A General Model of Dynamic Asset Allocation with Incomplete Information and Learning
48 Pages Posted: 28 Feb 2005 Last revised: 20 Jan 2011
Date Written: February 8, 2006
This paper develops a general and flexible multivariate discrete-time model of dynamic asset allocation with incomplete information and learning in the case of timevarying investment opportunity sets. The state variables are described by a vector autoregression and the investor is assumed to have normally distributed and possibly correlated priors on the values of the state variables. We apply the model to an investor who learns about the mean returns on the market and Fama-French SMB and HML portfolios when the size and value premia disappear (possibly stochastically) over time due to trading by other investors. The portfolio implications are shown to be substantial.
Keywords: Portfolio choice, learning, VAR, predictability, hedging demands
JEL Classification: G11
Suggested Citation: Suggested Citation