Investigatino The Relationship Beween Maret Values And Accounting Numbers For 30 Selected Australia Listed Companies

Investigating the Relationship between Market Values and Accounting Numbers for 30 Australian Listed Companies ABSTRACT In capital market research CMR studies of the value relevance of

Investigating the Relationship between Market Values and Accounting Numbers for 30 Selected

Australian Listed Companies

Victoria Jane Clout

B Business (Hons.) (Queensland University of Technology)

A dissertation submitted for the degree of Doctor of Philosophy within the School of Accountancy, Faculty of Business, Queensland University of

Technology

April 2007

Investigating the Relationship between Market Values and Accounting Numbers for 30 Australian Listed Companies

ABSTRACT

In capital market research (CMR) studies of the value relevance of accounting numbers are founded upon the concept that, in equilibrium, the book values are equal to or have some long-term relationship with the market value and that market returns are related to book returns

This thesis seeks to resolve a gap in the CMR by examining 30 selected individual firms listed on the Australian stock market during the period 1950

to 2004, using equilibrium correction modelling techniques Even these limited prior works used cross-sectional techniques rather than the long-run, time-series, analysis used in this study Moreover, dynamic analysis in the CMR has tended to focus on indexes or portfolio data rather than using firm-specific case study data of the type modelled here No prior research has taken this approach using Australian data

The results of this thesis indicated that an equilibrium correction relationship between market values and book values for firms listed on the Australian Stock Exchange (ASX) could be determined by using accounting and macroeconomic regressors The findings of the thesis were consistent with the literature in terms of the variables suggested and important in the firm’s valuation from the three main approaches, the analysts (industry) approach, the finance and accounting theory (textbook) approach and the CMR literature approach The earnings, dividends and book value variables are significant in their relationships with the firm’s market values The models constructed were typically more informative and had an increased forecasting

performance compared with the a priori models tested, based on theory and

the literature

Keywords

Capital market research; Value Relevance; Dynamic Modelling Equilibrium

correction; Error correction models; Forecasting; Sufficiency

Contents

Chapter 1 Introduction

2.4.1 Cross-sectional studies – fundamental analysis 22

3.7 Benchmarking in final ‘best’ model selection 40 3.8 Economic theory at the foundational level of CMR 42

3.10 The sufficiency of accounting numbers for market value 44

Chapter 4 Research methods

4.5.2 Testing for co-integration using an equilibrium

4.6 Benchmarking of models and procedure for the

4.6.2 Procedure for the selection of the ‘best’

Chapter 5 Results

5.4.1 Statistical equilibrium correction models

5.4.2 Rationale for discontinuing untransformed (raw)

5.5 Results at the logged level for each of the 30 firms 107

5.5.1 Statistical equilibrium correction models

5.3.3 Summary of the ‘best’ models phase for the 30 firms

5.5.4 Graphical analysis of model performance and model

5.6 Comparison of the untransformed and

Chapter 6 Discussion and analysis

6.3.1 Statistical model example - Burns, Philp & Co Ltd 140

6.3.2 Real statistical model example - Southcorp Ltd 144

6.3.3 Random walk with trend model example -

6.3.4 Earnings capitalisation model example -

6.3.5 Ohlson-type model example - Coles Myer Ltd 151

6.4 Comparative analysis of firms with non-statistical ‘best’

models 156

6.5 Analysis of the consistency, value relevance

6.7.3 Accounting textbook approaches to valuation 187

Appendix B – Accounting & market value and data levels

Appendix C – Benchmarking results for all 30 selected firms 293

Appendix D – Results used untransformed (raw) data – change in

market value as the dependent variable (estimated for 4-year

Appendix E – The ‘best’ model performance and recursive graphics

for all 30 firms – based a 10 year hold out forecasting period 354

Appendix F - companies with alterative models to the statistical

List of Tables

Table 4.1 Number of companies on the Sydney Stock Exchange Official List as

Table 4.4 Exact periods raw share price obtained from the CRIF

AGSM Annual Report Record (ARR) database (from the master dataset

Table 4.5 Australian Foundation Investment Company Ltd raw share

Table 4.7 Exact periods accounting numbers obtained from the CRIF

AGSM Annual Report Record (ARR) database (from the master dataset

Table 4.8 Australian Foundation Investment Company Ltd accounting

Table 4.9 Example of currency conversion for Australian Foundation

Table 5.1-1 Descriptive statistics of financial data for the 30 firms –

Table 5.1-2 Descriptive statistics of financial data for the 30 firms –

Table 5.1-3 Descriptive statistics of financial data for the 30 firms –

Table 5.3-2 Augmented Dickey Fuller (ADF) tests on individual firms

– FHF to SOL (tested with a trend and a constant and at the first

Table 5.3-3 Augmented Dickey Fuller (ADF) tests on individual firms –

SRP to WPL (tested with a trend and a constant and at the first

Table 5.4-1 Untransformed statistical model results for the first

Table 5.4-5 Combinations of financial variables in the short (SRD)

