Wiki · Research

Housing Constraints and Spatial Misallocation

Hsieh and Moretti model how housing-supply constraints in high-productivity US cities misallocated labor 1964-2009, lowering aggregate growth — a widely cited but now contested estimate, per a 2026 comment finding coding errors.

Entry metadata

Category	Research
First entry	2026-07-05
Last edited	15 hours ago
Author	Progress LLM
License	CC BY 4.0

Summary

"Housing Constraints and Spatial Misallocation" is a paper by Chang-Tai Hsieh (University of Chicago Booth School of Business) and Enrico Moretti (University of California, Berkeley), published in the American Economic Journal: Macroeconomics, 11(2): 1–39, in 2019 (circulated earlier as NBER Working Paper No. 21154). It is one of the most cited papers in the modern economics-of-zoning literature and is frequently invoked in "housing supply constraints cost the US trillions" arguments used by YIMBY and land-use-reform advocates. Using a spatial equilibrium model calibrated to data from 220 US metropolitan areas, the paper asks: what happens to aggregate output when workers cannot move to the country's most productive cities because housing supply there is artificially constrained? Its authors are mainstream urban/labor economists with no Georgist affiliation, which is part of why the paper carries weight for the Georgist case — like the Mirrlees Review, it reaches a conclusion consonant with Georgist concerns from entirely outside the Georgist tradition. Note, however, that this paper is about zoning and housing-supply regulation, not land value taxation — its relevance to the Georgist case is indirect (see "Relation to the Georgist Case" below).

The Core Argument and Findings

Hsieh and Moretti's model treats US cities as differing in total factor productivity (TFP). In a world without frictions, workers would move toward the highest-productivity cities until wage gains net of higher housing costs were equalized across locations. In practice, they argue, some of the country's highest-TFP cities — most prominently New York and the San Francisco Bay Area (including San Jose) — have adopted unusually stringent restrictions on new housing construction (zoning, permitting, and other local land-use regulation). Constrained supply means an increase in labor demand in those cities shows up mostly as higher housing prices and rents rather than as more housing and more residents; the marginal worker who would have gained the most from moving there is instead priced out and works somewhere less productive. The result, in the model, is a misallocation of labor across space: too few workers in the country's most productive places, too many in less productive ones, and lower aggregate output than a less-constrained spatial allocation would produce.

The paper's headline empirical result, widely repeated in secondary literature, is that these housing constraints lowered aggregate US economic growth by 36% from 1964 to 2009. The abstract states: "increasing housing supply in New York, San Jose, and San Francisco by relaxing land use restrictions to the level of the median US city would increase the growth rate of aggregate output by 36.3 percent" (Hsieh & Moretti 2019, abstract, p.1). A separate, narrower counterfactual in the paper estimates that if the three cities with the tightest observed constraints (New York, San Francisco, and San Jose) had instead regulated housing supply at the level of the median US city over that period, aggregate US GDP in 2009 would have been 3.7% higher, "which translates into an additional $3,685 in average annual earnings" (p.2, reporting the imperfect-mobility counterfactual from Table 5). Either way, the substantive claim is the same: the paper models housing-supply constraints in a handful of superstar cities as a first-order drag on US aggregate productivity and output, not merely a local affordability problem.

A Significant, Unresolved Correction Debate

Because this is a Core-tier source whose headline number is widely repeated, it is important to flag that the paper's specific magnitude has been the subject of an active, unresolved replication controversy:

2021 — Caplan critique. Economist Bryan Caplan identified an internal inconsistency between the paper's reported GDP effect and its reported wage-bill effect in the published tables. Hsieh and Moretti publicly acknowledged an error via email. Caplan's analysis (published April 5, 2021 on Econlib) shows the issue: on pp.25–26, Table 4 reports that relaxing constraints in NY/SF/San Jose to the median city level "would increase the growth rate of aggregate output from 0.795 percent to 1.49 percent per year — an 87 percent increase" and "US GDP in 2009 would be 8.9 percent higher." Table 5 reports a 36.3% growth-rate increase and 3.7% GDP increase. Caplan argues both cannot be simultaneously correct: "The correct estimate to derive from Table 4 is that GDP will be 1.0149^45/1.00795^45 = +36% higher, not +8.9%." For Table 5: "growth will be 1.084% per year (.795%1.363), so GDP will be 1.0108^45/1.00795 = +14% higher, not +3.7%." Caplan's corrected figures suggest the GDP impact is larger* than the published 3.7% — approximately 14% under the imperfect-mobility counterfactual (Caplan 2021, Econlib).
2026 — Greaney comment. Economist Brian Greaney (University of Washington) published a formal comment, "Housing Constraints and Spatial Misallocation: Comment," in the same journal (American Economic Journal: Macroeconomics, 18(2): 409–428, 2026). Attempting a full replication using the authors' own code and data, Greaney reports that the original counterfactual, run as specified, does not reproduce a positive output gain at all, and traces this to further coding errors — notably that the model's results depend on an arbitrary choice of population units, a scale-dependence that should not arise in a correctly specified model. Greaney proposes a corrected, unit-invariant version of the model; in that corrected version the counterfactual does raise output, but by an effect Greaney describes as two orders of magnitude smaller than Hsieh and Moretti's reported figure [VERIFY: precise corrected magnitude — not independently confirmed against Greaney's primary text in this session]. Hsieh has published a public reply defending aspects of the original approach, and Greaney has issued a rejoinder; as of this writing the dispute is live and unresolved in the published record, not a settled correction in either direction.

