w—wa

.- “v...— - .
_ ~ ‘ - - A . -- . .—
-_ . _ , ._ . ‘
.A — - . . -
.-u - , . _ , V .
:2 U v ._ . - ..

“inn“ ..

.uu w—

 

  
 
   
   
      
    

.—

-
‘ P“ ‘
.._.-

,
:2:::::"”
_' _ 1,4
_,-._.~_.,fl _

 
 

t:;::==:=:. -‘
-.,-
WV, .
k, 5 d _ ==, _’
w W- —.—.-
EL.:
::r

::==f"

4...... '
'.
-

 
 

filth

- .z:::;=-
. .u
,
-- -

AA_.—‘...—-r-

R

W

   
 
 
 
 
 

- ‘ ‘l" — -' ‘ I-— - — :
O
-. o ’ - ' ' - ‘
z ‘ . ~ .-__ —_ x
I a — I _ ' n
' v " - - . ‘ _
HA7?» - - ,
‘ J8 - -
_ 4 .' -
- . ' __ _ --

‘

._._...
‘”
M.
“—3
M

,
.-

a...

.. “.500-

J
“_:7

-
4

4

.— ' ,

'MM

.vv-o M

w-
u

'31 “

-.

'-
-0

-
4
. - -,
.
3- u -_.
W-.-._a‘u
. ‘
.M‘Jap ,
.-.
.—r
fi‘f '.
.
"v «1'
- V; N
H»; fi‘fk"
k 1‘
, -
' I

~—-

 
   
 
 

 

 

 

   
   
   
 

'.

E:- ., l
fir. -‘ v“ ~
__ I}x’-vn

,
Ja
‘U

”1‘51:
'- ‘
v.

  

a“ . a
.4- ‘ '-' - - J ' 4 o
“" "-.. 4.
.- v ~ ' y - '
73:}. -u“ ;‘ . ‘,
t'.‘ do. 9. '
Fr... '. , .. o
_ - :r- "
0 V" .

  
  

  

 

 

.. A "' .4
.L..~.-.sov""' ‘4
.40!
n

“g ' Kw? i915 {a‘ih‘ 1
{:1 ‘16"‘1 i 1““
I" it ”‘14:." {-43. ‘H‘fir

-- 2,1.-
-m.- {£214-

- . Mg?”
7 "' M" u-
‘ as“ n . '3
.- , ' .. 0
_ v.
‘ " o ‘
.._ 4‘
. V .W

,4 _..
4 -
:Ei'vdf

¢=:
n
{r

at

m
I‘
' I
n
-F
1‘
.
1w
,-
. 4
W‘ 4.. ~ .
.
V

.-

04-“

In”

fi

H m
J

u... ‘
“5'.
u.-

_r :. 'Y.
, a
-I
"' -- 4 ->
‘ .. ..
:4

‘;vum::§l“
-
__ a-.. <0
. .>,. . ‘
. .
, .

. ._"o~
~*3::SE”

:qpo
:
.
‘3 ‘

w
on

 

LIBRARY
Michigan State

University J

 

 

This is to certify that the

dissertation entitled

EFFECTS OF ACCOUNTING TECHNIQUES ON THE

 

STUDY OF MARKET POWER

presented by

Kiran Verma

has been accepted towards fulfillment
of the requirements for

Ph.D. . Accounting
degree In

 

 

Major professor

 

Date 8/17/87

 

MS U is an Affirmative Action/[q ual Opportunity Institution 0-12771

 

 

 

 

MSU

LIBRARIES
m

 

 

RETURNING MATERIALS:
Place in book drop to
remove this checkout from
your record. FINES will
be charged if book is
returned after the date
stamped below.

 

 

 

 

 

EFFECTS OF ACCOUNTING TECHNIQUES ON THE STUDY
OF MARKET POWER

by

Kiran Verma

A DISSERTATION

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

DOCTOR OF PHILOSPHY

Department of Accounting

1987

ABSTRACT

EFFECTS OF ACCOUNTING
TECHNIQUES ON THE
STUDY OF MARKET POWER

by

Kiran Verma

The purpose of this study is to examine the effects
of alternative accounting techniques on inferences drawn
from the relationship between firm profitability and its
market power. Accounting data have been widely used as
sample evidence in such analyses. As accounting rules
allov firms considerable leeway in their chioce of
accounting techniques. it has been alleged that
inferences drawn fram such analyses are potentially
misleading. This is particularly true for time series
analyses at the firm level where interfirm differences due
to accounting techniques are not aggregated out.

This study is conducted in two parts. The first part
deals with the development of a model linking a firm's

accounting profit to its market share. capital stock. and

raw materials used in production. The second part
examines the robustness of this model to estimation using
accounting data that are generated by alternative
accounting techniques (e.g. LIFO vs. FIFO. Straight line
vs. Double Declining Balance. and Historical cost vs.
Replacement cost).

Unlike model developed in previous research. the
model in this study specifically incorporates the presence
of inventories. This is done in order to describe more
closely the economic phenomenon of measuring the
accounting cost of goods sold and hence the accounting
profit.

Robustness of the model developed above is tested
with respect to two parameters. (1) its explanatory power.
and (2) inferences drawn from it about the firm's market
power. The results show that both of these parameters are
sensitive to the choice of certain accounting techniques.
This effect is most pronounced for the change in inventory
techniques and almost nonexistant for the change in
depreciation techniques. The results are mixed for the

change from Historical cost to Replacement cost.

To Sumi. Charu. and Priya

iv

ACKNOHLEDGEHENTS

Special thanks to Ed Outslay. Ron Marshall. and
Stephen Martin for their numerous suggestions and
support. Also thanks to Ernst and Hhinney for financial

support.

Page
LIST OF TABLES V
I. INTRODUCTION I
II. BACKGROUND 3
2.1 Effects of Alternative Accounting Techniques 3
2.2 Purpose of the Dissertation 6
2.3 Profitabilitxntnd-Market Structure 7
III. THE MODEL 11
3.1 A Simple Model of a Profit Maximizing Firm II
3.2 Factors of Production and Accounting Values I3
3.3 Effects of Accounting Techniques on the
Coefficients of Interest 28
IV. NYPOTEESES 23
V. METHODOLOGY 27
5.1 Retatement procedures 27
5.2 Inventory Restatement 28
5.3 Restatement for Depreciation 38
5.4 Restatement for Replacement Cost 31
5.5 Data 28
5.6 Methods of Analysis 29
5.6a Analysis of Variance 35
5.6b Kendall's Coefficient of Concordance 39
5.6c Spearman's Rank Correlation Coefficient 41
VI EMPIRICAL RESULTS AND ANALYSIS 43
6.1 General Results 43
6.2 Parametric t test 43
6.3 Analysis of Variance 45
6.4 Adjusted R2 46
6.4b 46
6.5b Spearman's Rank Correlation Coefficient 46
6.5 Market power test criterion 47
6.5a Kendall's Coefficient of Concordance 48

TABLE OF CONTENTS

6.5b Spearman's Rank Correlation Coefficient 48

vi

VII
VIII
IX

CONCLUS I ONS
APPENDI X
FOOTNOTES
BIBLI OGRAPNY

vii

Page
97
III
189
113

LIST OF TABLES

Table

1.
2.

10.

11.

12.

13.

14.

15.

16.

17.

Combinations of Accounting Techniques.

Pairwise comparison of alternative accounting
techniques.

Firms. Industry SIC codes. and reported accounting
methods.

Results of the time series OLS regression with 36
observations for the Armco Inc.

Results of the time series OLS regression with 36
obsevations for the USX Corporation.

Results of the time series OLS regression with 36
obsevations for the Lukens Corporation.

Results of the time series OLS regression with 36
obsevations for the LTV Corporation.

Results of the time series OLS regression with 36
observations for the General Motor Corporation.

Results of the time series OLS regression with 36
observations for the Ford Motor Company.

Results of the time series OLS regression with 36
observations for the American Motors Corporation.

Results of the time series OLS regression with 36
observations for the Chrysler Corporation.

Adjusted R2 from estimation of equation (15) using
combinations of accounting techniques.

Test Criterion. ao+alMS from estimation of
equation (15).

Significance (Z-statistic) of individual
coefficients.

't' test for difference in means of individual
coefficients.

Adjusted R3. analysis of variance for the
'Iron and Steel' industry.

Adjusted R2. analysis of variance for the
'Automobile' industry.

viii

Page
28
22

49

SI

51

52‘

53

54

SS

56

S7

57

58

59

Table

18.

IS.

26.

21.

22.

23.

24.

25.

26.

27.

Adjusted R2. analysis of variance for the
'Iron and Steel' and the 'Automobile'
industries.

Adjusted R3. firm rankings. and Kendall's
coefficient of concordance for the combined
sample of the 'Iron and Steel and 'Automobile'
industries.

For the comparison LIFO vs.FIFOlSLNC. Adjusted R3.
firm rankings. and Spearman's rank correlation
coefficient for the 'Iron and Steel' and
'Automobile' industries.

For the comparison LIFO vs.FIFOlDDBNC. Adjusted R3.

firm rankings. and Spearman's rank correlation

coefficient for the 'Iron and Steel' and
'Automobile' industries.

For the comparison SL vs. DDBlLIFONC. Adjusted R3.
firm rankings. and Spearman's rank correlation
coefficient for the 'Iron and Steel' and
'Automobile'industries.

For the comparison SL vs. DDBlFIFONC. Adjusted R3.
firm rankings. and Spearman's rank correlation
coefficient for the 'Iron and Steel' and

'Automobile' industries.

For the comparison RC vs. HCtLIFOSL. Adjusted R3.
firm rankings. and Spearman's rank correlation
coefficient for the 'Iron and Steel' and

'Automobile' industries.

For the comparison RC vs. HCIFIFOSL. Adjusted R2.
firm rankings. and Spearman's rank correlation
coefficient for the 'Iron and Steel' and
'Automobile' industries.

For the comparison RC vs. NCOtLIFODDB. Adjusted R3.
firm rankings. and Spearman's rank correlation
coefficient for the 'Iron and Steel' and

'Automobile' industries.

For the comparison RC vs. NC lFIFODDB. Adjusted R2.
firm rankings. and Spearman's rank correlation
coefficient for 'Iron and Steel' and

'Automobile' industries.

61

62

62

63

63

64

64

65

65

Table

28.

29.

38.

31.

32.

33.

34.

35.

36.

37.

38.

35.

48.

Adjusted R3. Spearman's rank correlation
coefficient for the combined sample of the 'Iron
and Steel' and 'Automobile' industries.

Adjusted R8. firm rankings. and Kendall's
coefficient of concordance for the Iron and Steel
industry.

For the comparison LIFO vs. FIFOtSLNC. Adjusted R1.
firm rankings. and Spearman's rank correlation
coefficient for the Iron and Steel industry

For the comparison LIFO vs. FIFOIDDBNC. Adjusted
R3. firm rankings. and Spearman's rank correlation
coefficient for the Iron and Steel industry.

For the comparison SL vs. DDBtLIFONC.Adjusted R3.
firm rankings. and Spearman's rank correlation
coefficient for the Iron and Steel industry.

For the comparisonSL vs. DDBtFIFONC. Adjusted R3.
firm rankings. and Spearman's rank correlation
coefficient for the Iron and Steel industry.

For the comparison RC vs. HCLIFOSL. Adjusted R3.
firm rankings. and Spearman's rank correlation
coefficient for the Iron and Steel industry.

For the comparison RC vs. NCFIFOSL. adjusted R3.
firm rankings. and Spearman's rank correlation
coefficient for the Iron and Steel industry.

For the comparison RC vs. NCLIFODDB. Adjusted R2.
firm rankings. and Spearman's rank correlation
coefficient for the Iron and Steel industry.

For the comparison RC vs. NCFIFODDB. Adjusted R1.
firm rankings. and Spearman's rank correlation
coefficient for the Iron and Steel industry.

Adjusted R2. Spearman's rank correlation
coefficient for the Automobile industry.

Adjusted R3. firm rankings. and kendall's
coefficient of concordance for the Automobile
industry.

For the comparison LIFO vs. FIFOlSLHC. Adjusted R3.

firm rankings. and Spearman's rank correlation
coefficient for the Automobile industry.

X

Plat
66

67

78

78

71

71

72

73

74

Table

63.

64.

65.

66.

67.

78.

71.

72.

73.

74.

For the comparison SL vs. DDB:FIFONC. Test
criterion. firm rankings and Spearman's rank
correlation coefficient for the Iron and Steel
industry.

For the comparison RC vs. NCLIFOSL. Test criterion.
firm rankings and Spearman's rank correlation
coefficient for the Iron and Steel industry.

For the comparison RC vs. NCFIFOSL. Test criterion.
firm rankings and Spearman's rank correlation
coefficient for the Iron and Steel industry.

For the comparison RC vs. NCLIFODDB. Test criterion.
firm rankings and Spearman's rank correlation
coefficient for the Iron and Steel industry.

For the comparison RC vs. RCFIFODDB. Test criterion.
firm rankings and Spearman's rank correlation
coefficient for the Iron and Steel industry.

Market power test criterion. firm rankings
and Spearman's rank correlation coefficient for
the Iron and Steel industry.

Market power Test criterion. firm rankings
and Kendall's coefficient of concordancefor the
Automobile industry.

For the comparison LIFO vs.FIFO:SLHC. Test
criterion. firm rankings and Spearman's rank
correlation coefficient for the Automobile industry.

For the comparison LIFO vs. FIFO: DDBNC. Test
criterion. firm rankings and Spearman's rank
correlation coefficient for the Automobile industry.

For the comparison SL vs. DDS: LIFOHC. Test
criterion. firm rankings and Spearman's rank
correlation coefficient for the Automobile industry.

For the comparison SL vs. DDBIFIFOHC. Test
criterion. firm rankings and Spearman's rank
correlation coefficient for the Automobile industry.

For the comparison RC vs. HCLIFOSL. Test criterion.

firm rankings and Spearman's rank correlation
coefficient for the Automobile industry.

xiii

Page
87

91

92

92

93

93

94

Table Page

75. For the comparison RC vs. NCFIFOSL. Test criterion. 94
firm rankings and Spearman's rank correlation
coefficient for the Automobile industry.

76. For the comparison RC vs. NCLIFODDB. Test criterion. 95
firm rankings and Spearman's rank correlation
coefficient for the Automobile industry.

77. For the comparison RC vs. ECFIFODDB. Test criterion. 95
firm rankings and Spearman's rank correlation
coefficient for the Automobile industry.

78. Market Power Test criterion. firm rankings 96

and Spearman's rank correlation coefficient for the
Automobile industry.

xiv

I INTRODUCTION

Establishing methods for assessing the desirability
and effects of alternative accounting techniques has long
been a goal of accounting researchers and accounting
policy makers. Despite an abundance of empirical and
theoretical research in the area. several of the issues
are still unresolved. The purpose of this study is to
examine the effects of accounting techniques in the
context of a specific economic analysis and with respect
to a specific class of users. The economic analysis
involves the study of market performance in the area of
industrial organization by researchers and antitrust
policy makers.

Two important reasons for analyzing the impact of
different accounting methods on the relationship between
profitability and market structure are the following.
First. as mentioned before. the issue of accounting
techniques is of considerable importance for accounting
policy makers. Second. the study of determinants of market
structure has relevance not only for economists. but also
for regulators. legislators. and lawyers dealing with
antitrust litigation (Adams (1979)). In this context.
accounting rates of return are frequently used as a
measure of profitability and treated as an index of

monopoly power.l

Accounting numbers are among the most frequently used
data in empirical work on industrial organization. The
systematic properties of these numbers are used as a basis
for making inferences about market structure and market
performance factors (such as the presence of monopoly
power and above normal profits). Thus. potential effects
of accounting techniques on such inferences are an
important factor to be considered by researchers in this
area. According to Gonedes and Dopuch [1978]. this role
of accounting numbers and the large cost of making
avoidable. unwarranted inferences are sufficient grounds
for being interested in the effects of accounting

techniques.

II BACKGROUND

ZEl_EiEE91I_21_A1I8IDIIA¥£_AQSQNDLADS_I£SDDAEN£E

Research examining the effects of alternative
accounting techniques has taken several approaches. Prior
to the mid 1868's most attempts to deal with the question
used 'a priori' models which purported to establish
criteria for identifying the 'best' or 'optimal'
accounting procedures (e.g. Paton and Littleton (19481.
Chambers (1966]. and Edwards and Bell [1965]).
Subsequently the 'a priori' approach came under heavy
criticism because (I) it lacked theoretical background.
and (2) using this approach one could declare just about
any set of accounting procedures 'superior'. These
failings have led to an increased emphasis on the use of
empirical analyses to assess the effects and desirability
of alternative accounting techniques.

