Entrepreneurship May Work Like A Clock, But It Still Needs Winding: Exploring the Kauffman Study on New Firm Formation
Like others in the tech-journalism business, we here at Xconomy tend to pore over the latest statistics about the entrepreneurial economy pretty obsessively: how much money venture firms are raising and investing from quarter to quarter; how much they dole out to each new startup in their portfolios; how much these portfolio companies eventually return to their investors through mergers, acquisitions, or public offerings.
But what if none of this really matters? What if it turned out that the number of new companies created by entrepreneurs is pretty much the same every year—and that things like how much money venture firms are handing out, or how many companies are achieving lucrative exits, or how many students are graduating from business school, or how many startup incubator programs are springing up, make no difference whatsoever to the nation’s overall levels of entrepreneurial activity? Would this mean that all the conferences and white papers and blog posts about the best ways to boost innovation and entrepreneurship are, in the end, pointless?
Well, that’s a serious question now—because the last three decades of data, according to a new study from the Ewing Marion Kauffman Foundation in Kansas City, MO, show that the number of new businesses incorporated in the United States holds steady at about 700,000 per year, give or take 50,000. It’s as regular as clockwork. In fact, it’s as if American entrepreneurs were programmed to start 700,000 new ventures every year—in the same way that, say, American parents pass on the genes for red-headedness to roughly 170,000 newborns every year.
You can read all about it in Exploring Firm Formation: Why Is the Number of New Firms Constant?, by Kauffman Foundation senior analyst Dane Stangler and senior fellow Paul Kedrosky. (Kedrosky, a San Diego-based investor, entrepreneur, and essayist, is also an Xconomist, and to complete the disclosures, the Kauffman Foundation is an underwriter of Xconomy’s Startups Channel.) When I first met Stangler at a Kauffman Foundation function last October, he and Kedrosky were still puzzling over the numbers they’d been digging up from places like the Census Bureau, the Small Business Administration, and the Bureau of Labor Statistics, which all seemed to show the same thing: Americans start the same number of businesses every year, come hell or high water.
That is a remarkable and, at least on the surface, counterintuitive finding. As Stangler and Kedrosky point out in their final report, which was published Wednesday, a casual observer might guess that the number of new firms would fluctuate from year to year in response to such major forces as economic recessions or expansions, technological change, and the availability of capital and credit. (We certainly hear the howls of local technology innovators every time venture firms scale back the number or size of Series A rounds.) But these things don’t seem to make any difference in the big picture.
“It’s a real puzzle, and it didn’t appear as if anyone else had noticed it or written about it,” Stangler told me by phone yesterday.
He and Kedrosky might not have noticed the phenomenon themselves if they hadn’t already been examining, for a different study, the question of survival rates for new companies. The percentage of the companies founded in 1990 that were still in business in 1995, they’d found, is almost exactly the same as the percentage of companies founded in 2002 that were still around in 2007. (It’s about 50 percent.) “That was interesting,” Stangler says, “and one of the possible inputs to that is that the number of new companies founded each year is remarkably similar”—which turned out to be the case.
Nobody had noticed this fact before, Stangler speculates, because it’s about constancy, not change: “You don’t stop to think, ‘Why is there not a trend here? You have to recognize the absence of something.”
Being good scientists, Kedrosky and Stangler first checked to see whether there might be something wrong with their instruments—that is, that the data might be wrong or incomplete. But all the datasets they checked showed … Next Page »