Summary in English

In this thesis, I provide new insights in the field of asset pricing and portfolio optimization by investigating the role of information and factor decompositions. In particular, I investigate the effect of disentangling specific risk factors as used in the earlier asset pricing and portfolio allocation literature into their different sub-components. In Chapter 2, I propose an extension to the two-beta model of Campbell and Vuolteenaho (2004) and the two-beta model of Ang, Chen, and Xing (2006) to account for investor preferences in two respects. First, based on the assumption that investors are loss-averse, I distinguish between co-variation of stock returns with market returns in up and down markets, respectively. Second, I distinguish different investor reactions to market cash flow news and market discount rate news, particularly for long-term and short-term investors. Based on decomposing the risk factor space in all these different directions, I propose a four-factor asset pricing model. The new model distinguishes the co-variation of stock returns with cash flow and discount rate news in up and down markets. By disaggregating the risk factors into these different components rather than leaving them in their aggregated form as in Campbell and Vuolteenaho (2004) or Ang, Chen, and Xing (2006), I find that downside cash flow and discount rate betas typically carry the largest premia in the cross section. Downside cash flow risk is priced most consistently across different samples, periods, and return decomposition methods. It is also the only component of beta with significant out-of-sample predictive ability. I also find that for small stocks, mainly the downside risk components carry the premia. For the larger companies, the priced components of risk become more symmetric (both upside and downside) and are cash flow related. In Chapter 3, I examine whether investors can benefit from return predictability in their asset allocation decisions by decomposing the variation in commonly used state variables into long-term and short-term components. Based on a variety of filtering techniques and a semiparametric approach of Ait-Sahalia and Brandt (2001) to model the dependence of asset allocation decisions on state variables, I find that for short term investors the short-term component in state variables matters more for asset allocation decisions. This result, is obtained while accounting for short-sale restrictions into the regular GMM moment conditions by using the appropriate Lagrange-Kuhn-Tucker multipliers. For longer horizons, the long-term components