Wednesday, February 18, 2009

R-squared and return predictability.

The R-squared of a regression of mutual fund returns on various measures in the so-called 4 factor model is a measure of how much of the fund's returns can be explained by the model. The 4 factor model is pretty close to the "state of the art" for return measurement. A lower R-squared implies that the fund's returns are less well explained by the 4 factor model.

A recent working paper argues that this R-squared has useful forecasting ability. Specifically that lagged R-squared is negatively correlated with future fund alphas. Even more surprising is that the negative relation also extends to information ratios which is commonly defined as alpha divided by the funds idiosyncratic risk. This is important because you could boost alpha by just sacrificing diversification. The information ratio result suggests that this is not the case.

The authors interpret their findings as evidence of "selectivity or active management".

Interesting stuff. Especially after I just finished teaching my MBAs that idiosyncratic risk is not priced. Hmmm the tangled web we weave...

The paper in question is...

"Mutual Fund's R2 as Predictor of Performance" Free Download

YAKOV AMIHUD, New York University - Stern School of Business

RUSLAN GOYENKO, McGill University - Faculty of Management


and appears on SSRN here