The world is awash in bad, murky, misleading, and misused statistics, and nothing is going to change that. (Of course, the world is also awash in plenty of awesome statistics.) What we can try to do is point out when shoddy statistics affect public policy debates.
Today, the Senate is debating the Jumpstart Our Business Startups (JOBS) Act and appears set to vote soon. Plenty of arguments have been in made in favor, against, and for amending the legislation, but that's not my topic here. The JOBS bill is part and parcel of a long-running debate about the IPO market in the United States and why IPOs have not returned to the levels of the 1990s. One of the statistics making the rounds on Twitter, the blogosphere, and in various reports recently is that "92 percent of job growth occurs after a company's initial public offering." The ultimate source is a study done by IHS Global Insight for the National Venture Capital Association and promulgated most prominently in last autumn's report, "Rebuilding the IPO On-Ramp," by Treasury's IPO Task Force. (Some of the JOBS bill proposals come from that report.)
In that report, the relevant chart is this:
Accordingly, measures taken to boost the number of IPOs will spur job creation.
This is bad, albeit seductive, statistics. The "92 percent" assertion horribly conflates causality--by attributing (implicitly by some, explicitly by others) job growth to the fact of going public, purveyors of this number mix together all sorts of causes and effects. If it was the case that IPOs do in fact cause job creation then, as my colleague Paul Kedrosky has observed, we could easily solve unemployment by--presto!--having every single private company go public. Job growth after a company has gone public occurs for all sorts of reasons, very few of which may be related to the fact of having gone public.
Job growth at a specific company can also occur in different ways--either organically or through acquisitions. Organically, it counts as net employment growth for the economy. Via acquisitions, it doesn't, just a transfer from one company into another. Ideally, a merger or acquisition will end up creating value and thus more jobs in the long run, but initially it is counted statistically as the same number of jobs, not a net addition. Thus, citing a public company's job growth as net job creation for the economy as a whole is not entirely accurate.
Finally, the decennial trend in the chart above should offer a clue to its purveyors: post-IPO employment growth has been falling and pre-IPO employment growth rising. There are obviously many factors responsible for this, including the changing age at which companies go public, which went from 7 in the 1980s, to 8 during most of the 1990s, falling to 5 in 1999-2000, and jumping up to 12 years from 2001 to 2003. Older (and larger) companies will, predictably, be larger at the time of going public. It doesn't mean that the subsequent IPO is somehow less causal with regard to job creation, just that the firms are at different stages of growth.
The IPO market is determined by many things--regulation, market conditions, technological change, etc. Jay Ritter and colleagues have recently shown that the fall in IPO volume is due more to changing market structures and falling profitability among small companies rather than regulatory burdens, which are the most frequent scapegoats. Such changes are often unnoticed (or ignored) by the "92 percent" purveyors.
Yes, going public is expected to allow a company greater access to financing for expansion and innovation that, ideally, will create jobs. Boiling all of the underlying factors and changes into one number, however, and tangling all sorts of causal chains in the process, helps no one, least of all entrepreneurs.