IA“ Li‘u In..,
(‘3? ACCO‘JNHNG FOR LGNG-T mm mum.--
1!“ COMMON STOCK

Dissertation for the Degree 0! PM).

M‘CHEGAN STAYE HNNERSWY

CLAUD D‘N AWE DOWELL

1974

 

«J I .
LIBRARY :
Prithiga‘ Shite g
J”. Univcuity 0}

 

1' Of, 'Fxn 9"K‘VY . 1'3"... ’3;

This is to certify that the

thesis entitled

AN EMPIRICAL EVALUATION OF THE COST AND EQUITY METHODS
OF ACCOUNTING FOR LONG—TERM INVESTMENTS
IN COMMDN STOCK

presented by

Claud Dwayne Dowell

has been accepted towards fulfillment
of the requirements for

Ph.D. degree inBusiness Administration—
Accounting

  

Z7

Major professor

   

Date 4'37'7‘7‘

0-7 639

 

 

 

 

 

.cl ‘lnr:\

 

 

 

} a
' ‘ Vfl *tfilf v t - . ' , 1‘. 2 , _ ., a
.. .

tuna-rm. ,- C

     
  
  
   
  
 

'1” 637-3": . . Q"
‘0:- on m4 ,- ‘ ‘ -. 7 " -— ‘ ‘vn unmcn
.
‘ Catlin-r ctr", In“ . .
C . ’nl‘31- flip it ‘ - ‘ 4; ' .n r r; . “C +3

t

01'! ccrzv-‘p‘. t' ' ‘ ~ir':

:gmifl "YUM-‘- ”m 13- . ' .s-J. _. -‘ (um ' ..e
w thee-r, tilt"?! .- .l : -v'..'....:; 5 .t " v renni‘uer‘s:

‘.”.flmM’i'-u '.’.:"‘u;.t,"j i” .. a ,A: an: r... exact” 1'5"

H 'm ocwwric artery.

u have: (F1 mr' .1 .} ans-h “dept that F1 uses

‘zwlg w" till! par; <. 1 W? U .rwanrxn animate

» ~ 3 (’35!) n! We net sorrows as they, tho

. -. for 5’1 u; m using a; put a! mutant

7 C :- W1: comings museum (on) an
_ "3‘ 'v‘

 

ABSTRACT
AN EMPIRICAL EVALUATION OF THE COST AND EQUITY METHODS
OF ACCOUNTING FOR LONG—TERM INVESTMENTS
IN COMMON STOCK
By

Claud Dwayne Dowell

   
  
  
  
  
   
   
   
  
 
 
 
 
 
  
  
  

The purpose of this research effort is to provide empirical evi-
dence of whether cost method generated data appears to have more, less,
or the same usefulness potential as equity method generated data for
investors in corporations with material unconsolidated common stock
investments. The cost and equity methods are alternative methods of
accounting for an investor corporation's long—term investments in common
stock of other corporations.

The first step in attempting to accomplish the purpose is the
development of a conceptual framework in which to interpret observed
investor reactions to cost and equity data. The second step is applying
the conceptual framework to actually observed reactions.

The conceptual framework is based on the semi-strong form of the
efficient market theory (EMT) and the following six ideal conditions:

(1) Two alternative accounting methods (A1 and A2) exist for
measuring the same economic events.

(2) Two identical firms (F1 and F2) exist, except that F1 uses
A1 and F2 uses A2.

(3) At the beginning of time period 1 (01), investors estimate
the certainty equivalents (GEES) of future net earnings as they, the
investors, define net earnings for F1 and F2 using as part of the input

to the prediction process the quarterly earnings announcement (QEA) data

m_' ”fin—w -.._ __
-r

fl};,~'

7III_'____—___I__________—___'"""""""""'1PI'!

Claud Dwayne Dowell

  
   
 
  
  
   
   
  
  
  
  
  
  
  
  
  
  
  
  
  
  
  

released at the end of time period 0 (Q0). Inputs to the prediction
process at the beginning of Q1 other than those attributable to A1 and
A2 are identical for both firms.

(A) No events occur during Q1 that stimulate investors to alter
the predictions made at the beginning of Q1 except the QEAS released by
F1 and F2 at the end of Q1.

(5) Investors are rational and have no capital constraints.

(6) The GEES made at the beginning of Q1 for F1 were perfect, but
the corresponding GEES for F2 were imperfect, although investors were
unaware of this at the time of estimation.

Now obviously A1 provides data with greater usefulness potential
for investors than A2. Furthermore, since the GEE of Q1 earnings for F1
was perfect, there is no stimulus at the end of Q1 for investors to
revise GEES of net earnings for subsequent F1 quarters. If the GEES of
net earnings for subsequent F1 quarters are not revised then the instan-
taneous, unbiased price adjustment predicted by the EMT for the F1 QEA
released at the end of Q1 is zero. In contrast, since the GEE of Q1
earnings for F2 was imperfect, there is a stimulus at the end of Q1‘

for investors to revise GEES of net earnings for subsequent F2 quarters.
Furthermore, if all GEES of net earnings for subsequent F2 quarters are
revised in the same direction, as argued in the thesis, then the instane
taneous, unbiased price adjustment predicted by the EMT for the F2 QEA
released at the end of Q1 is nonzero. Accordingly, one of the two
conclusions of the conceptual framework is that under the assumed
conditions, the smaller the price adjustment immediately following the
Q1 QEA, the greater the usefulness potential of the Q0 output of an .

accounting method to investors on average. I

Glaud Dwayne Dowell

A second conclusion of the conceptual framework is concerned with

  
  
  
  
  
  
  
  
  
  
  
  
  
  
  
  
  
  
  
  
  
  
   
 
  

the interpretation of investors' responses to accounting data when those
responses are measured in terms of share trading volume. Within the
same time frame as described above share trading volume is a reflection
of individual investors' earnings expectational changes (EECS) whereas
price change is a reflection of the market's average EEC. But many
different distributions of individual investors' EECS may result in the
same average (or market) EEC. For example the investors' EEC distri—
bution associated with A1 may be normal with small dispersion and the
investors' EEC distribution associated with A2 may be normal with large
dispersion but both distributions may have the same mean. Now assume
that all investors who experience EECS different from the mean EEC
adjust their portfolio position by buying or selling the security and
all investors who experience EECS equal to the mean hold their portfolio
position. If the common stock market is a market of many small inves-
tors each wishing to hold the same number of shares of the security and
the EEC distributions associated with A1 and A2 are as described above,
then, the volume of F1 Shares traded would be smaller than the volume of
F2 shares traded. Furthermore, it should be obvious that the degree of
consensus among investors about F1 EECS is greater than about F2 EECS.
Accordingly, the second conclusion of the conceptual framework is that
under the assumed conditions, the smaller the volume of shares traded
immediately following the Q1 QEA the greater the degree of consensus
among investors regarding the usefulness potential of the Q0 output of
an accounting method.

The conceptual framework is applied, i.e. a test application of the

conceptual framework is made, by comparing, via a Mananhitney U test,

-'!-F‘T

 

 

Glaud Dwayne Dowell

investors' responses to QEAS of a sample group of cost method associated
firms and a sample group of equity method associated firms, selected on
the basis of ten criteria designed to control and/or compensate for
differences between real world and ideal conditions. The criteria
include:

(1) A control for competing information;

(2) A control for profitability;

(3) A control for liquidity;

(A) A control for accounting policy characteristics; and

(5) A compensation for industry differences.

_The response measurements are made in percentage terms over the five
day trading week immediately following a QEA and are adjusted for
uncontrollable and/or uncompensatable factors such as the general
market effect, capitalization changes, and the effect of nonpublic
information.

Unfortunately, the use of the selection criteria both severely
limited the sample sizes (eight firm—quarters in each group) and made
the samples nonrandom. Accordingly, the statistical results are incon—
CIusive. Intuitively, however, the empirical findings appear to
indicate that on average equity method data is more useful than cost
method data and that the degree of consensus among investors regarding
'the usefulness of cost data is greater than that of equity data.

Finally, the implications and limitations of the research effort

are noted and several suggestions for further research are made.

 

   

AN MIRICAL EVALUATION OF THE COST AND NET! METl-DDS
OF ACCOUNTING FOR LONG-TERM INVE'IMENTS
IN COMMON STOCK

By

Glaud Dwayne Dowell

A DISSERTATION

Submitted _to
Michigan State University
in partial fulfillment of the requirements
for the degree of

DOOM OF PHILOSOPHY

1971»

 

di‘:" '7 . 2,

    
   
  
    
   

.L uh’ 1151"" , ‘ r

‘1) TCT_.~. ‘ . ..‘

<
a

‘ ‘3 ‘2) To my w: -, _ .‘ '~ . -‘ ;.- ." .‘ . 1;": _ ' s « 1-.
. .' ‘IM contains. . .
w‘~“.x:, ~

the res»_~».L ~.'t"‘..

."“;'
"13) 5‘21"! 1. ,. n ' , . - ~-

‘ tr) 3.1 T1125; cc twp. . -. i”. (‘4. 1 Ian) “wads
‘
. I“, .
:W runes: f : .,

a

r
.2". x, . . . r" a
. I . , .. ..' . .» _ a ,1. - N. n“. _-J .-’ ..,v_r
. " ufr'h .".‘, ~14 ..._ . . _ ,
J.” To y l.
.
i ‘f

1- guidance during we en's-1 verve-:1- - ‘,.",Y‘[.

C;

3'“ 1;;2 1‘: r}. "‘11:” inn. rer tax 3!; u:

we, any errors r in
V l

. , C,
ties .u'u". any wists.

 

v
«H.» - . .
‘ 11 ‘
‘ ‘ ~ 1‘ I
‘ |
.-- 'ta _
~ “‘ . $ _ ‘

ACKNOWIEDWTS

     
 
 
  
  
    
  
  

Not unlike other researchers, I am indebted to many individuals
.2 for their aid in the completion of this research. However, I am
_ I '. particularly indebted to the following individuals:
.5 (1) To my parents for their initial inspiration;
(2) To my wife, Linda, for her patience, cooperation, encourage-
‘ in; meat, and continuing inspiration as well as her typing services
. ‘ throughout the research effort;
(3) To aw daughter, Carin, who although yet unaware has been
compelled to sacrifice the compamr of her daddy on many occasions
.I 1flaring the research effort; and
_ (1+) To my connfittee members—Professor George Mead (Chairman),

ghofessor Gardner Jones, and Professor Phillip Carter—for their

j“ fiflbtaiculous guidance during the entire research effort.

g! v or course, any errors or inadequacies regarding the research are

‘-. ’ ins responsibilities and only mine.
' ‘ .‘i: ‘ . ,

 

 

 

TABLE OF CONTENTS

Chapter Page
I O INTRODUC 'I'ION O I O C O 0 O O O O O I O I O 0 0 O O O O O O 1
Purpose and Approach of This Research Effort . . . . 1

Background of the Problem Under Consideration
in This ResearCh Effort 0 o o c o o o o o o c I o o
The Organization Of the TheSiS o o o o c o c o c c a

The Controversy Reviewed . . . . . . . . . . . . . .

IntrOductionococo-0.00.0000...

3
7
II. LITERATURE REVIEW AND RESEARCH JUSTIFICATION c o o o o o o 8
8
8

TheArgumentScoo-00.00.00.000. 9

Cost Versus Equity Empirical Research . . . . . . . . 12

IntrOdUCtion o c c o o o c o o c a o c o o o o o 12
Barrett's ResearCh o o o o o o .o o o o c o o o o 12
COHCIUSion c o o c o o o o o o c c o o o o c o o 17

other Related ResearCh o o o o o o o c o o o c o c o 17

IntrOduction o o o c o o I o c o o o o o o o o o 17
The Review . . . . . . . . . . . . . . . . . . . 18
ConCIUSionS o c o o o o o a c o o o o a o o o o o o c 30

III. METHODOLOGY-—CONCEPTUAL CONSIDERATIONS . . . . . . . . . . 32

TheEffiCient Market meow-00.0.0000... 32
Quarterly Earnings Announcements as New Public

Irlformationooooooococoa-coco... 33
The Content of Quarterly Earnings Announcements

and the Primary Criterion for Evaluating the

Alternative Methods of Generating the Content

of Quarterly Earnings Announcements . . . . . . . . 33
The Locus and Definition of Usefulness Potential . . 3h
Measuring and Comparing the Usefulness Potential

of Alternative Accounting Outputs in an

EfficientMarket.o............... 35

 

iii

 

 

 

 

Extending the Conceptual Framework to
VOlumeMeasureSo-ooooooo.0000.

Interpreting Price Change Measurements and
Volume of Shares Traded Measurements
Together 0 I O O O I O O O O O I O O I O O 0

Combination 1
Combination 2
Combination 3
Combination A
Combination 5
Combination 6
Combination 7
Combination 8
Combination 9

Sumam O O O O O C O O O I O I O O C O O O O C
IV. METHODOLOGY-—APPLICAIION CONSIDERATIONS . . . . . .

IntroduCtj-on O I O O I O I O O O O I O O O O 0
Ideal Versus Real World Conditions . . . . .
Selection Criteria Implied by the Ideal Condi-
tions and Procedures for Determining Compli—
ance With These criteria 0 n o o o o o u o 0
TWO Additional SeleCtion criteria 0 o o o o o 0
Time Period Covered by the Research . . . . . .
Measurement Interval and Measurement Forms for
Investors' Reactions to Quarterly Earnings
Announcements . . .

Data Adjustments o o o o o o o o n o o o c o o
The Comparison . . . . . . . . . . . . . . . .
smar'y o o o o o o o a o a o o o o o o o o o I

V. EMPIRICAL FINDINGS, IMPLICATIONS, LIMITATIONS
AND SUGGESTIONS FOR FURTHER RESEARCH . . . . . . . o

IntrOduction o o o I o o 0

Empirical Findings 0 o o o o o o c o o o o o 0
Implications o o o o o o o o o o o o o o o o 0
Limitations . . . . . . . . . . . . .
Suggestions for Further Research . . . . . . .
COHClUSion o o o o o o o o o o o o o o o o I o

BIBIIIOMI'IY I O C O O C C I O O C I C l I O 0 O I O C C 0

iv

0

Page

90
90

101
104

109
110

LISTOF TABLES

  
   
   
   
   
 
  

Tabular Arrangement of Barrett's Experimental Design

Tabular Summary of Barrett's Statistical Tests-—
ContrastsandResults......o.......o

Possible Combinations of Results of Price Change Com-
parisons and Volume of Shares Traded Comparisons .

Composition of Cost and Equity Firm Quarter Groups .

Selection Criteria Severity in Terms of Sequential
Sample Size Reduction Rates 0 e o o o o o e o e e 0

mice Change comparison 0 o o o o o e e e e o o e o e
VOlume comparison 0 o e_ o e o o o o o e e I o e o e 0
Effects of Price Data Adjustments . . . . . . . . . .

Effectsof Volume Data Adjustments . . . . . . . . .

Page

15

93

9h
96

98

 
       
     
  

.7—

Own,»

1‘ ) J1
_,5
'.'

 

(A

z: ‘79-
'_~ ‘I

a
A.

9-5:
‘-:,..
,_,‘

..
4“-

:0

 

LIST OF FIGURES

Figure Page

1. Schematic of Information Set Evaluation From Two
Different Levels of Observation——Investors' Level
of Observation and an Independent Agent's Level
Of Observation I I I I I I I I I I I I I I I I I I I I I 38

2. Illustrative Earnings Expectational Change Distri—
butions I I I I I I I I I I I I I I I I I I I I I I I I I #5

3. Possible Results of Comparisons—~Combination 1
SChematic I I I I I I I I I I I I I I I I I I I I I I I I “9

A. Possible Results of Comparisons-—Combination 2
SChematiC I I I I I I I I I I I I I I I I I I I I I I I I 50

5. Possible Results of Comparisons—-Combination 3
SChematj-C I O O I O C O O I O O O O O O O I O O O O O O I 51

6. Possible Results of Comparisons——Combination 4
SC hematic O O O O O I O I O O O I I O O O I O C 0 C O O I 5 2

7. Possible Results of Comparisons-—Combination 5
Schematic I I I I I I I I I I I I I I I I I I I I I I I I 53 ‘

8. Possible Results of Comparisons-Combination 6
SChematic I I I I I I I I I I I I I I I I I I I I I I I I 5A

9. Possible Results of Comparisons—~Combination 7
Schematic I I I I I I I I I I I I I I I I I I I I I I I I 55

10. Possible Results of Comparisons—-Combination 8
SChematiC I I I I I I I I I I I I I I I I I I I I I I I I 56

11. Possible Results of Comparisons-—Combination 9
SChematiC I I I I I I I I I I I I I I I I I I I I I I I I 57

 

 

 

 

CHAPTER I

INTRODUCTION

In this chapter is a brief statement of the purpose and approach
of this research effort, the background of the problem under considera-

tion, and the organization of the thesis.

Eggpgse and Approach of This Research Effort

The purpose of this research effort is to provide empirical evi-
dence of whether there is a difference in the usefulness of alternative
accounting methods' outputs from the viewpoint of investors in publicly
owned corporations (one class of accounting data users); specifically
Whether cost method generated data appears to have more, less, or the
same usefulness potential as equity method generated data for investors
in corporations with material unconsolidated common stock investments.
The cost and equity methods are feasible alternative methods of
accounting for an investor corporation's long-term investments in
common stock of other corporations.1 The relative usefulness potential
of data generated by alternative accounting methods is, according to
the American Accounting Association Committee to Prepare A Statement

of Basic Accounti Theo , the most important criterion against

1Harry Simone and Wilbert E. Karrenbrock, Advanced Accounting

 

...

I_..

 

,...

'4‘.
v

"V

,.I

 

 

 

 

2

which to evaluate feasible alternative accounting methods.2 The more
potentially useful the data generated by an accounting method the
better the accounting method.3

The general framework of this research effort, based on the semi-
strong form of the efficient market theory, involves comparing inves—
tors' market based reactions (i.e. price change and share trading
volumeh) to cost method generated data and equity method generated data
and drawing inferences from the comparative reactions. According to
this framework, developed in Chapter III, the alternative accounting
nwthod associated with the smaller price change immediately following
an earnings announcement is accorded the distinction of generating data
With the greater usefulness potential for investors on average. The
alternative accounting method associated with the least share trading
volume immediately following an earnings announcement is accorded the
distinction of generating data for which there is greater consensus
among investors regarding the usefulness potential of the method's out-
puts.—

The particular empirical approach of this research effort, devel—
oped principally in Chapter IV, involves comparing (after abstracting
EXOgenous influences) both the total price change and the total share

trading volume in the earnings announcement week (the first five trading

*—

2American Accounting Association, A Statement of Basic Accounting
Theogy (Evanston, 1966), p. 3.

31bid.

“William H. Beaver, "The Information Content of Annual Earnings
Announcements," 'rical Research in Accounti : Selected Studies,
1268é9a Supplement to Velume 5 of the Journal of Accounti Research,

-70.

 

 

 

 

E

 

 

3

days following an earnings announcement) associated with two sample
groups of investor corporations, a group using the cost method and a
group using the equity method. The two groups are similar except for
the accounting alternatives on which the dichotomous classification is
based.

It should be noted at the outset, however, that application of a
set of selection criteria considered important in obtaining similarity
between the sample cost and equity groups could, and did, result in
samples of severely restricted size and questionable randomness.
Consequently, the use of statistical methods of comparing the groups
was inappropriate. However, to complete the steps in the research
design, a statistical method of comparison is illustrated, i.e.
applied as if the groups had been selected randomly. The resulting
conclusions though are stated simply as intuitive conclusions and not
as statistically valid ones.

Backgnound of the Problem Under Consideration in
This Research Effort

 

Although thirty—two feasible alternative methods of accounting
for an investor corporation's long-term investments in common stock
of other corporations exist, by far the two most widely used, during the
Period covered by this research, were the cost method and the equity

method.5 The two methods will generally produce very different results,

 

5Melvin C. O‘Connor and James C. Hamre, "Alternative Methods of
Accounting for Long-Term Nonsubsidiary Intercorporate Investments in
Common Stock," The Accounti Review, XLVII (April, 1972), pp. 308-319;
and American Institute of Certified Public Accountants, Accounting
Trends and Technigues (New York, 1950-1970).

 

 

 

 

A

in terms of their effect on the investor corporation's reported earnings
and financial position.6

Using the cost method, the investor corporation initially records
the investment at its acquisition cost. The investor corporation's
investment account is then adjusted only for permanent decreases in the
value of the investment or for dividends received in excess of the
investee corporation's earnings subsequent to the date of investment.
Income from the investment is recognized by the investor corporation
only to the extent that dividends from investee corporation earnings
subsequent to the date of investment are received by the investor
corporation.7

Using the equity method, the investor corporation also initially
records the investment at its acquisition cost. However, the investor
corporation's investment account is adjusted periodically for its
Share of earnings or losses of the investee corporation (subsequent to
the date of investment and after elimination of all intercompany trans—
actions), for all dividends received from the investee corporation, and
for permanent decreases in the value of the investment. Income from
the investment is determined by the investor corporation's share of
earnings or losses of the investee corporation (qualified as above) and
by amortization of any difference between the cost of the investment
and the equity in the underlying net assets of the investee corporation

at the date of investment.8

 

6Simons and Karrenbrock, pp. 3A7-392.
71bid-. pp. 377-392.
81mm, pp. 347—365.

