Addressing regional differences in startup activity
In the United States there is a certain regionalization of reputation when it comes to startup activity. Silicon Valley does internet, Boston biotech, New York City fintech, and so on. We wanted to know if these disparate regional reputations were backed up by the data. Here is a breakdown of geographic concentrations of companies in our sample (Fig. 1):
For the purposes of our analysis, we will focus on those geographies with ten or more companies present, while recognizing that sample sizes limit our analytical capabilities beyond the top four regional centers of Silicon Valley, New York, Boston, and Los Angeles. We will also include an aggregated analysis of companies located outside of larger innovation hubs (those designated as “Other Cities”); while this will not give us insight into any particular geographies, it will inform us as to what kind of activity is occurring in smaller cities that have had fewer companies break the $500 million threshold since 1998.
As noted in our previous post, there are some limitations to our data. CB Insights’ methodology uses “Internet” as a catch-all for most tech-oriented activities that are not mobile app-based. See below for the top 20 most valuable companies in our dataset along with their respective industries (Fig. 2):
Here we can see some of the challenges of working with CB Insights’ classifications. Airbnb and Twitter are both categorized as Internet companies, while Uber and Lyft find themselves in the Mobile & Telecommunications bracket. This despite the fact that Airbnb, Uber, and Lyft all have far more in common with one another than they do with Twitter, which, ironically, evolved as a product that was essentially a mobile app.
While it is important to note that these sectoral categorizations may not perfectly represent companies on an individual level, we still found that they are generally representative in the aggregate, particularly when you get away from Internet and Mobile & Telecommunications, which have significant overlap in activities. Companies from these categories also have very similar business needs in terms of access to talent and expertise, and the line between them will likely continue to blur as time goes on. Nonetheless, we found it useful to maintain them as distinct units of analysis for the purposes of our project.
Looking at the breakdown of companies by sector, we see that those companies classified under the Internet umbrella make up nearly half of our sample, with Healthcare and Mobile & Telecommunications rounding out the top three (Fig. 3):
When we analyze the sectoral makeup of specific geographies, it largely supports our preconceived notions about regional startup activity. First we will look at Silicon Valley (Fig. 4):
Internet, Mobile & Telecommunications, and Healthcare activity dominate the region in the same way it dominates our larger sample. Mobile & Telecommunications, third in our sample at large, take the second rank from Healthcare, perhaps due to the skills and expertise overlap between Internet and Mobile businesses, and Silicon Valley’s importance as a hub for Internet startups.
Moving beyond Silicon Valley, it is sensible to compare startup activity in the New York Tri-State with that in the Greater Boston area, as they rank neck and neck behind Silicon Valley for companies that met our baseline criteria, and each have well-developed innovation ecosystems, both in terms of talent and available financing (Figs. 5 and 6).
What stands out here relative to other geographies of interest is that Healthcare is uniquely preponderant in Boston. Beyond that, we see that proportional to Internet companies, Mobile & Telecommunications firms make up a much smaller share of total companies in these cities than they do in Silicon Valley, suggesting that the latter region has an edge in mobile application development. The lesson here is that if starting a health or biotech company in the Northeast, it makes sense to look to Boston, as that region’s reputation for leadership in this sector is well-deserved. At the same time, the New York Tri-State is probably a better region for starting a company oriented towards the internet (recognizing that CB Insights’ Internet category is quite broad, as noted above).
Interestingly, our fourth-ranked geography, Los Angeles, has a ratio of Mobile & Telecommunications to Internet firms that is similar to that of Silicon Valley, albeit with a much smaller sample size (Fig. 7):
Interestingly, we see that Los Angeles’ industries break down on very similar lines to those of San Francisco. While the sample is too small to draw any conclusions about causality from this, it bears mentioning that California’s leading startup centers appear to share similar compositions of industries, as it could be due to shared talent and funding networks.
After Los Angeles, we get into cities with fewer data points. We see this as being potentially useful information for our readers, but we do caution against making conclusive arguments off of such small sample sizes (Fig. 8):
For the most part, company composition in these cities is similar to what we saw in our aggregated sample. The main relationship that stands out is that between Healthcare and San Diego and Philadelphia. Both of these cities have fewer overall companies, but Healthcare predominates among those companies that do exist there. As noted, we want to avoid making any unjustified claims based on this data, but it could warrant additional investigation.
Below is a breakdown of what we found was happening outside of larger innovation hubs (Fig. 9):
The only thing that stands out here is a disproportionately small share of the companies coming from the Internet sector, balanced by a somewhat larger share of companies belonging to Healthcare. Again, there are potential sample size issues in play here, but many of the smaller cities included in the Other Cities designation have based much of their economic development around hospitals as anchor institutions, which could play into the success of healthcare-oriented companies within them.
While we failed to identify anything groundbreaking with the above analysis, we found it useful to examine whether our preconceived notions as to where certain industries thrive is supported by data analysis. Ultimately we found that Silicon Valley’s activity is fairly representative of what is happening at the national level, but does not necessarily mirror what is happening in some of its larger tech-hub competitors in New York and Boston.
