CoChat
Sign Up
Cited by 70Housing Market and EconomicsUrban and Rural Development ChallengesUrban Planning and Governance

The impact of upzoning on housing construction in Auckland

Open access
Greenaway-McGrevy, Ryan, Phillips, Peter C.B.·2023·Journal of Urban Economics
DOI 10.1016/j.jue.2023.103555
Open this paper in CoChat — free account

AI snapshot

Written from the paper’s full text.

TL;DR

This paper uses Auckland’s 2016 large-scale upzoning under the Auckland Unitary Plan as a quasi-experiment and finds strong evidence that upzoning increased housing construction. The estimated spillover-robust effect is 27,183 additional dwellings permitted over the five years after upzoning, about 5.13% of Auckland’s dwelling stock.

Key findings

  • Upzoning stimulated construction: conventional DID estimates show 21.52 additional permits on average in upzoned areas versus non-upzoned areas in each of 510 Statistical Areas by 2021, corresponding to 10,975 additional permits citywide; cumulating 2016–2021 yields 32,869 additional permits.
  • Treatment effects remain statistically significant even under implausibly large displacement: the preferred specification remains significant when counterfactual sets span about seven times the extrapolated pre-treatment trend in control areas (7.84 = 8,768/1,118).
  • Spillover-robust estimates imply 27,183 additional dwellings were permitted over the five years following upzoning, approximately 5.13% of the Auckland region’s dwelling stock; completed construction is likely between 4.87% and 5.08% given 95–99% completion rates.
  • Effects are mostly attached dwellings: 2021 treatment effect is 16.83 for attached and 4.69 for detached dwellings; by zone, spillover-robust cumulative totals are 8,290 permits for THA, 11,769 for MHU, and 11,215 for MHS.

Methods

  • Quasi-experimental difference-in-differences design using annual geocoded building permits for new dwelling units issued by Auckland Council from 2010 to 2021, matched to pre- and post-AUP planning zones.
  • Treatment areas are residential land upzoned to Terrace Housing and Apartments, Mixed Housing Urban, or Mixed Housing Suburban; control areas are non-upzoned residential land, with 510 Statistical Areas as the geographic unit.
  • To address negative spillovers/displacement, the paper adapts Rambachan and Roth’s set identification approach, using pre-treatment trends in control-group outcomes to construct counterfactual sets and spillover-robust confidence intervals.

Limitations

  • The paper states it lacks precise measures of dwellings demolished when properties are redeveloped because demolition permits are only required for buildings less than three storeys.
  • The paper states permits in upzoned areas were still trending upwards as of 2021, so the full impact of the policy will likely not be known for several more years.
  • The paper states it does not explicitly identify factors that mediate the efficacy of zoning reforms in different contexts; evident from the text, the analysis is short-run, uses permits rather than completed construction, and the allocation of spillovers across treatment zones relies on baseline construction weights.

Model-written summary. Check it against the paper before relying on it.

Integrity signals

Registry

Crossref · 10.1016/j.jue.2023.103555

Cites

39 of 39 references verified0 citation errors · 0 not checkable

Statistics

Nothing to recompute.

Claims the paper makes

3 of 16 cited sources checked · 2 supported · 0 flagged · accuracy not yet measured

  • Premisecitation
    § Introduction

    Present techniques for addressing spatial spillover effects require the spillovers to be localized, with the magnitude of spillover between geographically distant areas assumed negligible.

    “Present techniques for addressing spatial spillover effects require the spillovers to be localized, in the sense that the magnitude of the spillover between geographically distant areas is assumed to be negligible (Clarke, 2017; Butts, 2021; Huber and Steinmayr, 2021).”
    Quote in the paperSubstantial claim1 of 3 sources checked
    • cites Clarke, 2017
    • cites Butts, 2021 source
    • cites Huber and Steinmayr, 2021 source
  • Findingown data
    § Results

    By 2021, six years after the policy was introduced, some 21.52 additional permits are issued, on average, in upzoned areas compared to non-upzoned areas in each of the 510 SAs.

    “By 2021, six years after the policy was introduced, some 21.52 additional permits are issued, on average, in upzoned areas compared to non-upzoned areas in each of the 510 SAs.”
    Quote in the paperSubstantial claim
  • Findingown data
    § Spillover Effects

    Permits in upzoned areas trend upwards from 2016 onwards.

