Noise valuation through the shock of airport closure: testing a new causal identification strategy for hedonic pricing based on comparables

Difference-in-difference (DiD) exploit exogenous shocks to capture causal effects in hedonic pricing (HP) models but rely heavily on fixed effects (FE) to, both, define the quasi-experimental settings and satisfy the assumptions of no spatial and temporal omitted variable bias (OVB). Bishop et al (2020, p:271-2) explain the function of time and space FEs: a)“generalizing the DiD model by interacting price function parameters with time-period dummies, allows the shape of the price function to change over time”; and b) “the geographic scale for these [spatial] dummy variables presents a bias-variance trade-off: defining neighborhoods to be smaller reduces bias by better controlling for omitted amenities but increases variance by relying on less within-neighborhood variation in the amenity of interest”. To help address the reliance on FEs and bias-variance trade-off, we develop a causal identification strategy that integrates DiD and the “Differenced-Price-Peers” (DPP) (Thanos and Dubé, 2022) approach. The DPP approach is based on comparable sales occurring in close spatial proximity and tight time-window before a house sale. By differencing out common spatial observables and unobservables between comparables, DPP delivers a spatiotemporal OVB treatment without relying on FEs or sacrificing any variance, which is especially useful in cases of sparse and challenging data. We first evidence the advantages of our approach by replicating the influential Linden and Rockoff (2008) results with improved statistical significance. We procced to employ the rare exogenous shock of airport closure in order to construct a quasi-experimental setting for aviation noise valuation, which is otherwise not possible due to the bias-variance trade-off. We find a 0.73% price increase per decibel of noise reduction for houses above the background noise level of 55 decibels, which suggests, on average, a €14,000 increase to affected properties due to the airport closure and aviation noise termination. Our approach also allows to examine nonlinearities, as for instance we find a 2% price increase per decibel for noise reductions above 65 decibels. This translates to a substantial € 43,000 increase to properties affected by aviation noise levels of 70 decibels.