You are here: Knowledge Center Reading Room

Survivorship Bias

E-mail Print PDF

Survivorship bias refers to the phenomena whereby the past records of existing mutual funds are examined to determine various trends.  The problem lies in the fact that you are only examining the past records of currently existing funds — funds which ceased existence in the past are not included in your data.  This tends to cause one to (falsely) conclude that the average mutual fund has performed better than is actually the case (because the funds which cease to exist and are therefore removed from the sample universe tend to be the poor performers).  Due to survivorship bias, it is actually possible to (falsely) conclude that the average dollar invested in mutual funds performed better than average!  Also, see mutual fund persistence.

Mark M. Carhart, Jennifer N. Carpenter, Anthony W. Lynch, and David K. Musto, "Mutual Fund Survivorship," Review of Financial Studies, Issue 5 2002, pp. 1439-1463 (371kb).  Also here.  This paper shows that mutual fund survivorship bias is about 0.07% over one year periods, but about one percent for periods of 15 years of longer (in other words, if one studied mutual fund performance over a 15 year period, the annual return of the average fund dollar would be overstated by about one percentage point annually).  An excellent comprehensive analysis of survivorship issues in mutual fund performance studies.

Stephen J. Brown, William N. Goetzmann, and Stephen A. Ross, "Survival," Journal of Finance,  July 1995, pp. 853-873 (1.9mb). This paper concludes that survivorship bias increases "the probability of false rejection of temporal independence."  In other words, survivorship bias tends to cause one to conclude that phenomena such as mutual fund persistence exists, when it may not.

Back to Reading Room