Stock value is dependent on the market environment. Yet many stock valuation models have fixed, hidden biases that implicitly represent a forecast market scenario and prevent the model from being responsive to top-down investment information. The popular dividend discount model (DDM) is a case in point: The DDM provides a framework for efficient management of bottom-up investment information, but it has inherent biases that cause it consistently to favor high-yield, low price-earnings ratio stocks.
An approximate mathematical decomposition of the information coefficient of standard DDM valuations confirms the existence of the yield bias and reveals a second bias—a negative correlation between the model’s forecast components—that represents an internal inconsistency and affects performance. The analysis shows, however, that the biases in the model are independent of the underlying discounted cash flow framework. This independence provides an opportunity to create a scenario-dependent dividend discount model, while maintaining reliance on near-term analysts’ forecasts for relative valuation.
A scenario-dependent generalization of the DDM—the “conditional valuation” or CV-DDM—controls the observed biases. CV-DDM valuations may be explicitly conditioned to reflect an institution’s investment philosophy and available top-down investment information. The enhanced technology thus eliminates self-defeating inconsistencies, providing a tool for bridging the gap between top-down investment information and bottom-up analysts’ forecasts. A more realistic return structure may also lead to an increase in the level of information that can be derived from analysts’ forecasts.