 

 

 

5

Advocates of the cost method criticize the equity method severely
for its departure from generally accepted principles of revenue recog-
nition, its mixture of different elements in the investment account,
and its special analyses required to determine retained earnings legally
available for dividends and income subject to taxation. They support
the cost method as one consistent with generally accepted accounting
principles applied to other assets.9

Supporters of the equity method emphasize the substance of the
investor corporation—investee corporation relationship. They criticize
the cost method for preventing the recognition of all earnings which
are related to the substantive entity and for the possibility of invest—
ment income manipulation when the cost method is used. They hail the
equity method because it produces results similar to that of consoli-
dation.10 Consolidated statements are assumed to be more meaningful

than separate statements.11

 

91bid., p. 390; and Walter B. Meigs, Charles E. Johnson, and '
Thomas K. Keller, Advanced Accounting (New York, 1966), pp. 297-298.

10Meigs, Johnson, and Keller, p. 298; and American Institute of
Certified Public Accountants, Accounting Principles Board, "APB Opinion
NO. 18: The Equity Method of Accounting for Investments in Common
Stock," The Journal of Accountancy (June, 1971), pp. 63-68.

11American Institute of Certified Public Accountants, “Accounting

Research Bulletin No. 51-—Consolidated Financial Statements," nag

Accounti Princi les: Ori 'nal Pronouncements as of Febru 1 l 1
Chicago, 1971 , II, 091.

4..

 

 

 

6

This is the controversy which is the impetus for this research
effort.12 Advocates of both methods have argued at length for their
position.13 Others have argued that either method is equally justi-
fiable.”+ However, these arguments have been in the majority a priori
in nature. Indeed, neither the cost advocates nor those contending
that either method is equally justifiable have presented any evidence
supporting their positions, particularly in terms of the usefulness
potential criterion. Equity advocates have done little better.15 As
noted previously, this research effort is designed specifically to

provide the heretofore missing evidence bearing on this question.

 

12Although a controversy still exists over which of the accounting

alternatives, cost or equity, should be used (See for example O'Connor
and Hamre, pp. 308-319), it should be emphasized that the Accounting
Principles Board of the American Institute of Certified Public Account-
ants has settled the issue of which accounting alternative must be used
in its "Opinion No. 18: The Equity Method of Accounting for Investments
in Common Stock." This research effort, however, is designed to provide
SVidence bearing on the question of which accounting alternative should
be used, and will therefore provide evidence either in support of or
against the Accounting Principles Board's opinion.

 

13See for example, Simone and Karrenbrock, p. 390; American
Institute of Certified Public Accountants, Accounting Principles Board,
Pp. 64~66; P. T. Driscoll, "Consolidated Financial Statements and the
EQuity Method of Accounting,” The New York Certified Public Accountant
(August, 1971), pp. 567—571; M—aurice M"o'o"ni't'z',"'Th”-e E_ntity Th' e'ogz —of
Consolidated Statements (Madison, 19th), p. 59.

1l"'0'Connor and Hamre, pp. 308-319; and Meigs, Johnson, and Keller,
PP- 297-298.

15A review of the accounting literature published since January,
1952, revealed only one research effort attempting to evaluate the cost
and equity accounting methods, and even it is subject to criticism.
See the section entitled "Cost Versus Equity Empirical Research" in
Chapter II I

 

I11: E

nu.-.

! o
I -

 

um sation of the Thesis

  
 
  
  
  
  
  
  
  
 
  

,3, '7". Chapter I is devoted to a statement of the purpose and approach
.' "ofslthie research effort, a discussion of the accounting methods under
“amideration, a synopsis of the arguments regarding the merits and

4" dinerits of each method, and a description of the organization of the
Tithsis. Chapter II contains an in—depth specification of the arguments
“of each party to the controversy, a review and critique of empirical
Work which has been done on the controversy, and a review of other
-emp:l.rical work bearing on the market based research design for this

‘ -._effort presented in Chapters III and IV. The research findings and

A their implications are presented in Chapter V along with a discussion

9‘ 'M the limitations of the research and suggestions for further research.
, J... w

"(W :‘.(-'. . ., ‘ ,‘. V’ ‘

 

 

 

CHAPTER II

LITERATURE REVIEW AND RESEARCH JUSTIFICATION

The Controversy Reviewed

Introduction. Thirty—two feasible alternative methods of account-
ing for a corporation's long—term investments in common stock of other
corporations exist, and even a cursory review of the accounting litera-
ture reveals that a controversy exists over which of these thirty-two
alternatives is best.1 It is beyond the scope of this paper to review
the arguments pro and con related to all thirty-two alternatives.
However, a significant portion of the controversy centers around the
question of whether the cost method is better than the equity method.2
(The cost method and the equity method were the two most widely used of
the thirty-two feasible alternatives during the period covered by this

research.)3 This is the question under consideration here. The

 

1Melvin C. O'Connor and James C. Hamre, "Alternative Methods of
Accounting for Long—Term Nonsubsidiary Intercorporate Investments in
Common Stock," The Accountin Review, XLVII (April, 1972), pp. 308-319.

2Ibid.; American Institute of Certified Public Accountants,
Accounting Principles Board, "APB Opinion No. 18: The Equity Method of
Accounting for Investments in Common Stock," The Journal of Accountanc
(June, 1971), pp. 63-68; M. Edgar Barrett, "APB Opinion Number 18:
A Move Toward Preferences of Users," Financial Analysts Journal (July-
AuguSt) 1972), Pp. 47-540

3O'Connor and Hamre, pp. 308—319; Barrett, p. A7; American
Institute of Certified Public Accountants, Accounting Trends and
Technigues (New York, 19 50-1970) .

8

 

 

 

 

 

9

arguments of both the cost method advocates and the equity method
A

advocates are summarized below.
The Argnnents. Cost advocates argue that

'. . . the equity method involves basic departures from
generally accepted practice in the recognition of revenue.
Earnings of the subsidiary are viewed as earnings of

the parent, an economic View which, it can be maintained,
finds proper expression on the consolidated statement

and not on the statement representing the separate legal
entity.”5

”Income of a subsidiary is not realized by the parent
until there has been a transfer of liquid assets (that
is, dividends declared or paid)."

Equity supporters reply that

“An exchange of liquid assets is not the only acceptable
evidence of realization. The change in the net assets of
a subsidiary as a result of Operations is based on
evidence equally as good as that used to establish the
realized growth of net assets which produces the income of
the parent. Since the parent has a controlling interest
in the subsidiary, there is no question of ultimate control
over the increase in net assets; therefore, recognition

on the parent's books of its share of any undistributed
income of the subsidiary is warranted."

 

hThe arguments presented in this section are, almost without
exception, stated in the context of one corporation having an absolute
Controlling interest (greater than fifty percent ownership) in
another corporation. However, it would appear that since the Accounting
Principles Board of the American Institute of Certified Public Account—
ants has structured its opinion on cost versus equity accounting around
the concept of effective control (presumed to exist, in absence of
eVidence to the contrary, with only a twenty percent interest, and in
some cases with less than a twenty percent interest) that these argu-
ments would apply in that context (effective controls as well. (See
American Institute of Certified Public Accountants, Accounting Prin-
ciples Board, pp. 63-68.)

5Hilbert E. Karrenbrock and Harry Simons, Advanced Accounting
(Cincinnati, 1961), p. 385.

6Walter B. Meigs, Charles E. Johnson and Thomas F. Keller,
Advanced Accounting (New York, 1966), p. 297.

7Ibid.

 

 

 

10

"[Furthermore,] . . . in statements prepared from the legal
entity viewpoint, the issue of proper asset valuation is
still present. The original cost of an investment in an
unconsolidated subsidiary becomes an outdated historical
figure as time passes and the subsidiary grows through the
retention of earnings. Since the accountant has objective
evidence (in the subsidiaries' financial statements) of the
amount of this growth, he should use this information to
update the valuation of the investment."8

The second objection that cost advocates have to the equity method

is as follows:

tion

”The valuation of an investment in a subsidiary when it is
accounted for on an equity basis is a hodgepodge. . . ."

". . . Investment balances fail to offer either costs or
current values of subsidiary holdings, since they are
composed of a mixture of different elements-costs (or
investment values) as of stock acquisition dates, modified
by adjustments based upon the changes in capital reported
on the books of the subsidiaries following acquisitions."lo

“Any difference between cost and the book value of the
subsidiary's net assets at date of acquisition remains, and
the equity basis valuation of cost plus or minus changes
in the underlying book value of subsidiary net assets
defies interpretation."11

The equity advocates answer this second objection to their posi—
as follows:

"The price paid by the parent for its investment is more
likely to reflect current value at that time than the

book value of the parent's equity in the subsidiary. Thus
the equity method begins with a valid basis of valuation
and this valuation is then adjusted for changes which can
be objectively measured on the basis of the same accounting
concepts as used by the parent."12

 

 

81bidI’ ppI 297-2980

91pm. , p. 298.
10Karrenbrock and Simons, p. 385.

11Meigs, Johnson, and Keller, p. 298.
12mm.

 

 

 

11‘

The third objection that cost advocates lodge against the equity

method is:

“. . . the equity method calls for analysis of the accounts

and special adjustments in arriving at the amount of

retained earnings legally available for dividends and the

amount of income that is subject to income taxes."

The response from the equity supporters to this argument could be
that the analyses of accounts and special adjustments to arrive at
retained earnings legally available for dividends are little, if any,
more difficult than cost behavior analyses and, furthermore, that the
worth of a better valuation of assets on the statement of financial
position far outweighs the relatively insignificant difficulty asso-
ciated with the analyses and adjustments. Special analyses with respect
to income taxes must be prepared anyway due to the complexities of the
income tax laws.

Finally, equity advocates indict the cost method on the grounds
that when it is used the possibility of dividend, and consequently
income from investment, manipulation exists.

Cost supporters can only affirm this indictment stated in terms
of a possibility. However, they can contend, with evidence, that this

POSSibility has not historically been used to grossly distort the

reported income of the investor corporation.u+

‘—
13Karrenbrock and Simons, p. 385.

1[‘Nicholas Dopuch and David F. Drake, "The Effect of Alternative
Accounting Rules for Nonsubsidiary Investments,“ Mirical Research in

A(Ecounti : Selected Studies 1966, a Supplement to Volume 1. of the
ECNEEEI 0% Iccountin. Research, pp. 192—227.

 

 

 

12

Cost Versus Eguity Empirical Research

Introduction. The arguments in the accounting literature reviewed

above are a priori in nature. It would appear from the review of these
arguments that resolution of the controversy via a priori argumentation
is improbable. In this vein, some accountants have suggested that
resolution of the controversy is, most likely, to be obtained via
empirical research addressing the controversy.15 However, it seems
that only one such attempt at empirical resolution has been made.
Indeed, an extensive systematic review of the accounting literature
covering the period of January, 1952, through December, 1972, revealed
onhy one empirical research effort addressing the question of whether
the cost method is better than the equity method on the basis of any
evaluation criterion. And even the one empirical research effort
unearthed is subject to criticism.

‘The single attempt at empirical resolution of the controversy
referred to above was conducted by M. Edgar Barrett of the Harvard
Business School and was reported in the July-August 1972 Financial
Agnlysts Journal under the title, "APB Opinion Number 18: A Move Toward
Preferences of Users."16 This research is summarized and criticized
below.

Barrett's Research. According to Barrett,

”The technique employed in the study consisted of asking
selected members of the Financial Analysts Federation to

‘—
15O'Connor and Hamre, p. 318.

16M. Edgar Barrett, ”APB Opinion Number 18: A Move Toward
Preferences of Users,” Financial Analysts Journal (July-August, 1972),

pp. [+6-53I

 

 

 

13

put a price on the common stock of companies offering stock
to the general public for the first time.”

Furthermore, according to Barrett,

"The general thrust of the inquiry was to determine whether
financial analysts would place a higher estimated value on
the common stock of a company simply as a function of (1)
the method employed in accounting for long term investments
in common stock or of (2) the amount of information included
in the notes to the published financial statements about

the effects of using the alternative method.”18

The methodology used in the study is described by Barrett as
follows:

“The methodology included creating prospectuses for two hypo-
thetical petroleum companies. These hypothetical companies
(Inglewood Oil Company and Miller 011 Company) represented a
crude oil producer with extensive Middle East holdings and
an integrated petroleum company with both domestic and
foreign facilities, respectively.

“Each financial analyst invited to participate in the study
received a prospectus for each of the two 011 companies.

“The 750 financial analysts asked to participate were
divided into three groups of 250 analysts. Members of the
first group (experimental class one) received prospectuses
which contained less total information about the inter~
corporate investments of the two hypothetical oil companies
than was contained in the prospectuses provided to members
of either of the other two groups. While members of the ‘
second and third groups (experimental classes) received
prospectuses with the same total amount of information about
the intercorporate investments of each oil company, the
information was provided in a different form to each of the
two classes.”

This experimental design was summarized by Barrett in the follow;

irug tabular arrangement.20

 

17Ibido, pI [bee
18Ibid.

191bid., pp. 48-h9.

zoIbidI, p. 19.

 

 

1h

TABLE 1

TABULAR ARRANGEMENT OF BARRETT'S EXPERIMENTAL DESIGN

 

 

Experi— Accounting .
mental Method Am°u§2 ggogizgigsure
Class Employed
1 Cost Quoted Market Value
Only
2 Cost Quoted Market Value,

Annual Earnings Under
the Equity Method, and
Cumulative Investment
Under the Equity Method

3 Equity Quoted Market Value,
Annual Earnings Under
the Cost Method, and the
Original Cost of the
Investment

 

The financial analysts participating in the study were asked to
respond to the following question:
”Given only the information provided, what dollar value

per share would you place on the common stock of these
two companies on their prospectus dates:

Inglewood Oil Company $. . . . . . per share
Miller Oil Company $. . . . . . per share"21 a
Then,

"Appropriate statistical tests were applied to the entire
number of responses received, as a group, for each of the
two oil companies."

The results of the statistical tests may be summarized in the

fOllowing tabular form.23

5
211bid., p. 50.
zzIbid.

23Ibid., p. 52.

 

 

l5

TABULAR SUMMARY OF BARRETT'S STATISTICAL
TESTS-—CONTRASTS AND RESULTS

 

 

Class , _ .
Contrast Inglewood 011 Company Results Miller 011 Company Results

 

1 vs 3 Reject HO that the class 1 Reject HO that the class 1
and class 3 responses are and class 3 responses are
statistically similar. statistically similar.

1 vs 2 Reject HO that the class 1 Reject Ho that the class 1
and class 2 responses are and class 2 responses are
statistically similar. statistically similar.

2 vs 3 Do not reject HO that the Do not reject Ho that the
class 2 and class 3 re— class 2 and class 3 re-
sponses are statistically sponses are statistically
similar. similar.

 

Barrett places the following interpretation on the above statis-
tical results:

”The fact that a statistically significant difference exists
between the responses to experimental classes one and three,
while none exists between the responses of classes two and
three, indicates that professional security analysts-in our
experimental setting—-prefer data prepared under the equity
method to data prepared under the cost method. This is
suggested, of course, by the fact that members of experimental
class two (which had enough information in the footnotes to
enable their recasting the financial data in terms of either
the cost or equity methods) submitted value estimates more
in line with those of class three than those of class one.'2h

Now to criticize, it can be contended that Barrett has overstated
his case. Even Barrett warns that

'. . . the difference between experimental class one and
the other two experimental classes was the total amount of
audited information about the intercorporate investments,
while the difference between the last two experimental 25
classes was in the method of presenting such information."
(Emphasis added)

 

 

2A1bido, p. 530
251bid., p. 1.9.

 

 

.r-

1»

 
 
 

 

16
Since the difference between experimental classes one and three was
the total amount of data provided and no difference existed between
experimental classes two and three with respect to the amount of data
provided, a more plausible interpretation of the statistical results
is simply that financial analysts respond differently to different
amounts of data and similarly to similar amounts of data, regardless
of the presentation form. In other words, it is not necessary to
assume (as Barrett has done in deducing his interpretation of the
statistical results) that the financial analysts in experimental class
two recast the cost method generated data provided them in terms of
equity method generated data and then responded to this recast data.
The assumption is, of course, a feasible one. However, it is also
feasible that members of both experimental class two and experimental
class three (1) recast the data provided them in terms of some common
intermediate form, i.e. a form other than that provided by either cost
or equity, or (2) used both the cost generated data and the equity
generated data in a joint sense. Either of these alternative behavior
Patterns, certeris paribus, would likely result in the same response
Pattern as that observed by Barrett. If Barrett wished to determine
whether financial analysts preferred equity method generated data to
coat method generated data, i.e. whether or not his assumed behavior
INIttern was accurate, given the design of his questionnaire-based field
enEjperiment, he should have included a fourth experimental class in which
the equity method was the accounting method employed and only the quoted
“Hirket value was disclosed in footnotes. Then, if a statistically
significant difference existed between responses in classes one and two,

One and three, and one and four, but no statistically significant

 

 

17
difference existed between responses in classes two and three and three
and four, the conclusion that financial analysts prefer equity method
generated data to cost method generated data would be more defendable.
Conclusion. Once again, it can be contended that Barrett has
overstated his case. And since no other empirical research addressing
the controversy was revealed by the literature review, it would appear
that if resolution is to be obtained empirically additional empirical

research is necessary.

Other Related Research

Introduction. If additional empirical research is necessary to

 

resolve the controversy, as argued above, a question concerning the
nature of the research immediately follows. After reviewing and
analyzing eleven frequently cited research efforts—~including two
controlled laboratory experiments, one combination questionnaire-
simulation experiment, and eight market oriented experiments-Hagerman,
Keller, and Petersen in a March, 1973, article published in The Journal
Of Accountancy concluded that market oriented studies are preferable to
other types of studies.26 Furthermore, these authors must not be alone
141 their opinion, for an increasing number of market oriented research
efforts are being conducted and published.27 In this context, then,
tell selected market oriented research efforts, the results of which

hears a direct bearing on the research design to be developed in the

 

n 26Robert L. Hagerman, Thomas F. Keller, and Russel J. Petersen,
Accounting Research and Accounting Principles," The Journal of
Accountanc! (March, 1973). pp. 51—55. "

27For examples examine issues of the Accounting Review and Journal
Of Accounting Research from 1968 forward.

 

18
next two chapters, are reviewed briefly below.28 Note, however, the
emphasis on briefly. The reviews are not intended to be detailed cri—
tiques; indeed, the purpose of this section is not to criticize
selected scholarly research efforts, but rather to emphasize the impli—
cation of the research results for the methodological design of any
subsequent market oriented research.

The Review. King argued that a significant portion of the price
movement associated with any individual security can be attributed to
the effect of market wide influences and industry wide influences.29 He
used what he called a "quick and dirty" form of factor analysis on the
price behavior of sixty—three securities continuously listed on the
New York Stock Exchange from June, 1927, through December, 1960, to
test his argument. And, indeed, he found that for his sample for the
time period examined that about half of individual security price move-
ment can be attributed to market and industry wide influences. However,
he also found that market and industry influences apparently decreased

over time and concluded:

 

28Market oriented studies are, in general, those studies which

attempt to draw inferences about a research issue based on an exami—
Iflltion of security price histories and/or volume of shares traded
1iii—stories. They are based on the assumption that security prices fully
?efflect all publicly available information and that security prices
JJIStantaneously and unbiasedly adjust for any new information which
becomes publicly available. (See Chapter III for more discussion.)
Ho‘uwor, even though market oriented studies are based on such an
aBemmption, it should be mentioned that an impressive amount of research
}“is been done to test the assumption. The results of this extensive
l'-'esea:rch almost uniformly support the reasonableness of the assumption.

See Eugene F. Fama, "Efficient Capital Markets: A Review of Theory
3:33 Empirical Work," The Journal of Finance, XXV (May, 1970), pp. 383-

7.

29Benjamin F. King, "Market and Industry Factors in Stock Price
nighavior,” The Journal of Business, mix (January, 1966), pp. 139-
0.

 

 

  

19
". . . the apparent diminution of the influence of the
market implies an increase in relative importance of the

unique parts of the variances, or at least those parts
that are e33c81ained by factors in addition to market and

industry."

