A new method for forming asset pricing factors from firm characteristics

Commonly used asset pricing models do not successfully account for both time-series and cross-sectional variations of asset returns. In this article, we propose a new method for forming pricing factors that are intended to capture the time-series as well as the cross-sectional variations. The new pricing factors are constructed by utilizing a set of basis assets that are associated with firm characteristics. Compared with popular extant asset pricing models, empirical results show that the new model can parsimoniously and successfully account for both time-series and cross-sectional variations of asset returns and significantly improve model performances in terms of various measures: the Jensen's , root mean squared alpha, time series regression R-2, cross-sectional regression R-2 and the HJ-distance measure.