Table 5.5-1 Summary of statistical models (model 1) developed for

Table 5.5-2 Summary of statistical models (model 1) developed for

Table 5.5-3 Inclusion of variables in the SRD and LRD (log-transformed

Table 5.5-4 Combinations of financial variables in the short (SRD)

Table 5.5-5 Diagnostic testing of the statistical models

Table 5.7 Best models, coefficients on variables in the SRD and OLS

Table 5.10 Comparison of untransformed (additive) and

Table 6.11 Comparison coefficient magnitudes with average of

List of Figures

Figure 3.1 Models compared for each of the 30 firms in the benchmarking

Figure 5.1 Australian money supply and total credit market debt for the

Figure 5.2 The Australian All Ordinaries Index for the period of

Figure 5.3 Australian Foundation Investment Co Ltd (AFI) market

Figure 5.6 Permanent Trustee Co Ltd (PMT) model performance additive

model 106 Figure 5.7 BHP Billiton Model Performance in sample and 10-year hold-out

Figure 5.8 Gowings Ltd Performance in sample and 10-year hold-out

Figure 5.9 Movements in the growth rate of market value for selected

Figure 5.13 Movements in the growth rate of macro-economic variables 137

Figure 6.1 Burns, Philp & Co Ltd model performance of statistical model

Figure 6.2 Southcorp Ltd model performance of real statistical

Figure 6.3 Permanent Trustee Co Ltd model performance of real

Figure 6.4 Australian Gas Light Co Ltd model performance of real

Figure 6.5 Coles Myer Ltd model performance of real statistical model

Statement of Original Authorship

“The work contained in this thesis has not been previously

submitted for a degree or diploma at any other higher education

institution To the best of my knowledge and belief, the thesis

contains no material previously published or written by another

person except where due reference is made.”

Acknowledgements

I would like to express my sincere gratitude to Professor Roger Willett, my principal supervisor Roger offered his hard work, patience and support throughout the writing of this thesis

I would also like to express my thanks to the academic and administration staff of the School of Accountancy at QUT for their resources, support and advice Their encouragement and kindness made me feel most welcome during my time at QUT I would also like to acknowledge the financial support of the QUT scholarship, the QUT Postgraduate Research Award, and the School of Accountancy Scholarship I gratefully received which enabled me to undertake this thesis

My appreciation is also extended to my fellow research students Steve Su, Eko Swardi, Sabri Hasan, Teruyo Omura and Chun Wei Huang for their genuine friendship and support I would also like to thank Dr Elizabeth Webster of the University of Melbourne for her suggestions regarding share price collection sources and Dr David Simmonds, CRIF manager, for data collection assistance

Finally, I would like to thank my parents, Robyn and John, my brother Richard and my sister Elizabeth, for their support, encouragement, patience and understanding over the past three years

This thesis has been edited

Chapter 1

Introduction 1.1 Purpose of the study

The purpose of this thesis is to investigate the dynamic relationship between market values and accounting numbers for 30 selected firms that have been listed continuously in Australia for at least 50 years Studies investigating the role of accounting information in capital markets began with the pioneering work of Ball and Brown (1968) This is now one of the most popular research areas in the accounting literature There have been numerous studies investigating this topic, especially in the US, for example, Beaver (1974), Beaver and Ryan (2000), Ohlson (1990; 1995) This research can be broadly classified into three main areas, (i) studies of market reaction to newly released

accounting information (Brown, 1970; Brown et al 1977; Easton, 1991), (ii)

studies of the long-term association between stock returns and accounting

numbers (Bartholdy et al., 2004’ Easton & Harris, 1991; Penman, 1992;) and

(iii) studies devoted to the use of accounting data by investors and to the

impact of market pressure on accounting choices (Ball et al., 2000; Barth &

Clinch, 1998) This study analyses the issue of the nature of the fundamental long-run relationships between market values of firms and their main accounting aggregates and the net book value of assets, earnings and dividends

This area of research is vital with regard to capital markets because accounting information is thought to facilitate the prediction of a firm’s future cash flows and to help the investors assess future securities’ risks and returns For this reason, many studies have been conducted during the last three decades (Kothari 2001; Richardson & Tinaikar, 2003)

