2012 county & metro area estimates released today

[jimrtex]

The ten fastest-growing counties with a population over 10,000 and the likely reason for their growth were:
10) Franklin County, WA (Pasco)  - Tri-Cities area has been growing

All but Franklin County experienced double-digit growth.  Williams County, ND grew by almost 20%.

The growth in the Tri-Cities has been decidedly eastward (away from Hanford), and part of the Urbanized Area extends into Walla Walla County.    In 1980 the Tri-Cities were Richland 34K, Kennewick 34K, and Pasco 17K.  Pasco wasn't really a city, and it was a stretch to call it "Tri-"

In 2010, it was Kennewick 73K, Pasco 59K, and Richland 48K.

[jimrtex]

Texas Counties that have grown more than 5% between the 2010 Census and July 1, 2012

Andrews      16  9.0%   Permian Basin
Williamson  456  7.9%   Austin Suburbs
Kendall      35  7.6%   San Antonio Suburbs
Hays        168  7.6%   Austin Suburbs
Hemphill      4  7.2%   Panhandle
Fort Bend   627  7.2%   Houston Suburbs
Midland     146  7.1%   Permian Basin
Travis     1095  7.0%   Austin
Denton      707  6.7%   Fort Worth, Dallas Suburbs
Collin      834  6.7%   Dallas Suburbs
Montgomery  485  6.4%   Houston Suburbs
Guadalupe   139  6.3%   San Antonio Suburbs
Rockwall     83  6.0%   Dallas Suburbs
Comal       114  5.5%   San Antonio Suburbs
Lipscomb      3  5.4%   Panhandle
Ector       144  5.2%   Permian Basin
Gaines       18  5.1%   Permian Basin

[jimrtex]

They added Benton County MS to the Memphis metro. That decision belongs in the deluge. It's an hour outside of town and at least a half hour beyond any signs of civilization. We're not Atlanta. People don't commute across three counties here.

The procedure for defining core-based statistical areas is automated.

The Census Bureau starts by defining Urban Areas which are densely populated areas (500+ per square mile).  If 50% of a county population is in urban areas of at least 10,000 population; or the county has 5,000 persons in single urban area with more than 10,000 persons it is a "Central County".

Shelby, TN: 97% of population is in Memphis UA or Arlington UC.
Tipton, TN: 19,000 population in Atoka UC
Fayette, TN: Not a central county, 7K in Oakland UC, tiny bits in Memphis UA and Arlington UC

DeSoto, MS: 80% of population is in Memphis UA
Tunica, MS: Not a central county, 4K in Tunica UC
Tate, MS: Not a central county, 7K in Senatobia, UC
Marshall, MS: Not a central county, 6K in Holly Springs, UC
Benton, MS; not a central county, no Urban Areas

Crittenden, AR; 79% of population in Memphis UA

The definition of an Urban Area includes hops and skips which permit an UA to extend along highways, and in this case across the Mississippi to West Memphis.  The definition of central counties is done for the entire country.  I just showed the counties that ended up in the Memphis MSA.

So you have the following Central Counties:

Shelby TN, DeSoto MS, Crittenden AR for Memphis Metropolitan Statistical Area
Tipton TN for a hypothetical Atoka Micropolitan Area

Then it is determined whether Outlying Counties should be added to the CBSA.  This is based entirely on commuting patterns:

If 25% of the employees resident in the county work in a central county, the county is an outlying county; or if 25% of those employed in a county come from a central county, the county is an outlying county.

Tipton TN: 60% of employed residents work in other Tennessee counties, and probably most of these in Shelby.

Fayette, TN: 64% in other Tennessee counties.

Tate, MS: 32% in other Mississippi counties, and 19% in other states.   Since most of those working in other states probably work in Shelby County, only about 1/5 of the Mississippi intercounty workers have to be to DeSoto County.

Marshall, MS: 44% in other states, 22% in other Mississippi counties.

Benton, MS: (3153 employed workers): 22% in Benton County; 49% in other Mississippi counties, 29% in other states.  Basically gets included because there are very few jobs in the county, and the population is not high enough to support much of a commercial or service sector.  Likely no Walmart, few doctors, etc.   So you have school district employees and farmers.

50% of Benton employees worked in a M/MSA (this would include both outlying and central counties, but most of the jobs are going to be in Shelby and DeSoto counties, and places like Jackson, TN and Tupelo, MS are a long haul).

It wouldn't surprise me if there were a lot of people who worked for FedEx or other jobs at the airport.  Many of these would not be 9-5/M-F jobs, so the commute would be at off hours (it is right at 60 minutes).  If you wanted to own 40 acres in the country, it could be affordable, and if you were working 4 nights a week, have daylight to keep up on it.

