The Accuracy of Parsimonious Equity Valuation Models - Empirical tests of the Dividend Discount, Residual Income and Abnormal Earnings Growth model

In many contexts there is a need for parsimonious valuation models, i.e. technically less complex models that are based on few and readily available input variables. Previous research indicates that the accuracy of such models can be weak, and that the choice of a model specification involves striking a balance between the need for valuation accuracy and modelling simplicity. In this paper we investigate the accuracy of four well-known valuation models - the dividend discount (PVED) model, the residual income valuation (RIV) model and two versions of the abnormal earnings growth (AEG) model – using financial data from a Scandinavian (Denmark, Finland, Iceland, Norway and Sweden) capital market setting. Measuring the valuation accuracy in terms of both precision and spread, we find that the RIV model in general allows for the best parsimonious model specifications. Incorporating complexity adjustments (longer explicit forecast periods, bankruptcy risk adjustment of discount rates, and elimination of transitory income items) in our most parsimonious models, we find that the valuation accuracy of all models improve. RIV modelling still comes out as the best valuation approach, but the gap to PVED modelling decreases. Despite our complexity adjustments, the AEG models generate poor valuation results and basically cannot be used as valuation benchmarks for our sample of Scandinavian firms.