closer look

Morningstar Takes a Closer Look at Survivorship Bias

closer look

One of the problems with commonly cited mutual fund returns data is that it ignores the issue of survivorship bias, the tendency for mutual funds with poor performance to be merged into other funds or shut down. Specifically, if we look at the ten-year average returns of all domestic large cap value equity funds as of 12/31/2013, we are limited to funds that survived to the end of the period. Clearly, survivorship bias substantially overstates the average return of all investors who started at the beginning of the period. Furthermore, for those investors who chose the low-cost indexing route, it understates how well they did relative to their peers. Morningstar recently published an article that sheds light on this important topic. The author, Michael Rawson, points out that not only is survivorship bias a concern, but so is look-ahead bias which results from funds style drifting into a particular category and then being compared to funds that were in that category the entire time.

Working through the particular example in the article will help us understand the impact of these biases. We can start with the average return of all large blend funds at 8.9% for the 20-year period ending 3/31/2014. Since an S&P 500 ETF (SPY) got 9.4%, this is nothing to write home about, but it gets worse because this number excludes the performance of funds that did not survive the period, and it includes the performance of funds that were not in the large blend category at the beginning of the period. If we adjust for both of these issues, the average return drops down to 8.1%, and the ranking of SPY jumps from the top 24% to the top 11%.

Rawson notes that the survivorship rate of index fund share classes in the large blend category was substantially higher than the rate for actively managed funds (55% vs 34%). The index funds that did close tended to have higher-than-average costs, so smart investors would have dismissed them.

The other important predictive variables for non-surviving funds are fund assets and performance relative to peers. Funds that do not gather a minimum level of assets find it impossible to compete against other funds that can take advantage of economies of scale. Since most investors focus on short time frames such as 1, 3, and 5 years, funds that have a period of subpar performance are likely to suffer withdrawals until they realize that the most sensible economic decision is to shut down the fund.

In looking at returns over the last ten years, Rawson found that survivorship bias has a bigger impact on growth funds. For large cap stocks, the impact on growth funds was double the impact on value funds. From what we have seen, this makes a great deal of sense. Since growth funds tend to hold companies that are held in high regard, investors in these funds have a natural expectation that they will outperform the rest of the market. When this does not happen, they express their disappointment by withdrawing their money. Only 42% of the large growth funds survived to the end of the ten-year period.

The lesson we should draw from all of this is clearly spelled out in Rawson’s conclusion:

“This exercise serves as a helpful context for investors when presented with fund company marketing material. Naturally, fund companies are going to pitch the funds with the best performance, hide the funds with poor performance, and not even mention the fact that some of their funds no longer exist.”

We could not agree more.