THE IMPACT OF TOWN CENTRE ACCESSIBILITY ON INTRA-TOWN HDB PROPERTY VALUES

Abstract

Whether accessibility differences within a HDB new town have an impact on house prices is a prime area of contention. Qualitative research, theory and industrial expert opinion suggest that there is a positive relation, whereas some prior quantitative research like Ibrahim, Cheng, and Eng (2005) has churned out findings that indicate otherwise. The reliability of such quantitative models is thus an issue. This study was conducted with the objective of improving on existing hedonic house price models and the incorporation of accessibility factors into their analysis, so as to ascertain whether intra-town accessibility differences impact HDB house prices. In particular, different ways of specifying accessibility factors were considered. Using data of resale transactions in Tampines, four related hedonic price models with different specifications of accessibility were constructed and compared. House price was found to be statistically significantly positively correlated with town centre accessibility in all four models. The impact of accessibility can be considered substantial. Secondly, based on empirical evaluation of the models, the models which use a continuous distance or travel time variable substantially outperformed the models which use conventional discrete distance representations. Using discrete distance representations is also shown to produce unreliable coefficients for locational factors in a regression. Lastly, accounting for negative externalities from overland MRT operation does not improve the explanatory power of any of the models. This means any non-positive correlations of house prices and accessibility found in other researchers’ models are likely to be from a model misspecification than because of the presence of noise nuisance at the MRT stations. This study’s critique of the associated literature and ability to produce a model with results which align with real-world understandings of the market suggests that those particular prior studies were wrong about their findings about accessibility and house prices. It also recommends that using a continuous distance or time variable instead of conventional discrete distance variables would improve the performance and reliability of hedonic modelling as well as the quality of inferences generated from such models.