WSJ Interactive, April 4, 1997

Dead-and-Buried Funds Would Alter 'Averages' -- If Included


The average stock fund has trailed the Standard & Poor's 500-stock index benchmark in recent years, prompting plenty of grousing. But did funds actually do even worse than the usual measures of "average" performance? Some fund researchers and finance professors believe the answer is yes. To bolster their case, they've been prowling around the mutual-fund graveyard, gathering the names and vital statistics of the many mutual funds that have been merged into others or liquidated out of existence. Lousy performance is a prime reason that funds are killed off, they explain. So the usual industry averages, figured without deceased funds, may be somewhat exaggerated and "not representative of what investors actually got," says Don Phillips, president of Morningstar Inc., the Chicago fund-data company. However, the graveyard hasn't yielded all its secrets -- or conclusive findings. Some industry data and academic research show that "average" fund results for particular periods have been boosted significantly by the absence of dead funds. But other numbers suggest that there was no significant upward distortion of "average" fund results over the past decade. If such distortion exists, it would be yet another reason for investors to favor unmanaged index funds that simply track the S? 500 or another benchmark, say index-fund advocates such as Vanguard Group Chairman John C. Bogle. While the results aren't necessarily clear, it is obvious that lots of funds disappear each year. In 1996, for example, 242, or 5%, of the 4,555 stock funds tracked by Lipper Analytical Services Inc., of Summit, N.J., were merged or liquidated. As a result of such yearly disappearances, some of Lipper's fund-performance averages, like wine and cheese, have improved markedly with age. Consider 1986: A decade ago, Lipper reported that 568 diversified U.S. stock funds had delivered an average 1986 return of 13.39%. Today, Lipper, which supplied the data for this section, puts the average 1986 return at 14.65%. Why the improvement? The new number is based on the performance of only the 434 funds from the 1986 group that are still in business today. If you look at results for dead funds as well as survivors, says Princeton University economics professor Burton Malkiel, "you show that mutual-fund returns are a lot lower than most of the published statistics." Mr. Malkiel did just such a study, using Lipper data, for the decade from 1982 through 1991. His conclusion, published in the Journal of Finance: While funds around for the whole period gained an average 17.09% a year, the average return from all funds in existence each year was a lower 15.69%. The adjustment makes the average fund look even worse compared with the S? 500's average annual gain of 17.52%. "The main lesson is that active management has not done super-well," says Mr. Malkiel, the author of "A Random Walk Down Wall Street." (That's not a surprising conclusion from Mr. Malkiel, a self-described "believer in index funds" and also a board member at Vanguard, a big index-fund seller.) But the argument that average fund returns are inflated by survivorship bias isn't airtight, particularly for more recent periods. Using an admittedly crude calculation based on Lipper numbers, Mr. Bogle of Vanguard estimates that survivorship bias added only three-tenths of a percentage point to the average annual stock fund return over the past decade. That's not much when the average annual return was over 13%. However, in looking at the 15 years through 1992, Mr. Bogle found an upward bias of just under one percentage point a year. Underscoring the issue's ambiguity, officials at Morningstar got a surprise this week when they looked at preliminary results of a campaign to eliminate survivorship bias from their database. Over the past decade, it appears that adding dead funds back in with survivors wouldn't change the average U.S stock fund performance at all. That jibes with data from Micropal Inc., a fund-research firm based in Britain that keeps some performance records on U.S. funds both with and without an adjustment for survivorship bias. Performance differences are slight over the past decade, says David Masters, a senior funds analyst, and some stock-fund categories actually performed better when nonsurviving funds are included. In any event, the survivorship-bias debate highlights the approximate nature of some numbers that investors may assume to be precise. Fund ads always warn that past performance is no guarantee of future results. When it comes to the "average" stock fund or "average" small-stock fund, it seems, past performance doesn't even guarantee an exact indication of what happened in the past. Distortions in fund-performance averages could become more pronounced in future years. The stock market could see some tough times, Mr. Bogle of Vanguard says, and an extended downturn might lead to lots more funds being merged or liquidated. He figures that survivorship bias was far greater in the decade through 1981, a period that included the painful 1973-1974 bear market. Meanwhile, data gatherers at Lipper and Morningstar fret about a different threat to the accuracy of their fund-performance averages -- what Morningstar's Mr. Phillips calls "creation bias." Regulators are allowing some new funds to count, as part of their track record, performance achieved when the funds existed as limited partnerships or other nonfund vehicles. Fund companies will only want to cite such records if they are good, the data gatherers say. So including that "preinception" performance in fund averages will tend to pump up results. Mr. Phillips says Morningstar will probably add such performance to its database but highlight it in some way as hypothetical. Lipper won't put those early results in its standard database. "To me, this is a much bigger issue than any sort of survivor bias," President A. Michael Lipper says.

Copyright (c) 1997 Dow Jones & Company, Inc. All Rights Reserved.


  1. What is the survivor bias in mutual fund performance and why does it affect average returns?
  2. How would you correct for survivor bias in measuring mutual fund performance?
  3. Can you think of another example of how survivorship bias can affect the perceived performance of an investment strategy?