Citywide housing costs
Evidence guide
What we know—and how confidently we know it.
Housing research is not one paper or one model. The most reliable conclusions recur across places, datasets, and research designs. Each claim below keeps the takeaway, limitations, studies, and relevant reforms together.
Established evidence
Broad consensus; exact magnitude varies.
Strong evidence
Multiple credible studies converge.
Mixed evidence
Results depend materially on context.
Emerging evidence
Promising but still limited evidence.
Nine conclusions
The weight of the evidence.
Open any claim for the concise answer, the honest limitation, related policy reforms, source studies, and available research figures.
01
Established evidence
02Scarcity turns neighbors into bidders.
When more households want in than there are homes available, the highest bidders win. Everyone else pays more, accepts less, moves farther away, doubles up, or leaves.
Strong evidence
03A city cannot build homes it has made illegal.
Legalize real projects across enough land and builders respond. Auckland did this at city scale and saw a large permitting boom; narrower reforms have usually delivered less.
Strong evidence
04More homes give renters more leverage.
Across citywide reforms and individual projects, more housing means lower rents than continued scarcity. Sometimes rents fall. In fast-growing places, they may simply rise less.
Strong evidence
05One new home can open many doors.
The first resident leaves another home behind. The next resident often does too. Those moving chains carry vacancies into older homes, cheaper homes, and more neighborhoods.
Strong evidence
06Saying no to new homes does not make newcomers disappear.
When newcomers cannot rent new homes, they bid for existing ones. Project-level studies generally find that nearby construction reduces rent and displacement pressure rather than increasing it.
Strong evidence
07Legal on paper is not the same as buildable in practice.
A reform can promise density and still bury it under tiny maps, parking ratios, setbacks, fees, hearings, and delay. Good reform makes ordinary projects legal, feasible, and predictable.
Emerging evidence
08Set the rules once. Stop retrying every home.
California's SB 35 gave qualifying projects a faster, more certain path than years of discretionary review. Affordable-housing builders became some of its most visible users.
Established evidence
09Free parking is paid for through everything else.
A parking space consumes land, concrete, and money. Mandates hide that bill inside every home—even when a resident would rather have lower rent than a place for a car.
Strong evidence
Rent control can protect covered tenants while shrinking the rental supply.
San Francisco's 1994 expansion made covered tenants more likely to stay, but it also led affected landlords to convert or redevelop regulated rentals. Good policy must account for both incumbent security and housing availability.
Visual evidence
Six findings worth seeing.
These charts span rents, construction, moving chains, displacement, and reform. Each opens to a shareable version with the source and caveat attached.
New apartments → nearby rents
Nearby rents fall after large apartment buildings open.
Asquith, Mast & Reed · 2023
Upzoning → housing supply
Larger upzonings produce larger—and delayed—housing responses.
Büchler & Lutz · 2024
New homes → neighborhood change
Rent and displacement effects fade with distance from new housing.
Pennington · 2021
Featured studies
Inspect methods, findings, and limitations.
These studies have page-referenced evidence records. Filter them here, or use the Library for the complete 113-resource reading directory.
Loading study filters…
Search the complete libraryOur standard
Advocacy should make us more careful, not less.
We distinguish association from causation, permits from completed homes, land values from rents, and nominal change from change relative to a counterfactual. Results that find little or no effect stay visible. Working papers are labeled. Figures disclose whether they reproduce data or only visualize an estimate reported by the authors.