Do outliers matter? The predictive ability of average skewness on market returns using robust skewness measures

摘要

We used robust skewness measures to revisit a recent theory that the average asymmetry (measured by the average monthly skewness values across firms) can negatively predict future market returns. Skewness measures employed in previous studies are moment-based which are normally sensitive to outliers of returns. We thus consider a quantile-based robust skewness measure and find that the predictive power of the average skewness to market returns no longer exists. Instead, we find a negative relation between the average expected (or ex-ante) skewness and market returns, suggesting that investors' average expectation on skewness can negatively predict market returns.