great american “metro areas”

When any US study or journalist refers to “metro areas,” they probably mean this:







These are all photos of US Metropolitan Statistical Areas (MSAs).  Many, many national studies — most recently the Brookings study on “transit and jobs in metropolitan America” — mean “MSA” when they say “metro area.”

MSAs, however, are aggregations of counties. They’re the red patches on this map:

Counties come in all kinds of weird sizes, and are usually irrelevant to anyone’s lived experience of a metro area.  Eastern US counties are mostly small, so MSAs there are often credible.  But western counties are often huge, so MSAs have to be huge too.  Almost two-thirds of California’s land area is a metro area by this defintion, including the “Riverside-San Bernardino-Ontario MSA,” which contains most of the Mojave Desert.  Metro areas in America include the Grand Canyon,  a big chunk of the Everglades, and the vast Voyageurs wilderness of northern Minnesota, accessible only by canoe or snowmobile.

So when the Brookings Institution, for example, declares that Riverside-San Bernardino is doing poorly on transit travel times to work, they’re referring partly to travel times from Needles to Riverside, a distance of about 230 miles (370 km) over open desert.  They’re also implying that there ought to be intense transit service between the Riverside area and the Palm Springs area, even though locals largely experience these as two different metro areas.  (Their centroids are 50 miles apart, the towns between are mostly semi-rural in nature, and if those facts don’t convince you, there’s also a 10,000-foot mountain in the way.)  What matters to the MSA is that the two metro areas are in the same counties as Riverside-San Bernardino, so nothing else about their lived geography can possibly matter.

A deeper problem arises when all the demographic statistics of an MSA are declared to be features of a “metro area.”  Consider the Visalia-Porterville MSA, site of the top photo above.  The MSA, identical to Tulare County, has a 2000 population of 368,000.  All of these people are counted in MSA-based statistics about “metropolitan America,” but only about half of them live in a city over 50,000.  The other half live in much smaller towns and in rural areas.  (The rural areas also have high labor needs, so they support semi-mobile populations, validly picked up by the census, that have no relationship to any city.)  A fundamentally rural and small-town culture, indistinguishable from many other entirely rural counties, is being described as metropolitan whenever the Visalia-Porterville MSA is referenced as part of generalizations about “metropolitan America.”  This culture is not just small and easily dismissed statistical “noise.”  It’s half of the population of the MSA.

This is one of those absurdities that we’re trained to think of as eternal.  Many weird and misleading boundaries (e.g. some counties, city limits etc) are going to persist even if they have no emotional or cultural meaning, simply because influential people are attached to them as a matter of self-interest.  But how many people are really attached in this way to MSAs?  And is it really impossible, with all the increasingly detailed information in the census, to describe metro areas in a more subtle and accurate way?

Even if we’re stuck with them, is it really appropriate to keep saying “metro area” when you mean MSA?  It’s statistically convenient given how much data is organized by these crazy units.  But are you really misleading people about what a metro area is?

In the sense that usually matters for urban policy, “metro area” means “the contiguous patch of lights that you can see at night from an airplane or satellite.”  You can approximate this with census blocks.  Their technical definition is something like “any agglomeration of contiguous census blocks that all have a non-rural population or employment density.”  Census blocks are small enough that they can aggregate in a way that follows the geography, connecting what’s really connected and separating what’s really separate.  Defining “metro area” in that way would finally mean what ordinary people mean by “metro area.”

What’s more, it would really cut down on bear attacks in “metropolitan America.”


  1. Kaweah Gap, Sequoia National Park, Visalia-Porterville MSA, California.  Credit: Davigoli, Wikipedia.
  2. Lake Mead National Recreation Area, Las Vegas-Paradise MSA, Nevada.  (my photo)
  3. Duck Lake and High Sierra, John Muir Wilderness, Fresno MSA, California.  (my photo)
  4. Matanuska Glacier, Anchorage MSA, Alaska.  Credit: Elaina G, via Google Earth.
  5. Melakwa Lake, Alpine Lakes Wilderness, Seattle-Tacoma-Bellevue MSA, Washington.  Credit: Wikipedia.
  6. Joshua trees in open desert southwest of Las Vegas, Las Vegas-Paradise MSA, Nevada.  (my photo)




17 Responses to great american “metro areas”

  1. Cascadian May 20, 2011 at 11:48 am #

    The thing is, it’s a good enough approximation and it’s comparatively hard (for a layperson anyway) to collect statistics at the census block level instead of county level.
    Just taking King County (the core of the Seattle MSA that includes Melakwa Lake–nice hiking/camping spot, by the way): There are almost 2 million people in the county and most of them live in the metro (city/suburb) part. Those who live in rural or near-wilderness areas are such a small portion of the total that you don’t lose much analyzing things at the MSA level.
    I don’t really see the point of changing county boundaries–most of the metro government functions are served by municipal or metro regional government. For example, Sound Transit includes only areas within the urban growth area, and King County Metro, while nominally countywide, has very little service in the rural areas outside of the ST district. I also think that it’s helpful to include surrounding rural and wilderness areas because ecologically and socially they function as a unit and all of the parts are necessary for the whole to function.

  2. Eric Fischer May 20, 2011 at 11:51 am #

    Excellent article!
    And on the other end of the spectrum, you also have cases like San Jose being treated as a separate metropolitan area from San Francisco and Oakland, even though these areas have been connected by continuous urbanization for decades.

