Illlllll'llllll Illllllllllllllllll

THESIS 3 12930 01089 3893

   
 

1513.52.43; estate '
University

I‘ , A 'pff-Tlf‘vr

This is to certify that the
thesis entitled

An Inquiry Into the Prediction of Mergers
Using Discriminate Analysis On Financial
Ratios

presented by

David C. Distad

has been accepted towards fulfillment
of the requirements for

Ph.D. Finance

_ degree in

 

é//
Date 7M7X}

Major professor

0-7 639

 

MSU

LIBRARIES
.—;—_

 

 

 

 

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.

 

 

 

 

#3565 19;:

 

 

 

  
  
  
  
  
  
  
  
  
  
 
  

'Ht ’

AN INQUIRY INTO THE manrc'rrm OFMEGES
USING DISCRIMINANT WEBB on THE
FINAICIAL RATICB 0F ACQUIRED FMS

by
David Charles Disbad

". .'- .
'01} mm“ ‘93: L"; .'.‘ ‘- '- ,. -' x - . ->.-, in"..’"!
#1:, RN” avfi‘i‘a". r, _. . .. 3",

, ,r -; ‘ vv “”0

‘ mm C“. “‘2‘ "\“t"‘ ‘ " \r 7 ' ‘ ; _I - - 2‘)?"

_.‘_.v£lue Of the “8' "l "" ,_, 4:” . . '-,S?~~"J

Q’I‘CPa‘fifi thy)??? '74 5,12; s , l u- u- '--’ . . ”-‘y

'31” 7”. 197%.

A DWATIQW
”plea Here" n. 2'. . . ‘m-
311-1171;! to

Azmred-trm 2mm .. w?"

.‘ a1 fulfillment of the acquit-cents
M from the marsh mutter . {woe

V .V‘ l':( 1' yr
~ LAW-143' ‘

 

 

 

ABSTRACT
AN INQUIRY INTO 'JIHE PREDICTION

OF lemmas USING DISCRIMINANT
ANALYSIS 0N FINAICIAL RATIOS

.,~< by

David Charles Distad

   
  
  
  
    
     
 
  
  
  

:1”; F11»
Wombat the history of mergers in the United States, share

5‘ cs3praniuns awarded to existing shareholders have exceeded market

1 .31‘ :38. Because of the contradiction between share praniuns and the
i 1 “' "i 1".

M”sting value of the acquisition and because of the increasing snphasis

.1.“ .b 1

k‘ corporate growth through acquisition, it is appropriate to study
‘ m the 19703.
‘ "re "
. ”Wanna were extracted from the Compustat data bases. may
.. t1; rid-hmrdred-thirty—five firms acquired in the period 1970-
‘211 lute iron the same time span for three-Inmdred-twenty-three

153-. ‘

silt-figured as of July, 1981. 3. Fifty films acquired in 1979.

Wred-forty—flve unacquired firms as a control group for

A A gent"

My addressed financial characteristics presuming a
‘Zasqutsition's attributes are reflected in its financial

 

 

 

David (harles Distad
statements.

his study incorporated financial information to three years
preceding acquisition.

'Nenty types of accounting information were used to calculate
thirty-seven financial ratios for each firm in the first two samples.
'mose ratios were used to calculate three discriminant functions.
Multivariate discriminant analysis (MDA), an acceptable statistical
procedure for these applications was used.

Next, the nunber of ratios used in the discriminant functions was
reduced, but attempts to reduce the nunber of ratios below thirteen
impaired the models' ability to classify firms correctly.

All three models separated firms better than by an other process
at high statistical levels of significance in the ex post tests.

Acquired firms tended to have greater than average operating
incane profitability ratios, but lower than average after-tax ratios.
mey were slightly more levered but, cash to interest expenses was more
important than any of the leverage ratios. Liquidity and asset activity
ratios here generally insignificant.

The MBA models were used ex—ante to predict merged firms fran
those fifty firms fran Sample 3. Preliminary evidence suggests ratios
canmt be used to predict mergers, however, prediction success rates are
highest using financial information three years prior to a finn's
acquisition.

Finally, equal sample sizes fran wples three and four produce
high statistically significant positive results for the ex ante model.

 

 

 

 

ACKNOWIEM’IEITS

I wish to thank the members of 11w thesis committee, Professors
Allen Grunewald, Myles Delano and Jeffrey Towle for their contributions
to this research. I especially wish to thank Professor Grunewald whose
interest, encouragement, and enthusiasm were of great help to me.

I also wish to thank Dr. David Gabhart at Michigan State
University who was very helpful to me in the initial stages of this
study.

Much of the research was canpleted at California State University
at Hayward. There, I am indebted to Peter Chamberlin, the computer
applications faculty liaison, who wrote my computer programs. Also at
Hayward, I would like to thank Professor Craig Johnson, who reviewed
early versions of the first three chapters. I also wish to thank the
administration at (BUH for various financial awards to help canplete
this.

Most of this paper was written when I was a visiting lecturer at
the University of Michigan. I wish to thank the administration of their
graduate school of business for their support, and there I wish to thank
Cheryl Strickland who typed most of the first manuscript.

Finally, I wish to thank my wife Cynthia for her support over these
manymonths, an! to my children, Eric and Annie, whose presence provided
the hunor and the sense of urgency required to canplete something of this
scale.

11

TABLE OF CONTENTS

 

 

page

1 ListofTables.......... ........... ...vi
'1 Introduction. . . . . . . ....... . ....... . . . 1
Chapter I: Hypotheses and Their Significance . . . . . . . . 3

Chapter II: he Theory Underlying The Hypotheses . ..... 6

A. Managerial Theory . . . ....... . . . ..... 6

B. Monopolistic Theory . . . . ......... . . . . 6

C. Efficiency Considerations and Synergy . . . ..... 7

1. Operating Effects ..... . . . ........ 7
2.MarketShareEffects.. ...... 9

3- Financial Effects . ........ . . . . . . . 10

a. CorporateDethapacity...........10

b. Undervalued Assets — Tobin's q ..... . .

c. Accounting Treaiment of Mergers . ......

d. Price-EarningsRatios........... 16
4-ConglanerateEffects...............l7
a. Foreign Acquisitions. . . . . . . . . . . . .

D. Managerial Theory - A Review of the Literature. . . . 20

c

o
[\J
U'I

E. Siareholder Returns - A Review of the Literature.
F. COHClUSiOIl. o c I o c u o c o ...... o o o o o c 31
ChapterIII:'JhePlanofResearch.............. ’40

 

A. Rationale Fbr The Use of Financial Ratios . . . . . . ”0

iii

   

t

0
I (ma III (Contimxad) page I
RatiosIncluiedInThisStmy............"2 ?
as f
I

Etionale For he the of A Discriminant Emotion.
" Statistical Limitations or This Stuiy . . . . . . . . “3

   
   
 

mtionale for a Dimension Redmtion . . . . . . . . . "9 ;
anter Iv: Data AnalysisandSanple Design. . . . . . . . . 54

A.IhtaSelection....................5’4

D. CharacteristicsoftheData............. 69 T
‘Glapterv: FindingsoftheStuiy.............. 71

’A.‘ Distad's Reduced Dimension Model Using Discriminant i
Analysis.......................71
B. Distad's Reduced Dimension Model Using Rector
C. Analysis of Stevens' Factor Analysis Procedures . . . 82
”- Distad—Stevens Comprisons of Results Using Easter
.‘9Analysis.......................93 E:
WE RepliestionoftheStevmsl/bdels.........: 95
A ‘u‘ir. Analysis of the Simkowitz-Monroel‘liodel. . . . . . . .111 ‘
Riff-G. 00nc1usion......................1114 a
VI: Validation of Mtivariate Discriminant 1;.
Malysis....................118 I

 

Jill: Implications of ihis deel and Its Uses . . . . 123

1v

 

    
  
  
  
  
  
  
  
   
   
  
  
  

f ‘T
Q
W
. YW-
.“ ‘ ‘Hr' adapt, or VII (Continued) page
2. Dista's was C O O I U 0 C I O I O O I I I O I 125 ‘
BC 001101151011. 0 o o o o o o o o o I o I o o o o o o l I 135 i
“ ~ ‘ mapter VIII: Conclusions ard Evaluations, Limitations and "
Future ksearch Suggestions. . . . . . . . . .138 A
A. StmaryoffllleResultsofThisStlfly.........138 !
B. PredictionofMergers . . . . . . . . . . . . . . . .1"?
C. Confirmation of Ratio mpotheses. . . . . . . . . . . 153 t

D. Statistical Significance of the Models. . . . . . . . 157 I.

E.DimensionReduction.................157 A;
1.FactorAnalysis.................158 ‘
if}. A 2. Sequential Univariate Analysis. . . . . . . . . .158 2
r.RetioSeleotionorer-Time..............159 ‘
~I}.Idmitetions.....................160 i
Futm'eResearchSuggestions.............150

 

 

LIST OF TABIES

RatiosSttfiied..............
Additional htios Studied . . . . . . . .
Fisher‘s Boa: M Test for Equal Covariances
Classification Results — Hold Olt Group,
Split Sanple Validation, LAG 2. . . . . .
Classification Resflts - Hold Olt Group,
Split Sanple Analysis and Validation,
IAG1
Classification Results - Hold Out Group,
Split Sample Analysis and Validation,
IAGO
Descriptive Statistics, Method Direct,
Standardized Coefficients . . . . . . . .
Sine (tiaracteristies Comparisons, Merged
ardlbn—mergedFirms...........
Mined Dimension Descriptive Statistics,
Standamdized Coefficients . . . . . . . .

    

Dis-tars Redmed Dimension Discriminant

groutinmezn...........
,_‘ QMM'ES Mmed Dimension Discriminant
it‘mon,ne1.............
. l . ~w3'sfbdmed Emulsion Discriminant
vi

Page

”3

1:5

60

61

65

66

66

68

69

72

73

7a

 

   

5—3 (a)

-

5-3 (b)

5-3 (c)

Fianctim,IAGO..............
Factor Analysis, Eigenvalms, Percent of
Variame,IAGO..............
Easter Analysis, Eigenvalues, Percent of
Variance,IAG1..............
Factor Analysis, Eigenvalues, Percent of
Variance,IAG2..............
First Factor Factor Ioadings, IAG O . . . .
Stevens' Factor - Percent of Variance
'Jhble.
Distad's 19705 {Best Factor loadings Per

FinanciallhtioByBactor,IAGO . . . . .

Dietad's 197% Test Factor Loading Per
Fimncial Iatio DyFactor, IAG1 . . . . .
Distad's 1970s 'Iest Factor Ioadirm Per
Financial Ratio HyFactor, IAGZ . . . . .
Stevens' Factor Characteristics Compared to
Distad's Factor Characteristics, LAG O. . .
Final MDA ani Factor Analysis Key Ibtios
Compared, (Fran Gables 5-1 and 5-6), IAG o.
EDA: Classificatim Matrix, Cbserved

I'W30.000looooeoooooooo

Distal Ep‘lication of Stevms' Coefficients

Shevm' New Classification of Pbrger Groups.

_. Wim'efflvpe I ani II Error levels .
Mention lattices for,_Stevens'

,4?

'- Ir'; .1
.3. P411, . V11
‘ :3: V

Page
75

78

79

80
81

8h

86

89

91

93

91!

96

97

102
103

 

}' fbfomulatedrlbdel...............

I ', 5—16 Stevens' Coefficient Comparison . . . . . . . .

. 5-17 81:13va Model vs. Distad's Replication

Results

6-1 (a) Distad's 1970 Data Test: Univariate Tests of

l \ ' Significance on m Selected Key Financial

l'r'. Ratios,IAG2.................

1 "‘L 6-1 (b) Distad‘s 1970 Data 'Dest: Univariate ’Uests of

" ZR"- ‘ Significance on MDA Selected Key Financial

 

  
  
  
  
  
  
  
  
  
  
  
  
  
  
  
  
  
 

Ratios,IAG1.................
8241 (c) Distad's 1970 Data Test: Univariate Tests of
'I ‘ l Significance on IVHJA Selected Key Financial
Ratios,IAGO.................
6-2 Key Ratios Selected Using Univariate F {Bests
Significantat.05 ..............
7-1 (a) Success Rates From Ex Ante Prediction
Models - Stevens' Original Model. . . . . . . .
7-1 (19) Success Rates From n: Ante Prediction
Models - Stevens' Rsfomulated Model.

: 7—1 (c) Success Fates From Ex Ante Rediction
I Lbdels-Stevens' Article Model . . . . . . . .
Success Rates Ran E: Ante Prediction
‘ Models - Distad's Models - Reduced Dimension. .
g 1(9) Success Rates Iii-cm Ex Ante Prediction
Models-MethodDirect.............

3‘ gl E; into classification Resnts, Distad's Reduced
4: , . viii

  
  
  
  
  
  
  
  
   
  
  
  
  

DimensionModel,IAG2,1976............130
7-2 (b) Ex Ante Classification Results, Distad's Reduced
DimensionModel, IAG1,19'T7. . . . . . . . . . . . 131
7-2 (c) Ex Ante Classification Results, Distad's Reduced
DimensionModel,IAGO,1978. . . . . . . . . . . . 132
7-3 Bypofiietical Classification thtrix, LAG 2, i!
,. Asauning a 94.6% Maximun Chance Rate, and a 1005 1‘
Success Rate Predicting Nbrged Firms. . . . . . . . 131l '

7-4 Hypothetical Enlarged Sample Classification
mun, IAG 2 U C I O O U I O I I O O O O O O O I O 135

1
I

1

1 1
l

881 (a) Final. Classification thrices, LAG 2. . . . . . . . 1119
' 8-1-1 (11) Final Classification Vatrices, IAG 1. . . . . . . . 150
'- ‘\ 5—.1 (c) Final Classification tatrioes, IAG 0. . . . . . . . 150

ix

 

 

 

 

INTRODUC TION

Elle purpose of this study is to generate a set of models which will
predict mergers. ’Jhis will be accanplished using publicly available
data to develop a financial profile of firms more likely to be acquired
in subsequent time periods.

The financial profile will be developed using multi—variate
discriminant analysis (MDA) . In previous studies researchers have
found that MDA provides better predictors for such events as
banln'uptcies, credit rating, bond ratings, and mergers than the
analysis of individual financial ratios analyzed sequentially, in other
words, univariate analysis.

Throughout the history of mergers in the United States, share
price premiums awarded to existirg shareholders of the cannon stocks of
target firms have been large, usually in excess of 25% [1 ,4,22].

In recent years that preniun has increased particularly when a
target firm has been the object of a bidding contest between two or more
potential acquirers [51]. Moreover, a dramatic increase in the nunber
of mergers has been occurring. Finally, the nunber of large firms,
whether measured by sales or by total assets is increasing [51].

