Ex-post identification of geographical extent of benefited area by a transportation project: Functional data analysis method

This study proposes an approach for ex-post identification of the geographical extent of an area benefiting from a transportation project, using functional data analysis methods. Our approach focuses on real estate (land) price data for the ex-post spatial evaluation. First, we prepare a panel of land prices observed before, during, and after the project in the areas that are potentially impacted. Second, using functional data analysis, movements of land prices in each observed site during the target period are approximated by continuous functions. Third, using the functional ordinary Kriging technique, the functions for land price movements in each micro district are spatially predicted. Lastly, by employing the functional clustering (functional K-means) technique, potential areas of benefit may be identified. Different from exiting before-and-after methods, including difference-in-differences method, the proposed procedure based on functional data analysis can describe a map with a complex spatial distribution pattern of benefit rather than using distance bands (ring buffer) from transportation cores, such as railway stations, bus stops, and highway interchanges. Then, the proposed procedure is empirically applied to a large-scale Japanese heavy railway project. The obtained result shows so-called redistributive effect, that is, land prices decrease around exiting stations and increase around new stations. In addition, interestingly enough, the spatial distribution pattern of the identified areas of benefit using this procedure are fairly similar to that of ex-ante predicted areas of benefit by the hedonic approach. Thus, capitalization is observationally confirmed with regard to accessibility improvement.