The United States (US) went through three major stages of urban spatial structure evolution. In the 1840s, the American dream indicated spreading construction to the suburbs [1
]. In the 1950s, the suburban movement began as many urban residents moved to the suburbs, encouraged by affordable automobiles and government subsidies [2
]. Between the late 1950s and early 1970s, many states (e.g., Kentucky, Oregon) established urban growth boundaries (UGB) to curb urban expansion [3
]. According to a survey in 1991, roughly a quarter of cities used UGBs to limit urban growth areas [4
]. From the American dream, suburban movement to UGBs, urban policies swung back and forth to adjust to the urban spatial structure. As a result, some policies may have positive impacts on some regions, while having negative impacts on the others.
The planning literature has identified employment centers, and the growth within and outside of them. In other words, the spatial structure of metro areas (interchangeably “metros” or “cities” in this paper) may evolve towards concentration or dispersion (including decentralization). Earlier studies [5
] have disagreed on whether cities are edged or edgeless. The former suggests that a re-concentration process occurs at a city’s edge during expansion, while the latter suggests that urban development is largely scattered—cities are edgeless. More recent studies [8
] found both concentration and dispersion in different metros, and also in different employment sub-centers within a single metro. These are mainly case studies that cannot demonstrate their theories’ applicability to a larger sample of metros.
Given the literature review, two big questions arise. The first question is in regards to the origin and destination of the employment concentration and dispersion (including decentralization). Where does the employment move to when it spreads from the Central Business District (CBD)? Does it go to sub-centers, non-center areas, or does it move outside of the metro? The second question involves employment clusters that are smaller than an employment center (typically defined as a minimum size of 10,000 workers), but that display certain agglomerative power. These areas have largely been neglected in empirical studies. Is it really safe to say that agglomeration economies at the submetro level are no longer important? Or, maybe agglomeration economies are presenting themselves in a new form? In fact, a few studies [15
] point out the complexity of the metropolitan spatial structures. They describe the structural complexity as being “beyond polycentricity” [19
], a network of “trading places” [20
], and “complex variation in density” [21
]. After all, the formation of an employment center is a gradual process; it inevitably generates employment clusters of varying sizes. A metropolitan spatial structure is more than a dual system of employment centers and non-center areas.
Studies on employment centers implicitly include four submetro sections: the CBD, sub-centers, urban areas, and rural areas. The CBDs are easy to define for small-sized (e.g., case-study based, descriptive) studies. It is also common to define the largest employment center as the CBD [11
], but we can cautiously use the term “main-center” in the case that some metro areas’ CBDs are smaller than their sub-centers. There are also recognized methods to delimit urban and rural areas [11
], and to identify employment centers [23
]. However, no study gives equal attention to all of these submetro sections; instead, most studies only analyze employment dynamics within and outside of the employment centers.
This paper investigates the dynamics of employment growth in five submetro sections that include four implicit ones (the main-center, sub-centers, non-cluster urban areas, and rural areas) in the literature and the non-center clusters. We use the entire country as the study area, and the longest time frame available. Next, we divide the metros into three size categories, since metros of different sizes may behave differently, e.g., the formation of sub-centers requires a population size threshold [24
]. Then we explore the employment shares in 2000 and 2010, as well as the correlations of employment growth in these five submetro sections. In this study, we make two major assumptions. First, we assume that the growth of the metros is independent, e.g., two metros’ main-centers are not correlated with one another. Second, we assume that the metros are mature economic units that are highly diverse in industrial composition. Although different employment clusters in a metro may change in different directions (e.g., decline, remain) [7
], in this study we look at the average effect of each submetro section, e.g., the sum of all sub-centers as a submetro section.
The significance of this study is twofold. First, the dynamics of employment growth in five submetro sections can provide policymakers with more accurate knowledge for proactive planning. Especially, we add the section of non-center clusters, which represent the gradual process of forming a sub-center; these clusters may also play a significant role in capturing agglomeration economies. Second, most studies on US metropolitan structure focus on the major metros (e.g., Los Angeles, Chicago); the repeated emphasis on the same cities results in biased understanding regarding the US metro structure as a whole. Additionally, those case studies provide little insight for small and midsize metros’ growth patterns. Our detailed analysis of all US metros paints a more complete picture of the US metro structure, which may serve as a benchmark for national and regional policy reference.
4. Discussion and Conclusions
This paper investigated US metros’ spatial structures and evolution patterns. We use employment shares in five submetro sections to denote a metro’s spatial structure and use submetro growth patterns to describe the metro’s evolution process. The measurement methods are adopted from literature with minor modifications. We aim to make the methods easy to understand and applicable to the entire country. This study takes a closer look into the metropolitan spatial structures for all 361 metros. Policymakers can clearly see which stage their metro is currently at, and which stage it is heading to, as well as identifying similar metros. Additionally, other countries may use the same methods to conduct their own countrywide study.
We found that metros’ evolution patterns vary greatly across the three size categories. First, small metros’ main-centers agglomerate employment from the other four submetro sections: sub-centers, non-center clusters, non-cluster urban areas, and rural areas, in decreasing the order of likelihood. Non-center clusters tend to disperse into non-cluster urban areas. Therefore, for metros with a population of less than a quarter million, policymakers should focus the resources on creating jobs in the main-centers.
Second, midsize metros seem to be in the critical period to form sub-centers and reduce urban expansion (to rural areas). Midsize metros’ sub-centers have the largest mean growth share among the three categories, and not only compete with the non-center clusters but also the rural areas. Additionally, unlike large metros, midsize metros have a relatively large room for urban development. Therefore, for metros with a population of over a quarter but less than a million, policymakers should provide more convenience for sub-center development.
Third, large metros’ evolution patterns are of high regularity: Growths in the five submetro sections positively associate with one another, except for the main-center (that competes with the sub-centers). Additionally, large metros are much more likely to have growth in the low-density areas (i.e., non-cluster urban areas and rural areas) than the high-density areas. In fact, the growths in the sub-centers and non-center clusters are extremely small, indicating that the metros have already completed sub-centering before entering into the large metro category. This further confirms the importance of laying out a proactive plan at the midsize metro stage. After a metro’s population exceeds one million, large-scale changes to its spatial structure become difficult, but policymakers can still put efforts in increasing the development density and renewing the main-center.