Sharpe's 1963 extension of the pioneering work of Markowitz on
portfolio analysis suggested a relatively simple manner of adjusting
for or eliminating the market and/or industry effects on security price
changes, found by King above, so that the unique elements of price
change, as King called them, could be studied.31 This adjustment
technique involved the following steps:

(1) Estimate the coefficients AJ. and Bj in the simple regression
model below using data on both sides of (i.e. before and after) the
study or examination period:

P =A +B.M +R.
3t 3 at at

S + D.
Where P jt = ln(—g-t—'E);
jt-l
Sjt = the closing price of the jth stock at the end of
period t;

D. t = the dividend on the jth stock during the period the
J stock went ex dividend;

 

Nt
Mt = ln(N );
t—l
Nt = the closing value of some market and/or industry index

of period t;

and Rjt = the unique part of Pjt'

\______
3°Ibid., p. 157.
3liliiLlialm F. Sharpe, “A Simplified Model for Portfolio Analysis,"

Hm ement Science, IX (January, 1963 , pp. 277-293; and Harry M.
Herfiowitz, Portfolio Selection Efficient Diversification of Invest-
ments (New York, 1959). t

 

20

(2) Using the estimates of A and Bj from step (1) and the data

J
from the examination period, the unique element of security j price
change in examination period t is identified for study by solving the

equation below for Rjt:

R =P —A.-eM
it it J J t

where R3 t = the unique element of P j t for security j in the
examination period t;

S + D.
Pjt = 1n(—-g-E—AE);
jt—l
S't = the closing price of the jth stock at the end of the
J examination period t;
DJ t = the dividend on the jth stock if the stock went
ex dividend during the examination period;
N
Mt = ln(N—t—);
t-l
and Nt = the closing value of some market and/or industry

index of the examination period t.

Fama, Fisher, Jensen, and Roll, in a paper directed principally
at a study of common stock price response to stock splits, went to
Considerable effort to evaluate the above Sharpe model for general
market effect adjustment.32 They did so essentially by using the Sharpe
nloclel to adjust for the general market effect on a sample of forty—
seven securities during the period January, 1927, through December,

1959, and then observing the statistical properties of the residuals

 

01‘ unique elements of price change. They concluded:

 

   

3:Zlifugene F. Fama, et al., "The Adjustment of Stock Prices to New
Information,“ International Economic Review, X (February, 1969), pp. 1-
21.

 

 

 

 

 

21

“In sum we find that regressions of security returns on
market returns over time are a satisfactory method for
abstracting from the effects of general market conditions
on the monthly rates of return on individual securities."33

Beaver attempted to determine if the information content of annual

earnings announcements is useful to investors. He argued that if the

information content of annual earnings announcements is useful to inves-
tors, then, the market response of investors in the earnings announce—

ment week should be greater than the average response in the eight

Inreceding and eight succeeding nonreport weeks. Responses were measured

irr two forms: percentage price change and percentage volume of shares
‘trraded, each adjusted via the Sharpe technique for the market effect

fkrund by King above. He tested his argument on a sample of 1A3 firms

oxrer the period 1961 through 1965 selected on the basis of six criteria:

"(1) The firm must be on the COMPUSTAT tape; (2) the firm
must be a member of the New York Stock Exchange; (3) the
fiscal year must end on a date other than December 31; (A) no
dividends were announced in the same week as the earnings an-
nouncement; (5) no stock splits were announced during the 17
week period surrounding the announcement of earnings; and

(6) there were less than 20 news announcements per year
appearing in the Wall Street Journal.”

115—8 statistical results dramatically support his argument and he cone

<3luuded:

"The dramatic price and volume reaction indicates that
investors do look directly at reported earnings and do not
use other variables to the exclusion of reported earnings.
The evidence also indicates that news announcements

 

33Ibid., p. 7.

3“William H. Beaver, "The Information Content of Annual Earnings
‘Announcements," nnpirical Research in Accounting: Selected Studies,
.1268, a Supplement to Volume of the Journal of Accounti Research,
13p. 67-107.

35Ihiu., p. 70.

 

 

 

 

22

occurring prior to the earnings report do not entirelyfi>
preempt the information content of reported earnings."

May used Beaver's methodology above, with investor response
measured only as price change, in an attempt to determine if the infor—
mation content of quarterly earnings announcements is useful to inves—
tors.37 His study, however, was based on a sample of American Stock
Exchange firms because American Stock Exchange firms typically have
fewer news releases in any arbitrary period of time than New York Stock

Ebcchange firms. He concluded in a fashion similar to Beaver; i.e. that

investors do respond directly to reported earnings, in his case quar—
terly earnings, and that the information content of quarterly earnings
announcements is not entirely foreshadowed by competing and prior news

announcements. He further concluded, on the basis of some additional

empirical testing, that the five trading days immediately following a
cIllalrterly earnings announcement is a sufficiently long time, if not the
Optimal time, over which to observe investors' reactions to the infor-
Inattion content of a quarterly earnings announcement.

Ball and Brown, at about the same time Beaver was doing his
I‘eSearch above on the impact of annual earnings announcements on invest—
ment decisions, attempted to evaluate not only the impact but also the
timeliness of annual earnings announcements with respect to investment

de(kisions.38 To do so they used a sample of 261 firms whose (1)

y.—
361bido, p. 81+.

37Robert George May, "An Empirical Evaluation of the Significance
or Quarterly Accounting Data in Investor Decision Making" (unpublished
P114). dissertation, Michigan State University, 1970).

Bailey Ball and Philip Brown, "An Empirical Evaluation of Account-

ing Income Numbers," Journal of Accounting Research, VI (Autumn, 1968),
Pp. 159—177.

 

 

 

 

23

earnings data was available on COMPUSTAT tapes for each of the years
19h6—1966, (2) fiscal year ended December 31, (3) price data was avail-
able on CRSP (Center for Research on Security Prices) tapes for at least
100 months, and (A) annual earnings announcement dates were available
by reference to the Wall Street Journal. The first step of their
research methodology involved forecasting earnings for these firms,
grouping the firms on the basis of the forecast error sign, and then
observing in retrospect the signed pattern of the monthly price changes
of the firm groups for the forecast year (with all data being net of
any general market effect). The second step of their research
methodology involved, in essence, comparing the price changes of all
firms in the annual earnings announcement month with the yearly price
Changes of all firms (where the yearly price change was measured over
'the twelve months of which the annual earnings announcement month was
the last). Their findings were as follows:

(1) ". . . the information contained in the annual income

number is useful in that if actual income differs from

expected income, the market typically has reacted in the
same direction.

(2) “However, most of the information contained in reported
income is anticipated by the market before the annual report

is released.

(3) "About 75 per cent . . . of the value of all informa-
tion appears to be offsetting, which in turn implies that
about 25 per cent persists.

(A) ”Of the 25 per cent which persists, about half . . .
can be associated with the information contained in
reported income.

(5) "0f the value of information contained in reported
income, no more than about 10 to 15 per cent . . . has not
been anticipated by the month of the report.

 

 

 

2L,
and (6) I'The value of information conveyed by the income
number at the time of its release constitutes, on average,

only 20 per cent . . . of the value of all information

coating to the market in that month."39
However, even Ball and Brown admit that findings (3), (1+), (5), and
(6) above are due in large part, if not in total, to the existence of
competing new publicly available information about a firm frequently
entering the market throughout a firm's fiscal year. Therefore, to
amplify the value of and, consequently, the response to earnings
announcement data immediately following an earnings announcement, com-
peting new publicly available invormation about a firm must be
controlled between earnings announcements, as Beaver above attempted
to do, in part, by selecting for study only those firms with 20 or less
news releases during the examination year.

Beaver and Dukes used the Ball and Brown methodology (with minor
variation) without controlling for competing information sources in an
attempt to determine whether the "unexpected" portion of income when
using interperiod tax allocation or income determined without inter—
Period tax allocation, or "unexpected'' cash flow (defined as earnings
before tax allocation plus depreciation, depletion and amortization)
reIDOI‘ted via the annual earnings armouncement was most closely asso-
ciated with price changes.’+0 Both unexpected income and cash flow were
determined under five different forecast models. They assumed that the
“neXpected income or cash flow variable above most closely associated

with price changes was the variable with the greatest usefulness for
‘—
39Ibid., pp. 169—176.

hoflilliam H. Beaver and Roland E. Dukes, ”Interperiod Tax Allo—
cation, Earnings Expectations, and the Behavior of Security Prices,"

.The Accounti_ng Review, XLVII (April, 1972), pp. 320—332.

 

\s

0 1'11

. :I'Ih

25

the aggregate of investors and sirrdlarly for the associated total income

   

or cash flow concept. The sample of firms on which the study was based

consisted of
". . . 123 NYSE Compustat firms whose fiscal year end was
December 31 and whose methods of depreciation for tax and
for reporting purposes could be identified . . ."l‘

for the years 1963 through 1967. Their results indicated the following

ranking of the three accounting variables:

 

(1) income determined using interperiod tax allocation,

(2) income determined without interperiod tax allocation, and

(3) cash flow.

It should also be mentioned, although Beaver and Dukes did not, that

a close examination of the statistics for what they referred to as their
"best" forecast models and accounting variables would imply, in part, I ‘
a confirmation of the Ball and Brown results above, particularly Ball
and Brown results (1) and (2).

Benston attempted primarily to determine whether the following
tlrpes of information were of importance to a large number of investors
as measured by their market price response: (1) current period measure-
ments of net sales; net income before depreciation, amortization,
income taxes, and nonrecurring items; net income before nonrecurring
exZpense or income; net income net of tax before nonrecurring expense ,7
or income; and net income; and (2) prior period measurements of each ‘

0f the income variables in (1) such as are frequently reported in the .

\_—.—_—————
hillbido, p. 3260

 

   

 

  

26

ten or twenty year summaries of annual reports.42 He used as his sample
the intersection of firms on the CRSP and COMPUSTAT tapes. He used

five different multiple regression models to analyze the data. The

models differed primarily in terms of the assumptions about how inves-

tors use accounting data. Each model was of the general form:

where P't = stock prices of the common shares of company j in
J period t,

Ajt = published accounting data of company j in period t,
when the data became 'known'. . .,

D distribution to stockholders of company j of assets

or claims over assets (such as cash dividends) during

period t,

jt

(Stat = changes in dividends of company j in period t, that
affect investors' expectations,

(SMt = changes in general market conditions in period t,
that affect the market valuations of companies during
this period,

information that affects the market valuations of all
firms in a given industry, K, that becomes 'known' in
period t,

IKt

th = economic income generated by the assets of the company j
‘during period t, that change the present value of the

company,
81nd U

other information about company j that becomes 'knzgn'
ll

3t in period t that affects investors' expectations

Benston concluded, in part, from his extensive study that ". . . the

iJlformation contained in published accounting reports is a relatively

 

hzbeorge J. Benston, “Published Corporate Accounting Data and
stock Prices," 'rical Research in Accounti : Selected Studies,
JEEQZ, a Supplement to Volume 5 of the Journal of Accounti Research,
pp. 1-51;.

ABIbido' P. 1+.

 

 

a’

 

.nl

 

 

o
v

'v-

\a.

3.

~..‘

 

 

 

 

27
small portion of the information used by investors"hh and that ”within

this limited use, it appears that . . . the information is used prima—
#5

rily in the period in which it is made public. . . ." However as
with the Ball and Brown study above, part of Benston's results are due
to the existence of competing information, control for which should
amplify the importance of, and consequent response to published
accounting data. But it also would appear that such control is only
necessary within close time proximity of the announcement date of any
new information to which one wishes to measure investors' responses(s).
Gonedes attempted to determine whether the following accounting
variables reported in the annual earnings announcement had an impact
on investors' decisions during the period 1947 through 1966: (1) net
income/common equity, (2) current assets/total assets, (3) total debt/
total assets, and (h) net income/sales.)+6 He tested investors' reac-
tiorm to these data using the eighty firm intersection of COMPUSTAT and
CRSP firms during the period 1947 through 1966 and three reaction
"Kniels: (1) a linear model, (2) a disjunctive model, and (3) a COD!
Junctive model. A linear model assumes that a low impact with respect
tO <3ne ratio may be compensated for by a high impact with respect to
al’lo‘ther ratio. A disjunctive model assumes that high impact ratios

dknninate low impact ratios and high impact ratios are those in which

 

“‘Ibid. , p. 28.
”fluid.

héNicholas J. Gonedes, "Some Evidence on Investor Actions and
Accounting Messages—Part I," The Accounti Review, XLVI (April, 1971),
‘Pp. 320-328; and Nicholas J. Gonedes, "Some Evidence on Investor Actions
and Accounting Messages——Part II," The Accounti Review, XLVI (July,

1971). pp. 535-551.

 

28
large differences between expected and reported ratios exist. A
conjunctive model also assumes that high impact ratios dominate low
impact ratios, but high impact ratios are those in which small differ-
ences between expected and reported ratios exist. He concluded from
his analysis:

". . . All or virtually all of the impact on investors'
action that is associated with the contents of annual
accounting reports precedes the period during which the
report is issued and the three-month period prior to the
period during which the annual report is issued, i.e., the
time period encompassed by our tests.

"The model that we employed in order to secure surrogates
of investors' expectations with respect to accounting data
does not allow for adjustments of investors' expectations
in response to messages and events that appear before the
time periods encompassed by our tests (e.g., quarterly
accounting reports, messages prepared by brokerage houses,
and economic events). Such messages and events may facili-
tate investors' anticipatory behavior vis a via the contents
of annual accounting reports. Evidently, adjustments in
investors' expectations occur more frequently than the
frequency of adjustment permitted by the design of our tests,
tests that ignore interim messages and focus upon annual
accounting messages. It appears that any changes in inves—
tors' expectations associated with the contents of annual
accounting messages are impounded in investors' networks

of beliefs and expectations before the test period of our
analysis.

“Note that the above interpretation is consistent with the
existence of sources of information that compete with the
accounting process qua a supplier of information. Thus, the
annual accounting report may have no impact on investors'
expectations because of: (1) accounting reports other than
the annual report (e.g. quarterly reports) and (2) informa-
tion provided by sources that compete with thfi accounting
process in its role of information supplier.‘I 7

Tile implication of these results and conclusions on market based re-
search designs are the same as those for Ball and Brown, Beaver and
Ihlkes, and Benston above: To amplify the value of and, consequently,

\—————_—

AqGonedes, pp. 5h9—550.

 

 

29
the response to earnings announcement data, competing new publicly
available information about a firm must be controlled.

Baskin attempted to determine whether consistency exceptions
reported in the auditor's certificate, an integral part of the annual
report financial statements (not the annual earnings announcement),
had an effect on investors' decisions in 1965 through 1968.1+8 He
selected three samples of firms from the American Institute of Certified
Public Accountants' annual publication.Accounting Trends and Techniques.
The first sample, a control group, made no accounting changes in the
report period, and, consequently, were not subject to a consistency
exception. The second sample, experimental group I, made accounting
changes in the report period but notified investors of the change
(i.e. inconsistency) at the time of the annual earnings announcement
as well as at the time of annual report release. The third sample,
fnqperimental group II, made accounting changes in the report period and
notified investors only at the time of annual report release. Baskin

aI‘gued that

”. . . firms without accounting principle changes and firms
notifying investors of their accounting changes at annual
earnings announcement date report no 'new' information,
ceteris paribus, at annual report date. But firms that
notify investors about their accounting principle changes

in their annual reports rather than in earnings announcements

divulge 'new information.""+9
Accordingly he argued that the reaction of investors to annual reports
(If experimental group II should be different than the reactions of

\

thlba F. Baskin, "The Communicative Effectiveness of Consistency
Esameptions,” The Accounting Review, XLVII (January, 1972), pp. 38-
1.

“9mm. , p. at.

V 30
investors to the annual reports of experimental group I and the control
group whereas reactions of investors to the earnings announcements
should be the same for experimental group II and the control group but
different for experimental group I.

He concluded from his data analysis that ”. . . the consistency
exception Opinion [in the annual report] does not appear to have infor-
mation content for most investors."5O However, he further pointed out
that when the consistency exception for experimental group II involved
a large dollar effect the limited (not statistically significant)
reaction by investors at annual report date was larger than when the
consistency exception involved only a small dollar effect. Presumably
investors react to the majority of accounting information at earnings
announcement date rather than at annual report date and any response
at annual report date is elicited only by large inconsistencies with

the earnings announcement data.

Conclusions

A controversy exists over the question of whether the cost method
(If accounting for a corporation's long-term investments in common stock
‘Jf other corporations is better than the equity method of accounting
ier a corporation's long-term investments in common stock of other
("erorations. In the Opinion of some accountants the controversy
E3‘IXists because of insufficient empirical research addressing the contro—
“Versy. Furthermore, some accountants are of the Opinion that the

iresearch necessary to resolve the controversy should be market based.

SQIbid., p. 50.

31

In this vein ten market based research efforts, the results of which
have implications for the design of market based research, were
reviewed. The implications of these research efforts for subsequent
market based research efforts were:

(1) Investors' market based responses to accounting data must be
adjusted for general market influences if the unique elements are to

be studied;
(2) This adjustment can be accomplished effectively with the

Sharpe model;

(3) Investors' market based responses to accounting data may be
studied in the five trading days immediately following an earnings
announcement but may not be effectively studied in a corresponding
period following the publication of the annual report; and

(A) To amplify the importance of, and, consequent response to
accounting data reported via the earnings announcement, competing
iJIformation releases must be controlled but the control need only be
Inade for competing information releases within relatively near time

Proximity of the earnings announcement.

CHAPTER III

METHODOLOGY--CONCEPTUAL CONSIDERATIONS

The Efficient Market Theory

There is in existence a large bank of research evidence to support
what is referred to in the literature of finance as the semi-strong form
of the theory of efficient markets.1 This rather well developed theory
is that existing security market prices fully reflect all existing
publicly available information and that the security market instanta-
neously and unbiasedly adjusts these prices for any new information

made publicly available.2
3

Operationally, "instantaneous adjustment” is a random variable.
The value of the random variable for any given security is the time

<iLfference between the initial perception of the information and com-

Iilete adjustment for it.’+ Its expected value is very small.5

¥Eugene F. Fama, ”Efficient Capital Markets: A Review of Theory
:tkiiEmpirical Work,” The Journal of Finance, XXV (May, 1970), pp. 383—
17. .

2Ibid.

3Robert George May, "An.Empirical Evaluation of the Significance
(Df Quarterly Accounting Data in Investor Decision Making" (unpublished
?PruD. dissertation, Michigan State University, 1970).

thid.
5Ibid.
32

33
Operationally, ”unbiased adjustment" means that the total price
change associated with a given security is also a random variable whose
expected value is an unbiased adjustment for the new information.

The new public information referred to in the theory is defined

as anything newly perceived by the market in general.7

Quarterly Earnings Announcements as New Public
Information

Quarterly earnings announcements published in the financial press
are, under certain conditions, one type of new public information as
defined above. Assuming that all of a firm's news releases published
by the financial press are perceived by the market in general, these
conditions are that no such releases, predating the quarterly earnings

announcement, communicate the data contained in the quarterly earnings

announc ement .

fie Content of Quarterly Earnings Announcements and
fine Primary Criterion for Evaluating the Alternative
Mithods of Generating the Content of Quarterly Earn-

ings Announcements ,

Quarterly earnings announcements are essentially a summary of the

measured quarterly Operating results of an economic entity. The

announcements are issued ". . . to provide investors and others with

9

timely information as to the progress of the enterprise. . ." and

6Ibid.

7Fama, pp. 383-1.,17.

8American Institute of Certified Public Accountants, Accounting
Principles Board, OLinion 28 Interim Financial Reporting (New York,

1973). pp. 520-533.

91bide, pp. 522-5230

34
because the announcements ”. . . should be viewed primarily as an
integral part of an annual period [report]"10 they are ”in general . . .
based on the accounting principles and practices used by an enterprise
in the preparation of its latest annual financial statements. . . ."11
However, in many instances alternative measurement methods (accounting
methods) do exist. When this is the case, the American Accounting

Association Committee to Prepare A Statement of Basic Accounting Theory

has suggested that the alternative which provides the most potentially

useful measurement should be used. 12

The Locus and Definition of Usefulness Potential

The determination of usefulness potential of any measurement (in
this case, an accounting output) lies ultimately with the user and is
dependent upon the use(s) to be made of the measurement. 13 In the
<uontext of investment activity, the user of accounting outputs is the
inrvestor. The investor uses accounting outputs (1) to aid in predicting
.future net earnings as he, the investor, defines net earnings and (2) to
Etidin appraising the degree of certainty (or conversely the risk or
uncertainty) associated with these net earnings predictions.1 In the
theory of finance these two uses are combined in the concept of cer-

tétinty equivalents in which case it can be said that the investor uses

lolbido’ p0 5230

11Ibid.

12AmericanAccounting Association, A Statement of Basic Accounting
'Theony (Evanston, 1966), p. 3, p. 22, and p. 24.

13Ibido, p. 3 and p. 70

1“Ibid., pp. 23-24.

on

,D

'5'

‘1'

o.

