Big City Advantage in Innovation Not What It Used to Be

Image credit: “Cities and Ideas,” National Bureau of Economic Research.

by James A. Bacon

Maybe the Internet is allowing innovation and creativity to break free from the confines of geography after all. Economists conventionally argue that large metropolitan areas are better incubators of inventions and innovations than smaller cities and rural areas. However, a new study, “Cities and Ideas,” by Mikko Packalen and Jay Bhattacharya, finds that the relationship between city size and inventiveness is not as strong as it once was.

I partially jest when I refer to the impact of the Internet. In the 1990s, starry-eyed dreamers theorized that the Internet would enable people to plug into global commerce from a mountain cabin or small town coffee shop. As rural America continues to empty out and population migrates to the bigger cities, that promise now seems a cruel joke. But something is changing. As Packalen and Bhattacharya demonstrate, big cities are far less dominant than they were a century ago. Furthermore, the geographic de-concentration of invention long preceded the rise of the Internet. Other trends — the proliferation of telephones, the spread of roads and the automobile, the rise of land-grant universities in out-of-the-way places — may have played equally critical roles in diffusing the capacity for invention.

Scholars first theorized about the correlation between city size and innovation, which they called an “agglomeration effect” in the 1920s. There was a solid basis for the theory then — large cities were the dominant incubators of innovation; rural areas were backwaters. But even as agglomeration-effect theory became more deeply rooted among scholars studying urban geography, the reality upon which the theory was based was steadily eroding.

To measure invention, Packalen and Bhattacharya built a database of U.S. patents between 1836 and 2010, identifying the inputs for each patent from previous patents, how old those inputs were, and where the inventions took place. The study gives great weight to the age of the patents, distinguishing between patented inventions that draw upon new ideas and inventions that draw upon older ideas. The authors explain:

If we find that inventors in large cities build on fresh ideas more often than inventors in smaller  cities, the evidence would quantify a specific benefit to locating inventive ideas in large cities. On the other hand, if we find that inventors in large cities are no more likely to try out new ideas in their work than inventors in smaller cities, the evidence would suggest that location may be largely irrelevant for inventive performance.

The dominant theory in academia today is that size and density matter. The bigger and denser a metropolitan region, the greater the number of people who can interact on a face-to-face basis. Proximity to other people allows innovators to conceive, discuss and test new ideas, and commercialize them in the marketplace. As can be seen in the chart above, which compares idea inputs of patents between cities in the 95th and 50th percentiles (large versus midsized cities) that was certainly true a century ago. But the dominance of big cities has declined, despite a few ups and downs, since then. Today, adjusted for the margin of error, there is very little difference at all.

Bacon’s bottom line: I am not equipped to dissect the statistical methodology employed to reach these conclusions, although I do have a couple of questions. Why the focus on the newness of the ideas behind the patents? Are patents based on newer ideas necessarily more consequential than those based on older inputs? Why not measure the frequency of patents? Surely the number of patents, adjusted for population size, is also an important indicator — perhaps the most important indicator — of inventiveness.

Those questions aside, “Cities and Ideas” would seem to provide hope for America’s small towns and rural regions. In this blog, I have frequently written about the tremendous disadvantages facing smaller communities when competing with big cities for human capital and corporate investment. The odds seem hopelessly stacked against the little guys. But if it turns out that it’s just as easy to keep up with the latest technology in Small Town USA as it is in Big City USA, a lot of people — and that includes me — may have to adjust their thinking.

Share this article


(comments below)


(comments below)


  1. Hamilton Lombard Avatar
    Hamilton Lombard

    I suspect there are two post-WWII trends influencing the chart. One is sprawl, which means large cities like Houston don’t necessarily have higher density levels than medium or small cities. The other trend is the rise of research universities which are more likely to be in small and medium sized cities like Charlottesville and Blacksburg.

    1. virginiagal2 Avatar

      Also, with the Internet allowing connectivity to anywhere, you are getting more and more well educated people who can live anywhere, who are choosing to live both in places like Charlottesville and Blacksburg and, stepping out one level further, in the rural areas outside of those small cities.