Other approaches taken by researchers in this area
include the 'materiality' studies and the studies based on
the connection between market equilibrium and accounting
techniques. The 'materiality' research examined whether
the process generating an accounting number according to
one accounting technique is significantly different (say
by a percent rule) from the process generating the same
number according to a different technique. This test of
'significance' usually involved point by point comparison

3

of observed values conditional on alternative accounting
techniques (see Rosenfeld (1869]. and Peterson (1973)).
These studies. however. ignored the time series issues
involved. and as such provided no basis for inferring that
the two stochastic processes corresponding to alternative
accounting techniques were different just because
contemporaneous drawings from each process were different.

In other studies (e.g. Dopuch and Watts [1872).
Dopuch and Drake [19661. and Dopuch and Pincus (19841) it
has been argued that significance (materiality) could be
determined by assessing the effect of a change in
accounting technique on parameters of the time series
process generating the accounting number of interest.
These studies do not answer the question of whether
predictive ability. or variability. is a sufficient basis
for assessing the effects or the desirability of
accounting techniques. Similar deficiencies characterize
the simulation studies (examples of the simulation studies
are Simmons and Gray (1969). and Greenball (1968 and
1969]).

On the other hand. market equilibrium studies are
based on explicit theoretical models of resource
allocation under uncertainty (see Ball and Foster (1982]
for a review of such studies). These studies suffer from
two common problems. (I) When do capital market agents

learn about a firm's choice of accounting methods (date of

event). and (2) how does one control for confounding
information which is generally released along with the
"main event“? These models provide some underpinnings
for empirical work on techniques (see Gonedes and Dopuch
(19741 for a discussion). solid underpinnings are also
available in other areas. such as the area of industrial
organization.

The approach used by a number of researchers working
in the area of industrial organization can be categorized
as the 'True' or "Ideal" number approach (see e.g. Ayanian‘
[1975]. Bloch [1976). Clarkson (19771. and Neiss [1969]).
These studies are based on the premise that accounting
techniques lead to 'biased' income numbers because they
are not computed according to procedures asserted to
provide values of 'Ideal' or 'True' income.

In some of these studies it was found that inferences
based on restated values differed from those based on
reported values. These studies. nonethless. did not
specifically assess the potential effects of accounting
techniques on inferences that were based on values of
accounting numbers. More importantly. in several of these
studies it was asserted that the restatement techniques
used were somehow superior to conventional techniques in
that they led to 'more informative' numbers. The
restatement techniques used in these studies. however. are

just as arbitrary as conventional techniques since the

models they are based on are not descriptive of a world
with uncertainty. imperfect markets. and multiperiod

horizons.

Z-ZW
There are two basic problems of using accounting
based variables in economic analyses which can be

conceptualized as follows (see Nair [1977]).

A s 8

here A is the unobservable economic variable such as
economic profit. B is the accounting surrogate such as
accounting income. and B1 and 82 are accounting income
numbers computed under alternative accounting techniques
such as LIFO vs. FIFO. The first problem deals with
whether the accounting variable is a good surrogate for
the economic variable. i.e.. whether the same modeled
relationship holds for the economic and the accounting
variable. The second problem is that even after picking
the surrogate accounting variable. the problems continue
because of the several different ways of producing the
accounting number. Most of the previous studies have only

examined the second problem.2

In the proposed study. both of these problems are
dealt with. The first problem is overcome by developing a
structural model which specifically links accounting
numbers with parameters of market structure such as the
elasticity of demand. product differentiation. and market
share. This model also attempts to deal with the issue of
multiperiod horizons. which according to Gonedes and
Dopuch [1879] (footnote 18) is one of the basic concerns.

Based on this model. the second problem is examined
by forming and testing hypotheses about the potential
effects of alternative inventory (LIFO vs. FIFO).
depreciation (st. line (3L) vs. double declining balance
(DDB)). and asset valuation (historical cost (NC) vs.
replacement cost (RC)) techniques on inferences about
market power. Instead of presuming that the same
relationship exists between independent and dependent
variables when estimated using accounting instead of
economic values. the present study will model the
relationship between accounting values and the parameters
of interest directly. and based on it will hypothesize the

effects of alternative accounting techniques.

2-3 EI9111Ih1li12.lflfl.fll£&ll.§1£fl£$fl£l
Collusion in oligopolistic industries that is

potentially damaging to consumers has not only been the

cause of legal restrictions and increased regulation. but

has also given rise to an enormous amount of research
attempting to relate market concentration to the exercise

3 Most such studies have measured

of monopoly power.
market structure by an index of concentration‘ (e.g. the
four firm concentration ratio as a surrogate for market
power). and performance by accounting profit (such as
gross profit/sales as a surrogate for the Lerner Index).s
In general. these studies found a positive relationship
between concentration and accounting profitability
measures.

A major criticism of these studies has been the
potential introduction of biases by the use of Accounting
Rates of Return (ARR). Benston [1985] discussed several
studies that have used the ARR as a surrogate for the
Internal Rate of Return (IRR). He pointed out that if the
differences between accounting measures and economic
measures were randomly distributed with respect to
variables of interest. the result would not seriously
alter inferences about the parameters of interest.
Opinion is divided. however. as to whether this error is
actually randomly distributed. Nagerman and Senbet
(19761. and Long and Ravenscraft [1984] have claimed that
there is no correlation between choices of accounting
techniques and concentration. while others (e.g.. Watts

and Zimmerman (1978]. Dhaliwal. Salamon and Smith (1881].

and Bagerman and Zmijewski (1979]) have found systematic

differences in the accounting methods adopted by firms of
different sizes. This latter result implies that the
relationship between concentration and profits may be
present because accounting mismeasurement is positively
correlated with size (which in this case is presumed to be
a measure of concentration).

Several other researchers (see Harcourt [1965).
Livingstone and Salamon [1878]. and Fisher and McGowan
[1983]) developed theoretical models of firms' investment
patterns and showed that the ARR is a poor surrogate for
the firms' IRR. and therefore according to Fisher and

McGowan.

Accounting Rates of Return even if properly measured
provide almost no information about economic rates
of return. (See Fisher and McGowan (1983). pp. 82)
Salamon [1985] came to a similar conclusion using
conditional IRR estimates and said that it was the nature
and the strength of the measurement error in the ARR.
rather than the correctness of the underlying economic
argument. that accounted for the observed positive
association between firm size and firm profitability.
In his discussion of the Fisher and McGowan article.
Martin (1984] pointed out that in economics literature
there are different definitions of economic profit. only

one of which is the discount rate that equates the present

18

value of its expected net revenue stream to its initial
outlay (the definition of the IRR used in the theoretical
studies mentioned before). Another definition that arises
normally in the formal models of profit maximizing firms
is the Lerner Index of monopoly power (or the price-
marginal cost margin that was defined in footnote 5).
Martin further goes on to state that.
It is doubtful whether any measure of profitability
can be unambiguously identified as "correct” to the
exclusion of all others. for the purposes of economic
analysis. Fisher and McGowan's discussion of what
they call the "economic rate of return" does not
establish that measures of profitability based on the
Lerner index are inappropriate for economic analyses.
(Martin (1985]. pp. 582)

The model developed in the proposed study is based on
the Lerner Index of monopoly power. Unlike earlier
models. however. where accounting values are treated as
surrogates of economic values. this model relates market
power directly to accounting values of gross profit. plant

and equipment. and the value of raw materials used in the

production of the goods sold.

III THE MODEL

The model relating accounting profit rate to market
power is a variation on the model developed by Martin
[19851. and Clarke. Davies and Waterson [1876]. Model
development is divided into three sections. The first
section gives the background to a simple model of a profit
maximizing firm in an oligopolistic environment. In the
second section. the model is developed further to
encompass factors of production. and relate the firm's
profit rate (specifically its accounting profit rate) to
its market power. In the third section. effects of
accounting techniques on the key parameters of the model

are-analyzed.

3-1 A_IiIRlE_IQdIl_9I_I_R£Q‘1I_leilliinl_iill
Market power can be thought of as the ability of a

firm to influence the price of its own product. This
notion of market power can be captured in the context of a
profit maximizing firm in a monopolistic environment
(i.e..the case of a single seller economy) as.

f ' P(O)O-C(Q) (1)
where t is the profit. P is the price that the firm
charges for its output. 9 is the quantity sold. and 0(0)
is a cost function. Note that here the price P is a
function of the quantity sold. 0.

11

12

In this simple case. it is assumed that the quantity
of goods sold is equal to the quantity of goods
manufactured. Assuming the usual conditions for a
production function that are neccessary for profit

maximization. the first order conditions give.6

11:..89 . _l_ (2)
P ‘09
where MC is the Marginal cost. and £OP is the price

elasticity of demand for the firm. By adding and
subtracting average cost (AC) from both sides of the

equation.

LLAQ._l_ l.-AQ (3)
+ i I)
P 6 FC P
0P

where FC . AC/MC is the function coefficient is a measure
of the economies of scale of a firm. The function
coefficient will be large when the cost of production of a
marginal unit is small as compared to the average cost of
production. A function coefficient of one implies
constant returns to scale.

Extending this case of a single seller to an
oligopolistic environment means that now the price will
depend on not only the output of the firm but also on the

output of all rival firms. Defining a as a measure of

collusion among firms as.

“9299.: (4)
0i dO-i

where Oi I firm's own output. and Q-i I the output of the

rival firms. Making the assumption that a is a constant.

13

and maximizing the profits of the firm in an oligopolistic

environment. eqn. (3) becomes.7
M.s_:r_fiifl_:_sl+tl.ubfi (5)
P (i FC P

here Si 8 Qi/D-i+0i is the firm's market share. and (i is
the price elasticity of demand for firm i. For the simple
model given by eqn. (5). the test for market power
involves the intercept term Iu+Si(I-«)I/ 6i. When the
firm is in pure competition. the price elasticity of
demand (i 9 a» and so the intercept should 4 8. Therefore.
the presence of market power would imply an intercept that
is not equal to 'zero'.

In order to estimate eqn. (5) for a profit maximizing
firm. an evaluation of the optimal factors of production
or AC is needed. This is done in the next section by
maximizing the one period economic profit (change in

value) of a firm.

3-2 WWW

To evaluate the cost of the factors of production. AC
in eqn. (5). the present discounted value of firm is
maximized over one period. It is assumed that the firm
invests in capital at the beginning of the period. and
produces and sells at the end of the period. Then the
present discounted value (PDV) of a firm can be written as
the discounted value of the firm at the end of the period

less the value at the beginning of the period. i.e.

R -
PDV - - PIKE! + 11 - b.(Srt + Rt) + (___l___)(P(s

here 0

IR.

Hy %an‘€01'fl

.1
I

14

K ._._
.Q)Q
l+r* t+l

- WL - c’(st+1) - c’fls:t + at) + Pftl - T)(Kz + I)
a ._ ._
+ bl(1 - a)(Srt +nt> +P(St+l.Q)St+l) (5)

Per unit price of capital stock other than raw
materials (plant and equipment) at the beginning of
the period.

Endowment of Plant and Equipment inherited from the
last period.

Investment in Plant and Equipment during the time
period.
Price per unit of capital consisting of raw

materials.

Stocks of raw materials at the beginning of the
period( beginning inventory of raw materials).

Raw materials purchases.

8 Stocks of finished goods inventory at the beginning
f the period.

Discount rate for the period.

Price of goods sold.

Quantity of goods sold.

Per unit labor cost.

Total units of Labor used during the period.

Per unit holding cost for finished goods inventory.

Per unit holding cost for raw materials inventory.

Per unit price of Plant and Equipment at the end of
the period.

Depreciation rate Plant and Equipment.

15

? I Per unit price of raw materials at the end of the

period.

b

o I Rate of usage of raw materials.

Note: a I 1 implies no ending raw materials inventory.

It is important to note the following points about

eqnation (6) for the PDV.

I. This equation differs from the PDV formulation in
earlier models. in that. both raw materials and
finished goods inventories have been incorporated in
the model. It is no longer assumed that all of the
goods manufactured are sold. This inclusion of
inventories in the model has been done in order to
analyze specifically the effects of alternative
accounting inventory techniques(LIFO vs. FIFO) and to
separate their effects from the effects of
alternative long lived asset valuation techniques
(e.g. HC vs. RC. and SL vs. DDB)

2. In this model price P is a function of the quantity
of finished goods sold 9. and the ending finished

goods inventory S . This formulation is based on the

t+1
argument that the presence of finished goods inventory
helps meet consumer demands more efficiently. and
therefore. the firm should be able to charge a higher
price for its product which includes the provision of
this service.

3. Last period's finished goods inventory. S is not

t8

16

explicitly included in the model because its value to
the firm or its opportunity cost is not determinable.
This is because it is assumed that the firm can only

produce the good and is not able to buy it in the

outside market.8

4. The per unit inventory holding costs. CF. and CR for
finished goods and raw materials respectively. are
assumed to be constants. Also it is assumed
that all costs to hold last period's finished goods
inventory were paid out in the previous period and
hence do not feature in this period's decision

problem.

Therefore. the decision problem is to maximize PDV or

equivalently (1+r*)PDV.

Equation (6) can be simplified by substituting PT and "a

in the equation where.

“T I r* + T + (1 I T)( )

bl - b
and Pa I r* + a + (1 - o)( n__)

17

Substituting PT and ”a in eqn. (6). the decision problem
becomes.

Max P(St+l.9)9 - WL - c'(s

)
t+1
[L'I'Rt'st+l '0]

- c"<s:t + at) - 1a,, 9:th + I)
R
- ”a b'(Srt+Rt)

* (1/1+r*) P('t+1'°)st+l

s.t. St + O‘KE + I. L. Srt + Rt) ' O + sttl

here. O‘KE + I. L. Srt + Rt) I Quantity of finished goods

manufactured.9

Note that in this simplified form. P? can
be thought of as the implicit rental for services of a
unit of capital consisting of Plant and Equipment that
costs a dollar. and PA is likewise the implicit rental for

capital that consists of raw materials.

A detailed mathematical analysis of the following
section is given in the appendix. The first order
conditions (eqns. (A.1). (A.2). (A.3). (A.4). and (A.5) in

the appendix) are.

H . no (a)
sp/sst+l(o + (1/l+r*)st+l) - cF + u . no (9)
HYPE/MFR . p - no (10)
R - 8
(CR + Hobo)/MPR - y no (11)

18

W/MPL ' P I MC (12)

here MFR. NPR. and MPL are the marginal products with
respect to capital (R and R) and labor respectively.

Using eqns. (18). (11). and (12) to evaluate the
factors of production we get eqn. (A.9) from the appendix.
PQ-WL-bR -CRR

PO

t

 

I (1 - FC + 5E?) + (l-E-B)FC.Si +

K i i
+ ”We“ + I”'t p
p9 -;5(h a - 1)

 

(13)

here h is a parameter that characterizes an inventory
policy under which the ending inventory of raw materials
available for use are kept at a constant proportion to the
optimal quantity of raw materials to be used in production
(see the appendix for a complete discussion).

Equation (13)in terms of accounting Sales Revenue. and
accounting Cost of Goods Sold is (from eqn. (A.12) from
the appendix).

19
WW . (1 _ so + 95%;) + (1_(;_g)rc.31
i .

 

Sales
P _l_
+ T!“ +1 1 5.1"
CGS
Change in the values raw
of raw material material
+ (bra - 1)[_.l_, (Marin + Mean) (14)
gl + 1 Sales

CGS

So the equation to be estimated is of the form.
M = I. + I151 + I2[_L_,

 

Sales 91 + 1 Sales
CGS
Change in the values raw
of raw material material
+I _1_ IM + WJ+2. (15)
3 I S I
+ 1 ales
CGS

here oI is the change in finished goods inventory and ei
is a stochastic error term.
For the evaluation of market power. the coefficients of
interest are.
an I 1 - FC + («PC/(i). (16)

and. al I ((1 - «)/€i)IFC (17)

From eqns. (16) and (17). [octSiU-«H/fi can be shown
to be.

(«+S.(1-o<)l a + a S.
l a 1 l _ 8 1 l (18)

6i F0

 

Now for the pure competition case. as 6i I m. 1 - (I -

(a0 + alSi)I/FC should I 8. Therefore. the test criterion

28

for the presence of market power depends on. 1 - (I - (a0
+ alSi)J/FC being # 8. For the case when a > 8. the test
criterion for the presence of market power is of the

form.10

1 - FC < a. + '13: (19)
Hence. the critical coefficients for the test of market

power are '8' and '1'

3-3 Effss1s_2f_As99nn1ina_Tssbninnss_2n_ibs_sesffisisnta__

9f_inisrsst

In the model given by eqn. (15) it is assumed that
sales S. and market share Si. can be measured without
error 12. while the CGS. Plant and Equipment. and
inventories are susceptible to errors introduced by

accounting techniques. The possible estimations of the

model are given in Table 1.

