Cemetery Crosses Graveyard

Survivorship Bias – Things Are Not As Good As They Look

Cemetery Crosses Graveyard

"Survivorship bias" is one of the many reasons that stock pickers' returns look better than they actually are. Survivorship bias is when mutual fund managers tout their fund's performance based on comparisons with an “average” mutual fund. This average is calculated from a list of funds that have survived during a particular period. Funds that did not survive the period are not included in the calculation. According to the Center for Research on Securities Prices (CRSP) at the University of Chicago, if only data from surviving funds is considered, the growth of a dollar for the surviving funds appears to be 19% better since 1962. If only "live growth and income funds" are considered over this period, $100 appears to grow to about $2,500. However, the only way to properly account for all active managers is to include those mutual funds that did not survive. When taking these dead funds into account, CRSP found that the average stock picker’s $100 investment grew to only about $2,100.

The Center for Research on Securities Prices (CRSP) at the University of Chicago has the only complete database of both live and dead mutual funds.

Mark M. Carhart, currently Co-Head of Quantitative Research, Goldman Sachs Asset Management, New York, developed this unique database for his 1995 Ph.D. dissertation at the University of Chicago Graduate School of Business. In his dissertation, Survivor Bias and Persistence in Mutual Fund Performance, he noted that the explosion in new mutual funds has been "...accompanied by a steady disappearance of many other funds through merger, liquidation, and other means...this data is not reported by mutual fund data services or financial periodicals and in most cases is electronically purged from current databases. This imposes a selection bias on the mutual fund data available to researchers: only survivors are included.."

In estimating the performance of an equal-weight index of equity mutual funds, Dr. Carhart found that analyzing only surviving funds biases performance upward by about one percent per year.

The CRSP database includes approximately 39,000 mutual funds from 1961 to 2012. In that time, approximately 13,000 of the 39,000 mutual funds died. That means that more than 33% percent of mutual funds data is not included in the average returns of active managers. The active managers who run them happily bury most of the data about dead funds. Is it possible that the 13,000 dead mutual funds had high returns for their investors?

For advanced students, print out a more detailed fifty-one page analysis of Survivorship Bias, by Mark Carhart: Mutual Fund Survivorship. Also see this abstract by Elton, Gruber and Blake and this article by Horst, et al.

The financial press goes to great lengths to inform the public about active managers with good luck, e.g., "Last Year's Top Ten Mutual Funds." This kind of media reporting provides no data about those managers who lose money taking chances in the market, then shutting their doors and erasing their bad returns from the record. Well, we have bad news for those mutual fund managers: we are exhuming their results. We paid CRSP $1,000 (it is expensive to dig up old corpses) to prepare a list of the top 200 worst performing dead mutual funds going back to 1961. To our surprise, this had never been done before. Utilizing more recent data from Morningstar Direct, the table below shows the top twenty of the worst performing dead mutual funds: