Download files
Abstract
This thesis investigates two aspects of the residential property market: the existence of rational speculative price bubbles in the Sydney housing market and the role of views in the determination of individual residential property prices. Both of these issues are analyzed from a spatial perspective in two novel ways. Chapter 2 inspects the geographic distribution of results from two separate bubble testing methodologies based upon the cointegration test of Diba and Grossman (1988b) and the regime-switching test of Shaller and Van Norden (1997) and Van Norden and Vigfusson (1998) using quarterly median house price data for individual postcodes in Sydney from March 1985 to September 2009. Although evidence is found for the existence of a property price bubble in Sydney, only 14.5 percent of individual postcodes within the city showed consistent evidence for the same price bubble across the various testing methods deployed. Additionally, those regions that demonstrate consistent evidence for a property price bubble over the sample period are concentrated around the central-western region of the city and are primarily composed of postcodes in the 10th and 20th lowest median price percentiles of the city. This result challenges the common presumption that house price bubbles are a city-wide phenomenon. In Chapter 3 a specific type of topographical map called a Digital Elevation Model is used in conjunction with the land parcel coordinates of 24,491 residential properties in Sydney to generate four novel measures of a residential property's view. Each of these new view variables is measured in continuous, metric units and is intended to produce estimates of a property's view that are more accurate, consistent and have more desirable characteristics than those methods currently used in the literature. In order to evaluate the contribution that different types of view have on property price levels, these view variables, along with nine attribute control variables, are entered into a selection of nine separate hedonic regression models. Each of these view measures is shown to have both a statistically and economically significant impact upon property prices which is generally commensurate with the results found in the view-based literature and confer a notable improvement in out-of-sample forecasting.