Table 1. Combinations of Accounting Techniques.

 

RC(LIFO)13 HC(LIFO) HC(FIFO)
SL 1 2 4
one - 3 s

 

By comparison of the parameter estimates in each of the

21

above cases. it is possible to test the effect of the
alternative combinations of techniques on the test
criterion given by eqn. (19). the magnitudes of the
individual coefficients ('8' a1. a2. a3). and on the
fitted R2 of the estimated equations.

From econometric theory it can be hypothesized that
even though the coefficients of interest. a. and a1. do
not involve variables with systematic errors. the errors
in other variables. however. can impact the coefficients
of interest. The magnitude and direction of this bias
will depend on the relative magnitudes of the covariance
matrices of errors with respect to the covariance matrices

14

of the true variables. In eqn. (15). it can be assumed

that the error in the term (1/((oI/CGS) + II) is very

15 Therefore. the errors in the last two variables

small.
on the right hand side of eqn. (15) are primarily due to
errors in plant and equipment. and raw material inventory
measurements respectively. The effects of these errors
can be highlighted by comparing different estimations as

shown in the following table.

22

Table 2. Pairwise comparison of alternative accounting
accounting techniques.

 

Comparison of Highlights the t conditional on
effect of
A (1) with (2) HC vs. RC 2 SL
B (2) with (4) LIFO vs. FIFO 3 HC and SL
C (3) with (S) LIFO vs. FIFO 3 NC and DOS
D (2) with (3) SL vs. DDB : HC and LIFO
E (4) with (5) SL vs. DDB : NC and FIFO

 

IV NYPOTNESES

The following hypotheses are formulated in order to
test the robustness of the estimated model with respect to
alternative accounting techniques. The hypotheses are

divided into three parts.

As a preliminary test. the first set of hypotheses can
be formulated in order to test the robutness of the
induvidual estimated coefficients. The hypotheses can be

stated as.

(1a) N 1!: a I 8

here superscript c I 8.......3 corresponds to the

coefficients '8' .....a3.

(1b) HBIb: a 3:” I a .j:" for all j. j.'m. and c

° 3*f'

Kalb: ac

J-m * 'cj'"

here (j.j') denote a pair of alternative accounting

techniques(e.g. LIFO. FIFO). and m denotes a particular

combination of techniques that j and j'are conditional on.
The null hypothesis 1a states that the individual

estimated coefficients are insignificantly different from

zero. The rejection of this hypothesis and proper signs on

23

24

the coefficients would indicate that the model is robust
with respect to estimation.

The null hypothesis (1b) states that there is no
effect of accounting techniques on the estimated
coefficients. This would imply that the model is robust
with respect to accounting techniques.

Hypotheses (1a) are tested by examining the 't'
statistics of the estimated coefficients ( see

Christie (1986)). Hypotheses (1b) are tested by using
the 't' test for difference in means with unknown
variances.

A second set of hypotheses can be formulated based on
the fitted R3 of the estimated equations.

(2) N 1: R2 I R2

I
l

38

HA]: R2 j'

+ R2
here j. and j' refer to different sets of accounting
techniques.

These hypotheses are tested in two ways. First. the
form of analysis outlined in section 3.3 can be viewed as
a single factor experiment with 8 subjects or firms being
measured under combinations of 5 levels of factors
(accounting methods. combinations 1.....5 in Table 1).
The proposed technique is the repeated measures technique

described by Myers (1972 (pp. 186-187)]. Second. the

25

above hypotheses are tested using nonparametric tests such
as the Kendall's coefficient of concordance and the
Spearman's correlation coefficient.

The third set of hypotheses are formulated in order
to test the senstivity of the test criterion given by
eqn. (19). Under the assumption of constant returns to

16

scale. the test criterion(T) for market power is.

T I a' + alSi > 8

Using this test criterion. the firms can be ranked by
market power. The stability of this ranking system with
respect to estimation techniques is an important policy
issue. The hypotheses based on the test criterion can be
stated as.
(3a) Ho3a: Tj I Tk for all j.k

j+k

HA3a: Tj * Tk
where j.k I 1......7 denote the combinations of techniques

given in Table 1.

(3b) 1
Ho3b: rj" . r3 ‘" for :11 j. j'. and m
3*3'

HA3b: Tj‘“ * 13"”

26

The null hypotheses state that the test for market power
is stable with respect to alternative accounting
techniques. whereas the alternative hypotheses state that
it is not. Hypotheses (3a. and 3b) are tested by
examining the shifts in firm rankings due to alternative
accounting techniques. This shift in ranks is tested by
using a non parametric test such as the Kendall's
coefficient of concordance and the Spearman's correlation
coefficient.

A rejection of the alternative hypotheses (3a. and
3b) would indicate that the estimation of the model is

robust with respect to accounting methods.

V METHODOLOGY

The methodology consists of estimating equation(15)
with time series data at the firm level. This method is
preferable to using industry level crossectional data for
the following two reasons. 21511. as the risks faced by
different firms in an industry are different. it is
unlikely that the implicit rate of rental for capital
stock (P) is the same for every firm in the crossectional
sample. Therefore. the use of crossectional data is not
appropriate for the estimation of this model. The 339931
reason is that the issue of firm specific market power is
aquestion with direct policy implications. and therefore
its stability with respect to estimation techniques is
worth investigating.

In. summary. the research method is to restate
reported inventory and capital stock figures for a sample
of eight firms. This restatement yields a set of five
financial statements for each firm. i.e. 48 in all. The
model is first estimated with the reported data and then
with the restated data. The results are then compared in
order to examine the effects of alternative accounting

methods on the relevant estimated coefficients.

27

28

5.1 W

One of the ways of examining the effects of
accounting techniques on economic analyses proposed by
Gonedes and Dopuch [19791 was to restate reported
accounting numbers of the firm and repeat the economic
analyses with the restated numbers.

The restatement routines used in this study for
inventory and depreciation amounts are the methods
proposed by Nair (19771 which basically make use of the
Dollar Value LIFO method for inventory and asset layering A
techniques for depreciation restatements respectively.
The restatement to Replacement Cost is done using the

method proposed by Falkenstien and Wiel (19771.

5-2 W

The restatement method used for inventories is the
Dollar-Value LIFO method. This method essentially removes
from the inventory effects that are solely due to price
changes. These methods are illustrated in detail by
Nair (19771. As Nair pointed out. in order to apply these
, restatement procedures the following simplifying
assumptions have to be made. Though these assumptions are
fairly restrictive. they can be relaxed if more detailed

data were available. The assumptions are.

29

1. For firms which use combintions of techniques for
inventory. it is assumed that the primary method of
inventory valuation is used for all inventory. This
simplifying assumption is neccessary because not every
firm in the sample discloses the precise breakdown of the
different inventory methods in use for that firm. If more
precise data were available. it would be better to restate
only the relevant portion. For example if a firm has 88
percent of its inventory on FIFO and 28 percent on LIFO.
restatement to LIFO should be done only on 28 percent of
the inventory.

2. In restating LIFO inventory to FIFO it is assumed
that all inventory is aquired 5 years before the test
period. Similarly for FIFO to LIFO restatement. the
ending FIFO inventory 5 years before the test period is
assumed to be the beginning LIFO inventory. This is done
in order to approximate what the inventory might have been
had the firm always been using LIFO.

3. The indices used in this study are the wholesale

price indices for various commodity groups that are

published on a monthly basis by the fiygxgy_gf_gngrgn1
Business-

4. For firms that were divesified into several lines
of business. the price index used is the one for its

primary line of business.

38

5-3 BEI1IL£I£D1_£Q£_DER££EAILLQD

The method outlined by Nair for restating
depreciation figures is based on layering assets by using
capital expenditure data. These layers are depreciated
according to the alternative depreciation techniques and
the resultant depreciation amounts aggregated to yield the
total depreciation expense for the period. Since in this
study it is the restated capital stock series that is used
as one of the independent variables. this series is
reconstructed using restated depreciation figures
beginning from a period three years before the test
period. For example. the restated capital stock figure

for the year 19x is computed as shown below.

Restated Capital Stock 19xICapitol Stock 19(x-3)
+£aquisitions- Sretirements
IZdepreciation+8adjustments to

accumulated depreciation
account
Where the aquisitions etc. are summed over the three
periods prior to the test period. The following basic

assumptions are made in the restatement of depreciation

amounts.

1. The first-in-first-out flow is assumed with
respect to property acquisitions and retirements.

2. The salvage value is assumed to be zero.

31

3. Multiindustry firms are restated using the useful
life most appropriate for their primary line of

business.

5-4 W

The method used for restatement to Replacement cost
is proposed by Falkenstien and Wiel (19771. The price
indices used for restatement are published by the Bureau
of Labor Statistics's Wholesale Sale Price Index (WPI)
for various commodities.

According to Falkenstien and Wiel. FIFO inventory
method is a good approximation for replacement cost of
inventories. and LIFO cost of goods sold with adjustments
for depleted inventory layers is a good approximation of
the replacement cost of goods sold.

The procedure is somewhat more complicated for
restating the Capital Stock series. The method requires
breakdown of the Plant and Equipment account into separate
building and equipment accounts and determining the
average age of depreciable assets. This is estimated by
dividing the accumulated depreciation balance by the
repective depreciation charges for that period and then
averaging over the test period. This estimated average
is a reasonable proxy for the base year from which to

measure the replacement cost of fixed assets. Finally

32

appropriate price indices are applied to the net building
and equipment accounts to yield the replacement cost of
Plant and Equipment.

The basic assumptions and caveats for restatement to
replacement cost are.

1. A major approximation in this restatement
procedure is the overestimation of the average age of
Plant and Equipment which results in overestimates for the
increments used to adjust historical cost to replacement
cost. As mentioned above. the formula used for
estimating the average age is.

Average age of . Accumulated Depreciation

depreciable assets depreciation charge for the year

 

However. if the firm is using fully depreciated
assets. then the balance sheet shows the historical cost
of these assets while deducting accumulated depreciation
equal to that amount. Fully depreciated assets add
nothing to the balance sheet totals nor to the
depreciation charges. Yet. because they remain on the
balance sheet. they increase the numerator of the above
equation without adding to the denominator. thus
overestating the average age of depreciable assets.

This overestimation is impossible to remove without

33

detailed knowledge of the fully depreciated assets profile
for each firm.

2. Another approximation where replacement cost can
be different from that computed by the firm is due to the
fact that the firm's internally developed price indices
could be different from the price indices that are
available from the Bureau of Labor Statistics.

Due to the above reasons the restated values of Plant
and Equipment are at best an approximation of the actual
replacemnt costs. The above mentioned overstatement could.
bias the results towards significant effects of the change
to replacement cost. This potential bias is discussed

more fully in the results and analysis section.

5-5 m.

The time series data for 48 quarters was obtained
from the quarterly SEC reports. The sample consists of
eight firms in two durable goods industries that have
traditionally been interesting to researchers and policy
makers as to their levels of concentration. The
industries. firms. and their reported methods of

accounting techniques are given in Table 3.

34

Table 3. Firms. Industry SIC codes. and reported
accounting methods.

 

 

Firm Industry SIC code Firms Reported
accounting
methods

Armco Corp. 3318 LIFO SL HC

U.S. Steel Corp. LIFO SL HC

Lukens Corp. LIFO SL HC

LTV Corp. LIFO SL HC

General Motors Corp. 3711 LIFO DDB HC

Ford Motor Co. LIFO DDB HC

American Motors Corp. FIFO DDB HC

Chrysler Corp. FIFO DDB HC

 

There was only one firm. U.S. Steel. which had a
significant merger (over 48 percent of total assets) with
Marathon Oil. a firm that is in a different Indsustry.
This neccessitated netting out the effects of this merger
for the years 1982-1984. In order to compute market
shares. industry data for the Iron and Steel was obtained
from the publication 'Iron and Steel Institute' which
publishes monthly statistics on shipments and imports of '
Iron and Steel. The source of industry data for the

automobile industry is the 'Ward's Automotive Book'.

35

5-6 Nstbeds_ef_Analxais
As mentioned before. the main purposes of this study

are twofold. (1) to examine the performance of the
analytical model developed in section 3.1. and (2) to
evaluate the effects of alternative accounting techniques
on the performance of this analytical model. For the
first purpose. the results of estimation are examined with
respect to significance and meaningfulness of the
estimated coefficients. For the second purpose. the
results due to estimation with alternative accounting
techniques are compared with respect to the overall
goodness of fit (R3). the significance of the individual
coefficients. and the magnitude of the test criterion
(ao+alSi) for market power. For this set of analyses the
null hypothesis is that for a particular firm. change in
accounting methods will cause no difference in either the
goodness of fit of the equation. or in the estimated
coefficients that are relevant for the purposes of
estimating market power.

In order to test the first set of hypotheses. the
parametric 't' test is used to test the significance of

individual coefficients.

S-SI Analzsi:_9f_!arisnes

The second set of hypotheses formulated in section IV

are tested by using a specific version of the Analysis of

36

Variance technique called the Repeated Measures technique
(Meyers 1972 pp. 164-861. This technique is particularly
appropriate for an experimental design where different
treatments are applied repeatedly to the same subjects.
This technique reduces error variances thereby improving
the power of the test. but results in observations that
are dependent. The form of analysis outlines in section
3.3 can be viewed as a single factor experiment with 8(or
four per industry) subjects (firms) being measured under
combinations of five levels of factors(combinations of
accounting techniques). The dependent variable in this
analysis is the adjusted R2 of the estimated equations
which is an accepted measure of the explanatory power of a
model.19 The proposed model underlying the analysis of
the dependent variable R2 is as follows.

‘31: - 11 + «j + ’1: + (3" with jIl.....5. and

kII....8

here “j I Pj - P is the main effect of the accounting

method factor.

Bk I ”k - P is the main effect of the firm factor.

37

The null hypotheses of interest can be stated as.
H.2.: «3 I 8 for all j
n82b: 'k I 8 for all k

The assumptions underlying the Repeated Measures

technique are that.

1. ejk are i.1.d normal.

2. E(€jk) I 8 and E(€jk)3) I 03 i.e. that the

variance of all €'s is the same for all subjets.

Though these assumptions are restrictive and hardly
met for most populations. Box (19531 points out that in
the case where group sizes are equal. the analysis of
variance is not very sensitive to nonnormality or
inequality of variances. It is important to point out.
however. that the condition for randomness of sample
selection is not completely met. The sample selection
could be considered random to the point that the sample of
8 firms (4 in each industry) were chosen out of the 181
firms for which requests had been sent for data. The two
industries chosen for the sample were the ones for which
atleast four firms complied with the request for data.

Therefore. the selection of a particular firm is not

38

expected to be correlated with the changes in the
dependent variable due accounting techniques.

The third set of hypotheses formulated in section IV
can be tested by using nonparametric statistics. Under
the assumption of constant returns to scale. the test
criterion for market power given by equation(19) under the
assumption of constant returns to scale is ao+a1Si.

Therefore. the null can be stated as.

H : T I T

j for all j.k. j*k

k
where T denotes the test criterion. and j and k denote
different combinations of accounting techniques.

The significance of changes in estimated market power
due to different accounting techniques is tested by
examining the significance of the change in the rankings
of firms by market power. The ranking of firms by market
power within a particular industry or within the larger
set of all firms. is a relevant question when considering
policy issues such as merger guidlines or antitrust
litigation. A statistically significant shift in the
ranks due to estimation with a different combination of
accounting techniques would indicate a significant effect
of accounting techniques on the estimation of market

power. The test used to measure this shift in ranks is

39

the Kendall's coefficient of concordance W (Siegel (1956).
pp. 229-381 and the Spearman's rank Correlation
Coefficient (Siegel. 1956. pp. 285-131. These methods are
particularly appropriate for small samples. The same
methods are used to test the stability of the rankings of
firms by the best goodness of fit measure for the
estimated model. Also. Kendall's Coefficient of
Concordance is used to test the rankings of accounting
methods for the best fit (i.e. which method leads to the
results with the most expanatory power for a particular
firm). These tests are performed in two parts. first with
the partial sample at the industry level(i.e. for the four
firms in each industry). and ggggnd_. with the combined
sample of both industries. This two part testing of
results is done in order to highlight the inter as well
the intra differences in results between the two

industries incorporated in the sample.