Tunica, MS.  8% of workers in other Mississippi counties, 8% of workers in other states.  So Tunica is included based on the reverse flow of employees in the casino and related businesses (restaurants, hotels, bars) driving in. Tunica itself only has 3800 locally-employed employees.

There is also a contiguity requirement, but both Fayette, TN and Marshall, MS supply this to Benton, MS.

There is also the Combined Statistical Area, the lovely-named Memphis-Forrest City CSA.  Forrest City Micropolitan Statistical Area is sort of a satellite of Memphis Metropolitan Statistical Area with 15% employment interchange.

[jimrtex]

Wayne county Michigan still losing people but only down to 1,792,000 from 1,820,000 in 2010. So Wayne county is on pace to lose 140K by 2020. Also on the bright side Michigan only lost 275 people in the last 2 years compared with 55k between 2000 and 2010. So Michigan will still lose a congress seat in 2020 unless we have some huge growth in the next 8 years but are we at risk to lose 2 seats. Does anyone know, would negative growth cause it.

To lose an N^th seat, a state has to have a growth rate about 1/N slower than the country.   So to lose a 14th seat, would mean that Michigan would have to grow about 7.1% slower than the country as a whole.  Between 2000-2010, the country grew about 10%, but will probably grow around 8% between 2010-2020.  So if Michigan had no change for the decade it would be just a bit less than 7.1% slower.   So if Michigan has zero change the rest of the decade, it will be at a rate to lose 1 seat and no more.

But there is a complication of rounding.  In 2010, Michigan was entitled to 13.9 representatives.  Projecting the changes forward to 2020, would result in 12.9 representatives, which would round to 13.   If Michigan were to lose population, it would edge toward the 12.5 mark and be subject to loss of an additional representative.   This would require a loss of 3.8% or around 338,000.   Because rounding is not done independently, there is a fuzzy zone, perhaps from a loss of 2.5% to a loss of 5.0%.  At 2.5% Michigan would have to be lucky to lose the extra seat, at 5.0% it would be lucky keep the seat.

[jimrtex]

Great map, Sheliak.

And thanks for that explanation of metro areas, jimrtex. I'd wondered before why the census's metro areas often seem to include a layer of rural counties that wouldn't be counted by the everyday understanding of the metro area.

It is somewhat schizophrenic.   An extreme example is Armstrong County which is part of the Amarillo metropolitan area and has a population just under 2000.  It also has an interstate highway, so it is an easy commute into Amarillo.

[jimrtex]

As an aside, it blows my mind that 500 people per square mile is considered "urban".  I have a hard time applying that adjective to areas ten times as dense sometimes.  (This is what happens when you grow up in a suburb whose density is roughly 9000 persons per square mile.)

Historically, any town with a population above 2500 was classified as "urban", while anything else was considered rural.   Around 1950, the Census Bureau started defining urbanized areas to reflect that the fringe around larger cities wasn't necessarily being formed into incorporated cities.

In 2000, the delineation of urban areas was switched to be entirely based on density, to recognize that city limits quite often do not reflect land use - often including large swaths of undeveloped land, and other areas that are not legally defined as cities or town.  An area must have 2500 persons to be classified as an "Urban Area".  Urban Areas with population above 50,000 are "Urbanized Areas" while those between 2500 and 50,000 are "Urban Clusters".

9000 per square/mile is pretty dense.   An area with single family houses might have half that, without any land being used for parks, schools, or commercial areas.

The 500/square mile is not the average for the entire area, but rather for each area that is added.  Agricultural land won't come close to qualifying, even if it is on the edge of a town and some lots have been carved off and a few houses built.   And any formally subdivided land will easily surpass ir - though it might not reach the 2500 total population limit.

An area with 3 and 4 acres lots would be near the threshold, but it would difficult to sustain over large areas.

[jimrtex]

So I actually bothered to look it up with my own eyes, and while 500 people per square mile is emphatically the exurban fringe, the ratio of subdivisions to rural uses is higher than I had imagined.

Inside the dotted line is Buckingham Township, in Bucks County PA.  497 people per square mile.



And here's a view from the ground: new houses on one side of the road, corn on the other.



...

I think the lesson here is that 500 people per square mile is a very unnatural density- the largest-lot subdivisions will be higher, anything purely rural (even including hamlets, and in wet climates) will be lower.  You need a hodgepodge to hit that mark. 