 

35

accounting outputs to aid in predicting the certainty equivalents of
future net earnings.15 Therefore, the usefulness potential of account-
ing outputs from the vieWpoint of the investor may be defined, in the
abstract, as the degree to which these outputs aid in the prediction of
the certainty equivalents of future net earnings.
.Measuring and Compgring the Usefulness Potentials of
Alternative Accounting Outputs in an.Efficient Market

After the certainty equivalents of future net earnings are pre-
dicted, investors transform them into common stock prices by discounting
the certainty equivalents to the present at the pure rate of interest.

Now recall that the efficient market theory predicts an unbiased
adjustment of common stock prices for whatever new information becomes
jpublicly available, including quarterly earnings announcements.17 It
follows that if the common stock market is efficient, as the evidence
indicates,18 then the usefulness potentials of accounting outputs
reported via quarterly earnings announcements are fully realized, i.e.
Ilsed, and are reflected in common stock price behavior. (If the useful-
Iness potentials are not fully realized, the resultant price adjustment
:is likely not unbiased for the information available.)

If the usefulness potentials of accounting outputs reported via
Cluarterly earnings announcements are reflected in common stock price

behavior, as just argued, it should be possible to make inferences about

\

1SJames C. Van Horne, Financial Management and Poligy (Englewood
Cliffs, 1968), pp. 66-690

16Ibid., p. 66.

17Fama, pp. BBB-#170

18Ibid.

36
the relative usefulness potentials of alternative accounting outputs
by comparing the common stock price behavior of firms using one account-
ing method with those using an alternative accounting method. To see
the nature of the comparison and the resultant inference assume the
following set of conditions:

(1) Two alternative accounting methods (Al and A2) exist for
measuring the same economic event or events.

(2) Two identical firms (F1 and F2) exist, except that F1 uses
A1 above and F2 uses A2.

(3) At the beginning of time period 1 (Q1), investors estimate the
certainty equivalents of future net earnings (GEE) as they, the inves-
tors, define net earnings for F1 and F2 using as part of the input to
the prediction process the quarterly earnings announcement data (EA)
released at the end of time period 0 (QO). Inputs to the prediction
process at the beginning of 01 other than those attributable to A1 and
A2 are identical for both firms. These other inputs are referred to
as I.

(4) No events occur during 01 that stimulate investors to alter
the predictions made at the beginning of Q1 except the quarterly earn—
ings announcements (EA) released by F1 and F2 at the end of Ql.

(5) Investors are rational and have no capital constraints.

(6) The certainty equivalent estimates of the investor defined
net future earnings (GEE) made at the beginning of Ql for F1 were
perfect, but the corresponding estimates for F2 were imperfect, although
investors were unaware of this at the time of estimation. (Notice that

certainty equivalent estimates of the investor defined net future earn-

ings are perfect if these estimates exactly equal reported earnings

s
‘o
v

’\

\"

I‘M"

r:

a...

'7!

-.~

g.
‘I

0..

 

.
n.
I.

 

 

37
after reported earnings are transformed to correspond to the investors'
concept of earnings.)

Refer now to Figure 1. Figure 1 demonstrates schematically that
an independent agent or researcher unable to observe investors' security
valuation processes directly can, under the aSSWption that the effi-
cient market theory is valid and the additional assumptions above,
evaluate the usefulness potentials of alternative accounting methods'
outputs by observing the ultimate actions of investors' in the market
place and reach the same conclusions as he would if he were able to

observe the investors' valuation processes directly. 19

To explain, it must first be emphasized that Figure 1 relates to
a usefulness based evaluation of the accounting data released at the
end of Q0 g the information set I (assumed to be the same for both

firms, F1 and F2) existing before the end of 00, not the accounting
data released at the end of 01. Now obviously A1 provides data with
greater usefulness potential for investors than A2. (The usefulness
POtentials of I plus F1 EA at 00 and I plus F2 EA at Q0 were fully
utilized according to the efficient market theory, the only difference
between I plus F1 EA at 00 and I plus F2 EA at 00 was that attributable
at A1 and A2, I plus F1 EA at 00 resulted in a Erfect GEE whereas I

19Notice that the reasoning depicted in Figure 1 is not dependent

upon knowledge (assumed or otherwise) of am particular investor
expectational (valuation) mode1(s). This characteristic would at least
appear to be one advantage of this reasoning over the so call API
reasoning developed by Bell and Brown and used by Beaver and Dukes to
examine similar issues. (See Ray Ball and Phillip Brown, "An Empirical
Evaluation of Accounting Income Numbers," Journal of Accounting
Research, VI (Autumn, 1968), pp. 159-177; and William H. Beaver and
I{Gland E. Dukes, ”Interperiod Tax Allocation, Earnings Expectations,
and the Behavior of Security Prices,” The Accounting Review, XLVII
(April. 1972). pp. 320—332.)

I a ; u
no.4; WU

F

..~

All! II.

he.

8
3

coapm>nmmoo mo Hm>cq m.nccm< pcmocommocH as new soapm>nmmoo mo Ho>cq .mp0pmm>cH

IlGOflpw>hmmoo no ch>mA pcopmmmflm 039 Eoum_cOHpmsam>m pom soapmshomcH mo prmsonom

on oncomcsnoo mcmeHpmm one

.H enemas

.mwcflcnmo mo pdcocoo pecan op mmcflchmo oophomoh anommcmsp mnopmmth
**

.mmcficnmo mo vdcocoo m.nopmo>cw can

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

J D
qm as as me
o a mq _ _ mead H
_ AAOH>mnmm hopmo>cH _
. stonoo 8. 323a. 65 _
.ul _
u _ xom xomam
_ _ _ 8 so an E.
_ o u an? _ mafia H
iLW _ _
N<.
Geno Hammad _
once mH H¢_ _
m.:.|.|l _
cosmflomm_ _
coseOo “ _
u _ 8 can «We a
Oncucoz ma HMO mm 0 pm <m N
o x m<¢_lmocmnmm.HHQ HG pm .5 pm _ made H
, _
_ as we 1 am we _
_ mmooonm_c0fin *mmooonm
_ Imfiommcmne coapmsflpmfl. _
_ 3 as 3 sh _
n onmN me «Q Hm <m Hm +T
o AGAIcOcOHOMHflQ ac cam 00 pm «a an
a F E

h

.mmcflcnwm onspdm no: mo mascam>fisvo hpcfimphoo mmeHpmc whopmmth

.x.

mommmoonm
comfiamQEOo mmcflchmm
new cessapmm mmcflcamm
m.aopmm>sH poops soap
Im>ncmoo mo Hm>mq
m.pcmm< eccoccdmocH

mommooonm
comflhmasoo
mmcflcnmm now
opmeapmm mmcflchwm
adonm soapm>hcmoo
no .859
.mhovm®>flH

39
plus F2 EA at Q0 resulted in an imerfect CE, and all other things were
equal.) Furthermore, since the certainty equivalent estimate of Ql
earnings for F1 was perfect, there is no stimulus at the end of 01
for investors to revise certainty equivalent estimates of net earnings

for subsequent F1 quarters.20 If the certainty equivalent estimates of

net earnings for subsequent F1 quarters are not revised, then the
instantaneous, unbiased price adjustment for the F1 quarterly earnings
announcement released at the end of Ql is zero. In contrast, since

the certainty equivalent estimate of 01 earnings for F2 was imperfect,
there is a stimulus at the end of Q1 for investors to revise certainty
equivalent estimates of net earnings for subsequent F2 quarters.
FUrthermore, Weston and Brigham suggest, implicitly, and Francis cites
evidence that all certainty equivalent estimates of net earnings for
subsequent F2 quarters will be revised in the same direction, the direc—
‘tion corresponding to the directional difference between the transformed

reported Q1 earnings and the certainty equivalent estimate of Q1

earnings for F2.22 If the certainty equivalent estimates of net

20Daryl J. Bem, Beliefs, Attitude_s_L and Human Affairs (Belmont,
1970), p. 24. Bem argues and presents evidence that ". . . men possess
a drive toward cognitive consistency, and that, therefore inconsistency
acts as an irritant or stimulus which motivates individuals to change
their beliefs. . ." If inconsistency does not exist, then presumably
Peliefs will not change. In the situation under consideration above,
inconsistency does not exist (expected and transformed reported earnings
Correspond exactly), therefore subsequent earnings expectations

beliefs) should not change.

21Ibid. In this situation inconsistency does exist (expected and
transformed reported earnings do not correspond), therefore subsequent
ea~?l:'nings expectations (beliefs) should change.

F . 22J. Fred Weston and Eugene F. Brigham, Essentials of Managerial
lAlnance (Dallas, 1968), p. 238; and Jack Clark Francis, Investments:
@ysm and Management (St. Louis, 1972), p. 333.

4O
earnings for subsequent F2 quarters are revised as suggested, then the
instantaneous, unbiased price adjustment for the F2 quarterly earnings
announcement released at the end of 01 is nonzero.

Generalizing, under the conditions assumed above, the smaller the
price adjustment immediately following a quarterly earnings announcement,
the greater the usefulness potential of the output of an accounting
method to investors. However, this generalization needs to be quali-
fied slightly. The argument above implicitly assumes that all investors
experience the same earnings expectational change. This is highly
unlikely. It is more likely that investors will experience different
earnings expectational changes. If investors do experience different
expectational changes, then the observed price change is an average
response (discussed in more detail in the next section). ‘When the
observed price change is recognized as an average response the gener-
alization becomes as follows: The smaller the price adjustment immedi—
ately following a quarterly earnings announcement, the greater the
usefulness potential of the output of an accounting method to investors
on the average.

An interesting implication of the above argument is that under the
assumed ceteris paribus conditions (i.e., that F1 and F2 are identical
except that F1 uses A1 and F2 uses A2) and investor rationality if A1
' provides a smoother contribution to reported earnings than A2, then
A1 provides data with greater usefulness potential on average than A2
because the difference between expected and transformed reported earnp
ings will be minimized. However, one should not make more of this
inplication than is intended. It is not, for example, an argument in

favor of ”smoothing" or ”managing" earnings in the sense that these

#1

terms are generally used in the accounting literature. These terms, as
generally used, denote the manipulation of reported earnings by changing
accounting methodologies from period to period to produce a ”smooth”
earnings trend.23 The argument above, on the other hand, is based on
the assumption of consistent use of a chosen accounting methodology or
set of methodologies from period to period and is that, assuming consis—
tent use, the methodology or set of methodologies providing the
smoothest contribution to reported earnings is the best or has the
greatest usefulness potential for investors.

Finally, the generalization above, resulting from the reasoning
depicted in Figure 1, appears to be in conflict with the research of

Beaver and.Dukes reviewed in Chapter II.25

However, reconciliation can
be obtained.

Although not emphasized in the Chapter II review, because it was
not important for the purpose intended by that review, Beaver and Dukes
made the following two critical assumptions: (1) Investors on average

are able to make the appropriate (or at least approximately appropriate)

transformation from one accounting measurement (accounting method

 

23See for example, M. J. Gordon, "Postulates, Principles and
Research in.Accounting,” The Accounting Review, XXXIX (April, 1964),
pp. 261-262; and Russell M. Barefield and Eugene E. Comiskey, "The
Smoothing Hypothesis: An.Alternative Test,” The Accounting Review,

2[Within this implication is a suggestion for an alternative
research methodology for the issue under consideration in this research
project as well as an additional research project to determine if
smooth earnings trends really do elicit smooth price behavior, both of
which are beyond the sc0pe of this thesis. See the section entitled
”Suggestions for Further Research” in Chapter V.

25William H. Beaver and Roland.E. Dukes, “Interperiod Tax Alloca-
tion, Earnings Expectations, and the Behavior of Security Prices,"
The Accountinijeview, XLVII (April, 1972), pp. 320—332.

A2
output) to alternative accounting measurements?6 and (2) Investors on
average are omniscient to the extent that they know and will use the
alternative measurement in (1) which is most useful.27 Using five
different earnings expectational models and Sharpe's market return
model,28 they, then, examined the relationship between the unexpected
behavior of various earnings measurements (really earnings and cash
flow measurements) and an index of the behavior of associated security
returns (prices) computed over 12 months, the 11 months preceeding the
earnings announcement and the earnings announcement month. They cone
tended that the earnings measurement whose unexpected portion demon-
strated the highest positive algebraic sign correlation with the index
of unexpected security return behavior was the measurement most used
and consequently the measurement most useful to investors on average.29
However, this contention is critically dependent upon the above assump-
tions.

Now to begin reconciliation of the apparent conflict, relax
assumption (2). Then, since investors are not presumed necessarily to
use the most useful measurement, from the correlation analysis it can
only be said that the earnings measurement whose unexpected portion
demonstrates the highest positive correlation with the indexof unex-
pected security return behavior is simply the measurement most used.

However, if the index magnitude is examined it can be said ex post that

 

261bid., p. 320.

271bido, p. 3210

28William F. Sharpe, "A Simplified Model for Portfolio Analysis,"
Management Science, IX (January, 1963), pp. 277-293.

29Beaver and Dukes, p. 321 and pp. 331-332.

#3
the accounting measurement associated with the smallest absolute index
value was the most useful on average since the index is an indicator of
unexpected return, itself presumably a function of unexpected earnings,
and minimization of unexpected earnings presumably indicates greater
usefulness of estimation data inputs than maximization of unexpected
earnings. This is particularly true when as in the Figure 1 reasoning
no events (news releases, etc.) specific to the security other than
the earnings announcements are allowed to influence the security's
:return behavior for, as Ball and Brown point out, in that situation the

index used by Beaver and Dukes is expected to be affected only, or at

least only appreciably affected, by the security return behavior in the

month of earnings announcement.30 Furthermore, the inference resulting

:firom.an.examination of the index magnitudes just described can also be
made in the absence of assumption (1) above if investors can be iden-
tified as using alternative measurements associated with different
securities, and in the absence of any assumed earnings expectational
mOdels as well.

To summarize, Beaver and Dukes made two critical assumptions:
(1) Investors on average are able to make the appropriate transformation
from one accounting measurement to alternative accounting measurements;
and (2) Investors on average are omniscient to the extent that they know
and will use the alternative measurement in (1) which is most useful.
When these assumptions are relaxed and the Beaver-Dukes model reexamined
with the measurement of usefulness in mind the apparent conflict

\

1 30Ray Ball and Phillip Brown, "An Erupirical Evaluation of Account-
11g Income Numbers,” Journal of AccountingResearch, VI (Autumn, 1968),

’ 16A.

 

AA
disappears, for then, it can be said from their model that the account-
ing measurement associated with the smallest security return index
value is the most useful on average as it has been said from the
Figure 1 reasoning—~the accounting measurement associated with the

smallest price change (return) is the most useful on average.

Extending the Conceptual Framework to Volume Measures
The conceptual framework develOped above is couched in terms of
observing investors' reactions to new information as price changes.

However, one can also observe investors' reactions as the volume of

31

shares traded. In fact, at least one researcher has argued that the

volume of shares traded may be a more sensitive measure of individual
investors' reactions than is price change, while price change is a more

sensitive measure of the total market reaction (reaction of investors

32

in the aggregate) than is volume of shares traded. He argues as

follows:

"An important distinction between the price and volume tests
is that the former reflects changes in the expectations of
the market as a whole while the latter reflects changes in
the expectations of individual investors. A piece of infor-
mation may be neutral in the sense of not changing the
expectations of the market as a whole but it may greatly
alter the eXpectations of individuals. In this situation,
there would be no price reaction, but there would be shifts
in portfolio positions reflected in the volume."3

 

1
3 William H. Beaver, "The Information Content of Annual Earnings

Announcements," Empirical Research in Accounting: Selected Studies,
1968, a Supplement to Volume 6 of the Journal of Accounting Research,
pp. 67-101.

321bid., p. 69—70.

33Ibid.

#5
This argument, implicitly, but accurately, indicates that price change
is a reflection of the market's average earnings expectational change.
But many different distributions of individual investors' earnings
expectational changes may result in the same average (or market) earn~
ings expectational change. TWO such distributions are illustrated in
Figure 2 below.
Number of Investors

Experiencing a Particular
Expectational Change

Distribution 1

‘\\\\\\‘\\_‘ Distribution 2

    

 

  

 

r--—--—-

 

 

Lit Continuum of
Mean.(Average) Possible Earnings
Expectational Change Expectational
(Uniquely associated with Changes

the observed price change)

Figure 2. Illustrative Earnings Expectational Change DistributionsBL’

It should be obvious that the degree of consensus about earnings
expectational changes is greater if individual investors' earnings
expectational changes are described by distribution 1 than if they are

described by distribution 2.

 

3L'Notice that the peak of distribution 1 is at a higher frequency
than the peak of distribution 2. The reason for this phenomenon is the
implicit assumption that the sum of investors associated with distri-
bution 1 is the same as the sum of investors associated with distri-
bution 2. Furthermore, this assumption would appear reasonable, even
in the real world, since all market participants (investors) are potenp
tial buyers or sellers of any given security.

#6

Now assume that all investors who experience earnings expectational
changes different than the mean earnings expectational change adjust
their portfolio position by buying or selling the security.35 If the
common stock market is a market of many small investors each holding or
wishing to hold the same number of shares of the security, then, the
volume of shares traded would be smaller if the distribution of indi-
vidual investors' earnings expectational changes were described by
distribution 1 than if it were described by distribution 2. (Small
investors and large investors are defined in the microeconomic sense.
Small investors are those who hold or desire to hold securities of a
sufficiently small number as to render them unable to independently
affect the security price. Large investors, in contrast, are those
whose holdings or desired holdings render them able to independently
)36

affect a security price. The number of investors holding their

portfolio positions in distribution 1 is larger than in distribution 2

 

35This would probably not occur in the real world. A more likely
situation would be that investors experiencing earnings expectational
changes in a small range around the mean earnings expectational change
would hold their portfolio positions, whether that be as a holder or
nonholder of the security. An example of such a range is illustrated
in Figure 2 as R. The assumption above, made for the sake of illu-
stration, is simply that this range is zero.

36The composition of the market in terms of small versus large
investors is one reason the American Stock Exchange is used in this
research effort rather than the New York Stock Exchange. See the sec-
tion entitled ”Selection Criteria Implied by the Ideal Conditions” in
Chapter IV. A Wall Street Journal article of May 23, 1973, reports that
”. . . individual investors make up 80% of AMEX volume.” (See Richard E.
Rustin, ”AMEX Board Divided on Whether to Follow Big Board for General
Broker—Fee Rise,” Wall Street Journal, May 23, 1973, p. A.) Also notice
that the assumption that all small investors hold or wish to hold the
same number of shares of the security is stricter than necessary. The
necessary assumption is that the shares held or desired by investors
with expectational changes to the left of the mean expectational change
are the same as those to the right of the mean.

47

leaving fewer shares to be traded in distribution 1 than in distribu-
tion 2. In this vein, the following generalization is made: The
smaller the volume of shares traded immediately following a quarterly
earnings announcement, the greater the degree of consensus regarding
expectational changes, and, consequently, the greater the degree of
consensus among investors regarding the usefulness potential of an
accounting method output.
Interpreting Price Change Measprements and Volume
of Shares Traded Measurements Together

In preceding paragraphs two forms of investor response measurements
were discussed. These two response measurement forms were, respec-
tively, price change and volume of shares traded. It was argued, under
a set of very strict conditions that, of two alternative accounting
methods' outputs, the output associated with the smaller price change
immediately following a quarterly earnings announcement has the greater
usefulness potential for investors on the average; and the output asso-
ciated with the smaller volume of shares traded immediately following a
quarterly earnings announcement is the output for which there is greater
consensus among investors regarding the output's usefulness potential.
Now if either of these two measurement form comparisons are used inde-
pendently in evaluating two alternative accounting methods' outputs
(A1 and A2), three possible results exist. The price changes asso-
ciated with A1 and A2 could be equal; the price change associated with
A1 could be greater than the price change associated with A2; and the
price change associated with A1 could be smaller than the price change
associated with A2. Entirely analogous results could exist for the

volume of shares traded measurement form comparisons. But when the

A8

two measurement form comparisons are considered together nine possible
result combinations exist. These nine possible result combinations are
arrayed in Table 3. Each possible combination is numbered in Table 3

and discussed below with reference to its table number.

TABLE 3

POSSIBLE COMBINATIONS OF RESULTS OF PRICE CHANGE COMPARISONS
AND VOLUME OF SHARES TRADED COMPARISONS

 

Possible Results of Volume of Shares
Possible Results of Price Traded Comparisons
Change Comparisons

 

VA1 = VA2 VAl > VA2 VA1'< VA2

 

 

 

A PAl = A PA2 Comblilatlon Comblgatlon Comblgatlon
A PAl > A P A2 Comblzatlon Comblglatlon Combigation _
A P A1 < A P A2 Comblpatlon Comblrelatlon Comblgatlon

 

 

Combination 1. The results designated combination 1 are displayed
schematically in Figure 3. The figure indicates that Al and A2 outputs
have the same usefulness potentials for investors, on average, and that
the degree of consensus regarding the usefulness potentials of A1 and A2
outputs are similar. In the event that this combination of response
comparisons is obtained, the evidence would indicate that the choice

between using A1 or using A2 is of no consequence to investors.

