"Why is it our data on small business are so bad that we can't get a handle on straightforward questions like how many jobs small business produces?" laments Timothy Bates, a professor at Wayne State University, in Detroit. As it turns out, there is no easy answer.

Such numbers are simply hard to find. And the ones that are out there aren't particularly reliable. Bates tells the story of a fellow researcher who undertook a study using both Census Bureau data and Bureau of Labor Statistics (BLS) numbers. The researcher found that the two data sets disagreed by 20% on the number of self-employed people in the United States. "We're talking about the two major statistics-gathering organizations," grumbles Bates.

The situation is worsened, Bates continues, by dickering with classification criteria. Woman-owned businesses, for example, lose that designation when they grow big enough to register as C corporations, thus making the average woman-owned company look artificially small. The same is true for black-owned businesses.

Think that's absurd? It gets worse. A 1988 study showed that government data missed nearly all manufacturing companies with fewer than five employees. Meanwhile, the government required that all individuals who claimed a piddling $50 or more in self-employment revenues on Schedule C of their tax form be counted as "small businesses." The mind boggles.

What exactly is the problem? For one thing, there's little official support for a standardized national database that tracks such basics as company births, growth, and deaths. The Small Business Administration, which developed such a database in 1979, pulled the plug on it in 1991. "Since it was discontinued," says Bruce Kirchhoff of the New Jersey Institute of Technology, "there hasn't been much information around." What's worse, reports Kirchhoff, is that apparently no one bothered to save the old database. A year expired, and the machine erased it. Just like that, the nation's primary source of small-business information vanished forever.

The federal government has shown no eagerness to develop or promote alternative sources. Catherine Armington, formerly a consultant with the BLS, cites a prototype database built by the BLS that was based on annual reporting of unemployment-insurance tax payments for U.S. businesses from 1989 to 1994. "It's by far the best data on employment," claims Armington. Yet when a government researcher prepared an article based on it last year, she says, the bureau--unwilling to defend the "experimental" data publicly-- "squashed it" and has no plans to make any further use of the data.

All of that probably doesn't alarm many business owners, but it should. "Research has effects," says Bruce Phillips, data czar at the SBA's Office of Advocacy. "It filters its way down into public policy." He points out that many of the policy innovations that cleared the way for the recent entrepreneurial explosion--such as small-business-friendly clauses in the 1986 tax reform--came as the result of SBA-funded studies during the 1980s, when the agency's research budget was almost 10 times what it is now.

Unfortunately, data that might help further an innovative agenda may grow even scarcer. Congress withheld funding for SBA research for much of this year and toyed with the idea of wiping out the SBA's Office of Economic Research entirely. Ultimately, the research office may be killed for the same reason the SBA database was: because its custodians, according to Kirchhoff, "couldn't find any small-business owners or politicians who thought there was any value in it."

--Jerry Useem