Equity Valuation Employing the Ideal versus Ad Hoc Terminal Value Expressions*


  • *

    Accepted by Jerry Feltham. This paper was presented at the 2000 Contemporary Accounting Research Conference, generously supported by the CGA-Canada Research Foundation, the Canadian Institute of Chartered Accountants, the Society of Management Accountants of Canada, the Certified General Accountants of British Columbia, the Certified Management Accountants Society of British Columbia, and the Institute of Chartered Accountants of British Columbia. We would like to thank workshop participants at the 2000 American Accounting Association meetings; 2000 Canadian Academic Accounting Association Conference; 2000 Contemporary Accounting Research Conference; 2000 European Accounting Association Conference; HEC, Laval; University of Queensland, University of Technology-Sydney; and University of Waterloo for their comments.

    Special thanks are extended to Sati Bandyopadhyay, Joy Begley, Brian Bushee, Peter Clarkson, Steve Fortin, Kin Lo, Russell Lundholm (the discussant), Pat O'Brien, Terry O'Keefe, Steve Penman, Ranjini Sivakumar, Theodore Sougiannis, Ken Vetzal, and especially Jerry Feltham (the editor) for their helpful comments and suggestions on earlier versions of the paper; Kendrick Fiorito and Mort Siegel at Value Line for their advice on the project; Nick Favron for programming assistance; and Daniel Roy and Nicole Sirois for their excellent research assistance.

    The research is supported by the Social Sciences and Humanities Research Council of Canada and the Canadian Academic Accounting Association. Jennifer Kao also receives financial support from Canadian Utilities Fellowship for this project. All remaining errors are the authors' sole responsibility.


Recently, Penman and Sougiannis (1998) and Francis, Olsson, and Oswald (2000) compared the bias and accuracy of the discounted cash flow model (DCF) and Edwards-Bell-Ohlson residual income model (RIM) in explaining the relation between value estimates and observed stock prices. Both studies report that, with non-price-based terminal values, RIM outperforms DCF.

Our first research objective is to explore the question whether, over a five-year valuation horizon, DCF and RIM are empirically equivalent when Penman's (1997) theoretically “ideal” terminal value expressions are employed in each model. Using Value Line terminal stock price forecasts at the horizon to proxy for such values, we find empirical support for the prediction of equivalence between these valuation models. Thus, the apparent superiority of RIM does not hold in a level playing field comparison.

Our second research objective is to demonstrate that, within each class of the DCF and RIM valuation models, the model that employs Value Line forecasted price in the terminal value expression generates the lowest prediction errors, compared with models that employ non-price-based terminal values under arbitrary growth assumptions. The results indicate that, for both DCF and RIM, price-based valuation models outperform the corresponding non-price-based models by a wide margin. These results imply that researchers should exercise care in interpreting findings from models using ad hoc terminal value expressions.