Paul Kedrosky, a Canadian venture capitalist (and loquacious tweeter), recently asked a very good question: How many tech start-ups actually matter?

In an 11-page study released by the Kauffman Foundation earlier this week, Kedrosky parses through economic data to dissect a couple of rather large tech start-up clichés.

"Specifically, it is often claimed that there only are fifteen to twenty information technology companies created per year in the United States that turn out to 'matter,' where matter is defined as the company (relatively) promptly going from founding to $100 million in revenues," Kedrosky writes. "Further, and of real consequence to cities and regional economies, is that most such companies founded in any given year are thought to be in California."

"This paper," he writes, "tries to find out if the preceding is true."

First, and perhaps most importantly, Kedrosky lays out the precise definition of a company that matters. First, it must be scalable, and grow to at least $100 million in revenue. Second, it must be a "disproportionate creator of jobs." And third, the company must be a "disproportionate creator of wealth," which is to say, they not only make its founders wealthy, but put wealth back into the company's ecosystem. (Also important to note: Kedrosky isn't saying that companies with less than $100 million don't matter at all; while they're certainly vital for the economy, "even if high growth is not in their future.")

Below, Kedrosky shows the number of $100 million companies in the United States from 1980 to 2012. There are between 125 and 250 companies that reach this level of scale in any given year. Now, it's important to note that private company data is notoriously hard to verify, and the numbers represented here may be slightly undercounted due to survivorship bias.

But considering there are 552,000 new employer firms that start each year, this is a sobering reminder of how few companies ever actually become really sizeable.

"It is, as one might expect, a very small percentage, even if often by design, given that most companies are not growth firms, with little expectation to grow beyond a level that supports its founder's needs or objectives," writes Kedrosky. "Nevertheless, this number is important, and worth keeping in mind, both in the context of this paper and in the context of the economy itself."

Next, Kedrosky asks another important question: Do information technology companies really take up the lion's share of the country's highest-growth companies?

The verdict: Not really.

"Unsurprisingly, but contrary to some rhetoric, while information technology is important, it is not the most important contributor in percentage terms to the $100-million firms in the United States on a founding cohort basis," he writes.

The largest contributors, he found, were consumer discretionary companies (i.e. nonessential goods and services firms) and industrials. That would make sense, too: "After all, the consumer discretionary and the industrial sectors are the largest non-government segments of the U.S. economy, so it stands to reason they produce more companies, many of which, in turn, go on to become large and successful."

And what about the claim that high-growth companies are found in California? Sort of. Kedrosky analyzed the number of companies that reach the $100 million mark by region, and found the highest amount in the Southeast--comprising of states like Louisiana, Georgia, and Kentucky. The Pacific region, including California, comes in second. But what's actually important here is the sectoral skew (seen below): California still reigns supreme for creating hugely profitable IT companies. 

"Putting it in relative perspective, the Pacific region's production of $100-million companies alone would, in standalone terms, be larger in $100-million company creation than are the U.S. Midwest or Mountain regions in all sectors," he writes. "That is fairly remarkable."