The most prosperous American cities have been able to establish a foothold in innovative, technology-driven industries. This economic development path tends to be paved by young, high-growth companies that achieve technological breakthroughs, create new markets, and yield quality job growth. Iconic examples include Amazon in Seattle, Qualcomm in San Diego, and Epic Systems in Madison, Wis. More often, this development occurs via thousands of less recognizable 50- to 250-person companies whose growth enhances prosperity across the American landscape every year. As a result, cities and states have placed technology entrepreneurship at the center of their economic development strategies.
The challenge for leaders guiding these types of strategies is that it’s difficult to identify the set of local companies on the cusp of achieving significant, valuable product and process breakthroughs—what we call the “economic development frontier.” But understanding that frontier is critical for strategy because it is increasingly the key source of new innovation, growth, and prosperity for any city or state.
This exploratory brief offers a new analysis that seeks to fill this knowledge gap by better measuring the “economic development frontier” of U.S. metropolitan areas, analyzing Crunchbase data of high-tech startups using principles and methods advanced by the field of economic complexity.
What is the economic development frontier, and why use startups to measure it?
The economic development frontier is, by definition, future-oriented and very difficult to measure, especially in an advanced economy such as the United States.
Yet startups provide a useful metric for several reasons. First, recent evidence demonstrates the importance of young, dynamic firms in driving net job creation and productivity growth.1 Second, frontier activities at the local level are currently hard to measure using other traditional economic datasets. Industry datasets have a time lag and classify economic activities according to somewhat outdated definitions that do not fully capture the technological progress of recent years. Patent data measure technology inventions—an important catalyst for innovation-led growth—but typically via inscrutable definitions that do not coincide with how economic policymakers or business leaders think about the economy. Both classification systems, for instance, tend to miss new sources of economic value such as analytics, artificial intelligence, or cybersecurity.
This analysis deploys crowdsourced data from Crunchbase, a continuously updated platform of technology-based startups. Companies on the Crunchbase platform are tagged with hundreds of categories, ranging from cutting-edge technologies such as autonomous vehicles, neuroscience, and 3D technology, to niche markets including gamification, career planning, and content delivery networks. The benefit of Crunchbase is that it provides more up-to-date information on young firms, and these tags better align with the realities of the modern tech frontier. We identified 27,415 innovative, young U.S. firms on the Crunchbase platform, operating in 421 metro areas across 424 technology categories.2
The benefits of a dataset such as Crunchbase also come with challenges. While Crunchbase conducts quality checks on its data, its crowdsourced nature ensures that it will not include every relevant startup, and that it likely suffers from some selection bias issues. As such, this analysis should be treated as a preliminary look at the technology frontier, but in no way a definitive analysis of it.
Why measure the frontier using “complexity”?
Discussions on what drives regional economic growth often center around specialization, the degree to which a local economy concentrates its output in a limited variety of goods and services. From David Ricardo’s comparative advantage to Michael Porter’s industry clusters, a common belief among the economic development community is that creating conditions that foster specialization is key to future economic prosperity.
However, if specialization is all that matters, and (nearly) every economy has comparative advantages in some products, why are some places thriving while others are struggling? The answer is that not all specializations have equal value. It matters whether an existing specialization can easily be replicated in many other regions, thus making it not-so-special. This central insight informed Hidalgo and Hausmann’s development of the “economic complexity index,” which showed that underlying economic capabilities could explain the productivity differences among countries.
What makes an economy complex? A metro area’s economic development frontier can be defined by the diversity of technological capabilities it possesses, as well as the rarity of those capabilities it possesses.
Researchers have used various data to understand how economic complexity relates to regional economic outcomes. Our colleagues at Brookings calculated the complexity of U.S. metro areas using industry and occupation data, and showed that higher complexity is associated with higher population growth. Balland and Rigby analyzed patent data to determine the “knowledge complexity” of U.S. metro areas, revealing that it is highly correlated with long-term economic performance.