    “permits in upzoned areas trend upwards from 2016 onwards.”
    Quote in the paperSubstantial claim
  • Findingown data
    § Set Identification

    We actually observe 0.44 fewer permits in the control group in 2021 relative to 2015.

    “However, we actually observe 0.44 fewer permits (since ˆ ϕT = −0.44) in the control group.”
    Quote in the paperSubstantial claim
  • Findingown data

    Permits in non-upzoned areas trend upwards until reaching a peak in 2015, a year before the policy change in late 2016, before trending downwards to a nadir in 2019.

    “permits trend upwards until reaching a peak in 2015, a year before the policy change in late 2016, before trending downwards to a nadir in 2019.”
    Quote in the paperSubstantial claim
  • Classified as probably not a substantial claim. 20
  • May combine two claims. 35

  • Classified as probably not a substantial claim.

    Premisecitation
    § Introduction

    The Rambachan and Roth set identification approach extrapolates pre-treatment trends to generate a set of counterfactual outcomes in the treatment group.

    “RR extrapolate pre-treatment trends to generate a set of counterfactual outcomes in the treatment group.”
    Quote in the paperSubstantial claim1 of 1 sources checked
    • cites Rambachan and Roth, 2023 source

  • Classified as probably not a substantial claim.

    Premisecitation
    § Robustness Checks

    Abstracting from external effects of increased density or new construction, upzoned land should be priced higher than non-upzoned land due to its greater potential floorspace capacity, all else equal.

    “Abstracting from external effects of increased density or new construction, upzoned land should be priced higher than non-upzoned land due to its greater potential floorspace capacity (Turner et al., 2017; Greenaway-McGrevy et al., 2021), all else equal.”
    Quote in the paperSubstantial claim1 of 2 sources checked
    • cites Turner et al., 2017

      Couldn’t be matched to a paper in Crossref, so it can’t be checked.

    • cites Greenaway-McGrevy et al., 2021 source

      The Introduction reports that Auckland relaxed land-use rules to enable more intensive redevelopment and that the change appeared to favor land-intensive housing relative to existing capital-intensive structures. It does not state that upzoned land should be priced higher than otherwise comparable non-upzoned land because of greater potential floorspace capacity. (LLM, conf=0.77)

      Experimental — accuracy not yet measured

  • May combine two claims.

    Findingown data
    § Spillover Effects

    In the all dwellings sample, the period-effect coefficients trend upwards until 2015; thereafter they trend down until 2019.

    “In the all dwellings sample, the coefficients trend upwards until 2015; thereafter they trend down until 2019.”
    Quote in the paperNot a substantial claim

  • May combine two claims.

    Findingown data
    § Spillover Effects

    There is a mild upward trend in permits built in non-upzoned areas until 2015, one year prior to policy implementation; thereafter, there is a mild decrease until 2019.

    “There is a mild upward trend in permits built in non-upzoned areas until 2015, one year prior to policy implementation. Thereafter, there is a mild decrease until 2019.”
    Quote in the paperSubstantial claim

  • May combine two claims.

    Findingown data
    § Spillover Effects

    The coefficients for the remaining variables are smaller in magnitude and statistically indistinguishable from zero in all periods.

    “The coefficients for the remaining variables are smaller in magnitude and statistically indistinguishable from zero in all periods.”
    Quote in the paperSubstantial claim

  • May combine two claims.

    Findingown data
    § Dataset

    The difference in permits between upzoned and non-upzoned areas remained remarkably consistent even as permits in non-upzoned areas more than doubled between 2011 and 2015.

    “This difference remains remarkably consistent even as permits in non-upzoned areas more than doubled between 2011 and 2015.”
    Quote in the paperSubstantial claim
    • Rests on an assumption: Building permits are a valid proxy for housing construction.

  • May combine two claims.

    Findingown data
    § Dataset

    By 2019, there were more attached dwelling permits than detached in upzoned areas, consistent with the upzoning goal of incentivising more capital intensive structures.

    “By 2019, there were more attached dwelling permits than detached, consistent with the upzoning goal of incentivising more capital intensive structures.”
    Quote in the paperNot a substantial claim
    • Rests on an assumption: Building permits are a valid proxy for housing construction.

  • May combine two claims.

    Findingown data

    The most appropriate alternative treatment date, 2016, results in larger estimated treatment effects because permits in non-upzoned areas are lower in 2016 than 2015.

    “the most appropriate alternative treatment date, 2016 (the year the policy is enacted), results in larger estimated treatment effects because permits in non-upzoned areas are lower in 2016 than 2015.”
    Quote in the paperNot a substantial claim

  • Probably not a substantial claim: may not pass S3 (load-bearing). May combine two claims.