S-6b MW

This statistic is used to measure the strength of
the agreement among different rankings of certain entities
by differnt ranking mechanisms. In this study the entities

are the firm's Test Criterion and Adjusted R1 and the

48

ranking mechanisms are the accounting techniques. The

statistic W can be defined as

H m s
34‘,- l”(113 - 111

Where S I Sum of the squares of deviations from the mean
of the sum of ranks assigned to each entity
being ranked.

k I Number of sets of ranks I number of ranking
mechanisms
N I Number of entities ranked.

This method is illustrated by using the data in Table 59

 

 

Firms Armco USS Lukens LTV
Acct.

Methods

LIFOSLHC 2 1 3 4
LIFODDBHC 2 1 3 4
FSH 2 4 3 I
FDH 2 4 3 1
R 2 1 4 3
Sum of the

ranks 18 11 16 13

Mean of the ranks I 58/4 I12.5

S I (18-12.5)2 + (11 I 12.5)! + (16 I 12.5)2
+ (13 I 12.5)2 I 21

Here k I 5. N I 4

Therefore W I Z] 3 I 8.168

(1/12) (5)2 (4 I 4)

 

The Coefficient of Concordance W. expresses the degree of
agreement in rankings among the five methods of
estimation. It is important to note that a W equal to 1
indicates perfect agreement among the five methods of

estimation and therefore implies that accounting

41

techniques have no effect on the estimated results. The
Null hypothesis for this statistic is that of complete
independence of rankings. 1.e. W I 8. therefore. by the
nature of this statistic. in order to show significant
effects of accounting techniques we need to accept the
Nullzo. In order to be statistically significant. the
value of estimated S has to be >- the critical value of S
that is availabe from the tables for this technique.
Critical value of S from table P (Siegel pp. 286) is 62.6.
Since the estimated S I 21 is not )I 62.6. the null .
hypothesis of independence of rankings is not rejected.
Therefore in the case of Iron and Steel Industry.

Accounting techniques have significant effects on the

ranking of firms by market power.

5-5c WWW

Tests using this coefficient are appropriate when the
agreement between two sets of rankings is being analyzed
(e.g. for the pairwise comparisons like
LIFO vs. FIFO: SLHC. This method of analysis is
illustrated by using the data for the Test Criterion in
table 68.

42

 

Ranks
Firms Armco USX Lukens LTV
Combination
of
Techniques
LIFOSLHC 2 1 3 4
FIFOSLHC 2 4 3 1
Difference di 8 -3 8 3
Edit I 18

Then the Spearman's Rank Correlation Coefficient is given)
by.

Since N I 4. therefore r8 I -8.88. Here rs I 1 would
indicate perfect agreement between different systems of
rankings or in other words no effects of accounting
techniques. From the table of critical values of rs
(Siegel pp 284) critial value of rs for N I 4 is 1.88.
Therefore -8.8 591 being >- 1.88. indicates that the null
hypothesis of independence of rankings cannot be
rejected. Once again here. accepting the null implies
that accounting techniques have significant effects.
Therefore in the comparison for the results of LIFO vs.
FIFO: SLHC. the results show that the effects of the
change from LIFO to FIFO on the rankings by market power

are statistically significant.

VI EMPIRICAL RESULTS AND ANALYSIS

The results in this section are divided into five
parts. Part (1) consists of general results about the
performance of the model. Part (2) gives the results of a
parametric 't' test which is done in order to test the
effects of accounting techniques on the individual
estimated coefficients. Part (3) gives the preliminary
analysis of variance results about the effects of
accounting techniques on the goodness of fit of the
estimated equations. Part (4) gives the results of
nonparametric tests such as the Kendall's Coefficient of
Concordance and the Spearman's Correlation Coefficient on
the rankings of adjusted R2 of the estimated equations. .
Finally part (5) gives the results of the nonparametric
tests for the robustness of firm rankings by the market
power test criterion. These nonparametric tests are
conducted in order to analyze the effects of accounting
techniques on the estimated results jointly ( using
Kendall's coefficient of concordance) as well as on the
basis of pairwise comparison (e.g. LIFO vs. FIFO or SL

vs.DDB). (using Spearman's correlation coefficient).

6-1 W

The results of regressions with reported and restated
variables are presented in Tables 4 to 11.

43

44

The F statisticis significant for every estimation of the
model. A summary of these tables for the adjusted R2 and
the test criterion T are given in Tables 12 and 13. The
adjusted R3 of the fitted equations ranges from 8.31 for
Lukens (LIFODDBHC) to 8.98 for General Motors (LIFOSLHC
and LIFODDBHC). The Test Criterion (T I ao+a‘MS) ranges
from 8.25 for LTV (LIFOSLDDB) to 1.298 for Ford Motor Co.
(FIFODDBHC).

Table 14 gives the results for hypotheses (1a). The
Z-statistic ( see Christie (19851) for individual
coefficients is significant for the constant term. so.
the coefficient for plant and equipment. a2. and the
coefficient for raw materials.a3. The market power test
criterion. T. is also highly significant. Contrary to
model predictions. the coefficients for market share and
plant and equipment frequently have negative signs. The
negative sign for the market share variable . however. is
insignificant. The reason for the insignificance of the
market share coefficient may be due the fact that since
market share for a firm changes very slowly. it is
probably highly correlated with the constant term. As for
the negative coefficient for the plant and equipment
variable. the negative sign persists only for the Iron and
Steel industry. This may be due to the fact that for the
time period under analysis. this industry has been

documented as making large capital expenditures for

45

inefficient and outdated plant and equipment ( see Adams
(1986) pp. 182). which could be a possible explanation for
the observed negative returns to plant and equipment

increases.

6-2 W
This test is done in order to test Hypotheses (1b).
The results are given in Table 15. As can be seen each of

the comparisons is highly significant.

5-3 AM

As mentioned in section 5.5. the dependent variable
for the analysis of variance is the adjusted R2 of the
estimated equations. The results for the analysis of
variance are given in Tables 16. 17. and 18. From the
significant F statistics it can be seen that there is an
inninniiignni effect of accounting techniques for either
the subsamples of the two industries or for the combined
sample. It can be noted that even though statistically
insignificant. the effect of accounting techniques is
greater for the Iron and Steel industry than for the
Automobile industry. On the other hand the firm effect is
highly significant in all three cases thereby indicating
that the expalanatory power of the model varies accross
firms or that there are additional firm specific

characteristics that are not captured by this model. This

46

last result is in line with what is to be expected for a

simple model such as the one developed in this study.

5-4 W

The adjusted R2 of the fitted equation is a measure of
the explanatory power of the model. This variable has
been studied by researchers in order to rank models (
Jorgenson [ 1) and its robustness with respect to
accounting techniques has implications for the kinds of

inferences drawn from such analyses.

6-4I W

The Kendall's coefficients of concordance for the
rankings by the adjusted R3 are given in Tables 18. 28.
and 38. the results indicate that the rankings of firms
by the explanatory power of the model are not affected by
estimation with alternative accounting techniques. This
result is true for the combined sample ( Table 18) as well
as for the industry subsamples for the Iron and Steel

(Table 28) and Automobile (Table 38) industries.

6-4b W

The consolidated results for the Spearman's
correlation coefficient are given in Tables 27. 37. and

47. For this pairwise analysis the result of no effects

due to accounting techniques only holds consistently for

47

the SL vs. DDB. All other comparisons show significant

effects.

6-5 W

The market power test criterion. whicn as mentioned
earlier is a measure of market power of the firm. A
statistically significant shift in the rankings of firms
by their market power will indicate a significant effect
of accounting techniques. This in turn would indicate the
possible introduction of a bias in the estimated
coefficients due to alternative accounting techniques.

The test statistics used to examine the significance of
shifts in these ranks are the Kendall's coefficient of

concodance. and the Spearman's correlation coefficient.

15-53:! Was:

The Kendall's coefficient of concordance for rankings
by market power are given in tables 49. 59. and 71. the
reasults in Tables 49. and 71 indicate that the ranking of
firms by market power is not affected by alternative
accounting techniques for the combined sample as well as
for the subsample for the automobile industry. However
for the subsample of the iron and steel industry. there is
a significant effect. This result would seem to indicate
that the estimation procedure is more robust towards
accounting techniques for the automobile industry than for

the iron and steel industry.

48

6-5b W

The results of the Spearman's Correlation Coefficient
test are given in tables 58 - 78 for the test criterion.
It can be seen that in the case of ranks based on the Test
Criterion. the change from SL to DOE has and insignificant
effect. and the change form RC to HCFIFO has a significant
effect across all samples. All other changes are
statistically significant at the individual industry level
but are insignificant at the combined level. It is
interesting to note that by rank ordering the rs values.
the effect is the most pronounced for RC vs. HCLIFO
(smallest rs ). and the least for LIFO vs. FIFOSL.

49

 

Table 4. Results of the time series OLS regression with 36
observations for Armco Inc.
Test
Market Raw Adj. F Criterion Durbin
Constant share FEE Material R1 Stat. T 1 Watson
a. a1 a2 a3 ao+alMS
LIFOSLHC .a63‘* .Oasss -.a.67* -.a516* .asg 21.2 .e592* 2022
(6098) (-1015) (-3a23) (.4021) (7a.8)
LIFODDBHC 8.643* 8.638 -8.862* -8.535* 8.67 19.7 8.596 2.13
(Sage) (.1a37) (.2a94) (-4a19) (5.85)
FIFOSLHC 8.999* -8.855* -8.858* -1.829* 8.88 38.1 8.936* 2.55
(13-8) (-2003) (.3a22) ('Uo31) (IZaI)
FIFODDBHC 1.881* -8.888* -8.859* -1.831* 8.79 36.7* 8.936* 1.83
RC 8.568* -8.728* -8.366* -8.375* 8.75 28.4* 8.514* 1.88
(7a3‘) (-2a65) (-6a6.) (.3all) (7a54)
( ) t values in parenthesis

Average Market Share 8.874

* significant at 8.81

 

58

Table 5. Results of the time series OLS regression with 36

observations for USX Corporation.

 

Test
Market Raw Adj. F Criterion Durbin
Constant share PSE Material R2 Stat. T 1 Watson
a0 a1 a2 a3 a.+alMS
LIFOSLHC 8.875* -8.189 -8.889 -8.948* 8.88 62.8* 8.829* 2.82
(13.5) (I1.52) (I1.49) (I12.9) (4.13)
LIFODDBHC 8.883* -8.187 8.811 -8.958* 8.87 62.8* 8.838 2.83
(13.88 (I1.49) (I1.35) (I13.8) (3.99)
FIFOSLHC 8.819* -8.197 I8.813 -8.858* 8.71 22.4* 8.771 1.85
(7a16) (.Oasa) (.0a77) ('6a‘.) (18.2)
FIFODDBHC 8.823* -8.192 -8.818 -8.846* 8.74 22.4* 8.774 1.85
(7e15, (.oaSS) ('8.U3’ (.GaS.) (10a4)
RC 8.885* '8.885 '8.818 '0a865* .076 3.e3* .a78‘ 2.88
( ) t values in paranthesis

1 Average Market Share 8.243

* significant at 8.81

 

51

Table 6. Results of the time series OLS regression with 36
observations for Lukens Steel Corporation.

Test
Market Raw Adj. F Criterion
Constant share PSE Material R1 Stat. T 1
a. a1 a2 a3 ao+alMS

Durbin
Watson

 

LIFOSLHC 8.426* 3.168 I8.833* I8.348* 8.32 5.2* 8.442*

LIFODDBHC 8.4185 3.433 I8.838* I8.344* 8.31 5.8* 8.435*
(5.38) (8.55) ('2.82) {-3.491 (5.85)
FIFOSLHC 8.854* I5.578 I8.821 I8.938* 8.53 21.8* 8.848*
(9.45) ('1.18) ('1.32) ('7.34) (8.83)
FIFODDBHC 8.859* 5.858 I8.158 8.955* 8.55 15.5 8.848*
RC 8.432* I8.555 I8.881* I8.358* 8.35 5.8* 8.428
(5063) ..el3. (-2006) (-3079) (6043)
( ) t values in paranthesis

Average Market Share 8.885

* significant at 8.81

2.182
2.197
2.21
1.98

2.17

 

52

Table 7. Results of the time series OLS regression with 36
observations for LTV Corporation.

Test
Market Raw Adj. F Criterion Durbin
Constant share PSE Material R3 Stat. T Watson
a. a1 a2 a3 ao'tal MS

 

LIFOSLHC 8.295 8.467 I8.892 I8.288 8.67 18.1* 8.251 2.85

LIFODDBHC 8.331* '8.855 8.188* 8.145 8.73 24.2* 8.325* 2.85
(2a12) (-0012) ('506.) ("073) (2038)

FIFOSLHC 1.858* I8.121 I8.859* I1.288 8.82 38.7* 1.858 2.85

FIFODDBHC 1.231* I1.851 8.818* I1.217 8.85 45.9* 1.228* 2.42
(9096) (-1044) (-406) (.130'2) (1506)

RC 8.488* 8.452 8.855* 8.837 8.54 15.7* 8.443* 2.87
(2.18) (8.871 ('3.52) (‘1.45) (2.55)

( ) t values in paranthesis

Average Market Share 8.8927

* significant at 8.81

 

53

Table 8. Results of the time series OLS regression with 36
observations for General Motor Corporation.

Market
share
'1

Test
Raw Adj. F Criterion Durbin
PLE Material R3 Stat. T Watson
a2 a3 a.+alMS

 

Constant
'11
LIFOSLHC 1.878*
(22.5)
LIFODDBHC 1.858*
(22.5)
FIFOSLHC 1.158*
(12.8)
FIFODDBHC 1.158*
(12.9)
RC 1.879*
(17.1)
( ) t values

I8.178
(I8.17)

I8.158
('1063)

I8.218
(-1046)

-0021.
(I1.45)

I1.125
(..093)

I8.885 I1.898 8.98 168* 8.997* 2.13
(I8.72) (I69.9) (46.1)

8.881 I1.878* 8.98 128* 8.984* 1.74
(8.89) (I78.28) (74.86)

8.883 I1.188* 8.92 138* 18.58* 2.15

8.884 I1.182 8.91 138* 1.864* 2.15
(8.36) (I16.95) .(28.4)

I8.888 1.868* 8.93 185* 1.827* 2.87
(I8.46) (145.8) (55.8)

in paranthesis

Average Market Share 8.418

* significant at 8.81

 

54

Table 9. Results of the time series OLS regression with 36
observations for the Ford Motor Company.

 

Test
Market Raw Adj. F Criterion Durbin
Constant share PSE Material R8 Stat. T 1 Watson
a. a1 a2 a3 a.+alMS
LIFOSLHC 1.828* -8.133 -8.811 -1.879* 8.94 184* 8.993* 1.96
(19.7) ('0093) (‘0.57) (.1801) (2‘02)
LIFODDBHC 1.828* (-8.248 (-8.833 -1.886* 8.95 188* 8.956* 2.19
(20a9) ('1085) (.1087) ('1408) (21.8)
FIFOSLHC 1.328* -8.778* 8.873 -1.186* 8.82 53.8* 1.128* 2.65
(11.2) (I2.12) (I1.48) (I8.31) (7.69)
FIFODDBHC 1.298* -8.675 -8.856 -1.l7* 8.81 51.7* 1.123* 2.56
(11.1) (I1.91) (I1.21) (-7.22) (7.78)
RC 8.946* -8.223 -8.838* 8.983* 8.88 88.5* o.888* 2.58
(13.7) (-1.15) (I2.69) (-18.7) (15.2)
( ) t values in paranthesis

Average Market Share 8.268

* significant at 8.81

 

55

Table 18. Results of the time series OLS regression with 36
observations for American Motors Corporation.

Test
Market Raw Adj. F Criterion Durbin
Constant share FEE Material R8 Stat. T 1 Watson
a0 a1 a2 a3 a.+a1M

 

LIFOSLHC 8.826* 8.896* 8.858 I8.889* 8.72 31.7* 8.828* 2.68

(7.28) (2.31) (1.59) ('8.35) (9.63)
LIFODDBHC 8.825* 8.474 8.835 I8.918* 8.72 31.8* 8.834* 2.58
FIFOSLHC ).59* 2.858 8.885 I8.748* 8.73 32.9* 8.717* 2.72

(6.92) (1.94) (8.15) (I7.94) (9.83)
FIFODDBNC 8.575* 2.158 I8.887 I8.735* 8.73 33.8* 8.715* 2.71
RC 8.855* 2.214 8.887 I8.989* 8.92 142 8.987* 2.93
( ) t values in paranthesis

1 Average Market Share 8.8187

* significant at 8.81

 

56

Table 11. Results of the time series OLS regression with 36
observations for Chrysler Corporation.

