Start-up Nation

If you think it's tough to take a census of small businesses, think for a moment about the difficulty of enumerating new small businesses. You'd like to do it in a timely fashion, when the information would be most useful to marketers, policymakers, and even prospective entrepreneurs themselves. You'd like your count to be complete, accurate, and detailed -- to tell you which parts of the country are hottest for start-ups right now, which industries are attracting the most entrepreneurs, and so on.

Good luck. Researchers of various stripes -- academic, governmental, even commercial -- have been working on this challenge for a while now, and their results have yielded only a succession of approximations. One fundamental problem: agreeing on what constitutes the birth of a company. Is it the opening of an office? The first sale? The first tax return? Different definitions lead to vastly different counts. A second difficulty: recording and tabulating any of those occurrences. The government records events like hiring (since the company now owes payroll tax) and sales (since the business now must file an income-tax return), but doesn't get around to reporting the data for a couple of years. Private organizations such as Dun & Bradstreet record the creation of a business when they have reason to do so -- when someone requests a credit report on the company, for example. Meanwhile, untold numbers of fledgling businesses fly under everybody's radar. The advent of E-commerce makes it even harder to spot businesses in their earliest stages. (See "And Can We Ever Really Know?" below.)

So how many businesses started up last year? Here's a guide to some of the wildly different approximations we're forced to rely on:

Government estimates. The U.S. Small Business Administration's Office of Advocacy updates census figures for companies with employees, by using state payroll-tax information gathered by the Department of Labor. The SBA's most recent estimate:

588,900 companies with employees started in 1999

Is 2000 or 2001 likely to look much different? "My guess is that the number of employer businesses is still growing but at a slower rate," says Robert Berney, chief economist for the Office of Advocacy.

Dun & Bradstreet estimates. Since government agencies gather data so slowly, plenty of people rely on private-sector number crunchers like D&B, which until last fall issued a week-by-week summary of business starts. What constituted a business start in the now discontinued summaries was a new entry in D&B's credit-and-marketing-information database for a company whose birth date was in the previous 36 months. "Entry to the D&B file generally coincides with the point at which a business begins to actively compete in the marketplace," according to the company.

Well, yes. D&B probably caught most companies that had established themselves, hired employees, and began recording substantial sales. But there must have been plenty of others too small or shaky to make it onto the company's records. D&B tallied only 151,016 business starts in 1999. The figures for 2000, through October, when the series was discontinued, show starts down 14.2%. If that rate held true through the end of the year, the number of new companies could be estimated at --

129,572 companies started in 2000

New Business USA. This Omaha-based company gathers state-by-state data from public agencies, including new-incorporation filings, new permits to do business, and so on, and then markets the data to businesses. New Business USA's latest estimate:

1,573,256 companies started in 2000

That estimate includes 910,361 home-based businesses and 662,895 commercial-site businesses, says New Business USA account executive Joe Addink.

And Can We Ever Really Know?

Talk to Jerome Katz about the problem of counting new companies and you will hear laughter. How can you even begin to find new companies in the age of the Internet? asks Katz, a professor of management at St. Louis University's John Cook School of Business.

"If you have a credit card, you can create a business in about two hours. You buy a dot-com name, locate it on a server, contact a wholesaler, have them ship the goods to a fulfillment center that handles billing, returns, etc. You can go from nothing to operational in two hours. But it isn't registered anywhere! It may not even have a telephone number.

"And if you want to push this, think about all those people on eBay. We used to say maybe one-quarter of self-employed people were self-employed part-time. They moonlighted. But what about the people on eBay who have 10, 15, 20 auctions going at once? They're pretty involved in their business -- and that's millions of additional people. But nobody's counting them.

"We've probably increased the number of self-employed people in America by 25%. It could be 50%. But we can't track it."

Please e-mail your comments to