Because of the apparent contradiction between share premiuns of
acquisitions and the theoretical value of the firm, and because of the
increased anphasis on corporate growth by acquisition in the last twenty
years, it is appropriate to conduct a study of mergers. Any technique

1

 

2
able to predict mergers will be of great benefit to managanents either as

an invesiment aid in identifying potential acquisitions, or to assist a
firm in exposing its weaknesses in sufficient time to correct the
deficiencies am prepare a defense against its future acquisition. In
the event that merger candidates can be identified prior to their share

price appreciation, rewards to investors would be of considerable

 

magnitude. Perhaps of as much importance such a predicting model would
have implications for the existing literature on the seni—strong form of
the efficient markets hypothesis. Toe ability to predict mergers would
open the possibility of risk arbitrage and risk adjusted excess returns.
Finally, models of this nature maybe useful to regulators in anti-trust
litigation because of the information provided regarding the capital
structures, profit margins, and rates of return on equity or assets

employed of both acquirers and acquisitions. Any contributions of this

 

nature would in turn be of benefit to the public.

 

 

 

 

 

CHAPTER I
HYPOTHESFS AND THEIR SIGNH‘ICANCE

This dissertation defines a set of ratios which is useful in
identifying those firms most likely to be acquired at sane future time.

Previous studies [118,135] have not always provided clear results.
Different studies have produced contradicting sets of most important
ratios.1 Furthermore, the possibility may exist that the set of most
predictive ratios can change over time, independent of econcmic
factors.2 One critic suggests the results of one test had substantially
less financial significance than statistical signficance.3

Much of the literature and empirical work on mergers was done
during tin 1960s. Enpirical studies were generally characterized by
anall sample sizes. There have also been changes in accounting rules
regarding mergers. Acquiring firms, prior to the enactnent of AFB 15,
were not required to calculate a fully diluted earnings per share
reflecting the coumon stock equivalents (convertible preferred and
convertible debentures) so commonly used in the 1960s to finance
mergers. Iewellen,4 Nielsen and Mel icher5 and others6 believe that
merger preniuns and valuations have changed significantly in the ensuing
years. Finally, it appears that with the exception of Singh's [20]
study of mergers in the United Kingdom, no studies on industry
1Footnotes appear at the end of each chapter.

3

 

 

 

 

M
characteristics of acquired firms have been conducted. CL‘ne pertinent

questions are: 1 . "Are there characteristics prevalent within certain
industries which make them more susceptible to takeover?", and 2. "Can
these characteristics be identified using this statistical methodolog
on financial ratios?"

The intent of this study is to provide timely insights into these
issues and to resolve confl icts where possible. Ellie methodolog of this
study extends the prior literature in that new data, much larger sample
sizes, and new ratios are enployed. Fran this, new evidence will
develop as to which ratios are most important in predicting mergers, is
there any continuity of a best set of ratios over different time spans,
and finally, are there any identifiable industry characteristics?

Evidence fran earlier literature, to be discussed in the next
chapter, suggests that acquired firms are likely to have excess
liquidity, unutilized debt capacity, possess favorable earnings
prospects, and tend to be priced at lower levels relative to their book
value per share, than are unacquired firms. Another null hypothesis in
this study is that an MDA designed ratio analysis can not assist in the

prediction of mergers.

 

 

 

Footnotes
Chapter I

For exanple consider the conflicting evidence as to the importance
of price-earnings ratios and dividend payout percentages found in
Michael Simkowitz and Robert Monroe, "A Discriminant Analysis
Fimction For Conglanerate Targets, " Southern Journal of Business
(Novenber, 1971): 13, and Donald Stev "Financial
Characteristics of Merged Firms, A Multivariate Analysis," Journal
of Financial and Quantitative Analysis 8 (Nb.rch,19T5):1SZ.

Donald L. Stevens, "A Multivariate Analysis of Financial
Characteristics of Acquired Firms in Industrial Mergers, " (Ph. D.
dissertation, Michigan State University, 1972)' 116.

Robert J. Monroe, "Comment: Financial Characteristics of Phrged
Firms: A Multivariate Analysis," Journal of Financial and
Quantitative Analysis 8 (March, 1973): 164.

Wilbur G. Iewellen, "A Pure Financial Rationale for the Conglanerate
Merger," Journal of Finance 26 (Play, 1971): 523.

Janes F. Nielsen and Ronald W. Melicher, "Financial Analysis of
Acquisition and Merger Premiuns," Journal of Financial and
Quantitative Analysis 8 (March, 1973): 139, 146

'ihree articles are of interest here. They are:

H. Kent Bker, Thanas 0. Miller, and Brian J. Ransparger, "An Inside
Look)at Corporate Mergers and Acquisitions," IVSU Topics 829 (Winter,
1%1 4967.

Iewis Bernard, "What We Learned Fran 'Ihe Great Merger Frenzy,"
Fortune, April, 1973. PP. 70-73, 144, 148—150.

"Ira Harris: Chicago's Big Dealmaker," Business Week, June 25, 1979,
m' 70.78-

 

 

 

CHAPTER II
THE 1mm UNDERLYING THE HYPOTHEBES

There are no universally accepted reasons why firms acquire other
firms. The reasons are not always obvious because desired objectives
are often unfulfilled, or because the premiun paid for the acquisition
exceeded the desired gains.ll

However, separating mergers according to general theories of
mergers is useful in explaining the process. There are three motives
for a merger most commonly described in the literature: [1, 3].

A. Ma__n_a&rial Theory2

This theory holds that

. . . the interests of the stockholders are
subordinated to the managers who control firms and who seek
growth maximization which in turn maximizes the salaries and
enolments of managers. This managerial theory predicts

that merger activity should have. an unfavorable impact 3on

the market values of the securltles of acquiring firms.

If the managerial theory has validity it has mad or ramifications
in merger literature. Because of its importance an extensive review of
the literature is provided in a later section of this chapter for each
theory.

B. Monopglistic Theog

This theory predicts that either or both of the acquiring and
acquired firms should benefit from mergers. Here the maj or issues are

6

 

 

 

 

whether the public is served econanically and socially, and do the
pranised benefits accrue to both of the firms?

C. Efficiency Considerations and Synergy

 

Generally, any canbination of assets producing a new firm
whose value is greater than the sum of the values of the pre—merged firms
produces a synergy, which may be attributable to any nunber of
considerations including portfolio effects.

These considerations may be categorized as:

1. Operating Effects
2. Market Share Effects
3. Financial Effects
4. Conglanerate Effects

1. firatig effects

The most commonly cited motive is that a merger will produce a
"fit" between two canpanies. An example might be two similar firms; the
first has strong production capabilities; the second has strong research
capabilities. The canbination of the two firms produces a canpany
strong in both areas, and hence potentially stronger than the sum of its
parts.

Another operating effect involves the use of factors of
production. For example, in manufacturing operations, larger firms
produce longer, more efficient production, and increasing returns to
scale.

While both of these exanples are forms of horizontal mergers,
operational efficiencies also are found in vertical mergers. Vertical
mergers enable canpnies to acquire others at different levels in the

distribution chain of a given iniustry. An example is an oil refinery

 

 

8
acquiring a chain of service stations, thereby extending its control
over the flow of a product to the consuner.

Two other operational effects, management and tax considerations
are included in Alberts' discussion of bargains which encourage merger
activity. rihey are poor forecasts by existing management and inaccurate
cost of capital measurenents. Alberta argues the ability to use
operational effects in bargaining may affect the price of the
acquisition, particularly if the shares are thinly traded.4

Whmgenent bargains abound because, in the absence of any other
information, firms are valued under the expectation that the existing
management will remain in control. If the managenent of firm X is
superior to the managenent of firm Y, then Y' s efficiency should improve
after it is acquired by X, which in fact was observed by Mandelker.‘5
Copeland and Weston6 and Iewellen7 suggest that firms most likely to be
identified as misrnanaganent bargains would be within the same industry,
because managers there would be more capable of detecting less-than—full
performance by the other firm's managenent. Misnanaganent bargains are
not restricted to horizontal mergers; in practice, they may be vertical
or conglomerate mergers.

Two other operationm. effects are tax effects of mergers and the
value of accunulated tax losses. With regard to the former :8

One such tax consideration is to substitute capital

gains taxes for ordinary incane taxes by acquiring a growth

firm with a snall or no dividend payout and then selling it

to recognize capital gains. Also, when the growth of a firm

has slowed so that earnings retention cannot be justified to

the Internal Revenue Service, an incentive for sale to

another firm is created. Rather than pay out future

earnings as dividends subject to the ordinary personal
incane tax, future earnings can be capitalized in a sale to

another firm. Most substantial mergers are tax-free
exchanges. Not only is a lower capital gains tax

 

 

 

 

9

applicable, but it is also postponed until the securities
received in the tax-free exchange are liquidated for cash.

(lice the assumptions of perfect capital markets are lifted,
individuals or firms may have difficulty selling their tax losses in the
marketplace.

If losses are not salable in the market, then a merger

allows a firm with substantial losses to benefit immediately

fran the loss (by cancelling the loss against the profits 0

a profitable firm with which the losing firm has merged) .~

2. Market Share and Market Power Effects

These motives for mergers are not as clearly defensible as
operating effects, for a nunber of reasons. Manne, in citing Supreme
Court Ruling, questions any merger between carpeting firms, suggesting
they are " . . . at least suspect and perhaps per se illegal. The latter
result seens especially likely when one of the canbining firms already
occupies a substantial position in the relevant market."10 He concedes
an econanic interpretation of the failing firm defense is that a firm's
tendency toward bankruptcy suggests it is no longer a canpetitor. A
failing firm defense of a merger is, " . . . a civilized alternative to
bankruptcy . . . that transfers assets from falling to rising firms."11

The consensus of the courts and many econanists is, " . . . there
are no important econanies of scale [that] can be attained through a
[horizontal] merger which cannot be gained either by internal growth or,
at worst, by a cartel, if that were legal."12

Furthermore:

Takeovers of corporations are too expensive generally

to mks the 'purchase' of managenent canpensation an

attractive proposition. It is far more likely that a second

kind of reward provides the primary motivation for most

take—over attempts. The market price of shares does more

than measure the price at which the normal canpensation of

executives can be sold to new individuals. mare price
. . . measures the potential gain inherent in the cannon

 

 

10

stock. 'Jhe lower the stock price, relative to what it could

be with more efficient managenent, the more attractive the

takeover becomes . . . and the potential return from

successful takeovers and revitalization of a poorly run

canpnycanbeenormws. . . . we cansee how. . . taking

over control of badly run corporations is one of the most

impor t 'get rich quick' opportunities in our econany

today.

3. Financial Effects of ‘Mergrs

a. Comrate Debt Capacity
The financial consideration most conmonly cited for

mergers involves the alteration of an acquiring firm's existing debt
structure and the resulting reduced probability of bankruptcy.
Lintner14 found mergers beneficial for four reasons: (1) reduction of
the lender's risk in bankruptcy losses, (2) reduction of scale
diseconcmies in credit investigations of what was formerly a snaller
finn, (3) lowering flotation costs of public issues enhancing their
marketability, and (4) producing combined larger issues of debt.14

Elidence exists that one of the most important factors influencing
the quality of a bond rating is the size of the offering,15 which
provides sane support for Lintner's fourth rationale for mergers.
Concurring, Iewellen believed the increased size of the combined firm,
in conjunction with a reduced joint probability of failure, would create
additional debt capacity, and lower lenders' risks. {The latter should
result in lower interest rates, assuning that ratings agencies perceive
the reduction in risk.16
Copeland and Weston attack Iewellen's logic:

While reducing the unused debt capacity will increase
the value of the firm, it has not been established that a
merger is the necessary and only method capable of bringing
about this method. [the firm with the unused debt capacity

is perfectly able to increase the anount of its borrowing
without the merger. 7

 

 

ll
Replying to Copeland and Weston, Ihley and Schall contend:
If the merger reduces the variability (uncertainty) of

firm cash flow, the newly merged firm may wish to raise its

total debt above the total debt of the urmerged firms. 'Ihe

merged firm will set its debt at the level which maximizes

firm value, and this level of debt will pgten be higher than

the total debt of the premerger firms.

Copeland and Weston also fail to discuss excessively leveraged
firms. As Stevens indicates:

. . . in the case of the acquiring firm with excess
debt, the merger would lower financial risk and move
the new £11161 in the direction of an optimun capital
structure.
b. Undervalued Assets - Tobin's q
Mergers are often undertaken because a potential acquirer
believes the market misvalues the assets of a possible acquisition. 'Ihe
hypothesized misvaluation may be due to unusually large anounts of cash,
unutilized debt capacity, or future cash inflows estimates perceived by
the potential acquirer to be different than market estimates.

Ratios should assist the potential acquirer in the screening
process to find those firms possessing the above attributes or any other
quantitative attributes deaned to be important by the acquirer.

In the last few years more emphasis has been placed on the
possibility of asset misvalmtions. 'Ihe misvaluation contention is
that the acquirer will be \mable to build new facilities, whether it be
an asset or an entire canpany for a canparable price; that the market
V9113 of an acquisition is less than its replacanent cost.

Any mergers stuiy should address the misvaluation problen, and
this stuiy has done so, by testing thirty seven ratios assessing
liquidity, leverage, asset managenent, and profitability. Incluled in

them ratios in Tables 3-1 and 3-2 is fixed asset turnover which is

 

 

l2
discussed in greater detail in Chapter 5. Another possible ratio for

measuring misvalued assets, Tobin's q, was considered but not included.
Because of its wide acceptance, the decision not to use that ratio should
be explained.

To begin, valuation disequilibriuns are illogical to financial
theorists who believe the market pricing mechanisn should restore an
equilibriun to market values and replacanent values. 'lhere are many
reasons for a misvaluation of physical assets ranging fron the obvious
potential of obsolescence to the less obvious discount rate used to
value the increnental cash flows of that asset by a host of potential
acquirers possessing varying risk levels, to the range of values from
going concern values to liquidating values. A theorist will contend the
above three reasons ought to apply to equity share values as well, but
will intuitively concede a misvaluation is more likely to exist in the
appraisal of a partially depreciated physical asset than for an equity
share actively traded in a secondary auction market. Such a potential
misvaluation introduces the availability of risk adjusted excess returns
to the arbitraguer.

James Tobin developed a ratio to measure the discremncy between
market values and reproduction values of assets which is now referred to
as the q ratio. 'ihe mnerator of his ratio is the market valuation, the
prevailing market price for exchanging existing assets. His
denaninator is the replacement or reproduction cost, the market price
for newly produced assets.20

Tobin believed that disequilibriuns existed for more than brief
time periods for two reasons: 1. in exanples of improvenents on real

property a time lag exists because of construction time, and 2.

