The Cross-section of Long-run Expected Stock Returns
80 Pages Posted: 26 Feb 2021
Date Written: December 10, 2020
Abstract We predict cumulative stock returns over horizons from 1 month to 10 years using a tree-based machine learning approach. Cumulative stock returns are significantly predictable in the cross-section over all horizons. A hedge portfolio generates 250 bp/month at a 1 year horizon and 110 bp/month at a 10 year horizon. Individual stock returns are significantly predictable at all horizons in panel data. Cashflow and momentum related predictors are mostly important at shorter horizons while dividend yield and value related predictors are more important at longer horizons. By contrast, variables related to turnover and volatility are influential at all horizons.
Keywords: Long-Run Discount Rates; Return Predictability; Cross-Section of Returns; Machine Learning; Gradient Boosting; Variable Importance; SHAP Values
JEL Classification: G12, G17, C53
Suggested Citation: Suggested Citation