  2. Cville Resident Avatar
    Cville Resident

    Agree Hamilton. Are “university towns” and their communities “small” or “rural”? If so, I don’t see a lot of utility in the study. Of course, places like Charlottesville, Eugene, Bloomington, Chapel Hill, Ann Arbor, Ithaca, Hanover New Hampshire, New Haven, Princeton, and Blacksburg (even though they’re small) have exploded their patent acquisition in the past few decades. At one time, Blacksburg was touting itself as the most wired city in America (not sure about the validity of the claim). But does anyone actually lump these places in with “small town America”? I haven’t read the study, so perhaps it makes allowances for communities like that?

    1. virginiagal2 Avatar

      FWIW, it seems like a lot of people use small cities like Blacksburg or Charlottesville as a place to live near – so they’re no longer just living in C’ville or even Ivy, they’re out to Waynesboro or Nellysford or Palmyra or Gordonsville – or they go to Giles County instead of Blacksburg – and drive in.

      I’ll also comment that a lot of well educated and well to do people choose this because they want to have horses or a mini-farm, and have jobs that are either self-employed to one degree or another, or can work remote – or they have family from the area and want to keep the family property.

  3. LarrytheG Avatar

    Not all internet is the same and I would assert that it changes the definition of “rural” for all practical purposes.

    “rural” today is where the internet is nonexistent or weak and/or unreliable and places that have true high-speed internet are no longer “rural” by modern standards.

    I’d argue that the definition of rural has changed – again – more akin to when rural meant – with or without electricity/phone.

    1. virginiagal2 Avatar

      Depends on context. If you mean, can I work remote, there is no rural any more. I get emails occasionally from a friend who does charitable work, while he’s in remote areas of the DRC, and that’s rural by any definition.

      OTOH, if you mean, can I keep goats and pigs, rural isn’t defined by how wired I am. It’s defined by neighbors and zoning.

  4. LarrytheG Avatar

    re: the most wired …

    in the ordinary scheme of things why would you run high speed internet to any rural area just because it had a college?

    if you can justify it for a college – why not other things like k-12 schools?

    1. virginiagal2 Avatar

      All colleges and universities have to have high speed Internet. Most (I don’t claim all, I don’t know all) rural public schools I know of also have high speed Internet.

      Which is actually a good way to get decent Internet in the country – choose to live near a school, university, or govt office.

  5. Cville Resident Avatar
    Cville Resident

    Interesting experiment if someone could find the numbers. What % of patents acquired by Virginia residents/businesses in a particular year were from the Urban Crescent? Then, what % of patents acquired by Virginia residents/businesses in a particular year were from the Urban Crescent, Charlottesville, and Blacksburg?

    I’d imagine at least 98% of all patents acquired in the state go to the Urban Crescent (NoVa, Richmond, Hampton Roads), Charlottesville, and Blacksburg.

  6. LarrytheG Avatar

    One thing that many folks may not know and that is govt employees get patents also… The govt owns the patents. And there are a significant number of them emanating from R&D work.

  7. Blackbird Fly Avatar
    Blackbird Fly

    Disproportionately weighting “newer” ideas devalues the advantage large, dense cities have of offering exposure of established ideas to more people; hence the increased likelihood that they will in some way improve the “old” idea. In other words, it seems apparent to me that ideas in large, dense settings are more likely to take a course of evolution while the relatively disconnected small communities may be more likely to create something novel being that they may be less familiar to other ideas.

    1. virginiagal2 Avatar

      I’m not so sure that you really get exposed to more people. I collaborate with a lot of people, and it’s not rare for me to collaborate via email both with folks in the same office and on different continents.

      As much time as most nerds spend on the Internet, reading and talking about technical things, you get exposed to a lot of new ideas every day. You’re not going to be intellectually isolated, regardless of location, unless you choose to be.

Leave a Reply