Industry Signals and the Dynamics of Stock Market Predictability

Abstract

ABSTRACT We investigate whether industry portfolios contain information that forecasts aggregate stock market returns. Earlier research reports evidence of their usefulness as predictors, yet later studies find weaker and unstable effects. Using updated U.S. data, we confirm that industry portfolios offer little forecasting power when market excess returns are modelled directly. We then evaluate two component‐based designs: a return‐component decomposition and a sum‐of‐the‐parts framework. Both generally improve out‐of‐sample performance, with the sum‐of‐the‐parts forecasts producing the strongest results. This pattern is consistent with the gradual diffusion of information across industries and the aggregate market, while showing that the findings are not confined to one decomposition method. We extend the analysis with forecast combination methods and trading strategy evaluations, showing that component‐based industry forecasts can improve statistical performance and generate economic gains.