Systematic use of dividend discount models has the potential to add value to the portfolio management process. Using consensus forecasts of earnings growth, the authors tested four valuation models ranging in complexity from a simple price-earnings ratio model to a three-period dividend discount model. Each model was used to rank 150 stocks from the S&P 400 into five portfolios, and in each case the model’s ranking was borne out by subsequent portfolio return. Furthermore, all the top-ranked portfolios outperformed the market average.

Portfolio returns improved considerably, however, as the complexity of the model used to rank them increased. The top-ranked portfolio from the three-period dividend discount model, for example, had an annual return 3.5 per cent greater than the return on the top-ranked portfolio from the P/E model.

Ability to discriminate between under and overperformance also increased with model complexity. The portfolios ranked best and worst by the P/E method differed in return by 22.26 per cent; this difference increased to 35.63 per cent when the three-period model was used for ranking. Adjustment for risk improved model performance, especially when used in conjunction with the more complex models.