Abstract:
Consistent with neoclassical models with investment lags, we find that a bottom-up measure of aggregate investment plans, namely, aggregate expected investment growth, negatively predicts future stock market returns. with an adjusted in-sample R2 of 18.5% and an out-of-sample R2 of 16.3% at the 1-year horizon. The return predictive power is robust after controlling for popular macroeconomic return predictors, in subsample periods, as well as in other G7 countries. Further analyses suggest that the predictive ability of aggregate expected investment growth is more likely to be driven by the time-varying risk premium than by behavioral biases such as extrapolative expectations.