英國(guó)南安普頓大學(xué)留學(xué)生財(cái)務(wù)會(huì)計(jì)碩士dissertation由英國(guó)dissertation網(wǎng)提供,本文通過(guò)在臺(tái)灣證券交易奧爾森模型的實(shí)證分析,即對(duì)奧爾森估值模型對(duì)賬面價(jià)值與不正常的收入及其他信息的統(tǒng)計(jì),探索固定效應(yīng)回歸模型。The University of Southampton Faculty of Law, Arts and Social Sciences School of ManagementMSc Dissertation
The Information Content of Accounting Numbers ---Empirical Evidence by Adopting Ohlson Model in Taiwan Security Exchange
Presented for MSc. Accounting and Finance
This project is entirely the original work of student registrationnumber 22625658. Where material is obtained from published orunpublished works, this has been fully acknowledged by citation inthe main text and inclusion in the list of references.i
Abstract
This dissertation examines the validity and usefulness of other informationin the Ohlson valuation model in Taiwan’s security market. Due tothevague definition of “other information” in the Ohlson valuation model, thisdissertation extends prior empirical investigations to discuss the otherinformation. These other information proxies are as follows: the qualifiedinstitutional investors’ holding rate, the directors’ and supervisors’ holdingrate, the directors’ and supervisors’ pledging rate, the interest rate, theefficient exchange rate index, and the consumer price index. Six industriesin the Taiwan Security Exchange were selected as empirical samples, andthirty-six quarters of accounting and other information data were involved inthis study. The correlation between book values, abnormal earnings, otherinformation and market prices was examined by a fixed-effect regressionmodel. The result indicated that “other information” proxies were valid anduseful to be incorporatedin the Ohlson valuation model in terms of theapplications of Taiwan’s six industries. Therefore, further empiricalapplications of the Ohlson valuation model in Taiwan securities should notomit other information from the studies.
Keywords: Ohlson valuation model, book value, abnormal earning, other information, fixed-effect regression model
ii
Acknowledgments
I would like to express my gratitude to my supervisor, Dr. Robert Day. Hehelped me and encouraged me while I met some problems during thedissertation research. I appreciate Dr. Robert Day’s knowledge and skill inresearch areas. I would also like to thank my parents, my younger brothers,and Notus Lin, without whose love, I would not finished this dissertation.
iii
Contents
1. INTRODUCTION .......................................................................... 1
1.1 OBJECTIVES OF RESEARCH ............................................................................. 1#p#分頁(yè)標(biāo)題#e#
1.2 RESEARCH STRUCTURE .............................................................................. 4
2. LITERATURE REVIEW .................................................................................. 5
2.1 INTRODUCTION ......................................................................................... 5
2.2 ACCOUNTING INFORMATION CONTENT ................................................................. 6
2.3 OHLSON MODEL ............................................................................ 11
2.4 EMPIRICAL APPLICATIONS ON OHLSON’S VALUATION MODEL IN TAIWAN ............................... 17
2.5 SUMMARY OF THE LITERATURE REVIEW ........................................................ 20
3. RESEARCH DESIGN .......................................................................... 22
3.1 METHODOLOGY ............................................................................ 22
3.2 RESEARCH QUESTION ........................................................................... 26
3.3 THE OHLSON’S VALUATION MODEL ........................................................ 26
3.4 DATA COLLECTION .......................................................................... 35
3.5 SAMPLE COLLECTION ..................................................................... 36
3.6 VARIABLE MEASUREMENT .................................................................. 37
4. RESULTS ........................................................................................ 40
4.1 ORIGINAL OHLSON’S MODEL WITHOUT OTHER INFORMATION VARIABLE ............................. 42
4.2 REVISED OHLSON’S VALUATION MODEL ..................................................... 45
4.2.1 Comparison among six possible other information proxies ................................ 46
4.2.2 Comparison in Groups of Unsystematic and Systematic Risk Factors ............. 55
iv
4.2.3 Revised Ohlson’s Model .................................................................... 60
5. ANALYSIS OF RESULTS ...................................................................... 67
6. CONCLUSION ............................................................................... 70
REFERENCE ............................................................................. 73
v
Table Contents
TABLE 1 ........................................................................................................... 25
THE OTHER INFORMATION PROXIES ...................................................................... 25
TABLE 2 ........................................................................................................................................... 36
SAMPLE FIRMS ................................................................................................................................ 36
TABLE 3 ........................................................................................................................................... 41#p#分頁(yè)標(biāo)題#e#
REVISED OHLSON MODEL BY ADDING DIFFERENT OTHER INFORMATION PROXIES .......................... 41
TABLE 4 ........................................................................................................................................... 43
ORIGINAL OHLSON MODEL WITHOUT OTHER INFORMATION VARIABLE (MODEL 1) ...................... 43
TABLE 5 ........................................................................................................................................... 43
THE EFFECT ON MARKET PRICE IN MODEL 1 .................................................................................... 43
TABLE 6 ........................................................................................................................................... 46
OHLSON MODEL WITH QFII HOLDING RATE (X1T) (MODEL 2) .......................................................... 46
TABLE 7 ........................................................................................................................................... 47
OHLSON MODEL WITH DIRECTORS’ AND SUPERVISORS’ HOLDING RATE (X2T) (MODEL 3) .............. 47
TABLE 8 ........................................................................................................................................... 49
OHLSON MODEL WITH DIRECTORS’ AND SUPERVISORS’ PLEDGING RATE (X3T)(MODEL 4) ............... 49
TABLE 9 ........................................................................................................................................... 50
OHLSON’S VALUATION MODEL WITH INTEREST RATE (X4T) (MODEL 5) ........................................... 50
TABLE 10 ......................................................................................................................................... 52
OHLSON MODEL WITH EFFICIENT EXCHANGE RATE INDEX (X5T) (MODEL 6) ................................... 52
vi
TABLE 12 ......................................................................................................................................... 56
OHLSON VALUATION MODEL WITH UNSYSTEMATIC RISK INFORMATION (MODEL 8) ..................... 56
TABLE 13 ......................................................................................................................................... 58
OHLSON VALUATION MODEL WITH SYSTEMATIC RISK INFORMATION (MODEL 9) .......................... 58
TABLE 14 ......................................................................................................................................... 60
REVISED OHLSON’S VALUATION MODEL (MODEL 10) ..................................................................... 60
TABLE 15 ......................................................................................................................................... 62
THE EFFECT ON MARKET PRICES IN MODEL 10 ................................................................................. 62#p#分頁(yè)標(biāo)題#e#
TABLE 16 ......................................................................................................................................... 65
COMPARE VALUES OF ADJUSTED R2 AMONG TEN MODELS .............................................................. 65
1
1. Introduction
1.1 Objectives of Research
Following one after another fraud scandals across the world, investors haverealised that is not adequate to evaluate a firm’s value only by its marketprice or technical analysis. Enron is the most representative scandal of the2000s, when executives fabricated financial statements in order toencourage the public to invest in the company and raise its market value. Ifinvestors only observe the market value dynamics, and the surfaceinformation in financial reports, the evaluation of the investment may fail.
The 1990s saw a wealth of research being conducted into thecharacteristics of accounting-based valuation models, and how well thosemodels could be applied to the real market. More recent advances whichhave been made in these accounting-based valuation models will result inreducing any biased justification for investing in securities and build ahealthier and prosperous capital market.Due to the fact that financial reports are substantial decision-makingindicators for potential investors, governments and principle accountingprinciple regulatory institutions have endeavoured to improve the quality offinancial statements, reports and presentations. More conscientious andrigorous accounting principles will lead to more unbiased financialstatements for fundamental analysis. An extensive discussion of thusefulness of accounting information has inspired scholars to conductseveral pieces of research on estimating market value by means ofaccounting information, such as traditional financial ratios, andaccounting-based stock valuation models. This paper will place emphasis
on accounting-based models, and choose the Ohlson model (1995) as thebasis of its empirical studies.
Ohlson model is different from financial-based valuation model, and it doesnot take investors’ minimum required rates or beta value1 into consideration.Ohlson model depends on bottom-line items in accounting data and the“other information” relative to entity operation. The Ohlson model adoptsbook values, abnormal earnings (residual earnings), and the other
information (non-accounting information) to evaluate a firm’s value. It isremarkable because the Ohlson model introduced the “other information” inthe valuation function which is different from traditional accounting–based
valuation models. The contribution in the other information is that currentaccounting variables (accounting information) can evaluate current firms’operating ability and profitability, but the other information variables
(non-accounting information) can enhance the interpretation in forecastingthe future operating ability and profitability. However, Ohlson model onlyprovided conceptualised ideals on other information variables, and followingstudies (Subramanyam and Wild, 1996; Frankel and Lee, 1998; Myer, 1999)argued dissimilar opinions and definitions on other information in empirical1 The Beta value is the correlation between firm’s market value and the whole marketperformance (Fama and French, 1992).applications. Thus, this paper aims to conduct Ohlson valuation model with#p#分頁(yè)標(biāo)題#e#
variables of book values, abnormal earnings, and other information. Bookvalues and abnormal earnings empirical investigations are based onOhlson’s work and other information empirical research will refer to relativeempirical papers. According to Wu, Lin and Cheng (2004), otherinformation proxies can be replaced by operating risk factors. The reason isthat current accounting information illustrates a firm’s current performanceand historical events but operating risks can forecast the possibility offuture events. Market price contains not just what a firm’s performance inthe past but also investors’ expectations, so this paper will observe firm’soperating risks in order to determine the other information proxies.
A firm’s operating risks can be defined as systematic risks and unsystematicrisks. Systematic risks are caused by politics or economic events; firmscannot diversify risks by this kind of influences. For instance, the exchangerates, the interest rates, the inflation rates, and volatility of economic growthare recommended indicators to evaluate systematic risks which might haveeffects on firms’ market values (Abell and Krueger, 1989). On the otherhand, unsystematic risks are individual contingent events in a firm which willcause loss, unfavoured operating performance, or even bankruptcy. Forinstance, unexpected exchange losses will affect those firms which trade byin the real world it is hard to be wholly eliminated. From prior research,investors can evaluate fair market value of a firm through itsoperating risks.When the possibility of the operating risks is higher, the fair market price
will be lower. From several inside trading scandals, scholars havediscovered that the corporate governance is a useful indicator to evaluate afirm’s unsystematic risks (Jensen and Melking, 1976; Patton and Baker,1987). Therefore, the other information proxies in this empirical study willcontain two groups of operating risks and they will be conducted in the
Ohlson’s valuation model analysis.The purpose of this paper is to operate an empirical application of theOhlson’s valuation model in order to test the magnitude of the validity ofOhlson model. The empirical data session was between 2000 and 2008and the sample collection included six industries in Taiwan Security
Exchange. More specifically, this study was undertaken to understand whatthe extent of the Ohlson’s valuation model can achieve in evaluating
Taiwan’s securities.