    Findingnone given
    § Institutional Background

    Parcels zoned as THA, MHU or MHS collectively comprise 75.1% of residential land, while SH accounts for the remaining 24.9%.

    “Parcels zoned as THA, MHU or MHS collectively comprise 75.1% of residential land, while SH accounts for the remaining 24.9%.”
    Quote in the paperNot a substantial claim

  • May combine two claims.

    Findingown data
    § Set Identification

    The confidence intervals for the set-identified treatment effects sit above zero for 2018, 2019, 2020 and 2021, meaning we can reject the null of no treatment effect at the 95% level when allowing for a counterfactual set defined under equation 2 with M=5.

    “Notably, the confidence intervals sit above zero for 2018, 2019, 2020 and 2021, meaning the we can reject the null of no treatment effect at the 95% level when allowing for a counterfactual set defined under 2 with M = 5.”
    Quote in the paperSubstantial claim

  • Probably not a substantial claim: may not pass S3 (load-bearing).

    Findingown data
    § Spillover Effects

    The patterns in permits are also evident in the estimated period effects depicted in Figure 7.

    “These patterns are also evident in the estimated period effects (i.e., n ˆ ϕs o ), which are depicted in Figure 7.”
    Quote in the paperNot a substantial claim

  • May combine two claims.

    Findingown data
    § Spillover Effects

    The initial upward trend in the control group, and subsequent downturn from 2015 onwards, is most evident in the detached dwellings sample, which is also consistent with a negative spillover.

    “The initial upward trend in the control group, and subsequent downturn from 2015 onwards, is most evident in the detached dwellings sample, which is also consistent with a negative spillover.”
    Quote in the paperNot a substantial claim

  • May combine two claims.

    Findingown data
    § Results

    The estimated treatment effect for 2021 is 16.83 for attached dwellings and 4.69 for detached dwellings.

    “The estimated treatment effect for 2021 is 16.83 for attached dwellings and 4.69 for detached dwellings.”
    Quote in the paperSubstantial claim

  • May combine two claims.

    Findingown data
    § Residential Zones

    For MHS, the corresponding estimates are 0.0, 1.6, 4.1, 4.5, 5.5 and 6.2, yielding a cumulative total of 22.0 permits per Statistical Area, or 11,215 in total.

    “Finally, for MHS the corresponding estimates are 0.0, 1.6, 4.1, 4.5, 5.5 and 6.2, yielding a cumulative total of 22.0 permits per Statistical Area, or 11,215 in total.”
    Quote in the paperSubstantial claim

  • Probably not a substantial claim: may not pass S3 (load-bearing).

    Findingnone given
    § 5.5.2 Alternative Treatment Dates and Samples

    Permits in non-upzoned areas peaked in 2015.

    “permits in non-upzoned areas peaked in 2015”
    Quote in the paperNot a substantial claim

  • May combine two claims.

    Findingown data
    § Detached Housing

    For detached housing, the counterfactual extrapolated trend is 0.915 in 2018, while the counterfactual set spans (-0.585, 2.415).

    “The counterfactual extrapolated trend is 0.915 in 2018, while the counterfactual set spans (−0.585, 2.415).”
    Quote in the paperSubstantial claim

  • May combine two claims.

    Findingown data
    § Set-Identified Treatment Effects by Zone

    Over 2010-2014, 16.0% of permits were in areas upzoned to THA, 31.0% were in upzoned to MHU, and 53.1% were in upzoned to MHS.

    “Over 2010-2014, 16.0% of permits were in areas upzoned to THA, 31.0% were in upzoned to MHU, and 53.1% were in upzoned to MHS.”
    Quote in the paperSubstantial claim

  • Probably not a substantial claim: may not pass S3 (load-bearing).

    Findingown data
    § 8.6.1 Alternative Treatment Dates

    The upzoned differential for the 2013 treatment date is 34,784 permits.

    “The corresponding upzoned differentials for these three treatment dates are 34,784, 33,464 and 35,789 permits, respectively.”
    Quote in the paperNot a substantial claim
    • The authors state this as an assumption: In the absence of the upzoning policy, dwelling permits in upzoned and non-upzoned areas would have followed the specified counterfactual trend (e.g., linear trend).

  • Probably not a substantial claim: may not pass S3 (load-bearing).