 

Test
Market Raw Adj. F Criterion Durbin
Constant share PSE Material R2 Stat. T Watson
a0 a1 a2 a3 ao+alMS
LIFOSLHC 8.597* 8.925 8.831 -8.744* 8.87 68.9* 8.679* 2.42
(8.47) (1.73) (1.12) (-8.58) (8.83)
LIFODDBHC 8.598* 8.615 8.812 -8.681* 8.89 53.1* 8.653* 2.48
(8.95) (1.32) (8.36) (I8.28) (9.82)
FIFOSLHC 8.533* 8.812 -8.832 -8.481* 8.87 58.1* 8.534* 2.78
(7.88) (8.821) (I1.21) (-5.21) (7.82)
FIFODDBHC 8.525* 8.814 -8/858 -8.468* 8.87 68.1* 8.526* 2.78
(8.13) (8.83) (I1.53) (I5.23) (8.86)
RC 8.768* 8.728* 8.814* -8.93* 8.89 69.7* 8.825* 2.86
( ) t values in paranthesis

Average Market Share 8.889

* significant at 8.81

 

57

 

 

Table )2. Adjusted R2 from the OLS estimation of equation (15).
Firms

Combination of
gichniques ARMCO USX LTV LURENS GM FORD AMC CHRYSLER
LIFOSLHC 1‘53, fiiflfl, 1‘51, 1‘11, 8.98 8.94 8.72 8.87
LIFODDBSL 8.69 8.87 8.73 8.31 1‘11 .1225 8.72 8.89
FIFOSLHC 8.88 8.71 8.82 8.63 8.92 8.82 8.73 8.87
FIFODDBNC 8.79 8.74 8.85 8.66 8.91 8.81 1‘11, 1‘11
RC 8.75 8.75 8.66 8.36 8.93 8.88 8.92 8.89

denotes reported combination of techniques.

 

Table 13. a +a MS from the OLS estimation of

Test Criterion. 8 1

equation (15).

 

 

Firms

Combination of
Techniques ARMCO USX LTV LURENS GM FORD AMC CHRYSLER
LIFOSLHC Qififiz_ fiifizayfli151‘1igfiz 8.997 8.993 8.828 8.679
LIFODDBSL 8.596 8.838 8.326 8.435 Qifififi‘gigfifi 8.834 8.653
FIFOSLHC 8.936 8.771 1.858 8.848 1.868 1.128 8.717 8.539
FIFODDBHC 8.936 8.784 1.228 8.848 1.864 1.298 8.215 Qifizfi

RC 8.514 8.784 8.443 8.428 1.827 8.888 8.987 8.825

denotes reported combination of techniques.

 

58

Table 14. Significance (Z-statistic) of individual

 

 

coefficients.
Coefficients

Combination Constant Market PSE Raw Test
of Share Material criterion
Techniques a0 a1 a2 a3 T I ao+al
LIFOSLHC 29.27** -8.26 -3.38* -43.38** 36.84**
LIFODDBHC 29.61** -8.28 -4.88* -41.97** 44.88**
FIFOSLHC 26.78** -1.16 -3.49* -24.84** 38.45**
FIFODDBHC 27.21** -1.79 -3.38* -22.97** 33.44**
RC 27.69** -8.51 -4.87* -33.48** 43.76**
Reported
Combination 29.45** -1.13 -5.43* -48.98** 45.64**

* Significant at .81.
** Significant at .881.

 

59

Table 15. 't' test for difference in means of individual

 

 

coefficients
Coefficients
Comparison 11. al a2 a3
LIFO vs. FIFO xx an an ax
SL v3. nnn we at at at
RC vs.HC xx an as as

** significant at 8.81

 

Table 16. Adjusted R1. analysis of variance for the 'Iron and
Steel' industry.

 

Effect Estimated F Critical value Level of significance

 

statistic of F
Firm 11.5 F3.12 I 5.95 8.81
Accounting 1.52 ’4 12 I 3.25 not significant

Method

 

Table 17. Adjusted R3. analysis of variance for the
'Automobile' industry.

 

Effect Estimated F Critical value Level of significance

 

statistic of F
Firm 6.22 F3.12 I 5.95 8.81
Accounting 1.21 F4 12 I 3.25 not significant

Method

 

Table 18. Adjusted R2. analysis of variance for the 'Iron and
Steel' and the 'Automobile' industries.

 

 

Effect Estimated F Critical value Level of significance
statistic of F

Firm 23.97 F7.28 I 3.36 8.81

Accounting 8.54 F4 28 I 4.87 not significant

Method

 

61

 

 

Table 19. Adjusted R3. firm rankings. and Randall's
coefficient of concordance for the combined sample
of the 'Iron and Steel and 'Automobile' industries.

Firms

Combination

of

Techniques ARMCO USX LTV LURENS GM FORD AMC CHRYSLER

Adusted R2

LIFOSLHC 1‘51 ‘Qififi, 1‘51, [‘11 8.98 8.94 8.72 8.87

LIFODDBSL 8.69 8.87 8.73 8.31 1‘21_ [‘35, 8.72 8.89

FIFOSLHC 8.88 8.71 8.82 8.63 8.92 8.82 8.73 8.87

FIFODDBHC 8.79 8.74 8.85 8.66 8.91 8.81 1‘11, [‘11 .

RC 8.75 8.75 8.66 8.36 8.93 8.88 8.92 8.89

Ranks

LIFOSLHC 6 4 8 7 I 2.5 5 2.5

LIFODDBHC 7 4 8 5 1 2 6 3

FIFOSLHC S 7 8 3.5 1 3.5 6 2

FIFODDBHC S 6 8 3 l 4 7 2

RC 5.5 5.5 7 8 1 4 2 3

Kendall's coefficient of concordance W I 8.7865
Value of S I 736.5
Value of I 24.5*

* significant at 8.81

 

62

 

 

Table 28. For the comparison LIFO vs.FIFO:SLHC. Adjusted R2.
firm rankings. and Spearman's rank correlation
coefficient for 'Iron and Steel' and 'Automobile'
industries.

Firms

Combination

of Armco USX Lukens LTV GM FM AM Chrysler

Techniques

Adjusted R3
LIFOSLHC 8.69 8.88 8.32 8.67 8.98 8.94 8.72 8.88
FIFOSLHC 8.88 8.71 8.63 8.82 8.92 8.82 8.73 8.87
Ranks
LIFOSLHC 5 3.5 7 6 1 2 4 3.5
FIFOSLHC S 7 8 3.5 1 3.5 6 2

Spearman's Rank Correlation Coefficient rs I 8.785*
* significant at 8.85

 

Table 21. For the comparison LIFO vs.FIFO:DDBNC. Adjusted R2.

firm rankings. and Spearman's rank correlation
coefficient for 'Iron and Steel' and 'Automobile'
industries.

Firms

Combination

of

 

Techniques Armco USX Lukens LTV GM FM AM Chrysler

 

LIFODDBNC
FIFODDBHC

LIFODDBNC
FIFODDBHC

Adjusted R2

8.67 8.87 8.31 8.73 8.98 8.95 8.72 8.88
8.79 8.74 8.66 8.85 8.91 8.81 8.73 8.87

Ranks

7 4 8 5 1 2 6 3
S 6 8 3 l 4 7 2

Spearman's Rank Correlation Coefficient rs I 8.786*
* significant at 8.85

 

63

Table 22. For the comparison SL vs. DDBtLIFOHC. Adjusted R3.
firm rankings. and Spearman's rank correlation
coefficient for 'Iron and Steel' and 'Automobile'
industries.

Firms
Combination of

Techniques Armco USX Lukens LTV GM FM AM Chrysler

 

Adjusted R2

SLLIFOHC 8.69 8.88 8.32 8.67 8.98 8.94 8.72 8.88
DDBLIFOHC 8.67 8.87 8.31 8.73 8.98 8.95 8.72 8.87

Ranks
SLLIFOHC 5 3.5 6 1 2 2 4 3.5
DDBLIFOHC 7 4 8 S 1 2 6 3

Spearman's Rank Correlation Coefficient ts I 8.875*
* significant at 8.81

 

Table 23. For the comparison SL vs. DDB:FIFOHC. Adjusted R3.
firm rankings. and Spearman's rank correlation
coefficient for 'Iron and Steel' and 'Automobile'
industries.

Firms
Combination
of

Techniques Armco USX Lukens LTV GM FM AM Chrysler

 

 

Adjusted R2

SLFIFOHC 8.88 8.71 8.63 8.82 8.92 8.82 8.73 8.87
DDBFIFONC 8.79 8.72 8.66 8.85 8.91 8.81 8.72 8.87

Ranks
SLFIFOHC 5 7 3.5 1 3.5 1 3.5 2
DDBFIFOHC 5 6.5 8 3 1 4 6.5 2

Spearman's Rank Correlation Coefficient rs I 8.958*
* significant at 8.81

 

54

Table 24. For the comparison RC vs. Hc:LIFOSL. Adjusted R2.

firm rankings. and Spearman's rank correlation

 

 

coefficient for 'Iron and Steel' and 'Automobile'
industries.
Firms
Combination
of
Techniques Armco USX Lukens LTV GM FM AM Chrysler
Adjusted R2
RC 8.75 8.76 8.36 8.66 8.93 8.88 8.92 8.89
HCLIFOSL 8.69 8.88 8.32 8.67 8.98 8.94 8.72 8.87
Ranks
RC 6 5 8 7 1 4 2 3
HCLIFOSL 6 3 8 7 l 2 5 4

Spearman's Rank Correlation Coefficient rs I 8.786*

*significant at 8.85

 

Table 25. For the comparison RC vs. HC:FIFOSL. Adjusted R2.
firm rankings. and Spearman's rank correlation
coefficient for 'Iron and Steel' and 'Automobile'

industries.

Firms
Combination

 

of
Techniques Armco USX Lukens LTV GM FM

AM Chrysler

 

Adjusted R2

RC 8.75 8.76 8.36 8.66 8.93 8.88 8.92

HCFIFOSL 8.88 8.71 8.63 8.82 8.92 8.82 8.73
Ranks

RC 6 5 8 7 1 4 2

HCFIFOSL 5 7 8 3.5 1 3.5 6

Spearman's Rank Correlation Coefficient r I

3 8.598

8.89
8.87

65

Table 26. For the comparison RC vs. HCO:LIFODDB. Adjusted R3.
firm rankings. and Spearman's rank correlation
coefficient for 'Iron and Steel' and 'Automobile'
industries.

Firms
Combination

of
Techniques Armco USX Lukens LTV GM FM AM Chrysler

 

 

Adjusted R2
RC 8.75 8.76 8.36 8.66 8.93 8.88 8.92 8.89
HCLIFODDB 8.67 8.87 8.31 8.73 8.91 8.81 8.73 8.87
Ranks
RC 6 5 8 7 1 4 2 3
HCLIFODDB 6 2.5 7 5.5 1 4 5.5 2

Spearman's Rank Correlation Coefficient rs I 8.735*
* significant at 8.85

 

Table 27. For the comparison RC vs. HC :FIFODDB. Adjusted R2.
firm rankings. and Spearman's rank correlation
coefficient for 'Iron and Steel' and 'Automobile'
industries.

Firms
Combination
of

Techniques Armco USX Lukens LTV GM FM AM Chrysler

 

 

Adjusted R2

RC 8.75 8.76 8.36 8.66 8.93 8.88 8.92 8.89

HCFIFODDB 8.79 8.74 8.66 8.85 8.91 8.81 8.73 8.87
Ranks

RC 6 5 8 7 1 4 2 3

HCFIFODDB 5 6 8 3 1 4 7 2

Spearman's Rank Correlation Coefficient r I 8.476
Critical value of rs at 8.85 level of sigfiificance I 8.643

 

66

Table 28. Adjjusted R3. Spearman's rank correlation
coefficient for the combined sample of the 'Iron
and Steel' and 'Automobile' industries.

 

Comparison Spearman's Rank Correlation
Coefficient ts

 

LIFO vs. FIFO: SLHC 8.785*
LIFO vs. FIFO: DDBHC 8.786*
SL vs. DDB : LIFOHC 8.862I
SL vs. DDB : FIFOHC 8.958*
RC vs. HCLIFOSL 8.786*
RC vs. HCFIFOSL 8.598

RC vs. HCLIFODDB 8.735*
RC vs. HCFIFODDB 8.476

* significant at the 8.85 level
** significant at the 8.81 level

 

TABLE 38. For the comparison LIFO vs. FIFO1SLHC. Adjusted R2.
firm rankings. and Spearman's rank correlation
coefficient for the Iron and Steel industry

 

 

Firms

Combination
of
Techniques Armco USX Lukens LTV

Adjusted R2
LIFOSLHC 8.69 8.88 8.32 8.67
FIFOSLHC 8.88 8.71 8.63 8.82

Ranks

LIFOSLHC 2 1 4 3 FIFOSLHC
2 3 4 1

Spearman's Rank Correlation Coefficient r I 8.68

Critical value of r' at 8.85 level of sigiificance I 1.88

 

Table 31. For the comparisonLIFO vs. FIFO:DDBHC. Adjusted R2.
firm rankings. and Spearman's rank correlation
coefficient for the Iron and Steel industry.

 

 

Firms
Combination
of
Techniques Armco USX Lukens LTV

Adjusted R2

LIFODDBHC 8.67 8.87 8.31 8.73
FIFODDBHC 8.79 8.74 8.66 8.85

Ranks
LIFODDBHC 3 1 4 2 FIFODDBHC
2 3 4 1

Spearman's Rank Correlation Coefficient r I ~8.48

Critical value of rs at 8.85 level of sigfiificance I 1.88

 

69

TABLE 32. For the comparison SL vs. DDB:LIFOHC.Adjusted R3.
firm rankings. and Spearman's rank correlation
coefficient for the Iron and Steel industry.

 

 

Firms
Combination
of
Techniques Armco USX Lukens LTV
Adjusted R2
SLLIFOHC 8.69 8.88 8.32 8.67
DDBLIFOHC 8.67 8.87 8.31 8.73
Ranks
SLLIFOHC 2- 1 4 3 DDBLIFOHC
3 1 4 2

Spearman's Rank Correlation Coefficient r I 8.88
Critical value of rs at 8.85 level of sigRificance I 1.88

 

TABLE 33. For the comparison SL vs. DDB:FIFOHC. Adjusted R2.
firm rankings. and Spearman's rank correlation
coefficient for the Iron and Steel industry.

 

 

Firms
Combination
of
Techniques Armco USX Lukens LTV

Adjusted R2
SLFIFOHC 8.88 8.71 8.63 8.82
DDBFIFOHC 8.79 8.74 8.66 8.85

Ranks

SLFIFOHC 2 3 4 1 DDBFIFOHC
2 3 4 1

Spearman's Rank Correlation Coefficient rs I 1.88*
* significant at 8.85

 

78

Table 34. For the comparison RC vs. HCLIFOSL. Adjusted R2.
firm rankings. and Spearman's rank correlation
coefficient for the Iron and Steel industry.

 

 

Firms
Combination
of
Techniques Armco USX Lukens LTV

Ajusted R3

RC 8.75 8.76 8.36 8.66
HCLIFOSL 8.69 8.88 8.32 8.67

Ranks
RC 2 1 4 3 HCLIFOSL
2 1 4 3

Spearman's Rank Correlation Coefficient rs I 1.88*
*significant at 8.85

 

Table 35. For the comparison RC vs. HCFIFOSL. adjusted R3.
firm rankings. and Spearman's rank correlation
coefficient for the Iron and Steel industry.

 

 

Firms

Combination
of
Techniques Armco USX Lukens LTV

Adjusted R2
RC 8.75 8.76 8.36 8.66
HCFIFOSL 8.88 8.71 8.63 8.82

Ranks

RC 2 1 4 3 HCFIFOSL
2 3 4 1

Spearman's Rank Correlation Coefficient r I 8.68

Critical value of rs at 8.85 level of sigfiificance I 1.88

 

71

Table 36. For the comparison RC vs. HCLIFODDB. Adjusted R3.
firm rankings. and Spearman's rank correlation
coefficient for the Iron and Steel industry.

 

 

Firms

Combination
of
Techniques Armco USX Lukens LTV

Adjusted R2
RC 8.75 8.76 8.36 8.66
HCLIFODDB 8.67 8.87 8.31 8.73

Ranks

RC 2 1 4 3 HCLIFODDB
3 1 4 2

Spearman's Rank Correlation Coefficient r I 8.88
Critical value of rs at 8.85 level of sigaificance I 1.88

 

Table 37. For the comparison RC vs. HCFIFODDB. Adjusted R2.
firm rankings. and Spearman's rank correlation
coefficient for the Iron and Steel industry.

