Comparing the Degree of Urbanization to the US Census Bureau’s Urbanized Areas, Urban Clusters, and Rural Areas Michael Ratcliffe, Michael Commons, and Jennifer Zanoni US Census Bureau United Nations Expert Group Meeting on Statistical Methodology for Defining Cities and Rural Areas New York, NY USA 28-30 January 2019

Census Bureau Urban Areas Urbanized areas: 50,000 or more population. Urban clusters: at least 2,500 and less than 50,000 population. Defined primarily based on residential population density measured at the census tract and census block levels. Initial urban core: at least 1,000 per square mile (386 per km2) Remainder of urban area: at least 500 per square mile (193 per km2) 2010 Census Population 2010 Percent Total 308,745,538 100.0 Urban 249,253,271 80.7 Urbanized Area 219,922,123 71.2 Urban Cluster 29,331,148 9.5 Rural 59,492,267 19.3 The Census Bureau defines two kinds of urban areas: urbanized areas of 50,000 or more population and urban clusters of at least 2,500 and less than 50,000 population. Both areas are defined using the same criteria, which primarily focus on population density used to identify the continuous area of densely settled population, but also include measures to account for non-residential urban land uses as well as densely settled areas that are separated from the main body of the urban area by relatively short distances. We apply two different population density thresholds: 1,000 persons per square mile (386 per square kilometer) to define the initial urban core and 500 persons per square mile (193 per square kilometer) to delineate the remainder of the urban area contiguous with the core. The difference in population density thresholds applied by the Census Bureau and those used in the EU classification (1,500 per square kilometer/3,884 per square mile for urban centres and 300 per square kilometer/777 per square mile) are critical for understanding the differences in results. In addition, the different minimum population size thresholds (2,500 versus 5,000) also are critical.

Distribution of urbanized areas and urban clusters in the United States: urbanized areas in purple (full extent shown); urban clusters in green (point locations only). Map included only to provide quick reference.

Degree of Urbanisation

Findings and Results The European Commission’s (EC) overall classification of urban and rural population and area compares favorably with the Census Bureau’s urban/rural classification. We see close comparability in the urban centre/urbanized area categories and in the rural categories: 98% likelihood that an EC-urban centre cell is urban in the Census Bureau’s classification. 98% likelihood that an EC-rural cell also is rural in the Census Bureau’s classification. Differences at all scales can be explained by: Differences in population density thresholds and other criteria. Differences in geographic units of analysis: grid cells v. census blocks. For the sake of time, I’ll provide an overview of our findings and results before highlighting a few examples. The EC’s overall classification of urban and rural population and area compares favorably with the Census Bureau’s urban/rural classification. In particular, we see close comparability between the EC’s urban centre category and the Census Bureau’s urbanized area category as well as between the two rural categories. In other words, there is a high likelihood that a cell defined as urban centre in the EC model also will be urban in the Census Bureau’s model. We see a similarly high degree of likelihood for rural cells. That said, we do see differences between the two classifications at all scales, most of which can be explained by differences in density thresholds and other criteria as well as differences in geographic units of analysis.

Findings and Results In terms of definitions, the EC’s Level 1 categories are similar to the Census Bureau’s categories. In terms of spatial extent, the EC’s urban categories differ from the Census Bureau’s. The EC’s urban centre category generally covers a smaller spatial extent than Census Bureau urbanized areas, but generally corresponds to the initial cores of urbanized areas. The EC’s urban cluster category differs substantially from the Census Bureau’s urban cluster category. Encompasses territory within Census Bureau urbanized areas that might be considered “suburban.” Includes areas defined as urban clusters by the Census Bureau. Does not include some smaller Census Bureau urban clusters. On a cautionary note, while the EC’s categories are similar to the Census Bureau’s categories in terms of definitions, there are substantial differences in spatial extent. The EC’s urban centre category more closely resembles the initial urban cores defined as part of the Census Bureau’s urban area delineation process; however, the higher population density threshold employed by the EC has a tendency to underbound the urban centre relative to the Census Bureau’s urbanized area definition. The EC’s urban cluster category is more expansive than the Census Bureau’s urban cluster category, encompassing territory within urbanized areas (what might be considered “suburban”) as well as the Census Bureau’s urban clusters (at least those of 5,000 or more population).

Comparison of Urbanization Classifications: Washington, DC Area EC Global Human Settlement Classification Census Bureau Urban Area Classification

Comparison of Urbanization Classifications: Atlanta Area EC Global Human Settlement Classification Census Bureau Urban Area Classification

Degree of Urbanisation

Total area for corresponding classifications Percent of each Euro-classification that falls within each Census-Classification Percent of each Census-classification that falls within each Euro-Classification

Percentage of Urbanized Areas dominated by each of the five proposed categories Percentage of Urban Clusters dominated by each of the five proposed categories

HUD’s American Housing Survey Share of households describing their neighborhood as… Urban Suburban Rural Total Census Urban Areas Urbanized Areas 32% 63% 5% 100% Urban Clusters 28% 45% 26% 16% 79% Source: American Housing Survey, 2018 Footer area

Concluding Thoughts The EC’s overall classification of urban and rural population and area compares favorably with the Census Bureau’s urban/rural classification. Differences related to density thresholds, geographic units of analysis, minimum population thresholds, and other criteria are to be expected. Differences between the two classifications do not indicate that one is better than the other, but do raise questions about the Census Bureau’s criteria and classification that warrant further consideration. EC’s overall classification of urban and rural population and area compares favorably with the Census Bureau’s urban/rural classification. Differences related to density thresholds, geographic units of analysis, minimum population thresholds, and other criteria are to be expected. Differences between the two classifications do not indicate that one is better than the other, but do raise questions about the Census Bureau’s criteria and classification that warrant further consideration. Example: Should the Census Bureau apply different criteria at different scales to avoid overbounding of smaller urban areas?

Concluding Thoughts: Terminology Overall, the terms used in the EC’s Level 2 categories are appropriate and useful for the US settlement system. Terminology, however, is always a challenge. Terms such as “town,” “village,” and “suburban” have different meanings and perceptions depending on context. “Suburb” implies a spatial relationship with an urban center/city. Yet, in the EC’s classification, there is no explicit linkage. In the US settlement context, areas classified as “town” in the EC’s classification may have functional relationships with larger urban centers and are, in effect, suburbs. Should we talk instead about Settlement Intensity? Impact on the landscape is the more critical issue, along with questions of access/proximity to, or isolation from, larger concentrations of population and economic activity. EC’s overall classification of urban and rural population and area compares favorably with the Census Bureau’s urban/rural classification. Differences related to density thresholds, geographic units of analysis, minimum population thresholds, and other criteria are to be expected. Differences between the two classifications do not indicate that one is better than the other, but do raise questions about the Census Bureau’s criteria and classification that warrant further consideration. Example: Should the Census Bureau apply different criteria at different scales to avoid overbounding of smaller urban areas?

Thank you! Questions? Contact Information: Michael Ratcliffe Geography Division US Census Bureau 301-763-8977 michael.r.ratcliffe@census.gov