Firm Characteristics, Industry, Horizon and Time Effects, in the Cross-Section of Expected Stock Returns

We construct a panel data model to explain the cross-section of individual stock returns, using monthly data for 1,880 large US firms for 1985-2005. Model specification is geared towards multiple explanatory variables, poolability across industries, alternative forecast horizons, and the effects of unobserved heterogeneity among firms. We find that combining multiple firm characteristics increases the predictive power. High expected returns are mostly related to size, cashflow-to-price and turnover, and somewhat to earnings revisions and momentum. Diversified portfolios sorted on expected returns have moderate risk exposures and generate significant risk-adjusted returns over all horizons. Longer forecasting horizons drastically reduce portfolio turnover and hence lower costs