Our “Startup Complexity Index” (SCI) combines metrics of startup diversity and startup ubiquity. Startup diversity refers to the number of technological categories in which a metro area’s startups exhibit advantage, or a higher-than-average propensity to innovate.3 Of the 99 metro areas we analyzed, San Francisco has the highest startup diversity, registering an advantage in 288 technology categories. By contrast, 15 metro areas have an advantage in just one technology category.
Startup ubiquity refers to the total number of metro areas that have an advantage in a technology category. Crunchbase’s broadest technology categories such as software, health care, and information technology are the most ubiquitous. The least ubiquitous categories include enterprise resource planning, podcasts, and fleet management, among others. The SCI therefore captures the complexity of startup ecosystem based on the interaction between diversity and ubiquity.4
More diversified metro areas house startups in more exclusive technology categories
Startup complexity is highly uneven across U.S. metro areas. Not surprisingly, the metro areas with the highest complexity in their startup ecosystem are San Francisco, New York, Los Angeles, San Jose, Calif., and Boston. Collectively, these tech giants house more than half of all the companies in our sample, and dominate most of the technology categories. Among the least complex metro areas are Jacksonville, Fla., Louisville, Ky., Syracuse, N.Y., and Blacksburg-Christiansburg, Va. These metro areas exhibit an advantage in just one category—software, the most ubiquitous technological category.
The largest and “techiest” U.S. metro areas boast higher Startup Complexity
Source: Brookings analysis of Crunchbase data
SCI is a strong indicator of key prosperity metrics
A closer look at the Startup Complexity Index distribution is even more concerning. Outside the tech hubs, most metro areas have low startup complexity. Notably, the complexity scores of college towns such as Madison, Wis., Boulder, Colo., Ann Arbor, Mich., and Durham, N.C.—paragons of “rise-of-the-rest” dynamics—are dwarfed by the more established regional ecosystems.
The majority of U.S. metro areas have low startup complexity
Source: Brookings analysis of Crunchbase data
This highly uneven geography of startup complexity matters, because the SCI appears to be a strong indicator of local prosperity. In metro areas with a higher SCI, workers are more productive and earn higher wages, and households have higher incomes. To be clear, we are unable to determine whether these are causal relationships, but the SCI has very strong correlations with key metrics of regional prosperity.
Higher SCI is associated with higher wages, income, and output per job
Source: Brookings analysis of Crunchbase data
Note: Output per job is a measure of productivity
Applying the Startup Complexity Index
Similar to Rigby and Balland’s Patent Complexity Index and even more powerfully than higher education attainment, the SCI is likely capturing technology and business capabilities that are highly valuable in the modern economy. Understanding these capabilities is critical for leaders who are seeking to guide and invest in the entrepreneurship activities that drive the economic development frontier forward.
We also think these data may have practical applications for economic development strategies, since they capture a metro area’s relative position in key technology specializations much better than any source of which we are aware. In subsequent analyses, our goal is to practically apply these data to inform local development strategies, as forging new, more complex technological advantages is fundamental to advancing metropolitan prosperity.
Interactive by Alec Friedhoff.
Report Produced by Brookings Metro
- Ryan Decker et al., “The Role of Entrepreneurship in US Job Creation and Economic Dynamism,” Journal of Economic Perspectives 28 (3): 3-24.
- We define innovative young firms as private companies on Crunchbase platform that were founded in the last decade and have received at least one investment in the last five years. Please check the Crunchbase Knowledge Center for details on categories and funding types.
- We consider a metro area as having an advantage in one technology category if the metro area has at least four young firms in that category, and the location quotient of that technology exceeds one. Categories that are associated with fewer than 10 startups nationwide are considered atypical and removed from this analysis. According to our definition, 322 of the 421 metro areas didn’t have a revealed advantage in any technology category.
- Learn more about economic complexity methodology at: http://atlas.cid.harvard.edu/