This study constructs firm-specific models of the relationship between time series of annual market values and accounting values for 30 selected firms listed on the Australian Stock Exchange (ASX) from 1950 to 2004 It extends the previous literature in several ways First, the value-relevance theme is adapted to rely more on empirics and less on unproven theoretical assumptions Second, the analysis is dynamic (over time) as opposed to the predominately cross-sectional modelling approach used in prior research Third, the research method supplements the econometric analysis of data with accounts of the history of the firms in the sample, to introduce more formally the contextual information into the assessment of the market to the accounting value relationship Fourth, the study provides information about the unique Australian economic environment over an extended period of time

The selection criteria for the 30 selected firms was based on the firm being listed on an Australian stock exchange prior to 1955 and remaining listed continuously for at least 50 years The dynamics of the annual market value to the accounting number relationship for the companies are modelled individually using an Equilibrium Correction Model (ECM) over an extended period of time (Hendry, 1995) This tests for the existence of a long-run equilibrium between market and accounting values that is often only either assumed (Miller & Modigliani, 1961) or inferred on the basis of other theoretical assumptions (Ohlson, 1995) Value relevance research has been conducted predominately in the form of testing a specific theory through its interpretation as an estimated, cross-sectional econometric regression model The study contributes to the literature as one of only a few dynamic investigations into the long-run equilibrium between market and accounting values pertaining to the Australian data The dynamics of the relationship are

individually modelled as an ECM for each of the 30 firms between 1950 and

2004

1.2 Relationship to prior research

The existing approaches to company valuation in the literature can be classified into three types of approaches, industry (analyst), finance and accounting theory (textbook) and capital market research (CMR) The research of this thesis is within the context of these three approaches

At present, little direct evidence on the financial analyst approach to valuation

is published because of the proprietary nature of the information However, there are survey and behavioural researches — including Arnold and Moizer (1984), Barker (1999), Govindarajan (1980), Lee and Tweedie (1981), Previts

interested in earnings, earnings magnitude, earnings growth and earnings per share (EPS) Based on anecdotal, survey and behavioural research, the analysts’ perspective on earnings is that it is of considerable importance in firm valuations This variable is frequently investigated in capital market research (CMR)

Finance and accounting theory (textbook) approaches are widely published and have been examined in empirical studies in the literature The usual starting point in the textbook finance approach is the present value of expected dividends (PVED) The main disparity in this theory, amongst others in explaining ‘firm valuation’, is an issue within the literature that is open to empirical investigation The inclusion of dividends as a variable in the data set used to construct the firms’ models permits the consideration of the validity of the emphasis on the distribution of wealth to shareholders rather than on the creation of wealth (earnings)

In the accounting ‘textbook’ approach, the main valuation model is the residual income valuation (RIV) model Within the survey and behavioural literature, there is little evidence of this model being used in practice by analysts for firm valuation, although again it appears extensively in textbooks The accounting textbook approach, as opposed to the finance textbook approach, often adopts book value as an anchor for valuation models and is a key component of the original Ohlson (1995) model

Ohlson (1995) provided a theoretical structure on the relationship between the market value of common equity and accounting variables This seminal paper was followed by a strong revival of interest by empirical researchers who re-examined the relationships between firm value and accounting information using fundamental analysis There have been numerous empirical investigations to date of the Ohlson (1995) model using a cross-sectional

approach (Abarbanell & Bernard, 1995; Dechow et al., 1999; Francis et al.,

2000; Frankel & Lee, 1997, 1998; Hand & Landsman, 1998; Penman & Sougiannis, 1998) However, adopting a cross-sectional approach has limitations because it ignores the Ohlson model’s time-series nature (Lo &

Lys, 2000; Qi et al., 2000) The first empirical study to investigate the

time-series relationships between market value, dividends, earnings and book value

was by Bar-Yosef et al (1996) The study examined the lag structure of the

Ohlson model for a sample composed of stationary firms A number of recent empirical studies of the Ohlson model have adopted a time-series

approach (Ahmed et al., 2000; Ballester et al., 1999; Callen & Morel, 2000; Lee

et al., 1999; Morel, 1999; Myers, 1999)

Another theoretical perspective on the market accounting value relationship is provided by Penman (1992), who also argued for a “return to fundamentals” Empirical studies that have followed this approach include Abarbanell and

1989b) and Piotroski (2000), Stober (1992) Again, the majority of studies testing the hypotheses generated by Peman’s theory have been cross-sectional

To date, time-series studies into the relationships between accounting numbers and market values have been much fewer in number Examples exist more often in the finance literature (Schiller, 1981) and more recently in

accounting, for instance, Kothari and Shanken (1997), Qi et al (2000) Yang et

al (2000) examined the Australian market in a dynamic study of the relationship between prices and earnings at the firm level, producing results that supported the co-integration of prices and earnings Dynamic

econometric studies of firm level data have also been undertaken by Cooke et

al (2005), Omura (2005), Suwardi (2004) and Willett (2005) This thesis is based on a dynamic modelling approach to the fundamental analysis developed in the latter studies