1.2 Research Structure
The first section of this dissertation is to demonstrate the researchmotivation and purpose, and the second section is a review of the classicand pertinent literatures. Literature review addressed both theoretical andempirical aspects of the role of the Ohlson valuation model, and further thispaper analysed significant empirical studies as the underlying theories forempirical foundation. This is followed by methodology, data collection,sample collection, and variable measurements on accounting information,the other information, and historical market values. The results for thevarious analyses were presented in following descriptive and statisticsections. Implication and limitations were also demonstrated after resultsanalysis and provided a comparison for other preceding and pertinentresearch. Conclusions were illustrated in final section in this paper.#p#分頁(yè)標(biāo)題#e#
2. Literature Review
2.1 Introduction
In recent years, numerous studies have been carried out on theeffectiveness of the empirical applications on accounting-based valuationmodels. These accounting-based valuation works, reported in literature,can be classified into three main categories. The first is financial statementanalysis, which contains operating ratios, credit ratios, and investmentsratios. Analysts or investors can utilize these financial ratios to evaluatetheir investing targets by financial statements and reports. The secondcategory is discount cash flow models, such as the free cash flow model,the adjusted present value model, and the flow to equity model. These cashflow models discount a firm’s future expected cash flows in order to derivean intrinsic value of the firm. The third category is the discount dividendmodel, which calculates a firm’s present value of future expected dividendsas a firm’s intrinsic value. This dissertation will address on a derivativemodel of the discount dividend model. This model was introduced byOhlson (1995), and it is called the Ohlson valuation model (residual incomemodel). Although this paper will only apply the Ohlson valuation model,other accounting-based valuation models are also significant to the study,in order to understand the underlying theories of accounting informationcontent. Therefore, in the following literature review, the review will focuson accounting information content research paradigms, the Ohlsonvaluation model theory, and related empirical applications in this field.
2.2 Accounting Information Content
This section will demonstrate general accounting information contentliteratures before discussing literatures in dividend discount model and
http://www.mythingswp7.com/dissertation_writing/Accounting/Ohlson valuation model. Owing to no matter what accounting-basedvaluation models will be applied in empirical research or practiceapplications, the accounting information content is a fundamental andessential background to understand.
Early prominent seminal work on accounting information content wascarried out by Ball and Brown (1968). Ball and Brown examined therelationship between unexpected earnings and unexpected market returnsby event study, this was the first time to discuss the effectiveness ofaccounting information content. Ball and Brown employed the forecastingerror (unexpected earning) and Abnormal Performance Index (unexpectedreturn) to test whether the earnings announcement will influence the marketprice movement. The results in this paper indicated that accountinginformation related to an individual firm can be captured by the annualincome report, but not rating highly as a timely medium to gain abnormalprofits in the security market. It was the first article to prove the information
content of accounting numbers. However, the Ball and Brown (1968) onlyprovided a direction in this research filed and they did not test the natureand magnitude of the accounting information content.#p#分頁(yè)標(biāo)題#e#
Beaver, Clarke, and Wright (1979) continued Ball and Brown’s (1968)research and took the magnitude of the unexpected earning information
content into consideration. The finding illustrated that the abnormal returns
in each sample group was positive to the unexpected earning. Both of thesetwo papers established three fundamental hypotheses: (1) time-series ofearning movement is random walk (2) earning numbers influence themarket price and the earning information can sufficiently interpret thechange of the market price (3) in different firms or different fiscal years, the
Earning Response Coefficient is a constant.Utilizing a single variable (accounting earning) to interpret market pricemovement is arbitrary and impractical, so following researchers weredevoted to loosen restrictions in the empirical investigation and improve thepropositions in the accounting information content theory. Ball and Watts(1972), and Watts and Leftwich (1977) continued the information contentresearch on the time-series properties of earnings. They addressed thecharacteristics on the relationship between earning time-series and marketreturns, and suggested these characteristics can be used to measureexpected earnings in forecasting. Most research on accounting informationcontent in 1970s provided little illustration on complete framework ofaccounting-based valuation properties. Thus, Ohlson (1979), and Garmanand Ohlson (1980) established a more concrete and throughout aspectabout accounting information content and valuation models. They
addressed that information link is the relationship between accounting dataand future benefits from an equity investment. Meanwhile, valuation link isthe relationship between the future benefits from an equity investment andmarket price movement. Therefore, the accounting data can be a validforecasting resource to evaluate the market return.
However, the future stream of benefits from an equity investment is abstract,so Easton (1985) assumed that the present value of future expecteddividend stream can be a proxy for future benefits of an equity investment.
This is an early point of view on dividend discount model, and Eastonconducted an empirical test on cross-sectional regression of futureexpected dividend stream to affirm his assumption. The resultdemonstrated that there is a strong valuation link between market price andthe present value of future expected dividends. Meanwhile, Easton settledup another hypothesis which is if market price relates to future dividendsthen the market price will relate to accounting information. This empiricalstudy also tested the cross-sectional correlation between the earnings(accounting information) and present value of future expected dividends.The empirical results moderately supported Easton’s assumption.After confirming the relationship between accounting information andmarket price, scholars attempted to develop practical valuation function inorder to apply this technique on real market. Kormendi and Lipe (1987)demonstrated their research questions on what is the nature of theinformation in reporting earnings (accounting data) and how does it relate toa firm value. Their query brought the research into a more accurate andrigours process at the measurement in the accounting information content.In addition, they discussed about how the magnitude of the relation betweenearnings and returns relates to the time-series properties of earnings.#p#分頁(yè)標(biāo)題#e#
Following are main hypotheses according to Kormendi and Lipe’s works: (1)
the present value of the revision in expected future earnings approximatesthe present value of the revision in expected future benefits (2) the
univariate time-series model of earnings is equivalent to market expectation.
The significant contribution of this paper is the finding of the magnitude inearnings time-series models, and this can make a big progress indiscussing the accounting information content. Moreover, Kormendi andLipe (1987) considered that no evidence in empirical studies has provedmarket returns is excessively sensitive to the earnings innovations.
英國(guó)南安普頓大學(xué)留學(xué)生財(cái)務(wù)會(huì)計(jì)碩士dissertation由英國(guó)dissertation網(wǎng)提供,本文通過(guò)在臺(tái)灣證券交易奧爾森模型的實(shí)證分析,即對(duì)奧爾森估值模型對(duì)賬面價(jià)值與不正常的收入及其他信息的統(tǒng)計(jì),探索固定效應(yīng)回歸模型。Previous research in the information content of accounting numbersgenerally adopted the earning per share as the accounting variable,however, Easton and Harris (1991) addressed on the level of earning. Thereason is that Easton believes the residual error from a regression model ofannual abnormal returns might mitigate the effect of measurement error byincluding both earning level and earning change. Another reason to inspirethis view is the pervasiveness of low R2 statistics in the valuation of marketreturns by earnings, and Lev (1989) also recommended the earning levelcould be a substitution for improvement. Easton and Harris (1991) assumedthat earnings level divided by price at the beginning of the market returnperiod is relevant to market returns, and the model was based on the theorywhich the book value is relevant to market value. Subsequently, the secondmodel was assumed that the market price as a multiple of earnings, andfinally the results for these two assumptions were affirmative. The earninglevel and earning change were associated with market returns and thecoefficient on earnings level was statistically significant in all testing periodsin this empirical study. From this paper, the earning level and earningchange were proved to be better proxies than unexpected earnings in thevaluation function.
Before 1990s, most discussion between accounting information content and
market price was based on the dividend discount model. In the dividend
discount model, it assumes the firm value equals to the present value of
future expected dividends, and the market price in the capital market
represents a firm’s fair value. In this dissertation, it will discuss empirical
investigation on Ohlson valuation model which was revised from discount
dividend model, and the related works will be illustrated in the following
text.#p#分頁(yè)標(biāo)題#e#
11
2.3 Ohlson Model
In 1995 an article was published by James Ohlson and this article has been
the most imperative paper in the accounting-based valuation subject.
Ohlson (1995) developed a valuation model by introducing three substantial
variables: book values, abnormal earnings, and the other information.
Ohlson (1995) pointed out that accounting data is an important integrative
function to value the change in owner’s equity. Ohlson (1995) added clean
surplus relation (CSR) assumption into the valuation function, and the clean
surplus relation means the dividend payment today will reduce book value
and leave the current earnings unaffected. This clean surplus relation also
supports Modigliani and Miller (1958) properties which address dividend
policy is irrelevant to the firm value. Three basic assumptions were
established to create a robust framework for Ohlson valuation model. First,
firm value equals to the present value of future expected dividends. The
second is dividends can be replaced with earnings and book values based
on the clean surplus relation. Then, the final assumption is stochastic
behaviour of accounting data leads to multiple-date uncertainty. According
to these assumptions the model derived a firm’s value from weighted
average current book values and capitalised current abnormal earnings
(adjusted for dividends). Ohlson substituted the current book value and
abnormal earnings for the present value of future expected dividends in
Ohlson valuation model. Moreover, Ohlson utilized abnormal earnings in
the function which was based on Peasnell’s (1981) finding. Peasnell
implied that goodwill equals to the present value of future expected
12
abnormal earnings. The abnormal earning in Ohlson’s valuation model is
also called residual earning. Thus, the Ohlson valuation model is also
called Ohlson residual income model (RIM).
However, the abnormal earnings contain unobservable and future
information which will result in invalid condition for this accounting-based
valuation model. Ohlson (1995) improved this characteristic by assuming
liner information dynamics (LID), which solved the discretionarily evaluate
the abnormal earnings. According to linear information dynamics in the
Ohlson valuation model, book values can represent a firm’s current
operating performance, and the abnormal earnings can interpret a firm’s
future profitability. These two accounting information variables improve the
accounting-based valuation technique. The other remarkable distribution is
that Ohlson (1995) introduced “other information” or “non-accounting
information” in the valuation function. The other information is information
that beyond abnormal earnings and book values in current period. These#p#分頁(yè)標(biāo)題#e#
value-relevant events may affect future profitability but the accounting data
captures this information in a time delay. Thus, the other information can
enhance the valuation accuracy. Bernard (1995) examined the intrinsic
value of firms by Ohlson’s model2, and the result illustrated the explanatory
power was between 68%~80% which supported Ohlson’s theoretical
development by empirical evidence.
2 “Ohlson’s model” in this paper means Ohlson’s earlier work in 1995, the later work with
Feltham in 1995 will be described as “Feltham-Ohlson model”.
13
Feltham and Ohlson (1995) continued Ohlson’s earlier work, and they
adopted measurement which adjusted operating cash flows for accruals in
determining a firm’s value. The use of accounting conventions for accruals
has leaded to a difference between a firm’s market value and book value.
This discrepancy is unrecorded goodwill or abnormal earnings in the
current financial status. They discovered that operating and financial cash
flows provide distinct accounting measurements in valuation function.
Depending on three basic assumptions in Ohlson model, Feltham and
Ohlson (1995) explored the relation between market value and expectations
in future cash flows. This article addressed that operating activity is a
residual activity, because the operating activity plausibly reflects the
outcome of all activities except for those relate to pure borrowing and
lending activities. Meanwhile, Feltham and Ohlson (1995) refined Ohlson’s
earlier work and established a robust linkage among book values, abnormal
earnings, and other information by applying the model on operating and
financial activities. Only the little specific definition in the “other information”
has been debated lasting for several years.
Ohlson’s theory is not without its flaws. Myers’s (1999a) paper constituted a
powerful attack on Ohlson’s linear information dynamics assumption.
Myers (1999a) outlined a method for modifying the linear information
models to preserve internal consistency, and argued that Feltham and
Ohlson‘s (1995) weighted accounting numbers are inefficient. In addition,
Myers (1999a) argued that the median conservatism parameter of Feltham
and Ohlson (1995) is significantly negative, so Myers (1999a) replaced
14
income parameter and the book value parameter in Ohlson’s valuation
model. The outcome of Myers work was failed to capture the true stochastic
relationship among accounting variables in the valuation model. In addition
to the disagreement with the linear information dynamics in Ohlson’s model,
later work (Myers, 1999b) asserted that the estimations of Ohlson’s
residual valuation model understate a firm’s market value. Ohlson’s theory#p#分頁(yè)標(biāo)題#e#
depends on conservative accounting data, in the theory the conservatism
will reverse in a long horizon. Thus, the residual income will grow with time
and this can assume estimations of book value, abnormal earnings, and
the other information equals to a firm’s market value. However, Myers
(1999b) criticized that terminal income in firms will be hidden from archival
databases, because average firms’ survival is ten years in the empirical
evidence, this is not long enough for the accounting conservatism reverses.