 

 

 

13
valuations of existing assets will be more volatile than for the price of

21 However, ultimately the nunerator and denoninator

reproductions.
will be brought into equilibriun. Indeed, a market value greater than
replacenent value provides an incentive to create more of the asset in
pursuit of an economic profit which will persist until either market
values fall, reproduction costs rise, or both.

The significance of q to the merger movenent is not clear because
its enpirical test are inconclusive, in spite of the contentions of its
advocates.

'mroughout the 19605, the ratio was greater than one, based on
Tobin and Brainard's original enpirical stuiies.22 Q rose from 2.21 in
1960 to 2.54 in 1968, dropped to 2.12 in 1969, and plunged to 0.97 in
1974. Q levels higher than one are expected to stimulate investment and
in response, the 19605 are now characterized as the decade of
conglonerate mergers.

Tobin attributed the drop in q from 1973 to 1974 as being due to a
spectacular rise in the discount rate applied to earnings brought about
by tight anti-inflationary monetary policies rather than to declines in
earnings.23 'ihe drop in the value of q also coincided with the end of
the great merger era.

Q fell to .8 in 1974-75 and according to one researcher is now at
.73 in spite of the record levels of merger activity.24 If a high q is
expected to stimulate investment and produces record levels of mergers,
ought not we to expect an absence of mergers if q is at levels below 1,
given the disincentive to invest?

According to Ciccolo, q levels less than one, accanpanied by low
prices of equities, make it cheaper for firms to acquire other firm,

 

 

lu
invest in treasury bills, or increase cash dividends rather than make

new capital invesirnents.25 He contends it is the low level of q
currently that accounts for the record merger activity which began in
the latter half of the 197%. Ciccolo's theory does not explain the
high q levels associated with the merger experience of the 19605, nor
does he define low stock prices or when they cease being "low."

Other enpirical studies of q produce conflicting results. Von
Furstenberg, Malkiel,ani Watson [127] concluded q was a powerful
influence on levels of investment. Fbwever, testing for q at the firm
level, rather than in the aggregate, Chappel and Cheng produced
contradictory results, generally as a function of the industry in which
a canpamr functioned.26

'ihere also appears to be a measurement problem in quantifying q.
Ciccolo calculated q to be 1.65 in 1965,27 well below the 2.5 calculated
by Tobin.28 At the sane time, the Council of Econanic Advisers which
also calculates q, determined its level to be 1.25.29

In 1974, Tobin calculated q to be 0.97,30 Ciccolo calculated it to
be 0.8,31 and the Council of Fconanic Advisers determined q to be
0.663.32

Because Ciccolo contends that at any time span q levels vary from
firm to firm, it would seen that care must be taken in selecting a
"repeaentative" sanple.

Furthermore, the potential for misvaluation exists in spite of
sample selection. Because Tobin and Brainard used a set of valuation
assunptions, those assumptions have been followed in subsequent tests,
doubtlessly to pranote "canparability." Tobin's q ratio nunerator, in
alpirical tests, is not the market value of assets, it is the market

 

 

 

15

value of the firm's cannon stock, preferred stock and long term debt.
Ironically, the first step in the testing process to measure
disequilibrium of asset values and stock values is to assune that they
are in equilibriun. It is difficult to test for misvalmtions by
presuning there are none. Because the available data bases do not
provide market values for bonds and preferred stocks, their market
values must be estimated. Canpanies are assuned to have Pea twenty year
maturity long term debt (Chappel and Cheng assumed twenty years, but
matched yields according to ratings classes),33 and their preferred
stock is valued at the year and Standard and Poor index of preferred
stock yield, regardless of the firm's risk level.34 Those estimates do
not seen critical however, when compared to the potential errors from
estimations used in the denaninator. ’I'he denaninator, replacenent
cost, " . . . is the sum of the book values of canmon stock, preferred
stock, and long term debt, connected by a cannon annual index of the
ratio of replacement cost to book value."35

'Ihe replacement cost of assets is an attempt to adjust for
depreciable assets at the rate of five percent per year as opposed to
accounting depreciation. It is also to revalue inventories at
replacement cost and adjust for inflation using the price deflator for
the fixed investment canponent of GNP. Strangely, the category "other
assets" is left at book value. Unfortunately, "other assets" incluies
marry of the potential misvaluations, especially land and intangible
assets.36

Perhaps most puzzling is Ciccolo' s contention that individual
firms have varying q ratios}.7 That implies varying degrees of
disequilibriun at the firm level, and more specifically that those firms

 

b0.

M ,

16
are not being valued at the correct discount rate. Both investment

professionals and the acadenic community must be interested in his
perceptions of the correct discount rates for various firms.

To conclude, intuitively the prospect of some assets beirg
misvalued by potential acquirers are more probable than for other
assets. Bit anpirical tests to date do not clearly demonstrate that
entire firms are being valued incorrectly given their existing
managenent and current econanic states of nature.

c. Accountirg Treatment of Mergers

The accounting treatment of mergers, which changed in the
1970's has probably lost its presumed sign.ificance.38r3§9 A stuiy
conducted by Hong, Kaplan and Mandelker indicates the pooling-of-
interest method of accounting for mergers did not lead to abnormal stock
returns.40 Mthemore, they contend that the efficient capital market
should be able to see through the particular accounting convention used
to describe amerger and respond to the "econanies of the merger, not its
accounting description."41

In a separate stuiy, Phndelker found no evidence of accounting
treaiment of mergers influencing the profitability of a mergerfl'2

Nevertheless, misconceptions persist. Ferguson and Pbfidn's
recent article disregarded pro-merger price appreciation of target firms
and the presence of negative risk adjusted excess returns accruing to
acquirers in the post merger period, contending that existing accounting
conventions for depreciation encouraged mergers.43

d. Price-Earning Ratios
Vergers occurring in the 19605 were often "justified" by
the expectation that a parent firm's price-earnings (p/e) ratio may

 

 

 

17
remain unchanged after it acquired a firm with a lower p/ e, and presuned

lower growth prospects. Such an acquisition, in theory, should depress
the price-earnings ratio of the acquiring firm. 'Jhe new price—earnings
ratio should reflect this change in outlook for the growth of the
canbined firm. Aside fran the more obvious defects in a price-earnings
analysis (vis—a—vis the quality of earnings], it is the value of the
canbined firm, not its earnings, which is the relevant test.
***************

Some acquisitions have less econanic suitability than others. It

is those which are discussed next.
4. Moments Merger Effects

 

Conglanerate mergers are defined as those mergers which
lack any of the operational, market-share, or financial advantages
previously discussed.

In the absence of those advantages, any two firms whose returns
over time are less than perfectly positively correlated will benefit
fran diversification creating a superior risk adjusted investment due to

the redmtion of the costs of bankruptcy.

Levy and firnat observe:

. . . diversification can be expected to produce a

true econanic gain owing to the fact that the canbination of

the financial resources of the two firms making up the

merger reduces lenders' risk while canbining each of the

individual shares (fl the two companies in investors’
portfolios does not.

Most of the other reasons for conglomerate mergers are categorized
as operating, managenent, or financial effects which are not exclusive
to conglanerate mergers.

Alberta, in illustrating the reduction of risk, erroneously

concluied investors could diversify their own portfolios. However, he

am

3 18
[ failed to consider bankruptcy costs.45
‘ Haley and Schall concede that once bankruptcy costs are

considered, " . . . there may be benefits from reducing financial

distress costs as a consequence of the mergers.”6

Rubinstein states:

Bankruptcy penalties, . . . , create an incentive
to merge since mergers st invariably diminish the
probability of bankruptcy. Synergistic benefits could
clearly cause the same project to have diffefient marginal
dollar returns and costs to different firms.

levy and Sarnat suggest:

. . . mergers may create financial advantages. For
exanple, large firms have better access to the capital
markets and also enjoy significant cost savings when
securing their financial needs.

Iogue and Naert advocate conglomerate mergers not only for the
previously mentioned reasons but also because of an extension of the
diversification argunent, called the "resource allocation and resource
utilization synergflfio

Baugen and Iangetieg, in discussing synergy, observe that it:

. . . may result because the merger makes possible
entry into new product lines which change the level,
stability and cyclical nature of the firm's profitability.
Vertical canbination may reduce the risk of fluctuations in
the price and availability of raw materials. It is possible
that managanent of the acquired firm is replaced by
individuals who are more aggressive in nature. Merger may
raise the profitability of a depressed firm in poor
financial coalition and significantly reduce the risk of
bankruptcy. Fbr the relatively snall firm, it opens new
sources of capital and it may also reduce the possibility of
insolvency due to an unfavorable liquidity position. It
may also significantly affect the volatility of the market
value of their common stock. In any case . . . these
factors should manifest thanselves in a change in the
distribution of rates of return to stockholders of the
merged firm.

a. Foreig Aguisitions
Another motive for mergers which might be categorized in a

 

l9

conglanerate context is an extension of the risk reduction argunent for
diversification; the acquisition of foreign based assets to reduce
further the systematic risk in a portfolio [24,36,EQ,115,122]. Such
redmtion of systenatic risk is due to the less than perfect positive
correlation of the respective asset returns. Aside from diversi-
fication benefits, there are additional risk reducing factors explaining
this activity in the United States. Those factors incluie political
instability and foreign exchange. "There exists a desire to shelter
capital by moving it into the canparative safety of the United
States.”2

What is perceived as an increase in socialian in Eirope53 and in
Canada54 has made the United States appear to be one of the last bastions
of a capitalist-oriented free enterprise system.55 Regarding foreign
exchange:

Foreign buyers have special reasons of their own for playing

the takeover game. With both the dollar and the stock market

on the mat, U. S. canpanies look enchanti%ly cheap, even

though their earnings are also in dollars.

Carberry canmented on the relative "cheapless" of U. S.
corporations [40]. Regarding Real Estate Invesiment Trusts:

In Elrope there aren't as many properties available for

investment; it takes longer to arrange financing, and most

deals are done with at least 50% cash. An added allure for

foreigners is the low value of the dollar relative to

foreign currencies and the cgmparative security of

investments in the United States.
Rout also comments on foreign acquisitions of United States canranies:

'Ihe foreigners appetites for U. S. concerns have

been whetted during the last two years by a variety of
coniitims, ranging from the weak U. S. dollar . . . 'Ihe
foreigners are cashing in on a canbination of cheap
canpanies and cheap dollars. . . . with the weak dollar has

meant the U. S. canpanies are undervalued absolutelg and
even more so given the currencies of the acquirers.l5

 

 

 

 

 

20
Fran an untitled article in a European publication came the
following qwtation: " . . . movenents in exchange rates and recent
rates of inflation have made the United States again a relatively low
cost producing area . . . "59

***************

D. Literature Review — Manfirial Therm

 

The Managenent Theory of Mergers led Weller to explain
abnormally low returns of firms during the post-merger period canpared
to the pro-merger period. He sugests that corporate managenent may be
at cross purposes with the broad goals of the corporation in that while a
corporation seeks maximization of the firm's value, managers often are
canpensated for their ability to produce asset expansion.

Managerial salaries, bonuses, stock options, and pranotions all
tend to be more closely related to the size or changes in size of the firm
than to its profits.60

In one observation of mergers policy within A. H. Robbins, the
acquisitions staff was split into teens. The head of one team wanted to
find the acquisition so he could put another picture on his office wall
to accanpany pictures of other acquisitions for which he was
responsible, according to a critic on one of the other acquisition
teens.61

Mueller also contends that snaller growth oriented firms have
higher marginal rates of return than do more mature firms.62 A similar
statement can be made about their marginal cost of capital.
Furthermore, the invesiment opportunities of the mature firm, within its
traditional business activities, are more restricted. Because of that,
the schedule of marginal rates of return on invesiments is likely to

 

 

 

 

 

 

21
intersect well within the horizontal segment of its marginal cost—of-
capital function.

If Mueller's contentions are correct, the results would have
enormous ramifications in merger theory and in the valuation of
acquisitions. His article produced several rebuttals and observations
[84,3,73,86] as well as his subsequent reply [102].

Mueller's argunent that mature firms have lower costs of capital
drew criticisn fran Copeland and Weston:

If the managers were making investment decisions
using an investment hurdle rate below a market equilibriun
rate and therefore below the alternative returns to
stockholders, stockholders would shift their investment to
firms offering higher rates of return. Basic capital market
forces would not permit differe firms to follow a 'two-tier'
investment hurdle rate policy.

The reconciliation of these views is reasonably simple. Mueller,
in discussing conglomerates, seems to have missed the correct rationale
for his own argunent. Acquiring firms do not use lower discount rates
in order to maximize growth rather than shareholder wealth. As Iogue
and Naert observed:

. . . a firm . . . (holding a group of diversified
assets with significant overall variance reducing covariance
effects) will be able to establish a broader clientele of
investors and ézzius because of the greater clientele, sell at a
higher price. . . . we would rather argue that a lower
discount rate is us because of managenent's expectation of
synergistic effects. 5
However, it is Wizeller's first contention-that firms maximize

growth, not shareholder wealth maximi zation—that draws the most
criticism.

Iewellen and Huntsnan's study of fifty of the top ninety-four
canpanies in the Fortune 500, from 1942—1963, regressed managerial

canpensation on profits and sales. then to avoid debate over whether

 

 

 

 

 

22

profits maximization ms the same as stockholder wealth maximization,
they ran a second regression replacing profits with market value of
equity. In both instances, the regressions produced high coefficients
of determination. The coefficient of sales was insignificant in both
regressions, and coefficients of both profits and market value of equity
were large in their respective regressions.66

Haley and Schall devoted an entire chapter to the concept that
firm objectives, in imperfect markets, might produce coalitions where
firm—value maximization is inconsistent with shareholder-wealth
maximization. The most likely of the two causes for divergence exists
if a firm adopts investments which change the risk of the firm's
outstanding bonds.67

If management pursues policies that maximize the

total value of the firm's securities, under sane

circunstances those policies may result in lower values for

the firmég shares, but higher values for the firm's debt

[bonds].

rihe second cause for a divergence is less likely. Ibre, the firm
issues new bonds which are not of a lower priority than the old bonds.
It ignores corporate taxes, relies heavily upon perfect capital markets,
and assunes that the firm's value will not change, regardless of its
debt-equity ratio.59

Regarding managerial canpensation, Cort contends, " . . . merge
frequencies vary greatly anong industries [and are] highly concentrated
in some industries.”0 He continues:

. . . if mergers are a consequence of the personal ambitions of

managers to manage large firms, why is it, then, unless

anbitious men tend to exist in certain industries, that you

have oonceptraticns of merger activity within certain

industries? 1

Still the case that firms maximize shareholder wealth is not

23
universally clear, according to Conn and Nielsen;72 as well as others.
Reid, in his book on mergers, discusses several stuiies which contend
otherwise:73 Those studies reviewed include Eaunol's stuiy which
states:

Indeed, in talldng to business executives one may easily
cane to believe that the growt of the firm is the main
preoccupation of top managenent. . . . managenent's goal
may well be to max ize 'sales' [total revenue] subject to a
profit constraint.

