Accounting and Finance 44 (2004) 223–255
© AFAANZ, 2004. Published by Blackwell Publishing.

Blackwell Publishing, Ltd.Oxford, UKACFIcc unting and Fi ance0810-5391AANZ, 2003. Publishe by Blackwell Publishing442

RIGINAL ARTICLEG. Richardso , S. Tinaikar / Accounting and Finance 44 (2004) 00–00 Accounting based valuation models:
what have we learned?

Gordon Richardson, Surjit Tinaikar

Rotman School of Management, University of Toronto, Toronto, Canada

Abstract

The present survey article formed the basis of a presentation by G. Richardson
to the 8 July 2003 plenary session of the Accounting and Finance Association
of Australia and New Zealand Conference in Brisbane, Australia. The present
article reconciles the historical and forecasting branches in the published
accounting literature. Prior survey articles have primarily focused either on the
historical branch or the forecasting branch. While these approaches have
yielded useful insights, they do not attempt to synthesize the link between the
two branches of the published literature. An obvious link between the two
branches is that the Ohlson model begins with the Residual Income Model as
an initial assumption. We believe that there are other links that need further
emphasis. In the process, we also review the empirical issues and the evidence
within these two branches. We know of no paper to date that has surveyed the
empirical evidence on both the historical and forecasting branches of the pub-
lished literature. In particular, we draw inferences on the following question: on
balance, what have we learned from nearly a decade of research on accounting
based valuation models and its applications?

Key words

: Accounting based valuation; Residual income; Linear information
dynamics, Conservatism; Accounting based measures of expected returns

JEL classification

: G12; G14; M41

1. Introduction

The role of accounting numbers in valuation has been of fundamental interest
to analysts, investors and researchers alike. Much of the empirical research in

We gratefully acknowledge comments from participants at the AFAANZ conference in
Brisbane, Australia and the following individuals: Stephen Taylor, Jeffrey Callen, Peter Easton,
Ole-Kristian Hope, Jennifer Kao, Peter Pope, Theodore Sougiannis and Florin Vasvari.

Received 13 October 2003; accepted 4 December 2003 by Stephen Taylor (Associate Editor).

224 G. Richardson, S. Tinaikar / Accounting and Finance 44 (2004) 223–255

© AFAANZ, 2004

accounting based valuation has revolved around analysing historical and fore-
casted accounting numbers. The Ohlson (1995) and Feltham and Ohlson (1995,
1996) models spawned considerable interest in the role of historical accounting
numbers in valuation. Empirical applications of these models include studies
examining the value relevance of historic accounting numbers in both the levels
(e.g., Collins

et al

., 1997) and the changes (e.g., Easton and Harris, 1991). The
Ohlson (1995) and Feltham-Ohlson (1995, 1996) models also revived an inter-
est in the Edwards-Bell-Ohlson residual income model (RIM) which found
applications in valuation based on forecasted accounting numbers (Penman and
Sougiannis, 1998), fundamental analysis (Frankel and Lee, 1998) and cost of
capital studies (Gebhardt, Lee and Swaminathan, 2001).
The fact that both the historical and forecasting branches of the published
literature have evolved from the residual income model leads us to believe that
a link exists between the historical and forecasting branches of the published
empirical accounting literature. The present survey article reconciles the his-
torical and forecasting branches in the published accounting literature. Prior
survey articles have primarily focused either on the historical branch or the
forecasting branch. Lo and Lys (2000) analyse the Ohlson (1995) model and its
empirical implications, and therefore focus on the historical branch. Bernard
(1995) focuses on the forecasting branch, in particular the RIM and its inherent
advantages over the dividend discounting approach to valuation. Lee (1999)
argues that the essential task of valuation is forecasting. Therefore, historical
accounting numbers in a fundamental analysis exercise are not sufficient stat-
istics for the stream of expected payoffs. While these approaches have yielded
useful insights, they do not attempt to synthesize the link between the two
branches of the published literature. An obvious link between the two branches
is that the Ohlson (1995) model begins with RIM as an initial assumption. We
believe that there are other links that need further emphasis. For example,
empirical tests of the Ohlson (1995) and Feltham-Ohlson (1995, 1996) models
point to the conclusion that analyst forecasts capture future expected abnormal
earnings (i.e., goodwill) far better than do historical accounting numbers
combined with linear information dynamics. As a second example, tests for
accounting conservatism in the historical branch based on Ohlson (1995) and
Feltham-Ohlson (1995, 1996) models have been disappointing, yet the Ohlson
(1995) and Feltham and Ohlson (1995, 1996) theory gives us sharp insights in
the forecast branch regarding the joint implications of conservative accounting
and growth for continuing value expressions given finite forecasting horizons.
As yet another example, in an attempt to resolve the omitted variables con-
undrum facing historical type models, several empirical studies employ analyst
forecasts to proxy for other information. In the limit, historical type Ohlson
(1995) and Feltham and Ohlson (1995, 1996) empirical studies, with enough
supplementing by forecasts, become pure forecast type RIM models.
In the process, we also review the empirical issues and the evidence within
these two branches. We know of no paper to date that has surveyed the empirical