In Myers (1999b) investigation, he adopted the Ohlson’s model by using
11036 delisted firms and the results of this investigation demonstrated that
this accounting model generally understates firms’ market prices. Myers
(1999b) provided suggestions that while using Ohlson’s model in security
valuation should also estimate price and book value premiums3 and take a
firm’s market price into consideration. The purpose of this recommend is to
reduce the risk of understatement while applying Ohlson’s valuation model.
Except for Myers’s doubtful statement, another issue has also caused
extensive research debates on Ohlson’s valuation model. There is no
3 The price and book value premium is the difference between price per share and book value of
equity per share at the end of the fiscal year.
15
general consensus on what is the “other information” in the Ohlson’s
valuation model. In Myers (1999a) paper, it applied the other information as
a constant proxy in the valuation model which is not the original belief in
Ohlson’s theory (1995). In Ohlson’s theory, the other information is
time-delay information beyond the current accounting information
(information data), so the other information should be other reference which
is not included in the current financial statements. Although the vague
definition of other information in the Ohlson’s valuation model. Liu and
Ohlson (1999) illustrated that the other information should not be omitted,
and the other information can be observed by expected growth in earnings
or the size of the business. Liu and Ohlson (1999) also disagreed that some
empirical studies of Ohlson’s valuation model neglect the “other information”
or deal with the other information on ad hoc basis. Moreover, Dechow,
Hutton, and Sloan (1999) utilised analysts’ forecasts of earnings in the
information dynamics and the results increased forecast accuracy. Thus,
their research generally supported Liu and Ohlson’s (1999) paper and
highlighted the important role of the other information in the Ohlson’s
valuation model.
In contrast to these debates and arguments, Ohlson’s valuation model has
some empirical supports both at the theoretical and applied levels. Dechow#p#分頁(yè)標(biāo)題#e#
et al. (1999) carried out an empirical assessment of the Ohlson’s valuation
model. They concluded although existing empirical results on Ohlson’s
model were similar to empirical results on dividend discounting model, the
modest improvements in explanatory power by adding analysts’ earnings
16
forecasts in the other information in the Ohlson’s valuation model. Dechow
et al. (1999) concluded that Ohlson’s model is a parsimonious framework
for incorporating information in book values, abnormal earnings, and
earnings forecasts in empirical research but is a significant guide for
development of accounting-based valuation models. Lo and Lys (2000) did
reviews on numerous empirical studies which implemented research on the
Ohlson (1995) model and Feltham-Ohlson (1995) model. The aim of this
paper was to derive better understand on Ohlson’s valuation model and its
limitations. Lo and Lys (2000) analysed that formal linkage between
Ohlson’s valuation function and accounting numbers, versatility of the
Ohlson’s valuation model, and higher R2 in cross-sectional price empirical
results. Meanwhile, the results showed higher correlation in the valuation
model and market prices while applying “other information” in the Ohlson’s
valuation model. Lo and Lys (2000) believed that these merits are the
reasons for the enthusiasm of the Ohlson’s valuation model. They revisited
the Ohlson model and discovered some misapplied problems in prior
research, such as ignorance of scale-effect, usage of level data, or neglect
of firms’ specific discount rates. Lo and Lys (2000) considered those
misapplied manners will results in overstatement of the validity of the
Ohlson’s valuation model. They suggested that the Ohlson model or
Feltham-Ohlson model should be refined and tested its empirical validity
with these limitations. In Lo and Lys’ paper, they recommended that the
Ohlson’s valuation model can be refined by improvement of financial
literature and the application should conquer the perfect market hypothesis
conditions.
17
2.4 Empirical Applications on Ohlson’s Valuation Model in Taiwan
In Taiwan, most researchers or scholars put their emphasis on how to apply
the Ohlson model in real market. Book values and abnormal earnings are
not difficult to implement in research, but the “other information” is not as
easy as book values and abnormal earnings to observe in the empirical
application. The other information is an unequivocal component in the
Ohlson’s valuation model. Some researchers defined the other information
as operating risk factors or operating techniques in entities, such as
research and design expenditures, values of copyright, institutional#p#分頁(yè)標(biāo)題#e#
investors’ holding rates or good will (Su, 2002; Teng, 2002; Chen, 2002;
Guo, 2002). Ting (2003) addressed that abnormal earnings in the Ohlson’s
valuation model satisfy investors’ requirement of capital returns, which is
consistent with the idea of corporate governance mechanism. The
corporate governance mechanism is to ensure the reasonable returns for
capital creditors, and the extreme aim for corporate governance is to
sustain a firm’s value in long horizon which is also consistent with the
assumptions in Ohlson’s valuation model. Therefore, Ting (2003) utilized
CLSA4 corporate governance quality ranks as the other information proxy
and calculated other proxies 5 from financial statements as the other
information components. Findings in Ting’s work illustrated that the quality
of corporate governance is positive to firms’ market values. This finding
4 CLSA is one of Asia’s largest independent equity brokers and CLSA’s headquartered is in Hong
Kong.
5 Ting (2003) recalculated the abnormal earnings divided by weighted average ordinary shares
and book values divided by weighted average ordinary shares as the other information proxies.
18
implies that individual entity can improve their corporate governance quality
in order to increase a firm’s wealth, and this will also encourage more
executives to adopt higher-standard governance conditions. Ting’s paper
(2003) accepted the Ohlson’s valuation theory and proved the other
information is significantly relevant to evaluate a firm’s market value.
However, this empirical research only applied on Electronics industry and
the tested duration was only three years, it is not ample to confirm the
significance of governance mechanism as an ”other information” proxy in
Ohlson’s valuation model.
Wu, Lin and Cheng (2004) extended Ting’s (2003) work and suggested that
the other information should contain information which is not only relative to
a specific firm or industry but also relative to the macroeconomic
environment. Their point of view illustrated that a firm’s operating risks are
not only the uncertainty in an individual entity, sometimes, economic or
politics environment will also affect on a firm’s operating performance or
security value. In Wu et al. (2004) paper, they enhanced the other
information proxies by adding systematic risks (the interest rates, the
exchange rates, and the volatilities of consumer price index) and the
qualified foreign institutional investors’ holding rates6 into the information
dynamics. The results of Wu et al.’s (2004) research supported Ohlson’s
theory and illustrated that systematic risk factors and governance
mechanism factors are statistically significant in Ohlson model more than#p#分頁(yè)標(biāo)題#e#
6 In 1991, the Taiwan Security Exchange permitted qualified foreign institutional investor (QFII) to
directly invest in Taiwan’s security market.
19
abnormal earnings. In Wu et al.’s (2004) work, they improved the empirical
investigation by utilizing eight industries and the data collection duration
was five years. This improvement enhanced the explanatory power in the
correlation between accounting-based valuation information and market
values.
Hsieh (2004) continued Wu et al. (2004) research and added technological
progress indicator as the other information proxy. Hsieh’s work tried to build
a more functional valuation model with internal and external risk factors to
improve the explanatory power in the empirical results. The results
demonstrated that governance mechanism has negative effect on market
values; meanwhile, technological progress indicator and systematic risk
factors have obvious influence on market prices. Moreover, Hsieh (2004)
tested substitutions for the book value and abnormal earnings by traditional
financial indicators7, and the results demonstrated the original accounting
information variables equip better explanatory power in interpretation of
market values. The empirical results generally supported the Ohlson’s
valuation model in Hsieh’s work because the technological progress
indicator was highly related to market prices of Electronics firms. Therefore,
further empirical studies attempt to improve the Hsieh’s work by adopting
Ohlson’s valuation model, other specific characteristics which are pertinent
to a specific entity’s operation should be verified and added into the
empirical investigation.
7 These traditional financial indicators are returns on assets, returns on equity, and returns on
sale.
20
2.5 Summary of the Literature Review
The extensive literature review has been presented in order to clarify the
relationship between accounting information content and market price
movement. Meanwhile, empirical applications, by adopting Ohlson’s
valuation model and implications in academic research, are also discussed.
In the early stage of research on accounting information content, earnings
has been utilized to conduct the empirical research, the concept was linking
earnings and expected future dividends to estimate the expected firm value
and market price. After rigorous and pertinent investigations, scholars took
time-series properties and earning persistence characteristics into
consideration. Researchers intended to analyse financial statements and
data on market prices through information perspective before the 1990s but
they have improved by adopting measurement perspective as a more
comprehensive approach in refining the valuation model. Nowadays, more#p#分頁(yè)標(biāo)題#e#
researchers have devoted their efforts to enhancing other applicable and
significant factors into empirical studies, in order to derive a true picture of
the relationship between accounting information and market prices.
Accounting-based valuation models, such as the dividend discount model
and Ohlson’s residual income model, are frequently discussed while
applying accounting information content research. According to the
conservatism and growth characteristics of accounting data, Ohlson (1995)
introduced a residual income model, which equips the same core aspects
with the dividend discount model. However, the Ohlson valuation model
has been improved in several dimensions. Firstly, Ohlson (1995) added
21
linear information dynamics into the fundamental assumptions, and this
linear information dynamics assumption resulted in being able to estimate
future abnormal earnings in the valuation function. Another contribution is
adding the other information into a valuation formula, and the other
information enhances the explanatory power in interpretation of the
valuation function. Finally, the third assumption in Ohlson’s model is the
stochastic behaviour of accounting data. These assumptions in the Ohlson
residual income model have developed this accounting-based valuation
model, to make it more practical and applicable. Since 1995, Ohlson’s work
has obtained a great deal of admiration and appreciation. More and more
researchers are devoted to examining the validity and usefulness of
Ohlson’s residual income model. Dechow et al. (1999), revisited the Ohlson
valuation model and concluded that it is a guide for a more robust
accounting–based valuation model. Myer (1999a), suggested that the
linear information dynamics and the understatement of market prices, is
due to the design defects in the Ohlson valuation model, which needs more
empirical studies to refine it. Moreover, Lo and Lys (2000) arranged
Ohlson’s valuation model applications and discussed this model in depth.
They considered that if the Ohlson valuation model can be modified to
adjust the scale-effect and levels of data influence, the Ohlson valuation
model will be more flawless and practical. Applications in Taiwan put more
emphasis on what proxies are appropriate for the other information
variables. Scholars have combined the operating risk concepts with the
Ohlson valuation theory. They have discovered that unsystematic and
systematic risk factors could be applicable proxies for the other information
22
variables. Results of these relative hypotheses support Ohlson’s work. In
addition, Hsieh (2004) took specific operating risk factors as other
information proxies into consideration in the Electronics industry in order to
test the usefulness of the Ohlson valuation model in a specific industry. The#p#分頁(yè)標(biāo)題#e#
results of Hsieh’s (2004) work also agreed with the conclusion of Ohlson’s
valuation theory. This literature has given subsequent researchers clues
and guidance about how to apply the accounting-based valuation model
and what limitations should be further refined.