A subsequent test by McGuire, et. al., stlxiied correlations
between executive incanes, sales, and profits of 45 of the largest 100
industrial corporations from 1953-1959. 'Ihe results indicated:

. . the evidence presented would seen to support the
likelihood that there is a valid relationship between sales
and execligive incanes but not between profits and executive
incomes.

In another study, Roberts concluded the relationship between sales
and executive canpensation appeared to be stronger than the relationship

between canpensation and profits?7

 

Patton studied 420 canpanies in the period 1953-1964 and
determined that canpany size was the chief determinant of top executive
pay.78 Additionally:

Since sales growth is such an important variable in

determining top managenent incane, there is a basis for

conflict between the personal intenfgsts of top managenent

and the interests of shareholders."

Wanna has created such a scenario:

 

Generally speaking, managers' incentives and interests
coincide with their shareholders in every particular except
one: They have no incentives as managers to keep managenent
services for the company at the lowest possible price. To the
extent that the sane individuals are also shareholders, their
motivation will reflect a conflict. Even if the market for
corporate control is working perfectly, so long as the cost to
the corporation of the incunbent manager's inefficiency is
below the cost to an outsider of taking over control, the

   

 

2L!

insiders will renain secure in the positions with high

salaries. This may furnish sane proof for thifiwtion that

executive canpensation is a function of size.

Dean and Snith are very critical of the Baunol and the McGuire,
Chin and Elbing studies. Regarding Baunol, they criticize his
"uncritical acceptance" of executive interviews and their public
statenents that "growth has a new and independent status."81 Dean and
Snith assert that "in a time that values growth, annual reports and other
corporate utterances will stress anything that can be used to show that
the firm is in progressive alliance with the trend of the times."82

They also criticized the McGuire, Chiu, and Elbing stuiy
addressing the lack of standard definitions of managenent and
compensation to mean the same thing for different size and industry
groups.85 Also:

The managenent job is in part the efficient administration

of a collection of assets; and the profitability of these

assets, especially in the short run, will usually be

determined far more by externals thaéhr by the special
contributions of a current managenent."

Much of the preceding was based on manager activities in the 19505
and 19605. Fhrbar, in what may be indicative of different
circunstances, discusses merger activity in the 19705. He indicates
shareholders have been suing directors for neglecting fiduciary
duties.85

Gerber's board is being sued by shareholders who
believe they should have been the ones to accept or reject
Anderson Clayton's offer last year. Courts historically
have been hesitant to second—guess a board of directors, but
Gerber's shareholders have a chance of winning their suit.

The board spurred a friendly feeler fron Unilever while it
was fighting Anderson Clayton; the directors' motives

seemed to be to keep the company independe rather than get
the best possible deal for shareholders. Similarly,

 

 

 

25

Universal Leaf is being sued for thwarting Congoleum's
offer, and Marshall Field is being sued over its successful
defense of Carter Hawley Hale's takeover attempt. However
the suits are decided, the prospect of having to answer
shareholders in court has alrgfidy begun to influence the
behavior of outside directors.

Such legal actions reinforce wfarme's contention that mergers or a
threat of takeover by more efficient management acts as an incentive for
inefficient managers.88

***************

E. Literature Review — Shareholder Returns

 

Firms are acquired on the theoretical grounds of a "fit." The
question is whether or not the acquisition produced the desired
synergistic effect.

Dodd and Hiback assert:

The effect of mergers upon the value of the firm has

been a contentious issue in the literature and empirical

investigations have presented conflicting results. Many of

122?? 52:33: ifenefiieifcgfiiiiififiirmf§f$%°%is and

What stuiies of post-merger returns behavior of the acquiring firm
suggest that returns are normal or less than normal in the post-merger
period and any gains accrue to shareholders of the acquired firms.

Melker's study is cited here because it is one of the more
recent stuiies (1974), and uses the two factor market model [87]. He
concluies that the acquiring firm' s monthly returns are normal in the
post-merger period and shareholders earn abnormal returns of
approximately 14% on the average in the seven months preceding the
merger.90

Dodd and Ruback contend that andelker did not consider what may

be an important consideration in measuring any potential abnormal

 

 

26
returns to stockholders of acquiring firms.

Mandelker reports that shareholders of acquired firms
earn abnormal positive returns over the seven months before
the merger month. If the mergers in his sample are preceded
on average by a tender offer or similar announcement, the
prenerger gains could reflect the market reaction to the
earlier release of this information.

Using tender offer dates, Dodd and Ruback discovered:

. . in the month of the announcement, target firm
stockholders earn large and significant abn8§mal returns of

20. 58 percent for successful offers. . . .

The paper also contains the first enpirical assessment of the
market reaction to unsuccessful takeover attanpts. The stockholders of
bidding firms which initiate unsuccessful tender offers neither gain nor
lose—-they earn normal returns in the offer period. Unsuccessful target
firms, however, earn large significant positive abnormal returns of
18.96 percent in the month the offer is announced. Furthermore, the
price change is permanent since they earn normal returns for five years
after the offer.93

Mandelker does not indicate what percent of his sample experienced
prior tender offers, but the potential for a bias exists. There
probably are many more tender offers now than there were prior to
Marflelker's study.

According to Ehrbar, in his discussion of recent characteristics
of mergers:

The current takeover wave got started in 1974 [the year
Walker's dissertation was published], and has been
building ever since. More and more of the takeovers sea to
be hostile, i.e. , opposed by the target's management.

Eirbar asserts that the targets are a lot larger in asset size than

they used to be 95 and "Most important the bidders have been paying

 

   
 

 

27
higher praniuns than ever."96 Historically, according to Benjanin

Graham, tender offers have been made at prices averaging around 20—50%
above the market value of the target stock prices.97 Nerjos, in her
stuiy (1977), found premiuns to be 40-50%.98
Recently the premiums have been averaging more than

60%, and the contested deals have been even hotter. The

average premiun in the ten $100 million-plus contests last

year was more than 80%. In a few ases the raiders paid

more than double the market value.98

How can there be such a disparity between existing prices and
either tender offer prices, or merger prices? Is there any evidence in
the literature indicating that those premiuns were warranted? Most
stuiies conclude that the gains, if any, generally accrue to the
shareholders of the acquired firm. The existence of positive postnerger
gains accruing to shareholders of acquiring firms is less certain.

Conn and Nielsen, in an enpirical stuiy of an earlier model by
Iarson ani Gonedes (Ir-G) [77] found a significant nunber of mergers
resulted in losses to both acquirer and acquired firms.100 However, the
wealth loss ms much greater for acquiring firms than acquired . "Rarely
did the acquiring firm's stockholders gain while the acquired firm‘s
stockholders lost."101 Regarding the interests of the shareholders, " .
. . at least 40% of the mergers [in their study] do not conform to the
rationality assunption that the bargaining process is constrained by
each firm maintaining at least its stockholders' wealth status in the
period of the merger."102 One flaw acknowledged by the authors in the
study is that the L-G model made no provision for the possible change in
the risk-return profile of the merger participants.103 AS the authors
contend, "If a merger results in reduction of systematic risk that is

unobtainable for existing stockholders . . . their risk return position

 

T"

 

 

28
may actually improve even if return declines.104

Dodd [1976], in a study of Australian equities, found shareholders
of canpanies receiving takeover offers benefited whether or not the
acquisition was completedf

earnirrespective (185 the outcome, [those] shareholders

gains. . . . 0n the other hand, shareholders of
acquiring firms suffered significant losses after
takeovers. It appears that any gains arising from the
merger-werewon b66the acquired firms at the expense of the
acquiring firms.

In a more recent study of NYSE firms in the 1970s, Dodd (1980)
found results similar to those of his Australian study. Here the
acquired firm earns returns of 13%, and the non-acquired firm earns
almost 11%, even though their managenent vetoed the proposed takeover.
If the potential acquirer called off the merger, prices returned to
previous levels. Finally, losses of 7% accrued to stockholders of
successful bidders in the post—merger period.107

Lev and Manialker (IN), using a paired canparison sanple and
annual returns, concluded (1) there is no evidence of risk reduction
fron mergers and (2) the acquisition produced a decrease in the growth
rate in the post—merger period, relative to non-merging firms, which he
referred to as a "digestion effect."108

Other results in the study indicated there were no tax effects,

and no evidence of accounting treatment of mergers influencing the

profitability of merger. Instead the accounting treatments tended to

*Emis paper also provides a fifty year summary of merger stuiies in
the United States beginning with Dewing‘s stuiy.

Dewing, Arthur S., "A Statistical Test of the Success of
Cmaolidaticns." Quarterly Journal of Economics 36 (1921): 84-121.

 

 

29
Imderstate the profitability of mergers.109 Because of the use of

annual returns, their study " . . . did not investigate short run
effects."110

Both Reid (1975) and Ibneycutt criticized the L"! stuiy. Reid
because, "426 of the non-merger' s [control group] had as many or more
mergers . . . than the acquiring firms.111 Honeycutt argues that an
acquisition would change the primary industry classification of the
acquiring firm, while the control firm would not change, reducing the
value of a paired canparison.112

In another study, Reid found that companies growing internally
grew at a rate of 601% versus 307% for canpanies actively involved in
mergers during the period 1951—1961. His study involved 478 of the
nation's largest oorporations.”3 He attributed this finding to the
substantial preniuns over market value paid for assets acquired.114

Halpern concluied, in a study of mergers between large and snall
firms, that adjusted gains were positive to both groups and divided
eveniy.“5 His study of '78 mergers in the period 1950-1965 did not
include firms actively acquiring other firms, nor did he make any
attenpt to separate buyers fron sellers.

. . . this distinction [separating buyers from sellers] is

arbitrary and has no econanic justification. Since we do

'1‘? hfi‘iii‘éfifiia’r‘i it?iiiirt’iiff’gexfgé’fii‘““521 *3

13°53 1% We

He also disregarded diversification argunents, citing investor
diversification and attributed the positive gains to good "fits."1"'7

Haugen and Iangetieg stuiied 59 mergers in the period 1951-1958,
using a 72 month time series and concluied:

. . . a merger fails to produce economically significant

changes in the distribution of rates of return to the
stockholder. Orr attention centers on the risk attributes

 

 

 

30

of the distribution, and we do not address dollar benefits

of canbination which are capitalized in the stock price with

the armouncflignt and subsequent consunmation of the

combination.

Hogarty studying 43 mergers in the period 1953-1964 were " . . .
judged unsuccessful" to the acquirer using annual returns.119 But, a
few " . . . obtain very large returns, and the prospect of these large
returns tempts other firms to engage in merger activity."120

In Iangetieg' s couprehensive three factor model (a zero beta with
an industry factor and a matched non—merging control group) using
monthly data seventy two months before and after amerger of 149 mergers,
he observed returns of 12% to the acquired stockholders over the
interval 6 months to 1 month prior to amerger. These results replicate
Mandelker's study [87]. Regarding shareholders of acquiring firms,
Iangetieg found positive pre-merger returns:

However, the gain is clearly too small to conclude that

enhancanents of stockholder welfare is the sole motive for

merger. . . . perhaps another motive . . . managerial
welfare 1 gmay have been the instrunental cause of the
merger.

(he enpirical study supports the managerial motive. The
explanation is the difference in risk perception of managers and
stockholders. Stockholders can diversify asset portfolios, but
managers do not have portfolios of enployers. Managers may acquire
firms not to enhance profits but rather to make the firm less risky,
making jobs more stable.122

Nblicher and Fhrter found that prices of acquiring canpanies are
"bid-up” before rather than after merger benefits, especially when the

acquired canpany was more than 1/2 the size of the acquiring company,

[measured by total assets]."123

 

 

r

 

31
Nielsen and Melicher found no support for " . . . instantaneous or

real financial synergy."124 When financial gains were obtained, it
occurred in instances where acquiring firms were able to pay below
prmiuns for their acquisitions.”5 Bradley's stuiy of one-hundred
sixty one successful tender offers produced " . . . conpelling evidence
for a. synergy theory of tender offers."126 Perhaps Bradley's results
are contradicted though by the initial experiences of tendering
stockholders fron thirty firms which experienced sizeable share price
reductions in the post—tender period 1969—1970.127

***************

F. Conclusion

The evidence generally indicates that market price appreciation of
merged firms takes place seven to thirty months prior to the event. The
gains going primarily to the acquired firms suggests little support for
the monopoly theories since there was rarely evidence that both parties
profited. (he recent article contends that as many as seventy percent
of all mergers are unsuccessful [110].

Sane evidence exists to support the managerial theory, that there
is a. cmflict between shareholders and managenent. Weston and Rice
[130] anong others, conclude:

The inefficient utilization of econanic resources by the

prior managenent leads to their being acquired by firms with

a. prior record of above average performance. Hence the

evidence leans in the dirfigion of efficiency and/ or synergy

as the explanation . . .

(he survey of one Imndred seventy five chief financial officers involved
in 1978 mergers iniicated their perception was the horizontal mergers

were more successful than vertical or conglomerate mergers.129

 

32

The fact that the acquisitions have had "subnormal" performance
previous to the merger coupled with the efficiency explanation provides
the incentive to determine that such subpar performance will be
reflected in the firm's financial statements.130

To sunnarize this chapter, it was necessary to discuss the
theories of mergers as motives appearing in the literature, as well as
other motives advanced by both arbitrageurs and by mergers staffs. By
understanding the motives for mergers, it may becane possible to predict
then. If firms are acquired because of their financial characteristics,
such characteristics should be reflected in the financial statenents of
the acquired firm.

The purpose of this study is to generate three models which use
financial ratios to predict mergers.

In the next chapter, various ratios are examined which purportedly

measure incentives for mergers and hence provide predictive content.

 

 

 

Chapter II

Footnotes

See: "Asset Redeployment: Everything Is For Sale Now," Business
Week, 24 August 1%1, p. 71, for a discussion of incorrect future

discount rate to be used on forecasted future cash inflows.

See: William M. (hrley, "Airline Service Reakdown Persists Iong
After Pan Am, National Merge," Wall Street Journal, 10 March 1981,
p. 25.

"During one November week, 154,000 calls weren't
answered. 'We'll never know how mmh revenue we lost.‘ says
William Waltrip, executive vice president of Pan m."

"In buying National, Pan Am got four M105 it didn't
need, and hasn't been able to sell then because ofaglut of
used aircraft on the market. Pan Am also has a brand new
$45 million 11310 sitting on the ground at the NcDonnell
Douglas plant near Ios Angeles. The plane, which isn't
needed now, had been ordered by National."

