Population Growth in CSAs and Metropolitan Areas, 1900-2016

[cinyc]

Below you'll find maps of the population change in 2012 Primary Statistical Areas (PSA), i.e. Combined Statistical Areas (CSA) and separate Metropolitan Areas from 1900-2016.  Combined Statistical Areas combine Metro and Micropolitan areas into larger areas.  Not every Metro is in a CSA.  I will try to do these maps in PSA Population Order.

PSA Rank (2014)
1.  New York-Newark, NY-NJ-CT-PA
2.  Los Angeles-Long Beach, CA
3.  Chicago-Naperville, IL-IN-WI
4.  Washington-Baltimore-Arlington, DC-MD-VA-WV-PA
5.  San Jose-San Francisco-Oakland, CA
6.  Boston-Worcester-Providence, MA-RI-NH-CT
7.  Dallas-Fort Worth, TX-OK
8.  Philadelphia-Reading-Camden, PA-NJ-DE-MD
9.  Miami-Fort Lauderdale-Port St. Lucie, FL
10. Houston-The Woodlands, TX
11. Atlanta–Athens-Clarke County–Sandy Springs, GA
12. Detroit-Warren-Ann Arbor, MI
13. Seattle-Tacoma, WA
14. Phoenix-Mesa-Scottsdale, AZ
15. Minneapolis-St. Paul, MN-WI
16. Cleveland-Akron-Canton, OH
17. Denver-Aurora, CO
----------
23. Pittsburgh-New Castle-Weirton, PA-OH-WV
----------
42. New Orleans-Metairie-Hammond, LA-MS

Metro Rank (2016)
1. New York-Newark-Jersey City, NY-NJ-PA
2. Los Angeles-Long Beach-Anaheim, CA
3. Chicago-Naperville-Elgin, IL-IN-WI
4. Dallas-Fort Worth-Arlington, TX
5.  Houston-The Woodlands-Sugar Land, TX
6. Washington-Arlington-Alexandria, DC-VA-MD-WV
7. Philadelphia-Camden-Wilmington, PA-NJ-DE-MD
8. Miami-Fort Lauderdale-West Palm Beach, FL
9. Atlanta-Sandy Springs-Roswell, GA
10. Boston-Cambridge-Newton, MA-NH
11. San Francisco-Oakland-Hayward, CA
12. Phoenix-Mesa-Scottsdale, AZ
13.  Riverside-San Bernardino-Ontario, CA
14. Detroit-Warren-Dearborn, MI
15. Seattle-Tacoma-Bellevue, WA
16. Minneapolis-St. Paul-Bloomington, MN-WI
19. Denver-Aurora-Lakewood, CO
21. Baltimore-Columbia-Towson, MD
26. Pittsburgh, PA
32.  Cleveland-Elyria, OH
35. San Jose-Sunnyvale-Santa Clara, CA
38. Providence-Warwick, RI-MA
46. New Orleans-Metairie, LA
57. Bridgeport-Stamford-Norwalk, CT
58. Worcester, MA-CT
65. New Haven-Milford, CT
67. Oxnard-Thousand Oaks-Ventura, CA
69. Allentown-Bethlehem-Easton, PA-NJ
77. Stockton-Lodi, CA
80. Akron, OH
107. Santa Rosa, CA
111. Port St. Lucie, FL
122. Vallejo-Fairfield, CA
129. Reading, PA
131. Flint, MI
132. Manchester-Nashua, NH
134. Canton-Massillon, OH
143. Trenton, NJ
146. Ann Arbor, MI
155. Boulder, CO
163. Greeley, CO
173. Olympia-Tumwater, WA
174. Santa Cruz-Watsonville, CA
176. Atlantic City-Hammonton, NJ
182. Bremerton-Silverdale, WA
183. Hagerstown-Martinsburg, MD-WV
207. Barnstable Town, MA
216. Athens-Clarke County, GA
221. Gainesville, GA
222. St. Cloud, MN
233. Kingston, NY
238. Dover, DE
252. East Stroudsburg, PA
266. Chambersburg-Waynesboro, PA
267. Vineland-Bridgeton, NJ
271. Sebastian-Vero Beach, FL
278. Monroe, MI
290. Napa, CA
297. Winchester, VA-WV
303. Hammond, LA
306. Sherman-Denison, TX
317. Mount Vernon-Anacortes, WA
325. Weirton-Steubenville, WV-OH
337. California-Lexington Park, MD
338. Michigan City-La Porte, IN
339. Kankakee, IL
363. Ocean City, NJ

[cinyc]

The New York-Newark CSA has grown every decade since 1900 except the 1970s:



But not every component Metro/Micro area has been so lucky:



In particular, the Kingston MSA (Ulster County) significantly lost population in the 1910s, and the MSAs on the outer fringes of the New York CSA have been losing population this decade.  As you can see from the map, the population loss of the 1970s was largely due to a decline in the core New York City MSA.

The county-level 1900-2016 map shows even more variation (the white numbers are population change):


Many NYC boroughs had been losing population before 1970, but the bulk of the NYC Metro's 1970s population loss was due to losses in New York City and its immediate suburbs.  Gains in the outer suburbs were not enough to make up for this population loss.