  3. G-Man (Type E) May 20, 2011 at 12:01 pm #

    Why doesn’t more urban research use “Urbanized Area” which is defined by density and adjacency of census blocks (at least AFAIK)? I have had some difficulty getting data at that level, but it seems like it should be easy to aggregate from the census if there was a market of people interested in using it. I suppose MSA has better comparative value in temporal analysis because county boundaries don’t often change, while UA is redefined with each census…

  4. Winston May 20, 2011 at 12:03 pm #

    The Census Urbanized Areas are more useful, and I wish they were more used for this kind of thing. While they occasionally draw arbitrary boundaries (including the one Eric Fischer mentioned above) they are a meaningful way to compare cities without being confused by artificial boundaries (for example, just as bad as talking about the LA CMSA is talking about the City of Los Angeles, which has some of the oddest borders of any city in the country.

  5. Justin N May 20, 2011 at 12:33 pm #

    I think it’s key to understand the unit of analysis you’re talking about when doing an analysis based on MSA’s. If you want to say something about *people* in metro areas, the MSA is probably a fine measure- the minor statistical noise introduced by the handful of people who live outside of the “metro” bits of the MSA is probably negligible, and the improvement in your results in using some finer-grained measure probably isn’t worth the time.
    However, if you want to talk about *geography* in metro areas- as we often do when we talk about transit- they have some significant disadvantages as detailed above.

  6. Al Dimond May 20, 2011 at 2:07 pm #

    Riverside-San Bernardino, as a “Metro Area”, shows up on a lot of top-10 or bottom-10 lists for various things. I never realized its geography was so extreme… that’s surely a reason for its extreme results.

  7. Morgan Wick May 20, 2011 at 5:57 pm #

    One disadvantage of urbanized areas, as far as I know, is that they don’t have the equivalent of PMSA’s for polycentric regions, which might not be useful for talking about the general agglomeration of people, but which is useful for talking about any level of transit more granular than commuter rail or BART.
    Use of urbanized areas and demand for data involving them would probably double if they were called “urbanized metropolitan areas”.

  8. Alon Levy May 20, 2011 at 7:24 pm #

    Brookings’ methodology, despite its many faults, does not actually compute things based on dirt. It computes things based on what percentage of people and of jobs are in their view transit-accessible.
    The Palm Springs-Riverside problem is real, but any competent study (which Brookings’ isn’t) should resolve this by weighting things based on actual commute markets.

  9. JN_Seattle May 20, 2011 at 8:14 pm #

    The above pictures of wilderness do not represent the focus of the long-standing Brookings Metropolitan Policy Program! In this study did Brookings do computations and create findings and conclusions that include averages encompassing absurd applications of urban transit, such as travel times from Needles to Riverside in California? I doubt it, but I’ll check and be back at some point later.

  10. Jarrett May 21, 2011 at 1:50 pm #

    NOTE: In response to comments up to this point, I’ve added the new paragraph beginning “A deeper problem …”

  11. Enviro_writer May 22, 2011 at 8:48 am #

    FiveThirtyEight wrote a pretty interesting critique of the Brookings study too:
    (BTW, thanks for visiting my blog, Jarrett. I didn’t know about the prickly pear legacy. And I love “radioactive cuteness of the koala”!).

  12. Enviro_writer May 22, 2011 at 8:53 am #

    Blech, sorry for repeating a link from an earlier post…

  13. Alan Kandel May 25, 2011 at 9:29 am #

    To me, this is all mumbo jumbo, especially when it comes to trying to interpret what metro area population in any meaningful way. Just look at some of the definitions Webster provides for “metropolitan,” “community,” and “city.”
    Webster defines metropolitan as: “1. characteristic of a metropolis or its inhabitants, esp. in sophistication. 2. of or pertaining to a large city and its surrounding communities: the New York metropolitan area.” (“the New York metropolitan area ” is italicized).
    Community, meanwhile, is defined by Webster as: “1. a group of people who reside in a specific locality, share government, and often have a common culture and historical heritage. 2. a locality inhabited by such a group. 3. a social, religious, occupational, or other group sharing common characteristics or interests: the business community.” (“the business community” is italicized).
    City, on the other hand, as defined by Webster is: “1. a large or important town. 2. (in the U.S.) an incorporated municipality, usu. governed by a mayor and council.”
    When traveling along the highway and upon entering a metropolitan area, signs are posted identifying the metro area about to be entered, and provide the population and elevation. The population statistic listed doesn’t take into consideration the population in the county outside the immediate metropolitan area.
    As for “Metropolitan Statistical Area,” how does this term help? It sounds like another way of saying “county”. And if this is the case, why not just say “county”?
    “Conurbation” is in my view one of those words that leaves nothing to interpretation. As defined by Webster, conurbation is: “an extensive urban area resulting from the expansion of several cities or towns so that they coalesce but usu. retain their separate identities.”
    Some of the definitions Webster provides for town, meanwhile, are: “1. a thickly populated area, usu. smaller than a city and larger than a village, having fixed boundaries and certain local powers of government. 2. a densely populated area of considerable size, as a city or borough.”
    Like I said, mumbo jumbo.

  14. Alan Kandel May 25, 2011 at 9:44 am #

    Please forgive. The word “culture” contained in the definition of “community” above, should instead be “cultural.”

  15. Alon Levy May 25, 2011 at 3:02 pm #

    As for “Metropolitan Statistical Area,” how does this term help? It sounds like another way of saying “county”. And if this is the case, why not just say “county”?
    Because in the East, counties are small and MSAs consist of many of them. The New York metro area defined in terms of small municipalities is quite close to the county-based CSA.

  16. Alan Kandel May 26, 2011 at 8:15 am #

    Food for thought: In the interest of conciseness and clarity, it may be more appropriate just to classify the extreme inner cores or downtowns of cities as “urbs,” while the outermost areas could either be classified “exurbs” or “rural areas,” depending on situation. Meanwhile, the area between the extreme inner cores or urbs and the outermost exurb or rural areas, these would be designated “suburbs.”

  17. Compare cities December 26, 2014 at 3:31 am #

    Traveling experience in the metropolitan area is amazing.