Methodology
In approaching this project, our initial goal was to get a better picture of the implications of geography on startup activity in the United States. We wanted to see where highly successful startup activity was concentrated, and what forces drove it. To this end, we found that CB Insights offered the data set that best represented what we needed to find.
To define a “highly successful” startup, we set a benchmark of $500 million. We saw this as a number that would allow for venture capital firms to engage with companies on a profitable basis. Note that for privately held companies these are CB Insights’ proprietary estimates, and we do not have insight into the metrics used to arrive at these conclusions, but this still represents the best publicly available dataset.
Based on that $500 million benchmark, we then selected for US-based companies that were private, had been acquired, or had gone through the initial public offering process. We wanted to sort for companies founded since 1998, and also wanted to know where those companies were founded. Both of these issues were problematic, as CB Insights does not provide this data. To solve for this, we utilized open resources like Crunchbase and Wikipedia to determine when and where companies were founded, filtering out for those companies founded prior to 1998. This left us with a sample of 582 startups.
Those 582 companies were based in a number of different cities scattered around urban innovation cores. For example, Palo Alto and Oakland can both be considered to be a part of the Silicon Valley startup ecosystem in terms of factors like financing and education that impact regional network effects. To clean up our data, we elected to segregate cities into innovation hubs. We avoided using US Census-based designations, as in some cases these might represent multiple innovation hubs that overlap with one another. We instead used a subjective standard to assign founding cities to those urban centers with which they are most likely to be associated. This included:
· Merging New Jersey and Connecticut with New York to form a New York Tri-State region. Given the region’s integrated transportation and education networks, as well as New York City’s preeminence as a regional — and national — financial center, this seemed appropriate.
· Assigning companies based in Washington, DC’s Virginia and Maryland suburbs to Washington, DC.
· Associating New Hampshire’s companies — all located in the southeastern portion of the state — with Boston, due to the city’s regional hegemony.
· Terming the entire Bay Area “Silicon Valley” rather than dividing it into different sections. We deemed this to be sensible given the integrated nature of funding and talent networks in the region.
These regional assignations left us with the following breakdown of companies (Fig. 1):
Note that 34 of our companies — one of our larger groupings at approximately 6% of the total sample — are located outside of cities that have large numbers of companies that met our filter criteria of $500 million. This does not meant that these cities are not innovation centers of their own — in many cases they are — but instead that there were not enough companies in our filtered data set to provide any significant insight into the activity in those cities. Startups from these cities have been assigned the “Other Cities” designation (Fig. 2):
Cities included under the “Other Cities” assignation include our own home of Pittsburgh, Nashville, and Oklahoma City. All of these places have interesting innovation activity going on, but have yet to see a large number of highly successful startups, at least according to our guidelines. The fact that these cities have been less “successful” by our metric may be due to a lack of developed venture capital networks and an associated herd mentality — and perhaps a coastal bias — associated with valuing startups.
Going forward, we will seek to delve deeper into regional differences in startup activity, particularly as it pertains to different industry sectors predominating in specific geographies.
