Modelling housing markets : issues in economic measurement

Housing markets are an integral part of an economy and each and every one is affected by changes in house prices either directly by owning a property or indirectly by renting. Housing markets are hence watched by policy makers, investors, house owners and urban planners. House price indices are an important instrument to assess housing markets although standard indices are probably not enough to fully understand and capture the complicated dynamics in these markets.This dissertation proposes new methods to enhance the assessment. The techniques are applied to house prices and rents in Sydney between 2001 and 2014.The first essay shows that standard house price indices are prone to suffer from an averaging effect which can lead to wrongly placed turning points and wrong price levels. As an alternative, this essay proposes a time-continuous index based on adaptive smoothing. This index is not affected by an averaging effect and has favourable robustness properties. The second essay highlights the rich set of variation in price changes even within urban areas and suggests the construction of location- and segment-specific imputation indices based on quantile regression models. Empirically, this essay finds that the Sydney housing boom peaking in 2004 was mainly driven by sharply rising prices of suburban, low-priced houses. The third essay jointly analyses the rental and housing market and constructs quality-adjusted and quantile-specific rental yields based on a combination of propensity score matching and quantile sampling. The empirical analysis shows that rental yields are consistently downward sloping when moving from the low end of the distribution to the top end confirming theoretical considerations about the user cost of housing. Additionally, a severe sample selection bias is detected when relying on houses that were sold and rented within a short period of time.