    Findingreasoning
    § Attached Housing

    The 0.374 additional permits per SA implies an additional 191 permits in total across the 510 statistical areas under the extrapolated counterfactual scenario.

    “This implies an additional 191 (= 0.374 × 510) permits in total across the 510 statistical areas under the extrapolated counterfactual scenario.”
    Quote in the paperNot a substantial claim

  • May combine two claims.

    Findingown data
    § Set-Identified Treatment Effects by Zone

    Treatment effects in all three treatment zones (THA, MHU, MHS) are generally positive and statistically significant in the post-treatment period.

    “Treatment effects in all three areas are generally positive and statistically significant in the post-treatment period.”
    Quote in the paperSubstantial claim

  • May combine two claims.

    Findingnone given
    § Set-Identified Treatment Effects by Zone

    Upzoned to MHS is allocated 53% of the spillover, whereas MHU receives only 31%.

    “However, upzoned to MHS is allocated 53% of the spillover, whereas MHU receives only 31%.”
    Quote in the paperNot a substantial claim

  • Probably not a substantial claim: may not pass S3 (load-bearing).

    Findingown data
    § 8.6.1 Alternative Treatment Dates

    Constraining the counterfactual to the linear trend results in 33,191 additional permits when 2016 is used as the treatment date.

    “Constraining the counterfactual to the linear trend results in 29,866, 27,786 and 33,191 additional permits when 2013, 2014 and 2016 are used as the treatment date, respectively.”
    Quote in the paperNot a substantial claim
    • The authors state this as an assumption: In the absence of the upzoning policy, dwelling permits in upzoned and non-upzoned areas would have followed the specified counterfactual trend (e.g., linear trend).

  • Probably not a substantial claim: may not pass S3 (load-bearing).

    Findingnone given
    § 6 Discussion

    Permits for attached dwellings are still trending upwards.

    “Permits for attached dwellings are still trending upwards”
    Quote in the paperNot a substantial claim

  • May combine two claims.

    Findingown data
    § 8.6.5 Land Prices

    Land prices in upzoned areas increase relative to non-upzoned areas soon after announcement, but not before.

    “Land prices in upzoned areas increase relative to non-upzoned areas soon after announcement, but not before”
    Quote in the paperSubstantial claim

  • May combine two claims.

    Findingown data
    § Residential Zones

    For MHU, the corresponding estimates are 0.3, 1.4, 3.1, 5.5, 5.8 and 7.0, yielding a cumulative total of 23.1 permits per Statistical Area, or 11,769 across all statistical areas.

    “For MHU, the corresponding estimates are 0.3, 1.4, 3.1, 5.5, 5.8 and 7.0, yielding a cumulative total of 23.1 permits per Statistical Area, or 11,769 across all statistical areas.”
    Quote in the paperSubstantial claim

  • Probably not a substantial claim: may not pass S3 (load-bearing).

    Findingown data
    § 8.6.1 Alternative Treatment Dates

    Constraining the counterfactual to the linear trend results in 27,786 additional permits when 2014 is used as the treatment date.

    “Constraining the counterfactual to the linear trend results in 29,866, 27,786 and 33,191 additional permits when 2013, 2014 and 2016 are used as the treatment date, respectively.”
    Quote in the paperNot a substantial claim
    • The authors state this as an assumption: In the absence of the upzoning policy, dwelling permits in upzoned and non-upzoned areas would have followed the specified counterfactual trend (e.g., linear trend).

  • Probably not a substantial claim: may not pass S3 (load-bearing).

    Findingown data
    § 8.6.1 Alternative Treatment Dates

    The upzoned differential for the 2014 treatment date is 33,464 permits.

    “The corresponding upzoned differentials for these three treatment dates are 34,784, 33,464 and 35,789 permits, respectively.”
    Quote in the paperNot a substantial claim
    • The authors state this as an assumption: In the absence of the upzoning policy, dwelling permits in upzoned and non-upzoned areas would have followed the specified counterfactual trend (e.g., linear trend).

  • May combine two claims.

    Findingown data
    § Attached Housing

    For attached housing, there is only a slight upward trend in attached dwelling permits prior to treatment, with the extrapolated trend yielding only an additional 0.374 permits in 2021 compared to 2015.

    “Evidently, there is only a slight upward trend in attached dwelling permits prior to treatment, with the extrapolated trend yielding only an additional 0.374 permits in 2021 compared to 2015.”
    Quote in the paperNot a substantial claim

  • Probably not a substantial claim: may not pass S3 (load-bearing).

