Two essays on the value premium

This thesis includes two essays. Essay 1 concentrates on the use of conditional asset pricing models to speculate on the nature of the value premium. Previous findings seem to be strictly dependent upon the state variables used to define the "good" and "bad" states of the world. I investigate this issue in a conditional CAPM framework allowing for alternative sets of conditioning information. Contrary to earlier research, my results suggest no evidence in favor of a risky interpretation of the value factor. Moreover, some of the previous conclusions based on these conditional models, seem to be sample specific. Essay 2 represents a natural extension of this research finding related to the value premium and led me to investigate other empirical questions such as the presence of a clientele effect. If book-to-market is a risk factor, then it should on average affect all stocks the same in the cross-section. If it is not a risk factor, and instead it arises from behavioral effects, then there may exist a systematic "clientele effect" in which subsets of stocks are sensitive to B/M, and other are not. My findings confirm this conjecture.