Journal of Economic Geography, Vol 22 (6), 1309-1352

Twitter summary

  1. Background/Motivation: Ongoing debate over whether market-rate supply raises nearby house prices through an amenity effect, or lowers them through a supply effect. E.g. Diamond & McQuade 2019 finds that amenity effect of LIHTCs dominates in low-income neighborhoods, but has a dis-amenity effect in high-income neighborhoods. Plenty of papers study market-wide effects (Glaeser & Ward 2009, Anenberg & Kung 2018, Gyourko & Molloy 2015), less look at localized impacts.

  2. Data: Prior studies have suffered from a lack of data on rents, but this paper uses Notices of Property Value (NOPVs) which include gross rental income (after concessions and vacancy). Building Permits are used for approval date, COs for completion. Considers buildings within 500ft of a new high-rise completed between 2003-2013. Data caveats: a) only rental buildings with >5 units, b) some years are missing, c) includes some commercial rents.

  3. Empirics/Key Threats/Challenges: Key challenge is overcoming endogeneity: houses are most likely to be supplied in locations where prices are already rising.$\small{ln(Rent_{it})=\alpha+\sum_{t}^{\tau}\beta_{\tau}YearSinceCompletion_{it}(\tau)+\sum_{k}^{\kappa}\gamma_{\kappa}YearSinceApproval_{it}(\kappa)+\delta Borough_i*Year_t+\mu_i+\epsilon_{it}}$$\beta$s are of interest, with time-since-completion providing “plausibly-exogeneous” variation in supply, after controlling for time-since-approval (with the latter essentially stripping-out the underlying trends in rents which attract new development). Key threat to causality here is if constructions become delayed in locations where rents are slowing, however this would bias the coefficient-of-interest towards zero, such that results represent a lower bound for the size of the effect.

  4. Results: Rents close to completed and not-yet-completed high-rises follow similar paths before completion, and then after completion, rents decline by 1.6% within 500 feet. Converted to elasticity, this means that for every 10% increase in stock within a 500-foot buffer, rents decrease by 1%.

    Sale prices also decline right after high-rises complete, without anticipation effects. “For every 10% increase in condo stocks, condo sales prices decrease by 0.9%”.

    Each new high-rise also attracts 0.11 new restaurants, suggesting that even with positive amenity effects, the supply effect dominates.

  5. Discussion/Conclusions: Market-rate apartments in NYC reduce both rents and prices nearby: a 10% increase in supply lowers both rents and prices by 1% within 500 feet. Caused by completion of construction, not planning approval. Clearly a supply effect: effect is larger for more similar buildings. Effect is smaller in more central locations (possibly more elastic demand; more people willing to in-migrate). Smaller for lower-rent buildings (somewhat consistent with Damiano & Frenier, although the effect here is always negative). Some evidence of amenity effects: new apartments also induce new restaurants. This is consistent with gentrification, but the question remains whether it’s really a problem if gentrification occurs alongside declining rents?