In Full-Information Estimates, Long-Run Risks Explain at Most a Quarter of P/D Variance, and Habit Explains even Less

We develop a model in which asset prices depend on long run growth, long run volatility, habit, and a persistent residual. We estimate the model using Bayesian methods which account for the entire likelihood of the data on consumption growth, dividend growth, and the price-dividend ratio. The residual is dominant, accounting for 60% of the variance of the price-dividend ratio. Moreover, the filtered residual tracks most of the recognizable features of the U.S. stock market, such as the late 1990's boom and bust. Long run volatility also plays a significant role, accounting for 30% of the variance, but it contributes primarily in rare crises. Long run growth and habit contribute 15% and 1%. These results show that while long run risks play a non negligible role, something else is driving the bulk of stock market fluctuations. Estimations under alternative priors show that the low correlations between asset prices and conditional moments of consumption growth underlie the large role for the residual.