49

Number of Investors
Experiencing a Particular
Expectational Change

Distribution associated with
both A1 and A2

 

 

L
I

 

Continuum of Possible
Mean (Average) Expectational Expectational Changes
Change (Uniquely associated -

with the observed price change

for both A1 and A2)

Figure 3. Possible Results of Comparisons——Combination 1 Schematic

Combination 2. Combination 2 is illustrated in Figure A. Since
the observed price changes associated witthl and A2, in this situation,
are equal, apparently the outputs of A1 and A2 have equal usefulness
potentials for investors on average. In contrast since the expecta-
tional changes associated with A2 are less widely dispersed than those
associated with A1, apparently there is greater consensus among inves-
tors regarding the usefulness potential of A2 outputs than there is
regarding A1 outputs. Now, certeris paribus, greater consensus appears
more desirable than less consensus. Therefore, in this situation, the
evidence would appear to indicate that accounting method A2 should be

used rather than accounting method A1.

50

Number of Investors
Experiencing a Particular
Expectational Change

Distribution associated
with A2

 

Distribution associated
with A1

\\\\\\\“~— Continuum of Possible

Mean (Average) Expectational Expectational Changes
Change (Uniquely associated

with the observed rice change

for both A1 and A2‘

 

 

 

‘b--——

Figure A. Possible Results of Comparisons-—Combination 2 Schematic

Combination_3. Figure 5 illustrates the situation labeled combi-
nation 3. This situation, its analysis, and the resulting conclusion
are entirely analogous to combination 2 except that Al and A2 are
interchanged. Therefore only the conclusion for this situation is set
forth: The evidence would appear to indicate that accounting method

A1 should be used rather than accounting method A2.

51

Number of Investors
Experiencing a Particular
Expectational Change

Distribution associated
with A1

Distribution associated
with A2

\\\\\\““—‘ Continuum of‘Possible

Mean (Average) Expectational Expectational Changes
Change (Uniquely associated

with the observed price change

for both A1 and A2)

 

u-----

 

Figure 5. Possible Results of Comparisons——Combination 3 Schematic

Combination A. Combination A exists if the price change associated
with A2 is less than the price change associated with A1 whereas the
volume of shares traded associated with A1 is equal to the volume of
shares traded associated with A2. This is depicted in terms of inves-
tors' expectational change distributions in Figure 6 and would appear
to indicate that accounting method A2 should be used rather than
accounting method A1 since on average A2 outputs have greater usefulness
potential than A1 outputs, as indicated by the comparative price changes
while the degree of investor consensus regarding the usefulness potenr
tials of the outputs is equal, as indicated by the comparative volume

measurements.

52
Number of Investors
Experiencing a Particular
Expectational Change

Distribution associated with A2

Distribution associated with A1

 

qp-———--—
q-—-’--——---

Continuum of Possible

 

Mean (Average) Mean (Average) Expectational Changes
Expectational Change Expectational Change

(Uniquely associated (Uniquely associated

with observed price with observed price

change for A2) change for A1

Figure 6. Possible Results of Comparisons-—Combination A Schematic

Combination 5. The results designated combination 5 are displayed
schematically in Figure 7. The figure indicates that A2 outputs have
greater usefulness potential for investors, on average, than A1 outputs.
Furthermore, the figure indicates a greater degree of consensus among
investors regarding the usefulness potential of A2 outputs than of
A1 outputs. The evidence, then, would appear to indicate that when
faced with a choice between using accounting method A1 or using account—

ing method A2, accounting method A2 should be chosen.

53
Number of Investors
Experiencing a Particular
Expectational Change

Distribution associated with A2

 

Distribution associated with A1

 

 

qb-—--~—-

 

I
I
I
I
L

 

. Continuum of Possible

Mean (Average) Mean (Average) Expectational Changes
Expectational Change Expectational Change

(Uniquely associated (Uniquely associated

with observed price with observed price

change for A2) change for A1)

Figure 7. Possible Results of Comparisons-—Combination 5 Schematic

Combination 6. Combination 6 is illustrated in Figure 8. Since
the observed price change associated with A2 is smaller than the
observed price change associated with.A1, the outputs of A2 apparently
have greater usefulness potential for investors, on average, than the
outputs of A1. However, since the volume of shares traded associated
with A2 is greater than the volume of shares traded associated with
A1, there is apparently a smaller degree of consensus among investors
regarding the usefulness potential of A2 outputs than there is regard-
ing the usefulness potential of A1 outputs. Now it would appear
reasonable to assume that both greater usefulness potential and greater
consensus among investors regarding usefulness potential are more
desirable than lesser usefulness potential and lesser consensus. How&
ever, assuming that the usefulness potential measure, i.e. price change,
represents the consensus Opinion of a ”substantial" group of investors

the American.Accounting Association Committee to Prepare A Statement of

54
Basic Accounting Theogy contends that the usefulness criterion has

37

primacy over all other criteria. Therefore, in this situation it
appears that accounting method A2 should be used rather than accounting
method Al even though there exists a lesser degree of consensus among
investors regarding the usefulness potential of A1 outputs than exists
regarding A2 outputs.

Number of Investors

Experiencing a Particular

Expectational Change

r

Distribution associated with A2

Distribution associated with A1

 

qp—-—--—

4b-—-—

 

Continuum of Possible

Mean (Average) Mean (Average) Expectational Changes
Expectational Change Expectational Change

(Uniquely associated (Uniquely associated

with observed price with observed price

change for A2) change for A1)

Figure 8. Possible Results of Comparisons-Combination 6 Schematic

Combination 7. Combination 7, illustrated in Figure 9, is entirely
analogous to combination A with A1 and A2 interchanged. Therefore, only
the conclusion is set forth here: It appears that accounting method A1

should be used rather than accounting method A2.

 

37AmericanAccounting Association, p. 3, p. 22, and p. 2A.

55
Number of Investors

Experiencing a Particular
Expectational Change

Distribution associated with A1

Distribution associated with A2

 

I
l
I
I
I
I
I
I
l
1

 

. % Continuum of Possible
Mean (Average) Mean (Average) Expectational Changes
Expectational Change Expectational Change
(Uniquely associated (Uniquely associated
with observed price with observed price
change for A1) change for A2)

Figure 9. Possible Results of Comparisons-—Combination 7 Schematic

Combination 8. The situation designated as combination 8 is
depicted schematically in.Figure 10 and is entirely analogous to combi—
nation 6 with Al and A2 interchanged. The discussion of combination 6
indicates that the usefulness potential criterion has primacy over all
other criteria, and, therefore, in this situation, it appears that
accounting method A1 should be used even though the degree of consensus
among investors regarding the usefulness potential of A1 outputs is
smaller than the degree of consensus regarding the usefulness potential

of A2 outputs.

56
Number of Investors

Experiencing a Particular
Expectational Change

Distribution associated with A1

Distribution associated with A2

 

 

I
I
I
I
I
l
I

. Continuum of Possible
Mean (Average) Mean (Average) Expectational Changes
Expectational Change Expectational Change

(Uniquely associated (Uniquely associated

with observed price with observed price

change for A1) change for A2)

 

Figure 10. Possible Results of Comparisons-Combination 8 Schematic

Combination 9. Finally, Figure 11 illustrates combination 9, a
situation entirely analogous to combination 5 with A1 and A2 inter-
changed. The conclusion, in the event that combination 9 results are
obtained, is that accounting method A1 should be used rather than

accounting method A2.

57
Number of Investors
Experiencing a Particular
EXpectational Change

Distribution associated with A1

 

 

Distribution associated with A2

 

 

--——----

 

1 %** Continuum of Possible
Mean (Average) Mean (Average) Expectational Changes
Expectational Change Expectational Change
(Uniquely associated (Uniquely associated
with observed price with observed price
change for A1) change for A2)

Figure 11. Possible Results of Comparisons-—Combination 9 Schematic

Summgpy

Assuming that a given firm's accounting outputs are used in the
prediction of certainty equivalents of that firm's future net earnings
which when discounted at the pure rate of interest equal the current
share price of that firm's common stock, the usefulness potential of
the firm's accounting outputs to investors, defined as the degree to
which the outputs aid in the prediction of future earnings certainty
equivalents, should be reflected in the firm's common stock price
behavior. Specifically, if the certainty equivalent estimates of future
net earnings made using accounting outputs at the beginning of period
one are perfect, then the earnings reported at the end of period one
via the quarterly earnings announcement (assuming no intervening changes
in certainty equivalent estimates and transformation of reported earnp
ings to correspond with investors' concept of earnings) exactly equal

the certainty equivalent estimate of earnings for period one. In this

58
situation, there is no stimulus for revising subsequent earnings cer-
tainty equivalent estimates, and, therefore, the expected price adjust-
ment at the end of period one is zero. In contrast, if the certainty
equivalent estimates of future net earnings made using accounting
outputs at the beginning of period one are imperfect, then the earnings
reported at the end of period one via the quarterly earnings announce-
ment (qualified as above) will not exactly equal the certainty equiva-
lent estimate of earnings for period one. In this situation, there is
a stimulus for revising subsequent earnings certainty equivalent
estimates. If the revisions are all in the same direction, as suggested
by Weston and Brigham and Francis, the expected price adjustment
immediately following the quarterly earnings announcement released at

38

the end of period one is nonzero. Generalizing, the smaller the

price adjustment immediately following a quarterly earnings announcement
the greater the usefulness potential of the output of an accounting
method to investors on average.

When the conceptual framework, initially deveIOped in terms of
observing the effect of usefulness potential on common stock price
behavior, is extended to observing the effect of usefulness potential
on common stock share trading volume, another decision rule results for
the behavior of share trading volume. The smaller the volume of shares
traded immediately following a quarterly earnings announcement the

greater the degree of consensus among investors regarding the usefulness

potential of the outputs of an accounting method.

 

38Weston and Brigham, p. 238; and Francis, p. 333.

59
Finally if greater usefulness potential and greater consensus are
regarded as better than lesser usefulness potential and lesser consensus
and the two decision rules are used together, the usefulness potential
criterion has primacy over the consensus criterion, the consensus
criterion essentially being used only in the event that alternative
accounting methods' outputs have the same usefulness potential for

investors on average.

CHAPTER IV

METHODOLOGY-APPLICATION CONSIDERATIONS

Introduction

As stated in Chapter I the primary purpose of this research effort
is to provide empirical evidence of whether cost method generated data
appears to have more, less, or the same usefulness potential as equity
method generated data for investors in corporations with material
unconsolidated common stock investments. Under ideal conditions this
purpose could be pursued by direct empirical application of the gener-
alized decision rules developed in Chapter III. However, ideal
conditions do not always exist in the real world. To the extent that
real world conditions and ideal conditions are not the same, compensa-
tion must be made as much as possible for the difference. Accordingly,
in an effort to make possible the pursuit of the primary purpose of this
research effort and at the same time make possible a test of the
empirical applicability of the Chapter III conceptual framework, several
real world application considerations are developed in this chapter,
including sample firm selection criteria, adjustment techniques for
uncontrollable exogenous influences on investors' market based
responses, and a statistical method of comparing investors' market based

responses to accounting data.

60

61
Ideal Versus Real World Conditions

The ideal conditions under which the decision rules should be
applied are specified by the assumptions made in Chapter III. They
are:

(1) Two alternative accounting methods exist for measuring the
same economic event or events.

(2) Two identical firms exist, except that one firm uses one of
the accounting alternatives in (1) above and the other firm uses the
other alternative.

(3) At the beginning of the testing period, investors estimate
the certainty equivalents of future net earnings for each of the firms,
the only difference in the inputs to the estimation process associated
with the two firms being that attributable to the two accounting
alternatives under consideration.

(A) No events occur during the period between quarterly earnings
announcements that stimulate investors to alter the estimates made at
the beginning of the period.

(5) Investors are rational and have no capital constraints.

(6) Market participants are small investors rather than large
investors.

The first ideal condition above is compatible with real world
conditions for this research effort. The cost method and the equity
method were during the period covered by this research allowable
feasible alternative methods of accounting for a corporation's long-term

O I O 1
investments in common stock of other corporations.

 

1See the section in this chapter entitled "Time Period Covered by
the Research? for additional discussion.

62

The second ideal condition above is likely not compatible with real
world conditions for this research effort. It is, indeed, highly
unlikely that two firms exist which are identical except that one firm
uses the cost method and the other firm uses the equity method. Two
measures will be taken to compensate for the difference between ideal
and real world conditions. First, two groups of firms will be used
instead of simply two individual firms. The use of two groups of firms
as a compensating measure is based on the assumption that whatever the
difference between any pair of firms, one from the cost group and one
from the equity group, the difference between another pair of firms
will counterbalance it, thus resulting in two very similar, even if not
identical, groups. Second, in order to minimize the differences between
any pair of firms, each firm included in either of these groups will be
selected in compliance with a rigorous set of criteria to be discussed
later.

The third ideal condition above has two parts—~(1) investors
estimate the certainty equivalents of future net earnings and transform
these estimates into common stock share prices, and (2) the only differ-
ence in the inputs to the estimation process associated with the two
groups of firms is attributable to the two accounting alternatives under
consideration. Part one of this ideal condition is based on the theory
of finance and will be assumed to hold.2 Part two of this ideal condi-

tion, however, likely is not consistent with real world conditions for

 

2Jerome B. Cohen and Edward D. Zinbarg, Investment Analysis and
Portfolio Management (Homewood, 1967), p. 2A9; James C. T. Mao,
Quantitative Analysis of_ Financial‘Decisions (London, 1969), p. A7A;
and James C. Van Horne, Financial Management and Policy (Englewood
Cliffs, 1968), pp. 66—69.

63
any arbitrary, or randomly chosen, pair of firms. Four measures will
be taken to compensate for this inconsistency between ideal and real
world conditions.

First, as discussed above, two groups of firms will be used. The
reason is similar to that already given for using two groups rather than
two individual firms.

Second, also discussed above, each firm included in either of the
groups will be selected in compliance with a rigorous set of criteria.
The use of selection criteria as a compensating measure is based on the
assumption that if firms are very similar then the inputs to the esti-
mation process associated with each firm will be very similar. Since
two of the most important inputs to the estimation process are
investors' appraisals of profitability and liquidity, the selection
criteria will include specifications for both of these areas.3

Third, assuming that the more current information inputs to the
estimation process are the more important inputs and consequently that
the effect of differences in older information inputs is less than the
effect of differences in newer information inputs, each firm included
in either of the groups will be one for which no news releases specific
to the firm occurred in the quarter immediately preceding the testing
A

period except the quarterly earnings announcement for that quarter.

The result of this proviso is that the most important input to the

 

3Cohen and Zinbarg, pp. 219-355.

hGeorge J. Benston, ”Published Corporate Accounting Data and
Stock Prices," Empirical Research in.Accounting;f Selected Studies,
1967, a Supplement to Volume 5 of the Journal of Accounting Research,

pp- 1-5h-

6b,
estimation process is the quarterly earnings announcement immediately
preceding the testing period.

Finally, since quarterly earnings announcements result from com-
bining outputs of many accounting methods each having a certain useful—
ness potential, if firms are also selected such that accounting methods
other than the cost method and the equity method are the same for each
group, then differences in usefulness potentials of the most important
input to the estimation process associated with the firms in each group
should be primarily attributable to differences in the usefulness
potentials of cost method generated data and equity method generated
data.5 Therefore, firms included in either group will be limited to
those which use the same accounting method in each of five critical
accounting areas (e.g. inventory valuation) and do not disclose by
footnote, or otherwise, the results of applying an alternative. This
nondisclosure limitation will also be applied to the way firms account
for and report the consequences of holding long—term investments in

common stock of other firms.

 

sGeorge H. Sorter and Selwyn W. Becker, "Corporate Personality as
Reflected in Accounting Decisions: Some Preliminary Findings," in
Accountipg and Its Behavioral Implications, William J. Bruns, Jr.
and Don T. DeCoster (eds.)—(New'York, 1969), pp. 267-277. Sorter and
Becker have suggested that firms may exhibit a personality in their
choice of accounting methods. The implication of this suggestion for
this research effort is that firms which choose the cost method,
generally regarded as more conservative than the equity method, may
consistently choose other conservative accounting alternatives. If
this situation were to result in this research effort, the two groups
of firms might not be similar with respect to other accounting data
generating systems and any difference observed in the way investors
react to the quarterly earnings announcements of the two groups could
not be attributed solely to the differences inherent in the cost and
equity methods. To minimize this possibility, firms will be selected
for this research effort so that the accounting methods in five critical
areas other than the long-term investment accounting area are the same
for each group.

65

The fourth ideal condition is that no events occur during the
period between quarterly earnings announcements that stimulate inves-
tors to alter the estimates made at the beginning of the testing
period. Although this ideal condition is probably not entirely com-
patible with real world conditions, one measure may be taken to minimize
the difference between the ideal condition and real world conditions.
This measure is to include only firms in either group which have no news
releases specific to the firm during any quarter for which the firm is
included except quarterly earnings announcements. By taking this
measure the difference which does exist between the ideal condition and
the real world conditions may be attributed to a general market effect,
i.e. the effect of news releases by any firm on the prices of all other
firms, and to the effect of nonppublic information. Techniques used
to adjust for these two effects which cannot be controlled will be
discussed later in this chapter under the section entitled "Data Adjust—
ments.”

The fifth ideal condition is that investors are rational and have
no capital constraints. That investors are rational in the real world
as well as in an ideal one is simply assumed, although the evidence in
support of the efficient market theory tends to provide empirical
support for the assumption as well.6 The ideal condition that investors
have no capital constraints is obviously violated in the real world.

The existence of real world capital constraints is directly related to
the effect of nonppublic information (each investor's personal capital

constraint being one form of non-public information) discussed above.

 

6‘Eugene‘F. Fama, "Efficient Capital Markets: A Review of Theory
and Empirical Work,” The Journal of Finance, XXV (May, 1970), pp. 383-
A17.

66
As already indicated, the adjustment for the effect of nonppublic infor-
mation will be discussed later in this chapter under the section enti-
tled "Data Adjustments."

Finally, the sixth ideal condition is that market participants are
small investors rather than large investors, small and large being
defined in terms of the number of shares held or desired, however, with
no precise number or range being designated as the level of distinction.
This ideal condition is, at least, approximated in the real world

capital market known as the American Stock.Exchange.7

For this reason,
as well as others discussed later, firms included in either group will
be limited to American Stock Exchange listed firms.
Selegtion Criteria Implied by the Ideal Conditions
and Procedures for Determining Compliance With
TEESe‘Criteria

The discussion above indicates that firms (or more properly firm
quarters) included in this research effort will be selected in accord—
ance with the following set of criteria:

(1) Only firm quarters in which the associated firms use the cost

method or the equity method to account for long-term investments in

common stock of other corporations will be selected. Furthermore, no

 

7Richard E. Rustin, "am Board Divided on Whether to Follow Big
Board for General Broker-Fee Rise,” Wall Street Journal, May 23, 1973,
p. 40

67
disclosure of the results of applying an alternative accounting method
may be disclosed in footnote or otherwise.8

The medium to be used in determining whether firm quarters satisfy
this criterion is the associated firm's annual report for the year
containing the quarter under consideration. In applying this criterion
it is assumed that unless an accounting method change is indicated by
footnote in the annual report the firm has consistently applied in each
quarter of the year the method of accounting for long—term common stock
investments disclosed in its annual report. If an accounting method
change is indicated by footnote in the annual report, then quarterly
reports will be examined to determine which, if any, firm quarters may
be selected.

(2) Each firm quarter selected must have no news releases specific
to the associated firm occurring during that quarter except a quarterly
earnings announcement. The medium to be used in determining whether
firm quarters satisfy this criterion is the Wall Street Journal Index.

This medium lists by firm and date all news releases specific to any

 

81t should be recognized, however, that frequently the primary, if

not the only, financial statements of an investor corporation are
consolidated statements. Furthermore, these consolidated statements
may be the result of consolidating the accounting data of the investor
and all of its investees or the result of consolidating the accounting
data of the investor and only some of its investees. However, regard—
less of the extent of consolidation, all investments which are consoli-
dated are being reported, essentially, under the equity method
accounting concept, and it should be added will be considered as such
in applying this criterion. Finally, upon preliminary review of one
hundred forty annual reports of publicly owned corporations, it appears
that most investor corporations consolidate at least one investee;
therefore, for practical reasons (1) firms in the cost group may
include those with some investees consolidated, the remainder being
accounted for using the cost method and, (2) firms in the equity group
may include (a) firms with all investees consolidated and (b) firms
with some investees consolidated, the remainder being accounted for
using the equity method.

