Breaks and Trends in Factor Premia

This paper investigates structural breaks and regime-dependent premia in risk factors. We analyze the slope factor model (e.g., Fama and French, 2020; Smith and Timmermann, 2022), where characteristic loadings align with factor risk premia, and propose a predictive regression with time-varying coefficients to accommodate heterogeneous breaks and piecewise constant loadings. Our method captures time-varying trends in slope factor returns through detected breaks and regimedependent premia estimates. We establish the consistency of our time-varying parameter estimates and derive large-sample properties for the estimator. Using U.S. stock returns and numerous firm characteristics, we identify dynamic shifts in factor premia, especially during financial crises. Our results show that this method enhances investment performance for selected regime-dependent slope factor models. For instance, SUE (unexpected earnings) consistently predicts with high factor premia, while SP (sales-to-price) and ABR (abnormal returns) display regime-dependent patterns. Moreover, a regime-based out-of-sample factor timing strategy outperforms traditional buy-and-hold approaches for most slope factors, evidencing time-varying factor premia.