Research Discussion Paper – RDP 2006-03 Australian House Prices: A Comparison of Hedonic and Repeat-sales Measures
May 2006
Abstract
House prices are intrinsically difficult to measure due to changes in the composition of properties sold through time and changes in the quality of housing. I provide an overview of the theoretical nature of these issues and consider how regression-based measures of house prices – hedonic and repeat-sales measures – can control for compositional and quality change. I then explore whether these regression-based alternatives can provide accurate estimates of pure house price changes in the Australian context.
Using unit record data for Australia's three largest cities – Sydney, Melbourne and Brisbane – between 1993 and 2005, the results suggest that the two regression-based approaches provide similar estimates of the pure price change in housing. The measures are comparable in terms of statistical fit, with around half of the variation in prices growth (for those houses sold more than once) explained. The regression-based measures also produce similar estimates of pure price changes to those obtained by a mix-adjusted measure. However, all three measures behave quite differently from a simple median, implying that compositional change matters empirically. These results confirm that regression-based measures are likely to be a useful analytical tool when measuring pure house price changes in Australia.