The results from the 2020 census provided the opportunity to extend the urban patterns research forward by a decade, extending the coverage from 1950 to 2020. Rather than simply updating the original urban patterns dataset, I made the decision to create a new datset that benefits from the experiece gained in using the first version and that also makes makes use of new data. The new urban patterns 2 dataset has been made as consistent as possible with the 2020 Census Urban Area definition, greatly changed from the previous censuses.
A detailed description documents the new urban patterns 2 dataset and is provided in the first paper listed below.
Urban patterns 2 research papers
Urban Patterns of the Largest Urban Areas in the United States, 1950-2020: A New Dataset
A new dataset has been developed using housing units for census tracts to identify the 56 largest urban areas in the United States in 2020. These areas have been delineated following as closely as possible the procedures being used by the Census to define Urban Areas for 2020. Census tract counts and estimates for each census back to 1950 have likewise been used to define urban areas in the same way for each census year. The result is a dataset having urban areas defined in a consistent manner for each year with census tract data on housing units that can be used to study patterns within the urban areas and their change over time.
Change in the Size and Distribution of Large Urban Areas in the U.S., 1950-2020
A dataset with the numbers of housing units for the 56 largest urban areas in 2020, going back to 1950, is used to examine the changes in their sizes. While all of the areas more than doubled in size, the growth of some was astounding, adding over 99 percent of their 2020 housing units since 1950. The differences in rates of growth produced significant changes to the list of the largest 15 areas, with six new areas added compared only one change in the previous four decades. The areas in the South grew most rapidly. In 1950, the large urban areas in that region accounted for by far the smallest share of all of the housing units in the large urban areas but had the largest share in 2020. In the final decade, nearly half of all growth in housing units for the large urban areas was in the South. The distribution of the sizes of urban areas is often thought to conform to Zipf’s law, postulating that the size and rank are related by a power law with an exponent of one. In 1950 the exponent was slightly greater than one. The exponent dropped by 2020 such that the the smaller urban areas were larger than expected.
Density of Large Urban Areas in the U.S. and Barriers to Urban Expansion, 1950-2020
Housing unit densities are examined in 56 large urban areas in the United States defined in a consistent manner from 1950 to 2020. The mean density declined slightly over this period but this masks tremendous variation across the urban areas. Some of the most dense urban areas at the start experienced large drops, but substantial numbers of areas had increases in density, some large. Densities across regions changed dramatically, with mean densities for urban areas in the West rising from only slightly above the South to the highest by 2020, well above the Northeast and the Midwest which were highest in 1950. Density and density change are related to the size of the urban area (number of housing units), and change is also related to change in size and (negatively) to density at the start. The effect of potential barriers to expansion on density is investigated, with strong, significant effects of water and mountains on urban area densities.
Density and Sprawl in the Older and Newer Parts of Large Urban Areas in the U.S., 1970-2020
To examine housing unit density in the older and newer portions of large urban areas, the urban core is identified as the territory included in the urban area in 1940, with areas added subsequently designated the suburban periphery. And since urban sprawl is frequently associated with development since World War II, density in the periphery can be considered to be a measure of the extent of sprawl. The 56 large urban areas are classified into 4 groups based on low or high density in the core and the periphery. Areas in the two groups having low densities in the periphery are concentrated in the eastern half of the United States, while the areas with high periphery densities are in the South and West. Areas in the classes varied in the change in density over time and the size and rate of growth of the entire urban area. The latter two urban area characteristics along with the presence of water and mountains as barriers to urban expansion are associated with density and the change in density in the urban core and suburban periphery.