Note that the Bronx was part of New York County until 1914.  Some of the 1910-20 Manhattan (New York County) population loss was due to the separation.

Yearly county population estimates are available from 1980 onward.  I've mapped the yearly population change below.  I think there is a lot of noise in the 1999-2000 estimate change because the 2000 estimates were revamped after the census:



You can see how the NYC exurban counties started to lose population around 2009 or so.  This trend continues.

The last gif is of NYC CSA town population growth from 2000-16.  You've seen gifs like these in my Population Growth Patterns in Metro Areas, 2000-16 thread, but many of the maps weren't properly centered and didn't include the full CSA:



Finally, I've been asked to make some static maps of county and town population changes from 1980-2016 and 2000-2016.  Not every 2010 town existed in its current form in 2000, so there may be some discontinuities.  Those towns are marked in gold.

First, NYC CSA county population growth from 1980-16:


What's most stunning about this map is the growth of the Pocono exurbs and slight decline in Essex County (Newark).

Next, NYC CSA county population growth from 2000-16:


Compare that to NYC CSA town population growth from 2000-16:


Finally, an overall NYC CSA town population growth map from 2010-16.  This map might be good for redistricting purposes:

[cinyc]

Maps of the Los Angeles CSA are a bit more boring than New York, because there are only 5 counties in the CSA.

First, the CSA overview map:


The CSA has grown every decade, usually at a healthy clip.

Next, the map of the component metro areas:



Again, the three component metros grew every decade.

The decadal county growth maps show the same consistent growth pattern at the county level:


But the 1980-2016 yearly population estimate change maps show that Los Angeles County lost population a few years during the mid-1990s and mid 2000s.  Orange County also lost a little bit of population in the mid 2000s:



Next, the 2000-16 town population change map.  I've zoomed in to only include the major populated parts of Riverside and San Bernadino Counties.  The town population change maps pretty much confirm the OC and LA County slight population loss pattern:



Finally, two static town population change maps - 2010-2016 and 2000-2016.  I'm only going to make the county 1980-2016 and 2000-2016 maps on request, since there are so few counties in the LA CSA, and the maps wouldn't be that interesting:

2010-2016:


2000-2016:

[Tintrlvr]

Question for people who know more about SoCal/Orange County: How has Irvine grown so consistently faster than all its surroundings (including both more and less developed neighbors) for such an extended period of time?

[cinyc]

Like Los Angeles, the Chicago CSA has grown every decade since 1900:



But unlike Los Angeles, some of the component Chicago CSA Metropolitan/Micropolitan Areas haven't always grown.  The core Chicago-Naperville-Elgin Metro area has always grown, though:



The 1900-2016 decadal county map shows a somewhat erratic growth pattern.  Even Cook County has lost population at times:


And the 1980-2016 shows an equally erratic pattern, although the Collar Counties have generally grown - until recently:



The 2000-2016 town map gif shows early growth in the far-out exurbs, followed by growth inward, and now, not much growth at all, except perhaps in the far west and southwest suburbs:


The 1980-2016 static county growth map shows population losses in Cook, and gains in most of the Collar Counties:


The 2000-2016 County Map is similar:


The 2000-2016 Town Map shows an outward growth pattern:


And, as always, the 2010-2016 Town Map might be useful for redistricting:

[muon2]

Thanks for the Chicagoland analysis. My observation has been that the traditional suburban growth/city existed until the onset of the Great Recession. At the point the collapse of housing prices froze many people in place throughout the metro region. By 2013 housing prices had recovered enough to allow mobility, but the economy of the region was not experiencing the economic recovery of the rest of the nation. The effect is that the population declined overall. The maps seem to back up that observation.

One piece of conventional wisdom is that the outflow from the IL collar counties this decade was in part due to moves to neighboring states driven by tax policy. That CW seems to hold up to some degree in the northern collars of Lake and McHenry compared to Kenosha. It doesn't seem to hold for the eastern edge, since the neighboring areas of IN in the CSA aren't growing.

[cinyc]

Baltimore and Washington, DC are in the same CSA.  Like the federal government, the CSA has steadily grown since 1900:



But it really is a tale of two cities.  The current Baltimore Metro used to be larger than the DC Metro.  As the metro map gif shows, DC is now over twice as large:



The 1900-2016 decadal county growth map gif shows that both the cities of DC and Baltimore started steadily losing population in the 1950s.  But DC has been gaining population recently.  Baltimore generally has not:



With the raw growth numbers removed, you can see the percentage growth patterns of Virginia's independent cities more clearly:



The 1980-2016 yearly percentage population growth map gif shows that DC turned the corner around 2005 and is currently picking up steam. Baltimore continued to lose population, except for a few years:



Unfotunately, the 2000-16 yearly town map gif doesn't add a whole lot of data.  There are fewer incorporated areas in most of the south and west, and Census doesn't provide yearly population estimates for CDPs:



The static 1980-2016 map shows that DC has had net positive growth since 1980.  But the DC suburbs and exurbs have grown even more rapidly:



The 2000-16 static county map shows the same pattern:


For comparisons, the 2000-16 static town map is below:


And, finally, here's the 2010-16 static town map, which is marginally useful for redistricting purposes:

[cinyc]

One piece of conventional wisdom is that the outflow from the IL collar counties this decade was in part due to moves to neighboring states driven by tax policy. That CW seems to hold up to some degree in the northern collars of Lake and McHenry compared to Kenosha. It doesn't seem to hold for the eastern edge, since the neighboring areas of IN in the CSA aren't growing.

