The Evolution of the American Urban System: History, Hierarchy, and Contagion

Elizabeth A Dobis, Purdue University

Abstract

This dissertation analyzes the spatial and temporal evolution of cities in the American urban system. We explore these evolutions through analyses focusing on the history of the urban system and the flows among cities related to the effects of their hierarchical structure and contagion due to proximity. We create five datasets tailored to our research objectives, three based on incorporated places and two based on urban areas. These sub-county geographic units are rarely used in analyses of urban systems due to a paucity of data available but allow for a more direct analysis of the flows within the American urban system. The first analysis focuses on the history of the urban system by exploring three aspects of the urban system that affect change: the number, spatial distribution, and growth of settlements. We perform econometric and spatial statistical analyses that explore these three aspects separately. First, we determine how the age of a settlement is related to its physical location using a truncated linear regression and a cross-section of 2010 incorporated places. Settlement age is determined by the number of years since the date of incorporation. Second, we determine whether settlements in the United States are spatially random, and, if not, what physical geographic characteristics are associated with the intensity of settlement throughout the country. This is accomplished with the Clark-Evans ratio and a spatial Poisson regression of 25km grid cells that measure the intensity of settlement as the number of 2010 incorporated places within the boundaries of the grid cells. Third, we determine the relationship between a settlement's growth rate and its previous population and physical location using a random effects unbalanced panel regression of 2010 incorporated places. Growth is determined by the decennial population change within an incorporated place from 1890 to 2010. The history analysis makes four contributions to the literature. First, we tell a comprehensive, multifaceted story of the evolution of the American urban system that includes explicit statistical analyses of the three aspects of the urban system that influence change. This is more complete than previous work in the urban growth and city systems literature because it includes all aspects of the urban system that experience change, rather than focusing on one or two of them. Second, we perform an explicit analysis of the effect of physical location on a settlement's age. Third, we use the Clark-Evans ratio and test to determine whether incorporated places in the United States are spatially random. Finally, we perform an explicit analysis of the effect of physical geographic features on settlement intensity in the United States. The second analysis focuses on the flows among cities by exploring how hierarchical relationships and spatial contagion affect the population in cities in the urban system. The hierarchical structure leads to a ranking of cities according to the variety and specialization of goods and services available, while spatial contagion among cities refers to spatial spillovers that result from proximity. We use a spatial lag hierarchical level model to disentangle the effects of hierarchy and contagion on the population size of a cross-section of urban areas in 2010 and a balanced panel of metropolitan areas from 1990 to 2010. The hierarchy and contagion analysis makes two contributions to the literature. First, we disentangle the effects of hierarchy and contagion among cities by explicitly accounting for both using a spatial lag hierarchical model. Second, we capture the effect of including small urban areas with a population of between 2,500 and 50,000 inhabitants in an analysis, providing a more comprehensive view of the American urban system. We find four overarching conclusions about the roles of history, hierarchy, and contagion in the evolution of the American urban system. First, relative proximity is important to understanding relationships within the urban system. Second, city population size affects the rate of growth, both when accounting for spatial relationships among cities and when not. Third, there is weak support for the perception that natural amenities affect the distribution of population and change within the urban system. Finally, settlements in the United States are spatially clustered, particularly along the coasts, but that clustering has become less prominent over time.

Degree

Ph.D.

Advisors

Florax, Purdue University.

Subject Area

Geography|Economics|Demography

COinS