The predictive performance of multilevel models of housing sub-markets: A comparative analysis

Abstract

Much of the housing sub-market literature has focused on establishing methods thatallow the partitioning of data into distinct market segments. This paper seeks tomove the focus on to the question of how best to model sub-markets once they havebeen identified. It focuses on evaluating the effectiveness of multilevel models as atechnique for modelling sub-markets. The paper uses data on housing transactionsfrom Perth, Western Australia, to develop and compare three competing sub-marketmodelling strategies. Model 1 consists of a city-wide ‘benchmark’; model 2 providesa series of sub-market-specific hedonic estimates (this is the ‘industry standard’) andmodels 3 and 4 provide two variants on the multilevel model (differentiated by variationin the degrees of spatial granularity embedded in the model structure). Theresults suggest that the more granular multilevel specification enhances empiricalperformance and reduces the incidence of non-random spatial errors.