From the town maps I've already done, that seems to be the case in the St. Louis Metro.  Metro East Illinois is shrinking, while the STL Missouri exurbs are growing.

Right now in Chicagoland, it seems like only the far west and southwest exurbs are growing - along the newish I-355 corridor, perhaps?

[cinyc]

The Bay Area's CSA is the country's fifth-largest.  It is called San Jose-San Francisco-Oakland because San Jose is the largest city in the metro.  San Jose and San Francisco-Oakland are also in different metros.

The 1900-2016 CSA map gif shows a consistently growing San Francisco CSA:



The 1900-2016 metro map gif shows that every component metro grew every decade:



I had to use the slide effect instead of the flip effect on this gif because the latter was crashing my gifmaking program.

The decadal county map gif shows that the city/county of San Francisco lost population from the 50s through the 70s, probably due to suburbanization:



The yearly 1980-2016 yearly county estimate population change gif shows the effects of the dot com bust in the early 2000s.  Many counties lost population then, but have since rebounded, making the decade's growth in them net-positive:



The 2000-2016 yearly town estimate population change map gif shows how the dot com bust played out at the micro level:



The static 1980-2016 county percentage population change map shows good growth, particularly in the suburban/exurban counties:



And the 2000-16 static county map also shows CSA-wide growth:


Unfortunately, I noticed a rounding issue with my 2000-16 static town maps.  I have to re-update my databases in order to fix it.  I also have to fix at least some of the earlier maps (there's more near 0s than there should be, and county remainders, at least, are getting rounded up to the nearest tenth, instead of not being rounded). 

Here is a static map of town 2010-16 growth.  Everything is growing:

[Indy Texas]

Why is Ulster County PA part of the New York CSA?

Are there people in Ulster County who actually commute to NYC or spend a decent amount of time there?

[cinyc]

Why is Ulster County PA part of the New York CSA?

Are there people in Ulster County who actually commute to NYC or spend a decent amount of time there?

Jimrtex can probably give you the technical answer, but as I understand it, CSA borders are based on commuting patterns.  They change every 5 years, and were last changed for 2012.  We're due for a 2017 update soon, probably in January 2018.

Enough Ulster County residents probably commute to Dutchess County (in the NY MSA) or White Plains, Westchester County to justify CSA status, but not Metro status.  And there might even be a few Ulster County residents who commute all the way to NYC on Metro North from Poughkeepsie, which is right across the Mid-Hudson Bridge from southern Ulster.  Ulster is in the NYC TV market, so including it in the CSA isn't terribly strange to me, anyway.

The stranger county to me is Carbon, PA, which is in the Allentown-Bethlehem Metro being in the CSA.  I doubt many people commute to NYC from there.  But enough probably commute to Lehigh County, PA to put it in the Allentown Metro, and enough Lehigh/Northampton County, PA residents probably commute to NYC or its NJ suburbs to put the Allentown MSA in the CSA.

[cinyc]

The story of the Boston-Worcester-Providence CSA is one of fairly steady, unspectacular growth of about 400,000 people most years:



Most of the component metros have always grown, except on Cape Cod and the far reaches of New Hampshire:



The decadal County map gif shows Suffolk County (mainly Boston) lost population in the 1930s and 1950s-70s. Some of the losses were six-figures. Cape Cod rapidly grew, but is now stagnant:



The 1980-2016 yearly estimate map gif shows a pattern similar to San Francisco in the early 2000s, with much of the Massachusetts part of the population shrinking or stagnant.  Did the dot-com bust affect the Boston CSA, too, or is this due to people moving to the NH exurbs?



In the 2000-16 Town map gif, you can see that Cape Cod's population stagnation isn't exactly uniform or consistent town-by-town, and I think Bedford, MA losing population in the 2000s due to cuts at Hanscom AFB:



As for the static maps, the 1980-2016 county percentage population change map shows growth throughout - but higher percentage growth at the fringes of the CSA:



And the 2000-2016 county growth map shows Boston's county among the fastest-growing counties of the metro over that period, with losses on Cape Cod and urban Rhode Island:



I'm still trying to fix the 2000-16 Town growth maps.  The 2010-16 town version isn't terribly remarkable:

[jimrtex]

The New York-Newark CSA has grown every decade since 1900 except the 1970s:

Yearly county population estimates are available from 1980 onward.  I've mapped the yearly population change below.  I think there is a lot of noise in the 1999-2000 estimate change because the 2000 estimates were revamped after the census:

The base for the 1990-2000 estimates was the 1990 Census. Census estimates are based on a demographic model, where the census population is updated by aging, births, deaths, and migration in and out. Birth and death information is fairly accurate - though there may be challenges to associate it with a particular geography. But migration is more difficult. The Census uses aggregate data from Social Security and the IRS to get information on the number of persons who have moved into or out of an area. One reason why state estimates are typically more accurate is that they use the number of housing units, which they have good information on because they levy property taxes.

