6. Summary and Future Directions
Value added per worker in Auckland region is 30 to 50 percent higher than that of regions outside Auckland. The premium is even higher in Auckland CBD (120 to 150 percent higher). While these magnitudes seem large, they are similar to the premia documented by ONS (2007) for London's GDP per capita. London has GDP per capita that is 50 percent above the rest of the UK, and Inner London has a premium of 140 percent.27
Industry composition differences account for about half of Auckland's higher labour productivity. This finding is similar in magnitude to that of Rice and Venables (2004), who attribute around 40% of London's income per capita premium to job composition, as captured by occupational mix. We find that industries are positively selected into Auckland – the industries that are most concentrated in Auckland are the ones that show the highest Auckland premium.
Our analysis of productivity patterns within Auckland reveals considerable geographical variation, and a strong correlation between productivity and employment density. A significant fixed effect estimate of the relationship between density and productivity suggests that the cross-sectional pattern is not solely reflecting patterns of spatial equilibrium. However, the finding of a positive correlation of density and productivity over time within areas is not robust to alternative approaches of estimation such as first differencing or using differences across six-years. More detailed econometric analysis, taking into account spatial autocorrelation and lags may provide a fuller understanding of the links.
The cross-sectional estimates imply that having density that is twice as high is associated with productivity that is 8.6% higher. This is at the upper end of the 3% to 8% range from the studies summarised in Rosenthal and Strange (2004), and the 1% to 10% range reported by Graham (2005a) and Graham (2006a). The current paper's estimates differ from many of those reported in the literature in 2 key respects. First, the current estimates do not adjust for within-industry differences in capital intensity or labour quality, both of which are likely to be higher in dense areas. Some international studies use more sophisticated production-function approaches to control for such factors. Our estimates may therefore overstate true productivity differences. Second, our estimates cover all industries, whereas many of the studies in the literature look only at manufacturing. Graham (2006b) shows that the elasticity is lower for manufacturing than for the economy as a whole, and reports an economy-wide elasticity of 12.5% but only 5% for manufacturing.
We examine industry variation in effective employment density, both of overall employment and of industry employment. The industries that are most concentrated within Auckland, and also over-represented in Auckland, have the highest productivity premium from operating in Auckland (33%). The size of this premium is higher in areas of Auckland where employment density is high, and especially where own-industry density is high. We identify two groups of industries that show this pattern, together accounting for 48% of Auckland's employment. These groups include mainly Transport and Storage industries, Finance and Insurance, and a range of Wholesaling and Manufacturing industries.
There is a sizeable group of industries (40% of Auckland's employment) that are neither over-represented in Auckland nor particularly concentrated within Auckland. These industries mostly provide local goods and services. They do, however, appear to benefit from locating in denser areas of Auckland and from locating in areas of high own-employment density. Overall, localisation effects appear to be stronger than the effects of density per se. Own-industry density is more strongly related to productivity than is density, even for the most dispersed industries.
While the focus of this paper has been on documenting Auckland's productivity performance and providing an initial view into the links between density and productivity, a fuller picture of Auckland's productivity could be advanced by using more detailed firm-level information that is available from the LBD. It is possible to estimate industry-level production functions and obtain estimates of the contribution of agglomeration to multi-factor productivity (See Graham (2006b) Henderson (1986) Eberts and McMillen (1999) Rosenthal and Strange (2004) for examples of this approach). This approach would support a decomposition of Auckland's productivity premium, gauging the relative contributions of factors such as higher capital intensity, higher labour quality, proximity to local productive inputs, local industry and market scale, head-office concentration, and residual advantages. Obtaining additional information on local output price and quantity variation separately would allow a clearer analysis of the contributions of technical productivity and allocative productivity.
This paper helps to focus ongoing policy debates on the strength of agglomeration effects in Auckland. First, it provides estimates of the size of the Auckland productivity premium, with sub-analyses by territorial authority, by smaller area units, and by industry. Second, it confirms a cross-sectional link between productivity and density, but shows that this link is not evident over time within industries. Finally, it shows the diversity of agglomeration patterns across different groups of industries, suggesting that localisation and urbanisation effects operate differently for different industries. This delineation of patterns provides a foundation for more targeted analysis to identify which agglomeration mechanisms operate in specific sets of industries. It also emphasises that policies that aim to improve Auckland's productivity performance by increasing employment density are likely to have an uneven impact across industries.
Back to Top