Asymmetric Relation Between Firm-Level Characteristics and Returns

Abstract

This paper applies quantile regression to reassess the relationship between firm characteristics and future stock returns across the return distribution. Unlike traditional OLS methods, this approach captures heterogeneity and tail-specific dynamics. We show that characteristics such as beta, size, illiquidity and MAX exhibit asymmetry (i.e., sign reversals) across quantiles and that their correlation with return volatility can predict the direction of this asymmetry. The methodology improves out-of-sample forecasting relative to classic Fama-MacBeth regressions, especially for extreme returns and certain firm types. Our findings can inform more targeted investment strategies and highlight the importance of accounting for heterogeneity in cross-sectional analysis.