One reason that the Census estimates are retrospectively updated is that some of their data is delayed. They probably are just now getting IRS data based on filings for 2016.

When one census is happening, they continue the July 1 series based on the previous census. The data from the April 1, 2000 Census was still being processed when the estimates for July 1, 2000 was being released. Additional estimates based on the 1990 Census may have been made (or maybe not since the 2000 Census data would have been available (eg April 2000 Census is likely to be a better "estimate" for July 2001 than a continuation of the series based on the 1990 Census).

Eventually, a new estimate series based on the 2000 Census would be developed, including estimates for July 2000, July 2001, ... included, but no attempt was made to blend the two estimate series.

The 1990 Census is believed to have a fairly large undercount, particularly in large urban areas, particularly with large minority populations (NYC qualifies). There was resistance to using adjusted "counts" based on estimates of the census "undercount", but one outcome was to attempt to get more people to respond to the 2000 Census. So estimates based on the 2000 Census may implicitly provide a correction to the 1990s Census.

You can see the same phenomena in 2010. Look at the Google population data for Douglas County, Colorado (as well as other counties, particularly in the ski areas).

It shows an average 12K per year increase from 2000-2009, followed by a 2K decline in 2010, and then a renewal of growth from 2010-2015 at about 7.2K per year. There was a definite slowdown following the Recession/Housing Bubble, but the 106K increased from 2000-2010 was not from a 12K increase for 9 years, followed by a 2K drop, but rather by a definite slowing that started earlier.

[jimrtex]

Maps of the Los Angeles CSA are a bit more boring than New York, because there are only 5 counties in the CSA.

The decadal county growth maps show the same consistent growth pattern at the county level:

Incredible fact: Los Angeles County had a greater numeric increase from 1900 to 1910, than it did from 2000 to 2010, even though the base population in 1900 was 170K, and the base population in 2000 was 9.5M.

[cinyc]

The Dallas-Fort Worth Metro grew by about 100,000 residents pre-WWII.  Growth quickly ramped up after.  Now, explosive growth of 1,000,000+ is the norm:



The CSA is made up of 7 micropolitan and 2 metropolitan areas.  The vast majority of people live in the central Dallas-Fort Worth-Arlington Metro.  It's always grown.  The outlying metros and micros haven't been so lucky:



The decadal county map shows huge losses in outlying rural counties during the 1920s.  It also shows suburban Denton and Collin Counties growing fastest on a percentage basis these days:



The 1980-2016 county population estimates map shows growth practically everywhere every year.  Dallas County lost population during 2 years in the mid 2000s:



According to the town map, those Dallas County losses were largely due to population loss in Dallas and some southeastern Dallas County suburbs:



The 1980-2016 static county map show explosive growth, particularly in the collar counties:



The 2000-2016 county map confirms that this growth is continuing:



And the 2010-2016 town growth map shows where the growth is at the town level. Texas has some very weirdly shaped towns due to their annexation laws:

[cinyc]

The Philadelphia-Reading-Camden CSA has grown every decade since 1900 - though it only barely grew in the 1970s:



Its component metros weren't as consistent.  The core Philadelphia-Camden-Wimington MSA lost population in the 1970s.  The Ocean City MSA, a.k.a. Cape May, like Cape Cod has been stagnant recently:



On the county level, Philadelphia lost population in the 1930s and 1950-1990s.  Philadelphia's suburban counties grew in the post-WWII timeframe:



The 1980-2016 yearly county population estimate map gif shows Philadelphia turning the corner around 2006.  Now, the CSA's New Jersey counties are stagnant or losing population:



The county trends are confirmed by the 2000-16 town percentage population change maps.  Philadelphia's New Jersey suburbs are losing population, perhaps along with some Bucks County towns, while Philadelphia's Pennsylvania suburbs still grow a bit:



The 1980-2016 static county population change map shows that all but Philadelphia and Salem County growing in that time frame:



The 2000-16 static county map shows that Philadelphia has gained since then, while Salem and Cape May Counties have lagged behind:



The 2010-16 static town map confirms the New Jersey stagnation:

[cinyc]

The Miami-Fort Lauderdale-St. Lucie CSA has always grown since 1900.  The extent of the growth in the 1900s-20s is not easily calculable, though, due to county boundary changes. At the turn of the last century, Dade County covered most of Southeast Florida, including some or all of current Miami-Dade, Broward, Palm Beach and Martin Counties.  Further, the northern part of the CSA was in Brevard County, which is not in the CSA. Nevertheless, it is likely that fewer than 10,000 people lived in the CSA as of 1900.  But the calculated early numbers should be taken with a grain of salt:



Every component metro of the Miami CSA has always grown every decade since 1900.  The one component micropolitan area, Okeechobee, shrunk in the 1930s.  This was probably not due to a county boundary change, but I can't be sure:  



On the county level, practically every CSA county has been gaining population every decade:



