ABSTRACT
This chapter aims to investigate the spatial dynamics of an urban housing market, using Cardiff as a case study. Implicit in this aim is an attempt to incorporate space into the hedonic house price model. To prevent rounding errors due to large numbers, and to facilitate interpretation of the models, the continuous independent variables were deviated around their means, so the models were estimated with regards to the typical property an average sized mid-terrace house. Heteroscedasticity in the error term can be checked by plotting the residuals of the hedonic model against the predicted house prices. An elementary method of checking for spatial autocorrelation is by mapping the model's residuals and checking for spatial patterning. By taking into account the context of location, the multi-level specification will allow the locational attributes, ameliorating the problem of the spatial drift specification in which the social class effect is constant across the city, regardless of neighbourhood context.