Also, see: Thonas Petzinger, "To Win a Bidding War Doesn't Ensure
Slccess of Merged Companies." Wall Street Journal, 1 Septenber
1981, pp. 1, 12.

J. R'ed Weston and Eihard 1". Rice, "Discussion," Journal of Finance
31 (May. 1976): 745. ——

Ibid.
W. W. Alberta and J. E. Segall The Co rate Var r (Chicago:
University of Chicago Press, mesh—W .

German i’hndelker, "Risk and Return: The case of the Merging
Firm," Journal of Financial Economics 1 (Decenber, 1974): 305.

Thanas E. Copeland and J. Red Weston, Financial Theo and
Comrate Polig (Menlo Park: Addison Wesley, 1W9): :23.
Wilbur G. Iewellm, "A Pure Financial Rationale for the Conglanerate
Morgen: Journal of Finance 26 (May, 1971): 522.

Copeland and Weston, p. 423.

Charles W. Eloy and Iawrence D. Schall, The Theom of Financial
33

 

16.
17.
18.

 

311
Decisions, 21d ed. (New York: McGraw Hill Book Company, 1979): 447.

Henry G. Mamas, "Mergers and the Market for Corporate Control," Journal
of Political Econggx 73 (April, 1965): 110.

Ibid, p. 111.

Ibid.

Ibid, p. 113.

John Lintner, "Expectations, Nbrgers and Ehuilibriun in Purely
Cgfititixg. Securities Markets," American Economic Review 61 (Why,

George Finches and Kent A. Mingo, "A Multivariate Analysis of
Industrial Bond Ratings," Journal of Finance 28 (March, 1973):

Iewellen, "A Pure Financial Rationale," p. 522.

Copeland and Weston, p. 426.

Ihley and Schall, p. 446.

Donald Iee Stevens, "A Miltivariate Analysis of Financial
Characteristics of Acquired Firms in Industrial 1"Iergers" (Ph.D.
dissertation, Michigan State University, 1972), p. 19.

Janes Tobin and Willian C. Brainard, "Asset Markets and the Cost of
Capital, " Cowles Foundation Paper No. 440 (New Haven: Cowles
Founiation for Research in Econanics at Yale University, 1977): 235.
Ibid, pp. 236—237.

Ibid. pp- 254-255-

Ibid, p. 262.

Seymour flicker, "'Ihe Q—Ratio: Fuel for the Merger Fania," Business
Wed:, 24 August 1981, p. 30.

Ibid.

Enry W. mappell, Jr. and David C. Cheng, "E: T‘ctations, 'Ibbin's q and
Investment: A Note," Journal of Finance 37 ,1982): 231.
kicker, p. 30.

Tobin and Brainand, p. 254.

Adan l’eyerson, "Merger Mania and the High Takeover Premiuns," W_all
Street Journal, 20 July 1981, p. 14

ibbin and Brainard, p. 254.

 

41 .
42.
43-

45-

47.

 

35

Zucker, p. 30.

beerson, p. 14.

Chappell and Cheng, p. 232.
Tobin and Brainard, p. 249.
Ibid.

Snappell and Cheng, p. 233.
Zuclner, p. 30.

Iewellen among others believed some investors were deceived by
misvaluations of mergers in the 19603. See Iewellen, pp. 522-523.

Elier, et a1, believe that more conservative accounting resulting fran
the passage of APB 16 and 17, coupled with other factors reduced merger
activity but fostered more soundly conceived mergers. See:

H. Kent Baker, Thanas 0. Miller, and Brian J. Ransparger, "An Inside
Iook)at Corporate Mergers and Acquisitions," MSU Topics 29 (Winter,
1%1 : 49.

Bai Hang, Ibbert S. Vaplan, and Gershon Mandellner, "Pooling vs.
Purchase: The Effects of Accounting for Mergers on Stock Prices," The
Accountipg Review 53 (Janmry, 1978): 44.

Ibid, p. 31.

Mandelker, pp. 307, 330.

Robert Ferguson and Phillip Rafldn, "Pulling Ihbbits Oit of I-hts in the
Oil Easiness - and Elsewhere," inancial 11mm Journal 38 (Wareh—
April, 1%2): 25-

Beim levy and Marshall Sarnatt, "Diversification, Portfolio Analysis
and the Uneasy Case for Conglanerate Mergers," Journal o_f Finance 25
(Septanber, 19170): 801. — ---
Alberta and Segall, pp. 262-272.

lhley and Szhall, p. 446.

Mark E. Rubinstein, A Mean-Variance S thesis of Co rate
Financial Theo . ed. SEE C. MyTers, Raern Develo ents in
Comrafi financial Theory (New York: Praeger fiblisfiers, 1976),

p. 47.

Ibid, p. 51.

Ievy and arnatt, "Diversification, Portfolio Analysis,"
p. m1.

50.

51.

52.

53.

54.
55.

56.

57.

59.
60.

61.

62.
63.
64.
66.

67.
68.
70.

36

Dennis E. Iogue and Pnillippe A. T‘aert, "A 'I‘heory of Conglomerate
"ergers: Comment and Iktension," Quarterly Journal of Economics 87
(November, 1970): 665-666.

 

Robert A. Haugen and Terrence C. Iangetieg, "An Vmpirical 'f‘est for
synergism in T-"erger," Journal of Finance 30 (September, 1975):
1004.

 

Iawrence Rout, "Ewing American: Weak Dollar, Stocks Spur Foreigners
to Seek Acquisitions In The United States," Wall Street Journal 21
August 1979, p. 1.

"'Ihe Great Takeover Pings," Pusiness Week, 14 November 1977: 1'78-
179.

 

 

"A Financial T’mpire In The Taking," Business Week, 91"arch 1991: 87.

 

"Market Commentary: T'ergers and Acquisitions - How To Buy An
American Company," Furomoney, November, 1976, p. 141.

A. F. Ehrbar, "Corporate Takeovers Are Here to Stay," Fortune,
8 “Kay 1978: 91 .

James Carberry, "P.”any REITs Stage Comeback, Aided Py An Attraction of
Foreign Investors," Wall Street Journal, 6 August 1979, p. 6.

 

Rout, p. 1.
"I‘~'arket Commentary," Furomoney, p. 141.

Dennis C. Mueller, "A Theory of Conglomerate Vergers,"
Quarterly Journal of Economics 83 (November, 1969): 644.

Victor Zonana, "Planning A Plant: How A. H. Robins Co. Wade Decision
to Puild a Chemical Facility," Wall Street Journal, 22 October 1975,
p. 1.

 

1‘711eller, pp. 646-647.
Copeland and Weston, p. 429.
Iogue and I’aert, p. 664. 65. Ibid, p. 665.

Wilbur G. Iewellen and Elaine Huntsman, "Vanagerial Pay and Corporate

figfomance," American Economic Review 60 (September, 1970): 711-

Paley and Schall, p. 478.
Ibid, p. 482. 69. Ibid, p. 478.

Michael Cort, "An Phonemic Disturbance Theory Of T’ergers,"
Ellarterly Journal of Economics ‘35 November, 1969): 625.

71 .
72.

74.

75.
76.

(micago: University of Chicago Press, 1966): 8.

37
Ibid.

R. L. Conn and J. P. Nielsen, "An Empirical Test of the Iarson-Gonedes
Exchange Ratio Determination Model," Journal of Finance 32 (June,
1977): 754-756.

 

Samuel Richardson Reid, “-iergersLVanagers, and the Economy (New York:
ichraw-Hill Book Campanyf1968): 135.

 

William J. Baunol, "01 the Theory of the I‘x nsion of the Firm,"
American Economic Review 52 (December, 1962 : 1078.

 

Ibid , p. 1085.

Joseph W. P’bGuire, John S. Y. Chiu, and Alvas O. Elbing, "Executive
Incomes, Sales, ani Profits," American Economic Review 52 (September,
1962): 760.

 

D. R. Roberts, Executive Compensation (New York: The Free Press of
Glencoe, 1959).

 

Arch Patton, "Deterioration In To Ececutive Pay," Harvard Business
Review 43 (November-December, 1975 : 106.

Ibid, p. 113.

Henry G. Nanne, "l‘ergers and the Market for Corporate Control,"
Journal of Political Economy 73 ( April, 1965): 117-

Joel Dean and Winfield Smith, The Relationships Between Profitability
and Size, ed. W. W. Alberts and J. E. Segall, rI‘he Corporate Verger

 

Ibid.
Ibid, p. 7.
Ibid.

A. F. Ihrbar, "Corporate Takeovers Are Here to Stay," Fortune, 8 T‘Tay
1978, p. 96.

Ibid.

Ibid.

I"WIDE, p. 113.

Rater Dodd and Richard Ruback, "Tender Offers and Stocldmolder Returns,
An Enpirical Analysis," Journal of Financial Economics 5 (December,
19'77): 351 .
Valflelker, p. 303.

91 .
92.
93-
94.
95-

96.
97.
98.
99.
100.
101.
102.
103.
104.
105.

1%.
107.

1%.

109.
110.
111.

112.

113.

38
Dodd and Ruback, P. 352.
Ibid.
Ibid.
Emrbar, p. 91.
Ibid. Also, while there are not nearly as many mergers created in the
1970s, the dollar value of the latter are almost equal to that of the
former. See "’Ihe Great Takeover Binge," p. 176.
Fhrbar, p. 91.
Ibid.
Anna Pbrjos, "Takeover Targets," Parrons, 15 Way 1978, p. 14.
Ihrbar, p. 91.
Conn and Nielsen, pp. 755-756.
Ibid, p. 755.
Ibid, p. 754.
Ibid, p. 758.
Ibid.

Rater Dodd, "Company Takeovers and the Australian Equity i’arket,"
Australian Journal of Management 1 (October, 1976): 24.

Ibid.

Idem, "P“erger Proposals, Vanagement Discretion, and Stockholder
Wealth," Journal 2f Financial Economics 8 (T-’Tarch, 1980): 105.

Baruch Lev and Gershon thdelker, "The Nicroeconomic Consequences of
Corporate "ergers," Journal of Busine$ 45 (January, 1972): 97.

 

Ibid, p. 102.
Ibid, p. 103.

Samuel Richardson Reid, "'Ihe F-“icroeconamic Consequences of Corporate
Mer‘gers: Comment," Journal of Business 48 (April, 1975): 280-

 

T. Crawford Honeycutt, "The I‘icroeconomic Consequences of Corporate
Mergers: A Comment," Journal of Business 48 (April, 1975): 267°

"Growing From Within Way Pay Off 1Register," Business Week, 17
Saptenber, 1966, p. 44.

 

114.
115.

122.

125.

124.

125.
126.

127.

128.

129.

130.

39
Ibid, p. 46.

Paul J. Halpern, "Empirical Istimates of the Amount and Distribution
Gains to Companies in '"ergers," Journal of Business 46 (October
1973): 570.

Ibid, p. 557.
Ibid, p. 573.
Haugen and Iangetieg, p. 1003.

 

Thomas F. Hogarty, "(me Profitability of Corporate Tv'ergers," Journal
of Business 43 (July, 1970): 322. "““"

 

Economics 6 (December, 1978): 382.

Ibid, p. 326.

Terence Iangetieg, "An Application of a Three-Ractor Performance
Index to Yeasure Stockholder Gains From 1'erger," Journal of Financial

Yakov Amihud and Baruch Iev, "Risk Reduction as a Vanagerial Motive
for Conglomerate P’ergers," The Bell Journal of Economics 12 (Autumn,
1981 : 606.

 

Ronald W. '"elicher and Thomas R. Harter, "Abstract: Stock Price
Movements of Finns Engaging in Iarge Acquisitions," Journal of
Financial and Quantitative Analysis 7 ("arch, 1972): 1474.

 

James F. Nielsen and Ronald W. Velicher, "A Financial Analysis of
Acquisition and Merger Premiums," Journal of Financial and
Quantitative Analysis 8 (F’arch, 1973): 146.

 

 

Ibid.

Michael Bradley, "Interfirm Tender Offers and the Varket for
Corporate Control," Journal of Business 40 (October, 1980): 346.

 

"The Morning After - Most of the Stockholders Who Accepted Those Fancy
Tender Offers Now Have Reason to Regret Their Taste," Forbes, 1
February 1970, pp. 15-16.

Weston ani Rice, p. 746.

Baker, P~1iller,and Pamsparger, p. 56.

Stanley Block, "The Merger Impact on Stock Price F’ovements," "SU
Business Topics 17 (Spring, 1969): 11.

 

CHAPTER III

The Plan of Research

A. Rationale for the Use of Financial Ratios

 

Analysts find financial ratios to be appealing for several
reasons. 1) Financial information is publicly available for analysis of
publicly traded corporations because of Federal security disclosure
regulations. 2) Financial ratios provide measurements of a firm's
growth in earnings and asset size, debt capacity, efficiency, and
dividend policy. 3) Ratios compare a given firm's performance within
an industry, or to firms within other industries because the ratios are
designed to provide comparability by relating the above attributes to
the size of the firm measured by sales, earnings, asset size, or number
of shares of common stock outstanding. 4) Numerous investment
advisory services rely extensively upon various financial ratios.

Given that financial ratios are pOpular, widely used, and readily
available, how successful are they as an analytic device?

In appraising ratio usefulness to investors, O'Connor found that
univariate ratio analysis would not be useful in differentiating between
common stocks primarily because of the semi-strong form of the efficient
markets pricing mechanism. " . . . even on a multivariate basis ratios
might be found to be of questionable usefulness in the prediction of

return rankings for common stocks."1
110

111

Beaver believes the evidence "overwhelmingly" suggests a
difference in the ratios of failed and non-failed firms,2 that the
ratios can be useful in the prediction of failure for at least five years
before the failure,3 and that the most commonly used ratios would
possess little utility because they are most often manipulated by
managenent.4 Beaver also suspects that a multi-ratio analysis can
predict better than the single ratios, but his results did not verify
that hypothesis.5 Neter criticized Beaver's sample design suggesting
that the non-failed firms sample be as large as possible for more precise
information . 6

Horrigan contended that, regarding bond ratings, "ratios are not
likely to be efficient predictors of dependent variables which shift in
a random pattern over time such as stock market prices, because the
financial ratios tend to be highly correlated over time.7 However, he
did conclude that the general approach ought to be that of a multiple
regression, rather than a univariate analysis, and that accounting data
and financial ratios have been found to be useful for determination of
corporate-bond ratings. The multivariate approach used by Finches and
Mingo [111] in their bond rating study provided further support for his
conclusion.

Iev [13] has cautioned users of financial ratios in the
application of financial ratios. Summarizing these limitations he
cautions against univariate analysis, misaveraging ratios, misinter-
preting ratio changes, and other faulty uses of percentages.