The 1980-2016 county yearly percentage population change map gif shows that Miami-Dade County lost population in 1992 (likely due to Hurricane Andrew) and Broward County lost population from 2005-07 (likely due to Hurricane Wilma).  Martin and Okeechobee Counties have also periodically lost population. Otherwise, the map is a sea of deep red:



The 2000-16 town change map shows that the 2005-07 Broward County losses were pretty much countywide, except for a few inland communities.  Otherwise, the map shows that most other areas have consistently grown, including the city of Miami:



The static maps aren't terribly interesting.  The 1980-2016 county percentage population change map shows heavy growth throughout the CSA:



As does the 2000-16 county map:



The 2010-16 town map is a sea of red.  Every component town has gained population this decade:

[cinyc]

The Houston-The Woodlands CSA was 10th largest in 2014, according to Wikipedia, but has leapfrogged at least Miami since.  The CSA has grown every decade since 1900, largely at an accelerating rate.  Like Dallas, it added more than 1,000,000 people in the 00s:



The CSA is made up of the main Houston-The Woodlands-Sugar Land Metro and four micro areas.  The core Houston Metro has always grown since 1900.  The micros have not been so lucky:



The 1900-2016 decadal county map shows that the Houston CSA's core county, Harris, has grown every decade since 1900. Recently, suburban/exurban Fort Bend and Montgomery Counties are growing faster than Harris on a percentage, but not numerical, basis:



The 1980-2016 yearly percentage population change map shows that Harris County actually lost some population after the oil bust in the mid 1980s.  Otherwise, the picture is generally one of growth, except at the fringes:



Finally, the 2000-16 town map shows a slight decline in Houston in 2003.  The city of Galveston also seems to have lost population throughout the 00s.  Otherwise, it largely mirrors the county map, which shouldn't be surprising in a CSA that has relatively few incorporated places:



Like in most high-growth areas, the static maps aren't that interesting.  The 1980-2016 county percentage population change map shows growth throughout the CSA, except in Matagorda County:



The 2000-16 County map is virtually identical:



And the 2010-16 Town Map pretty much shows growth throughout the CSA, except in Trinity County:

[jimrtex]

Question for people who know more about SoCal/Orange County: How has Irvine grown so consistently faster than all its surroundings (including both more and less developed neighbors) for such an extended period of time?

Irvine is fairly new. The Irvine Ranch was huge (roughly 110,000 acres) and went from the mountains to the coast. Development was more planned, and could take into account changing(-ed) demographics. In the 1950s no one would conceive that any 30 YO would not be married without at least 3 children, so houses were quickly built that families could afford (these people who grew up during the Depression, and if they were from a rural area, might not have had running water, indoor plumbing, or electricity beyond lights). Couples who are childless, or have one child, or are single, might not be attracted to a 3- or 4-bedroom house near to schools. They might prefer a townhouse, condo, or apartment.

An area that was quickly filled up with houses filled with two parents and four children, might decline in population 20 to 30 years later as the children became adults and left home. The parents in their 40s and 50s were not ready to move to a retirement home, so there were not opportunities for new families.

Being developed later, Irvine provided a variety of housing, as well as commercial and business areas. They also were chosen for a new campus of the University of California (-Irvine), which now has about 30,000 students, many of which, probably most are counted as residents by the Census.

Irvine also recently annexed the former El Toro Marine Air Station. While some of the area is devoted to Orange County's Great Park, most of it can be developed. This may reason for the recent sustained increase.

[cinyc]

The Atlanta--Athens-Clarke County--Sandy Springs CSA generally grew by about 100,000 people per decade before WWII.  It slowly ramped up after WWII to gain more than 1,000,000 people per decade in the 1990s and beyond:



Note that the 1900-40s numbers might not be accurate due to county boundary changes. For example, Campbell and Milton Counties were absorbed by Fulton County in the 1930s.  That's why Fulton County has its strange, long shape.

At the metro level, the core Atlanta-Sandy Springs-Roswell Metro has always grown. The outlying metropolitan and micropolitan areas have been mixed.  The Athens-Clarke County Metro lost population in the 1920s:



The decadal county map shows that Fulton County lost population the 1970s. Atlanta's suburban and exurban counties started to grow after WWII, and really ramped up in the 1960s:



The 1980-2016 yearly county percentage population change map shows that core Fulton and DeKalb Counties lost population in the early 2000s. It has since recovered. Growth in some exurban counties cooled down after the great recession, but is picking up again, especially on the northern fringes of the metro:



The 2000-16 town percentage population change map shows that Fulton's early 2000s population loss was due at least in part due to the city of Atlanta losing population:



The static county maps confirm Atlanta's suburban/exurban growth.  For example, practically every suburban/exurban county on the 1980-2016 map is dark, dark red, particularly north of Fulton:



Even most counties in the 2000-16 county growth map are deep red:



The 2010-16 town percentage population change map shows which towns and county remainders have grown fastest since 2010:

[jimrtex]

Why is Ulster County PA part of the New York CSA?

Are there people in Ulster County who actually commute to NYC or spend a decent amount of time there?

Jimrtex can probably give you the technical answer, but as I understand it, CSA borders are based on commuting patterns.  They change every 5 years, and were last changed for 2012.  We're due for a 2017 update soon, probably in January 2018.