3. Research design
3.1 Methodology
The paper is a piece of market-based accounting research (Bernard, 1987),
and can be classified as a functionalist paradigm (Morgan, 1980) or a
positivistic paradigm (Eckberg and Hill, 1979). The normal process under a
positivistic paradigm is to study the literature, construct a hypothesis and to
design empirical research (Hussey and Hussey, 1997, p.55). This paper
firstly addresses its aim to confirm the validity and usefulness of the Ohlson
valuation model in empirical applications, and to synthesize literature
reviews of what is and what is not known. Subsequently, this paper will
establish Ohlson valuation model assumptions underlying the residual
income theory and three fundamental assumptions in Ohlson’s work, in
order to execute the empirical investigation. Finally, positivistic paradigm is
adopted by using highly specific and precise data (Hussey and Hussey,
23
1997, p.56), and also, the market data used in this paper belongs to
longitudinal data (Singer and Willett, 2003). Therefore, in data collection,
this investigation has chosen six industries: Building & Construction,
Chemicals & Biotechnology, Electronics, Foods, Plastics, and Textiles. The
duration of this longitudinal data collection is nine years and the collection
is quarterly data. The goal of this paper is to verify the usefulness of the
Ohlson valuation model’s practical applications in the Taiwan Security
Exchange.
The core concept in dividend discount model is analogy to the Ohlson
valuation model, and the reason for choosing the Ohlson valuation model,
rather than the dividend discount model, is that “other information” could
improve the accuracy of the valuation model, according to prior empirical
investigations. Another reason is that not every firm distributes cash
dividends regularly, and the result of which is that the dividend discount
model is hard to operate appropriately. This empirical study attempts to
supplement the findings of Wu et al. (2004), and Hsieh’s (2004) earlier
works. The methodology is similar to the previous studies discussed above,
but some processes in the empirical investigation are different. This paper
will expand the observation periods and the industries in the sample
collection, and it will improve one of the other information proxies. The
exchange rates which were utilized in Wu et al. (2004) and Hsieh’s (2004)
paper will be refined as the efficient exchange rate index. Because of#p#分頁(yè)標(biāo)題#e#
exports are not only exported to the United States or evaluated by US
dollars; this other information proxy should consider the purchasing power
24
and compare it to different significant currencies. Therefore, this paper will
adopt the New Taiwan dollar efficient exchange rate indices as one of the
other information proxies, and this improvement can describe appropriate
purchasing power of New Taiwan dollars. Moreover, this paper will ignore
the technological progress factors, which were introduced in Hsieh’s (2004)
paper in other information proxies, because these factors are less
important in traditional industries’ operating activities. Book values and
abnormal earnings measurements are as the same as in Hsieh’s (2004)
work. Below is the main comparison in adopting the other information
proxies in the Ohlson valuation model,
25
Table 1
The other information proxies
Wu, Lin & Cheng
(2004)
Hsieh (2004) This paper
Duration 1998Q2-2003Q3 1999Q1-2004Q1 2000Q1-2008Q4
Industry eight industries one industry six industries
Unsystematic
risk factors
Directors’ and
supervisors’ holding
rate
Managers’ holding
rate
QFII holding rate
Directors’ and
supervisors’
holding rate
Directors’ and
supervisors’
pledging rate
QFII holding rate
QFII holding rate
Directors’ and
supervisors’
holding rate
Directors’ and
supervisors’
pledging rate
Technological
progress
factors
none R&D activities8
Equipment rate9
none
Systematic
risk factors
Interest rate
Exchange rate
CPI10
Electronics index
Exchange rate
GDP11
Interest rate
Efficient
exchange rate
index
CPI
8 R & D activities represent the value of research & design expenditure divided by net revenue.
9 Equipment rate is net fixed assets divided by total number of employees.
10 CPI is consumer price index.
11 GDP is the amount of gross domestic production.
26
3.2 Research Question
The research question in this dissertation is whether other information in
the Ohlson valuation model is imperative or not, and what content should
be included in other information. According to the literature review, some
believe that other information could be retrieved from financial analysts’
forecasting information, constant proxies, corporate governance proxies,
and macroeconomic proxies (Liu and Ohlson, 1999; Myer, 1999a; Wu et al.,
2004; Hsieh, 2004). This paper applied the Ohlson valuation model with six
possible other information proxies, and test each of them in the Ohlson
valuation model in order to see whether or not the other information is#p#分頁(yè)標(biāo)題#e#
imperative in the valuation function. Furthermore, the empirical
investigation discussed the synergy of the Ohlson valuation model in the
Taiwan Security Exchange.
3.3 The Ohlson’s Valuation Model
Ohlson (1995) introduced the clean surplus accounting relation and residual
income valuation concept to enhance the equity valuation measurement.
Ohlson model was revised from dividend discount model and established by
three fundamental hypotheses. These three hypotheses result in market
value can be sufficiently interpreted by book values, abnormal earnings and
the other information. This paper will conduct Ohlson model to test the
validity of Ohlson’s valuation model in Taiwan Security Market.
27
The first assumption in Ohlson valuation model is a firm’s market value
equals to the present value of future expected dividends, and assuming all
investors are risk neutral and have homogeneous investment expectation.
Market risk-free rate satisfies investors’ minimum expected return and a
firm’s market value is calculated by this risk-free rate as its discount rate.
The present value of firm’s market value is,
Pt = Σ R
τ ∞
τ E dτ (PVED)(3-1)
Where,
Pt denotes a firm’s market value or stock price at time t;
dτ denotes net cash dividend distribution at time t;
Rf denotes risk-free rate pluses one;
Et denotes the expected value of available information at time t.
Second assumption is owner’s accounting data which has clean surplus
relation. Accounting data subsumes the clean surplus relation which
indicates that the changes in the book value are due to the changes from
cash dividends and earnings. Meanwhile, the dividends will reduce the
current book value but the dividend reduce will not affect the current
earnings. Using the clean surplus characteristics can substitute book values
and earnings for the present value of future expected dividends in PVED.
28
The clean surplus relation is illustrated by,
BVt = BVt-1 + Et - dt (CSR)(3-2)
Where,
BVt denotes a firm’s book value at time t;
Et denotes a firm’s earnings at time t;
dt denotes a firm’s net current dividends at time t.
According to PVED and CSR assumptions, the market value can be
illustrated by the residual income model,
Pt = BVt + Σ R
τ ∞
τ E Xτ
(3-3)
Where,
Pt denotes a firm’s market value or stock price at time t;
BVt denotes a firm’s book value at time t;
X
denotes abnormal earnings at time t;
Rf denotes risk-free rate pluses one;
Et denotes the expected value of available information at time t.
29
However, the present value of abnormal earnings contains future abnormal#p#分頁(yè)標(biāo)題#e#
earnings which is unobserved information (forecasting information), Ohlson
(1995) enhanced this valuation function by adding linear information
dynamics (LID) assumption which frames stochastic time-series behaviour
of the abnormal earnings. These two linear information dynamics are,
denotes abnormal earnings at time t;
v denotes available other information or non-accounting information at
time t, but the other information has not reflected on current
abnormal earnings;
ω is abnormal earnings persistent parameter of last period, 0 ω 1;
γ is other information persistent parameter of last period, 0 γ 1;
ε is disturbance terms;
ε is disturbance terms.
The formula (3-4) is modified autoregressive and the formula (3-5) is
regular auto regressive, so the value of ω and γ must be between 0 and 1.
In this case, the regression model will be stationary. The linear information
dynamics assumption aimed to avoid discretionary setting of the future
abnormal earnings and other information in the valuation model.
30
The refined formula is,
Pt = BVt + αX
+ αv (3-6)
Where,
Pt denotes a firm’s market value or stock price at time t;
BVt denotes a firm’s book value at time t;
X
denotes abnormal earnings at time t;
v denotes available other information at time t;
α is the parameter of abnormal earnings;
α is the parameter of the other information.
In formula (3-6), abnormal earnings can measure the current profitability but
the “other information” can further assess the future profitability of a firm.
However, Ohlson (1995) did not provide specific definition for the other
information, so most literatures built the other information proxies (or nonaccounting
information) by operating information, depending on individual
entity risk factors, economics conditions or even modifying the Ohlson
model to omit the other information(Myers, 1999a). Moreover, some relative
accounting-based empirical studies saw the other information as error items,
this aspect may result in estimated biases in the empirical research.
Therefore, this paper will refer to successful empirical investigations to
carry out appropriate the other information proxies for the valuation
application.
31
Except for accounting data from financial statements, enterprises’
operating risk factors can be utilized to interpret the possible volatility of
market values. Operating risk factors are classified as systematic risks (or
market risks) and unsystematic risks. Systematic risk is market wide, and
all enterprises will be affected in some degree. Also, systematic risk is
non-diversifiable, and it can influence all enterprises’ operating#p#分頁(yè)標(biāo)題#e#
performances and market values. Unsystematic risk is uncertainties about
a specific firm, and it is generally diversifiable. However, corporate
governance papers were proved that it can be significant unsystematic risk
which is hard to diversify. Therefore, this paper will apply the other
information proxies in these dimensions, one is unsystematic risk factors
and another is systematic risk factors. Thus, the concrete variables can be
expected to enhance the explanatory power of the Ohlson’s valuation
model.
Wu et al. (2004) adopted the qualified foreign institutional investors’ holding
rates (QFII) and directors’ and supervisors’ holding rates as the other
information proxies in the Ohlson’s valuation model. The results in Wu et
al.’s work (2004) with these two proxies were statistically significant;
moreover, the other information proxies were more significant than
abnormal earnings in some industries. Meanwhile, Hsieh (2004) also
supported that the directors’ and supervisors’ pledging rates were relative to
the valuation of firms’ market values in empirical finding. In systematic risk
factors, the proxies are the interest rates, the efficient exchange rate index,
and the consumer price index and these proxies will be carried out in the
32
Ohlson model application in this paper. These macro economic factors have
been proved that have statistical significance in relation to a firm’s market
price (Abell and Krueger, 1989; Li, 2002; Wu et al., 2004).
Following formula is added the other information proxies,
vδδ δ!!δ""δ##δ$$
(3-7)
Where,
vt denotes the other information in time t;
X1t denotes qualified foreign institutional investors’ holding rate in time t;
X2t denotes the directors’ and supervisors’ holding rate in time t;
X3t denotes the director’ and supervisors’ pledging rate in time t;
X4t denotes the interest rate in time t;
X5t denotes the efficient exchange rate index in time t;
X6t denotes the consumer price index in time t;
δ& is the parameters for the other information proxies;
' is error term.
33
The revised Ohlson’s valuation model in this paper is,
Pt = a1 + a2 BVt + a3 X
+ a4 X1t + a5 X2t+ a6 X3t + a7 X4t + a8 X5t + a9 X6t (3-8)
Where,
Pt denotes a firm’s market value or stock price in time t;
ai denotes parameters with each variables;
BVt denotes a firm’s book value in time t;
X
denotes a firm’s abnormal value in time t;
Xit denotes the other information proxies in time t.
This empirical study contains time-series and cross-sectional data, and this
analysis nature is also defined as panel data or longitudinal data. In#p#分頁(yè)標(biāo)題#e#
longitudinal data analysis, the fixed-effect (or dummy variable) regression
model is frequently adopted. The paper will also apply the empirical
research by utilizing the fixed-effect linear regression model and preceding
research have proved that the fixed-effect linear regression model is
appropriate in this research subject (Wu et al., 2004; Hsieh, 2004). The
random-effect regression model postulates that all data in sample group
have identical characteristic of the research subject (Allison, 2005).
However, this paper attempts to examine the correlation between Ohlson’s
valuation model and market prices in six industries, it is impossible for
every industry equips with identical characteristics in their operating
activities and performance. Thus, this paper conducts the fixed-effect linear
34
regression model to inference what characteristics this sample owns.
