Use of Geographically Weighted Regression and Multiscale Geographically Weighted Regression to Account for Spatially Heterogeneous Property Value Impacts of Heavy Rail Transit Stations

Abstract

This study provides evidence on the impact of a heavy rail-based suburban metro station in Fremont, California, on house prices using geographically weighted regression (GWR) and multiscale geographically weighted regression (MGWR). The dataset comprises sale price, sale date, and property and locational characteristics for single-family houses sold during the January 2000–April 2018 period within two miles of the Warm Springs Station on the San Francisco Bay Area Rapid Transit system. The results demonstrate that, relative to the referent period (2000–2001): (a) the station-led house price impacts are spatially heterogeneous—the house price increased in one pocket only, comprising less than one-tenth of the dataset, not across the entire 0–2 mile station area, (b) the price increase in that pocket began right after the project announcement period, and (c) the price impacts indicated by the MGWR model markedly differ from baseline ordinary least square regression (OLS) estimates. Overall, the study’s findings highlight the need to move beyond average estimation techniques, such as OLS, to those that account for spatially heterogeneous price impacts, such as GWR and MGWR.