Flat prices, cell phone base stations, and network structure : an instrumental variable approach to endogenous locations

The present paper analyses the effect of the distance to the nearest CPBS on listing prices of flats in the city of Nuremberg, and is related to a similar study of Brandt and Maennig (2012) in case of the city of Hamburg. Health effects of mobile phone radiation have been discussed vividly in the German public and there is still ongoing opposition against cell phone base stations (CPBS) in residential areas. Besides a negative health effect, a CBPS site is often perceived to be visually disruptive in a residential neighborhood. Therefore, a negative effect on property values due to the proximity to CBPS is expected. However, the empirical evidence is mixed and focused on rural areas (Bond 2007) while we consider an urban area. Furthermore, the endogeneity of CPBS locations and thus the identification of the effect is still an issue. The empirical approach applied here is based on a hedonic price function of housing prices. The hedonic regressions used to identify the effect of the distance to the nearest CPBS take account of housing and neighborhood characteristics but also a possible endogeneity of the variable of interest. Following the critique of Pinkse & Slade (2010) and Gibbons & Overman (2012) who advocate the use of the quasi-experimental approach, we develop an instrument for the estimation of local price effects of CPBS in an urban area. The instrument is derived from the spatial structure of the network and technical and regulatory requirements. Such a strategy could be also useful in other contexts in which location choice is endogenous but depends on an existing network structure. We find a significantly negative impact of nearby CPBS on flat prices. The discount amounts to 3.3% of a property's value when two similar flats at distances of 50 and 100 m to the nearest CPBS are compared. The effect size is comparable to findings of other studies. The results of the main OLS and IV specifications are robust to several modifications that include e.g. spatially lagged prices. The small difference between OLS and IV results suggests that the distance to the nearest CPBS is not endogenous, in opposition to Brandt & Maennig (2012). Both authors argue that CPBS are likely to be located on "visually disruptive" structures However, such structures are rare in city centres where the network is dense and distances to CPBS are smaller. Consequently, the endogeneity problem may be less relevant. Further theme: S_Y The Causal Impact of Infrastructure on Regional Economic Activity