The conclusion is that it seems the use of multivariate
discriminant analysis on financial ratios provides better results in

terms of predictive abilities than presumed. There are explanations for

112
this, which are discussed in part 0 of this chapter.
B. Ratios Included in This Study
The purposes of this section of the paper are threefold: 1)
review financial ratios used in the previous studies by S“, Stevens, and
Belkacui; 2) present an additional set of ratios; and 3) discuss their

expected importance in light of the preceding discussion of the theory

of mergers .

Profitability

Iev erage :

Activity:

Other :

T"inanc ial
Ratio

F81

FR2

FR?

FR4
W5
FR6
PR7
FR1 8
PR1 9

PR9

IRAQ
F'R11

FRZO

FR12
FR13
ER14

FR15

FR16

IR17

Table 3-1

Ratios Studied

 

Author
Ratio Employing

Earnings Before Interest
and taxes / Total Assets S
Cross Profits / Sales S
Farni 3 Before Interest and
Taxes Sales S
Net Income / Sales S
Earnings Before Taxes /Sales S
Net Income / Stockholders lbuity B,S
Net Income / Total Assets B,S
Cash Flow / Net hbrth B
Cash Flow/ Total Assets B
Long Term Debt / “-‘arket Value 8
Common Equity
Long Term Debt / Book Value
Common Equity
Iong Term Debt / Total Assets S
Total Liabilities / Total Assets S
long Term Debt 4» Preferred Stock B
/ Total Assets
Sales / Total Assets S
Cost of Goods Sold / Inventory S
Sales / Quick Current Assets S
Interest Expense / Cash Plus
Farketable Securities S
Dividend Payout Percentage S‘J,S
Price / Farnings Ratio s1 s

113

1111

FR21 Total Assets
liquidity: PR22 Current Assets / Total Assets B

FR23 Cash / Total Assets B
FR24 Net Working Capital /

Tbtal Assets B,S
FR25 QuiCk Current Assets / Current

Assets B
PR26 Current Ratio B
FR27 Acid Test Ratio B
FR28 Cash / Current liabilities B
FR29 Current Assets / sales B
PR3O Quick Current Assets / sales B
11131 Net Working Capital / Sales B,s*

 

*The capital letter indicates which of the authors discussed used these
ratios. B indicates the Belkaoui Study, SM indicates the Simkowitz-
MOnroe Study, and S indicates Stevens ratios used in.his dissertation,
published article, or in both.

145

The above ratios were selected because of their popularity in the
literature. Quoting Belkacui:8

They represent the 'traditional' categories in ratio

analysis: balance sheet ratios . . . income statement

ratios . . . , and mixed ratios.

. . . their possible relevance to the takeover
phenomenon; . . . [and]

. . . their appearance in the literature as indicators

of the ability of a firm to avoid takeovers.

Additionally one other set of ratios is introduced. "hat set
includes five ratios which examine specific attributes on a per share
basis relative to the year-end closing price of the firm' 8 common stock.
The closing price is incorporated into the ratio because the price of the
common stock should influence the likelihood of the acquisition of a
company. Those ratios are illustrated in Table 3-2.

The sixth ratio, sales to fixed assets, reflects the impact of
companies contending it is cheaper to buy existing assets than to build

new assets .

Table 5-2

Add it ional Rat ios Stud ied

FR32 Closing Price / Book Value Per Siare

IR33 Closirg Price / Cash Per Share

BR54 Closing Price / Net Working Capital
Per Share

FR35 Closing Price / Quick Net Working
Capital Per Share

FR36 Tax loss Carry Forward Per Snare /
Closirg Price

PR37 Sales / Net Fixed Assets

***************

116
C. Rationale for the Use of Discriminant Functions

 

There is substantial precedence for the methodology employed
in the present study. Piultivariate discriminant analysis (RDA
hereafter) was first employed by the scientists Fisher [56] for
taxonomic uses and Bernard [31] in measuring Egyptian skull sizes. The
first financial application was by Durand [48] to differentiate between
"good" and "bad" consumer loan applicants. Walter [128] classified
firms into high and low price earnings ratio groups. Ayers and Fbrgy
[103] developed a numerical credit evaluation system. Smith [121]
classified firms into standard investment categories. Finches and
Mingo [111] used FDA to determine bond ratings of long term debt.
Altman [26] predicted corporate bankruptcies through financial ratio
analysis. Since then Gabhart [134] used FDA to predict municipal
government insolvency in Michigan, and deister [50] attempted to
predict bankruptcies of small firms.

FDA has distinct advantages to other forms of analysis. First, in
using VEDA the researcher is able to analyze simultaneously the entire
variable profile of an object to be tested rather than by univariate
analysis. Univariate analysis is unable to provide two distinct
advantages of MDA: (1) MDA calculates covariances between the financial
ratios being tested and (2) also is capable of indicating which ratios
are more important. Most MBA studies of a financial nature provide
results which are dissimilar to results obtained from the use of
univariate analysis. In most instances there are clear differences in
the selection of most important ratios to predict an event. Examples of
the superiority of MDA in the literature are so common it is not helpful

to reference them here .

117
Another advantage of FDA, as opposed to multiple regression, is

MDA's ability to use qualitative dependent variables such as bankrupt or
solvent, good or poor credit risk, or is the firm an acquisition
candidate.

that is FDA, and how has it been used in merger studies? It is a
statistical technique which classified observations of several a priori
groupings based on certain characteristics. A linear MBA is an attempt
to derive a linear combination of these characteristics which best
"discriminates" between two groups. The discriminant function is of the
form Z = be + b1X1 + b2X2 . . . ann where bO is a constant used to
adjust for the grand means, (see necks [12], p. 443). hi, (i =1 . . .
n), are the discriminant coefficients of independent variable
characteristics, Xi’ (i = 1 . . . n) , and Z is the value which is then
used to classify the object into the two or more groups. See Altman,
[26] Morrison [99], or Greene [7] for excellent discussions of FDA.

There have been previous attempts to use FDA to identify merger
candidates. In Canada, Belkaoui [34] used discriminant functions on
data from 1960 to 1968 on twenty-five industrial firms listed on the
Toronto Stock l'kchange.

In a larger study, Singh used FDA and financial ratios to predict
mergers in the Unite Kingdom, from 1955-1967 [20,119].

In the United States, there were also to studies using NBA and
financial ratios to identify merger candidates. rI‘hese studies by
Simkowitz and Monroe (8‘1) [118] and by Stevens [1'55]. though done at
about the same time, reached dissimilar conclusions. It is possible
that the dissimilarities were attributable to the differing nature and

methodologies of their studies. Si". focused specifically on

118

conglomerate mergers and reached a ratio profile using a step wise
regression procedure. Stevens' study of mergers was not confined to
conglomerates, and he used a factor analysis to reduce the set of ratios
to a best set. Both SM and Stevens called for follow-up studies in a
different time period, to test for continuity of the best set of ratios.9
An attempt to resolve the dissimilar conclusions reached by the two
studies is overdue.

The new sample used in my study tests specifically for the
existence of such best set continuity, and will assist in the
determination of the better dimension reduction technique in this

application, factor analysis or discriminant analysis.

D. Statistical Limitations of This Study

 

There has been considerable discussion on the necessary
statistical assumptions and experimental design requirements using
financial applications of MDA [52,70].

NEDA requires that the discriminating variables have multivariate
normal distributions and the variables have equal variance-covariance
matrices within each of the groups. Morrison [99] and others have
indicated the importance of the equal variance-covariance matrices; it
is a prerequisite of the linear MDA model. 'Ihe test for equality is
Bartlett's Box M test, to be discussed in detail later. Rejection of
the hypothesis that the matrices are equal suggests use of a quadratic
EDA function, an extremely complex undertaking, which according to
Singh, " . . . raise[s] very awkward problems and [has] proved rather
intractable in practice."10

In a study by Marks and Dunn, the decision rule of unequal

covariance matrices precluiing a linear MDA function is clouded by

149
efficiency trade-offs with sample sizes. When sample sizes are small
and the number of variables relatively large , linear rules may give more
efficient estimates of the expected error rates than quadratic rules
even when the population dispersions are unequal.11 When sample sizes
are large, Klecka believes FDA to be robust to violations of the equal
covariance requirement [12].

'Ehe other considerations are misinterpretation of the
significance of the coefficients of the independent variables, reduction
in dimensionality, group definition problems, inappropriate a priori
probabilities , misestimation of classification error rates, use of split
samples, and misinterpretations of classification tables. All of these
will be considered in great detail in Chapter IV which discusses the

findings of this study.

E. Rationale for a Dimension Reduction
The chief purpose of a dimension reduction in this study is to

eliminate those financial ratios which do not contribute to the TTDA
function's overall ability to classify merger or non-merger candidates.
As Eisenbeiss observed: "This can be particularly important for
problems in . . . finance when it is often possible to generate a large
number of variables which need to be pared down to some manageable
size."11

Such an approach is not at all uncommon. Both the Stevens and 8-"
studies reduced substantially the number of ratios in their final model.
SP? began with twenty-four ratios, and eliminated twenty of them.”
Stevens began with twenty ratios and eliminated sixteen of then in a
factor analysis.”

Financial ratios of a firm tend to exhibit high correlations over

50

time. Because of this correlation, dimension reduction techniques are
able to eliminate many of the ratios without losing much, if any, of the
discriminating capabilities of the model. rl‘he difficult question is
which ratios to use. Critics of the process by which one begins with
many ratios, reducing their numbers to a few, fail to provide a rationale
for using specific ratios, ex ante. mr example, hypothesizing that
above average profitability improves the probability of a firm's
subsequent acquisition necessitates selecting one or more profitability
ratios. Should the researcher select net income after taxes divided by
common stockholders equity (return on equity)? Doing so disregards
asset size (rate of return on assets) , market value of common equity, and
efficiency (rate of return on sales). As a result you must begin the
study with all four ratios to determine which, if any, are most useful,
eliminating any or all of them in a dimension reduction process.

Another example: i-ferger literature suggests that firms likely to
be acquired have excessive liquidity. Which liquidity ratio is the best
to use? Care must be exercised since some evidence suggests that
managements manipulate the more popular ratios.15 That may provide one
explanation as to why neither current ratios nor acid-test ratios failed
to discriminate in Altman's study of bankruptcies. "Of the three
liquidity ratios evaluated, [net working capital to total assets,
current ratio, and acid-test ratios], this one [net working capital]
proved to be the most valuable."16 "The working capital/ total assets
ratio showed greater statistical significance both on a univariate and
multivariate 133,315."?7 “"ost financial texts emphasize acid test and
current ratios.

Finally, arbitrarily deciding which ratios to use ex ante may

51
reduce the utility of an MDA's most desirable feature, its ability to
classify data into subsets by analyzing its ratios simultaneously. The
two most commonly used dimension reductions are factor analysis and
stepwise regressions. Both of these techniques have several
variations. The Statistical Package for the Social Sciences (SPSS) [12]
has five variations of Stepwise methods. 'Ihe SPSS factor analysis
package contains at least nine combinations of factor analysis. Brief
discussions of fector analysis are in SPSS [12] and Greene [7]. An
extended discussion is in Harmon.[10].

An important part of this study will be to use both procedures as
part of an attempt to reconcile the differences between S!" and Stevens'
findings. One explanation for these contradictions could be the
different treatment of the multicollinearity problem in the studies. It
is not known whether differences in financial ratios due to differing
statistical procedures will: 1) weaken the general worth of the
financial model, or 2) diminish the userIness of'one of'the dimension

reduction techniques .

1.

2.

8.

10.

Footnotes

Chapter III

Melvin O'Connor, "()1 the Usefulness of Financial Patios to
Investors in Common Stock," The Accounting Review 48 (April,
1973): 351.

William H. Beaver, "Financial Ratios As Predictors of Failure,"
Journal of Accounting Research 4 (Supplement, 1966): 81.

Ibid, p. 102.

Ibid, pp. 79—81.
Ibid, p. 100.

John Neter, "Discussion of Financial Ratios as Predictors of
Failure," Journal of Accounting Research 4 (Supplement, 1966):
114.

James O. Horrigan, "Ellie Determination of long-Term Credit Standing
with Financial Ratios," Journal of Accounting Research
4 (Supplement, 1966): 48-49.

Ahmed Belkaoui, "Financial Ratios as Predictors of Canadian
Takeovers," Journal of Business Finance and Accounting 5 (Spring,

1978): 95.

Michael Simkowitz and Robert J. Monroe, "A Discriminant Analysis
Function for Conglomerate Targets," 6 Southern Journal of Business
(November, 1971): 14.

Donald lee Stevens, "A Multivariate Analysis of Financial
Characteristics of Acquired Firms in Industrial Mergers," (Ph.D.
dissertation, Michigan State University, 1972): 116.

Ajit Singh, Takeovers; Their Relevance to the Stock Market and
the Theory of the Firm, (Cambridge: Cambridge University Press,

 

1971). p- 91.

52

11.

12.

13.

140
15.
16.

17.

53

Robert A. Eisenbeiss, Problems in Applyng Discriminant Analysis

in Credit Scoring Y‘A’Todels (Federal Reserve Bulletin, [197fl),
p. 16.

Idem, "Pitfalls in the Application of Discriminant Analysis in
Business, Finance and Fcononics," Journal of Finance 32 (June,

1977): 885.

 

Simkowitz and Vonroe, "A Discriminant Analysis Function," pp. 9,
13’ 14.

Stevens, "A Multivariate Analysis," pp. 152, 154, 157.

Beaver, "Financial Ratios as Predictors," pp.78—79.

Edward I. Altman, "Financial Ratios, Discriminant Analysis, and the
Prediction of Corporate Bankruptcy," Journal of Finance 23
(September, 1968): 594.

Ibid.

 

Chapter IV

Data Analysis and Sample Design

A . Data Selection

 

This study investigates a sample of one thousand four hundred and
fifty five companies which have been divided into four subsets of data.
’Ihe first subset consists of two hundred thirty five firms which were
acquired in the time period 1970 through 1976. This data was extracted
from the Compustat Research File, a computer accessed data base
consisting of up to twenty years of financial information pertaining to
firms removed iron the Compustat Industrial File [133]. Del istings from
the Industrial File are caused by any of several financial events; the
appropriate event here is the acquisition of a firm which, prior to the
event, was included in the Industrial File.

The Industrial File is the source of the second data subset. It
is also a computer accessed data base providing up to twenty years of
financial data for several thousand publicly traded firms.

The second data subset consists of three hundred twenty three
companies. This subset, the control group, was randomly sampled and was
from the same time period as the data from the first subset.

The only modifications to the randomly sampled control group came
as a result of two problems.

