Average Skewness Matters!

Abstract

This paper investigates using the average skewness, which is defined as the average of the monthly skewness values across firms, to predict future market return. Although the empirical evidence is not conclusive about the predictive ability of the average volatility, we show that in contrast, the average skewness performs very well at predicting the future market return when it is introduced in conjunction with the current market return. This result holds for several alterations of the main specification: the average skewness can be computed as the value-weighted or equal-weighted average of firms' skewness. The result holds after controlling for the size or liquidity of the firms or for the current business cycle conditions, and it also holds when the skewness is defined as the cross-section skewness of the monthly firm's returns or when it is filtered for the market return (idiosyncratic skewness). We also find that the average skewness compares favorably with the other usual suspects at predicting subsequent market return. The average skewness generates better out-of-sample performances in an allocation strategy based on market return predictions.