Question about Survival Bias~

[no_ming]

Hi, Mr. Harper, its me again, for the following question, the explanation that poor performers could reduce average correlation of returns. Does it mean the poor performers out of database have higher correlation of returns?

But from my point of view, the poor performer seems have lower correlation to the good performers more reasonable as if the poor performers is highly correlated to the good performers, the difference between good & poor performers should be small and its not necessary to be dropped out of the fund.

 

[David Harper CFA FRM]

Hi @no_ming That looks astute to me. I agree. Although (a) is textbook best, I think you could make an argument for (c). To my knowledge, Bodie never discusses any implication of survivorship bias other than its implication on reported returns (performance). Vaguely, I tend to agree with you that removing the poor performance, as a hypothesis, ought to increase (overstate) correlation! Another angle is that we do know from history that hedge fund correlations increase during periods of lower volatility; which is almost by definition since correlation is covariance divided by the product of standard deviations. So, there is an inherent (at least partial) contradiction between "reduce the average volatility" and "reduce the correlation of returns;" i.e., ceteris paribus (same covariance), the reduction of volatility ought to increase correlations. Good insight, thanks!