The research methods used in this study adopt an empirically-driven, historical approach to understanding the relationships between book values and market values in the Australian financial markets The variables of book value, earnings and dividends have been incorporated into the valuation approaches of analysts, finance and accounting textbooks, as well as being included in both theoretical and empirical studies in the CMR Both the wider context of valuation models and the CMR literature provide a motivation for examining the variables, earnings, book value and dividend relationships to market values in the firms selected for this study Econometric techniques are used to provide the assessment of these relationships, rather than to test a central, single research hypothesis concerning the relationships The literature

on the mature economic markets tends to support the conventional assumptions of capital market research (CMR) such as the Capital Asset Pricing Model (CAPM), the efficient markets and portfolio theory as being appropriate when analysing the Australian share market (Kothari, 2001)

Nevertheless, there are doubts concerning the validity of the models on which these findings are based (Willett & Falta, 2007) Following the testing down approach of Hendry (1995), this study uses econometric models to provide a parsimonious statistical description of the relationship between market and accounting values, against which the models suggested by specific theories can be compared

1.3 Motivation

The motivation for the thesis is to improve the understanding of the relationship between market and accounting values in the context of Australian capital markets Australia is the world’s eighth largest equities market by market capitalisation and the second largest in the Asia-Pacific region after Japan (Crooke, 2003; Morgan Stanley Capital International (MSCI) World Index) In December 2005, the number of ASX listed companies was 1,8731 (ASX, 2005) In February 2005, the market capitalisation of the shares of the domestic companies in Australia was US$746 billion, the third highest in the Asia Pacific region (NSW DSRD, 2005)

Given the inconsistent findings regarding the relationship between market value and reported accounting numbers in cross-section research into fundamentals, it is an open question as to whether the disclosure of accounting information affects share prices or whether it is irrelevant to the behaviour of share prices because the market obtains the information through other sources (Beaver & Dukes, 1973) However, corporate collapses in the later half of the 1990s and in the first half of the 2000s in Australia2

and internationally have heightened awareness of the need for improvements in

1 The total number of companies was composed of 1,736 domestic and 71 foreign companies

transparency and the quality of disclosure in Australian financial statements The Federal Government’s formal response3

to these scandals in the political arena was to detail increases in auditor independence and transparency with the pursuit of other measures to improve the monitoring of corporate governance The Australian Securities and Investment Commission (ASIC) has announced a new accounting surveillance project directed at areas of accounting abuse of the type recently uncovered in the US, particularly capitalised and deferred expenses, recognition of revenue and recognition of controlled entities and assets Therefore, there is some interest in investigating the relationship between market values and accounting values in this context

as well as for purely intellectual reasons On the basis of the statistical record,

is there observable evidence that the market values of firms are dependent upon reported accounting numbers? This interest is served in this study by establishing a better understanding of the relationships between accounting net asset book values, earnings, dividends and share prices in 30 firms listed

on the Australian Stock Exchange

The previous two sections indicate the broad areas in which the research question is developed The main research question is: ‘At the level of the firm, what is the long-run nature of the relationship between accounting numbers and market value, certain macro-economic data and reported financial statements

of thirty Australian firms, based on the information drawn from the Australian

examination collected from a variety of sources of archival data supplemented

by historical data in the public domain

3 In a paper entitled: 'Corporate Disclosure: Strengthening the Financial Reporting Framework' now know as CLERP 9 (Corporate Law Economic Reform Program, Chapter 9)

1.5 Significance of the study

This study follows the tradition of CMR research by using econometric techniques to investigate and describe the relationship between market and accounting data However, it extends present knowledge of the relationship between market values and accounting information in novel ways In contrast

to prior research, the nature of this project is a general case study investigation into the nature of the dynamic relationship between market and accounting values in 30 different firms The econometric approach and manner in which qualitative and quantitative methods are used to address the

research question are unusual in the context of the CMR literature

Hendry’s (1995) general-to-specific method of testing econometric models is employed Consequently, within a general framework guided by economic theory, the approach is more empirically driven than is the usual ‘positivist’ approach taken when examining similar questions in the CMR Furthermore, the econometric analysis is carried out in a broader qualitative, contextual assessment of the history of each firm Much of the CMR is narrowly focused

on the statistical analysis of the sample used, often without sufficient attention being given to the issues of the model specification and also without enough consideration being directed towards understanding the broader context within which the statistical results are realised Other researchers contemporaneously conducting projects using a similar method with the data for other countries include (i) Suwardi (2004) in a ten-year study of

