A Factor Model of Company Relative Valuation

Abstract

Accurate company valuation is the starting point of value investing and corporate decisions. This paper proposes a statistical factor model to generate company valuation comparison across a large universe. The model scales the market value of a company by its book capital to generate a cross-sectionally comparable relative value target, constructs valuation factors by combining several descriptors from a similar category to increase coverage and reduce multicollinearity, and links industry classification and the valuation factors to the company relative value via a cross-sectional contemporaneous regression at each date. Historical analysis on U.S. publicly traded companies shows that the factor model explains a large proportion of the cross-sectional variation of company relative value and experiences little out-of-sample degeneration. The regression residual represents temporary company misvaluation, and can be exploited by both outside investors as attractive investment opportunities and internal management for market timing of corporate decisions.