Mutual funds don t lack for reporting requirements. Ask anyone who has struggled to un-wedge a bulky prospectus from his mailbox. An army of math-checkers fund company compliance officers, independent accountants and government auditors keep the numbers that fill those reports trustworthy. If a fund you invest in says it made 20% year-to-date, you can believe it.

Don t be as confident about the long-term returns boasted in advertisements. A new study makes clear how misleading they can be.

Investors have long known about incubator funds and how they distort returns through a phenomenon known as survivorship bias. A mutual fund family wishing to launch a new fund may quietly nurture several at a time. The funds that beat market averages are eventually marketed while ones that do poorly get closed. Like dead men, abandoned incubator funds tell no tales.

The result is an upward bias in returns for funds that made it out of incubation, and in average returns for fund families that incubate, as well as for the broad mutual fund universe. Past returns, more than any other factor, attract investor dollars, studied have long shown. And the number of dollars under management decides the profits of mutual fund families.

Government regulators frown on fund incubation but fail to prevent it. In response to a citizen query in 1997, the Securities and Exchange Commission stated that incubator fund performance should not be included in a mutual fund s prospectus in the absence of extremely clear disclosure explaining the sponsor s purpose in establishing the incubator fund. But it defined incubator funds as ones that aren t registered with the SEC and that exist for the purpose of generating performance records. Fund families can side-step the definition by registering incubator funds but keeping mum about them--not sending details to Morningstar and other data-reporting companies, and relying on employees for seed money. They can also launch a fund in each of several different asset categories, instead of several funds in the same category, to reduce the appearance of purposely launching more funds than they need.

Richard Evans, a professor at the University of Virginia s Darden School of Business, studied the effects of incubator funds over the decade ended 2005 and reported his findings in a paper scheduled for publication early next year in the Journal of Finance. He found that 23% of new funds were incubated, and that these outperformed non-incubated funds by an average of 3.5 percentage points a year, adjusted for risk. After the incubator funds were hatched to the public, however, the performance edge disappeared. This finding stands opposed to a longtime claim of the mutual fund industry that it uses the incubation process to identify promising managers and strategies. If they were indeed promising, the handsome returns ought to have continued. Moreover, fund families that incubated turned managers over more frequently that families that didn t, hardly suggesting that incubating families were cultivating long-term winners.

Three more findings from Evans s paper: First, fund families understandably prefer to incubate within fund categories where their existing offerings aren t attracting much new money. Second, as expected, investors chase after the higher returns of incubated funds with their deposits, oblivious to the tendency of the outperformance to fizzle. In other words, the process achieves its apparent aim. Third, incubation is most common among fund families that sell primarily through brokers.

That last finding should cast further doubt on the old broker argument that investors should choose funds based on performance and not sales charges.