Timeless Insights for Quant Researchers: An Oldie but a Goodie!

This paper titled “… and the Cross-Section of Expected Returns” by Campbell R. Harvey, Yan Liu, and Heqing Zhu looks at the many studies trying to explain the differences in expected returns of assets. The authors point out that there are now hundreds of factors being tested, and using the usual method to check if a factor is significant, like a t-ratio above 2.0, is no longer reliable. This is because so many factors have been tested that it’s likely some results are just due to chance.

To fix this issue, the authors propose a new method for testing factors that considers the fact that multiple tests have been done. They suggest using a higher standard, like a t-ratio above 3.0, to determine if a new factor is truly significant. They also look at how the standard for significance has changed since the first tests in 1967 and predict how it might change over the next 20 years if the current pace of research continues.

By analyzing over 300 research papers, the authors show that many of the factors published in financial studies might not actually be as meaningful as they seem. They suggest that most findings in this area are probably false and that researchers should use tougher criteria in the future to avoid false discoveries.