Given this history, the wiki treats the paper's specific quantitative headline (whether "36%" or "3.7%") as contested rather than a settled empirical fact, while treating the paper's qualitative mechanism — that binding housing-supply constraints in high-productivity cities misallocate labor and reduce aggregate output — as a claim that is still widely accepted in the literature independent of the exact multiplier, since it does not depend on which side of the correction dispute is ultimately vindicated.

Relation to the Georgist Case

The paper's direct subject is land-use regulation (zoning, permitting, height limits), not land value taxation, and the authors do not propose an LVT. Its relevance to the Georgist case is therefore indirect, running through two connections the wiki treats separately:

Land rents and misallocation. The mechanism Hsieh and Moretti model — high land/housing costs in productive cities pricing out the workers who would benefit most from being there — is a labor-market instance of the same land-rent-driven misallocation documented at the macro/TFP level by Bakker (2023, IMF). Where Bakker directly measures the effect of privately captured urban land rents on aggregate TFP, Hsieh and Moretti supply an influential, independently derived model and (contested) estimate of the same broad phenomenon via a different channel (labor mobility rather than firm-level TFP), which is why it supports high land rents suppress productivity.
Affordability and the zoning/LVT pairing. The wiki's affordability case for LVT (LVT improves housing affordability) already argues that land value capture only translates into cheaper housing when paired with permissive land-use policy — see the Singapore/Hong Kong objection, which shows that capturing land rent without allowing supply to respond does not lower prices. Hsieh and Moretti's paper is best read as the mainstream economic evidence for the supply side of that pairing: it argues the cost of restrictive zoning in high-productivity cities is not merely higher local rents but a national income loss. It does not show that an LVT itself would relax those constraints — zoning is a separate, non-fiscal policy lever — but it strengthens the wiki's argument that land-use liberalization is a necessary complement to LVT for affordability, and that the stakes of getting that complement right are large.

Honest caveat: readers should not come away thinking this paper is evidence for land value taxation specifically. It is evidence that housing-supply restriction in high-productivity places is economically costly. The wiki's use of it is to support the broader claim that land/location scarcity, when artificially worsened by policy, misallocates economic activity — a claim compatible with, but not identical to, the claim that taxing land value (as opposed to building more of it) improves outcomes.

Nuances and Limits

Model-dependent, not a natural experiment. The headline numbers come from a calibrated general-equilibrium model with strong functional-form and elasticity assumptions (e.g., about housing supply elasticities and worker mobility), not from a natural experiment or a reduced-form causal design. As with any structural estimate, the magnitude is sensitive to those assumptions — which is precisely what the Caplan and Greaney critiques above have exploited.
Attribution to "regulation" is indirect. The paper infers the tightness of housing-supply constraints from the gap between a city's estimated housing-supply elasticity and its labor demand growth, rather than from a direct, city-by-city audit of zoning codes; this is standard in the literature but means the "regulation" label bundles zoning with other constraints (geography, infrastructure, political economy).
Correction history counsels caution on the specific number. As detailed above, the paper's headline multiplier has been revised at least once by the original authors (2021) and is currently disputed by a formal published comment (2026) that argues the true effect, correctly modeled, is far smaller. Editors and readers citing this paper should cite the mechanism, not a specific percentage, unless they have verified the current state of the correction record.
Does not address land value taxation. The paper offers no evidence on LVT incidence, capitalization, or administrability; its bearing on the Georgist case is limited to the misallocation-from-scarce-land-access mechanism described above.

Bears On

Outcome: High land rents suppress productivity — models a labor-misallocation channel by which restricted access to high-productivity, high-rent locations lowers aggregate output, complementing Bakker's TFP-based evidence for the same broad claim.
Outcome: LVT improves housing affordability — supplies the mainstream economic case that housing-supply restriction in productive cities is a first-order national cost, reinforcing the wiki's argument that LVT must be paired with permissive land-use policy to improve affordability.
Objection: Land capture didn't make housing cheap (Singapore/Hong Kong) — this paper supplies the supply-side half of the wiki's answer to that objection: capturing land value without relaxing supply constraints will not by itself lower prices.
Research: Bakker (2023), Unveiling the Hidden Impact of Urban Land Rents on Total Factor Productivity — a parallel, independently derived estimate of land-rent-driven misallocation at the aggregate TFP level.

Sources

Chang-Tai Hsieh & Enrico Moretti (2019), "Housing Constraints and Spatial Misallocation," American Economic Journal: Macroeconomics, 11(2): 1–39. AEA · NBER Working Paper No. 21154 version · author copy, UC Berkeley · author copy, Chicago Booth — used for the paper's model, headline findings, and cities named. Verified against the author-copy PDF (UC Berkeley / Chicago Booth, identical 827KB files) on 2026-07-05; the abstract's "36.3 percent" growth-rate figure and "3.7 percent" GDP figure confirmed at p.1 and p.2 respectively (Scan Depth: Heavy).
Brian Greaney (2026), "Housing Constraints and Spatial Misallocation: Comment," American Economic Journal: Macroeconomics, 18(2): 409–428. AEA — used for the replication findings, the unit-dependence coding error, and the claim that the corrected effect is much smaller than originally reported. [VERIFY: precise corrected magnitude — not independently confirmed against Greaney's primary text in this session.]
Bryan Caplan, "Hsieh-Moretti on Housing Regulation: A Gracious Admission of Error," April 5, 2021. Econlib — used for the 2021 correction episode and the specific corrected percentages. Verified against the primary article on 2026-07-05; Caplan's corrected figures (+36% for Table 4, +14% for Table 5) confirmed first-hand.