    Findingreasoning
    § 5.5.2 Alternative Treatment Dates and Samples

    Estimates of the cumulative increase in the number of permits are slightly larger when alternative treatment dates are used.

    “Estimates of the cumulative increase in the number of permits are slightly larger when alternative treatment dates are used”
    Quote in the paperNot a substantial claim

  • Probably not a substantial claim: may not pass S3 (load-bearing).

    Findingown data
    § 8.6.2 Alternative Treatment and Control Areas

    In the design where downzoned areas are removed from the baseline sample, the upzoned differential is 34,124 permits.

    “The upzoned differential is 34,124 permits.”
    Quote in the paperNot a substantial claim
    • The authors state this as an assumption: In the absence of the upzoning policy, dwelling permits in upzoned and non-upzoned areas would have followed the specified counterfactual trend (e.g., linear trend).

  • May combine two claims.

    Findingown data
    § Set-Identified Treatment Effects by Zone

    The treatment effect for upzoned to MHU was 8.0, which is indistinguishable from the 8.0 treatment effect in MHS (rounded to 1 d.p.).

    “For example, the treatment effect for upzoned to MHU was 8.0, which is indistinguishable from the 8.0 treatment effect in MHS (rounded to 1 d.p.).”
    Quote in the paperNot a substantial claim

  • Probably not a substantial claim: may not pass S3 (load-bearing).

    Findingown data
    § 8.6.1 Alternative Treatment Dates

    The upzoned differential for the 2016 treatment date is 35,789 permits.

    “The corresponding upzoned differentials for these three treatment dates are 34,784, 33,464 and 35,789 permits, respectively.”
    Quote in the paperNot a substantial claim
    • The authors state this as an assumption: In the absence of the upzoning policy, dwelling permits in upzoned and non-upzoned areas would have followed the specified counterfactual trend (e.g., linear trend).

  • May combine two claims.

    Study design and conductreasoning
    § Introduction

    Adapting the Rambachan and Roth method to this application yields a confidence set of treatment effects that is robust to spillover effects and amenable to inference.

    “Adapting the RR method to our application yields a confidence set of treatment effects that is robust to spillover effects and amenable to inference.”
    Quote in the paperSubstantial claim

  • Probably not a substantial claim: may not pass S3 (load-bearing).

    Premisecitation
    § Introduction

    Limb and Murray (2022) argue that transit-oriented upzoning in Brisbane, Australia, generated no significant increase in housing construction.

    “Meanwhile, Limb and Murray (2022) argue that transit-oriented upzoning in Brisbane, Australia, generated no significant increase in housing construction.”
    Quote in the paperNot a substantial claim0 of 1 sources checked
    • cites Limb and Murray, 2022

      Couldn’t be matched to a paper in Crossref, so it can’t be checked.

  • May combine two claims.

    Interpretationown data
    § Set Identification

    The set-identified confidence intervals are centered at a point below βT, which is consistent with a negative spillover effect.

    “They are also centered at a point below ˆ βT , which is consistent with a negative spillover effect.”
    Quote in the paperNot a substantial claim

  • Probably not a substantial claim: may not pass S3 (load-bearing).

    Premisenone given
    § Institutional Background

    The final version of the Auckland Unitary Plan became operational on 15 November 2016.

    “the final version became operational on 15 November 2016.”
    Quote in the paperNot a substantial claim

  • May combine two claims.

    Interpretationown data
    § Spillover Effects

    The upward and downward trend in the period fixed effects evident in Figure 7 is gone, and the estimated coefficients are very close to zero, suggesting that the set of variables Xi explains time series variation in permits in non-upzoned areas prior to 2015.

    “The upward and downward trend in the period fixed effects that is evident in Figure 7 is gone, and the estimated coefficients are very close to zero, suggesting that our set of variables Xi explains time series variation in permits in non-upzoned areas prior to 2015.”
    Quote in the paperSubstantial claim

  • May combine two claims.

    Interpretationreasoning
    § Dataset

    The post-2015 decline in permits in non-upzoned areas is consistent with a negative spillover effect, as construction that would otherwise have occurred in non-upzoned areas may have been relocated to upzoned areas as a result of treatment.

    “This outcome is consistent with a negative spillover effect, as construction that would otherwise have occurred in non-upzoned areas may have been relocated to upzoned areas as a result of the treatment.”
    Quote in the paperSubstantial claim
    • Rests on an assumption: Building permits are a valid proxy for housing construction.

