richard florida’s supersized-airport theory

Bkk airport
Richard Florida has discovered that as cities get bigger, airport traffic goes up even faster than population.  Here's his interpretation:

As with other features of metropolitan life, such as productivity or invention, airport passenger traffic increases disproportionately with population size. This is yet another manifestation of the overall acceleration of social and economic life that occurs in urban population centers—a phenomenon that has been dubbed “superlinear scaling.”

OK, but how much of this is what transport planners call the network effect?  Just as transit planners try to locate connection points at already-active locations, airlines are motivated to locate their hubs in major cities, so that the city and the hub support each other in creating the largest possible market for flights in and out.

For example, ask: what's the biggest US metro area that is NOT an airline hub?  Using this table  (and setting aside greater Riverside because its airport is run by Los Angeles) I get all the way down to San Diego, the 17th largest, before I encounter one where hub effects have little role in the airport's function.  So it doesn't surprise me that the scaling of big city airports is "superlinear" (a fancy word for "curving upward") compared to population, for the same reason that boardings at major transfer points in a transit network are superlinear compared to their immediate surroundings.

I could interpret this as a nice example of the very common tendency of people with social science training to overread demographics and economic indicators, while underreading the effects of network structure and how it determines the actual utility of transport service.  This issue comes up in transit all the time. 

But another interpretation, more to Mr. Florida's liking, would be that the connection in a transport network is not really all that different from the social connections that occur in urban life.  Florida is famous for documenting how all these connections rise in non-linear ways as there are more people around to connect, thus leading to hotspots of creativity and innovation. 

When I suggest that transit agencies use the word connection, instead of the dreary word transfer, it is to invoke exactly this association.  A transport connection isn't just an analogy with the social, cultural, and intellectual connections that drive innovation.  In its multiplying power it's almost the same thing.

25 Responses to richard florida’s supersized-airport theory

  1. anonymouse May 27, 2012 at 3:32 pm #

    And San Diego has not one but two or three really good reasons why it can’t be a good airline hub: one is that it’s in a corner of the country, and not really a good transit point for domestic or international connections, especially with LAX being only 150 miles away. The other problem is that the airport is at the bottom of a cliff, and in a geographically constrained site with a relatively show runway that can’t be extended, which leads to weight restrictions, which can get especially severe when the winds are blowing from the “wrong” direction and the panes have to take off in the direction of the cliff.

  2. Eric O May 27, 2012 at 5:08 pm #

    This casual manner of interpreting factors abstractly without bothering to look at the actual physical systems that might be involved also seems to pervade the banter of economists.
    An amusing example of late is Ed Glaeser pinning Jane Jacobs’s “ideal” density for urbanity to 150 dwelling units per net acre. Ok. He then goes on to add that this means never building taller than 6 stories (reading: her ideal urban district is a low-rise environment). Keep in mind that transit-rich central Paris struggles to fit 90 dwellings per acre in 6 stories. But such Glaeser coolly surmises by multiplying 150 by 1600 sq.ft./apartment and dividing the total by an acre to get number of stories. Unfortunately, his conceptualization missed the fact that no one builds window-less apartments! That form might be an issue here.
    (Glaeser could have just asked his friendly neighborhood architect to point him to what 150 dwellings per acre looks like. An example here. Portland’s built out area of the Pearl District is probably in this ballpark).

  3. Joel N. Weber II May 27, 2012 at 8:45 pm #

    The primary census area table at considers Riverside a part of the Los Angeles area.

  4. Daniel Sparing May 28, 2012 at 12:12 am #

    Something I’d look at is the amount of transferring (connecting 😉 ) passengers vs. passengers with origin/destination in a given city.
    There are two distinct effects we are talking about:
    – as the city grows, it will gain more name and also it will make more and more sense to fly there, whether for leisure of business, and this growth can be bigger than population growth
    – a big city, and its big airport has a higher chance that an airline decides to use its airport as a hub, which brings loads of transferring passengers and airport (and tax) revenue, however, these people will not visit the city itself.
    So while both phenomena happen, it can very well be that one city has a predominantly transit airport (Frankfurt, Atlanta, Munich, Singapore) while another city fails to create a dominant airline hub but still has large traffic of its own (Brussels, Barcelona).
    I am not sure if my examples are the best ones but you get the point.

  5. M1EK May 28, 2012 at 5:50 pm #

    Yeah, transferring is not “connecting” in the sense the author tries to imply here. “connections” imply good things; nobody appreciates having to change planes in DFW just because that’s where American’s hub happens to be, and nobody appreciates having to change buses in a transfer barn on the outskirts of town just so we can say we are intermodal and have lots of connections.
    At least in the transit case if your ‘transfer barn’ is something good in the middle of the city, you might get something positive out of it as passengers who constantly go through there might get out and walk around someday. The same thing is never going to be true for the huge airport on the outskirts of the metro.