 

 

Firms
Combination
of
Techniques Armco USX Lukens LTV

Adjusted R2

RC 8.75 8.76 8.36 8.66
HCLIFOSL 8.79 8.74 8.66 8.85

Ranks
RC 2 1 4 3 HCLIFOSL
2 3 4 1

Spearman's Rank Correlation Coefficient r I 8.68
Critical value of rs at 8.85 level of sigfiificance I 1.88

 

72

Table 38. Adjusted R3. Spearman's rank correlation
coefficient for the Automobile industry.

 

Comparison Spearman's Rank Correlation

Coefficient rs

 

LIFO vs. FIFO: SLNC 8.68
LIFO vs. FIFO: DDBNC . I8.48
SL vs.DDB: LIFOHC 8.88
SL vs. DDB: FIFONC 1.88*
RC vs. HCLIFOSL 1.88*
RC vs. HCFIFOSL 8.68
RC vs. HCLIFODDB 8.88
RC vs. HCFIFODDB 8.68

* significant at 8.85

 

7

3

 

 

TABLE 39. Adjusted R2. firm rankings. and Kendall's
coefficient of concordance for the Automobile
industry.

Firms

Combination

of

Tecnnifques GM FM AMC Chrysler

Adjusted R2

LIFOSLHC 8.98 8.94 8.72 8.87

LIFODDBHC 8.98 8.95 8.72 8.88

FIFOSLHC 8.92 8.82 8.73 8.87

FIFODDBHC 8.91 8.81 8.73 8.87

RC 8.93 8.88 8.92 8.89

Ranks

LIFOSLHC 1 2 4 3

LIFODDBHC 1 2 4 3

FIFOSLHC 1 3 4 2

FIFODDBHC l 4 3 2

RC 1 4 2 3

Kendall's Coefficient of Concordance W I 8.664

Value of SI83*

Critical value of S at 1.88 level of significance I 88.5
* significant at 8.81

 

75

Table 42. For the comparison SL vs DDB :LIFOHC. Adjusted R2.
firm rankings. and Spearman's rank correlation
coefficient for the Automobile industry.

 

 

Firms
Combination
of
Techniques GM FM AMC Chrysler
Adusted R2
SLLIFOHC 8.98 8.94 8.72 8.88
DDBLIFOHC 8.98 8.95 8.72 8.88
Ranks
SLLIFOHC l 2 4 3
DDBLIFOHC 1 2 4 3

Spearman's Rank Correlation Coefficient rs I 1.88*
* significant at 8.85

 

Table 43. For the comparison SL vs. DDB: FIFOHC. Adjusted R2.
firm rankings. and Spearman's rank correlation
coefficient for the Automobile industry.

 

 

Firms
Combination
of
Techniques GM FM AMC Chrysler
Adjusted R2
SLFIFOHC 8.92 8.82 8.73 8.87
DDBFIFOHC 8.91 8.81 8.73 8.87
Ranks
SLFIFOHC 1 3 4 2
DDBFIFOHC l 3 4 2

Spearman's Rank Correlation Coefficient rs I 1.88*
* significant at 8.85

 

76

Table 43. For the comparison RC vs. HCLIFOSL. Adjusted R3.
firm rankings. and Spearman's rank correlation
coefficient for the Automobile industry.

 

 

Firms
Combination
of
Techniques GM FM AMC Chrysler
Adjusted R2
RC 8.93 8.88 8.92 8.89
HCLIFOSL 8.98 8.94 8.72 8.87
Ranks
RC 1 4 2 3
HCLIFOSL 1 2 4 3

Spearman's Rank Correlation Coefficient r I 8.28
Critical value of rs at 8.85 level of sigfiificance I 1.88

 

Table 45. For the comparison RC vs. HC: FIFOSL. Adjusted R2.
firm rankings. and Spearman's rank correlation
coefficient for the Automobile industry.

 

 

Firms

Combination

of

Techniques GM FM AMC Chrysler
Adjusted R2

RC 8.93 8.88 8.92 8.89

HCFIFOSL 8.92 8.82 8.73 8.87
Ranks

RC 1 4 2 3

HCFIFOSL 1 3 4 2

Spearmanis Rank Correlation Coefficient r I 8.48
Critical value of rS at 8.85 level of significance I 1.88

 

77

Table 46. For the comparison RC vs. HC: LIFODDB. Adjusted R2.
firm rankings. and Spearman's rank correlation
coefficient for the Automobile industry.

 

 

Firms
Combination
of
Techniques GM FM AMC Chrysler
Adjusted R2
RC 8.93 8.88 8.92 8.89
HCLIFODDB 8.98 8.95 8.72 8.88
Ranks
RC 1 3 2 4
HCLIFODDB 1 2 4 3

Spearman's Rank Correlation Coefficient r I 8.48
Critical value of rs at 8.85 level of sigiificance I 1.88

 

TABLE 47. For the comparison RC vs. HC: FIFODDB. Adjusted R2.
firm rankings. and Spearman's rank correlation
coefficient for the Automobile industry.

 

 

Firms
Combination
of
Techniques GM FM AMC Chrysler
Adjusted R2
RC 8.93 8.88 8.92 8.89
HCFIFODDB 8.98 8.95 8.72 8.88
Ranks
RC 1 3 2 4
HCFIFODDB 1 3 4 2

Spearman's Rank Correlation Coefficient r I 8.28
Critical value of rs at 8.85 level of significance I 1.88

 

78

Table 48. Adjusted R3. Spearman's rank correlation
coefficient for the automobile industry

 

Comparison Spearman's Rank Correlation
Coefficient rs

 

LIFO vs. FIFO: SLNC 8.88
LIFO vs. FIFO: DDBNC 8.88
SL vs.DDB: LIFOHC 1.88*
SL vs. DDB: FIFONC 1.88*
RC vs. HCLIFOSL 8.28
RC vs. NCFIFOSL 8.48
RC vs. HCLIFODDB 8.48
RC vs. HCFIFODDB 8.28

* Significant at 8.85

 

79

Table 49. Test Criterion. firm rankings. and Kendall's
coefficient of concordance for the 'Iron
and Steel' and 'Automobile' industries.

Firms

 

Combination
of
Techniques Armco USX Lukens LTV GM FM AM Chrysler

 

Test Criterion

LIFOSLHC 8.592 8.892 8.442 8.251 8.997 8.993 8.828 8.679
LIFODDBHC 8.596 8.838 8.435 8.326 8.984 8.956 8.834 8.653
FIFOSLHC 8.936 8.771 8.848 1.858 1.868 1.128 8.717 8.534
FIFODDBHC 8.936 8.777 8.848 1.228 1.864 1.298 8.715 8.526
RC 8.514 8.784 8.428 8.443 1.827 8.888 8.987 8.825
Ranks
LIFOSLHC 6 3 7 8 1 2 4 5
LIFODDBHC 6 3 7 8 1 2 4 5
FIFOSLHC 4 6 5 3 2 l 7 8
FIFODDBHC 4 6 5 2 3 1 7 8
RC 6 5 8 7 1 3 2 4

Kendall's Coefficient of Concordance W I 8.587

Value of SI532

Value of I 17.73 * .

Critical value of at 8.82 level of significance I 16.62
* significant at a I 8.82

 

Table 58. For the comparison LIFO vs. FIFO: SLHC. Test
Criterion. firm rankings. and Spearman's rank
correlation coefficient for 'Iron and Steel' and
'Automobile' industries.

Firms

 

Combination
of
Techniques Armco USX Lukens LTV GM FM AM Chrysler

 

Test Criterion

LIFOSLHC 8.592 8.829 8.442 8.251 8.997 8.993 8.828 8.679
FIFOSLHC 8.936 8.771 8.848 1.858 1.868 1.298 8.717 8.534

Ranks

LIFOSLHC 6 3 7 8 1 2 4 5
FIFOSLHC 4 6 5 3 2 1 7 8

Spearman's Rank Correlation Coefficient r I 8.262
Critical value of rs at 8.85 level of sigfiificance I 8.643

 

Table 51. For the comparison LIFO vs. FIFO: DDBHC. Test
Criterion. firm rankings. and Spearman's rank
correlation coefficient for 'Iron and Steel' and
'Automobile'industries.

Firms

 

Combination
of
Techniques Armco USX Lukens LTV GM FM AM Chrysler

 

Test Criterion

LIFODDBHC 8.596 8.838 8.435 8.326 8.984 8.956 8.834 8.653
FIFODDBHC 8.936 8.777 8.848 1.228 1.864 1.298 8.715 8.526

Ranks
LIFODDBHC 6 3 7 8 I 2 4 5
FIFODDBHC 4 6 5 2 3 1 7 8

Spearman's Rank Correlation Coefficient r I 8.895
Critical value of rs at 0.05 level of sigfiificance = 8.643

 

81

Table 52. For the comparison SL vs. DDB: LIFOHC. Test
Criterion. firm rankings. and Spearman's rank
correlation coefficient for 'Iron and Steel' and
'Automobile' industries.

Firms

 

Combination
of

Techniques Armco USX Lukens LTV GM FM AM Chrysler

 

Test Criterion

SLLIFOHC 8.592 8.829 8.442 8.251 8.997 8.993 8.828 8.679
DDBLIFOHC 8.596 8.838 8.435 8.326 8.984 8.956 8.834 8.653

Ranks

SLLIFOHC 6 3 7 8 1 2 4 5
DDBLIFOHC 6 3 7 8 1 2 4 5

Spearman's Rank Correlation Coefficient rs I 1.888 *
* significant at 8.81

 

Table 53. For the comparison SL vs. DDB: FIFOHC. Test
Criterion. firm rankings. and Spearman's rank
correlation coefficient for 'Iron and Steel' and
'Automobile' industries.

Firms

 

Combination
of

Techniques Armco USX Lukens LTV GM FM AM Chrysler

 

Test Criterion

SLFIFOHC 8.936 8.771 8.848 1.858 1.868 1.128 8.717 8.534
DDBFIFOHC 8.936 8.777 8.848 1.228 1.864 1.298 8.715 8.526

Ranks
SLFIFOHC 4 6 5 3 2 l 7 8
DDBFIFOHC 4 6 5 3 2 l 7 8

Spearman's Rank Correlation Coefficient rs I 1.888 *
* significant at 8.81

 

82

Table 54. For the comparison RC vs. Hc: LIFOSL. Test
Criterion. firm rankings. and Spearman's rank
correlation coefficient for 'Iron and Steel' and
'Automobile' industries.

Firms

 

Combination
of
Techniques Armco USX Lukens LTV GM FM AM Chrysler

 

Test Criterion

RC 8.514 8.784 8.428 8.443 1.827 8.888 8.987 8.825

HCLIFOSL 8.392 8.829 8.442 8.251 8.997 8.993 8.828 8.679
Ranks

RC 6 5 8 7 1 3 2 4

HCLIFOSL 7 3 6 8 1 2 4 5

Spearman's Rank Correlation Coefficient rs I 8.818*
* significant at 8.81

 

Table 55. For the comparison RC vs. Hc:FIFOSL. Test Criterion.
firm rankings. and Spearman's rank correlation
coefficient for 'Iron and Steel' and 'Automobile'
industries.

Firms

 

Combination
of
Techniques Armco USX Lukens LTV GM FM AM Chrysler

 

Test Criterion

RC 8.514 8.784 8.428 8.443 1.827 8.888 8.987 8.825

HCFIFOSL 8.999 8.819 8.864 1.869 1.868 1.128 8.717 8.534
Ranks

RCr 6 5 8 7 1 3 2 4

HCFIFOSL 4 6 5 2 3 1 7 8

Spearman's Rank Correlation Coefficient r I -8.848
Critical value of rs at 8.85 level of sigfiificance I 8.643

 

83

Table
Firms

 

Combination
of
Techniques Armco USX Lukens LTV GM FM AM Chrysler

 

Test Criterion

RC 8.514 8.784 8.428 8.443 1.827 8.888 8.987 8.825

HCLIFODDB 8.596 8.838 8.435 8.326 8.984 8.956 8.834 8.653
Ranks

RC 6 S 8 7 1 3 2 4

HCLIFODDB 6 3 7 8 1 2 4 5

Spearman's Rank Correlation Coefficient rs I 8.8452*
* Significant at 8.81

 

Table 57. For the comparison RC vs. HC: FIFODDB. Test
Criterion. firm rankings. and Spearman's rank
correlation coefficient for 'Iron and Steel' and
'Automobile' industries.

Firms

 

Combination
of
Techniques Armco USX Lukens LTV GM FM AM Chrysler

 

Test Criterion

RC 8.514 8.784 8.428 8.443 1.827 8.888 8.987 8.825

HCFIFODDB 8.936 8.777 8.848 1.228 1.864 1.298 8.715 8.526
Ranks

RC 6 5 8 7 1 3 2 4

HCFIFODDB 4 6 5 2 3 l 7 8

Spearman's Rank Correlation Coefficient r I -8.848
Critical value of rs at 8.85 level of sigfiificance I 8.643

 

84

Table 58. Market power test Criterion. and Spearman's rank
correlation coefficient for 'Iron and Steel' and

'Automobile'

industries.

 

Comparison

Spearman's Rank Correlation
Coefficient rs

 

LIFO vs. FIFO: SLNC
LIFO vs. FIFO: DDBNC
SL vs.DDB: LIFOHC

SL vs.
RC vs.
RC vs.
RC vs.
RC vs.

DDB: FIFONC
HCLIFOSL
HCFIFOSL
HCLIFODDB
HCFIFODDB

* significant at .85
** significant at .81

8.262
8.895
1.88**
1.88**
8.818*
I8.848
8.845*
I8.848

 

5

 

 

Table 59. Test criterion. firm rankings. and Kendall's
coefficient of concordance for the iron and steel
industry

Firms

Combination

of

Techniques Armco USX Lukens LTV

Test Criterion

LIFOSLHC 8.592 8.829 8.442 8.251

LIFODDBHC 8.596 8.838 8.435 8.326

FIFOSLHC 8.936 8.771 8.848 1.858

FIFODDBHC 8.936 8.777 8.848 1.228

RC 8.514 8.784 8.428 8.443

was

LIFOSLHC 2 1 3 4

LIFODDBHC 2 1 3 4

FIFOSLHC 2 4 3 1

FIFODDBHC 2 4 3 1

RC 2 1 4 3

Kendall's Coefficient of Concordance W I 8.168

Value of SI21

Critical value of S at 8.85 level of significance I 62.6

 

86

Table 68. For the comparison LIFO VS.FIFO:SLHC. Test
criterion. firm rankings and Spearman's rank
correlation coefficient for the Iron and Steel

 

 

industry.

Firms

Combination

of

Techniques Armco USX Lukens LTV

Test Criterion

LIFOSLHC 8.592 8.829 8.442 8.251

FIFOSLHC 8.936 8.771 8.848 1.858
Ranks

LIFOSLHC 2 1 3 4

FIFOSLHC 2 4 3 1

Spearman's Rank Correlation Coefficient r I -8.88
Critical value of rs at 8.85 level of significance I 1.88

 

Table 61. For the comparison LIFO VS.FIFO:DDBNC. Test
criterion. firm rankings and Spearman's rank
correlation coefficient for the Iron and Steel

 

 

industry.

Firms

Combination

of

Techniques Armco USX Lukens LTV

Test Criterion

LIFODDBHC 8.596 8.838 8.435 8.326

FIFODDBHC 8.936 8.777 8.848 1.228
Ranks

LIFODDBHC 2 I 3 4

FIFODDBHC 2 4 3 1

Spearman's Rank Correlation Coefficient r I -8.88
Critical value of rs at 8.85 level of significance I 1.88

 

87

Table 62. For the comparison SL VS. DDB:LIFOHC. Test
criterion. firm rankings and Spearman's rank
correlation coefficient for the Iron and Steel

 

 

industry.

Firms

Combination

of

Techniques Armco USX Lukens LTV

Test Criterion

SLLIFOHC 8.592 8.829 8.442 8.251

DDBLIFOHC 8.596 8.838 8.435 8.326
Ranks

SLLIFOHC 2 1 3 4

DDBLIFOHC 2 1 3 4

Spearman's Rank Correlation Coefficient rs I 1.88*
* significant at .85

 

Table 63. For the camparison SL VS. DDB:FIFOHC. Test
criterion. firm rankings and Spearman's rank
correlation coefficient for the Iron and Steel

 

 

industry.

Firms

Combination

of

Techniques Armco USX Lukens LTV

Test Criterion

SLFIFOHC 8.936 8.771 8.848 1.858

DDBFIFOHC 8.936 8.777 8.848 1.228
Ranks

SLFIFOHC 2 4 3 1

DDBFIFOHC 2 4 3 1

Spearman's Rank Correlation Coefficient rs I 1.88*
* significant at 8.85

 

Table 64. For the comparison RC VS. HCLIFOSL. Test criterion.
firm rankings and Spearman's rank correlation
coefficient for the Iron and Steel industry.

