Effects of Learning the Long-Run Asset Pricing Model.

Job Market Paper.

This paper solves a new learning problem in a model with long-run risk, where both, the level and persistence of consumption growth are unobserved. We introduce a new methodology to quantify the effects of learning about parameter uncertainty and latent variables. The representative consumer chooses state variables that are sufficient statistics of the learning problems and, conditioning on her information set, forms posterior distributions of the states and future consumption growth. We present a novel numerical approach, which approximates the agent's continuation-value and solves a series of nested dynamic programming problems. The problems are nested by imposing that solutions for learning problems converge to the solution of the problem without learning as uncertainty disappears. Keeping preference parameters constant, maximum Sharpe ratios increase significantly from .07 in the benchmark case without learning to .45 in the learning economy.

References

Lars P. Hansen (co-chair)	Ruey S. Tsay (co-chair)
Homer J. Livingston Distinguished Service Professor	H.G.B. Alexander Professor of Econometrics & Statistics
Department of Economics, University of Chicago	Booth School of Business, University of Chicago
lhansen@uchicago.edu	Ruey.Tsay@chicagobooth.edu
+1 (773) 702-8170	+1 (773) 702-6750
Pietro Veronesi	Kenneth L. Judd
Roman Family Professor of Finance	Paul H. Bauer Senior Fellow
Booth School of Business, University of Chicago	Hoover Institution, Stanford University
pietro.veronesi@chicagobooth.edu	judd@hoover.stanford.edu
+1 (773) 702-6348	+1 (650) 723-5866

November 19th, 2009