Identifying spatiotemporally-varying effects of a newly built subway line on land price: Difference and correlation between commercial and residential uses

Abstract

This study examines the spatiotemporal relationship between the newly built subway line and land price, focusing on difference and correlation between commercial and residential land uses. For a spatiotemporally sensitive analysis, Geographically and Temporally Weighted Regression (GTWR) method is used to test land price data in Daejeon, South Korea. A new approach is suggested in the bandwidth selection process of GTWR to consider the interactive relationship between commercial and residential land uses. With the proposed approach, this study develops models to measure the effects of a newly built subway line on commercial and residential land prices. The estimates of the models indicate that commercial and residential land uses have an interactive relationship, and the effects from the nearby environments indicate geographically and also temporally different patterns for each land use. The distribution of coefficients shows that the effect size is greater on the commercial land price than on the residential land price. However, the geographical effect is wider on the commercial land price, while the temporal effect is longer on the residential land price. This study provides a novel perspective on empirical research about the effects of newly built public transit on land prices. Furthermore, the results imply that land price estimation needs to consider the mutual relationship between land uses and to be differentiated by the type of land use.

Keywords:

• Public transit; subway; land price; land use; geographically and temporally weighted regression

Notes

1 Three other weighted models are used: the Temporal Weighted Regression (TWR) model, Geographical Weighted Regression (GWR) model, and basic Geographically and Temporally Weighted Regression (basic GTWR) model.

2 This test uses the method of comparing the AICc value of the original model with the AICc value of a model having the same coefficient of a certain variable for overall observations.

Additional information

Funding

This research was supported by a National Research Foundation of Korea grant funded by the Korean Government (MSIP) (NRF 2014R1A2A1A11052725).