The Reliability and the Applicability of the Residual Income-based Valuation Model : Theoretical Augmentation of the Linear Information Dynamics Model and Its Validity Compared with Ohlson (1995) and Edwards-Bell-Ohlson Approaches.

Choi, Young-Soo (2002) The Reliability and the Applicability of the Residual Income-based Valuation Model : Theoretical Augmentation of the Linear Information Dynamics Model and Its Validity Compared with Ohlson (1995) and Edwards-Bell-Ohlson Approaches. PhD thesis, Lancaster University.

PDF (11003605.pdf)
11003605.pdf - Published Version
Available under License Creative Commons Attribution-NoDerivs.
Download (9MB)

Abstract

Following the seminal theoretical works of Ohlson (1995) and Feltham and Ohlson (1995, 1996), many researchers have tried to investigate the linear information dynamics (LID) model's validity empirically. However, empirical applications of the LID approach to residual income-based equity valuation, such as Dechow, Hutton and Sloan (1999) (DHS) and Myers (1999b), have produced estimates of firm value that are substantially lower on average than corresponding observed market values. DHS's results that show the quite large downward bias of the value estimates based on Ohlson (1995), together with the work of Myers (1999b) that includes the RI intercept term in the pricing model, motivate me to augment the Ohlson model in order to capture the impact of the intercept terms on the residual income forecasts and firm values. I argue that the large negative bias in LID-based value estimates might be attributable to failure to deal fully with the effects of conservative accounting in projecting residual income. I term the augmented model, which incorporates residual income (RI) and 'other information' (OI) intercepts into the linear information dynamics, as the 'intercept-inclusive' LID model. I also show that the Feltham and Ohlson (1995) LID model as well as the Ohlson (1995) LID model are special cases of the 'intercept-inclusive' LID model. The main objective of the thesis is thus to examine whether the 'intercept-inclusive' LID model produces more reliable value estimates than the extant Rl-based valuation models: the Ohlson (1995) LID-type and the EBO-type valuation models. Using U. S. (Chapter 4) and U. K. (Chapter 6) data, I show that use of a LID that impounds the effects of conservative accounting, as reflected in analyst forecast-based residual income projections, produces value estimates that are substantially less biased than those extant Rl-based models. The thesis also addresses a potentially important issue of the different applicability under different conditions of different Rl-based valuation models in Chapter 7. This is based on the idea that the models' relative applicability can differ across various firm-specific characteristics and properties, because the implementation procedures and underlying assumptions of competing models are apparently different. Among some firm-specific ex-ante variables, eamings-to-price ratio, market-to-book ratio and analyst-based one-year ahead RI forecast-to-book ratio seem to be influential with regard to the applicability of models. Despite some contributions of this study, there are also several limitations that need to be explored in further research. In particular, value estimates based on the 'intercept-inclusive' LID approach are very sensitive to the assumed discount rate and growth rate. Moreover, the 'intercept-inclusive' LID model does not appear to improve the overall accuracy of value estimates. Together with the evidence of different applicability across firm-specific characteristics, how some firm-specific ex-ante variables can be used to modify the models and how to estimate firm-specific discount rates and growth rates could be important issues in further research.