Indonesian firms using quarterly financial reports, (ii) Cooke et al (2005) and

Omura (2005) examining five Japanese firms over a 50-year period and (iii) Willett and Falta (2007) modelling 30 US firms over a 50-year period The findings from this thesis, therefore, provide important and unique empirical results that can be usefully compared with the results from similar models applied to data from other countries, allowing for the consistency of the

approach and the cumulative development of understanding required for the growth of scientific knowledge

1.6 Outline of the thesis

The remainder of the thesis is organised as follows Chapter 2 reviews the literature relevant to the research question The focus of the chapter is the approaches to company valuation Currently, three different approaches exist including, the industry (analyst) approach, the finance and accounting theory (textbook) approaches and the capital market research (CMR) literature approaches Each approach advocates the inclusion of different variables for

a firm’s valuation This thesis is set within these competing valuation

approaches

The theoretical framework for the research methods used is explained in Chapter 3, which includes a discussion of two broad issues from the current CMR, the inadequacy of the method and the inadequacy of the theory The former is approached in discussions of the variable identification, functional form, dynamic specification, replication for benchmarking and the model selection strategy based on the benchmarking of the final ‘best’ model A detailed outline of the benchmarking process is presented Benchmarking is used here, as in Willett (2005), to describe the phase of testing during which the statistical models are compared based on the forecasting performance

with a priori models generated on the basis of theory and the prior literature

The latter issue is discussed in the examination of the economic theory at the foundation level of the CMR

Chapter 4 describes the research methods used in the study Details of the selection and the composition of the study sample are given with a description of the variables suggested from prior literature and the sources

from which the variables were obtained Different phases of the research method are described, exploratory, modelling and interpretation

Chapter 5 reports the results of the thesis from the application of the research methods in Chapter 4 General points relating to the 30 selected firms are covered The results of the modelling phase of the research method include the construction of the statistical equilibrium correction models, the construction of alterative models to be used in benchmarking and the selection of the ‘best’ performing models for each firm

Chapter 6 analyses and interprets the results reported in Chapter 5 in the context of what is known about each firm’s history and its exposure to the broader economic trends The overall results for the 30 firms are discussed for the resultant statistical equilibrium correction models and the ‘best’ models are selected during the benchmarking phase For each of the 30 firms,

a historical background is provided to embed the results of Chapter 5 with a specific discussion of the coefficients of the ‘best’ performing models as well

as an investigation of the fit of the models to the growth rate of the market value for each firm The ‘sufficiency’ of book for market value is discussed The results of the thesis are compared with those in the literature — as reviewed in Chapter 2 — with the consistent findings that this thesis reported

Chapter 7 concludes the thesis with a summary of the findings of the study, the contributions and the limitations of the study Suggestions are made for possible future research

is within the three main areas of financial analyst approaches, finance and accounting theory textbook approaches and the capital market research approach

2.2 Financial analyst’s approach

To date, little direct evidence has been published about the models used by financial analysts to value companies This information is considered proprietary

in nature However, analysts are known to be interested in earnings, earnings magnitude, earnings growth and earnings per share (EPS) through survey and behavioural research (Barker, 1999) It would be very unusual for an analyst to discuss a firms’ book value magnitude or growth as a strong indicator of the firm’s valuation This evidence indicates that a firm’s earnings are an important variable for study in this thesis

The survey and behavioural research of the market participants provides one of the few avenues for insights into the analysts’ valuation models (Barker, 1999) The most consistent findings in this area of research indicate that the valuation model commonly preferred by analysts is the price-earnings (P/E) ratio, while other approaches such as the dividend yield are also of relevance (Arnold &

Moizer, 1984; Barker, 1999; Govindarajan, 1980; Lee & Tweedie, 1981; Previts et

al., 1994; Yap, 1997) This provides evidence for the position that the firm’s earnings are an important variable for analysts in a firm’s valuation

The current literature investigations have indicated that analysts and market participants generally did not find discounted cash flow (DCF) models, technical analysis and beta analysis to be of practical importance for investment decisions

(Arnold & Moizer, 1984; Baker, 1999; Pike et al., 1993; Vergoossen, 1993) In the

case of other valuation models, such as the dividend discount model (DDM) (Baker, 1999), dividend yield, price-cash flow ratio (PCF) or net asset value (NAV), there is little direct evidence for their use in practice