  6. Joel N. Weber II May 28, 2012 at 8:24 pm # claims San Diego is a focus city for Southwest (which I’d always understood to be a mini-hub).
    Boston is not listed in nor is it a Southwest focus city, yet Boston is the 10th largest MSA and the 5th largest primary census area. Boston does get credited as a hub in the list of Essential Air Service routes at
    Boston does have moderately limited runway length, as well as being near a corner.

    • Tommy January 27, 2018 at 4:57 pm #

      Boston is a focus city for JetBlue.

  7. Peaton May 28, 2012 at 11:15 pm #

    Could you elucidate how network structure would play out in a causal structure or schematic framework? I ask because I think your observations are interesting, but I don’t clearly envision what your proposed alternative analysis would look like.

  8. Zoltán May 29, 2012 at 7:38 am #

    Oh look, it’s the transfer barn straw man again. Who exactly is advocating those here?

  9. EngineerScotty May 29, 2012 at 11:12 am #

    I remember back in the 1990s, there was some discussion in aviation circles about building wayports–airports located in the middle of nowhere, which would serve as transfer hubs but would not serve as an origin or destination for trips (or for only a minority of trips). The idea was to relieve overclogged hubs in urban airports, so that said urban airport’s capacity would be mostly allocated for travellers who were travelling to or from that city.
    Needless to say, no wayports have been constructed anywhere in the US–probably the closest one might come to a “wayport” are cities with multiple airports located nearby, often a big one geared towards long-distance travel, and smaller regional airports that serve more regional traffic (such as John Wayne or Burbank airports in LA).
    A couple thoughts as to why:
    1) Supporting an airport requires a lot of employees–i.e. there has to be a city of some sort nearby.
    2) Building airports is expensive. Airlines are happy to pass this cost off onto cities or port authorities; and cities and ports generally aren’t interested in overbuilding to attract transfer business.
    3) The existing hub model is good business for the cities that are hubs–they rake in the dough from landing fees and such, and having a big important airport is, as Florida notes, a big civic boost.
    4) It’s good business for the hub airline, who can jack up rates in and out of the hub, due to far less competition.

  10. Anthony May 29, 2012 at 12:12 pm #

    Good read.
    Aligning hubs at primary destination cities helps the airlines fill as many seats possible.

  11. Jarrett at May 29, 2012 at 5:20 pm #

    Scotty.  I do the "wayport" thinking had some influence on the siting of wildly distant airports, notably Denver,  But ultimately, diverse markets mean better markets, just as in transit.

  12. Wad May 29, 2012 at 5:52 pm #

    One factor against wayports may have been the culture shock from the Mirabel Airport failure in Montreal. Because Mirabel was built on a grand scale, and it failed spectacularly, airport planning became gun-shy at attempting another one like it.
    @EngineerScotty, the satellite airport market you are describing is akin to an express bus network in local transit.
    In the L.A. area, LAX is still the alpha airport, offering the most flights at lower prices. The secondary airports relieve the gate burden at LAX and allow suburban travelers to catch a flight closer to home. In a way, LAX is a wayport right now. It’s against the water at the western edge of L.A., making it far removed from most residences.
    The downside: Just as express buses charge a premium for travel, the satellite airport often has fares costlier than an LAX pair. For instance, LAX-SFO trips tend to be reasonably priced, but try any other permutation of a SoCal<->Bay Area trip and the ticket is more expensive, even though the planes are essentially going to the same place. I’ve seen $29 LAX<->SFO flights, but anything involving OAK, San Jose, Burbank, John Wayne, Long Beach or Ontario is at least triple that fare.

  13. chrismealy May 29, 2012 at 10:31 pm #

    I wish bus transfers were in nice places. The ones I’m familiar with in Seattle are on highways, in shopping mall parking lots, and where members of the lumpenproletariat shoot craps and screams at each other. Has anyone done deliberate placemaking for transfer points? It’d be cool to have Portland-style food cart pods where you wait. Waiting 10-20 minutes for a bus wouldn’t be as bad if you could sit in the shade and get a slice or some coffee.

  14. M1EK May 30, 2012 at 4:13 am #

    “Oh look, it’s the transfer barn straw man again. Who exactly is advocating those here?”
    Sorry. I mispelled “intermodal centers”.

  15. Ben Smith May 31, 2012 at 1:22 pm #

    I go to York University (the other university in Toronto, Florida teaches at the University of Toronto) and we have spent a fair amount of time critiquing Florida. Besides the fact that I disagree with his views as to what makes a good city, he really seems to ignore that correlation DOES NOT mean causation.

