Spatial concentration and dependence in labour market outcomes of Sydney and Melbourne

The last two decades have brought well-documented shifts in the spatial structure of Australia's economic growth not just in favour of Australia's cities, but within certain suburbs of these cities. Of interest is the extent to which spatial dependence in the form of neighbourhood interactions between closely situated suburbs, independent of other characteristics, may explain the diverse trajectories of ‘well-performing’ and ‘poorly performing’ suburbs over the period. This thesis examines the proposition of spatial dependence in suburb-level labour market outcomes of two Australian cities, Sydney and Melbourne, over the fifteen year period 1996-2011. Using four waves of census data the study begins by looking for significant clusters of socio-economic outcomes within Sydney and Melbourne. It employs Exploratory Spatial Data Analysis techniques (Anselin, 1995) capable of identifying significant global and local spatial autocorrelation. At the suburb-level in both Sydney and Melbourne positive spatial association in unemployment and labour force participation rates is found in both 1996 and 2011. Over our study period these ‘hotspots’ have shifted closer to the urban periphery, confirming Hulse et al.,’s (2014) finding regarding the spread of socio-economic disadvantage more generally. In Sydney and to a lesser extent Melbourne suburbs there is also evidence of strong positive global and local spatial association of education, industry and occupational variables, with spatial clustering increasing for a number of variables over the period 1996 to 2011. Having identified spatial clusters of labour market outcomes which represent significant deviations from a random pattern, we next examine the drivers of this clustering. Spatial econometric techniques (Anselin, 1988; Lesage and Pace, 2009 and Elhorst, 2014) offer a suite of models with endogenous and exogenous interaction effects, in addition to providing a framework for multivariate analysis. Results using a cross-sectional and panel approach in the context of unemployment rates, and the panel approach in the context of labour force participation rates reveal that commonly identified supply-side factors explain much of the variation in suburb-level labour market outcomes observed in both cities. Industry and occupation factors such as a suburb’s share of workforce in the manufacturing sector, and service-related and professional occupations are also significant. However we confirm the presence of significant spillovers between unemployment rates and labour force participation rates in neighbouring suburbs net of these controls. We also confirm spillovers operating on the basis of several notable socio-economic variables. The policy implications of our findings are considered in the concluding chapter.