Abstract

We compare the performance of macroeconomic factor models and Fama-French-type (fundamental) factor models in explaining and predicting stock returns at different frequencies through spectral analysis. Our analysis considers both long-term variation and business cycle variation. The results show that fundamental factors are more suitable than macroeconomic factors to explain the cross-section of equity returns in terms of R2 at multiple frequencies. The risk premia vary in magnitude over time and horizon, especially for macroeconomic factors. In terms of predictability, macroeconomic factors perform similarly to or better than statistical (PCA) factors based on 68 macroeconomic and financial variables.