Refining the stratification for the Established House Price Index
The House Price Index (HPI) team in the Consumer Price Index (CPI) section constructs an established HPI of quarterly estimates of the changes in price of the established housing stock in each Australian capital city and for Australia as a whole. The established HPI is based on grouping together "similar" suburbs within statistical sub-divisions (SSDs) in each capital city. The "similarity" of suburbs is determined by applying a stratification method based on attributes that can be broadly defined as the structural, locational and neighbourhood characteristics of the suburbs. Suburbs within an SSD are grouped together into strata, so that similarity within groups and dissimilarity between groups are maximised. In each period, a summary price measure based on house sales is calculated for each stratum and used to construct a stratum level price index. The aggregate indexes are then calculated by weighting together individual strata indexes using weights that represent the value of the housing stock in each stratum.
This stratification approach has been the subject of an ongoing study conducted by staff from the Analytical Services Branch and the HPI team in the Prices Branch. The study investigates ways to refine the stratification used for the construction of the established HPI to better control for compositional change. The study explores several avenues for improvement. The first avenue is to apply stratification within capital cities (rather than within SSDs), which may lead to more acceptable numbers of stratum in each capital city and each stratum containing a relatively broader grouping of similar suburbs with sufficient numbers of house sales every period to allow for the construction of reliable stratum summary price measures. Second, we can explore the inclusion of the long-term median house price of a suburb as a stratification variable and examine its potential for improving compositional adjustment - on the premise of controlling for economic strata - when combined with socioeconomic stratification variables. Third, we can investigate the variables used in the stratification to determine if more efficient results can possibly be achieved by using a simpler stratification method (for example, using fewer stratification variables). Finally, we can utilise improved data sets that have become available since the last stratification review was completed.
For more information, please direct enquiries to Alexa Olczyk on (02) 6252 5854 or Steve Lane on (02) 6252 7833.