Owing to the purpose of a remarkable accounting-based valuation model,
any empirical evidence should be conscientious. Moreover, Allison (2005)
stated that the fixed-effect linear regression model is possible to control for
all possible characteristics of individuals in the sample even without
measuring them or even these characteristics do not vary over time. The
formula (3-8) will be refined as formula (3-9),
Pit = a1i + a2 BVt + a3 X
+ a4 X1t + a5 X2t+ a6 X3t + a7 X4t + a8 X5t + a9 X6t (3-9)
Where,
Pit denotes an individual firm’s market value in time t;
a1i denotes a set of fixed parameters in the fixed-effect regression model;
ai denotes parameters for each variable item.
In formula (3-9), a1i is assumed as a set of fixed parameters which are
unrelated with book values, abnormal earnings, and the other information.
In the fixed-effect linear regression model, the a1i represents all stable
characteristics of individual variable (Allison, 2005). In this work, the data
analysis will be conducted by adopting formula (3-9) and the purpose of
formula (3-9) is to examine the validity and usefulness of Ohlson’s valuation
model.
35
3.4 Data Collection
The research centres on an empirical study involving listed firms in Taiwan
Security Exchange. Longitudinal data were collected primarily by public
financial statements and historical market prices records. Market values,
book values, abnormal earnings, the directors’ and supervisors’ holding
rates, the directors’ and supervisors’ pledging rates, the QFII holding rates,
the interest rates, the efficient exchange rate index and the consumer price
index were all collected from Taiwan Economic Journal Data Bank (TEJ)12.
The data collection sessions were during 36 quarters13 which were from
2000 Q1 to 2008 Q4. Six industries were participated in this research and
they were Building & Construction, Chemical & Biotechnology, Electronics,#p#分頁(yè)標(biāo)題#e#
Foods, Plastics, and Textiles. All required measurements in the fixed-effect
regression model will be extracted from Taiwan Economic Journal Data
Bank. The primary criterions for collecting data are as follows:
i. The selected firms must be listed on the Taiwan Security Exchange during
investigation period;
ii. The firm will be exempted if the firm changes it transaction manner (for
example: full-cash delivery stock);
12 Taiwan Economic Journal Data Bank was found in 1990, and the corporate website is
13 The first quarter is January to March, the second quarter is April to June, the third quarter is
July to September, and the final quarter is October to December. This definition is dependent on
Taiwan Security Exchange regulation.
36
iii. The firm provides complete financial statements and other required
information during investigation period;
iv. Banks, securities firms, and insurance firms are excluded due to their
specific characteristics;
v. Book value for every firm in any fiscal year should not be negative.
Table 2
Sample firms
Industry Number of firms Percentage
Building & Construction 42 12.24%
Chemical & Biotechnology 31 9.04%
Electronics 178 51.89%
Foods 22 6.41%
Plastics 21 6.12%
Textiles 49 14.3%
Total 343 100%
3.5 Sample Collection
Six industries14 in Table 2 were selected as inference materials of empirical
research. These six industries were also selected in Wu et al. (2004) paper,
and this dissertation would like to choose the same industries in order to
see the difference between the prior empirical investigation and this
14 Industry categories are in accordance with the industry categories in the Taiwan Economic
Journal Data Bank.
37
empirical investigation. These six industries include traditional industries
and high-tech industries which can represent sufficient features of the
Taiwan security market. 343 listed firms in six industries were involved in
the investigation and these 343 firms were regularly operated during data
collected sessions. Meanwhile, the sample duration is between 2000 and
2008 which is 4 years longer than Hsieh (2004) and Wu et al’ (2004) works
in order to test the validity of Ohlson valuation model in a longer horizon.
Moreover, the quarterly data was utilized in the empirical investigation due
to the other information proxies were generally measured by quarterly
intervals in data bank. Also, the quarterly data was adopted by both Wu et
al. (2004) and Hsieh’s (2004) works which were proved the quarterly data is
a proper interval for this subject.
3.6 Variable Measurement
Followings are concepts and the specifications of variables which are
implemented in the empirical test, and all numerical data are extracted from#p#分頁(yè)標(biāo)題#e#
Taiwan Economic Journal Data Bank:
(1) Stock price (Pt)
This paper adopts average quarterly closing market prices in time t as a
firm’s dependent variable in the Ohlson’s valuation model.
(2) Book value per share (BVt)
Book value or net value per share is determined by the firm’s net assets per
38
share in time t.
BVt = (assets – liabilities – preference capital)/ [(ordinary capital + reserves
for increasing capital – treasury shares (10) ( 10]
(3) Earning per share (EPSt)
Earning per share is normally an indicator of a firm’s operating performance.
Thus, earning per share is an investment return measurement for ordinary
shareholders. In this empirical test, earning per share is used to calculate
the abnormal earnings for quarterly data in time t.
EPS = (average quarterly earnings –average quarterly preference
dividends) / quarterly weighted number of ordinary shares
(4) Abnormal earning (X
)
Abnormal earning is the residual of current earnings per share at time t
minus an interest charge on the beginning book value (Ohlson, 1995).
According to Ohlson’s valuation theory (1995), while the abnormal earning
is positive, it represents the difference between a firm’s book value and
market value and the difference could be a performance measurement in
the valuation function in time t.
X
= EPSt – (three-month interest rate15 ( beginning book value balance)
15 The three-month risk-free interest rate is Taiwan First Commercial Bank three-month nominal
interest rate
39
(5) Other information proxies (Xit)
This paper will conduct six “other information” proxies and their
measurements are as follows,
i. Qualified foreign institutional investors’ holding rate (X1t)
X1t = average quarterly QFII holding shares in time t / (quarterly weighted
ordinary shares in time t– preferred shares in time t)
ii. Directors’ and supervisors’ holding rate (X2t)
X2t = average quarterly directors’ and supervisors’ holding shares in time t /
(quarterly weighted ordinary shares in time t– preferred shares in time t)
iii. Directors’ and supervisors’ pledging rate (X3t)
X3t = average quarterly directors’ and supervisors’ pledging shares in time t
/ quarterly weighted directors’ and supervisors’ holding shares in time t
iv. Interest rate (X4t)
X4t = Taiwan First Commercial Bank three-month nominal interest rate in
time t
v. Efficient exchange rate index (X5t)
X5t =New Taiwan Dollar efficient exchange rate index16 in time t
vi. Consumer price index (X6t)
16 The New Taiwan dollar efficient exchange rate index is settled that the rate in 2000 is 100
40
X6t = adjusted consumer price index17 in time t#p#分頁(yè)標(biāo)題#e#
4. Results
In this section, the paper has divided the analysis into three parts. The first
part tests the original Ohlson valuation model without other information
variables. This part aims to see the correlation between market prices,
book values, and abnormal earnings in different industries during thirty-six
quarters. Through examining the correlation between book values,
abnormal earnings, and market prices, the usefulness of the Ohlson
valuation model without other information can be presented. Subsequently,
the second part conducted regression analysis by adding different other
information proxies to the Ohlson valuation formula. This part attempted in
order to test which other information proxies can better improve the
explanatory power of the Ohlson valuation model, and six factors were
investigated separately, covering 343 firms over 36 quarters. Among these
other information, the analysis classified them into two groups:
unsystematic factors and systematic factors. Thus, the second part also
compared the interpretation power between different models, by utilizing
groups of unsystematic risk factors and systematic factors as other
information proxies in the Ohlson valuation model. Finally, it conducted the
revised Ohlson model with the following variables: book values, abnormal
earnings, and six possible other information proxies. Then, this section
demonstrated every fixed-effect regression model result in detail. In order
to investigate which other information proxy can be better applied in
17 The adjusted consumer price index is settled that the CPI=100 in 2006.
41
Ohlson’s valuation model; the following models will be analysed separately.
Table 3
Revised Ohlson model by adding different other information proxies
Model Formulas b
Model 1a Pit = a1i + a2 BVt + a3 X
Model 2 Pit = a1i + a2 BVt + a3 X
+ a4 X1t
Model 3 Pit = a1i + a2 BVt + a3 X
+ a4 X2t
Model 4 Pit = a1i + a2 BVt + a3 X
+ a4 X3t
Model 5 Pit = a1i + a2 BVt + a3 X
+ a4 X4t
Model 6 Pit = a1i + a2 BVt + a3 X
+ a4 X5t
Model 7 Pit = a1i + a2 BVt + a3 X
+ a4 X6t
Model 8 Pit = a1i + a2 BVt + a3 X
+ a4 X1t + a5 X2t+ a6 X3t
Model 9 Pit = a1i + a2 BVt + a3 X
+ a4 X4t + a5 X5t + a6 X6t
Model 10 Pit = a1i + a2 BVt + a3 X
+ a4 X1t + a5 X2t+ a6 X3t + a7 X4t + a8 X5t +
a9 X6t
42
a. Model 1 is the original Ohlson model without any other information proxies
b. X1t is the qualified foreign institutional investors’ holding rates in time t; X2t is
the directors’ and supervisors’ holding rate in time t; X3t is the directors’ and
supervisors’ pledging rate in time t; X4t is the interest rate in time t; X5t is the
efficient exchange rate index in time t; X6t is the consumer price index in#p#分頁(yè)標(biāo)題#e#
time t
The quantitative regression analysis utilized the STATA18 statistic software
package to implement fixed-effect multiple regression models. There were
343 firms and six industries in total in this empirical investigation, and the
test period was from first quarter in 2000 to last quarter in 2008.
4.1 Original Ohlson’s Model without Other Information Variable
According to the formula (3-9), a firm’s market value is relative to several
variables. In the Model 1, the underlying multiple linear regression formula
omits the other information (v). Therefore, this regression only examined
the relationship among book values, abnormal earnings and market prices
in 343 firms.
18 STATA was produced by Stata Corp LP, the official website is: http://www.stata.com/
43
Table 4
Original Ohlson model without other information variable (Model 1) a
Bui. & Con. b Che. & Bio. Electronics Foods Plastics Textiles
BVt
1.656***
(20.74)
1.585***
(16.72)
R2 0.3686 0.6058 0.4652 0.3886 0.7659 0.4517
N 1512 1116 6408 792 755 1763
a. The valuation formula is: Pit = a1i + a2 BVt + a3 X
b. *, **, *** denote statistical significance at the 10 percent, 5 percent and 1
percent levels respectively (two-tailed t statistics are provided in
parentheses)
Table 5
The effect on market price in Model 1
Industry Book Value a Abnormal Earning
Building & Construction + +
Chemical & Biotechnology + +
Electronics + +
Foods + +
Plastics + +
Textiles + +
a. + is positive to the market price, and – is negative to the market price
44
Table 4 reveals that the values of adjusted R2 were between 0.3686 and
0.7659, and the book value and abnormal earning were both statistically
significant in the fixed-effect regression models in every industries. The
best fitness regression model in this result was plastics industry which
presented fairly strong interpretation power in Ohlson valuation model. In
the Ohlson valuation model, book value information and abnormal earning
information are easier to collect by financial statements and accounting
publications. Thus, the adequate interpretation power in the original Ohlson
model can lead to apply this accounting-based valuation model more
practical. However, general applications in original Ohlson valuation model
cannot derive fairly strong interpretation results, more information should
be considered in the valuation function. Following analysis will add the
other information to enhance the empirical generalization.
Table 5 illustrates how book values and abnormal earnings affect the
market prices and all results shows that both book values and abnormal
earnings were both positive to the market prices in six industries. Therefore,#p#分頁(yè)標(biāo)題#e#
while book values and abnormal earnings increase the market prices
increase in fixed-effect regression Model 1. This result supports the Ohlson
valuation theory which defines the market value is positive to book values
and abnormal earnings (Ohlson, 1995).
