This thesis investigates neighbourhood effects and specifically whether these effects act in a non-linear manner in an Australian urban context. The hypothesis driving the research is that neighbourhood disadvantage and disorder must reach a critical threshold before the mechanisms of neighbourhood effects have an impact on residents’ economic outcomes, such as unemployment and financial difficulties, and on the overall dynamics and conditions of the neighbourhood.

Three lines of inquiry were pursued to test the hypothesis. The first (Chapter 2) used census data from the 1996, 2001 and 2006 Australian Census of Population and Housing to test whether a number of neighbourhood indicators had a threshold-like relationship with changes in the mean unemployment rate of neighbourhoods over five and ten years. A spline regression analysis technique was employed to uncover discontinuous ‘break points’ consistent with the threshold hypothesis. Statistically significant discontinuous ‘break points’ were found in all eight neighbourhood indicators but the indicators with the highest predicted changes were ‘in low status jobs’ and ‘without qualifications.’ Neighbourhoods above the 99th percentile in these indicators showed statistically significant increases in unemployment (10.41 per cent and 8.56 per cent, respectively, higher increase in the proportion of residents unemployed than neighbourhoods just below the 99th percentile over five years). Based on the analyses, a tentative conclusion is made that urban neighbourhoods with very high proportions of residents with particular features show more pronounced increases in unemployment over time; however, the results are neither entirely consistent nor straight-forward.