Thursday, 27 August 2015

USA MSAs urban vs suburban growth (2000-2010) - Part 6

 In earlier posts, I tried to measure urban core growth by defining urban cores using distance from city hall and 1960 densities.

The disadvantage of looking at areas that had a certain density in 1960 (5,000 pop/sq mi in this case) is that they often correspond to a very small part of the newer sunbelt metro areas. Although it's true that sunbelt metro areas are typically less urban, they still have post-war suburbs that are centrally located relative to job centers and have higher proportions of apartments and higher transit use than more far flung suburbs. In addition, there are efforts to urbanize some suburban areas near office parks and transit in metro areas like Atlanta.

Also, while many of these sunbelt cities are in the top 51 MSAs today (those with over 1,000,000 people), they were a long way from being top 51 in 1960. Some were just small towns with small town densities, so in cities like Raleigh and Charlotte, you could have neighbourhoods that were built pre-1960 but at densities below 5,000 pop/sq mi.

In this post, I'll modify the criteria to expand the urban cores of sunbelt cities so that they contain a proportion of the metro that is more comparable to those of the cores of older cities.

Mainly, I'll allow census tracts to qualify as urban if the non-auto mode share and density is above the metro area average. I've also included census tracts that have below average densities but have significantly higher than average non-auto commute mode share and vice versa (details at end of post). Where this is still insufficient to expand the urban core to 20% of the MSA population, I'll also include census tracts with 1960 densities of over 3,000 pop/sq mi.

Here are the results.

The cities are ranked according to the amount of urban core growth relative to their 2000 MSA population.

 The top cities are mostly coastal cities known for dense urban infill like Portland, D.C. and New York but there are a couple exceptions, like the #1 Raleigh and #2 Riverside-San Bernardino. Although these two cities have not had exceptional downtown growth, they have had growth throughout most of the metro area - downtown, inner suburbs and outer suburbs. Population growth in the downtown adjacent neighbourhoods and inner suburbs of these cities is a big part of the reason why they ranked so high.

Compared to the areas defined based purely on 1960 densities, many cities have seen the size of their urban cores expanded significantly, so here's some maps of the cities where the boundaries changed the most (in terms of absolute increase in population included). Again, the format is

MSA name
Urban core population change (rank)
Urban core % population change (rank)
Urban core population change relative to 2000 population of entire MSA (rank)

Dallas-Forth Worth

South Florida



Inland Empire


Tampa Bay



Las Vegas

More details on the methodology (bold are additions to the methodology from Part 3):

The primary city's downtown, defined as the business district with the highest walk score, must always be included in the urban core. The primary city is the one that's named first in the official metropolitan area name.

Any urban census tracts that are contiguous with the primary city's downtown census tracts also are considered part of the primary core.

Urban census tracts are defined as tracts that meet one of the following criteria:
  • a population density of at least 5,000 pop/sq mi in 1960
  • a population density of at least 6,000 pop/sq mi in 1970 if 1960 census tract data does not exist
  • a non-auto commute mode share in 2000 (m0) above that of the MSA's average (m) AND a 2000 density (d0) above that of the MSA's population weighted average density (d) 
    • if there is a secondary urban core (based on 1960 densities > 5000 pop/sq mi ONLY), AND that secondary urban core is in a separate county (or county equivalent) from the primary core AND has a larger population than any part of the primary core extending into the secondary core's county AND the secondary core has an non-auto commute mode share lower than the MSA average, the secondary core county's non-auto commute mode share is used instead of the MSA's for census tracts in that county
    • if the primary county has a  non-auto commute mode share that is lower than the MSA average, the primary county's non-auto commute mode share is used instead of the MSA's for census tracts in that county
  • m0 >= 2*m AND m0/m >= d/d0 OR d0 >= 2*d AND d0/d >= m/m0
  • a population density of at least 3,000 pop/sq mi in 1960 if the above criteria do not bring the urban core population to over 20% of the MSA
  • any census tracts completely surrounded by urban census tracts are also automatically considered urban.

Secondary urban cores are contiguous urban census tracts with a combined population of at least 25,000 in 2010. In addition, if the downtown of one of the MSA's secondary cities, ie a city that is in the MSA's name, is adjacent to a secondary core, it is also included in the secondary core.

Any census tracts with a 2010 density of over 15,000 pop/sq mi contiguous with primary or secondary urban cores are also included.

1 comment: