Full-Information Examinations of Long-Run Risks and Habit
60 Pages Posted: 1 Feb 2016 Last revised: 18 Oct 2018
Date Written: October 10, 2018
Many consumption-based models succeed in matching long lists of asset price moments. We show that the empirical fit is much worse in full-information estimations. We estimate a model with long-run risks, habit, and a residual using Bayesian methods. We find that long-run risks account for less than 25% of the variance of the price-dividend ratio. Habit's contribution is negligible. The residual is highly persistent, however. Filtered versions of our decomposition that focus on business-cycle frequencies find a much smaller residual. Business-cycle variation of the price-dividend ratio is 50% long-run growth, 20% long-run volatility, and 5% surplus consumption. These results are robust to the prior, including priors that assume long-run risks in consumption and highly persistent habit.
Keywords: Long run risks, Rare Disasters, Habit, Bayesian Estimation, Particle Filter, Time-Varying Beliefs, Time-Varying Preferences, Excess Volatility
JEL Classification: G10, G12, E21, E30, E44, C11, C15
Suggested Citation: Suggested Citation