Are exclusion factors capitalised in housing prices?

• Spatially adjusted hedonic models identify the value of accessibility in low socio-economic areas.
• Both lagged dependent variables and spatial error correlation are incorporated in the model.
• Proximity to services does not appear to be capitalised in housing prices in a car dependent city.

Social exclusion defines the degree to which an individual is limited in their access to the services and facilities to engage with their local and broader community. This paper investigates the relationship between exclusion and the level of accessibility to services provided by locality and transport. We provide household valuations of the factors that affect access and which can inform various policy directions.A Hedonic Price Analysis of an urban residential area is used to estimate implicit household monetary valuations on some key exclusion indicators. The value of access to schools, shops, parks and transport facilities is observed in the market price of the house. The application to social exclusion focuses on the outer suburbs of Perth, Western Australia with low socio-economic status. Locations are drawn from a prior cluster analysis that identifies local areas with distinct accessibility differences. Depending on the model structure, these evaluations may differ. Current results reveal a 6–8% premium for houses conveniently located near local shops, schools, railway station and to the CBD, a 20–25% premium for the quality of the neighbourhood, the remaining being embedded in the dwelling features.The analysis has both practical and academic implications: (i) it informs policies that aim to alleviate social exclusion. The implicit pricing is an important advance in this area because the household valuations may be imported into cost–benefit analysis of transport or service provision projects; (ii) the implicit prices are important inputs into the designing of experiments for follow-up stated choice surveys aimed at understanding residential choice; however, differences in evaluations lead to different designs, supporting the wider adoption of Bayesian designs, which can be more robust to variations of prior parameters. The models accounting for spatial effects provide more robust estimates, however the interpretation and prediction are not straightforward.