  • May combine two claims.

    Study design and conductown data
    § Results

    There is no evidence of anticipation in land prices or sales volumes prior to policy announcement, nor selection into treatment based on specifications that admit neighbourhood-specific trends.

    “we find no evidence of anticipation in land prices or sales volumes prior to policy announcement, nor selection into treatment based on specifications that admit neighbourhood-specific trends”
    Quote in the paperSubstantial claim

  • Probably not a substantial claim: may not pass S3 (load-bearing).

    Premisecitation
    § Introduction

    The Auckland Unitary Plan was motivated by housing affordability concerns.

    “Motivated by housing affordability concerns (Auckland Unitary Plan Independent Hearings Panel, 2016)”
    Quote in the paperNot a substantial claim0 of 1 sources checked
    • cites Auckland Unitary Plan Independent Hearings Panel, 2016

      Couldn’t be matched to a paper in Crossref, so it can’t be checked.

  • Classified as probably not a substantial claim.

    Premisecitation
    § Dataset

    The canonical Alonso-Muth-Mills spatial equilibrium model predicts that relaxing land use regulations will result in more housing close to the city center.

    “The canonical Alonso-Muth-Mills (AMM) spatial equilibrium model of the monocentric city predicts that a relaxation of Land Use Regulations (LURs) will result in more housing close to the city to the center (Bertaud and Brueckner, 2005).”
    Quote in the paperSubstantial claim0 of 1 sources checked
    • cites Bertaud and Brueckner, 2005 source

  • May combine two claims.

    Premisenone given

    The first version of the Unitary Plan, with clear notification of the intent for dwelling intensification, was released in 2013, more than three years before the final version became operational.

    “the first version of the plan, with clear notification of the intent for dwelling intensification, was released in 2013, more than three years before the final version became operational”
    Quote in the paperSubstantial claim

  • May combine two claims.

    Study design and conductown data
    § Institutional Background

    The parallel trends assumption holds remarkably well in the sample, indicating that demand-side policies did not have a differential impact on upzoned areas.

    “the parallel trends assumption holds remarkably well in our sample, affirming that these policies did not have a differential impact.”
    Quote in the paperSubstantial claim

  • May combine two claims.

    Study design and conductown data
    § Results

    There is no apparent trend in the estimated coefficients prior to treatment in any of the three samples, so the parallel trends assumption appears to hold.

    “Evidently there is no apparent trend in the estimated coefficients prior to treatment in any of the three samples, so the parallel trends assumption appears to hold.”
    Quote in the paperSubstantial claim

  • May combine two claims.

    Study design and conductreasoning
    § Introduction

    The study lacks precise measures of dwellings demolished when properties are redeveloped because demolition permits are only required for buildings less than three storeys.

    “Unfortunately we do not have precise measures of dwellings demolished when properties are redeveloped. This is because demolition permits are only required for buildings that are less than three storeys.”
    Quote in the paperNot a substantial claim

  • May combine two claims.

    Interpretationreasoning

    The result is a smaller spillover effect in the counterfactual set, and thus a larger treatment effect in the identified set, compared to that obtained with 2015 as the treatment date.

    “The result is a smaller spillover effect in the counterfactual set, and thus a larger treatment effect in the identified set, compared to that obtained with 2015 as the treatment date.”
    Quote in the paperNot a substantial claim

  • Probably not a substantial claim: may not pass S4 (based).

    Premisenone given
    § Introduction

    Observed trends immediately prior to the policy intervention are informative of the counterfactual scenario.

    “The intuition underpinning both the RR strategy and our strategy is the idea that observed trends immediately prior to the policy intervention are informative of the counterfactual scenario.”
    Quote in the paperNot a substantial claim

  • May combine two claims.

    Study design and conductcitation
    § Introduction

    The empirical design exploits temporal changes in zoning rules via a policy intervention and can mitigate many endogeneity concerns that afflict studies relying only on spatial variation in land use regulations.

    “Because the empirical design exploits temporal changes in zoning rules via a policy intervention, it has the capacity to mitigate many of the concerns stemming from the endogeneity of regulations that afflict studies which rely only on spatial variation in LURs (Gyourko and Molloy, 2015).”
    Quote in the paperSubstantial claim0 of 1 sources checked
    • cites Gyourko and Molloy, 2015

      Couldn’t be matched to a paper in Crossref, so it can’t be checked.

  • Probably not a substantial claim: may not pass S3 (load-bearing).

