A Spatial Econometric Analysis of Selected Local Labour Market Outcomes in New Zealand

Abstract

This thesis analyses several aspects of local labour market performance in New Zealand. Each of these aspects represents a different feature of the local labour market that together aim to provide a comprehensive understanding of how local labour markets respond to local or external shocks, taking spatial dependencies into account. The first aspect considered is the change in regional employment outcomes in New Zealand from the 1980’s to 2006. This is examined using shiftshare techniques supplemented by exploratory spatial data analysis. The analysis finds that in general, region specific factors were more important than industry structure in explaining total employment change in a region and that productivity and/or demand shocks spill over between regions. The relationship between homeownership and unemployment is the second aspect considered. Particularly, Oswald has argued that disparities in unemployment rates maybe attributable to differences in homeownership rates, higher homeownership rates being associated with higher unemployment rates. Using spatial panel models this claim is supported in the New Zealand context. The spatial estimation strategy adopted yields parameter estimates significantly lower than the standard fixed effects model. This indicates that in the non-spatial models some of the variation in regional unemployment rates has been incorrectly attributed to the explanatory variables rather than to the presence of some spatial spillover effects of unemployment across regions. Thirdly, the thesis investigates the extent to which the spatial-temporal variation in local labour market outcomes and social security benefit uptake can be linked to the composition of the local labour force. This is investigated using spatial seemingly unrelated regression. The results indicate that three factors in particular matter in the determination of labour market participation and social security benefit uptake: the age structure of the population, the past performance of the regional labour market and the proportion of solo parents. Lastly a simultaneous equations growth model of real income, population, land rent and public infrastructure investment is developed that allows the impact of local authority infrastructure spending in New Zealand to be assessed for the 1996-2006 period. The results of the estimation of this system of equations, using a spatial three stage least squares (3SLS) procedure, show that an increase in local infrastructure spending increases population growth, real income and land values, but is itself endogenous and spatially correlated.