12. Survivorship bias

The most common example used to illustrate the effects of the survivorship bias in investing is the evaluation of fund performance, when fund returns are calculated using only data from funds that are still in existence.

Naturally, this means that the returns of funds which, for whatever reason, had to shut up shop, are not included in the calculation.

This results in overall returns being skewed to the upside, simply because the funds that had to shut down probably posted negative returns, which would have lowered the overall performance number if they were included in the data set.

Consider an example where an investor wants to compare fund performance by investment strategy over the past five years. Since databases tend to stop tracking the performance of funds that no longer exist, it follows that these data sets, when used for the comparison, will lead to a conclusion that is skewed by the survivorship bias.

As returns for each investment strategy (value/growth/Emerging Markets, etc) are calculated from the pool of surviving funds in that particular strategy, the final, overall performance number will be skewed higher. 

This makes the overall number inaccurate, since it neglects to account for the funds who invested according to the strategy, but made losses that were large enough to push them into nonexistence.

Consequently, the overall return for that particular investment strategy is only reflective of the performance of those lucky and skillful enough to still be alive during the 5 year comparison period.

It isn’t reflective of the viability of the investment strategy as a whole.

Furthermore, the survivorship bias also affects the comparison of investment strategies on a more macro level. That is, the five year comparison period only reflects the performance of each strategy in relation to market conditions during those five years.

If growth stocks were rallying strongly during those years while value stocks languished, investment strategies that focused on the former will naturally outperform the latter. In this sense, growth stocks can be considered the “survivors” in the data set.

However, this might not be the case if the comparison period were to be extended to say, 15 years.

Since markets are not always conducive to a specific investment strategy, it is extremely important to calculate performance over a period of time that is long enough to encompass a variety of market cycles and conditions. Only then can we get an accurate picture of a strategy’s true performance.

Ultimately, fund managers can only perform as well as the current set of market conditions allows them to. It follows that as market conditions change, the fortunes of fund managers change along with them.

This means that when trying to avoid falling into the trap of the survivorship bias, we need to account for not just individual fund performance within an investment strategy, but also the performance of the strategy in the context of prevalent market conditions.