Stock return predictability and conditional asset pricing models

I examine the statistical and economic significance of the evidence on return predictability and departures from conditional asset pricing models when the decision maker faces uncertainty about the model parameters and dynamics.; In the first chapter, I introduce a Bayesian econometric framework to examine conditional asset pricing models that formally incorporates into the test the researcher's uncertainty about the model's pricing ability and its dynamics. I use a time series framework and examine the zero alpha restriction imposed by the model. Uncertainty about the model's pricing ability is incorporated by specifying priors for the alphas while uncertainty about the model dynamics is addressed by modeling betas as stochastic functions of explanatory variables. Using this econometric framework, I present a statistical evaluation of the departures from a conditional version of the CAPM. I find that the value-weighted index is not conditionally efficient and the relative inefficiency of the index depends on the researcher's priors about the extent of model mispricing, the stochastic nature of beta, and the value of the Sharpe ratio of the index.; In the second chapter, I evaluate the economic significance of the evidence on predictability and conditional asset pricing models using the econometric framework developed in the first chapter. I examine the portfolio choice of a Bayesian investor who allocates funds among several risky assets in the presence of return predictability. In addition to the predictability evidence, the investor uses conditional asset pricing models to guide him in the portfolio selection decision. I also explore how the uncertainty about the beta dynamics affects the investor's optimal portfolio. Using a market index and a small capitalization or a value portfolio, I find that the sample evidence on predictability plays a major role in the portfolio allocation decision. However, the extent to which the evidence affects the optimal portfolio depends largely on the investor's beliefs about the possible source of return predictability, namely model mispricing and/or time variation in risk premia and betas. Finally, I show that the portfolio allocation decision is also affected by the investor's uncertainty about the beta risk dynamics.