A new report by Vanguard found that exchange-traded funds that are successful in attracting assets may only do so because of the hypothetical performance data that is available on most ETFs.

As of March 31, 2012, more than $1.2 trillion was invested in 1,400 U.S.-listed ETFs, which sought to track more than 1,000 different indexes, according to "Joined at the hip: ETF and index development." Many of these launched using new indexes based on narrow market segments and alternative weighting methods that often lack live performance history, the paper said. Among the indexes created for use in ETFs, more than half included back-filled or back-tested data from before the date the indexes were first publicly available, and it is difficult for investors to discern which data are hypothetical and which are live.

The paper found that more than half of ETFs are launched with an index that has been in existence for less than six months. "Using an event-study analysis to look at the performance of indexes before and after ETF creation, we find that ETFs are most likely to be created with indexes that have performed well relative to the broad U.S. stock market before the inception date, but that such performance, on average, does not persist."