Arnold and Moizer (1984), Vergoossen (1993) and other researchers have provided evidence that analysts perceive themselves as being able to add additional value, including private information, by converting raw data into value-relevant information The process may be described as fundamental analysis, which incorporates the use of valuation models Currently, little of this

is explained in the existing empirical and theoretical literature (Baker, 1999) Owing to the lack of available evidence on the models used in practice by analysts, this thesis will follow an indirect investigation of this issue through the incorporation of earnings in the models tested in this thesis Until strong evidence exists of appropriate analyst models, this approach is the only option

to the analyst approach, is the present value of the expected dividends (PVED); this is essentially a consumption model (Ohlson & Gao, 2006) This conflicts with Modigliani and Miller’s (1958) dividend policy irrelevancy (DPI) in that the distribution of the wealth for a firm is not as important as the creation of that wealth It follows that distributing dividends is less difficult for a firm than is wealth creation This thesis seeks to investigate this conflicting set of ideas between the PVED model and the DPI indirectly by including dividends in the models tested The significance of dividends in the models (see Chapter 4 for model details) constructed in this thesis results (see Chapter 5 for results) will be compared with the support based in the PVED approach or the DPI (see Chapter 6 for discussion and analysis)

The finance approach known as the free cash flow (FCF) model examines the cash flow statement, in particular, the cash flows from operating activities and the cash flows from the financing activities Based on survey literature and anecdotal evidence, the FCF model is rarely used in practice by analysts while appearing extensively in finance textbooks Market players, such as security analysts, managers and the business media, are exceedingly interested in the accounting earnings The evidence of price responses to earnings announcements and earnings forecasts issued by analysts and managers supports

the argument that investors share this focus on earnings In contrast to current practice, modern finance textbooks generally promote a focus on ‘free cash flows’ and caution against the use of earnings in valuation models The majority

of textbooks on finance and financial statement analysis construct models focusing on the firm’s free cash flows as the present-value attribute (Ohlson & Gao, 2006) Little agreement exists in the literature on how the FCF model should be estimated in the empirical literature; thus, numerous competing measurements exist (Sloan, 1996)

Free cash flows are conceptually defined as all cash generated by operations that can be distributed back to shareholders without impacting the current level of

growth (Hackel et al., 2000) Thus, the firm can then distribute these free cash

flows back to the shareholders without affecting the market value of the firm, beyond the cash decrease experienced, or can use the free cash flows to pursue a new business opportunity By undertaking new investments, additional free cash flows will be generated in the future and will increase the firm’s market value An alternative use of the free cash flows is to pay off the debts of the firm Generally, free cash flows are estimated as the net cash flow from operating

activities minus capital expenditures (Hackel et al., 2000)

An FCF measure specified by Copeland et al (1994) and empirically investigated

by Francis et al (2000) is:

t t

t t

t t

Where SALES t is the sales revenue for year t, OPEXP t is the operating expenses

for year t, DEPEXP t is the depreciation for year t, t is the corporate tax rate, DWC t is the change in working capital in year t, and CAPEXP t is the capital

expenditures in year t.

The FCF valuation model (Ohlson & Gao, 2006) is as follows:

)/(

1 0

0 = fa +c R −γ

Where p 0 is share price at time zero, fa 0 is the financial assets of the firm net of

debt at time zero, c t is the expected cash flow from the operations for the period

t and R is equal to 1+r the discount factor (r = the cost of equity capital)

There are several problems with free cash flow estimation The first is the assumption that all capital expenditures made by the firm were necessary to maintain the current growth level of the firm This is often not the case in practice; some capital expenditures incurred by the firm are unnecessary and could more likely have been better used in being distributed back to shareholders, thus not impacting on future growth Second, part of the firm’s payments used in the continuing operations of the business may have also have been discretionary and thus, equally, could have been returned to shareholders without impacting on future growth (Jensen, 1986) An estimation of the FCF model will not be performed in this thesis An examination of the dividends in this thesis will explore the emphasis within the FCF model of the distribution of wealth to shareholders

The main accounting ‘textbook’ model for valuation is the residual income valuation model (RIV) Similar to the FCF model, little evidence in the survey literature supports the use the RIV model in practice by analysts, whilst appearing extensively in the finance and accounting textbooks Lo and Lys (2000) suggested that the earliest record of the RIV in the literature was in 1938 (Preinreich, 1938), yet there are indications that the relation was known even earlier As an alterative to translating the financial statement amounts into free cash flows, the RIV allows a more direct expression The RIV is based on a

hypothesis that asset prices represent the present value of all future dividends (that is, PVED):

pt = R−τEt(dt +τ

τ=1

∞

Where p t is market price of equity at the time t, d trepresents dividends or net cash

payments received at the end of the period t, R is unity plus the discount rate r, and E t is the expectation operator based on the information set at time t.