Enough Ulster County residents probably commute to Dutchess County (in the NY MSA) or White Plains, Westchester County to justify CSA status, but not Metro status.  And there might even be a few Ulster County residents who commute all the way to NYC on Metro North from Poughkeepsie, which is right across the Mid-Hudson Bridge from southern Ulster.  Ulster is in the NYC TV market, so including it in the CSA isn't terribly strange to me, anyway.

The stranger county to me is Carbon, PA, which is in the Allentown-Bethlehem Metro being in the CSA.  I doubt many people commute to NYC from there.  But enough probably commute to Lehigh County, PA to put it in the Allentown Metro, and enough Lehigh/Northampton County, PA residents probably commute to NYC or its NJ suburbs to put the Allentown MSA in the CSA.

Metropolitan Statistical Areas and Micropolitan Statistical Areas are collectively known as Core Based Statistical Areas (CBSA) - the distinction between Metropolitan and Micropolitan is population, their delineation is the same.

A CBSA is comprised of Central Counties and Outlying Counties. Counties qualify to be a central county by having half the population in urban areas, or containing 5000 persons of an urban area of at least 10,000. Urban areas are blobs of people (continuous areas of somewhat dense population). Urban Areas are either Urbanized Areas or Urban Clusters, again distinguished by their size, with Urbanized Areas having more than 50,000 people. Urban Clusters have to have 2500 persons to exist, but need 10,000 to form the basis for a Micropolitan Statistical Area.

Urbanized Areas, but not Urban Clusters, are grandfathered from one census to the next. If an Urbanized Area and Urban Cluster grow together, the Urban Cluster is absorbed. But when Urbanized Areas grow together, their identity is kept separate, typically at or near county lines.

The NYC Urbanized Area does quite reach the Delaware River. But it comprises 25% of Sussex, 40% of Hunterdon, 19% of Warren, and 8% of Mercer. Since it is the largest urban area in Sussex and Hunterdon counties, those two counties are Central Counties of the NYC Metropolitan Statistical Area.

More of Warren (29%) is in the Allentown-Bethlehem-Easton Urbanized Area, so it is a Central County of the Allentown Urbanized Area. Easton is traditionally associated with Allentown so there is a division in the county. The NYC Urbanized Area could conceivably grow such that it was the dominant Urban Area, and Warren would flip between MSA's.

Because of Trenton, most of Mercer is in the Trenton Urbanized Area, and blocks the NYC Metropolitan Area.

Pike County, PA is potentially the Port Jervis Micropolitan Statistical Area, since the Port Jervis Urban Cluster has 10,000 persons, including 5000 in Pike County.

To the north, the Bridgeport-Stamford Urbanized Area keeps the NYC UA out of Fairfield County, and the Poughkeepsie-Newburgh Urbanized Area keeps the NYC UA out of Dutchess and pretty much out of Orange. But the NYC Urbanized Area does extend into Putnam County, so that Westchester, Putnam, and Rockland are central counties to the north.

The Poughkeepsie-Newburgh UA does extend a bit into Ulster to reach New Paltz, but the dominant Urbanized Area in Ulster is Kingston UA.

So in New York, you have the Poughkeepsie-Newburgh proto-MSA with Dutchess and Orange serving as central counties.

Outlying counties are determined on the basis of commuter flows. If 25% of workers who live in a county work in the central counties of CBSA, the county is an outlying county. But a central county of one CBSA may not be an outlying county of another CBSA.

But once the initial CBSA are delineated, then one CBSA may be treated as outlying to another. This appears what pulls Orange+Dutchess as a unit into the NYC Metro area. Commuting into NYC, Westchester, Rockland, and Putnam, plus most of Northern New Jersey counts as commuting into the central counties.

The number one destination for Dutchess is Westchester (15.2K), Putnam is next at 5.5K, followed by New York (Manhattan) 5.3, Orange 5.1, and Ulster 4.3. For Orange it is New York 9.9K, Rockland 9.2, Bergen, NJ 7.1K, Westchester 6.9K, Dutchess 5.1K, and Bronx 4.4K. It appears that Orange is dragging Dutchess into the NYC MSA (but I can only account for 24.5% of Orange+Dutchess workers working in the NYC MSA).

After the CBSA's are delineated, they may be agglomerated into a CSA, with each CBSA being treated as a unit. While for a CBSA, commuting must be into a Central County, for a CSA it just has to be between CBSA. In addition the link can be weaker. That is what brings the Ulster (Kingston MSA) into the CSA. Ulster has 9K commuting into both Ulster and Orange counties, compared to about 5K for the entire rest of the NYC MSA.

Were Ulster not itself a Central County it would have been part of the proto-Poughkeepsie-Newburgh-Middletown (PNM) CBSA. I'm not sure why NYC and PNM are merged, rather than PNM and Kingston.

It may be possible for the merging of CBSA into CSA to be chained (e.g. Bridgeport-Stamford being pulled in, then brought New Haven, and Waterbury-Danbury in.

