Table of Contents:
Modern financial theory acknowledges that decision makers do not perfectly know the stochastic process of financial figures, in particular corporate cash flows. In other words, there is imperfect information in the sense of incomplete information. Incomplete information in the context of this thesis is modeled with the help of an unobservable underlying regime model, i.e., corporate dividends can assume several regimes. It is clear that incomplete information regarding dividends should be reflected in companies’ stock prices. It is not so clear how incomplete information translates into risk premia. Does incomplete information as a second source of risk in addition to “normal” stock price fluctuations (first source of risk) (i) automatically increase overall risk and, hence, call for an adequate compensation, or (ii) can incomplete information and “normal” stock price risk interact in a way that overall risk is reduced and risk premia decrease? Veronesi (2000) has shown that incomplete information leads to an unintuitive asset pricing outcome: the incomplete information risk premium (two sources of risk) is below its complete information counterpart (one source of risk) for typical values of risk aversion parameters. This result of Veronesi (2000) has been derived in a narrow model framework: CRRA utility, regimes in expectations only, and only one asset. This thesis analyzes the effect of incomplete information on asset prices in a more general framework (various types of utility functions, cash flow functions, regimes in both standard deviations and expectations, and several risky assets). The results can be summarized as follows: (i) Incomplete information exerts a substantial influence on risk premia for all models considered in this thesis - CARA and CRRA utility functions, richer class of regime processes, various forms of cash flow model, and more than one risky asset - as the analytical analyses demonstrate. Core of all pricing approaches is the covariance between stochastic discount factor and asset return. Incomplete information fundamentally alters this covariance. (ii) The numerical analyses illustrate that the theoretical pricing results are also relevant from an economic point of view: incomplete information risk premia are significantly different from complete information risk premia and the different model versions also translate into substantially different risk premia. By and large, the results found in Veronesi (2000) still hold in the more general setting that has been considered in this thesis.