1. Because the control group was of the period 1970-1976, some of

these firms were subsequently acquired. The control group has been
514

55
verified to ensure all merged firms have been deleted as of July, 1981.

2. Forty eight1 companies were removed from the randomly sampled
control group because their industries employed accounting methods which
were incompatible with some of the ratios used in this study. These
industries included banks, savings and loan institutions, insurance
companies, and utilities. Consequently, a deficiency in this test is
its inability to be generalized to include industries whose accounting
definitions are inapplicable to some of the thirty seven financial
ratios used in this test. Any test of those industries will require new
ratios which will conform with those accounting procedures.

The third subset is a sample of fifty two firms whose mergers
occurred in 1979. Information obtained from this subset is an important
contribution to this study for several reasons:

The fourth subset is a sample of eight-hundred—forty-five non-
acquired firms, used as a control group. Their financial data preceding
1979 is analyzed as a contrast to the third subset. The third and fourth
subsets provide a. test of the predictive power of the discriminant
fumtion generated from the combination of the first two subsets. Both
the Stevens and SW studies recommended updates to test their model and to
determine if the same ratios would be useful in any time period.

Second, rarely in FDA literature are there ex ante tests of a
discriminant function. Generally data are separated into two groups,
the first group is used to generate a discriminant fimction, and the
latter group is then classified, expost. Then the model is then
evaluated as to its predictive powers in an expost test. Joy and
Tollefson and others have criticized such previous MDA applied studies

stressing: "Ix post discrimination may provide a useful foundation for

56
explanation of the past, but it does not provide sufficient evidence for
concluiing that the future can be predicted ."2

Finally, this study evaluates ratios cross-sectionally over time,
armually for three years prior to the merger. One of the objectives of
this study is to determine if a few ratios will be sufficiently important
to appear in the discriminant functions from all three years prior to a
target firm's ultimate acquisition. Three years is suggested because
management may need considerable lead time to restructure its firm's
financial characteristics to fend off potential acquirers, and three
years also acknowledges Belkaoui's results [34].

There are other reasons for reducing the number of financial
ratios used to isolate merger candidates aside from the principle of
parsimony. Some articles have suggested that firms have obvious
financial characteristics which make them vulnerable. "Fanagement
should realize: many if not most of the takeovers or tenders could have
been foreseen by looking at the victim's published financial data."3

Vance claimed to have predicted seventeen of twenty one mergers by
using four ratios. His ex post study focused on price-earnings levels,
net working capital to total assets, long term debt to net worth, and
earnings per share growth rates.4 A deficiency of his study is that he
did not test his model ex ante.

Other researchers disagree with Vance, advocating other key
ratios, or that the merger selection card idate process cannot be
generalized, in other words, each merger has unique aspects and
therefore may not be easily predicted.5

A second objective of this study is to identify trends in the

importance of ratios over the three year span prior to a target's

57
acquisition.

Finally, this study will comment on the timing and significance of
the appearance of key ratios, update previous tests in other eras, and
resolve the contradictions in the literature.

Twenty accounting measurements have been selected, includirg such
items as sales, depreciation, long term debt, earnings per share, and
market price for the common stock.

The twenty items were selected because of their expected
importance in measuring the attractiveness of a potential acquisition.
Using this accounting data, the thirty seven financial ratios identified
in Chapter III were then calculated.

Because of the unique nature of this test, the cross sectional
ratio analysis, three, two and one year(s) prior to a merger, a computer
program written in Ibrtran IV was used to retrieve the accounting data
from the first two subsets and calculate the thirty seven financial
ratios.

Those ratios were accessed from a data file and, using a standard
packaged program, were used to produce three discriminant functions.
The discriminant analysis program was Subprogram Discriminant, written
for the Statistical Package for Social Scientists (eras) [12].

A necessary assumption in the use of FDA is that the populations
are multivariate normally distributed. If the two groups in this
particular case have different means (centroids), but identical
variance-covariance matrices, then a linear multiple discriminant

analysis provides an optimal solution to the classification problem.

Figure 4—1 is an elementary graphical illustration of a two group

analysis. A and B represent scatter diagrams of two groups, merged and

Figure 4-1

 

 

 

/
' l
A /
I B' l
D
ISCRIMINAN
T

FYI
NCTION Z

Graphical Illustration of Two-Group Discriminant Analysis
from Hair, et al [ 8].

58

59

nonmerged firms. In addition we have two measurements, X1 and X2 for
each member of the two groups. The ellipses drawn around groups A and B
usually enclose a predetermined prOportion of the observations,
generally 95% or more of each group. Because the ellipses usually
overlap, the objective is to draw a straight line which minimizes the
amount of overlap. The new axis Z expresses the two-variable profiles
of groups A and B as single numbers, the discriminant scores.

By finding a linear combination of the original variable X1 and
X2, we can project the result as discriminant Z scores on a single axis.
For a lengthier discussion of this see either [2] or [7].

It is important to emphasize that unless the variances and
covariance matrices between the two groups are equal, you cannot
construct a straight line to separate the two groups, because the
ellipses, also referred to as centours, will lack equal shape and
orientation, and unequally shaped centours may weaken one' s conclusions.

Inability to create a linear discriminant function suggests use of
a quadratic discriminant function. Several authors have countered that
large sample sizes will make linear discriminant functions robust to
violations of the equal variance-covariance requirements.6 In this
particular study the requirements are subjected to that debate.

The null hypothesis: Ho: “12 = 022

Results shown in Table 4-1 Show that the variance-covariance

matrices are clearly unequal.

Table 4-1

Fisher's Box 1"! Test
for Equal Covariances

arm STATISTICAL SIGNIFICANCE
LAG 2 (t-5) .oooo

IAG 1 (t—2) o.

LAG o (t-1) 0.

Table 4—1 results were obtained through the Statistic 7 of the
SPSS Subprogram DISCRD-‘DEART.

Given the large sample sizes, and their supposed contribution to
the robustness of violations of the equal covariance requirement it was
decided to temporarily disregard these Box FF scores and assess their
importance in the conclusion of this paper.

A more important statistical procedure is to split the total
sample into two groups. One group is used to generate a discriminant
function to be used to classify the members in the second group, the
validation group, because use of a single combined sample imparts an
upward bias in the overall classification success. In this test, the
firm' 3 Cusip number dictated the group. Bren numbered Cusip firms were
used to produce a discriminant function, and odd numbered Cusip firms
were then used as the validation sample in this split sample technique.
Readers may wish to read W'orrison [99] Frank, Tassey, and T-"orrison [57],
Joy and Tollefson [70] or Fisenbeiss [52] for additional discussion on

the problem of bias.

6O

Table 4—2 (a)

Classification Results - Hold Out Group
Split Sample Analysis and Validation
IAG 2 (t43)

 

 

Actual Predicted
Tie rged Eon-“ferged
1! of cases it (a) r (d)
Then Verged 101 74 (73.3) 27 (26.7)
Analysis
Group
mien lTon—‘Terged 150 39 (30.0) 91 (70.0)
total classification = (n11 + n22) / n.. = 71.49?
Acttel Predicted
Odd lierged 105 55 (52-4) 50 (47.6]
validation
Group
Odd Non-Merged 128 41 (32.0) 87 (68.0)

total classification = (55 + 87) / 233 = 60.9":

61

62

Results between the two sets of tests are consistent with the bias
discussion. The reader should note the lower total classification
success produced by the odd numbered merged firms in the validation
group when compared to the even numbered merged firms in the analysis
group (52.4% vs. 75.5%). Also note the slightly lower total
classification success, 60.991, in the validation (hold out) group
compared to the analysis group, 71.4%. The 71.4% produced in the
analysis group incorporates the bias inherent in the process of usiru: a
sample twice, once for a discriminant function, and then using the same
data in a classification process. The actual total classification
success is the 60.9%, the success of classifying observations not used
in the generation of the discriminant function.

The next step in the classification process is to determine if the
60.9% could have been obtained by a random process. There are three
methods used to test for statistically acceptable classification
accuracy. Total classificationfinn + n22) / mg] is one of the three
measures of classification, and is considered to be superior to the
other two measures, maximum chance and proportional chance
classifications .

Maximum chance is a naive model which assigns all observations to
the largest group, which in this study is non—merged firms.

Maximum chance in the IAG 2 model in Table 4-2(a) is:

[(n21 + n22) / n..] =[(41 + 87) /255]
= .549

frequency of non-merged firms
The maximum chance procedure avoids the central issues of

identifying both merged and non-merged firms. It may be useful if

63
sample sizes are clearly unequal, but that is not the situation in this
part of the study.

The other classification measure, proportional chance, does
attempt to identify merged firms. (loservations are randomly assigned to
either group with probabilities equal to group frequencies. The model
for this, using LAG 2 is
{um,+mp/nnfi+[mm+n2)/mjfi
{[(55 + 50) / 23312 + [(41 + 87) #23312}

0-505

p

The objective is to test total classification success against maximum
chance or proportional chance results.
The test statistic used to establish that a discriminant function

used on a validation sample produces results better than pure chance is:

0-?

p(1-p) ”2
n..,

 

 

where C = total classification success
and p = probability of success
n.. = total sample size
if p = maximum chance, then p =(n2. / n..): .549

or if p = pr0p0rtional chance, then

P =[(n1,/n--)2 + (“2./"")2]= .505

611
Z has a t—distribution with n—1 degrees of freedom.

To test the IAG 2 model first against maximum chance,

 

 

.609 - .549
Z = = 1.8404
(.549) x (.451) 1/2
233

and

t(120)(.05) = 1.658
t(°°)(.005) = 1.645
t(120)(.025) =1.9e
t(o°)(.025) = 1.96

So the model is significant at least at the .05 level.

Testing the IAG 2 model against the proportional chance model,

 

.609 '- 0505 .104
z = = = 5.175
(505) X (495) 1/2 0328
233

which is significant at .005.

IAG 2 is superior to random chance in classifying merged from non-
merged firms, evaluated either by maximum chance or by proportional
chance classification methods.

LAG 1 and IAG 0 model total classification results were then
tested. They are illustrated in Table 4-2(b) and Table 4-2(c),

respectively.

Analysis
Group

validation
Group

65
Table 4-2 (continued)
4-2(b)
IAG 1 (t—2)

Classification Results - Hold Out Group,

 

Actual Predicted
Her ed Non—Merged
4 of cases # 7%) # (34)
Even Verged 105 75 (71.4) 30 (28.6)
Even Non-Merged 131 42 (32.1) 89 (67.?)

total analysis classification rate = 69.5%

Odd Verged 109 60 (55.0) 49 (45.0)

Odd anr"erged 128 40 (31.2) 88 (68.8)
total validation classification rate = 62.4%

M'aximum chance significance

2.5946;

significant at .001

Proportional chance significance 3.7256;

.0005

significant at

Analysis
Group

validation
Group

66
Table 4-2 (Continued)
Table 4-2 (c)
LAG 0 (t—1)
Classification Results - Hold Out Group

Split Sample Analysis and Validation

 

Actual Predicted
P'er ed Non-Tierged
:14 of cases 21‘ 6(3) # (:17)
Even Verged 101 72 (71.5) 29 (28.7)
Ellen Non-"erged 119 35 (29.4) 84 (70.6)

total classification = (n11 + n22) / n.. = 70.99?

 

Actual Predicted
Odd .‘-"‘erged 108 54 (575.0) 54 (50.0)
Odd lTon-T'Terged 123 33 (26.8) 90 (73.2)

total classification e n11 + n22) / n.. = 62.5%

Faximum chance significance = 2.2542; significant at (.025)

Proportional chance significance = 3.1614; significant at (.005)

67

LAG 2 through IAG 0 models all demonstrate significant success in
classifying merged from non-merged firms. Because of this success
further analysis is warranted.

The discriminant functions generated were then used as an ex ante
predicting model to identify from the third subset the fifty two mergers
subsequently occurring in 1979.

Table 4q2 illustrates the three discriminant functions produced in
this merger study. Overall classification success is 60.94%, 62.4575,
and 62.34% respectively three, two, and one year prior to the merger.

The next step in the process was to establish that an T‘TDA
application would successfully separate merged from non—merged figures.
Table 4-3 illustrates that success. The process used was Method Direct,
an application of Subprogram DISCRBIINANT in SPSS. Method Direct
evaluates all thirty-seven ratios utilized herein, rejecting only those
ratios which do not contribute at all to the classification process. It

is not a dimension reduction technique.

68

Table 4-3

Descriptive Statistics

“Tethod Direct

Discriminant Functions.7

 

Standardized Coefficients

 

 

 

LAG2 LAG1 LAGO

LAGZFR1 -.061 91 LAG1FR1 . 23132 LAGOFR1
LAG2FR2 .33838 LAG1FR2 .48432 LAGOFR2
LAGZF'R3 -. 42686 LAG1FR3 -.44488 LAGO'F‘R5
LAGZFR4 .09824 LAG1FR6 . 31 565 LAGO'TR6
LAG2E‘R5 .07141 LAG1FR8 -. 10632 LAGOFR7
LAGZFR6 . 10458 LAG1FR9 .46799 LAGO’TRB
LAGZTR7 -. 521 a) IAG1FR1O .31837 LAGOTR9
LAGZTRB . 20954 LAG1FR11 -. 05036 LAGOFR10
LAGZFR10 -.38742 LAG1FR13 .08670 LAGOFR12
LAGZFR1 1 .05038 LAG1FR14 -. 45712 LAGO'TR13
LAG2FR12 -. 26097 G1FR15 . 2271 6 LAGOI-‘R1 4
LAGZFR13 -. 08102 LAG1FR16 . 30748 LAGOFR15
LAGZFR1 4 .00406 LAG1FR17 . 22564 LAGO'FI‘R1 6
LAGZFR15 -. 31 61 6 LAG1FR21 .74891 LAG07R17
LAG2FR16 -. 13294 IAG1FR22 . 41 224 LAGOFR21
LAG2FR17 . 17755 IAG1FR23 . 00488 LAGOFR22
LAGZFR21 -. 66853 IAG1FR25 .09492 LAGOFR23
LAG2T“R22 -. 10038 LAG1FR26 -. 67727 LAGOFRZS
LAGZ“R23 -. 281 29 LAG1FR27 .88073 LAGOT‘R26
LAG2FR25 .15351 LAG1FR28 . 14003 LAGOFR27
LAGZTR26 -. 2591 2 LAG1FR32 -. 38629 LAGOFR28
LAGZFR27 .04462 LAG1FR35 -. 21 1 8) LAGOFR29
LAG2’7R28 -. 00841 LAG1FR36 . 23947 LAGOFR3O
LAG-2FR29 -. 33457 LAGOFR31
LAG2"R30 -. 13142 LAGOFR32
LAG2FR31 . 1 1064 LAGOFR33
LAG273‘R32 . 21 876 LAGOI-‘R34
LAGZFR33 .46396 LAGOTR35
LAG2VR34 . 29540 LAGOT’R36
LAGZTR35 —. 10608 LAGO‘P‘R37
LAGZFR36 -.11580

LAGZ‘R37 -.02250

wILrs Lambda8 Chi Square9 (Significance)
Lag 2 leg 1 Lag 0 Lag 2 leg 1 leg 0
.8223 .8740 .8144 85.21 61.29 87.16
( .0000) ( .0000) ( .0000)

69
The three discriminant functions are all successful in classifying
merged and non—merged firms given their Wilks lambdas, Chi Square
Statistics and their resulting significance levels. Therefore, further
analysis, using the data and this methodology is warranted.
B. Characteristics of the Data
The data analyzed in this test came from publicly traded firms
which are, or were previously sufficiently large, to be included in the
Compustat Industrial File at some time during the 1970's.
To illustrate, Table 4-4 compares accounting data between merged

and non-merged firms:

Table 4-4
SIZE CHARACTIRISTICS COTPARISON
Averages and 953’: Confidence Intervals

Between '"erged and Non-lierged Firms

(Isa 0)
F-"erged Non-merged

Sales $195 million $595 million

955% CI ($144w‘5246 million) ($477-$711 million)
Total Assets $143 million $699 mill ion

95% CI ($10045175 million) (55569-829 million)
Net Income Before
Extraordinary Items 3‘6 million 3330 million

95% CI (($40>—$135 million) (($287>—$397 million)

Clearly, merged firms tend to be smaller, measured by sales and
assets. Income tended to be less for merged firms, but there is not a

clear separation here .