If you get fine-grained enough, down to the block, 500 people per square mile is too low I think.  But at a township level it's more defensible.

The township doubled in population between 1970 and 1990 and again between 1990 and 2010.  I don't see how it could not be considered urban (don't be confused between the Census Bureau definition of urban - which is essentially non-rural; and terms like suburban.

The 2010 density is 606 ppsm.  The Philadelphia Urbanized Area includes most of the western part of the township (west of York Road) expanding northward from (Doylestown).   I'll refer to northwest as "west", away from Philadelphia.   About 2/3 of the population in the eastern half is in the Philadelphia UA, but very little of the land.   The subdivisions are thicker to the west, so I suspect that enclaves are being closed up and added to the UA, while to the east you have exclaves.   There is also a strong of hops so that the UA just reaches across the Delaware into New Jersey at New Hope/Lambertville along US 202.   Note about 1/4 of the employees in some of the census tracts work out of state.   It is about an 1-1/2 into NYC via either I-78 or I-278 (which would be the quicker route into Brooklyn or Staten Island), so I'm guessing it would be more NJ than NY.

The numbering of the census tracts shows the development of the area.   When census tracts are modified, they are renumbered, so as to avoid comparing different areas between censuses.

The tracts are 1045.02, 1045.03, 1045.05, 1045.06 (with .01 and .04 missing).

So originally it was all tract 1045 (census tracts have a target population of around 5000).  The first split would have been between 1045.01 to the west, and 1045.02 to the east, which still exists.  Then it is likely that 1045.01 would have been divided between 1045.03 to the south, adjacent to Doylestown and 1045.04.   And then 1045.04 would have been split between 1045.05 and 1045.06.

1045.05 is the middle part of the western half somewhat pie-shaped.  The subdivision on the western edge is prominent in the satellite view because the trees haven't grown up around the houses yet, so probably 10 years or less.  This area has an average family size of 3.6, which means most families have children.  Since few families have 4 or 5 children any more, you have to have 2 in most families to get that close to 4.  Compared to 1045.02 to the east, this area is better educated, less German, and less likely to have been born in Pennsylvania.  So the area to the east which is less developed must have a residual population that is native to the area (at least Bucks County.

[jimrtex]

FWIW, I suspect the reason 500 sq/mile is used as a cutoff is because you have lots of counties that have a small city that has borders that extend far beyond the actual populated area. Lots of places in Alaska like that for example. Similar to how the county I grew up in has a population density of only 35 people per square mile but the city is more urban than anywhere in many counties with 20 times the density, it's just the city and LOTS of empty space.

They went to a density-based measure in 2000 specifically because formal boundaries did not correspond to land use, either the boundaries extended beyond the developed area, or vice versa.

Reading the federal register it appears that the 500 ppsm was to permit automated delineation.

Before 1950, the Census Bureau simply defined places (towns and cities) with population above 2500 as urban, and everything else as rural.  Beginning in 1950, the census bureau began defining urbanized areas around cities of over 50,000 to recognize that there were often unincorporated suburban areas adjacent to large cities.  The delineation of urbanized areas was not automated.

The 1990 census was the first to completely cover the country in census blocks.  In defining urbanized areas, analysts were permitted to identify non-residential urban uses of census blocks, and calculate a density based on the residential area - using a threshold of 1000 ppsm.

In 2000, the process was automated, and urban areas were defined based entirely on density, with no regard to city or town boundaries.  Urban areas were also defined for much small populations - replacing the old definition based on city boundaries in small town America.

To permit the process to be automated, a density of 500 ppsm was adopted - which avoided a need to determine land use.   A typical census block in a suburban area might have a density of a few 1000 (5 acres X 16 houses x 2.5 persons/household = 5120 ppsm).  Even an area with acre lots would have 1600 ppsm.

In addition, the area of moderate density must be adjacent to an initial core of over 1000 ppsm, or reachable through jumps and hops, which will disqualify isolated small areas that might reach the 500 ppsm.

[jimrtex]

They added Benton County MS to the Memphis metro. That decision belongs in the deluge. It's an hour outside of town and at least a half hour beyond any signs of civilization. We're not Atlanta. People don't commute across three counties here.

The lede story in the Memphis paper today is about Benton County and its inclusion in the metro area. I would provide a link, but it is behind a strict paywall. Amongst the highlights of Benton County pointed out:
No Wal-Mart
No stoplight
No movie theater
No golf course

No jobs. 

That is why they commute to Memphis.  I bet bunches work for Fedex.  They can work at night so they don't have to commute during rush hour, and then have a bit of an acreage out in the country during the daylight.