 

 

Firms
Combination
of
Techniques Armco USX Lukens LTV
Test Criterion
RC 8.514 8.784 8.428 8.443
HCLIFOSL 8.592 8.829 8.442 8.251
Ranks
RC 2 1 4 3
HCLIFOSL 2 1 3 4

Spearman's Rank Correlation Coefficient r I 8.88
Critical value of rs at 0.05 level of significance - 1.00

 

Table 65. For the comparison RC vs. HCFIFOSL. Test criterion.
firm rankings and Spearman's rank correlation
coefficient for the Iron and Steel industry.

 

 

Firms
Combination
of
Techniques Armco USX Lukens LTV
Test Criterion
RC 8.514 8.784 8.428 8.443
HCFIFOSL 8.936 8.771 8.848 1.858
Ranks
RC 2 1 4 3
HCFIFOSL 2 4 3 1

Spearman's Rank Correlation Coefficient r I -8.48
Critical value of rs at 8.85 level of significance I 1.88

 

Table 66. For the comparison RC vs. HCLIFODDB. Test criterion.
firm rankings and Spearman's rank correlation
coefficient for the Iron and Steel industry.

 

 

Firms
Combination
of
Techniques Armco USX Lukens LTV
Test Criterion
RC 8.514 8.784 8.428 8.443
HCLIFODDB 8.596 8.838 8.435 8.326
Ranks
RC 2 1 4 3
HCLIFODDB 2 1 3 4

Spearman's Rank Correlation Coefficient r I 8.88
Critical value of rs at 8.85 level of significance I 1.88

 

Table 67. For the comparison RC vs. HCFIFODDB. Test criterion.
firm rankings and Spearman's rank correlation
coefficient for the Iron and Steel industry.

 

 

Firms
Combination
of»
Techniques Armco USX Lukens LTV
Test Criterion
RC 8.514 8.784 8.428 8.443
HCFIFODDB 8.936 8.777 8.848 1.228
Ranks
RC 2 1 4 3
HCFIFODDB 2 4 3 1

Spearman's Rank Correlation Coefficient r I -8.48
Critical value of rs at 8.85 level of significance I 1.88

 

 

Table 68. Market power test criterion. firm rankings
and Spearman's rank correlation coefficient for the
Iron and Steel industry.

Comparison Spearman's Rank Correlation

Coefficient rs

 

LIFO vs. FIFO: SLNC
LIFO vs. FIFO: DDBNC
SL vs.DDB: LIFOHC

SL vs.
RC vs.
RC vs.
RC vs.
RC vs.

DDB: FIFONC
HCLIFOSL
HCFIFOSL
HCLIFODDB
HCFIFODDB

* significant at 8.85

I8.88
I8.88
1.88*
1.88*
8.88
.004.
8.88
.0 .48

 

91

 

 

Table 69. Market power Test criterion. firm rankings

and Kendall's coefficient of concordancefor the

Automobile industry.

Firms
Combination
of
Techni ques GM FM AMC Chrysler
Test Criterion
.LIFOSLHC 8.997 8.993 8.828 8.679
LIFODDBHC 8.984 8.956 8.834 8.653
FIFOSLHC 1.868 1.128 8.717 8.534
FIFODDBHC 1.864 1.298 8.715 8.526
RC 1.827 8 .888 8 .987 8 .825
Ranks

LIFOSLHC 1 2 3 4
LIFODDBHC 1 2 3 4
FIFOSLHC 2 1 3 4
FIFODDBHC 2 1 3 4
RC 1 3 2 4

Kendall's Coefficient of Concordance W I 8.88

Value of SI181*

Critical value of S at 8.85 level of significance I 62.6
Critical value of S at 8.81 level of significance I 88.5

* significant a

t .01

 

92

Table 68. For the comparison LIFO vs.FIFO:SLHC. Test
criterion. firm rankings and Spearman's rank
correlation coefficient for the Automobile industry.

 

 

Firms
Combination
of
Techniques GM FM AMC Chrysler
Test Criterion
LIFOSLHC 8.997 8.993 8.828 8.679
FIFOSLHC 1.868 1.298 8.717 8.534
BANE
LIFOSLHC 1 2 3 4
FIFOSLHC 2 1 3 4

Spearman's Rank Correlation Coefficient r I 8.88
Critical value of rs at 8.85 level of sigfiificance I 1.88

 

Table 71. For the comparison LIFO vs. FIFo: DDBHC. Test
criterion. firm rankings and Spearman's rank
correlation coefficient for the Automobile industry.

 

 

Firms
Combination
of
Techniques GM FM AMC Chrysler
Test Criterion
LIFODDBHC 8.984 8.956 8.834 8.653
FIFODDBHC 1.864 1.298 8.715 8.526
Ranks
LIFODDBHC 1 2 3 4
FIFODDBHC 2 1 3 4
Spearman's Rank Correlation Coefficient r I 8.88

Critical value of rs at 0.05 level of sigfiificance = 1.00

 

93

Table 72. For the comparison SL vs. DDB: LIFOHC. Test
criterion. firm rankings and Spearman's rank
correlation coefficient for the Automobile industry.

 

 

Firms
Combination
of
Techniques GM FM AMC Chrysler
Test Criterion
SLLIFOHC 8.997 8.993 8.828 8.679
DDBLIFOHC 8.984 8.956 8.834 8.653
Ranks
SLLIFOHC 1 2 3 4
DDBLIFOHC 1 2 3 4

Spearman's Rank Correlation Coefficient rs I 1.88*
* Significant at 8.85

 

Table 73. For the comparison SL vs. DDB:FIFOHC. Test
criterion. firm rankings and Spearman's rank
correlation coefficient for the Automobile industry.

 

 

Firms
Combination
of
Techniques GM FM AMC Chrysler
Test Criterion
SLFIFOHC 1.868 1.128 8.717 8.534
DDBFIFOHC 1.864 1.298 8.715 8.526
Ranks
SLFIFOHC 2 1 3 4
DDBFIFOHC 2 1 3 4

Spearman's Rank Correlation Coefficient rs I 1.88 *
* significant at 8.85

 

94

‘Table 74. For the comparison RC vs. HCLIFOSL. Test criterion.
firm rankings and Spearman's rank correlation
coefficient for the Automobile industry.

 

 

Firms
Combination
of
Techn i ques GM FM AMC Chrysl er
Test Criterion
RC 1.827 8.888 8.987 8.825
HCLIFOSL 8.997 8.993 8.828 8.627
Ranks
RC 1 3 2 4
HCLIFOSL 1 2 3 4

Spearman's Rank Correlation Coefficient r I 8.88
Critical value of rs at 8.85 level of sigfiificance I 1.88

 

Table 75. For the comparison RC vs. HCFIFOSL. Test criterion.
firm rankings and Spearman's rank correlation
coefficient for the Automobile industry.

 

 

Firms
Combination
of
Techniques GM FM AMC Chrysler
Test Criterion
RC 1.827 8.888 8.987 8.825
HCFIFOSL 1.868 1.128 8.717 8.534
Ranks
RC 1 3 2 4
HCLIFOSL 1 2 3 4

Spearman's Rank Correlation Coefficient r I 8.48
Critical value of rs at 8.85 level of sigfiificance I 1.88

 

95

Table 76. For the comparison RC vs. HCLIFODDB. Test criterion.
firm rankings and Spearman's rank correlation
coefficient for the Automobile industry.

 

 

Firms
Combination
of
Techniques GM FM AMC Chrysler
Test Criterion
RC 1.827 8.888 8.987 8.825
HCLIFODDB 8.984 8.956 8.834 8.653
Ranks
RC 1 3 2 4
HCLIFODDB 1 2 3 4

Spearman's Rank Correlation Coefficient r I 8.88
Critical value of rs at 8.85 level of sigfiificance I 1.88

 

Table 77. For the comparison RC vs. HCFIFODDB. Test criterion.
firm rankings and Spearman's rank correlation
coefficient for the Automobile industry.

 

 

Firms
Combination
of
Techniques GM FM AMC Chrysler
Test Criterion
RC 1.827 8.888 8.987 8.825
HCFIFODDB 1.864 1.298 8.715 8.526
Ranks
RC 1 3 2 4
HCFIFODDB 2 l 3 4

Spearman's Rank Correlation Coefficient r I 8.48
Critical value of rs at 8.85 level of sigfiificance I 1.88

 

96

‘Table 78. Market Power Test criterion. firm rankings

and Spearman's rank correlation coefficient for the
Automobile industry.

 

Comparison Spearman's Rank Correlation
Coefficient rs

 

LIFO vs. FIFO: SLNC 8.88
LIFO vs. FIFO: DDBNC 8.88
SL vs.DDB: LIFOHC 1.88*
SL vs. DDB: FIFONC 1.88*
RC vs. HCLIFOSL 8.88
RC vs. HCFIFOSL 8.48
RC vs. HCLIFODDB 8.88
RC vs. HCFIFODDB 8.48

* significant at .81

 

VII CONCLUSIONS

As is analyzed in the previous section. accounting
methods do not seem to have a significant effect on the
relevant parameters of estimation. when analyzed jointly
for the complete sample of firms (results from the
analysis of variance and the Randall's coefficient of
concordance tests). From the more disaggregate pairwise
analysis with Spearman's rank correlation coefficient.
however. it is very clear that changes in certain
accounting techniques have a very significant effect on
the ranking of firms both by market power as well as by
the goodness of fit of the model. In the context of the
model developed in the study. the techniques with the most
effects are the changes in inventory measurement
techniques i.e. from LIFO to FIFO. These results are
contrary to the results obtained by Nair (197911 in that
his study did not show significant effects due to
accounting techniques.

The results of this study emphasize to the importance
of paying attention to the uderlying accounting data when
intercomparisons of a test criterion or goodness of fit
are made. Since in no case did the same accounting
technique give the best rank for all firms. it points to
the fact that intercomparisons of results when different
techniques are used by the firms can be very misleading.

97

98

This study also points out an important fact that was
overlooked in the earlier study by Hagerman and
Senbet119751 in that it is not sufficient to look at
aggreagated crossectional results when the research
question is at the disaggregate firm level. So their
research finding that on averagae most firms in an
industry use the same accounting methods is not relevant
when examining firms in industries where different
accounting methods are used. In that case intercomparisons
of estimated data without regard to the underlyling
techniques can lead to misleading analyses. However. the
question as to which techniques need to be paid attention
to depends on the model being tested and the particular
firms involved in the intercomparison. Since the model
that was developed in this study dealt with the dependent
as well as the independent variables involving measurement
of inventories. it is not surprising that in most cases
the change of LIFO to FIFO and vice versa had the most
profound effect.

The effects of techniques on the goodness of fit of
the estimated model has implications when making
intermodel comparisons as Nair did. or when testing the
validity of a particular economic model. As this study
shows. the explanatory power of the model does depend on
the particular accounting techniques underlying the data.

In this study. an attempt has been made to examine two

questions. Referring to the diagramme on page 4. the two

questions of interest are. (1) whether the measurement
model B that is based on accounting measurements captures
the same phenomenon as the unobservable model A. and (2)
whether the accounting model is sensitive to alternative
accounting methods underlying the dependent and the
independent variables.

The normative answer to the first question is that if
the economic phenomenon is to be measured by accounting
numbers. it is important to model that phenomenon of
interest in terms of the definitions that accountants
use. For example inventories should be modeled and not
assumed nonexistant if accounting measurement of the Sales
Revenue and the Cost of Goods sold is involved.

The empirical answer to the second question is that at
the micro level of an individual firm where the
differences due to accounting techniques are not
aggregated out as in the case of industry level analysis.
it is very important to be aware of the accounting
techniqes that might impact on the dependent and the

explanatory variables.

Finally this study has attempted to analyze the
effects of accounting techniques from the perspective of
impact on decision rules in the case where accounting data

is actually used. In the light of the results from this

188

study it would be worthwhile to study and develope other
models that are more suited for measurement with

accounting variables.

VIII APPENDIX

The decision problem is to maximize eqn. (7) in the text.
1....

Max P(S +1.315 - WL - c'(st,l) - cncs:t + at)
[L'IDR Us to}
t 1+1

R .—
- ”r P:(RE + 1) - p6 b.(Srt + Rt)

+ (l/l+r*) P(St+1.Q)8t+l

s.t. st + otkE + I. L. Srt + Rt) I Q + St+1

Where. Q(Kn + I. L. Srt + Rt) I Quantity of finished goods

manufactured.
Forming the Lagrangian L*.
Max L* I

Max P(St+1.5)6 - WL - cFist+li - CR(Srt + Rt)
[L.I.Rt.St+l.Ql

R
- p? PEU‘E + I) I “6 210(Stt '1' Rt)

+ (1,1...) P‘5t+1'°’5i+1

+ 9(5t + oixE + 1. L. sit + fit) - E - st+11

First order conditions are.

E: sea/86 - p . 0
i.e. Marginal Revenue I P I Marginal cost

therefore. P I MC (A.1)

181

182

si+1‘ (6P/6st+l)(o + (1/1+r*)St+1) - c'- p - 0
or. ‘6P/55i+1"6 + (1/1+r*)St+1) - cP +9 (1.2)
Here 56P/6St+l is the marginal revenue due to holding an

extra unit of finished goods inventory.

1: ITPE/upx - P6Q/6I - 0

or. pTrfiruPR . u (1.3)
where MPK I 6Q/6I is the marginal product of capital other

than raw materials.
: - cR- » b: + PéQ/SR - 0

or. (cR + » b:)/MPR - p (1.41

where MPR I 6Q/6R is the marginal product of raw

materials.
L: - W + P6Q/6L I 8
or. W/MPL I P I MC (A.5)

Where MPL I 60/6L is the marginal product of labor.

Therefore. from the first order conditions (A.1). (A.3).

(A.4). and (A.5).
P I MC I PTPfi/MPR I (CR + 9°591/"PR I W/MPL (A.6)

It is important to note here that in the model
incorporating market power. the first order condition with

respect to St+1' the finished goods ending inventory. is

183

not used. This is similar to the treatment of advertising

expense in a model proposed by Sawyer (19831Al

in which
the advertising is not a part of the final model of market
power. Another way of thinking about this is that the
marginal cost of holding finished goods inventory is a
selling cost rather than a cost of production. and

therefore is not a part of the factors of production. AC
in eqn. (3).

Multiplying equations (A.3). (A.4). and (A.5) with
K I RE + I. L. and R I Rt + Srt(raw materials available).

respectively. and adding gives.

(mx + mark + 111?me - 141.1%: + 1:66:12 + c312 + 111.

pfpfix + HobfiR + CRR + WL

 

 

or. MC .9
Q (RM?K + RMPR + LMPL)
2A PTPfix + FabfiR + ORR + WL
or . MC I 1
Q FC

Substituting for MC in eqn. (2). we get.

P+PbRR+CRR+WL
p- .0 1
0 FC

 

 

9P

184

The above equation on simplification gives.

FTPEK + yob:n + CRR + WL

FC

 

Multiplying both sides of the equation by FC/PQ and adding

1 gives.

uvrxx + P 633 + one + WL
1 + FC - 1 + £9 + ° ° '
to? PO

”7P5“ + ”.513 + CRR + WL
or. 1 I 1 + FCI 1 _ 11 +
(Q? PQ
Multiplying throughout by 1 I PQ/PQ and subtracting from

 

 

both sides (WL + CRR + bRt). the cost of goods

manufactured.

 

 

 

Po - WL - CRR - bRt prr:x uabfin bRt
. -l-- 11 + + - -—
Po 1 I re: ‘0? PQ PQ Po

(A.7)

Here Rt is the quantity. and b is the price of raw
materials actually used in production. It is important to
note that if the price of raw materials R actually used

b I b5. the actual situation will approximate a FIFO

inventory measurement technique. and likewise b x b? will
approximate LIFO. Examining the FIFO caseA3 by

sustituting b for b5. eqn. (A.7) is equivalent to.

185

prefix PAbR bRt
M13. _1__ --
Pg 1 + FCiégP 11 + PQ + P9 P9

(A.8)

 

 

In order to extend this case of a single seller to a

differentiated product. oligopoly setting.

A4

(61m + S‘(1 - «)J/Gi is substituted for 1/6Pg in the

above eqn.. which gives.

Po - WL - one - bRt

 

. <1 - PC + %) + Ll_%_51FC 5i +
i

 

 

P9 i
x
P x p bR bR
1" + ° _ __E (A.9)
P0 P0 P9

Here I make a simplifying assumption that the total raw
materials availabIe are proportional to total materials
used in production. i.e. R I th where h is a constant.
This implies that the raw materials ending inventory

Srt+l I (h - 1)Rt. or that the raw materials ending
inventory is kept at some constant multiple of the optimal
quantity of raw materials available.