The RIV can be derived from the PVED when two additional assumptions are included The first assumption is that the accounting system that satisfies a clean surplus relation (CSR):

t t t

When used in accounting research b tis most commonly assumed to represent the

book value of equity at time t, x t represents earnings in period ending at time t and d t is dividends received at the end of the period t.

The second assumption is that the book value of the equity grows at a rate less than R, that is:

0 )

τ a t t t

t b R E x

Where a ≡ t − × t− 1

t x r b

RIV are mathematically equivalent (Lo and Lys, 2000)

The accounting textbook approaches use the book value as an anchor for the models while the finance theory holds the income statement as the corner stone and the key for the valuation Academic research on the RIV model has indicated that the model can be a useful tool (Lee, 1999), it assists in inferring equity capital and there is a need to understand the RIV to estimate the RIV-based regressions

The disadvantages of the RIV are numerous Firstly, there are measurement issues and a lack of agreement in the literature on measurements Secondly, the rejection of the single hypothesis that investors price securities as the expected present value of the future dividends leads to the conclusion that prices are not equal to the present value of the future dividends Thirdly, the testing, in itself, is made complicated by the necessity to truncate the infinite series of abnormal earnings and the use of proxies for investor expectations Based on these disadvantages, the RIV is not estimated in this thesis To investigate the potential

of the RIV model indirectly in an exploratory manner, this thesis will include the book value in the models tested (see Chapter 4 for the models constructed) Accounting approaches to valuation typically incorporate book value and thus this variable is worth investigating in this study Additionally, the RIV is a key component of the Ohlson (1995) model

The Ohlson (1995) model contribution stems from the modelling of information dynamics, with information dynamics postulating the time-series behaviour of abnormal earnings via two equations:

1

a t

x is defined as the current abnormal earnings, x t is the current earnings,

b t is the current book value of the common equity, w is the persistence of the

current abnormal earnings, νtis other information impacting future abnormal earnings with persistence g, εt+1and ηt+1are zero expectation disturbance terms, and 0 ≤ ω , γ <1 The lower bound of the restriction is dictated by economic reasoning and the upper bound is required to achieve stationarity

Combining the residual income valuation model with the information dynamics

in equations (2.7) and (2.8) results in the following valuation function:

t

a t t

Where α1 =ω /( 1+r− ω ) and α2 =( 1+r) /[( 1+r− ω )( 1+r− γ )] The Ohlson model’s main attraction for the empiricists is the provision of a testable pricing equation that identifies the roles of the accounting and non-accounting information, and requires only three accounting constructs to summarise the accounting component (Lo and Lys, 2000) Whether the separation is empirically valid is a testable proposition The Ohlson model is tested in this thesis empirically as a benchmark model (see Chapter 4 for details) against the models constructed in this thesis The models in this thesis incorporate the book value variable as suggested by the accounting approaches to valuation

At first appearance, there is little similarity between the Ohlson model and Gordon’s dividend growth model (Gordon & Shapiro, 1956) This is because the Gordon model is being stated in terms of dividends The two models are, in fact, quite similar Both models start with the PVED, with the addition of other assumptions including the specification of the processes for earnings and dividends The Gordon growth model assumes the relationships for the firms’ dividend policy and accounting rates of return are as follows:

t

t t

t t t

t t t

b g

b b

x b d

x b b

) 1 (

1 1

1 1

+

=

+

= +

=

− +

t

x g

b g b

x

)1(

)1

t

d g

x g x

d

)1(

)1)(

1()

a t

t t

a t

t t

a t

t t

t t

t

x g

rb b

g r x g

rb rb

x g

rb x g

rb x

x

) 1 (

) 1 ( )

1 (

) )(

1 (

) 1 (

1 1

1 1

+

=

− +

+ +

=

− +

+

=

− +

(2.15)

The above dynamic is similar to (2.7) above, except for Ohlson’s inclusion of

other information, n t To complete the similarity, replace (1+g) from above with

w In contrast to Ohlson, Gordon’s model typically assumed that g ≥ 0 With the

assumptions on dividend policy, accounting rates of return and the CSR, Gordon obtains the pricing equation as follows:

g r

d g g

x g

recognising that (1+g) in equation (2.15) corresponds to equation (2.17) when g=0 The Gordon pricing equation can then be rewritten as:

a t t t

The above equation is identical to Ohlson’s valuation function1without n t While there are similarities between the valuation functions of Gordon and Ohlson, there is one important distinction (Lo & Lys, 2000) This is the imposition of one more constraint on the model by Gordon or the specific linkages between dividends, earnings and book value via dividend policy and accounting rate of return, than there is on the Ohlson (1995) model This results in the Gordon model not satisfying the DPI This is noticeable from equation (2.19) above

because the growth rate (g=w-1) in the valuation function depends on the plough-back ratio parameter f Whereas, the Ohlson model specifies the abnormal earnings dynamic independently of the dividend policy, so w does not

depend on dividend policy This thesis does not explicitly test the Gordon model, yet the models constructed in this thesis (see Chapter 4) do examine the dividend variable A key contribution of the Ohlson model is information dynamics that provide a way to link the dividend discount model to the observable accounting variables, because of the revival of the existing Gordon model

