The OMB is proposing raising the threshold for a Metropolitan Statistical Area from 50 K to 100K. 144 cities would be reclassified as Micropolitan Areas if the proposal is adopted.
The Census Bureau is proposing a change to the threshold for an Urban Area from 500 people per sq mi to 385 housing units per sq mi. That is roughly equivalent to 1000 people per sq mi.
https://apnews.com/article/wisconsin-bismarck-census-2020-north-dakota-sheboygan-ad77e15f0f8cd13b8e398d2ca8339ca7
I missed this thread initially, so I merged my thread onto this one. One extra thing in the article is the proposed Census Bureau change. It seems confusing to up the OMB urban area population threshold while changing the urban area definition to housing.
Here is the federal register notice:
Urban Areas for the 2020 Census-Proposed CriteriaWorth noting is that the urban area classification is a Census Bureau standard, while the Core Based Statistical Area classification is an OMB standard, albeit with the data provided by the Census Bureau.
The Census Bureau definition is more a land use characterization - an area of dense residential population, and associated urban land use (determined by impervious ground cover, such as buildings and parking lots).
The rationale for switching to Housing Units from Population is two-fold:
(1) It permits updating urban areas mid-census. The Census Bureau maintains a Master Address File (MAF) intended to represent every Housing Unit in the country. This is used as the sampling frame for the ACS, and is continuously (periodically?) updated, as the Census Bureau becomes aware of new housing developments such as from the USPS list of mailing addresses.
One of the purposes of the Update/Leave phase of the Census, was to field check housing units that likely no longer exist. For example, Audubon County, Iowa had two small Update/Leave areas. One was a trailer park that satellite imagery showed had been declining over the years, and now has a building covering much of the site.
You can't determine non-existence by mailing a form, and getting no response. That is indistinguishable from a vacant housing unit or one with a non-respondent. Of the roughly 60 million Housing Units for the Update Leave phase, 10 million were found to be non-existent. The Census Bureau had a strong hunch that they weren't. The 150 million housing units for the Census never made sense, given that the current estimate for the ACS is around 135 million.
Since the list of Housing Units is maintained on a continuing basis, Housing Unit density can be calculated to a census block level on a continuing basis. 385 HU/square mile is equivalent to 1000 Persons/square mile at the national average household size of 2.6 (1000/2.6 = 385). The Housing Unit basis might also be better for areas with seasonal populations.
(2) The Census Bureau is intending to fuzz population counts, including for PL 94-171 data, for reasons of privacy disclosure. Since actual population counts will not be released, density can not be calculated in a transparent fashion at the census block level.
And in any event, an accurate population count is only determined at the Census.
The proposal would eliminate the distinction between urbanized areas and urban clusters. Prior to 2000, only urbanized areas, those with populations greater than 50,000 population were defined. Places (both incorporated and CDP) with more than 2500 person were classified as urban.
In 2000, the Census Bureau switched to an automated procedure based on population density that disregarded any political boundaries, and applied it to the entire country.
All areas were termed "urban areas". The name "urbanized area" was retained for those over 50,000 population that largely corresponded to the 1990 and earlier areas. The newly determined smaller areas, with population between 2500 and 50,000 were termed "urban clusters".
The 2020 proposal would eliminate the distinction. From a statistical perspective there is no reason for a distinction. Any programs based on the terminology distinction, would be better based on the actual size distinction.
Densely populated areas with populations less than 4,000 housing units/10,000 persons will no longer be considered urban areas, and the population will no longer be classified as urban.
This is more consistent with modern sensibilities as to what is urban, and also matched the OMB threshold for what constitutes the core of a micropolitan area.
It is a difference from long-standing Census Bureau practice. OTOH, the Census Bureau has also long characterized cities by population classes (more than 1,000,000, more than 100,000 etc.). In addition, the Census Bureau publishes detailed demographic information and estimates for all incorporated places.
The 2020 proposal would reduce the "jump" limit to 1.5 miles. Currently, urban areas may jump across non-residential areas for a distance of 2.5 miles, to reflect discontinuous development along highways which are part of the urban population.
Before 2000, the jump limit was 1.5 miles, even though urban areas were defined differently. The 2.5 mile limit was adopted perhaps out of concern that an automated process might produce bunches of urban clusters that were treated as independent of the nearby city even though they functionally were not. In 2010, the Census Bureau considered a return to the 1.5 mile threshold, but the response was equivocal. It appears that the Census Bureau is simply going back to what they wanted to do in 2010.
Linking census blocks will no longer be required. Urban areas may thus become discontinuous, but this would reflect actual land use.
The "hop" limit would remain at 0.5 miles. An urban area may only have one jump along a highway, but may have unlimited hops. These too may now be discontinuous.
Urban areas may continue to cross water and wetlands for distances up to five miles.
Indentations in urban area boundaries surrounded by urban areas would no longer be closed. Previously it was believed that these areas would:
(1) Become developed;
(2) Had non-residential urban uses; or
(3) Would make maps neater.
It turns out that these areas have not developed - and if they did, they would be included in the urban area. The Census Bureau has switched to a standard based on impervious ground cover (parking lots and buildings) to detect non-residential land usage. This is based on more recently available satellite imagery. Maps can be zoomed in or out. Larger scale cartographic applications can simplify boundaries.
Fully enclosed enclaves of less than 5 square miles would continue to be included.
The 385 HU/square mile would be used for the entire delineation process. Presently, an initial core of at least 1000 persons per square mile of at least 1000 persons is used to define the initial urban area. Extensions are only required to have a density of 500 persons per square mile.
I doubt the change will have much impact. Suburban areas with 1/4 acre lots have a population density of around 6700 persons/square mile. Even one acre lots produces blocks with about 1700 persons/square mile.
Blocks with group quarters adjacent to a densely populated residential area will be included in the urban area. Since there are no housing units associated with group quarters, the housing unit density can not be calculated.
I had wondered whether an artificial HU density could be calculated, but it might not be required. Larger facilities (100+) would qualify unless they were on a relatively large plot of land. Smaller facilities (20-) may qualify based on other Housing Units on the same census block. And for practical purposes, group quarters near urban areas are actually part of the urban landscape.
Truly isolated group quarters such as prisons won't qualify as reaching the 10,000 population threshold.
It is proposed that agglomerations of urban territory be merged or separated based on commuting patterns. Currently, urbanized area can absorb urban clusters, or urban clusters can merge. When two existing urbanized areas meet, they are separated at an isthmus of development or near the county line associated with their metropolitan areas.
The current separation is not based on current objective criteria, but is based on previous analysis which may not reflect current reality.
If 50% of the smaller urban areas workers work in the larger urban area, and 50% of the jobs in the smaller area are filled by workers from the larger area, the two urban areas will be merged. Essentially, the smaller area no longer maintains a separate identity.
Otherwise the boundaries can be adjusted.
If I understand the process, commuting basins are determined and smaller urban areas can be expanded to include territory commuting towards them.