45
4.2 Revised Ohlson’s Valuation Model
According to Table 4, the result was consistent with the Ohlson valuation
model theory, and this section adds other information proxies to enhance
the empirical research on the Ohlson valuation model. In order to compare
the usefulness among other information proxies in valuation function, the
first analysis process in this section testified different other information
proxies in the Ohlson valuation model. In the second analysis process, the
paper divided the six other information proxies into two groups; one is for
unsystematic risk factors and the other is for systematic risk factors. The
third part in this section applied the Ohlson valuation model with book
values, abnormal earnings and six other information proxies for
examination of the synergy of these possible other information proxies.
Through these three stages of examination, the analysis discussed in detail
the validity of the other information variables, and found out which other
information proxies could be better applied to specific industries in practice.
46
4.2.1 Comparison among six possible other information proxies
b. *, **, *** denote statistical significance at the 10 percent, 5 percent and 1
percent levels respectively (two-tailed t statistics are provided in
parentheses).
Table 6 presents the application of fixed-effect regression in Model 2. The
qualified foreign institutional investors’ holding rate represented the only
other information proxy in Model 2, and all industries were statistically
significant in the other information variable except Textiles industry. This
indicates that the qualified foreign institutional investors’ holding rate
information is not appropriate to evaluate market prices in Textiles industry.
Meanwhile, the values of adjusted R2 in Ohlson valuation model were
47
improved in Chemical & Biotechnology, Electronics, and Foods industries.
This improvement demonstrates that qualified foreign institutional investors’
holding rate information develops the Ohlson valuation model in these
industries. On the contrary, the information content of the qualified foreign
institutional investors’ holding rates does not improve the validity of other
information variable in evaluating market prices of Building & Construction,
Plastics, and Textiles industries.
Table 7
In Table 7, the Model 3 adds directors and supervisors’ holding rates as the
other information proxy. Through the regression analysis, the t tests in other#p#分頁(yè)標(biāo)題#e#
information variable were statistically significant in following industries
which were Building & Construction, Electronics, and Textiles. Namely, the
directors and supervisors’ holding rate might be not an essential
information variable in Ohlson valuation model in Chemical &
Biotechnology, Foods, and Plastics industries. Meanwhile, the
interpretation power of regression models were improved in Building &
Construction and Textiles industries. This result implicated that the
directors and supervisors’ holding rates were relative to the firm value in
Building & Construction and Textiles industries and the correlation between
directors and supervisors’ holding rates and market prices were positive.
Thus, the higher proportion in directors and supervisors’ holding rate
indicated better operating performance in Building & Construction and
Textiles industries in this empirical investigation.
49
b. *, **, *** denote statistical significance at the 10 percent, 5 percent and 1
percent levels respectively (two-tailed t statistics are provided in
parentheses).
Table 8 indicates that the influence by adding the directors’ and supervisors’
pledging rate into the other information proxy in the Model 4. The effect on
adding directors and supervisors’ pledging rate as other information proxies
were not statistically significant in Plastics and Textiles industries. Only in
Building & Construction, Chemical & Biotechnology, Electronics, and Foods
industries, the directors and supervisors’ pledging rate information
influenced the market prices. Meanwhile, the directors and supervisors’
pledging rates in Building & Construction, Chemical & Biotechnology, and
Electronics had negative effects on market prices. Thus, this result
50
illustrated that directors and supervisors’ pledging rates destroyed the
operating performance and market expectation in Building & Construction,
Chemical & Biotechnology, and Electronics industries. Moreover, the
Ohlson valuation models were improved by directors and supervisors’
pledging rate information in Building & Construction, Electronics, and
Foods industries.
Table 9
Ohlson’s valuation model with interest rate (X4t) (Model 5) a
b. *, **, *** denote statistical significance at the 10 percent, 5 percent and 1
percent levels respectively (two-tailed t statistics are provided in
parentheses).
51
In Table 9, the Model 5 inputted one of systematic risk factors into Ohlson
valuation model and this other information proxy was three-month interest
rate. The result showed striking effects of interest rates on Ohlson valuation
function in all industries. This empirical investigation also indicated that#p#分頁(yè)標(biāo)題#e#
interest rate is a rather appropriate other information proxy than prior three
unsystematic risk factors (Model 2, Model 3, and Model 4)in six industries.
Interest rates had positive effects on the market prices in all industries
except Building & Construction. That is to say, interest rate fluctuation had
opposite movement compared to market values of Building & Construction
during empirical observation. Meanwhile, after adding interest rate into
Ohlson valuation model, it improved the values of adjusted R2 of regression
models in Chemical & Biotechnology, Electronics, Foods, Plastics, and
Textiles. Moreover, interest rates were positively related to market prices in
these five regression models.
52
Table 10
b. *, **, *** denote statistical significance at the 10 percent, 5 percent and 1
percent levels respectively (two-tailed t statistics are provided in
parentheses).
In Table 10, the efficient exchange rate index was added as an other
information proxy in Model 6, and market prices were all associated
significantly with this single other information proxy. Efficient exchange rate
indices in all industries were positively related to market prices, except for
efficient exchange rate indices in Building & Construction industry. This
result showed that when the efficient exchange rate index was lower, the
higher the market price of the Building & Construction industry would be
during the observation period. This situation might indicate that when the
exchange rate rose, those investors believed more contingent exchange
53
losses would be recognised in financial statements which destroyed the
current operating performance in Building & Construction industry.
Nevertheless, an efficient exchange rate index seems to be positively
connected to market prices in the Chemical & Biotechnology, Electronics,
Foods, Plastics, and Textiles industries. All regression results in Model 6
were improved by adding the efficient exchange rate index as the other
information proxy, and this result also supports the empirical study of Wu et
al. (2004). Therefore, in this empirical evidence, the efficient exchange rate
index is an appropriate other information proxy.
Table 11
Ohlson valuation model with consumer price index (X6t) (Model 7) a
Bui. & Con. b Che. & Bio. Electronics Foods Plastics Textiles
b. *, **, *** denote statistical significance at the 10 percent, 5 percent and 1
percent levels respectively (two-tailed t statistics are provided in
parentheses).
54
Table 11 presents the fixed-effect regression results of the Ohlson valuation
model, which includes the consumer price index as the other information
proxy. The correlation coefficient turned out to be negative in the
Electronics, Plastics, and Textiles industries, and to be positive and#p#分頁(yè)標(biāo)題#e#
significant in the Building & Construction and Foods industries. Only in the
Chemical & Biotechnology industry, did the consumer price indices have no
significant influence on market prices in the regression results. This finding
indicated that when the purchasing power decreased, investors expected
market prices in Electronics, Plastics, and Textiles industries would raise. In
contrast, when the purchasing power increased, investors expect market
prices would rise in the Building & Construction and Foods industries.
However, the consumer price index was not appropriate as an “other
information” proxy in the Chemical & Biotechnology industry due to the
poor correlation with market prices in this empirical investigation. Also, the
consumer price indices information proved the values of the adjusted R2 in
regression models in all industries, except for Chemical & Biotechnology.
Moreover, according to the results of Model 5, Model 6, and Model 7, the
consumer price index was a worse information indicator in the Ohlson
valuation function among systematic risk factors.
55
4.2.2 Comparison in Groups of Unsystematic and Systematic Risk Factors
After presenting possible individual other information proxy in the Ohlson
valuation model, this section groups all discussed other information proxies
into two parts. Model 8 contains unsystematic risk factors in other
information proxies, and Model 9 contains systematic risk factors in other
information proxies. The purpose of this research stage is to compare
unsystematic risk factors with systematic risk factors, in order to
understand which group can be a better other information proxy in the
Ohlson valuation model in the Taiwan security market.
56
Table 12
Ohlson valuation model with unsystematic risk information (Model 8) a
Bui. & Con. b Che. & Bio. Electronics Foods Plastics Textiles
b. *, **, *** denote statistical significance at the 10 percent, 5 percent and 1
percent levels respectively (two-tailed t statistics are provided in
parentheses).
Table 12 provides the results of the Ohlson valuation model, which includes
book values, abnormal earnings, and unsystematic risk factors as other
information proxies. Unsystematic risk factors are qualified foreign
institutional investors’ holding rates (X1t), directors’ and supervisors’ holding
rates (X2t), and directors’ and supervisors’ pledging rates (X3t). Three
57
unsystematic risk factors were all statistically significant in the Building &
Construction, Electronics, and Plastics industries. However, the value of
adjusted R2 in the regression results was only improved in the Electronics
industry data. This result indicates that even unsystematic risk factors were
all significant in valuation functions in these three industries; the#p#分頁(yè)標(biāo)題#e#
interpretation power in the valuation model was still not good enough in the
Building & Construction and Plastics industries. In the Chemical &
Biotechnology and Foods industries, unsystematic risk factors were
significant in regression results and the values of the adjusted R2 were
improved by other information proxies. Thus, except for directors’ and
supervisors’ holding rates (X2t), unsystematic risk factors properly
enhanced the Ohlson valuation functions in the Chemical & Biotechnology
and Foods industries. Only one unsystematic risk factor was statistically
significant in the Textiles industry and the value of the adjusted R2 was not
improved by the unsystematic risk group (Model 8).
58
Table 13
Ohlson valuation model with systematic risk information (Model 9) a
Bui. & Con. b Che. & Bio. Electronics Foods Plastics Textiles
b. *, **, *** denote statistical significance at the 10 percent, 5 percent and 1
percent levels respectively (two-tailed t statistics are provided in
parentheses).
Table 13 lists regression results of the Ohlson valuation model, which
includes book values, abnormal earnings, and systematic risk factors as
other information proxies (Model 9). Here, systematic risk factors here are
the interest rate, the efficient exchange rate index and the consumer price
index. Values of R square in regression models were all improved by the
59
systematic risk group in six industries. Although, systematic risk factors
were not all significant in the Food, Plastics, and Textiles industries, the
interpretation power of the Ohlson valuation models were still improved in
these industries. Compared to unsystematic risk factors, systematic risk
factors better provided useful other information in the Ohlson valuation
model in empirical investigations. This result implies that macroeconomic
factors significantly influenced a firm’s value in the accounting-based
valuation function. Thus, according to this empirical research, systematic
risk factors could be proper other information proxies to enhance the
validity of the Ohlson valuation model in the Taiwan security market.
60
4.2.3 Revised Ohlson’s Model
Table 14
Revised Ohlson’s valuation model (Model 10) a
Bui.& Con. b Che. & Bio. Electronics Foods Plastics Textiles
b. *, **, *** denote statistical significance at the 10 percent, 5 percent and 1
percent levels respectively (two-tailed t statistics are provided in
parentheses).
In Table 14, Model 10 operated six possible other information proxies in the
Ohlson valuation model and presented the regressive results of six
industries. All variables were statistically significant, related in Model 10 in
the Electronics industry, and this finding indicates that six other information#p#分頁(yè)標(biāo)題#e#
proxies provide incremental information value in the Ohlson valuation
model. The qualified usefulness of the Ohlson valuation application in
Electronics was the reason why Hsieh’s (2004) article only examined the
Electronics industry in the Ohlson valuation discussion. However, in order
to test the validity of the Ohlson valuation model applications in the Taiwan
security market, it essential to include more than the Electronics industry. In
Model 10, the interest rate, consumer price index, and the qualified foreign
institutional investors’ holding rate were considerably important variables in
the Ohlson valuation model in six industries, and these other information
proxies provided incremental information content in the Ohlson valuation
function. Thus, values of R2 were all improved in regression results except
those for the Plastics industry. Although the values of adjusted R2 were
between 0.4114 and 0.7549, this examination of the usefulness of other
information proxies can still be used as guidance for further research in this
field.
