The upzoning paradox

The conventional argument runs like this. Cities have a housing affordability problem because cities have a housing supply problem. The way to add supply is to allow buildings to be bigger. Therefore, upzoning — increasing the legally permitted residential floor area — should reduce rents over time, by the simple economics of more product chasing the same demand. This argument is a near-universal premise of YIMBY policy. It informs zoning reform movements in Sacramento, Seattle, Minneapolis, Cambridge, and most other U.S. cities engaged in housing-policy reform right now.

NYC actually ran the experiment. The Bloomberg-era zoning revisions changed the legal capacity of large parts of the city between 2002 and 2010 — exactly the kind of natural experiment that urban economists fantasize about. Vicki Been and Constantine Kontokosta, working with PLUTO data, tracked what happened to those upzoned parcels over the next eight years. The result, in a 2019 path-analysis paper, was that upzoning had robust positive effects on observable physical density, on tax-assessed property values, and on visible greening (a proxy for neighborhood improvement). What it did not do was reduce rents.

This isn't a claim that supply doesn't matter. It's a more uncomfortable claim: that supply increases of the kind upzoning produces are absorbed faster than they relieve rents, and the relief isn't visible at parcel-level granularity in eight-year windows. The paper also found that upzoning didn't mitigate racial gentrification — properties that upzoned saw demographic turnover similar to comparable non-upzoned controls.

What PLUTO can show

The full counterfactual analysis requires longitudinal joins between PLUTO snapshots and rent-assessor data that we'd need to assemble over multiple deploys to access. But PLUTO's own structural distributions are themselves diagnostic. NYC's residential building stock, by year-built bucket:

The dominant epoch, by a wide margin, is 1925–1949. The Bloomberg-era upzonings opened capacity for new construction in 2000–2024, and that bucket is meaningfully populated — but nothing on the scale of the inter-war building boom. The supply increase that upzoning produced was real, but it was fractional against a stock built in a totally different policy regime.

The zoning code itself

NYC's zoning code, as expressed in PLUTO's zonedist1 column, is dominated by a small number of codes that account for most of the city's land area. The top dozen:

R-codes (residential), R-codes with letter modifiers (commercial overlays, contextual designations), and the C-codes (commercial) make up the meat of the city. The interesting detail isn't which codes are largest — it's that the system's complexity is what it is because every upzoning, every contextual rezoning, every special district designation since 1961 has added codes without removing them. The legal complexity itself is a frictional drag on construction that doesn't show up in academic analyses but is well-documented in practitioner accounts.

What the paradox suggests

None of this implies upzoning is wrong as a policy direction. The Been/Kontokosta paper found real benefits — physical density, tax base, greening, the ability to actually develop the parcel without a special permit. What it found was that the affordability theory of upzoning, in the form most policy advocates state it, oversimplifies. Adding density increases supply, and supply increases over the long term do bear on rent — but the time horizon is decades, the leakage to luxury construction is significant, and the income segments that need the rent relief most are not necessarily the ones that get it.

The honest version of the upzoning argument, post-paper, is structural rather than economic. Cities should have legally capable housing supply because the alternative is the housing politics we've had since 1961 — special permits, ULURP, contextual rezoning, and the cumulative legal complexity that you can see in the dataset above. The rent relief, if it comes, comes slowly. The structural improvement comes immediately. Both are true. The paradox is in pretending it's only one or the other.