    Premisecitation
    § 6 Discussion

    Recent studies show that zoning reforms have not had a substantial impact on housing construction.

    “recent studies that show that zoning reforms have not had a substantial impact on housing construction (Freemark, 2019; Limb and Murray, 2022)”
    Quote in the paperNot a substantial claim0 of 2 sources checked
    • cites Freemark, 2019 source
    • cites Limb and Murray, 2022

      Couldn’t be matched to a paper in Crossref, so it can’t be checked.

  • May combine two claims.

    Interpretationown data
    § Detached Housing

    Upzoning had a substantive impact on the construction of detached dwellings, albeit a smaller impact than that realized for attached dwellings.

    “We conclude that upzoning had a substantive impact on the construction of detached dwellings, albeit a smaller impact than that realized for attached dwellings.”
    Quote in the paperSubstantial claim

  • May combine two claims.

    Interpretationreasoning
    § How Many Additional Permits Did Upzoning Enable?

    The 27,183 additional permits imply that permits issued per year have approximately doubled as a result of the policy, which averaged 4,148 permits per year between 2010 and 2015.

    “This figure implies that permits issued per year have approximately doubled as a result of the policy, which averaged 4,148 permits per year between 2010 and 2015.”
    Quote in the paperNot a substantial claim

  • May combine two claims.

    Interpretationown data
    § How Many Additional Permits Did Upzoning Enable?

    Permits in upzoned areas are still trending upwards as of 2021, so the full impact of the policy will likely not be known for several more years.

    “It is important to note, however, that permits in upzoned areas are still trending upwards as of 2021, so the full impact of the policy will likely not be known for several more years.”
    Quote in the paperNot a substantial claim

  • May combine two claims.

    Interpretationreasoning
    § Residential Zones

    These figures imply that 26.5% of the overall increase in permits occurred in areas upzoned to THA, 35.8% in areas upzoned to MHU, and 37.6% in areas upzoned to MHS if we allocate the counterfactual change in permits equally to each of the three treatment areas.

    “These figures imply that 26.5% of the overall increase in permits occurred in areas upzoned to THA, 35.8% in areas upzoned to MHU, and 37.6% in areas upzoned to MHS if we allocate the counterfactual change in permits equally to each of the three treatment areas.”
    Quote in the paperNot a substantial claim

  • May combine two claims.

    Study design and conductown data
    § Robustness Checks

    Prior to 2013, coefficients for land prices are statistically indistinguishable from zero, indicating that there were no differential trends in land prices between upzoned or non-upzoned areas prior to policy announcement.

    “Prior to 2013, coefficients are statistically indistinguishable from zero, indicating that there were no differential trends in land prices between upzoned or non-upzoned areas prior to policy announcement.”
    Quote in the paperSubstantial claim
54 of 54

  • The paper relies on this without support.

    Building permits are a valid proxy for housing construction.

    “Our dataset is based on annual building permits for new dwelling units issued by the Auckland Council from 2010 to 2021.”
    “Throughout the paper, “permits” refers to permitted housing units or, in the New Zealand parlance, “consented dwellings”.”

  • The authors state this as an assumption.

    A plausible mapping of building height to storeys can be used to compute floor area ratio as the product of the storey limit and the site coverage ratio.

    “Under a plausible mapping of building height to storeys, the FAR is straightforwardly the product of the storey limit and the site coverage ratio.”

  • The authors state this as an assumption.

    There is no confounding variable generating a difference in permits between treatment and control areas prior to policy implementation.

    “that there is no confounding variable generating a difference in permits between treatment and control areas prior to policy implementation.”

  • The authors state this as an assumption.

    In the absence of spillover effects, the difference between the treatment and control groups under the counterfactual scenario is zero.

    “In the absence of spillover effects, δpost = 0.”

  • The authors state this as an assumption.

    The counterfactual change in permits in each area is equivalent to one-third of the counterfactual implied by the period fixed effects.

    “This is equivalent to assuming that the counterfactual change in permits in each area is equivalent to one-third of the counterfactual implied by ˆ ϕs.”

  • The authors state this as an assumption.

    In the absence of the upzoning policy, dwelling permits in upzoned and non-upzoned areas would have followed the specified counterfactual trend (e.g., linear trend).

    “Constraining the counterfactual to the linear trend results in 29,866, 27,786 and 33,191 additional permits when 2013, 2014 and 2016 are used as the treatment date, respectively.”
6 of 6

Self-check

Percentages, sums and reported differences re-computed from the paper’s own numbers.

