Abstract:
This thesis comprises three independent but inter-related essays on urban housing markets in New Zealand. The first chapter proposes a novel approach to quality adjusted hedonic imputation estimation based on a new type of Mallow’s Model Averaging estimator robust to spatial autocorrelation. Using model averaging we can simultaneously ensure a degree of flexibility in the hedonic specification over time and address the trade-off between model misspecification bias and overfitting variance when setting a headline hedonic function. The results of an application of the procedure to three real estate submarkets in Auckland, New Zealand, demonstrate that the approach exhibits several key advantages over standard ordinary least squares hedonic imputation index estimation. Namely these are tighter confidence intervals on index growth estimates, superior out of sample prediction of price growth, and superior accuracy of imputation estimates. The second chapter develops a novel measure of housing supply responsiveness which allows us to identify whether regions characterised by low levels of housing supply elasticity experience unique long run regional adjustment patterns. Unlike measures utilised in similar research in the North American context, our measure does not rely on detailed survey data. Instead it is created from estimates of price distortion arising from land use restrictions and detailed geospatial data. The results reveal evidence that regions characterised by low housing supply elasticity are synonymous with lower long run employment and population growth following positive economic shocks. Furthermore, these regions also maintain relatively smaller new housing supply responses and suffer from higher rental price growth. The final chapter models the long run housing supply and price effects of an urban rezoning policy implemented in Auckland, New Zealand. The chapter develops an empirical strategy for estimating the effects of policy shocks operating in time based on a panel vector autoregression approach augmented with conventional difference-indifferences treatment indicators for cross sections and time periods subject to policy shocks. Forecasts constructed from the model suggest that an upzoning announcement results in an immediate and persistent moderate price premium arising between low and highly intensified land parcels on upzoned land. The model also highlights that in residential areas upzoned to permit the highest level of site development, upzoning results in a 43% increase in construction of high-density dwellings over the long run.