Forecasting stock returns with large dimensional factor models

We study equity premium out-of-sample predictability by extracting the information contained in a high number of macroeconomic predictors via large dimensional factor models. compare well-known model with static representation common components Generalized Dynamic Factor Model, which accounts for time series dependence components. Using statistical and economic evaluation criteria, we empirically show that Model helps predicting premium. Exploiting link between business cycle return predictability, find accurate predictions also combining rolling recursive forecasts real-time.