Risk Horizon and Expected Market Returns

The paper proposes an equilibrium asset pricing model that accounts of the incomplete information on returns distribution and investors' preferences. Only moments up to order four of unknown unconditional distribution can be observed, and the model does not impose that portfolio diversi fication or moments preference should hold. Using Chebyshev-type of inequalities, an intuitive risk measure (risk horizon) is introduced with reference to the speed of convergence of a security's mean return to its expectations. By an arbitrage argument, this risk measure is related to the horizon of treasury securities in a system of equations that allows the calibration of the model parameters using term structure information. In particular, the expected return on the market portfolio can be endogenously estimated inside this system. The model calibration on U.S. market data provides plausible parameters estimates and interesting cyclical patterns in the time series of the expected return. The empirical relevance of these estimates is examined with tests of statistical and economic predictive ability for stock excess returns. The results provide signi ficant evidence on the added value of the estimates when compared to popular predictors found in the literature (see a.o. Lettau and Ludvigson, 2001; Rapach and Wohar, 2006; Goyal and Welch, 2008).