This dissertation consists of two essays exploring the often noted dispersion of economic activity within cities. Focusing in particular on the phenomenon of polycentricity, these essays explore the relationship between employment centers and spatial and economic outcomes of cities.
The first essay explores the implications of two common proposed criteria for identifying an employment center. Does the area represent a local concentration of employment? Does the area affect the local population density of the city? Using data on both place of employment and place of residence, I propose a new method for testing the relationship between concentrations of employment and population density within a metropolitan area. First a recently developed statistical method is used to identify concentrations of employment using data on place of employment. Second, I propose two methods for estimating the extent of the radius of influence for an employment center, using the relationship between tract of employment and tract of residence. Third, I propose a new specification for the entrance of distance into the polycentric regression. This new specification allows the impacts of the concentrations of employments on density to be positive, following the theoretical hypothesis. I use this new specification to jointly estimate the local gradients of 21 identified concentrations of employment in the Houston metropolitan area on their local population density. I find that not all identified employment concentrations have the expected significant positive gradients, and thus do not qualify as employment centers. I also find that the estimated gradients are sensitive to estimates for the radius of influence for each employment concentration, and that the level of employment in an employment concentration, alone, is not a strong predictor of significant local impact on population density or on the size of the estimated gradient.
The second essay tests for the theoretically predicted relationships between the number of employment centers in a city, and the city’s transport costs and wages. Urban area vehicle miles travelled rise with an increase in the number of employment centers in an urban area, while commute times are unaffected. These findings contradict the common hypothesis that additional employment centers lower transport costs by allowing workers to live closer to work. Instead, it appears that if transport costs are falling they do so through a fall in per unit distance price. I find that urban area average wages fall with an increase in the number of employment centers. I also find that average wages increase as a larger share of employment locates within employment centers. These two findings support the belief in the presence of agglomeration economies within employment centers that increases in concentration. In a competitive equilibrium the formation of additional employment centers have externalities in both the costs and benefits, thus it is not clear if the efficient number of employment centers will be formed within an urban area. This is explored through an investigation of the determinants of the share of urban area employment that locates in employment centers. I find that the predicted employment share maximizing number of employment centers increases with urban area size.||