68
given firm reported in the Wall Street Journal in a given year.
Although other media exist which report business news, indexes of all
news releases specific to any given firm by firm and date are not avail-
able for these media. Therefore, although a detailed examination of
these other media may be both desirable and feasible, for practical
reasons it shall not be attempted.

(3) Each firm quarter selected must have no news releases specific
to the associated firm occurring during the quarter immediately pre—
ceding the selected firm quarter except a quarterly earnings announce-
ment. The medium to be used in determining whether firms satisfy this
criterion is, like criterion (2), the Wall Street Journal Index.

(A) To obtain some assurance that all firm quarters selected are
those in which the associated firms are similar with respect to account—
ing policy characteristics each firm quarter selected must be one in
which the associated firm, when appropriate,

(a) Used and disclosed only the results of applying the tax
reduction method of accounting for the investment tax credit;

(b) Used and disclosed only the results of applying the
”normalized” method to account for differences in income taxes for book
purposes and tax reporting purposes;

(c) Used and disclosed only the results of applying the
straight—line method of computing depreciation;

(d) Used and disclosed only the results of applying the
lower of cost or market inventory valuation method; and

(e) Used and disclosed only the results of applying the

first-in, first—out inventory cost flow assumption.

69

The medium to be used in determining whether firm quarters satisfy
this criterion is the annual report of the associated firm for the year
containing the quarter under consideration. In applying this criterion
it is assumed that unless an accounting method change is indicated by
footnote in the annual report the firm has consistently applied and
reported in each quarter of the year only the accounting method dis-
closed in its annual report. If an accounting method change is indi-
cated by footnote in the annual report, then quarterly reports will be
examined to determine which, if any, firm quarters may be selected.

Accounting alternatives in other areas could be restricted also.
However, the five areas above are chosen for restriction because the
choice among the alternatives in each of these areas has a significant
influence on reported earnings of the firms making the choice.9
Alternatives in other areas, e.g. pension plan accounting and research
and development cost accounting, which also have a significant influence
on reported earnings of the firms making the choice, are excluded from
restriction primarily because firms do not typically disclose the

accounting methods used in other areas.1

 

9See for example, James Don.Edwards, Roger H. Hermanson, and
R. F. Solmonson, Accounting: A Programmed Text, Vol. 2 (Homewood,
1970), pp. 102-135, 170—171, 622, 630—632; Harry Simons and Wilbert E.
Karrenbrock, Intermediate Accounting (Cincinnati, 196A), pp. 199-22A,
335-352, 307-312, A21-A23, AA5—A60, 515-516; and Walter B. Meigs,
et al., Intermediate Accounting (New'York, 1968), pp. 107-108, 256-270,
356, 380-383 #39. 633—655.
1OSee for examples of the influence on reported earnings of other
accounting alternatives the entire texts of Edwards, Hermanson, and
Solmonson; Simons and Karrenbrock; and Meigs, et al. That firms do
not typically disclose the accounting methods used in areas other than
the five chosen for restriction is an observation made upon preliminary
review of one hundred forty annual reports of publicly owned corpora-
tions.

7O

(5) For every firm quarter associated with a given industry
selected for the cost group a firm quarter associated with the same
industry must be selected for the equity group and vice versa. Indus-
try classifications will be determined by reference to COMPUSTAT.

(6) Each firm quarter selected must be one during which the asso-
ciated firm was listed exclusively on the American Stock Exchange
(hereinafter referred to as AMEX). In addition to the reason given
earlier for using the AMEX, the AMEX is chosen rather than the New'York
Stock Exchange because May's research indicates that firms listed on
the AMEX typically have fewer news releases than firms listed on the
New York StockExchange.11 The AMEX is chosen rather than the Midwest
Stock Exchange, the Pacific Stock Exchange, or the informal exchange
called the Over the Counter Market for two reasons: (1) the AMEX pro-
vides many more firms and consequently firm quarters from which to se-
lect, and (2) the other exchanges do not have a market index identified
specifically with them to use in the adjustment process for the general
market effect to be discussed later.12 Like industry classification,
exchange listing status will be determined by reference to COMPUSTAT.

(7) For each firm quarter selected the associated firm must have
a current ratio that is within.the 95% confidence interval around an
estimate of the industry mean. The industry mean will be estimated
using the current ratios of all firms in a given industry which are

contained in the COMPUSTAT file, are listed on the AMEN, and have

 

11Robert George May, ”An.Empirical Evaluation of the Significance
of Quarterly Accounting Data in Investor Decision'MakingP (unpublished
Ph.D. dissertation, Michigan State University, 1970).

12Cohen and Zinbarg, pp. 9A~101.

71
long-term unconsolidated investments in common stock of other corpora—
tions. It should be noted that the data to be used in estimating the
industry mean do not satisfy the assumption of random selection under
which confidence intervals are ideally constructed and therefore may
(but do not necessarily) bias the estimate of the mean and the related
confidence interval. However, since the purpose of this criterion is
only to establish a desired degree of confidence in the similarity of
firm quarters of a given industry selected for this research with
respect to the liquidity function, any bias which may exist is of small
consequence. Therefore, the data used in estimating the mean and the
related confidence interval will be treated as if the random selection
assumption were not violated.

(8) For each firm quarter selected the associated firm must have a
profit margin that is within the 93% confidence interval around an
estimate of the industry mean. The sample data to be used in construc-
ting this interval are subject to the same limitation as the data to be
used in constructing the confidence interval around a mean industry
current ratio in criterion (7) above.

The satisfaction of these eight criteria is selecting firm quarters
for this research should result in two groups of firm quarters which are
similar except that one group is associated with the cost method and

the other group is associated with the equity method.

Two Additional Selection Criteria
Two additional criteria will be added to the list already dis-
cussed. The first is added because it facilitates the initial identi-

fication of firms having associated quarters with the potential for

72
satisfying selection criterion (1) above. This first additional
criterion is: Each firm, with associated quarters selected, is one for
which COMPUSTAT reports amounts under the captions ”Investments and
Advances to Subsidiaries” and/or “Investments and Advances-—Other" for
any of the years 1967 through 1969. Included under these COMPUSTAT
captions are long-term investments representing unconsolidated subsidi-
aries, affiliates, associates, and joint ventures.13

The second addition to the criteria list is made to give some
minimum assurance that the results of applying the chosen accounting
method has, in the accountants' terminology, some ”material" effect on
total net earnings of the investor corporation. This second additional
criterion is: For each firm quarter selected, the investees' contri-
bution to total investor earnings as a result of applying the chosen
accounting method must be at least 5% of investor net earnings. The
primary medium from which data will be obtained for determining satis-
faction of this criterion is the associated investor corporation's
annual report. If such data is not available in the annual report, the
secondary medium for such data will be that portion of the corresponding
SEC Form 109K required under Article A of Regulation SAX of the United
1A

States Securities and Exchange Commission. Finally, it will be
assumed that net earnings, other than dividends received from an inves-

tee, accrue uniformly through a year.

 

13Staff of Standard Statistics a division of Standard and Poor's
Corporation, COMPUSTAT'Manual (New York, 1972), sec. 5, p. 2A.

1L’United States Securities and Exchange Commission, Re ation SAX,
6) 7:3.

Form and Content of Financial Statements (Washington, 196 , pp.

73

Time Period Covered by the Research

The time period over which firm quarters will be selected for this
research is January 1, 1967, through December 31, 1969. January 1,
1967, is chosen as the beginning date because the market index to be
used in adjusting for the general market effect, to be discussed later,
was not developed until mid-1966.15 This market index is the AMEX
Index. The ending date is chosen as a result of ”Opinion No. 18" of the
American Institute of Certified Public Accountants' Accounting Prinp
ciples Board.16 That Opinion, issued in March, 1971, requires, in
essence, the use of the equity method for any investment representing
an ownership interest of twenty percent or greater in the investee. All
other investments are to be accounted for using the cost method unless
certain excepting conditions can be met. The effective date of the
Opinion was December 31, 1971, but earlier application was encouraged.
It is possible, although unlikely, that this encouraged early applica-
tion could have affected financial reporting practices for the year
1970. Therefore, in order to avoid the effect which any early applica—
tion might have on this research effort, the year 1969 is the last

year examined.

 

1BCohen and Zinbarg, p. 97.

16American Institute of Certified Public Accountants, Accounting
Principles Board, ”APB Opinion No. 18: The Equity Method of Account-
ing for Investments in Common Stock,” The Journal of Accountangy
(June, 1971), p. 66.

7A

Measurement_Interval and Measurement Forms for
InvestorsT Reactions to Quarterly Earnings
Announcements

As discussed previously the approach of this research effort is to
compare the instantaneous price change and share trading volume after
the release of quarterly earnings announcements associated with two
groups of corporations, a cost method using group and an equity method
using group. However, recall from Chapter III that operationally,
”instantaneous adjustment" is a random variable. The value of the
random variable for any given security is the time difference between
the initial perception of information and complete adjustment for it.
The research of both Beaver and May, particularly May, indicates that
the expected value of this random variable is less than or equal to
17

one trading week, i.e. five trading days. Therefore, this research
effort will employ a time interval of one trading week over which to
measure investors' reactions to quarterly earnings announcements. The
week will begin on the day of announcement and end four trading days
later. The day of the quarterly earnings announcement will be deter—
mined by reference to the Wall Street Journal Index.

Regardless of the time interval over which investors' reactions
to quarterly earnings announcements are measured, these measurements-—
price changes and share trading volume-can be made in either absolute
terms or percentage terms. Price changes will be measured in percent-

age terms in this research for two reasons: (1) When measured in

percentage terms, price changes are essentially transformed to a common

 

17May, pp. 100—126; and William H. Beaver, ”The Information Content
of Annual Earnings Announcements,” Empirical Research in.Accounting:
Selected Stpdies,_1968, a Supplement to Volume 6 of the Journal of
Accounting Research, pp. 67-101.

 

75
base facilitating comparative analysis. It is difficult if not impos-
sible to meaningfully compare a one dollar absolute change in price on
a twenty dollar share and a one dollar absolute change in price on a
fifty dollar share without transforming the changes to a common base.
However, when expressed in percentage terms no further transformation
is necessary for a meaningful comparison. The one dollar change in
both shares hypothesized above would be expressed as five percent and
two percent, respectively, allowing an immediate, direct, and meaninge
ful comparison. (2) An examination of nearly all standard finance
textbooks as well as the current finance literature indicates that the
market psychology, and consequently investor reaction, is in terms of
returns on investment expressed in percentage terms.18 It appears
reasonable that if the market psychology is oriented toward percentage
terms rather than absolute terms and if meaningful comparisons are
facilitated by the expression of price changes in percentage terms,
one should measure price changes in percentage terms.

Share trading volume will also be measured in percentage terms
in this research. Beaver argues for this measurement approach for no
less important a reason than that it facilitates meaningful compari-
sons.19 This is the same as argument (1) above in favor of measuring

price changes in percentage terms.

 

18See for example Henry A. Latane and Donald L. Tuttle, Security

Analysis and Portfolio Management (New'York, 1970).
19

Beaver, p. 73.

76
Data Adjustments

Regardless of whether investors' reactions to the outputs of
accounting methods are measured as absolute price changes, percentage
price changes, absolute share trading volume, or percentage share
trading volume, the raw observations of these reactions must be adjusted
before they can be used to make the inferences suggested previously.
Three types of adjustments are necessary: (1) an adjustment for
capitalization changes; (2) an adjustment for the general market effect;
and (3) an adjustment for nonppublic information effects.

For the obvious reason of comparability, it is necessary to adjust
price change measurements and share trading volume measurements for
capitalization changes such as stock dividends, stock splits, and new
issues. The technique to be used in making this adjustment is the one
suggested by COMPUSTAT:

”To adjust price to a current units basis, divide the indi-

cated price for each year by the corresponding adjustment

factor. To convert shares traded to an equivalent current

basis, the reported shares traded are multiplied by the

corresponding adjustment factor. . . . Adjustments to . . .

[dividends] are obtained directly by dividing the indicated20

leldend per share by the corresponding adjustment factor."
The COMPUSTAT adjustment factors are numbers which, when ". . . applied
to per share data for earlier years, in effect convert such data into
terms of the current share units.”21 For this research the current

share unit to which all per share data will be converted is that exist-

ing for each firm at December 31, 1969.

 

2OStaff of Standard Statistics, p. 25.

21Ibid.

77
The second adjustment of the data is for the general market effect.

The general market effect is, in essence, the influence of a news

release by any firm or agency on prices of all publicly owned firms.22

That this phenomenon exists is supported principally by the research
of King.23 In addition Beaver has argued as follows that the same
phenomenon may affect volume measurements:

”It is possible that the abnormally high volume may be
caused by market-wide pieces of information that are re-
leased at the same time as . . . {data generated by the
accounting alternatives under study]. Since the . . . [data

enerated by the accounting alternatives] are released . . .
fby individual members of each group] throughout the year,
this is not a very plausible explanation of the findings.
Nevertheless, removing the market-wide effects should allay
any fears that this unlikely situation does account for the
results. More importantly, the analysis will serve to
reduce 'noise' in the volume data. Noise is any movement
in volume due to unspecified factors, one of which is market-
wide events that would cause increases in the volume"

It is, therefore, necessary to remove from both the raw price and
volume data that portion of the movement attributable to the general
market effect. The residual is that which can be associated with the
individual securities. (King also found that industry effects are

relatively insignificant, accounting on the average for only eleven

 

22Benjamin'F. King, "Market and Industry Factors in Stock.Price Be-
havior," The Journal of Business, XXXIX (January, 1966), pp. 139-190.

2'3Ibid.

2l‘Beaver, p. 75.

78
percent of the movement of individual security prices while the market
effect accounted for thirty percent on the average.)25
The technique to be used in making the adjustment of individual
firms' price data is that proposed by Sharpe.26 The technique is as
follows:
(1) The regression coefficients Aj and Bj in the following

equation are estimated.

A. + B.M + R.

P.
at J J t Jt

S.t + D't
jt—l

where Pjt

Sjt = the closing price of the jth stock at the end of
week t;
D = the dividend on the jth stock during the week the

Jt stock went ex dividend;

M = ln(

);

 

t
Nt-l
N = the closing value of the AMEX Index at the end of
week t;

and R.t = a random term in the above equation representing the
3 non-market related part of Pjt.

 

25King, pp. 139-190. Although eleven percent may on first examina-
tion seem anything but insignificant, most if not all researchers regard
it as insignificant due to the difficulty of constructing an acceptable
industry index necessary to remove the industry effect. (For example
see Beaver, pp. 67-92; Elba F. Baskins, ”The Communicative Effectiveness
of Consistency Exceptions,” The Accounting Review, XLVII (January,
1972), pp. 38-51; Eugene F. Fama, et al., ”The Adjustment of Stock
Prices to New Information," International Economic Review, X (February,
1969), pp. 1-21; and Nicholas J. Gonedes, ”Some Evidence on Investor
Actions and Accounting'Messages-éPart II,” The Accountipg Review, XLVI
(July, 1971), pp. 535-551.) Furthermore selection criterion (5) above
should reduce any industry effect in this research effort to an even
lesser level of significance.
26William F. Sharpe, ”A Simplified Model for Portfolio Analysis,"
Management Science, IX (January, 1963), pp. 277-293-

79

The data used to estimate Aj and Bj are data from the twelve weeks
immediately preceding the quarterly earnings announcement week and the
ten weeks immediately succeeding the quarterly earnings announcement
weeks except as discussed below. Data from the two weeks immediately
preceding the quarterly earnings announcement are not used in the
regression. Beaver's work indicates that the price behavior of indi-
vidual securities in the two weeks immediately preceding an earnings

27

announcement are affected by the forthcoming announcement. Reasons
for this phenomenon may be that investors anticipate the forthcoming
announcement, that there is a slow leak in the communication channel,

or both.28 Whatever the explanation, an individual firm's security
price does slowly begin to respond to the firm's earnings announcement
about two weeks prior to its issue (although the majority of the
response occurs immediately after the announcement).29 Therefore to use
data from these twO weeks in the regression would bias the Aj and Bj
estimates.

(2) When estimates of A and Bj are found, the procedure is

J

reversed using data from the quarterly earnings announcement week.

R. = P.

-A. -B.M
Jt Jt J

j t
The Rjt now represents a price change measurement in percentage terms
for stock j over the quarterly earnings announcement week adjusted for

capitalization changes and with the estimated market effect eliminated.

 

27Beaver, pp. 67-101.

28Ibid.

29Ibid.

80

The market index to be used in the adjustment of price data for
the market effect is the AMEX Index. This index is used because it is
the only widely based index identified with the American Stock.Exchange
from which firms (or more prOperly firm quarters) are selected for this
research effort.30

The volume data will also be adjusted for a market effect. The
technique to be used in making the adjustment is that developed by
31

Beaver. It is quite similar to the technique which will be used to
adjust the price data.

(1) Beaver's technique ideally requires initially the computation
of the following two statistics for each firm in each week for which

32

volume data is collected.

number of firm i shares traded in the

five day trading_week t

jt number of firm i shares outstanding in
the five day trading week t

number of all AMEX firms' shares traded
V = in the five day ppading week t
Mt number of all AMEX firms' shares out-
standing in the five day trading week t
However, due to the unavailability of data with respect to the number
of firm i shares outstanding in week t and the number of all AMEX firms'

shares outstanding in week t, these statistics must be modified. ‘When

modified, they become:

 

30Cohen and Zinbarg, pp. 9A-101.
3J'Beaver, pp. 72—7A.
321bido, p. 730

81

number of firm i shares traded in the
V' five day trading week t
jt average number of firm i shares out-
standing for the quarter containing
the five day trading week t

number of all AMEX firms' shares traded
V' in the five day trading week t
Mt average number of all AMEX firms‘
shares outstanding for the quarter con-
taining the five day trading week t

vat and VMt represent approximations, respectively, of the percentage

of firm i's shares traded in the five day trading week t and the per—
centage of all AMEX firms' shares traded in the five day trading week t.
As approximations, they are, of course, less than perfect measures. How

imperfect is not determinable. Furthermore, the weighting scheme

33

implicit in computing V' introduces the possibility for bias. Howe

Mt
ever, Beaver defends the statistic, writing,
"While this feature is not entirely satisfying, its use is
defended on the grounds that this index is much easier to
obtain than an index that assigns equal weight to all firms
and because there is no reason to believe the use of this

index leads to either an upward or downward bias in the
findings. 0 o 0.3!";

(2) The regression coefficients a‘j and bj in the following equa-
tion are estimated.

1 = v
vjt a3 + ijMt + ejt

The data used to estimate a and b are data from weeks corresponding

3 J

exactly to weeks from which data is used to estimate A and B above.

3 J

(3) When estimates of aj and bj are found, the procedure is

reversed using data from the quarterly earnings announcement week.

 

33lbid., p. 7A.
3min.

82

= ' _ _ I
ejt vjt 83 ijMt

The ejt now represents a share trading volume measurement in percentage
terms for stock j over the quarterly earnings announcement week adjusted
for capitalization changes and with the estimated market effect elimi-
nated.

The final adjustment of the data that is necessary is an adjustment
for the effect of nonppublic information on the effect of common stock
price behavior. Even a cursory inspection of security price histories
leads one to the conclusion that security prices fluctuate or change
almost continually. Even in the absence of new public information this
phenomenon is apparent. For lack of a better explanation of these price
changes, they are simply called non-public information related price
changes. Since this research effort is specifically concerned with
isolating and comparing price changes that can be attributed to new
pppllp information, in particular the outputs of alternative accounting
methods, the portion of the total price change in the quarterly earnings
announcement week attributable to non-public information must be
removed. The technique to be used in making this adjustment is based on
the assumption that the average effect of non-public information on
security price behavior in the ten week period immediately preceding the
quarterly earnings announcement (excluding the immediately preceding
two weeks) is the same as the effect of nonppublic information on secu-
rity price behavior in the quarterly earnings announcement week. To
the extent that this assumption is fallacious, the technique will result
in an inadequate adjustment. The technique is as follows:

(1) Adjust the data from the ten week period immediately pre-

ceding the quarterly earnings announcement (excluding the immediately

83
preceding two weeks) for capitalization changes and for the market
effect. For practical reasons the adjustment for the market effect will
be made using the same A3 and Bj as used to adjust data from the quar-
terly earnings announcement week even though to some extent this is
technically inappropriate.

(2) Compute the average weekly price change over the ten week
interval in step (1) using the adjusted weekly data from step (1).

(3) Subtract the average weekly price change measurement in (2)
from the price change measurement in the quarterly earnings announcement
week after it is adjusted for capitalization changes and for the market
effect.

The result of this adjustment is a price change measurement in
percentage terms for the quarterly earnings announcement week for each
firm quarter selected for this research effort, adjusted for capitali—
zation changes, the general market effect, and the effect of non-public
information.

An entirely analogous adjustment will be made for the effect of

non-public information on share trading volume measurements for the

quarterly earnings announcement week.