2.4 Capital market approaches

An examination of accounting information — including earnings and book value

— with market information, both share price and market value, has

predominately taken a cross-sectional approach (Collins et al., 1997; Francis &

Schipper, 1999) However, most of the research was conducted on the US market Few studies to date have adopted a time series approach, examined the Australian market or incorporated an examination of earnings, dividends and book value as they relate to market value

1 Ohlson’s valuation function is: p t =b t +α1x t a +α2νt where α1 =ω (R− ω ) and

) )(

/(

2.4.1 Cross-sectional studies – fundamental analysis

Fundamental analysis is the use of available information to determine the value

of securities and with a focus on accounting information (Penman, 1992) The key motivation for fundamental analysis research and its main use in industry is the identification of mis-priced securities with a view to investment purposes (Kothari, 2002); while the aim of fundamental analysis is to estimate a firm’s intrinsic value

In their seminal study, Ball and Brown (1968), argued that unexpected increases

in income represented new information released to the market and that these increases would increase the value of the firm If the market was efficient, then the positive returns would cease at the announcement date because the information would be immediately incorporated into share price Thus, abnormal returns would not be earned from trading on publicly available information Similarly, unexpected decreases in earnings would give negative abnormal returns Their results indicated that the market anticipates favourable and unfavourable announcements, owing to the continuous release of information to the market and that price adjustments after the announcement are not instantaneous2 However, trading profits from this information are probably not sufficient to offset transaction costs Studies have since replicated the results of Ball and Brown (1968) and Beaver (1974) in different countries, using interim earnings and comparing them with annual earnings, using shorter earnings announcement periods The overall conclusion from these tests was that the correlation of the sign of the abnormal stock return in the month of an earnings announcement with the sign of earnings change over that firm’s previous year’s

2

Referred to in the literature as the post announcement drift (PAD) (see Foster et al., 1984; and more recent evidence on the PAD: Mendenhall, 2002; Liu et al., 2000)

earnings was significantly positive Invariably, in nearly all subsequent related research, both the sign and magnitude of the coefficient on the accounting variables are considered and assessed, as is the case in this study

Penman (1992) recognised the conflict between using the discount dividend and cash flow models in share valuation by accountants with the Miller and Modigliani (1961) dividend irrelevancy proposition As noted, there is a lack of relevance in the distribution of wealth to the creation of wealth; the latter was argued by Penman as being more aligned with share valuation The approach in CMR was also criticised for evaluating the information content of accounting numbers based on the reactions of the share markets on releases of accounting information According to Penman, it should be measured by the ability of the accounting numbers to signal information about the value of the firm

Another theoretical paper suggesting that capital market research should relate stock prices to fundamental accounting variables was by Brennan (1991) The study pointed to problems in valuation studies For instance, the regression coefficients were not stable owing to the possible misspecification of the models across the years Brennan argued that the regression per share measures created problems because all the variables were deflated by the number of outstanding shares Earnings served as an independent variable in explaining changes in stock prices, but did not contain information representing potential future earnings, thereby omitting an important variable In short, the lack of a sound theoretical background for the valuation studies was held to be a serious weakness in the literature relating stock prices to accounting earnings

Two typical regressions, adopted by researchers over the past 30 years for investigating price and earnings using the traditional regression analysis such as ordinary least squares (OLS), include price and the return regressed on earnings:

it it

it it

ignoring the long-run relationship (Yang et al., 2000)

2.4.2 Earnings

As discussed earlier, in practice, analysts prefer to examine earnings of firms when performing equity valuation Book value is largely ignored in practice as well as in finance approaches Pragmatically, the emphasis on earnings is a key motivation for examining the variable in this thesis A variety of studies in the accounting literature provide evidence that equity value is related to accounting

earnings (Ball & Brown, 1968; Barth et al., 1992; Collins & Kothari, 1989) An

examination of the literature for Australia, using a cross-sectional approach, was

carried out by Goodwin and Ahmed (2006) In an alternative approach, Yang et

al (2000) examined price and earnings relationship using a dynamic simultaneous equation approach at the firm level

Goodwin and Ahmed (2006) used cross-sectional analysis to examine Australian firm data for the period 1975 to 1999 The aim was to examine the longitudinal relationship between returns and earnings, and between price, earnings and book