Conditional Asset Pricing with Higher Moments

Diploma Thesis by Volker Ziemann
October 14, 2004

Abstract:

The Capital Asset Pricing Model (CAPM) is, despite its simplicity, still the most cited model in financial theory. However, empirically, it yields rather poor results in explaining cross-sectional returns of assets. The most successful extensions of the standard model often add factors that lack simple interpretations in terms of risk. In this study, instead of developing more factors, we will develop in detail the analysis (of dependencies of asset returns) with a single factor. We will concentrate on two main criticisms on the CAPM, questioning the hypothesis of normal distributed asset returns and the single-period character of the standard model. In the first step, we extend the model by taking higher moments into account, leading to risk premia for co-skewness and co-kurtosis. Second, we will allow these factors to vary over time in an autoregressive heteroscedastic context, using the new four-moment bivariate GARCH-in-mean model. This will lead us to the time-conditional four-moment Capital Asset Pricing Model.