The Comparison

After the price change measurements for the quarterly earnings
announcement weeks are adjusted as described above, the measurements
associated with the cost firms will be compared with the measurements
associated with the equity firms. Similarly, the adjusted share trading
volume measurements for the quarterly earnings announcement weeks asso-

ciated with the cost firms will be compared with the adjusted share

8A
trading volume measurements for quarterly earnings announcement weeks
associated with the equity firms. Finally, the decision rules developed
in Chapter III will be applied in evaluating cost method generated data
and equity method generated data.

However, the nature of these comparisons in an empirical setting
immediately arises. Specifically, to be able to draw convincing infer—
ences a reasonable technique for comparing many measurements in aggre-
gate is needed. Furthermore, the comparison technique needs to provide
some sense of the significance of any difference detected in the
measurements compared. Several statistical comparison techniques are
available which satisfy these needs. All of these statistical tech,
niques presume samples of some minimal size, randomly selected, or if
systematically selected, reasonable assurance that the selection cri-
teria do not introduce bias regarding the qualities about which
inferences are to be drawn. (In the absence of such samples, statis-
tical comparison techniques are invalid and one may, at best, make only
intuitive inferences about the pOpulations represented by the samples.)

Now assuming adequate samples the technique chosen for comparing

both the price change measurements and the share trading volume

35

measurements above is known as the nonparametric MannAWhitney'U Test.

 

3SSidney Siegel, Nonparametric Statistics for the Behavioral
Sciences (New'York, 1956), pp. 116—126.

85
When comparing the price change measurements, the test proceeds as
follows:36

(1) Determine the values of n1 and n2 where

(a) n1 = the number of observations (i.e., firm quarters)
in the smaller of the two groups, and
(b) n2 = the number of observations (i.e., firm quarters)

in the larger of the two groups.

1 + n2 observations. The

reason for using absolute values of the price change observations is as

(2) Take the absolute values of all n = n

follows: The decision rule developed in Chapter III is that the smaller
the price change immediately following a quarterly earnings announcement
the greater the usefulness potential of the output of an accounting
method to investors on average. The price change in percentage terms
can be either positive or negative depending upon whether the price
moves upward or downward from its starting point. However, with respect
to the decision rule, the direction of change is immaterial; only the
magnitude of the price change is important. Therefore, only the abso-
lute values of the changes are considered in the comparison process.

(3) Rank all n = n1 + n2 observations, assigning the rank of 1 to

the observation whose absolute value is the lowest. Ranks range from 1

to n = n1 + n2. Assign tied observations the average of the tied ranks.

 

36The steps enumerated in this section for making the MannéWhitney
U Test are those presented by Sidney Siegel in Nonparametric Statistics
for the Behavioral Sciences (New York, 1956), p. 126. Most of these
steps are extracted verbatum from Siegel's list without the normal
indication that the material was quoted. This is done due to the
nature of the material. Additions or modifications to Siegel's general
steps to make them particular for this research effort should be
obvious from the context, but the reader is encouraged to compare the
discussion in this section with that of Siegel's (p. 126) to determine
exact modifications.

86

(A) Sum the ranks assigned to the group with n1 observations.

Denote this sum by the symbol R1.

(5) Sum the ranks assigned to the group with n observations.

2
Denote this sum by the symbol R2.
(6) Compute the statistic U using either of the following formu-

 

 

lations.
n1(n1 + 1)
(a) U = nln2 + 2 - Rl
n (n + 1)
2 2
(b) U ‘ nln2 + 2 ‘ R2

(7) Determine the significance of the U statistic calculated above
according to the appropriate method below. The appropriate method
depends upon the size of n2.

(a) If n2 is 8 or less, the exact probability associated
with a value as small as the calculated value of‘U is determined by
reference to Table J published in Nonparametric Statistics for the

Behavioral Sciences by Sidney Siegel.

 

(b) If n2

culated value of U may be determined by reference to Table K published

is between 9 and 20, the significance of the cal-

in Nonparametric Statistics for the Behavioral Sciences by Sidney
Siegel.

(c) If n2 is larger than 20, the significance of the calcup
lated value of U may be determined by first transforming the U statistic
to a z statistic and then determining the significance of the z statis—

tic by reference to a normal deviate table such as Table A published in

Nonpnrametric Statistics for the Behavioral Sciences by Sidney Siegel.

 

The transformation of the U statistic to a 2 statistic is specified by

the following formulation:

nn

“VISi—fz-jflz [Ln—EJ- - 2T]

where U = the value of the U statistic calculated as indicated in
(6) above,

 

 

 

= the number of observations in the smaller of the two
groups,

= the number of observations in the larger of the two
groups,

 

where t the number of observations tied for a given rank.

(8) Finally, if the significance of the calculated value of U is
equal to or less than the desired significance, i.e. c( , reject the
null hypothesis H0 in favor of the alternative hypothesis H1. The
desired significance level for this research effort is 5%. HD for this
research effort is: The quarterly earnings announcement week price
changes associated with the cost group are stochastically the same as
the quarterly earnings announcement week price changes associated with
the equity group. Hl for this research effort is: The quarterly
earnings announcement week price changes associated with the cost group
are not stochastically the same as the quarterly earnings announcement
week price changes associated with the equity group. If H0 is rejected
in favor of H1, the direction of the difference will be determined by
comparing the rank sums associated with each group. For example, if
R is the rank sum for the cost group, R2 is the rank sum for the equity

1
group, and R21> R1, the conclusion will be that the quarterly earnings

88
announcement week price changes associated with the equity group are
stochastically larger than the quarterly earnings announcement week
price changes associated with the cost group. Then applying the deci-
sion rule developed in Chapter III, the conclusion will be that the cost
method generated data has greater usefulness potential for investors on
average than does equity method generated data; or equivalently that
the cost method is better than the equity method from the viewpoint of
investors on average based on the American Accounting Association's
usefulness criterion.

An entirely analogous comparison of share trading volume measure-
ments associated with the two groups will be made to determine the
degree of consensus among investors regarding the usefulness potentials
of cost method generated data and equity method generated data, except
that raw values for percentage of shares traded will be compared rather
than their absolute values because of the one directional concept

underlying share trading behavior.

Summgpy
The rules develOped in Chapter III for evaluating alternative

accounting methods based on the American Accounting Association's use-
fulness potential criterion by observing the behavior of common stock
prices and share trading volume were based on some assumed ideal
conditions. These ideal conditions are not in all cases consistent with
real world conditions. Therefore, several measures were discussed
which, it was argued, should reduce or compensate for the difference
between the ideal and real world conditions for this research effort.

Furthermore, three adjustments of the data for uncontrollable factors

89
were discussed-—(1) an adjustment for capitalization changes, (2) an
adjustment for the general market effect, and (3) an adjustment for the
effect of non-public information. Finally, a statistical method of
comparing the common stock price behavior and share trading volume
associated with cost method generated data and equity method generated
data was described with an indication of how the results of that
statistical comparison could be used in conjunction with the decision
rules develOped in Chapter III to evaluate the alternative accounting

methods under consideration.

CHAPTER V

EMPIRICAL FINDINGS, IMPLICATIONS, LIMITATIONS

AND SUGGESTIONS FOR FURTHER RESEARCH

Introduction

In Chapters III and IV a methodology was develOped for the purpose
of determining and appraising the relative usefulness potentials of
alternative accounting methods' outputs. That methodology was applied,
in part as a test of the empirical applicability of the methodology
itself, but obviously more fundamentally as an attempt to add convincing
empirical evidence to the (heretofore largely a priori) cost versus
equity debate. Accordingly, in this chapter the results of applying
the methodology to the cost and equity accounting alternatives, the
implications of the results, and the limitations of the methodology
and its application are presented and discussed. Finally, several

suggestions for further research are presented.

Empirical Findings

As indicated in Chapter IV, firm quarters were selected for this
research effort on the basis of a set of ten criteria. The ten criteria
are reproduced below in the sequence in which they were satisfied in the
sample selection process:

(1) Each firm, with associated quarters selected, was one for

which COMPUSTAT reports amounts under the captions ”Investments and
90

91
Advances to Subsidiaries” and/or ”Investments and Advances-—0ther" for
any of the years 1967 through 1969.

(2) Each firm quarter selected was one during which the associated
firm was listed exclusively on the American Stock Exchange.

(3) For each firm quarter selected the associated firm had a
profit margin that was within the 95% confidence interval around an
estimate of the industry mean.

(A) For each firm quarter selected the associated firm had a
current ratio that was within the 95% confidence interval around an
estimate of the industry mean.

(5) Each firm quarter selected had no news releases specific to
the associated firm occurring during that quarter except a quarterly
earnings announcement.

(6) Each firm quarter selected had no news releases specific to
the associated firm occurring during the quarter immediately preceding
the selected firm quarter except a quarterly earnings announcement.

(7) Only firm quarters in which the associated firms used the
cost method or the equity method to account for long-term investments in
common stock of other corporations were selected, except as modified
by footnote 8 (page 67) in Chapter IV.

(8) To obtain some assurance that all firm quarters selected were
those in which the associated firms were similar with respect to
accounting policy characteristics, each firm quarter selected was one in
which the associated firm, when appropriate,

(a) Used and disclosed only the results of applying the tax

reduction method of accounting for the investment tax credit;

92

(b) Used and disclosed only the results of applying the
”normalized” method to account for differences in income taxes for book
purposes and tax reporting purposes;

(c) Used and disclosed only the results of applying the
straight-line method of computing depreciation;

(d) Used and disclosed only the results of applying the
lower of cost or market inventory valuation method; and

(e) Used and disclosed only the results of applying the
first—in, first-out inventory cost flow assumption.

(9) For every firm quarter associated with a given industry
selected for the cost group a firm quarter associated with the same
industry was selected for the equity group and vice versa.

(10) For each firm quarter selected the investees' contribution
to total investor earnings as a result of applying the chosen accounting
method was at least 5% of investor net earnings.

The results of applying these criteria are presented in Table A
and an indication of the severity of each of the criteria in terms of
sequential sample size reduction rates is presented in Table 5.

The ten criteria in combination reduced the number of firm quarters
from 21600 firm quarters on the COMPUSTAT 1800 Industrials Tape for the
years 1967, 1968, and 1969 to 16 firm quarters on which to apply the
methodology develOped in Chapters III and IVA-8 firm quarters (and A
firms) in each group. The combined reduction rate for the ten criteria
in combination exceeds 99.99%, indicating that these criteria, argued
as important for a real world application of the conceptual framework
developed in Chapter III, are very severe in terms of limiting the

sample sizes. Furthermore, the samples were derived entirely from

93

TABLE A

COMPOSITION OF COST.AND EQUITY
FIRM QUARTER GROUPS

 

 

 

AMEX .
Firm Name Ticker Quarterly Earnings Industry
Announcement Date
Symbol
Cost Group:

Rowland Products Inc. RP Oct. 26, 1967 Chemicals-
Intermediate

Rowland Products Inc. RP Apr. 25, 1968 “

Rowland Products Inc. RP Oct. 23, 1969 "

Rogers Corp. ROG July 17, 1968 Plastic Products-
MiSCo

Shattuck Denn Mining SDE Apr. 7, 1969 Fabricated Metal

Corp. Products
Shattuck'Denn Mining SDE Aug. 15, 1969 "
Corp.

Greer Hydraulics GRH Dec. 3, 1968 Machinery—
Industrial

Greer Hydraulics GRH Apr. 30, 1969 "

Equipy Group:

Fields Plastics & Chem. FLP Oct. 2, 1967 Chemicals-
Intermediate

Fields Plastics & Chem. FLP March 18, 1968 ”

Fields Plastics & Chem. FLP Sept. 16, 1969 ”

Pioneer Plastic Corp. PPK June 13, 1967 Plastic Products-
MiSCo

Plant Industried Inc. PLD Jan. 2, 1968 Fabricated Metal
Products

Plant Industries Inc. PLD May 20, 1968 "

US Filter Corp. UFT Feb. 25, 1969 Machinery-
Industrial

US Filter Corp. UFT May 22, 1969 n

 

9A

TABLE 5

SELECTION CRITERIA SEVERITY IN TERMS OF SEQUENTIAL
SAMPLE SIZE REDUCTION RATES

 

 

Sequential
Selection Criteria POSSlble FlIm Reduction
Qtrs. Remaining Rate
(Rounded)
Initial Number of COMPUSTAT 1967, 1968,
1969 Quarters 21600
1. COMPUSTAT 1967, 1968, 1969 "Invest-
ment & Advances" Quarters 21336 1.A%
2. Exclusive AMEX listing 7AOA 65.3%
3. Profitability Index confidence
interval 2350 68.3%
A. Current Ratio confidence interval 2068 11.9%
5. No news announcements in quarter 17A1 15.7%
6. No news announcement in preceding
year 11.13 18.8%
7. Cost or equity 1350 A.A%
8. Accounting policy characteristics 1261 6.5%
9. Equal industry representation per
group 13A 89.l%
10. Materiality of investee contribu-
tion 16 85.5%

 

95
systematic application of the selection criteria above, with no unre-
stricted random selection at any stage. Finally, noting in retrospect
the number and nature of the selection criteria employed as well as the
fact that related observations (related by virtue of being associated
with the same firm even though separated by time) were included in each
sample group, it is clearly possible that the samples may be statis-
tically biased, i.e. not representative of a larger population of
investor firms (e.g. all COMPUSTAT AMEX investor firms, or more broadly
all investor firms).

Thus it appears inapprOpriate to carry out the statistical compari-
sons outlined in Chapter IV in an effort to develOp convincing, statis-
tically valid, empirical evidence. The data generated may be used,
however, in two ways: (1) To illustrate the statistical comparisons,
.2§.1£ the samples were unbiased; and (2) To draw intuitive inferences,
to the extent such may be suggested by the data.

Therefore, for the selected sample firm quarters the price change
comparison and volume of shares traded comparison in the quarterly
earnings announcement weeks were made as indicated in Chapter IV and
are presented in Tables 6 and 7, respectively, after adjustment for
capitalization changes, adjustment for the general market effect, and
adjustment for the effect of nonpublic information. The results of
these adjustments are tabulated and displayed in Tables 8 and 9 for
price data and volume data, respectively.

Assuming unbiased samples the null and alternative hypotheses for
the price change comparison and the volume comparison would be as

follows:

96

TABLE 6

PRICE CHANGE COMPARISON

 

 

 

 

 

 

 

 

Cost Group Equity Group
Quarterly Earnings Quarterly Earnings
Firm Announcement Week Firm Announcement Week
Ticker Percentage Price Rank Ticker Percentage Price Rank
Symbol Change Symbol Change
(Absolute Value) (Absolute Value)
RP ' 0. 16887 16 FLP 0. 03569 A
RP 0.07851 12 FLP 0.00198 3
RP 0.08377 13 FLP 0.05710 7
ROG 0.0A907 5 PPK 0.06205 9
SDE 0.07A51 11 PLD 0.00027 1
SDE 0.08610 1A PLD 0.00150 2
GRH 0.06A19 10 UFT 0.06003 8
GRH 0.05273 6 UFT 0.10391 15
(Rank Sum)1 87 (Rank Sum)2 A9
(nlxn1 + 1) (8)(9) ‘
U = (n1)(n2) + 2 - (Rank Sum)1 = (8)(8) + 2 - = 13

 

P(U s 13) = 2.5% => Reject null hypothesis at desired confidence
- level = 95%

 

(Rank Sum) 2 < (Rank Sum)1 Q Equity provides more useful data than cost.

 

97

TABLE 7

VOLUME COMPARISON

 

 

 

 

 

 

 

 

Cost Group Equity Group
Firm Quarterly Earnings Firm Quarterly Earnings
Ticker Announcement Week Rank Ticker Announcement Week Rank
Symbol Percentage Volume Symbol Percentage VOIume
RP -0.0063 1 FLP 0.00A8 13
RP 0.0001 7 FLP 0.0065 15
RP 0.0018 10 FLP 0.0032 12
ROG —0.00A2 A PPK -0.00AA 3
SDE -0.0060 2 PLD -0.0028 5
SDE 0.0011 9 PLD -0.0006 6
GRH 0.0025 11 UFT 0.0052 1A
GRH 0.0002 8 UFT 0.0223 16
(Rank Sum)1 52 (Rank Sum)2 8A
n2(n2 + 1) (8)(9)
U = (n1)(n2) + 2 — (Rank Sum)2 = (8)(8) + 2 — 8A = 16

 

P(U S 16) = 5.2% 2} Do not reject null hypothesis at desired confidence
level = 95%

 

P(U s 16) = 5.2% :5 Reject null hypothesis only at 9A.8% confidence
level or lower; if null hypothesis is rejected at
9A.8% confidence or lower then since
(Rank Sum) 1 < (Rank Sum) 2 consensus among investors
regarding usefulness of cost data is greater than
consensus regarding usefulness of equity data.

 

98

 

 

m0a.0u 000.0 400.0: 000.0: amm.0I HH0.0I 000.0: 00.H mo0.0n 00nd .mm as: as:
000.0 000.0: 0m0.0 sH0.0I saH.H s00.0I «no.0: 00.H «no.0: mesa .mm .boa ens
H00.0 0m0.0I 0m0.0n mm0.0 mmm.a am0.0 000.0 00.H 0m0.0 000a .00 an: 0am
000.0 mm0.0u m~0.0u m00.0u mnr.a «H0.0I am0.0u 00.a 000.0: mood .m .soe 0am
~00.0I 000.0 “no.0: 0N0.0 0sN.H s00.0u sm0.0n 00.m sm0.0u soda .ma once emu
sm0.0n HH0.0 mh0.0u 000.0 mom.~ aa0.0u 040.0: 00.H 0e0.0u 000a .0H .eaom new
H00.0 aa0.0u sa0.0I 000.0: mHH.0 0H0.0 000.0 00.H 000.0 moaa .mfl none: can
«no.0: m00.0 000.01 mH0.0 008.0 HH0.0I hm0.0n 00.H em0.0n soda .m .000 can
mm0.0I «00.0: rm0.0n mm0.0 0mm.0 0H0.0I 040.0: 00.H hr0.0n 000a .0m .eaa mew
400.0 ma0.0 0s0.0 mH0.0 00m.H 000.0: H0H.0 00.H H0H.0 whoa .m .oon 000
000.0: 000.0: ms0.0n mm0.0 sH0.0 m00.0u s00.0I 00.H s00.0n soda .mH .msa mam
1:10.01 80.0 20.0.. 80.0 400.0 000.0: 20.0.. 004 20.0.. 003 .s 5% mom
oao.0 000.0 u40.0 ~00.0I «mm.H m00.0u mm0.0I 00.H mN0.0I whoa .na sass 00m
000.0: s00.0n Hao.0u H00.0I mmm.0 000.0: H00.0I m0.a Ha0.0u soda .mw .eoo am
ms0.0u 000.0 Hs0.0u 000.0 0H0.a e00.0u AN0.0I m0.a sm0.0n mesa .mm .haa am
moa.0u 000.0 00H.0I 400.0: 4H0.0 000.0 0mH.0I m0.H 0mH.0I soda .0m .eoo mm
mm: omcmno
00.2“”an wmmmwm “0””an omcmso IHWWM we IHWWMMQ ”Mama“ Ipwflmfld owmmmzm ANWWWH Honeunm
.05 «m0 . .0200. E3. R . . . . . coapmn . . Roscocdoqfifi noxofl.
.0noa homcwwoa .exz .xz.<ma m.pmmw <.emmw cwmmms sheep .Mmoumw omseseom sham
Hanna .x: ago Iaaoaaaoo nacho menopause

 

2...ng 52d HOHmm .mO 2.0%

w 393.