Each contradiction between two passages listed here was confirmed by an independent check.

  • Special Housing Area robustness totals disagree

    Major

    The main text says including Special Housing Area permits lowers the estimated net increase to 24,767, whereas the appendix reports 27,073 for that same sample change.

  • Discussion reports a different total increase in permits

    Major

    The discussion reports 26,903 additional dwellings, whereas the results and abstract report 27,183 for the estimated policy effect.

  • Printed calculation of the upper completion percentage is wrong

    Minor

    Multiplying the printed 15.13% by 0.99 gives 14.9787%, not 5.08%. The stated 5.08% is consistent instead with multiplying 5.13% by 0.99.

    Reported 5.08 · recomputed 14.9787

  • Figure 7 identifies the wrong reference year

    Minor

    Figure 7 says its period effects are relative to 2016, but the results text identifies 2015 as their reference period.

  • Figure 18 notes describe land prices instead of permits

    Minor

    The appendix introduces Figure 18 as treatment effects under alternative dates and reports additional permit counts, but the figure notes describe estimates for log land prices.

  • Figure 15 describes a different counterfactual-set width

    Minor

    The text says Figure 15 presents the all-dwellings counterfactual set with M = 15, while its note specifies ±200 dwellings in the final period. With 510 statistical areas, M = 15 represents ±7,650 dwellings in total, not ±200.

  • Control-group counterfactual bounds conflict with the appendix

    Minor

    The main text gives bounds of −2.37 to 7.63 permits relative to 2015, while the appendix explicitly gives −2.81 to 7.19 for that counterfactual. The main-text bounds include the observed-versus-counterfactual difference of 0.44 permits.

  • Attached-housing trend has two different values

    Minor

    The attached-housing extrapolated trend is stated as 0.374 permits, but the later calculation uses 0.364. The stated total of 7,331 uses 0.374, not 0.364.

8 of 8

Cites

Flagged first, then by citations

Research works printed but not linked: 2

  1. Neumark, 2015, vol. 5
  2. Redding, 2015, vol. 5

Cited by

Sorted by citations

Related

Authors
Greenaway-McGrevy, R., & Phillips, P. C. B.
Affiliations
University of Auckland (NZ)
ORCID
Venue
Journal of Urban Economics
Publisher
Elsevier BV
Published
2023-07
Volume
136
Pages
103555
ISSN
0094-1190 (ISSN-L)
DOI
10.1016/j.jue.2023.103555
OpenAlex
W4378894993
Cited by
70
References
39
Citations by year
2026: 21 · 2025: 23 · 2024: 18 · 2023: 7 · 2022: 1
FWCI
14.08
Type
journal article
Language
en
Open access
Licence
Locations
  • Journal of Urban Economics (acceptedVersion)
  • RePEc: Research Papers in Economics (submittedVersion)
  • ePrints Soton (University of Southampton) (submittedVersion)
  • Singapore Management University Institutional Knowledge (InK) (Singapore Management University) (submittedVersion, cc-by-nc-nd)
Funders
  • National Science Foundation (1850860)
  • Royal Society Te Apārangi
  • Marsden Fund (UOA2013)
Topics
  • Social Sciences › Economics, Econometrics and Finance › Economics and Econometrics › Housing Market and Economics
  • Social Sciences › Urban Studies › Urban and Rural Development Challenges
  • Social Sciences › Urban Studies › Urban Planning and Governance
Keywords
upzoning, zoning reform, housing construction, housing supply, housing affordability, displacement effects, residential land use, Auckland, land use regulation, counterfactual estimation

About these checks

Registry, retraction, Cites and statistics results are computed, not judged, with one exception: a printed reference that no index lists is read once by a model before it counts as a citation error. Claims are extracted by a model and kept only when their quote appears verbatim in the text; each cited source is then checked, with a model’s help, for whether it says what the claim relies on. Those verdicts can be wrong — read the claim’s quote and the cited source before relying on one.

Claims read by deepseek-v4.1-flash under definition version 3 (extract-v3) on 2026-11-04.

Classified by jev-1.13-20260917 (gold-guideline-v0.2) on 2026-11-04.

Cited sources are checked by gpt-6-sol.

Statistics and arithmetic are recomputed, not judged.

Confidence bands are shown only for fields calibrated on our annotated test set. None is calibrated yet, so no band is shown.

What counts as a claim ›How we measure confidence ›How we check this ›

Are you an author, or do you see a problem with this page? Request a review