Update: December 4, 2010An
article published by the Huffington Post and posted on Yahoo News reports that I find "7,000 degrees within 7,000 square miles" in the Bay Area. This is simply not true. As the post below shows, I find that the city of San Francisco has slightly more than 7,000 college degrees per each of its roughly 47 square miles.
------------------------------------------------------------------------
Update: June 10, 2010If you arrived here via a link or news story that claims that this is a ranking of cities "from smartest to dumbest" please see my
follow-up post. These headlines are misleading, incorrect and jump to unfair conclusions about my analysis. In no way do I endorse the conclusion that having a low degree density makes a city "dumb".
------------------------------------------------------------------------
Original Post: March 23, 2010It's becoming increasingly accepted that there is real economic value to bringing a lot of smart and entrepreneurial people together in the same place. This can be tough to measure, unfortunately. Perhaps best proxy we have available is educational attainment - usually measured as the number of people in a particular place with bachelor's degrees or higher, as reported by the Census Bureau.
I have seen this done in two ways. The first approach reports the proportion of college degree holders in a particular city. Usually, college towns like Austin, Texas and other stereotypically "brainy cities" like San Francisco and Seattle do well. The second approach looks at the raw number of people with bachelor's degrees in a particular city. Using this approach, big cities usually do the best, as they should. New York City has a huge population, of course many of its residents will have college degrees.
Both of these approaches have flaws. The theory that there is economic value to having smart people together rests on the assumption that smart people collaborate with each other. You could have a whole bunch of smart people in one place, but if they don't interact with each other, what's the value?
That's why I propose we start using educational attainment density, measured as college degree holders per square mile. I went ahead and collected and analyzed the data; and I've broken the results down into four sections.
Part One: Clusters of Smart PeopleI started compiling the data when I discovered a problem: there is no reliable report on the land area of metropolitan areas. It could be pieced together using the OMB's
definitions of metropolitan areas, but that would have taken forever, and I simply don't have the time to invest.
Instead, I compiled the data at two geographic levels: first at the city level and second at the "urban county" level. I realize that comparing these geographies is not always entirely fair. That's why I'm giving away the spreadsheet with all of my work to anyone who wants to build upon this analysis (download it
here).
I picked these cities by looking at the 50 largest metro areas by population and pulling what I deemed to be the "primary city" from each. In two metro areas, the Twin Cities and Bay Area, I pulled two "primary cities".
At the city-level, college degree density breaks down like this:
(click to enlarge)
And at the county-level, it looks like this:
(click to enlarge)
Note: New York City is excluded from the county-level analysis because I had a really difficult time determining its urban county components.
Now, you can look at these graphs and say, "this merely reflects the overall population density in these cities" and you would be on to something...
Part Two: Predicting Degree DensityIf we regress degree density on population density we estimate a positive and statistically significant correlation result. At the city-level:
(click to enlarge)
And at the county-level:
(click to enlarge)
An interesting extension is to use a residual analysis to see which places have a higher or lower degree density than the population density predicts.
Part Three: Who's Doing Better or Worse Than ExpectedFirst, at the city-level:
(click to enlarge)
Essentially, what this means is that Nashville has more than twice as many degree-holders as its overall population density would predict. Detroit has less than half as many as its population density would predict. And Columbus and Cincinnati have about as many as we would expect.
Again, at the county-level:
(click to enlarge)
Part Four: Degree SprawlOf the cities and county pairs analyzed, population density is always higher (or the same) in the central city than in the urban county. However, in several cases, degree density is higher in the urban county than in the central city. These cities include:
- Louisville (113 fewer degrees per square mile in the city)
- Cleveland (108 fewer degrees...)
- Oklahoma City (26 fewer degrees )
- Jacksonville (13 fewer degrees...)
- Nashville (19 fewer degrees...)
- Indianapolis (7 fewer degrees...)
This data suggests some level of "degree sprawl" in these cities, where college degree holders and sprawling out into the suburbs rather than staying in the central city. While further research would be helpful, this preliminary result is particularly worrisome if you believe that metro areas need strong central cities and strong central cities need a lot of smart people.