  16. Jarrett at May 31, 2012 at 9:37 pm #

    Peaton.  Not sure what you're asking.  Simply observing that network structure identifies certain sites as nodes of connectivity purely in functional terms, and that these nodes will have traffic disproportionate to their local demand due to this effect. 

  17. EN57 June 1, 2012 at 12:30 am #

    Peaton. Following on from Jarrett, the airport has additional activity due to passengers transferring between flights within the airport, not directly related to the host city’s size or levels of economic or social activity.
    In attempting to measure accelerated economic and social activity due to population size – perhaps measuring the numbers of people moving between the airport and the city might have been a better measure in this case.
    I can see however, how the city’s size attracts sufficient numbers of flights to make transfers convenient at that city’s major airport; and that lots of passengers seeking/forced to make transfers through that city generates additional demand that makes more frequent flights viable.
    Cities with frequent flights are more accessible relative to their competitors, so that there is the potential for economic and social benefits to flow to the city – indirectly helped by lots of transferring passengers (in addition to the benefits of landing fees and airport jobs). There might be a network-city feedback thing happening.

  18. Alon Levy June 4, 2012 at 4:27 pm #

    There are three snags in this theory. First, if you tally everything together, the large-city effect disappears. New York has twice the air traffic of Dallas, despite a factor-of-four size difference.
    Second, partially correcting the first snag, O&D numbers look very different. These deemphasize Middle American hub cities like Chicago and Dallas, but then Las Vegas surges to near the top of the list. For O&D data, look at table 6 of the consumer airfare report, which gives domestic US O&D numbers for each city pair; tally them for each city to get O&D traffic.
    And third, because of global destination centralization, international hubs tend to be located where everyone wants to go (e.g. New York and London). This overstates their traffic, because of international-domestic connections. For example: JFK and Newark have 33 million annual international passengers per year, in both direction. This means 16.5 million people use them. But New York’s annual number of international visitors is only 6.2 million. (Americans using New York to travel abroad are much less likely to be New Yorkers than non-Americans using New York to traveling to the US are to be traveling to New York; this is the flip side of destination centralization.)

  19. Wad June 7, 2012 at 4:01 pm #

    @Chrismealy, on placemaking at transit centers, San Diego County has done amazing work.
    The Trolley has been able to attract whole hives of activity around its stations, especially the Green Line. Some stations are also busy bus connection points in or next to activity centers (San Diego State University, Fashion Valley mall, Old Town). Even in the north county, where the Sprinter DMU runs, important destinations just happened to fall into place along the rail line long before it even existed. Vista is next to busy shopping centers, San Marcos has its transit center at a community college, and Escondido’s and Oceanside’s stations are in built-up communities.

  20. Kenny June 9, 2012 at 7:33 am #

    Ben Smith – I’m glad to see someone else agrees. I love reading the Atlantic Cities site, but I hate it whenever one of Richard Florida’s posts comes up. I think in every post he says something like “Correlation does not mean causation, but…” and then proceeds to list a bunch of correlations between various ad hoc categories that he constructed simply to support his overall story. And it really frustrates me because his overall story is one that I largely like!

  21. Akshaya Gawarikar August 30, 2012 at 10:30 pm #

    I was searching the net on some information on transit choices made by people. I wanted to find out if its really true that people would choose and International transit point over a domestic transit point. Its an important question for India, We tend to loose our Australia bound passengers to airlines in South east Asia and hence Australian carriers do not find it profitable to fly directly. Further, to add the ongoing debate on transit patterns, Airline also form transit hubs over an airport because of airport and fuel cost advantages.Its valid in India, not sure about USA.

  22. Aidan Stanger September 3, 2012 at 5:59 am #

    Surely this theory would be pretty easy to test? Don’t we have stats that exclude interlining passengers?

  23. Harry Bill January 5, 2013 at 3:01 am #

    Good article. I think that it is really necessary to have more transit facility due to overgrowing population at a faster rate. The increasing number rise to more population at metros demanding more transport.

  24. matt439miller April 25, 2019 at 3:08 pm #

    ‘Super-linear scaling’ is the generic descriptive term that complex-systems folk use to talk about a wide variety of phenomena, including both network and non-network effects. (VMT:population is characterized by super-linear scaling, for example).

    The number of passengers at any airport has a number of causes, but can likely be attributed to one of two factors: local demand and the agglomeration economies offered by by massification. Local demand is probably itself non-linear: larger metro areas tend to be richer, richer areas tend to fly more. But massification is probably the main driver. Cost per flight rises with the square of area, but passengers per flight rise with the cube of area. Big planes are more efficient. So the hub-and-spoke model uses small planes to massify into large vehicles. The efficiencies are such that SLC-DC flights often take me through LA. That said, there is indubitably some advantage to being on the way: I spend a lot of time waiting for flights in Detroit, Chicago, Atlanta and DFW.