Kauffman Report: It Takes Decades to Build a Startup Hub

The Kauffman Foundation caused a stir in economic-development circles last month with a report showing that high-tech startups are more evenly spread across the United States than many people assume.

When researcher Ian Hathaway, an economic advisor with the advocacy group Engine, ranked metropolitan areas in terms of their per-capita density of high-tech startups, his top-25 list included a number of seemingly surprising names like Boulder, Fort Collins, Loveland, Colorado Springs, and Grand Junction in Colorado; Cheyenne in Wyoming; Corvallis and Bend in Oregon; Huntsville in Alabama; and Provo and Orem in Utah. (See map, above right.)

Blogs and newspapers in cities from Burlington to Cheyenne to Dover to Sioux Falls trumpeted the findings as proof that traditional high-tech hubs like Silicon Valley, Boston, and Seattle don’t have a lock on startup activity, and that it’s possible for smaller cities to come from behind and build their own thriving startup ecosystems.

Not so fast, says a second report from the Kauffman Foundation.

Dane Stangler, director of research and policy at the Kansas City, MO-based foundation, says that when he dug deeper into the data in the appendices of his colleague’s paper, he found that most of the cities being portrayed in news reports as novel startup hubs actually have a long history of high-tech industrial activity and a strong culture of entrepreneurship. The foundation released his follow-up article on the subject, “Path-Dependent Startup Hubs,” last week.

“Some of the coverage of Ian’s paper really annoyed me,” Stangler tells Xconomy. “Ian’s paper was a really good starting point, but one thing it didn’t show was a comparison over time between cities, and by size class. If you look at these ‘new’ startup places, a lot of them were already there in 1990.”

In particular, many commentators marveled over Hathaway’s data on Boulder, which boasts the nation’s highest per-capita density of high-tech startups. It’s 6.3 times greater than the national average, outpacing even the San Jose-Sunnyvale-Santa Clara metropolitan area (aka Silicon Valley), where startup density is only 2.6 times the national average.

Most media reports attribute Boulder’s prominence as a startup hub to the arrival of venture firm Foundry Group and the creation of Techstars, a startup boot camp, both in 2006. But when Stangler compared regional high-tech rankings from 1990 and 2010, the lists were strikingly similar. Among small- to mid-sized metropolitan areas, Boulder was already ranked first in the nation in 1990, likely thanks to the presence of high-tech institutions like the University of Colorado, the National Center for Atmospheric Research, and the National Institute of Standards & Technology, as well as the legacy of 1970s-era anchor companies like IBM and StorageTek.

“What you never hear people says is, ‘Yes, Boulder’s great, but it was also great 25 years ago,’” Stangler says.

Boulder topped the list of small- to mid-sized cities with high startup density in 1990 and again in 2010. Source: Dane Stangler, "Path-Dependent Startup Hubs: Comparing Metropolitan Performance: High-Tech and ICT Startup Density," Ewing Marion Kauffman Foundation, September 2013.

Boulder topped the list of small- to mid-sized cities with high startup density in 1990 and again in 2010. Source: Dane Stangler, “Path-Dependent Startup Hubs: Comparing Metropolitan Performance: High-Tech and ICT Startup Density,” Ewing Marion Kauffman Foundation, September 2013.

In fact, it’s not Techstars that’s boosting Boulder, Stangler suggests—it’s probably the other way around. The presence of startup accelerators and other entrepreneurship programs “is less a cause of startup activity than an indication of the underlying strength of the region and a base of talent that these regions can build on,” Stangler wrote in the paper.

Overall, Stangler found that all of the large cities that ranked among the top 10 startup hubs in 2010 were already among the top 20 in 1990. And only one city that was in the top 10 in 1990 had dropped off the top-20 list by 2010 (Houston). “The biggest observation that jumps out,” Stangler wrote, “is how much persistence there is across time—at least over this two-decade period.”

It’s easy to make too much of data that ranks regions by per-capita startup density, Stangler goes on to argue. One difficulty is simply that smaller cities simply have smaller populations—the denominators in density calculations. That means that if just a few new startups come to town, it can change the numerator and create the appearance of a dramatic increase in regional startup density.

Among large metropolitan areas, Stangler points out, the city showing the biggest increase in startup density between 1990 and 2010 was the Kauffman Foundation’s hometown, Kansas City. The area showing the biggest decrease was (perhaps surprisingly) Silicon Valley. But in absolute terms, Kansas City still has few tech startups compared to Silicon Valley, and the two cities are unlikely to trade places on the economic stage anytime soon.

Comparing places like Kansas City, Corvallis, OR, or Ames, IA, to San Jose “is kind of like apples and oranges,” Stangler says.

The other subtlety obscured by the startup density rankings is what Stangler calls “path dependence.” Put plainly, historical accidents count for a lot, and it’s hard to build something from nothing.

The seemingly permanent boomtown that is Silicon Valley, for example, was actually more than a century in the making, and might never have coalesced where it did if not for three unrelated facts: 1) Railroad magnate and politician Leland Stanford owned a country estate in Santa Clara County. It became the setting for Stanford University, which was transformed during and after World War II into one of the world’s most effective engines for university-industry collaboration. 2) In 1931, the U.S. Navy decided to build a major airfield adjacent to Mountain View and Sunnyvale, providing a nucleation point for dozens of aerospace and defense companies. 3) William Shockley, who co-invented the transistor at Bell Labs in New Jersey, decided to open Shockley Semiconductor Laboratory in Mountain View so that he could live closer to his aging mother in Palo Alto. Shockley’s firm was the progenitor of Fairchild Semiconductor, which in turn gave birth to Intel and most of the other companies that put the silicon in Silicon Valley.

Any region hoping to build an analogous high-tech ecosystem faces a chicken-and-egg problem, as Xconomy documented recently in an in-depth series on the startup scene in Santa Cruz, CA. In short: you can’t build fast-growing companies without a local talent pool, and it’s hard to attract talent without an existing base of high-tech employers.

“Most entrepreneurs, especially in the tech sector, have left an existing high-tech job,” Stangler says. “None of these places woke up and said ‘We suddenly have a technology sector.’ There was always some sort of internal dynamic they were able to capitalize on.”

Corvallis, for example, had Nike down the road in Beaverton, OR, and was a short drive from outposts of Intel and Hewlett-Packard. Boise, ID, had Micron. Santa Cruz had Borland Software.

When Stangler plumbed Hathaway’s data for common characteristics that might explain how regions achieve high startup densities, he didn’t come up with much. Some startup-rich areas have diverse and highly educated populations; others don’t. Some regions boast a lot of socioeconomic mobility; others don’t. Some are home to top research universities or federal facilities like national laboratories or military bases; others aren’t.

The only common thread seems to be what Stangler calls “entrepreneurial genealogy”—a tradition of established companies that train future entrepreneurs and spawn spinoffs.

But he also suspects that policy quirks play a role: Does the local university have a well-oiled technology transfer office? Do local laws allow large companies to enforce non-compete agreements, which are often used to deter former employees from starting companies in related sectors?

“There are so many things that are invisible but important,” Stangler says. “It’s like the ‘dark matter’ of the regional economy.”

Large Metropolitan Areas Ranked by Startup Density, 1990 vs 2010. Source: Stangler, "Path-Dependent Startups."

Large Metropolitan Areas Ranked by Startup Density, 1990 vs 2010. Source: Stangler, “Path-Dependent Startups.”

Wade Roush is a contributing editor at Xconomy. Follow @wroush

Trending on Xconomy