax already exceeded 10 percent of housing costs in Boston, New York, and other cities. In San Francisco, regulation accounted for more than half the price of a home.
It might seem that these costs are more concentrated in particular areas, and that is right. The Wharton Residential Land Use Regulatory Index, the standard academic measure of how tightly a metro area constrains new housing, finds the most restrictive markets clustered almost entirely along the Northeast Corridor and the West Coast—San Francisco–Oakland tops the list, followed by New York, Providence, Seattle, and Los Angeles—while the least restrictive are concentrated in the Rust Belt: St. Louis, Detroit, Cincinnati, and Rochester, places where regulation is loose largely because decades of population decline mean nobody is fighting over scarce permits in the first place. Charlotte is the one large, still-growing Sun Belt metro that shows up on the light-regulation end, which is part of why it has built enough housing to keep growing.
Data centers raise similar questions. The group Data Center Watch has done the equivalent tally: it counts $64 billion in data-center projects nationally either blocked outright ($18 billion) or delayed ($46 billion) by local opposition, organized by at least 142 activist groups across 24 states, with Virginia—already the most housing-restrictive state on the East Coast—the epicenter at 42 groups and tens of billions in projects tied up. Most neighborhoods are opposed to the construction of data centers, complaining about resource pressure, noise pollution, and the sacrifice of open space to large, windowless, concrete boxes. Nevertheless, data centers must be built somewhere if people are to enjoy the advantages of fast AI and streaming services, cloud storage, and other benefits.
What’s interesting is that housing-friendly places are not exempt from the data-center version of the same fight. Texas is the clearest case. This is a state with famously light housing regulation, weak state-level environmental review, and by-right development in most of its fast-growing metros. But none of that has stopped its legislature from opening hearings on data-center water use, or stopped Fort Worth from seeing roughly $750 million in projects delayed. Reporting on the broader backlash finds it crossing party lines just as readily as it crosses regulatory regimes: organized opposition has shown up in Virginia, Ohio, Pennsylvania, and Republican-leaning Georgia and Indiana alike, enough that in some of those places opposing new data centers has become, as one account put it, a prerequisite for running for local office in either party.
That is exactly what the Prisoner’s Dilemma analysis predicts: the trap is generated by the structure of hyper-local, concentrated-cost siting decisions, not by any one state’s general appetite for regulation. A state that has solved this problem for single-family houses—Texas, by keeping most housing approvals by-right and out of the discretionary hearing process—has not automatically solved it for industrial-scale infrastructure sited under a different set of rules, because the fix has to be applied issue by issue, not inherited wholesale from a state’s political culture. One wrinkle is worth flagging: some observers, including the writer Matthew Yglesias, argue that data-center opposition also draws on a source housing opposition mostly cannot—real public skepticism that the underlying technology, AI, is good for anyone at all—which may be why it is starting to produce statewide moratoriums rather than only the town-by-town attrition that has characterized housing and energy fights.
Blame the Game
We’ve covered a lot of ground; it’s worth summarizing the trip. First, the insistence on public hearings, the “gold standard” of democratic municipal decision, is not evil. In fact, it was a mostly necessary response to top-down and opaque choices by governments. Second, nonetheless, public hearings and the use of political permitting processes create and sustain a “prisoner’s dilemma” that resists change and imposes enormous costs. Third, the costs are real, and growing. A majority of people seem to want to solve the problem, as long as there is absolutely no impact on their neighborhood.
There’s no reason to believe that the people who show up to these hearings are bad actors. Einstein’s own research suggested that this is what you would expect from ordinary, decent people acting inside a bad game: most objectors describe themselves, honestly, as protecting their neighborhood, their trees, their kids’ schools, not as exercising raw self-interest. Russ Roberts’s “bootleggers and Baptists” aside in the podcast cuts the same way—people generally believe their own altruistic framing even when a more self-interested account would also explain their behavior. “Reject” is the dominant strategy in the local-development prisoners’ dilemma, regardless of anyone’s motives; a homeowner who shows up to oppose a fourplex next door is doing exactly what the incentives in front of her reward, the same way a driver who runs a yellow light before it turns red is not a bad person for responding rationally to a badly timed intersection. The people are not the defect. The rules are.
But rules that reward individually reasonable behavior with a collectively ruinous result do not get better because the people following them are sympathetic, and the cost of leaving those rules in place is compounding, not holding steady. A national shortage measured at 4 million units keeps growing every year that construction lags household formation; every additional cycle of three-month continuances is another cohort of renters priced further from opportunity; every additional Yucca Mountain is another forty years of spent fuel sitting in temporary storage at reactor sites rather than one permanent repository; and the data-center buildout now underway is happening on a timeline set by a global technology race, not by the pace at which a county planning board can schedule its fourth traffic study.
None of this is an argument that neighborhood defenders deserve contempt. It is an argument that the country cannot afford to keep letting a hyper-local, unrepresentative hearing decide questions whose costs of delay are now measured in the trillions and whose clock, on data centers and housing alike, is not particularly patient.