62
Table 15
significance in the quarter market
price; + means variables are positive to the market price; – means variables
are negative to the market price
Table 15 lists the impact of variables in the Ohlson valuation function, and
three symbols (+,-,*) represent different effect direction on market prices in
Model 10. The + symbol means that when the variable increased during the
observation period, the market price in the accounting-based valuation
function also increased. The – symbol indicates that the market prices were
negatively correlated with variables in the empirical regression models.
Finally, the * star symbol indicates that variables were not statistically
significant on market prices during the observation period in the empirical
regression models.
Book values and abnormal earnings were all positively correlated with
market prices in six industries, and this result sufficiently supported the
63
Ohlson valuation theory (Ohlson, 1995). The first item of other information,
qualified foreign institutional investors’ holding rates (X1t), was positively
related to the market prices in all industries except for the Textiles industry.
The reason is that the qualified foreign institutional investors’ holding rates
in Textiles were fairly low compared to other industries on average, so this
other information proxy was not significant to market prices in Textiles firms.
Nevertheless, the qualified foreign institutional investors’ holding rates were
all significantly and positively related to market prices in other industries
and this finding supports the theory that qualified foreign institutional
investors can be efficient external governance to supervise a firm’s#p#分頁(yè)標(biāo)題#e#
operating performance (Pound, 1988). Another information proxy, the
directors’ and supervisors’ holding rates (X2t), provided internal governance
mechanism information as incremental information in the Ohlson valuation
function. In the results of empirical regression models, coefficients were
positively correlated with market prices in the Building & Construction,
Electronics, and Plastics industries. This result supports that fact that
directors and supervisors are more willing to better execute a firm’s
business while they own more shares in this specific firm (Jensen and
Melking, 1976). In the Chemical & Biotechnology, Foods, and Textiles
industries, the directors’ and supervisors’ holding rates were not important
enough to be other information proxies in the Ohlson valuation model. The
third other information proxy is the directors’ and supervisors’ pledging rate
(X3t). The coefficients of the directors’ and supervisors’ pledging rates in the
Chemical & Biotechnology, Electronics, and Textiles industries were
negative in the regression results, which supports the theory that the
64
directors’ and supervisors’ pledging rate weakens the correlation in the
security valuation function (Kao, 2002). The directors’ and supervisors’
pledging rate was positively related with the market price in the Foods
industry and had no statistically significant influences in Building &
Construction and Plastics industries. Interest rate is the fourth other
information proxy (X4t), and the coefficients in Model 10 were positively
related to the market prices in all industries, except for Building &
Construction. However, the regression results in Building & Construction
supported a statement that interest rate and market price are related, but in
opposite directions (Abdullah and Hayworth, 1993). In efficient exchange
rate index information (X5t), empirical evidence indicated that the exchange
rate is positively related to the market price, especially in short-term
movement (Aggarwal, 1981). The statistical results for Building &
Construction, Chemicals & Biotechnology, and Electronics were consistent
with prior empirical evidence. The efficient exchange rate indices were not
statistically significant in the Foods, Plastics, and Textiles industries in
Model 10. Consumer price indices (X6t) were positively related to market
prices in the Building & Construction, Chemicals & Biotechnology, and
Foods industries, and negatively related to market prices in the Electronics
and Plastic industries. Moreover, the consumer price indices were not
statistically significant in the Textiles industry. The Ohlson valuation model
provides an internally consistent set of valuation equations, accounting#p#分頁(yè)標(biāo)題#e#
information, and other information which can be reasonably applied
separately (Lo and Lys, 2000). Therefore, even when this dissertation
applied different other information proxies in models, the results still
65
performed the characteristics of Ohlson’s theory. This empirical evidence
supports the fact that the Ohlson valuation model has validity and would be
useful to apply in the Taiwan Security Exchange.
Table 16
Compare values of adjusted R2 among ten models
Industry Bui.& Con. Che. & Bio. Electronics Foods Plastics Textiles
M1 0.3686 0.6058 0.4652 0.3886 0.7659 0.4517
M2 0.3472 0.6081 0.4923 0.4305 0.772 0.4507
M3 0.3808 0.6063 0.4416 0.3883 0.7633 0.4582
M4 0.3745 0.6032 0.4671 0.3866 0.7584 0.4438
M5 0.368 0.6567 0.539 0.4232 0.7766 0.4662
M6 0.3703 0.6402 0.5406 0.3946 0.7788 0.4628
M7 0.4119 0.6043 0.4836 0.4303 0.7843 0.453
M8 0.3666 0.6073 0.4735 0.4262 0.7246 0.4478
M9 0.4218 0.6575 0.5483 0.4789 0.7919 0.4663
M10 0.4114 0.6588 0.5563 0.5097 0.7549 0.4575
Through whole comparison among ten revised Ohlson valuation models
(M1~M10), the values of adjusted R2 can illustrate which revised Ohlson
valuation model was most properly adopted in six industries during the
observation session. Table 16 indicates that M9 is the most powerful
regression model in the Building & Construction, Plastics, and Textiles
industries, for interpreting the correlation between variables and market
prices in the Ohlson valuation function. The values of the adjusted R2 in M9
66
were 0.4218, 0.7919, and 0.4663 in the Building & Construction, Plastics,
and Textiles industries. M10 also displays a better explanatory role in the
Ohlson valuation theory in the Chemical & Biotechnology, Electronics, and
Foods industries. The values of the adjusted R2 in M10 were 0.6588,
0.5563, and 0.5097 in the Chemical & Biotechnology, Electronics, and
Foods industries. This result indicates that the revised Ohlson M10, which
includes six other information proxies, may not be the best forecasting
model for every industry. In this empirical investigation, market prices in the
Building & Construction, Electronics, and Foods industries were more
significantly relative to systematic risk factors. Although this empirical study
was only applied to six industries and nine years of quarterly data was
adopted, further research could extend these results to discover more
specific Ohlson valuation models for other industries in the long term.
Despite the fact that models M9 and M10 had higher explanatory power in
fixed-effect regression results, M7 was also considerably significant for the
Building & Construction and Plastics industries. According to this result, it
demonstrates that the fluctuation in the consumer price index influenced#p#分頁(yè)標(biāo)題#e#
significantly the market prices of the Building & Construction and Plastics
industries. Further study can design possible hypotheses to investigate
what is the relationship between the consumer price index and the Building
& Construction (or Plastics) industries, and to improve the design of the
other information proxy. For the Textiles industry, M5 presented a better
explanatory power than other revised Ohlson valuation models, except for
M9 in regression results. M5 demonstrated the impact of adopting interest
rates as the other information proxy in the Ohlson valuation function, the
67
more concrete relationship between interest rate and the market price
movement of the Textiles industry is also worth examining further. More
accurate and unbiased other information proxies in the Ohlson valuation
model will develop considerably the accounting-based valuation validity
and usefulness. Meanwhile, the results in Table 16 support previous
research findings that other information should not be omitted in the
valuation processes, because adding other information proxies improves
the interpretation ability of the Ohlson valuation model (Liu and Ohlson,
1999).
5. Analysis of Results
The research question in this dissertation is whether other information in
the Ohlson valuation model is imperative or not, and what content should
be included in the other information. Therefore, six other information
proxies were separately added into the Ohlson valuation model and applied
to six industries. The empirical evidence indicated that other information
proxies were generally significant in the valuation function in six industries.
In particular, systematic risk factors were most important other information
proxies in the Ohlson valuation models in this empirical study. Meanwhile, a
longer sample session improved the regression results in the Ohlson
valuation model. For instance, in this dissertation, two other information
proxies, which are the qualified foreign institutional investors’ holding rate
and the directors’ and supervisors’ holding rate, were as the same as Wu et
al.‘s work, but this dissertation derived dissimilar correlation results in these
68
two other information proxies. The qualified foreign institutional investors’
holding rates were not significant in the Building & Construction industry in
Wu et al.’s paper, but the qualified foreign institutional investors’ holding
rates became positively related to market prices in the Building &
Construction industry in this dissertation. In the Building & Construction,
Electronics, and Plastics industries, the directors’ and supervisors’ holding
rates were also not significant in prior research, but the results turned into
positively related to market prices in this dissertation. This improvement#p#分頁(yè)標(biāo)題#e#
can be explained by the fact that Wu et al.’s (2004) sample duration was
only 21 quarters of data, while this dissertation employed 36 quarters of
data. In the longer term, the regression analysis can more rigorously
examine the relationship between other information proxies and market
prices in the Taiwan security markets. Another finding in this dissertation is
that the significance of the interest rate proxy was more correlated with
market prices than book values and abnormal earnings. This finding
supports prior literature and proves that other information is imperative and
should not be omitted from the empirical research (Liu and Ohlson, 1999;
Wu et al., 2004; Hsieh, 2004). The importance of interest rates in the
accounting-based valuation model indicates that macroeconomic
conditions are also pertinent to a firm’s market value in the Taiwan security
market. Therefore, further research on the Ohlson’s valuation model
application may put an emphasis on searching systematic risk factors and
add these systematic risk factors into other information variables.
According to the empirical investigation, this dissertation supports the idea
that other information proxies in the Ohlson valuation model can be
69
adopted by analysts’ forecasting information, corporate governance
mechanism information, and macroeconomic information (Liu and Ohlson,
1999; Wu et al., 2004; Hsieh, 2004).
There are some limitations in this dissertation. The first limitation is that
quarterly data was adopted, rather than daily data, and the reason for this
is that Taiwan Economic Journal Data Bank only provides quarterly data in
these other information proxies- the directors’ and supervisors’ holding rate,
the directors’ and supervisors’ pledging rate, the efficient exchange rate
index, and the consumer price index. Therefore, this empirical investigation
can be further developed by utilizing daily data, which can observe every
variable in the Ohlson valuation model with daily market prices, in a more
rigorous research process. Another limitation is that there are twenty-three
industries19 in total in the Taiwan security markets, but only six core
industries were involved in this empirical study. The sample collection ofthis subject can be improved by increasing industries in the regressionanalysis, to examine the validity and usefulness in other industries, whichwere omitted in this dissertation. Moreover, this dissertation did notconsider specific discount rates in different firms in the Ohlson’s valuationmodel. This limitation can also be improved by more thorough research ofspecific characteristics of each industry or firm, and determined certaindiscount rate in the Ohlson valuation model (Lo and Lys, 2000).
19 The industry number is extracted from Taiwan Industrial Association and the official website is#p#分頁(yè)標(biāo)題#e#
70
Further research in applications of the Ohlson’s valuation model in TaiwanSecurity Exchange can add other industries which were not taken into
consideration in this dissertation. Moreover, the method of adopting thefixed-effect regression model can be substituted by other regressionmodels in order to compare which technology is the best tool in analysingthe validity and usefulness of the Ohlson’s valuation model.
6. Conclusion
Accounting-based valuation models or functions are imperative to investorsin the security markets. After growing agency problems and businessfrauds in the Taiwan security market, fundamental analyses of marketprices are in wide demand. Investors or funds managers have to collect notjust market price dynamic information, but also fundamental operatingactivities information in firms. Therefore, accounting-based models havebeen studied for years. Ohlson (1995) provided the residual incomevaluation model and created linear information dynamics into valuationfunction, which improved the accounting-based valuation models in severalways. Firstly, book values and abnormal earnings can properly evaluate themarket prices, and linear information dynamics can consider futureinformation in an accounting-based valuation. Owing to the linearinformation dynamics, the other information can be applied validly in theaccounting-based valuation model, which originally could not conclude anyfuture events in the valuation processes. The other information strengthensaccounting-based valuation characteristics and promotes a higher value ofthe adjusted R2 in the Ohlson regression model. These successes haveinspired more and more researchers do studies into the Ohlson valuationmodel.