Carbon being part of the Allentown-Bethlehem-Easton MSA is what bring it into the NYC CSA. Warren being part of the Allentown MSA and also including about 20% of its population in the NYC UA, likely pulls in Allentown. The commuting only has to anywhere into Northern New Jersey or southern New York, not into Manhattan or even Newark. It could even be into Sussex County.

A CSA might be considered to be more like a group of related CBSA, which is hard to see when comparing NYC and any CBSA, let alone Kingston. But the relationship is somewhat easier to understand with NYC and the Connecticut MSAs, or Allentown and Trenton and NYC.

Even easier to comprehend is Washington and Baltimore being in a CSA.

[cinyc]

Why is Ulster County PA part of the New York CSA?

Are there people in Ulster County who actually commute to NYC or spend a decent amount of time there?

Jimrtex can probably give you the technical answer, but as I understand it, CSA borders are based on commuting patterns.  They change every 5 years, and were last changed for 2012.  We're due for a 2017 update soon, probably in January 2018.

Enough Ulster County residents probably commute to Dutchess County (in the NY MSA) or White Plains, Westchester County to justify CSA status, but not Metro status.  And there might even be a few Ulster County residents who commute all the way to NYC on Metro North from Poughkeepsie, which is right across the Mid-Hudson Bridge from southern Ulster.  Ulster is in the NYC TV market, so including it in the CSA isn't terribly strange to me, anyway.

The stranger county to me is Carbon, PA, which is in the Allentown-Bethlehem Metro being in the CSA.  I doubt many people commute to NYC from there.  But enough probably commute to Lehigh County, PA to put it in the Allentown Metro, and enough Lehigh/Northampton County, PA residents probably commute to NYC or its NJ suburbs to put the Allentown MSA in the CSA.

Metropolitan Statistical Areas and Micropolitan Statistical Areas are collectively known as Core Based Statistical Areas (CBSA) - the distinction between Metropolitan and Micropolitan is population, their delineation is the same.

A CBSA is comprised of Central Counties and Outlying Counties. Counties qualify to be a central county by having half the population in urban areas, or containing 5000 persons of an urban area of at least 10,000. Urban areas are blobs of people (continuous areas of somewhat dense population). Urban Areas are either Urbanized Areas or Urban Clusters, again distinguished by their size, with Urbanized Areas having more than 50,000 people. Urban Clusters have to have 2500 persons to exist, but need 10,000 to form the basis for a Micropolitan Statistical Area.

Urbanized Areas, but not Urban Clusters, are grandfathered from one census to the next. If an Urbanized Area and Urban Cluster grow together, the Urban Cluster is absorbed. But when Urbanized Areas grow together, their identity is kept separate, typically at or near county lines.

The NYC Urbanized Area does quite reach the Delaware River. But it comprises 25% of Sussex, 40% of Hunterdon, 19% of Warren, and 8% of Mercer. Since it is the largest urban area in Sussex and Hunterdon counties, those two counties are Central Counties of the NYC Metropolitan Statistical Area.

More of Warren (29%) is in the Allentown-Bethlehem-Easton Urbanized Area, so it is a Central County of the Allentown Urbanized Area. Easton is traditionally associated with Allentown so there is a division in the county. The NYC Urbanized Area could conceivably grow such that it was the dominant Urban Area, and Warren would flip between MSA's.

Because of Trenton, most of Mercer is in the Trenton Urbanized Area, and blocks the NYC Metropolitan Area.

Pike County, PA is potentially the Port Jervis Micropolitan Statistical Area, since the Port Jervis Urban Cluster has 10,000 persons, including 5000 in Pike County.

To the north, the Bridgeport-Stamford Urbanized Area keeps the NYC UA out of Fairfield County, and the Poughkeepsie-Newburgh Urbanized Area keeps the NYC UA out of Dutchess and pretty much out of Orange. But the NYC Urbanized Area does extend into Putnam County, so that Westchester, Putnam, and Rockland are central counties to the north.

The Poughkeepsie-Newburgh UA does extend a bit into Ulster to reach New Paltz, but the dominant Urbanized Area in Ulster is Kingston UA.

So in New York, you have the Poughkeepsie-Newburgh proto-MSA with Dutchess and Orange serving as central counties.

Outlying counties are determined on the basis of commuter flows. If 25% of workers who live in a county work in the central counties of CBSA, the county is an outlying county. But a central county of one CBSA may not be an outlying county of another CBSA.

But once the initial CBSA are delineated, then one CBSA may be treated as outlying to another. This appears what pulls Orange+Dutchess as a unit into the NYC Metro area. Commuting into NYC, Westchester, Rockland, and Putnam, plus most of Northern New Jersey counts as commuting into the central counties.

The number one destination for Dutchess is Westchester (15.2K), Putnam is next at 5.5K, followed by New York (Manhattan) 5.3, Orange 5.1, and Ulster 4.3. For Orange it is New York 9.9K, Rockland 9.2, Bergen, NJ 7.1K, Westchester 6.9K, Dutchess 5.1K, and Bronx 4.4K. It appears that Orange is dragging Dutchess into the NYC MSA (but I can only account for 24.5% of Orange+Dutchess workers working in the NYC MSA).