While this may not be the exact scenario for every firm.
it is a plausible story for a feasible inventory policy.
Substituting th for R in eqn. (A.8) and combining the

last two terms on the right hand side.

186

 

-(1-rc+§1+u—;—°‘l’°si
i

 

’9 i
prefix bRt

* + (1111‘s - 1)__ (1.10)
PO Pa

The left hand side in the above equation is equal to gross
profit if all of the goods manufactured are sold. In
order to measure accounting gross profit which in terms of
the quantity of goods actually sold instead of quantity
of goods manufactured. both sides of the equation are
mutiplied by. (678)/(678).
Now the left hand side of eqn. (A.18) becomes.

‘_ Accounting Sales . Accounting Cost of

MIMI]: w—

Accounting Sales Revenue

 

P9
and the right hand side becomes.

1”ng bR

3+ titan-11
Q _ 0
P0

 

(1 I FC +lg) + (I I QIFC Si +
(i £1

PQ
To evaluate 878. ng 87Q. and bRt 679.

1. 0/0: Q I Q I St+1 I St

Multiplying both sides by AC

AC(Q - 8) I AC(St+1 I St)

I Change in finished goods inventories
(AI in dollars) when both inventories are
measured in current prices.

Therefore. a1 = 11c Eta/5 - 1)

187
On simplification this gives.

11 - J
0 (OI/CGS + 1)

2. PRK.6}Q:

8
Substituting for 570 from the last equation.

Pfix 15/0 1 PH

(oi/CGS + l) 0

I Amount of capital other than raw
materials (i.e. Plant and Equipment at
current prices at the beginning of the
period) to be allocated to the finished
goods actually sold. .

3. bRt in:
now , Srt + Rt I Rt + Srt+l.
‘and.Rt 670 I‘R I.qaantity of raw materials actually used
in the goods sold.
Therefore substituting for Rt and Rt 679

bR I b(Srt + Rt - Srt+l) Q/O

I (b(Srt - Srt+l) + bRtl Q/D
I (Change in value + Purchases of raw]
of raw materials materials Quantity manufactured

inventories

valued at b I bo

188

Substituting for 670. P R 5/0. and bRt 678. in eqn (A.11).
W . (1 _ so + :19, + (Lfgfcsii
i

 

Sales i
P __l_
+ TIO1.+ l 1 Sales
CGS
Change in the values raw
of raw material + material
+ (bro - l)[__.l_] [imam numbness) (It-12)
R1 Sales

cos *

1

 

IX FOOTNOTES

In 1972 and 1973. for example. Senator Phillip Hart
introduced an Industrial Reorganization Act under
which one of the indicators of monopoly power was
when a company's after tax return on stockholders'
equity exceeded 15 percent for five consecutive years
(Goldschmidt. Appendix B).

Hagerman and Senbet (19761. Nair[19791.

See Weiss (1974) for a review of 46 papers and
Benston (1973.19851 for a reveiw of about 28 more.

Concentration can be defined as the percentage of
total industry sales contributed by the largest
firms. ranked in the order of market share.

Professor Lerner proposed an index of monopoly power
as:

W

price

[here the degree of monoply power depends on the
divergence between price and marginal cost

The FOC's give.
st/BQ I P'(Q)Q + P(Q) - C'(Q) I 8
i.e. -P'(Q)Q/P I (P(Q) I C'(Q)1/P
or. -(dQ/Q)/(dP/P) I l/GQP I (P(Q) I C'(Q)1/P

a can be thought of as a kind of elasticity of

the rival's output with respect to firm's own. So
a I 1 implies that the rival firms imitate the
firm's actions in the same proportion. while u > 1
implies a greater degree of collusion (Martin119851)

Inclusion of the last period's inventory s at last
' period's price p (Qt-1.s ) will not change the
model as none of e varia8les feature in this
period's decision problem.

ti

The "one period“ model given by eqn.(6) gives the
same results as the dynamic version. This can be

189

 

18

11

12

118

seen by setting up the dynamic version of the PDV

as. a

-px‘x - 6x2 + J'(1/1.~.::'r1“’ P (s '5 )5 - w L
0 1 0 x2 t.1 t t+1' t t t t

R

' c1"stau’ ' c3“2.t+1’ ‘ Pt

R2

' ht (“2.1+1 ' (l ' °)K2.t’

here A and R are the two kinds of capital (i.e.
long lived asgets and raw materials respectively).
Collecting N and R terms inside the integration

1

and using thb'definitibhs of 6 and a' we get.
I J. t - - I -
pnv (1/1+:*1 Pt(St+1. otigt tht c’(:;+11
1
' cn‘xz.t+1’ ' ”s.tPt-1‘1.t ' “¢.tbt-1xz.t

For optimising purposes this is the same as equation'
(7). This result follows from Arrow'sil9641 Myopia
principle which states that.

III-the most striking feature of the optimal policy
is it's independence of future movements of the
profit function------.

This myopic property of the optimal capital policy
implies a considerable economy of imformation needs
in a firm's decision making process.”

(Arrow [19641.pp.27-8.)

Also when the model is developed to incorporate
imperfect markets. the test for market power
essentially is the same as the test that is developed
here. For a complete treatment see Martin (19851.

It is interesting to note here that if one was to
assume constant returns to scale. i.e. FC I 1. the
testfor market power would be of the form

a. + 'ISi > 8.

The correct definition and measurement of market share
is an unresolved issue in Industrial economics
literature. and is not in the scope of this study.

Errors in the context of the model in this study refer
to the differences that are introduced in variables
such as value of inventories. CGS. and plant and
equipment when alternative accounting techniques are
used.

13

14
15

16

17

18
19

28

111

Note that Replacement cost restatement implicitly
includes the LIFO method of inventory measurement.

See Judge pp. 515.

The error in this variable is only due to techniques
such as LIFO vs. FIFO. and HC vs. RC. This error is
equal to 'zero' if the percentage difference
introduced in the CGS is the same as the percentage
difference introduced in the ending inventory.

There is support in the literature for the existence
of constant returns to scale(see Scherer. chapter 4
for a good review).

The quarterly data filed with the SEC is peferable to
the quarterly compustat data as the latter can
introduce biases when trying to estimate

the ages of the long term assets (see Thies and
Revsinetl9771.

Source-Weiss (1974] (pp. 187).

Adjusted R2 I (1-(1-R3))(N-1)/(N-R)
where R3 I Unadjusted coefficient of the goodness of
fit of a model

N I Number of observations

K I Number of parameters estimated

This reversal of the null from the hypotheses
formulated earlier results in a loss of power of the
test statistic. However there is precedence for such
a procedure in the finance literature where the only
way to test for market efficiency is to have a null of
market efficiency (no information effect) and then
accept it if there is no information effect..

 

___9
KMP + LMP + RMP

112

FOOTNOTES APPENDIX

In the model proposed by Sawyeril983l. profits of a
firm are.

T I P(A.Q) I C(Q) I PA A. maximising the F.O.C. is.
aI/aQ I MR I MC which is equivalent to.

P(l - 1/() I MC. Assuming constant returns to scale
and substituting AC for MC in the above equation
gives. (P - AC)/PQ I 1/€ which is euqivalent to

(PO - C(Q))lPQ I 1/€. This is the Price cost margin
or the Lener index. This measure of market power
does not incorporate the advertising cost.

 

 

. 4
Total number of goods produced
assuming that each unit
produced was the marginal
unit. .

Multiplying both sides of the above equation with

1 I MC/MC gives.

Q.MC/Total cost I Q.MC/Q.AC II/FC .

R L R

For the LIFO case exactly the same results
follow except that a new implicit cost of rental for
capital consisting of raw materials is defined as.

. R n
”6' ”aha/b1

and now 5 I b is the cost of materials used in
productioh.

Same transformation as from eqn. (4)
to eqn. (5).

X BIBLIOGRAPHY

Ada-II- mut- We Nev
York. McMillan. 1970.

Ayanian. R. “Advertising and Rate of Return.” _Jgn;n.1_gfi
W (October unsung-see.

Ball R. and 6. Foster. ”Corporate Financial Reporting:
A Methodological Review of Empirical Research.“
. Supplement(1982):
161-234e

Beaver. H. 3.. and Demski. J. S. ' The Nature of Income
measurement.” 1hg_Agggnnting_Rgxgig_74(19791338-46. fl

Benston. G. J. “The Validity of Profits-Structure Studies
with Particular Reference to the FTC's Line of

Business" W (“Itch 1985):
37'67e

Bloch. a. "True Profitability Measures for Pharmaceutical
Firms.” In

lnngzlsign. edited by J. D. Cooper. pp. 147-57.
Washington. D.C.: American University. 1976.

Chambers. R. J.

W
thlxigng. Englewood Cliffs. M.J. : Prentice Hall.
1966.

Cl'tk'on- K- H- W-
Mashington D.C. : American Enterprise Institute for
Public Policy Research. 1977.

Cowling. K.. and Waterson. M. "Price Cost Margins and
Market structure." EanQ-igg 43(May 1976):267-74.

Dhaliwal. D.. Salamon. 6.. and Smith. E. D.. "The Effects
of Owner versus Management Control on the Choice of
Accounting Methods.”

EQQDQIiQi 4(July 1982): 41-53.

DoPuch. M.. and Drake. D. F. ”The Effect of Alternative
Accounting Rules for Monsubsdiary Investments."

5129133. 1255. Supplement to Jgnrngl gt Agggnnlinfl
Rgsgaggn 4: 192-219.

. and Pincus. M. "Uniformity vs. ComparabilitY'
Time Series Analysis of LIFO vs. FIFO Firms' Earnings
and Gross Margins.” Working paper. Washington
University(December 1984).

113

 

114

. and Watts. R. ”Using Time Series Models to
Assess the Significance of Accounting Changes.“

I9nrnal_9f_Assnnniin9_nsssnrsh litSpring 1972):
180-94.

 

. and Drake. D. F. ”The Effect of Alternative
Accounting Rules for Monsubsidiary InvestmentsJ

SIBQLIIL

Ennirisal_Rsssarsh_in_AssnnntinnL_Sslsstsd
1251 Suppl-Icnt to I9nrnsl_Qf_Acssuntins_Bsssarcb 4:
192- 219.

 

M..!“p Ea 0e. .nd 3.11) Pa "a

Ih1.1h19£2_lnfl_fllllfllllfln1
Berkley: University of California
Press. 1965.

 

Falkenstein. A.. and Weil. R. L. ”Replacement Cost
Accounting: What will Income Statements Based on the
SEC Disclosures Show?-Part I and II." '
(Jan-Feb 1977): 46- -57.and
(March-April):48-S7.

Fisher. F. M.. and McGowan. J. J. "On the Misuse of
Accounting Rates of Return to infer Monopoly

ProfitI-' AIsrisln_zsnn9nis_asxisx.73<March 1983):
92- 97.

Gonedes. M. J.. and Dopuch. M. “Capital Market
Equilibrium. Information Production and Selecting
Accounting Techniques: Theoretical Framework and
review of Empirical Work.“ Studies on Financial
Accounting Objectives: 1974. Supplement to lgngnnl
9f_hssnnnllnn_flslsmrsb 12: 48-169-

.__ "Economic Analyses and
Accounting Techniques: Perspective and Proposals.”

17(Autumn 1979):
384-419.

Greenball. M. "The Accuracy of Different Methods of
Accounting for Earnings: A simulation Approach."

Ianrna1.21.59999ntin9_assssrsh 6(Sprin9 1968):
114-29.

GI'Oenball. M. 'Appraising Alternative Methods of
Accounting for Accelerated Tax Depreciation: A

RIIItivc AccurIcy Appr0Ich-' Journnl_9f_hssnnntinn
figgggggn 7(Autumn 196913262-89.

n‘german. R.. and Senbet. L. ”A test of Accounting Bias
and Market Structure."
49(October 1976): 509- 13.

115

. and Zmijewski. M. “Some Economic
Determinants of Accounting Policy Choice."

W 3(August 1979): 141-61-

Harcourt./G.C. "The Accountant in a Golden Age." Oxford
Eggnglig_figpggg 17(March 1965): 66-88.

Kennedy. C. "Inflation Accounting: Retrospect and
Prams-h" W.
No. 4(1978): 59-64. Revised 1984.

 

Livingston. J. L.. and Salamon. G. L. "Relationship
Between the Accounting and the Internal Rate of
Return Measures: A synthesis and an analysis. "1921331

Wurst: Hutu-n 1970:1953-216-

Long. W. F.. and Ravenscraft. D. J. ' The Usefulness of
Accounting Profit Data: A comment on Fisher and

Macaw-m“ W311 “(June 1984):
494-511.

Martin. S. ”The Misuse of Accounting Rates of Return:

COIIInt- ”WNUWI 1984):
511- a.

”The Measurement of Profitability and the
Diagonosis of Market Power.” Working Paper: Michigan
State University. (Oct. 1985).

 

NIYIrIo J- L- W-
Boston: Allyn and Bacon. 1972

Hair. R.D. “Methods for Restating Inventory and
Depreciation Numbers.“ Working Paper No. 15.
University of Michigan (Oct. 1977)

“Economic Analyses and Accounting Techniques:

5" Empirical Study.” IQHInIl.2£.h£993fl11fl9.3£§£l££h
(Spring 1979). 225-42.

 

Paton. W. A.. and Littleton. A. C.

. New York: American
Accounting Association. 1940.

Peterson. Russell J. ”General Price-Level Impact on

Financial Information." Ihg_Aggggnting_figxigg
48(January 1973): 34-43.

Rosenfield. P. "Accounting for Inflation-A field test."

W (June 1969>=45-50-

116

Salamon. G. L. “Accounting Rates of ReturnJ 33311933
W 75 (June 1985): 495-584.

Sawyer. M. C. "On the Specification of Structure

RelltiOMhiPSJ' W 17(March
1983): 295- 386.

Scherer. F. M. 1nQHIlI1Il_HIIKIl.§1£l§1fl££.lflfl.fl£9nfllifi
_£g;fig;llngg. Boston. Houghton Mifflin Co.. 1988.

Seigel. S. ' '
fig1gnggg. York. Pa.: The Maple Press. 1956.

Simmons. J. R.. and Gray. J. "An investigation of the
Effect of Differing Accounting Frameworks on the
Prediction of Net Income.”

3311;! 44(October 1969L 757-76.

Thies. J. B.. and Revsine. L. ” Capital Expenditure Data
for Inflation Accounting Studies.” Ihg_Agann11ng.
_8121:1 (Jan. 1977): 216- 21.

Watts. R.. and Zimmerman. J. ”Towards a Positive Theory of
the Determination of Accounting Standards. '_1ng

Agggnn;1ng_figy1gg 53(January 1978):112-34.

Weiss. L. W. ”Advertising. Profits. and Corporate TaxesJ
WWWWW- 1969): 421- 30-

'The Concentration Profits Relationship in
Antitrust." in Harvey Goldschmidt et al. eds.
MW. Boston:

 

Little Brown and Co.. 1974:L 184-223.

 

116

Salamon. G. L. ”Accounting Rates of ReturnJ

_flggngn1g_flgx1gg 75 (June 1985L 495- 584.

Sawyer. M.C. ”On the Specification of Structure

Relationships.” W131 17(“33‘3'1
1983L 295-386.

Scherer. F. M. '
_Eg;‘g;|lngg. Boston. Houghton Mifflin Co.. 1988.

Seigel. S. ' '
fig1gnggg. York. Pa.: The Maple Press. 1956.

Simmons. J. R.. and Gray. J. 'An investigation of the
Effect of Differing Accounting Frameworks on the
Prediction of Net IncomeJ
3311:! 44(October 1969L 757-76.

Thies. J. B.. and Revsine. L. ” Capital Expenditure Data
for Inflation Accounting Studies.” 1hg_hgggnn11ng
_Rg¥1:1 (Jan. 1977):216-21.

Watts. R.. and Zimmerman. J. ”Towards a Positive Theory of
the Determination of Accounting Standards. '_Ih£

Agggnn11ng_flgx1gg 53(January 1978): 112- 34.

Weiss. L. W. ”Advertising. Profits. and Corporate Taxes.”

WWW- 1969)= 421-30-

"The Concentration Profits Relationship in
Antitrust.” in Harvey Goldschmidt et al. eds.
' . Boston:

 

Little Brown and Co.. 1974:L 184-223.

 

 

TY

 

”71171171111111!Ijfijitilfylflfzulmijvfyfiflfr“