The Negative Exponential Decline of Density in Large Urban Areas in the U.S., 1970-2020
The central density, density gradient, and goodness-of-fit for the negative exponential model are estimated using housing unit densities for census tracts for 40 large urban areas from 1970 to 2020. Nonlinear regression is used to estimate the model parameters. The values of the estimates in 2020 show extremely wide variation across the urban areas. Larger and older urban area, especially in the Northeast, tend to have the highest central densities, gradients and R2 values, while areas in the Sunbelt are among the lowest. These values have generally declined from 1970 to 2020, again with wide variation across areas. Mean R2 drops from 0.41 to 0.26, suggesting declining importance of the Central Business District associated with the decentralization of employment. The R2 values for some areas in 2020 are less than 0.10; the negative exponential model is accounting for a very small proportion of the variation in density. But despite an overall decline in central density over the fifty-year period, the mean increases in the most recent decades, with an especially large jump from 1970 to 2020, raising questions regarding possible shifts in the patterns of urban change.
The Extent of Centralization of Housing Units in Large American Cities, 1970-2020
The centralization of housing units in 56 large urban areas from 1970 to 2020 is measured using an index based on the ratio of the mean distance housing units are located from the center to the mean distance if the housing units were uniformly distributed across the urban area. Mean centralization declines from 1970 to 2010 and then increases in 2020. Highest levels of centralization tend to occur in old large urban areas, especially in the Northeast. Centralization increases with urban area size and declines as areas grow more rapidly. Measures based on units in the urban core and suburban periphery, and the central density from the negative exponential model are reasonable measures of centralization. The negative exponential density gradient is not. Several of these measures show patterns over time similar to the centralization index, first declining and then increasing in recent decades. This raises the possibility that long-standing trends in urban decentralization may be ending.
Racial and Ethnic Diversity in Large Urban Areas in the U.S., 1980-2020
The racial and ethnic diversity of 56 of the largest urban areas has increased dramatically from 1980 to 2020. The average share of the population White dropped from three-quarters to a little over half while the average Latino share nearly tripled. The mean value of an index of diversity jumped from 49 to 74 over the same period. Urban areas with the highest levels of diversity had much less than half of the population White with varying mixes of Blacks, Latinos, and Asians. Areas with the lowest diversity had high percentages of their populations White with one exception (El Paso with a large Latino population). In 1980, average diversity in the suburban periphery was far lower than in the urban core, especially for areas in the Northeast and Midwest. Suburban diversity increased substantially by 2020 in all regions, but remained far lower in the Northeast and Midwest.
Racial and Ethnic Diversity in Neighborhoods in Large Urban Areas in the U.S., 1980-2020
Racial and ethnic residential diversity in urban areas is measured using an index of exposure diversity, which is the mean of diversity within census tracts, and an index of evenness diversity relative to the diversity of the urban area. These indexes are used to examine neighborhood diversity among whites, blacks, Latinos, and Asians in 56 large urban areas in the United States from 1980 to 2020. Mean diversity across the urban areas increased dramatically. But substantial differences exist between exposure diversity and evenness diversity. An area with very low exposure diversity can have very high evenness diversity when the overall diversity of the urban area is also low. Diversity tends to be lower in urban areas in the Northeast and Midwest, highest in the West. Lower percent population white and decreases in the white population in the prior decade are associated with higher levels of exposure diversity with just the opposite for blacks, Latinos, and Asians. Average levels of the two measures of diversity increased in both the urban core and suburban periphery, but exposure diversity was higher in the core while evenness diversity was higher in the suburbs.
Changes in racial and ethnic diversity in neighborhoods in large urban areas in the U.S., 1980-2020
The shares of the population of census tracts white, black, Latino, and Asian are used to calculate a measure of neighborhood racial and ethnic diversity for the tracts in 56 large urban areas from 1980 to 2020. Diversity has increased dramatically over the period, with about three-quarters of all tracts increasing in each decade. Multilevel models are used to evaluate the association of tract and urban area characteristics with the change in tract diversity from 1980 to 1990 and from 2010 to 2020. Tract characteristics account for a much larger share of the variation in change in diversity than urban area characteristics. The change in the tract percent white is negatively related to diversity change while the percent white is positively related. For the other three groups the directions of the relationships are reversed. Tract diversity at the start of the period is negatively related to change in diversity (more opportunity for increase?) while urban area diversity is positively related (greater tolerance for diversity?). Population growth is positively and significantly related to diversity change, but the effect is smaller.
Urban Patterns 