After the CBSA's are delineated, they may be agglomerated into a CSA, with each CBSA being treated as a unit. While for a CBSA, commuting must be into a Central County, for a CSA it just has to be between CBSA. In addition the link can be weaker. That is what brings the Ulster (Kingston MSA) into the CSA. Ulster has 9K commuting into both Ulster and Orange counties, compared to about 5K for the entire rest of the NYC MSA.

Were Ulster not itself a Central County it would have been part of the proto-Poughkeepsie-Newburgh-Middletown (PNM) CBSA. I'm not sure why NYC and PNM are merged, rather than PNM and Kingston.

It may be possible for the merging of CBSA into CSA to be chained (e.g. Bridgeport-Stamford being pulled in, then brought New Haven, and Waterbury-Danbury in.

Carbon being part of the Allentown-Bethlehem-Easton MSA is what bring it into the NYC CSA. Warren being part of the Allentown MSA and also including about 20% of its population in the NYC UA, likely pulls in Allentown. The commuting only has to anywhere into Northern New Jersey or southern New York, not into Manhattan or even Newark. It could even be into Sussex County.

A CSA might be considered to be more like a group of related CBSA, which is hard to see when comparing NYC and any CBSA, let alone Kingston. But the relationship is somewhat easier to understand with NYC and the Connecticut MSAs, or Allentown and Trenton and NYC.

Even easier to comprehend is Washington and Baltimore being in a CSA.

Thanks for the explanation.  I could have sworn there used to be a separate Poughkeepsie-Newburgh-Middletown metro area in recent years. I was actually surprised to see Dutchess and Orange as part of the New York Metro Area again.

[cinyc]

With the Detroit-Warren-Ann Arbor CSA at CBSA #12, we finally get our first look at the population change patterns in a rust belt city - and it's not pretty. The Detroit CSA has seen better days. It was Atlanta or Dallas-like during the 1920s - adding over 1,000,000 people that decade at a time when the US population was much lower than it is today. The CSA has largely been in decline since the 1970s, with a brief respite during the 1990s:



The Ann Arbor Metro has helped prop up the Detroit CSA's numbers a little bit - it is still growing.  But the core Detroit-Warren-Dearborn metro and the Flint metro have generally been losing population or stagnant recently:



At the decadal county level, Detroit's Wayne County has lost population every decade since the 1970s.  Detroit-suburban Oakland and Macomb have gained population during that period, albeit sometimes anemically. Exurban Livingston County has grown fastest, percentage wise:



The yearly 1980-2016 county percentage population change map shows Wayne County losing population practically every year, except during the early-to-mid 1990s. Even Oakland County has lost population some years. Even exurban Livingston County lost a little bit of population in the late 2000s:



The 2000-16 town percentage population change map shows just how badly Detroit and most of its inner suburbs have lost population recently, while Detroit's outer suburbs generally gained population:



The 1980-2016 county static map shows Wayne 20-30% of its population, suburban Oakland and Macomb up, and Livingston growing at an 80% clip over that period:



The 2000-2016 county static map shows basically the same growth pattern, with Macomb growing a little more than Oakland:



And the 2010-2016 static town map shows most of Wayne County losing population, except a few towns at the outer fringes of the county:

[cinyc]

The Seattle-Tacoma CSA has grown each decade since the 1900. It had a big growth spurt in the 1900s and ramped after WWII:



Its component metros and micros have usually grown, too. The core Seattle-Tacoma-Bellvue Metropolitan Area has grown every decade since 1900:



King, Snohomish and Pierce Counties have grown each decade. Only the islands and far fringes have had loss decades:



1980-2016 yearly county map shows Snohomish County growing at a higher percentage rate than King - until recently. Tacoma's Pierce County likely had a slight down year in 2009:



The 2000-16 town map shows Seattle losing population in 2001-03. Now, it grows as fast as unincorporated Snohomish. Tacoma lost population in 2002-04 & 2009:



The 1980-2016 static county percentage population change map shows growth throughout, especially in suburban Snohomish County and Olympia's Thurston County:



The 2000-16 static county map shows the same growth pattern, albeit at a slower rate:



And the 2010-16 static town map shows that Seattle is one of the fastest-growing towns in King County:

[cinyc]

Phoenix-Mesa-Scottsdale was the 14th largest Primary Statistical Area in 2014, according to Wikipedia. It is the first PSA that is only a Metropolitan Area on the list. The Phoenix metro has grown every decade since 1900, adding over 1,000,000 people in the 1990s and almost as many in the 2000s:



The Phoenix metro only has two counties, Maricopa and Pinal. Both have grown every decade since 1900:



The 1980-2016 yearly percentage population change county map gif is equally boring. The only county that ever lost population was Pinal in 2010-11:



The 2000-16 town percentage population change map gif shows the city of Phoenix losing population in 2008-10, and that the Pinal County loss was pretty much county-wide:



Unsurprisingly, the 1980-2016 county percentage population change was off the chart:



The 2000-16 static county map shows that Pinal County grew faster than Maricopa County from 2000-16 in percentage terms, but Maricopa County is much larger:



The 2010-16 static town map shows a growing picture, with Buckeye and Pinal County leading the area in percentage population growth: