We analyse the impacts that infrastructure provision has on long run urban development. The topic is of major importance to policy-makers when deciding whether or not to invest in major infrastructure projects. The analysis helps policy-makers to understand the intended, and potentially unintended, long run consequences of their infrastructure investment decisions. Reflecting a spatial equilibrium approach, we maintain that population flows reflect people's overall rankings of urban areas; thus, through revealed preference, growing cities are shown to have had preferred attributes (wages and amenities combined) relative to other cities. Social infrastructure (such as higher educational institutions and hospitals) and transport infrastructure may have both productive and amenity value. Thus increased provision of such infrastructure within a city may enhance a city's attractiveness provided their benefits exceed costs of provision. Poor infrastructure provision linking an urban area to major cities and other amenities may, conversely, reduce the attractiveness of that urban area, curtailing its long run population growth. We first outline a new theoretical model that includes distance-related effects on individual utility, and hence on urban population development. The new derivation, while similar to that of a recent model by Duranton and Turner (Review of Economic Studies, 2012), avoids a convenient but questionable assumption in their approach in relation to the effect of distance on individual utility. Our theoretical approach includes the impact of amenities, and the effect of distance in reducing their attractiveness, while also accounting for potential distance-related effects on wages and living costs. We test our model using long-term (80 year) historical data (measured every 10 years from 1926 to 2006) that enables us to relate the populations of 60 New Zealand urban areas to early infrastructure provision and initial conditions (e.g. existence of a harbour, topography, climate, etc). The use of additional data from the start of the twentieth century as instruments enables us to test whether early (and subsequent) infrastructure provision affected the shape of city development over the 80 year period. As well as dealing with endogeneity issues through our choice of instruments, we use spatial-econometrics techniques to test for spatial spillovers between cities.