99

 

 

 

000.0 H00.0 000.0 0H0.0 000.0 400.0 H00.0 00.H H00.0 0000 .00 an: 000
000.0 H00.0 000.0 0H0.0 000.0 000.0 4H0.0 00.H 400.0 0000 .00 .boa 000
000.0 0H0.0I 000.0: 000.0 Hmd.0 400.0: 0H0.0 00.H 0H0.0 0000 .00 00: qua
000.0: 000.0 000.0: 400.0 004.0 000.0 000.0 00.H 000.0 000H .0 .000 Qua
400.0: 000.0: 000.0: 0H0.0 004.H 000.0: 000.0 00.0 000.0 soda .00 oss0 mam
000.0 000.0 000.0 000.0 404.0 000.0 000.0 00.H 000.0 000H .0H .0aom age
000.0 400.0: 000.0 000.0 040.H 000.0: 000.0 00.H 000.0 000a .ma been: new
400.0 400.0 000.0 4H0.0 H00.H 000.0 000.0 00.H 000.0 000d .0 .eoo 000
000.0 000.0 000.0 0H0.0 000.0 000.0 000.0 00.H 000.0 000a .00 .ea< 000
000.0 000.0 000.0 0H0.0 040.0 000.0 000.0 00.H 000.0 000H .0 .oon :00
000.0 000.0 H00.0 000.0 004.0 000.0 000.0 00.0 000.0 0000 .0H .000 mam
000.0: 000.0 000.0 Ha0.0 H0H.H 000.0 0H0.0 00.H 0H0.0 000d .0 .004 mam
400.0: 000.0 000.01 4H0.0 000.0 000.0: 400.0 00.0 400.0 0000 .sfl 00:0 000
H00.0 H00.0I 000.0 000.0 400.H 000.0 0H0.0 00.0 000.0 000d .00 .eo0 am
000.0 000.0: 000.0: 0H0.0 000.0 000.0: 000.0 00.H 000.0 0000 .00 .ea< 00
000.0: 000.0 000.0: 400.0 040.H 000.0 000.0 00.0 000.0 0000 .00 .eoo am
08502?
005.02? 00303 oboe 60.09
.0 m Gm 9:5 0

a saw? a messes IaWMowo Iaweowo . 00.00 1.00.04 as s :08 Hansen
.xr.400 . .0000 .enga . . . . a: coarse . esosoosaosc< across
. 0 Icoz . . m 000 0 000 0000 I a: 000 ms

0.00 000 .000 pa: 02.000 .00: .002 coaeos ease .000000 n apnea saga
Assam .x: 000 Iaaneaaoo 10000 saeoeeosa

 

Bgmg #33 28> .mO 80%

megs

100

(1) Price change comparison null hypothesis, PHO: The quarterly
earnings announcement week price changes associated with the cost group
are stochastically the same as the quarterly earnings announcement week
price changes associated with the equity group.

(2) Price change comparison alternative hypothesis, PH1: The
quarterly earnings announcement week price changes associated with the
cost group are not stochastically the same as the quarterly earnings
announcement week price changes associated with the equity group.

(3) Volume comparison null hypothesis, vHO: The quarterly
earnings announcement week percentage volume of shares traded associated
with the cost group are stochastically the same as the quarterly earn-
ings announcement week percentage volume of shares traded associated
with the equity group.

(A) Volume comparison alternative hypothesis, vH1: The quarterly
earnings announcement week percentage volume of shares traded associated
with the cost group are not stochastically the same as the quarterly
earnings announcement week percentage volume of shares traded associ-
ated with the equity group.

As indicated in Tables 6 and 7, respectively, PHO would be rejected
at the desired 95% confidence level but vHO would not be rejected at
that confidence level. And, in fact, PHO could be rejected at a confi-
dence level of 97.5% but VH0 could only be rejected at a confidence
level of 9A.8%. Furthermore, since the rank sum associated with the
equity group price changes is less than the rank sum associated with the
cost group price changes the results of the statistical comparison would
indicate that the quarterly earnings announcement week price changes

associated with the equity group are stochastically smaller than the

101
quarterly earnings announcement week price changes associated with the
cost group. Finally, even though the results at the desired confidence
level of 95% would be that the percentage volume of shares traded in
the quarterly earnings announcement week are stochastically the same for
the cost and equity groups, it would be obviously naive to ignore the
evidence to the contrary. For at the 9A.8% confidence level, only .2%
lower than that desired, VHO could be rejected. ~If vHO were rejected,
then, the quarterly earnings announcement week percentage volume of
shares traded associated with the cost group would appear to be stochas-
tically smaller than that associated with the equity group since the

cost group rank sum is smaller than the equity group rank sum.

Implications

When the empirical findings, reported in the immediately preceding
section, are interpreted in conjunction with the decision rules
develOped in Chapter III, the conclusions below emerge. If statistical
inferences were warranted, the evidence impounded in the price change
comparison would indicate at the 95% confidence level that for the firm
quarters included in the research the equity method provides more
useful data for investors on average than does the cost method. This
general inference would seem also at least intuitively plausible.
Similarly, the evidence impounded in the percentage volume of shares
traded comparison would appear to indicate that for the firm quarters
included in this research the degree of consensus among investors is
greater with respect to the usefulness of cost data than with respect

to the usefulness of equity data.

102

The most obvious implication of these results relates directly to
the Accounting Principles Board "Opinion 18."1 That opinion requires,
in essence, the use of the equity method for any investment representing
an ownership interest of twenty percent or greater in the investee. All
other investments are to be accounted for using the cost method unless
certain excepting conditions can be met. Since this research effort
was conducted in the context of effective control, operationally defined
as a twenty percent ownership interest or greater, it would appear,
based on the usefulness potential criterion, that the Accounting
Principles Board did make the appropriate choice between the cost and
equity methods in the effective control context. However, this impli-
cation as well as the others discussed below are, indeed, subject to
those limitations of this research effort already discussed and those
discussed in the next section. Furthermore, since this research effort
was conducted only in the context of effective control and in that
context examined only the relative usefulness potentials of cost and
equity data, no implications may be drawn from the results reported
above with respect to the apprOpriate choice among accounting alterna-
tives in a non-effective control context or with respect to the choice
among the equity method and the other thirty feasible alternatives in
any control context.

Another implication of the research results reported above is based

upon the relationship between the results of using the equity method and

 

1American Institute of Certified Public Accountants, Accounting
Principles Board, ”APB Opinion No. 18: The Equity Method of Accounting
for Investments in Common Stock," The Journal of Accountancy (June,

1971), pp. 63-680

‘.. ...- . '
«’_-.<_._

 

103
the results of consolidation. It was noted in Chapter I that one of the
arguments of equity advocates in support of their position was that use
of the equity method produced results similar to that of consolidation
and that consolidated statements are assumed to be more meaningful than
separate statements.2 Although the evidence is, indeed, indirect and
less than perfectly conclusive, it would appear that it does support
intuitively the assumption that, in the context of investment activity,
consolidated statements are more meaningful than separate statements,
particularly if the cost method is used in the separate statements of
the investor.

Still another implication of the research results reported above
is based upon the relationship among the cost and equity methods and the
principle of conservatism. In Chapter IV it was noted that the cost
method is generally considered more conservative than the equity method.
If, indeed, the cost method is more conservative than the equity method,
the research results reported above would appear to indicate that in
the context of investment activity the principle of conservatism is not
necessarily a sound foundation upon which to base a choice among
accounting alternatives, particularly if the ultimate evaluation of the
choice is to be based upon the usefulness potential criterion.

Finally, under the assumption that investors are rational, the

evidence reported above appears to indicate that, on average, investors

 

2Walter B. Meigs, Charles E. Johnson, and Thomas K. Keller,
Advanced Accounting (New York, 1966), pp. 297-298; American Institute
of Certified Public Accountants, Accounting Principles Board, pp. 63—68;
American Institute of Certified Public Accountants, ”Accounting Research
Bulletin No. 51—-Consolidated Financial Statements,” APB Accounting
Princi les: Original Pronouncements as of February 1, 1971 (Chicago,
19715, II, 6091.

10A
3

do prefer equity data to cost data, as Barrett concluded. However,
the congruence of Barrett's conclusion and the implication of the re-
search results reported above under the stated assumption does not
minimize the importance of the criticism of Barrett's research on a
conceptual level presented previously. Indeed, the implication just

stated is based upon the rationality assumption and that assumption may

or may not be sound.

Limitations

As with any research effort, the research effort reported in this
paper is subject to limitations. One obvious limitation of this re—
search is the small sample sizes. However small sample sizes are gen-.
erally regarded as limitations in the sense that they make rejection of
a null hypothesis difficult. It should be obvious that had the samples
been randomly selected the small sample size in this research effort
would not have had such an adverse effect. Indeed, the null hypothesis
related to the price test would have been rejected at 97.5% confidence
and the null hypothesis related to the volume test would have been
rejected at 9A.8% confidence.

A second obvious limitation of this research relates to the compo-
sition of the two samples. The observations included in each sample
were treated as independent observations. The fact that each observa—
tion was selected entirely in accordance with a stipulated set of

criteria as well as the fact that related observations (related by

 

3M. Edgar Barrett, ”APB Opinion Number 18: A Move Toward Pref—
erences of Users,” Financial Analysts Journal (JulyaAugust, 1972),

pp. A7-5A.

105
virtue of being associated with the same firm even though separated by
time) were included in each sample group makes adherence to the statis-
tical independence assumption suspect. Consequently, the implications
of the research findings may be viewed appropriately only as indications
and not as statistically valid inferences.
Another limitation of the research relates to the composition of

the two samples in terms of the satisfaction of criterion 7 above. It
was noted in Chapter IV that firms included in the cost group were those ~ 4

which carried at least one investment at cost, others possibly being

 

consolidated. It was also noted that the equity group included firms

that either fully consolidated all investments or carried one or more

 

at equity and consolidated the remainder. The reasons that criterion 7
was satisfied in this fashion were that most investor corporations
issue only consolidated statements rather than separate statements and
that the consolidation process and the equity method produce similar
Operating results which were presumed to be the data which influence
individual investor decisions, at least the individual investor deci-
sions on which this research effort was based. Now, to the extent that
investors do not perceive the results of consolidation and the equity
method as the same in terms of measuring Operations, the samples are
inappropriately constructed and any inferences, even intuitive ones,
are subject to even greater suspicion.

Still another important limitation of this research effort results
from a failure to test the regression lines in the general market effect
adjustment process for goodness of fit. If the regression lines are not
a good fit then the residuals for the quarterly earnings announcement

week, the primary data on which this research is based, likely contain

106
considerable noise. However, the goodness of fit test was omitted
(1) because other research efforts have indicated that the general
market effect adjustment technique used in this study is an adequate one
in most if not all cases,h (2) because the apprOpriate action in the
case of poor fit is elimination of the associated firm resulting in a
further reduction of sample sizes which would increase the severity of
the first limitation discussed above, an action which appears to be of
at least equal adversity as not making the goodness of fit test and
therefore retaining the firm or firms in the samples, and (3) because
the underlying assumptions of the goodness of fit test could not be met.

Other less than obvious limitations include the following:

(1) The raw price and volume data were subjected to three adjust-
ments (one of which was discussed above), which likely were less than
perfect; therefore, undoubtedly the data contains some noise;

(2) The relative size of firms was not taken into consideration
in selecting firms whose associated firm quarters were included in the
research;

(3) The attempt to adjust via a selection criterion for the

5

industry effect identified by King may have been less effective than
an alternative adjustment technique, namely a modification of the
market adjustment technique discussed previously;

(A) No consideration was given to the effect of general economic

cycles (i.e. up and down markets) and even a casual observer of the

 

AEugene F. Fama, et al., “The.Adjustment of Stock Prices to New
Information," International Economic Review, X (February, 1969), pp. 1-

21.
5Benjamin F. King, ”Market and Industry Factors in Stock Price

Behavior,“ The Journal of Business, XXXIX (January, 1966), pp. 139-190.

107
market must be aware that the economic trends in 1967 were different
than those in 1969;
(5) The current ratio confidence intervals and the profitability
index confidence intervals were wider than would be desired;
(6) The AMEX Price Index used in the general market effect adjust-
ment process is a simple average index which Latane and Tuttle suggest

is inferior to a geometric average index;

:0 I18"!

(7) The effects of short and long security transactions were not
considered; and
(8) As indicated previously, no attempt was made, due to practical

reasons, to test the apprOpriateness of the twenty percent rule con-

7

 

tained in the Accounting Principles Board ”Opinion No. 18.”
Finally, all of the measures discussed in detail in Chapter IV to
compensate for real world versus ideal world conditions represent limi-

tations in applying the conceptual framework develOped in Chapter III.

Suggestions for Further Research

The most obvious suggestion for further research is simply to apply
the methodology develOped in Chapters III and IV in comparing other
accounting alternatives, including the other thirty methods of account—
ing for long—term investments in common stock. In doing so, however,
it is suggested that when possible a longer time period be examined,

e.g. ten years rather than three years, thereby hOpefully increasing

 

6Henry'A. Latane and Donald L. Tuttle, Security Analysis and Port-
folio Management (New York, 1970), p. 167 and p. 188.

7American Institute of Certified Public Accountants, Accounting
Principles Board, pp. 63—68.

“I‘ : .
M's-L :-

108
the sample sizes. Additionally, the methodology herein develOped and
employed may be further refined particularly from an application view;
point. One such refinement (elimination of any unnecessary selection
criteria) implies a research project in itself, namely testing the
sensitivity of the results to the relaxation of any or all of the
selection criteria. Another possible refinement is to expand the

methodology to allow time series analyses.

.1, arm

The second most obvious suggestion for further research is to test
the appropriateness of the twenty percent rule referred to above.
It might also be suggested that the methodology implied in foot—

note 2A (page A1) in Chapter III be employed in a comparison of account-

 

ing alternatives.

All of the assumptions, both explicit and implicit, in this
research project, need to be tested empirically. These implied research
projects are not discussed here in detail. Those interested are
referred to Chapters III and IV in which most such assumptions are
stated explicitly.

The contention in Chapter II that control of competing public news
releases will amplify the importance of earnings announcement data is
subject to empirical verification.

Finally, probably the most important research which can be sug-
gested at the conclusion of a market based research effort such as this
is that research necessary to identify the real world expectational
(valuation) model or models used by investors. The lack of such
research makes all other market based research efforts, this one
included, less than conclusive even in the absence of any other limi—

tations.

109
Conclusion
In Chapter V, the empirical findings of this study were presented.
The implications of these findings were discussed and a considerable
number of limitations were indicated. Finally, several suggestions
for further research were made with an emphasis on the research

necessary to identify real world expectational models of investors.

I In. 53;!

 

BIBLIOGRAPHY

 

BIBLIOGRAPHY

American Accounting Association. A Statement of Basic Accounting
Theogy. Evanston: American Accounting Association, 1966.

American Institute of Certified Public Accountants. "APB Opinion E A
No. 18: The Equity Method of Accounting for Investments in Common ‘
Stock.” The Journal of Accountancy (June, 1971), 63-8.

. ”Accounting Research Bulletin No. 51-Consolidated Financial
Statements." APB Accounting Pplnciples: Origlnal Pronouncements
as of February 1, 1971, Vol. II. Chicago: Commerce Clearing
House, Inc., 1971.

 

. Accountinngrends and Techniques. New York: American Insti-
tute of Certified Public Accountants, 1950-1970.

. Opinion 28 Interim Financial Reporting. New York: American
Institute of Certified Public Accountants, 1973.

Ball, Ray, and Philip Brown. “An Empirical Evaluation of Accounting
Income Numbers.” Journal of AccountinglResearch, VI (Autumn,

1968), 159-77.

Barefield, Russell M., and Eugene E. Comiskey. "The Smoothing
Hypothesis: An Alternative Test.” The Accounting Review, XLVII
(April, 1972), 291.

Barrett, M. Edgar. ”APB Opinion Number 18: A Move Toward Preference
of Users.” Financial Analysts Journal (JulyeAugust, 1972), A7—5A.

Baskin, Elba F. ”The Communicative Effectiveness of Consistency Excep-
tions." The Accounting Review, XLVII (January, 1972), 38—51.

Beaver, William H. ”The Information Content of Annual Earnings
Announcements.“ Empirical Research in.Accounting: Selected
Studies,l968, a Supplement to Volume 6 of the Journal of Account-
ing Research, 67—101.

, and Roland E. Dukes. ”Interperiod Tax Allocation, Earnings
Expectations, and the Behavior of Security Prices." The Accounting
Review, XLVII (April, 1972), 320032.

Bem, Daryl J. Beliefs, Attitudes, and Human Affairs. Belmont:
Brooks/Cole Publishing Company, 1970.

110

111

Benston, George J. "Published Corporate Accounting Data and Stock
Prices." Empirical Research in Accounting: Selected Studies,
1967, a Supplement to Volume 5 of the Journal of Accounting
Research, 1—5A.

Cohen, Jerome B., and Edward D. Zinbarg. Investment Analysis and Port-
folio Management. Homewood: Richard.D. Irwin, Inc., 1967.

Dopuch, Nicholas, and David F. Drake. “The Effect of Alternative
Accounting Rules for Nonsubsidiary Investments." Empirical
Research in Accounting: Selected Studies, 1966, a Supplement to
Volume A of the Journal of Accounting Research, 192—227.

Driscoll, P. T. ”Consolidated Financial Statements and the Equity
‘Method of Accounting.” The New York Certified Public Accountant

(August, 1971) , 567-71.

Edwards, James Don, Roger H. Hermanson, and R. F. Solmonson. Account-
lng: A Prquammed Text, Vol. 2. Homewood: Richard D. Irwin,
Inc. , 19700

 

Fama, Eugene F. ”Efficient Capital Markets: A Review of Theory and
Empirical Work." The Journal of Finance, XXV (May, 1970), 383—A17.

, Lawrence Fisher, Michael C. Jenson, and Richard Roll. "The
Adjustment of Stock Prices to New Information.” International
Economic Review, X (February, 1969), 1—21.

Francis, Jack Clark. Investments: Analysis and Management. St. Louis:
McGrawaflill Book Company, 1972.

Gonedes, Nicholas J. ”Some Evidence on Investor Actions and Accounting
Messages-4Part I.” The Accounting Review, XLVI (April, 1971),
320—28.

. ”Some Evidence on Investor Actions and Accounting Messages-—
Part II.” The Accounting Review, XLVI (July, 1971). 535-51.

Gordon, Myron J. "Postulates, Principles, and Research in Accounting."
The Accounting Review, XXXIX (April, 196A), 251—63.

Hagerman, Robert L., Thomas F. Keller, and Russel J. Petersen.
”Accounting Research and Accounting Principles.” The Journal of
Accountancy (March, 1973), 51-5.

Karrenbrock, Wilbert E., and Harry Simons. Advanced Accounting, 3rd ed.
Cincinnati: SouthéWestern Publishing Company, 1961.

King, Benjamin F. ”Market and Industry Factors in Stock Price
Behavior.” The Journal of Business, XXXIX (January, 1966), 139—90.

Latane, Henry A., and Donald L. Tuttle. Security Analysis and Port-
folio Management. New York: The Ronald Press Company, 1970.

112

Mao, James C. T. Quantitative Analysis of Financial Decisions. London:
The MacMillan Company, 1969.

Markowitz, Harry M. Portfolio Selection, Efficient Diversification of
Investments. New York: John Wiley and Sons, Inc., 1959.

 

'May, Robert George. ”An Empirical Evaluation of the Significance of
Quarterly Accounting Data in Investor Decision.Making.” (unpub.
Ph.D. dissertation, Michigan State University, 1970).

Meigs, Walter B., Charles E. Johnson, and Thomas F. Keller. Advanced
Accounting. New York: McGrawhHill Book Company, 1966.

 

________J ________J , and A. N. Mosich. Intermediate Account-
ing. New York: MCGrawhHill Book Company, 1968.

Moonitz, Maurice. The Entity Theopy of Consolidated Statements.
Madison: Foundation Press, 19AA.

O'Connor, Melvin C., and James C. Hamre. "Alternative Methods of
Accounting for Long-Term Nonsubsidiary Intercorporate Investments
in Common Stock.” The Accounting Review, XLVII (April, 1972),
308-19.

Rustin, Richard E. ”AMEX Board Divided on Whether to Follow Big Board
for General Broker—Fee Rise.” Wall Street Journal. (May 23,

1973), A.

Sharpe, William F. ”A Simplified Model for Portfolio Analysis."
Management Science, IX (January, 1963), 277-93.

Siegel, Sidney. Non Parametpic Statistics for the Behavioral Sciences.
New York: McGraweHill Book Company, 1956:

Simons, Harry, and Wilbert E. Karrenbrock. Advanced Accounting, Ath ed.
Cincinnati: South4Western.Publishing Company, 1968.

, and . Intermediate Accounting. Cincinnati: Southp
Western Publishing Company, 196A.

Sorter, George H., and Selwyn W. Becker. ”Corporate Personality as
Reflected in Accounting Decisions: Some Preliminary Findings.”
Accounting and Its Behavioral Implications. Eds. William J.
Bruns, Jr. and Don T. DeCoster. New York: McGrawhHill Book

Company, 1969, 267-77-

Staff of Standard Statistics a division of Standard and Poor's Corpora—
tion. COMPUSTAT Manual. New York: Standard Statistics 00., 1972.

United States Securities and Exchange Commission. Regulation 50X, Form
and Content of Financial Statements. Washington: U.S. Government
Printing Office, 1966:

113

VanHorne, James C. Financial Manngement and Policy. Englewood Cliffs:

Prentice-Hall, Tnc. , 1968?

Weston, Fred J., and Eugene F. Brigham.

Essentials of Manngerial

 

Finance. Dallas: Holt, Rinehart and'Winston, Inc., 1968.

 

14'! I..." a... '71. ,1.
.

$.‘lxtlvv r.

.Iun. .ld

,.$\I.

ill.‘.$.l.‘.c . u

'I‘h I ‘.

I1”. .1

1.2. -,\I

O

. It’lt.

. I'D. If

.. 4.1.