Analytical Research on Other Dwellings Price Index
Dr Thai Nguyen, a postdoctoral researcher from the UNSW and temporarily employed at the ABS, is undertaking a suite of research analysis on assessing the best method for handling compositional change in units and apartments (collectively called 'other dwellings') as part of the investigation into the feasibility of constructing 'other dwellings' price index. (The scope of the existing House Price Index includes established detached houses only, not apartments and units). Outlined below are some exciting projects being undertaken within the NSW Office as part of a cooperation between Analytical Services Branch and the House Price Index section.
1. Exploring alternative clustering methods for Other Dwellings Price Index
A stratification method works well if all price observations in a stratified group are homogenous (i.e., having similar prices and price trends). The current stratification method for the House Price Index is based on suburbs’ long-term median prices and the SEIFA scores. Apart from the SEIFA scores, there may be other underlying characteristics that determine other dwellings' prices as well as price trends. Therefore, it may be of interest to include suburbs’ rates of change in dwelling prices as a stratified variable to capture the above-mentioned underlying other dwelling characteristics. This alternative stratification is currently being explored for the construction of the other dwellings price index.
2. Exploring repeat sales method for constructing Other Dwellings Price Index
Together with stratification, repeat sales method is a possible approach to constructing Other Dwellings Price Indexes, in the absence of detailed housing characteristics data. Essentially, the repeat sales method is a matched model approach since it utilises information on the same properties that have been sold more than once. This approach automatically controls for micro location (i.e. address) of houses and quality mix. Although the repeat sales approach has its own limitations, it may be useful to use it indirectly such as to align it with the benchmarking indexes or to compare the results with the repeat sales indexes produced outside the ABS.
3. Retrospective approximation of Other Dwellings Price Index for non-Census years (where weights data is not available)
House price indexes are currently constructed using a fixed-weighted Laspeyres price index, with the weights based on Census count. These weights remain constant until the release of new Census data, which can lead to increasingly unrepresentative weights as time passes by. When a new Census becomes available, it is desirable to use this information to retrospectively revise the indexes back to the previous housing Census date. The retrospective approximations of superlative other dwellings price index can help to reduce the well-known substitution bias associated with the Laspeyres index.
For more information please contact Thai Nguyen on (02) 9268 4449 or firstname.lastname@example.org.