The lack of definition of the other information variables in the Ohlsonvaluation model results in different discussions on the nature andcharacteristics of other information. The other information is pertinent
information in addition to book values and abnormal earnings in accounting
data. Therefore, this dissertation has discussed whether the other
information is imperative or not, and whether the other information should
be included in the Ohlson valuation model. The empirical investigation
designed the examination of the original Ohlson valuation model in the
Building & Construction, Chemical & Biotechnology, Electronics, Foods,
Plastics, and Textiles industries during the period 2000 to 2008. The result
of the original Ohlson valuation model analysis discovered that book values
and abnormal earnings were both positively correlated with market prices,
which supported Ohlson’s valuation theory. Subsequently, the empirical
investigation added six other information proxies which were: the qualified
institutional investors’ holding rate, the directors’ and supervisors’ holding
rate, the directors’ and supervisors’ pledging rate, the interest rate, the
efficient exchange rate index, and the consumer price index into the Ohlson#p#分頁(yè)標(biāo)題#e#
valuation model. The research applied other information proxies, one by
one, in six industries and the results illustrated that these other information
proxies can improve the values of the adjusted R2 in regression valuation
72
models. That is to say, adding these other information proxies can lead to
better interpretation ability, by accounting and non-accounting information
in security valuation functions. Moreover, the findings indicate that some
other information proxies in this empirical investigation were more pertinent
and significant than book values and abnormal earnings in the Ohlson
valuation model. This implication demonstrates that other information is
imperative in the Ohlson valuation model, and the other information should
not be omitted in. In Model 10, it shows these six other information proxies
are especially appropriate in the Plastics, Chemical & Biotechnology, and
Electronics industries. The values of the adjusted R2 in the Plastics,
Chemical & Biotechnology, and Electronics industries were 0.7549, 0.6588,
and 0.5563, respectively.
The result of this dissertation supports the Ohlson valuation theory (1995),
Wu et al.’s (2004) and Hsieh’s (2004) empirical applications in the Taiwan
security market. Meanwhile, the empirical investigations of the Taiwan
security market in this dissertation also proved the essential status of other
information in the Ohlson valuation model.
73
Reference
Abdullah, D., Hayworth, A. & Steven, C. (1993) ‘Macroeconometrics of
Stock Price Fluctuations’, Quarterly Journal of Business and Economics,
32, 50-68.
Abell, J.D. and Krueger, T. M. (1989) ‘Macroeconomic Influences on Beta’,
Journal of Economics and Business, 41, 185-93.
Aggarwal, R. (1981) ‘Exchange Rate and Stock Prices: A Study of the US
Capital Markets under Floating Exchange Rates’, Akron Business and
Economic Review, 12, 7–12.
Allison, P. D. (2005) Fixed Effect Regression Methods for Longitudinal Data
Using SAS, NC: SAS Institute Inc.
Ball, R. and Brown, P. (1968) ‘An Empirical Evaluation of Accounting
Income Numbers’, Journal of Accounting Research, 6,159-78.
Ball, R. and Watts, R. (1972) ‘Some Time Series Properties of Accounting
Income’, Journal of Finance, 27, 663-82.
Beaver, W., Clarke, R. & Wright W. (1979) ‘The Association between
Unsystematic Security Returns and the Magnitude of the Earnings Forecast
Errors’, Journal of Accounting Research, 17, 2, 316-40.
74
Bernard, V. L. (1995) ‘The Feltham-Ohlson Framework: Implication for
Empiricists’, Contemporary Accounting Research, 11, 733-747.
Bernard, V. L. (1987) ‘Cross-Sectional Dependence and Problems in
Inference in Market-Based Accounting Research’, Journal of Accounting#p#分頁(yè)標(biāo)題#e#
Research, 25, 1, 1-48.
Chen, C. H. (2002) ‘The Value Relevance of Patents’, Unpublished master
dissertation, Chung Yuan Christian University, Chung Li, Taiwan.
Dechow, P. M., Hutton, A., & Sloan, R. (1999) ‘An Empirical Assessment of
the Residual Income Valuation Model’, Journal of Accounting and
Economics, 26, 1-34.
Easton, P. D. (1985) ‘Accounting Earnings and Security Valuation: Empirical
Evidence of the Fundamental Links’, Journal of Accounting Research, 23,
54-77.
Easton, P. D. and Harris, T. S. (1991) ‘Earnings as an Explanatory Variable
for Returns’, Journal of Accounting Research, 29, 1, 19-36.
Eckberg, D. L. and Lester Hill, Jr. (1979) ‘The Paradigm Concept and
Sociology: A Critical Review’, American Sociological Review, 44, 6,
925-937.
75
Elliott, B and Elliott, J (2008) Financial Accounting and Reporting, England:
Pearson Education Limited.
Fama, E. F. and French, K. R. (1992) ‘The Cross-Section of Expected
Stock Returns’, Journal of Finance, 47, 2, 427-65.
Feltham, G. A. and Ohlson, J. A. (1995) ‘Valuation and Clean Surplus
Accounting for Operating and Financial Activities’, Contemporary
Accounting Research, 11, 2, 689-731.
Frankel, R. and Lee, C. M. (1998) ‘Accounting Valuation Market Expectation
and Cross-Sectional Stock Returns’, Journal of Accounting and Economics,
25, 283-319.
Garman, M. B. and Ohlson, J. A. (1980) ‘Information and the Sequential
Valuation of Assets in Arbitrage-Free Economies’, Journal of Accounting
Research, 18, 2, 420-440.
Hsieh, P. F. (2004) ‘The Effect of Financial Indices and Other Information on
Stock Price-An Extensive Application of Ohlson Model’, Unpublished
master dissertation, Chung Yuan Christian University, Chung Li, Taiwan.
Hussey, J. and Hussey, R. (1997) Business Research-A Practical Guide for
76
Undergraduate and Post graduate Students, UK Hampshire Basingstoke:
Macmillan Press LTD.
Jensen, M. C. and Melking, M. (1976) ‘Theory of the Firm: Managerial
Behaviour, Agency Costs and Ownership Structure’, Journal of Financial
Economics, 3, 306-360.
Kao, L. F. (2002) ‘The effects of shares collateralized by board of directors
on accounting earnings and firm performance’, Unpublished doctoral
dissertation, National Cheng Kung University, Tainan, Taiwan.
Kormendi, R. and Lipe, R. (1984) ‘Earnings Innovations, Earnings
Persistence and Stock Returns’, Journal of Business, 60, 3, 323-345.
Kuo, S. L. (2002) ‘The Relevance between Institutional Investors and
Financial Statements’, Unpublished doctoral dissertation, National Cheng
Kung University, Tainan, Taiwan.
Landsman, W. (1986) ‘An Empirical Investigation of Pension Fund Property#p#分頁(yè)標(biāo)題#e#
Rights’, Accounting Review, 61, 4, 662-691.
Li, S. C. (2002) ‘The Analysis of Exchange Rate Exposure--the Empirical
Studies of Panel TAR’, Unpublished master dissertation, Fu Jen Catholic
University, Taipei, Taiwan.
Lev, B. (1989) ‘On the Usefulness of Earnings and Earnings Research:
77
Lessons and Directions from Two Decades of Empirical Research’, Journal
of Accounting Research, 27, 153-202.
Liu, J. and Ohlson, J. (1999) ‘The Feltham-Ohlson (1995) Model: Empirical
Implications’, Working Paper, University of California and New York
University.
Lo, k. and Lys, T. (1999) ‘The Ohlson Model: Contribution to Valuation
Theory, Limitations, and Empirical Applications’, Conference of the Journal
of Accounting, Auditing, and Finance
Modigliani, F. and Miller, M. (1958) ‘The Cost of Capital, Corporation
Finance, and the Theory of Investment’, American Economic Review, 48,
261-297.
Morgan, G. (1980) ‘Paradigms, Metaphors, and Puzzle Solving in
Organization Theory’, Administrative Science Quarterly, 25, 4, 605-622.
Myers, J. N. (1999) ‘Conservative Accounting and Finite Firm Life: Why
Residual Income Valuation Estimates Understate Stock Price’, Working
paper, University of Washington, Business School
Myers, J. N. (1999) ‘Implementing Residual Income Valuation with Linear
Information Dynamics’, Accounting Review, 74, 1, 1-28.
78
Ohlson, J. A. (1979) ‘Risk, Return, Security-Valuation, and the Stochastic
Behaviour of Accounting Numbers’, Journal of Financial and Quantitative
Analysis, 317-316.
Ohlson, J. A. (1995) ‘Earnings, Book values, and Dividends in Equity
Valuation’, Contemporary Accounting Research, 11, 2, 661-687.
Patton, A. and Baker, J.C. (1987) ‘Why Won't Directors Rock the Boat’,
Harvard Business Review, 65, 10–18.
Peasnell, K.V. (1981) ‘On Capital Budgeting and Income Measurement’,
Abacus, 27, 52-67.
Pound, J. (1988) ‘Proxy Contests and the Efficiency of Shareholder
Oversight’, Journal of Financial Economics, 20, 237-265.
Singer, J. D. and Willett, J. B. (2009) Applied Longitudinal Data Analysis:
Modelling Change and Event Occurrence, New York: Oxford University
Press.
Soffer, L and Soffer, R (2003) Financial Statement Analysis – A Valuation
Approach, U.S.A: Pearson Education Limited.
79
Su, M. S. (2002) ‘The Corporate Value Relevance of R&D Expense in
High-tech. Industry’, Unpublished master dissertation, National Sun
Yat-sen University, Kaohsiung, Taiwan.
Subramanyam, K. R. and Wild, J. J. (1996) ‘Going-concern Status, Earnings
Persistence, and Informativeness of Earnings’, Contemporary Accounting#p#分頁(yè)標(biāo)題#e#
Research, 13, 1, 251-273.
Teng, C. C. (2002) ‘The estimation of trademarks value and its test on value
relevance’, Unpublished master dissertation, Chung Yuan ChristianUniversity, Chung Li, Taiwan.
Ting, L. P. (2003) ‘An Empirical Examination of the Relevance betweenCorporate Governance, Accounting Information, and Equity Market Value’,
Unpublished master dissertation, National Taiwan University, Taipei,Taiwan.
英國(guó)南安普頓大學(xué)留學(xué)生財(cái)務(wù)會(huì)計(jì)碩士dissertation由英國(guó)dissertation網(wǎng)提供,本文通過(guò)在臺(tái)灣證券交易奧爾森模型的實(shí)證分析,即對(duì)奧爾森估值模型對(duì)賬面價(jià)值與不正常的收入及其他信息的統(tǒng)計(jì),探索固定效應(yīng)回歸模型。Watts, R. L. and Leftwich, R. L. (1977) ‘The Time Series of Annual
Accounting Earnings’, Journal of Accounting Research, 15, 2, 253-271.Williams, J. B. (1938) The Theory of Investment Value, Cambridge: Mass.
Wu, P. C., Lin, J. C., & Cheng, C. H. (2004) ‘Corporate Governance
Mechanism, Systematic Risks, and Stock Valuation’, The 5th Annual
Conference of Taiwan’s Economic Empirics, Feng Chia University.
相關(guān)文章
UKthesis provides an online writing service for all types of academic writing. Check out some of them and don't hesitate to place your order.