Chapter IV

Fbotnotes

The breakdown was as follows:

SIC Code 4911-4931: 21 cases
6021 -6027: 16 cases

6120: 4 cases

6312: 6 cases

3714: 1 case

“aurice 0. Joy and John 0. Tollefson, "0n the Financial Applications
of Discriminant Analysis," Journal of Financial and Quantitative
Analysis 10 (December, 1975): 727.

 

Jack 0. Vance, "Is Your Company A Take-Over Target?" Harvard
Business view, May-June, 1969. P. 93.

 

Ibid, p. 94.

Fbr example, read Greenhill's remarks in "The Profit Potential in
Spotting Takeovers," Business Week, 24 October 1977, p.100.

Ibr example, see William R. ’{l ecka, "Discriminant Analysis," Chapter
23, Statistical Paclgge for the Social Sciences, 2nd ed., ed.
Norman H. Nie, C. ladley Bull, Jean G. Jenkins Varin Steinbrenner
and Dale R. Bent (mew York: 'TcGraw—Hill, 1975): 455.

The IAG 2 discriminant function reports standardized coefficients
(each coefficient is divided by its standard deviation) generated
from financial information three years prior to acquisition of the
firms in the sample group.

Wilks lambda is an inverse measure of the discriminating power of

the financial ratios in the discriminant functions which have not
yet been removed by the discriminant functions.

The Wilks lambda is distributed as a Chi Square statistic. Chi
square significance levels of .0000 are highly significant.

70

Chapter v

Findings of the Study

 

A. Distad's Reduced Dimension Model Using Discriminant

 

Analys is
Aside from the principle of parsimony, application of the

thirty-seven ratio discriminant function discussed in Chapter Iv would
be a cumbersome technique to isolate merger candidates. Therefore a
statistical screening process which removes financial ratios that do not
contribute significantly to the classification process is warranted.
The objective is to reduce the number of variables to a few "key" ratios.
Because there are several dimension reduction techniques, one of the
important aspects of this paper is to determine whether various
dimension reduction techniques produce conflicting sets of best ratios.
Stevens, using factor analysis, and Simkowitz and Fonroe (Sz‘i'), using
FDA, contributed studies of financial characteristics of merged firms
during the 1960's, but the two studies produced conflicting results.
Stevens suggested one possible explanation was the differing statistical
methodology. These methods are two of the most commonly employed
dimension reduction techniques used with discriminant analysis.

First using a stepwise discriminant function, the key ratios for

the 1970's test of the new mergers were reduced to the following:

71

72

Table 5-1

Reduced Dimension Descriptive Statistics

 

Reduced Discriminant Functions, Standardized Coefficients

 

LAG2 LAG1
(fl-3) (1)-2)
IAG2FR3 . 56385 LAG1E7R1 .42015
LAG2FR4 -. 63142 LAG1FR6 . 27666
LAG2FR7 . 56227 LAG1FR9 . 4571 1
LAGZFR10 .30075 LAG1FR15 .52839
LAGZE’R1 2 .30911 LAG1FR16 .30978
LAGZFR15 .33268 LAG1FR17 -. 21650
LAG2TR21 .70744 LAG1FR21.£{)609
LAGZIR23 . 26278 IAG1FR22 .16558
LAGZFR26 . 20596 LAG1FR26 . 21434
LAG2FR29 . 29725 LAG1FR32 -. 34656
LAGZFR32 -. 20991 LAG1T‘R33 -. 55826
LAGZF‘R33 -. 49384 LAG1FR34 . 28398
LAGZFR34 -. 26886 LAG11~‘R35 -.1E338
LAG1FR36 . 21 101
WILVS Lambda Chi quare
lag 2 leg 1 Leg 0 leg 2
.8524 .8615 .8243 (.0000)

 

 

 

LAGO

(t-1 )
LAGOVPJ .36CX)2
LAGO7‘F5 -. 251 37
LAGOVR7 -. 70662
LA COWS -. 27681
1151130789 -. 181 24
LAGO‘R1 4 -.15701
LAGOTTR1 5 -. 21962
LAGOT"R1 7 .86217
LAGOT‘R1 8 -. 42766
LAGOVR21 -. 66200
LAGCVR25 -. 25428
LAGOFR27 -.12971
IAGCFR36 -. 37275
LAGOF‘R37 .2761 7

(Significance)
, 1 leg 0
( .0000) ( .0000)

73
Table 5-2

Distad's Reduced Dimension

 

Classification "atrices

Table 5-2( a)

 

 

 

leg 2
(t-3)
Actual Predicted
Merged lion-
P’erged
.41 (a) 4 (so
Verged 121 (55) 99 (45)
lion—Tierged 57 (21) 213 (79)
Overall Classification Success: 68¢

l'aximum Chance Classification
Significance:
Proportional Chance Classification
Significance:
Type I Frror Rate:
Type II Error Rate:

5.786 vs. 3.373 at (.0005)

7.748 vs. 3.373 at (.0005)
45:13-
21%

74
Table 5-2(b)

Ias 1
(fr-2)
W Predicted
l-"erged Non—F-‘erged
# (%) # (5)
Merged 107 (47) 119 (53)
NOanerged 63 (24) 200 (76)
Overall Classification Success: 63%
1'8me Chance Classification
Significance: 4.02 vs. 3.373 at (.0005)

Proporational Chance Classification
Significance: 5.57 vs. 3.373 at (.0005)

Type I Error Rate: 5395
Type II Error Rate: 2495

 

 

75

Table 5-2( 0)

 

Leg 0
(ft—1)
Actual Predicted
P'erged Non—"erged
,3 (9‘4) (7‘ (£75)
T’erged 122 (56) 96 (44)
lion-'vierged 44 (18) 200 (82)
Overall Classification Success: 70"?

I’aximm Chance Classification
Significance:

Proportional Chance Classification
Significance:

Type I Iirror Rate:

Type II Frror Rate:

7.39 vs. 3.373 at (.0005)

8.51 vs. 3.373 at (.0005)
44%
18>?

76

Use of a stepwise discriminant function, Rao V from the SPSS
Subprogram DISCRBTIFAITT has produced a set of three models, LAG 2
through LAG 0, and variables which have been reduced to thirteen,
fourteen, and fifteen respectively from the original thirty—seven
ratios. All three models are significant at very high levels (.0000).

The next procedure is to employ a factor analysis, the version
used here was the SPSS Subprogram FACTOR.

B. Distad's Reduced Dimension lo’odel Using Factor Analysis

 

The objective of a factor analysis is to explore the
possibility of dimension reduction by constructing a set of new
variables on the basis of interrelations exhibited among the original
thirty seven financial ratios included in this study.

Using factor analysis, Stevens' original variable list was reduced
from twenty ratios to four or five ratios. Che objective of this paper
is to attempt to update his work using factor analysis as a comparison
with an I'IEDA originated dimension reduction process.

The factor analysis process transforms a set of variables into a
new set of variables which are uncorrelated with each other. The new
set of variables is a linear combination which accounts for more of the
variance in the data than any other linear combination of variables.

The general model is:

Zj = 83:11:11 + 83-21312 '1' o o o + ajn FYI

Where: j = 1,2,...,n observed variables
Z = variable j, in this case FRJ-
Rn = uncorrected components, each as a linear combination

of the n original variables

aji = factor loading (the standardized multiple

77

regression coefficient of variable j on factor i)

The SE8 Subprogram FACTOR was used for this test, factor loadings
were established using the principal components method (PA1), and these
factors were rotated by the varimax method. The principal components
process was selected because it is most commonly used; Cooley and Iohnes
believe the varimax solution is superior to other rotation procedures.1

Table 5-3(a) is the IAG 0 factor rotation and percent of variance
accounted for by each of the factors. In addition the table presents

the cumulative percent of variance by the sum of the factors. Tables 5-

3(b) is the IAG 1 analysis, and 5—3(c) is the IAG 2 analysis.

78

'T‘able 5-3(a)
Lid 0 (t-1)

Factor Analysis, Tigenvalues and Rercent of Variance

FACTOR FIGTVALUT PC? 0'" WP. CUT PCT
1 7.26371 19.6 19.6
2 4.82608 13.0 32.7
3 3.73049 10.1 42.8
4 3.12730 8.5 51.2
5 2.24859 6.1 57.3
6 2.19263 5.9 63.2
7 1.68810 4.6 67.8
8 1.49703 4.0 71.9
9 1.28674 3.5 5.3

10 1.26364 3.4 8.7
11 1.C6464 2.9 81.6
12 .97089 2.6 84.2
13 .88086 2.4 86.6
14 .85555 2.3 88.9
15 .77580 2.1 91.0
16 .62819 1.7 92.7
17 .60739 1.6 94.3
18 .44395 1.2 95.5
19 .38138 1.0 96.6
20 .31033 .8 97.4
21 .2249? .6 99.0
22 .17704 .5 93.5
23 .15511 .4 98.9
24 .11216 .3 99.2
25 .10357 .3 99.5
26 07408 .2 99.7
2 .06516 .2 99.9
28 .015fl .0 99.9
29 .01001 .0 99.9
30 .CO788 .0 100.0
31 .00414 .0 100.0
32 .C0331 .0 100.0
33 .C0202 .0 100.0
34 .00119 .0 100.0
3 .00025 .0 100.0
?6 .00007 .0 100.0
37 .00006 .0 100.0

VAC 6" 0‘?

\3‘»3CD\]O\\fl-¥>\Nl\)-‘

_.b_s—A
\NN"

._s_s_s
ONU'l-b

<1

I

18
1Q
20

21

22
23
24
25
26
27
2s
29
30
31

32
'37.
34
35
36
57

“ICVWALTT‘T‘

6.69565
5.46766
3 . 52596
2.7759
2.25127
2.12020
1.77167
1.67231
1.59537
1.24795
1.05129
.06539
.9071?
.77319
.68712
.54035
.50140
.44255
.4oo44
.31206
.29547
.2195;
. 1 7089
.14015
.12760
.0861?
.04902
.03965
.01334
. 00692
.00601
.00664
.00244
.00212
.0007:
.oooo4
0

. Vqu.

79

“able 5-3(b)

LAG 1

-"-‘-*-‘-*I\).'\.)1\)fO’xN-13~P->\n-17\-J\O4>;D
(300C)IDfDOOOA-‘N‘Nh-‘fi-x] U’.D-‘1\)—1>‘~J1‘O-*F\)C7\CDO\\NU‘I ‘Ofl-P-U‘lkflfll)‘

O O O O O O 0

19.1
32.8
42.4
49.9
56.?
62.0
66.8
71.3
75.6
79-2
92.1
94.7
96.9
90.0
no a

I‘JO-J

92.3
93.6
94.8
05.0
06.8
97.6
98.2
09.7
90.1
00,4
09.7
bd.e
99.9
60.0

100.0
100.0
100.0
100.0
103.0
100.0
100.0
100.0

VACMOR

KOCOflmU'I-FAKNNA

10

FIGS: WALUE

5.95461
4.56502
5.40170
2.95920
2.12170
1.85964
1.63050
1.49544
1.48218
1.23012
1.15165
1.07078

.97279
.84283
.74254
.69299
.61199
. 53597
.49777
.43507
.4060?
.38518
.27 ,3
.19460
.16506
.14390
.13587
.08600
.C4726
.04275
.02412
.00369
.00216
.00056
.00038
.00008

PCT 07 VAR

16.1
1

OOOOO—h—b-imaaammo-émwaqkoow O\-\]'~-O-*\.NOO-1>O<l-*~P-CD

—I_L_s_§_s_b..§ml\)1\)f\)1\)\fl\N-h-P-1>\nmwkoA

A—L—A—A—hu—b

CU’T'. PCT

moxmmhw NA

'J~1>O<J\-*\:1-J~Jm
O

L
o

59 :3 9)“ j

TO;
JJ
OOOOOO'~D1’T)\]J>-*-\lN-Q-‘OO'ID-P-O‘JJ‘J4> I\)U’1=.OO\.OU‘IU‘|C1—*-*~>\)J\O-‘

8888 8 88838118858859153188237895853891.

81

The purpose of Table 5-3 is to show how many factors are necessary
to explain the variance of the data. Hair et al have suggested an
arbitrary cut off when sixty percent of the variance has been
explained.2 ”he SPSS default procedure is to cease factor rotations
when eigenvalues fall below the one level, which is consistent with
Kaiser,35 the inventor of the varimax rotation. In this study any "cut
off" procedure indicates a large nunber of factors are necessary. Hair
et al's process necessitates six factors, and S 03 procedures advocate
twelve factors. Ecause factor 1 explains more of the variance than any
other factor it is considered to be more important than any other factor.

A portion of factor 1 is illustrated in Cable 5—4. r[he entire factor 1
list is in the appendix.

"able 5-4

LXG 0
Factor 1
T"actor loadings > H.3l
Distad's Tes-t

Ratio load ing
F82 . 99227
“F3 . 9821 6
T"R4 . 98208
YRS . 97058
77129 -.94864
FR30 -. 92035
2831 -. 96235
FR34 -. 63372

Stevens used a cut off of 137' or more in his factor analysis.4

Hair et a1 endorse |.5| as a cut off.5 Others contend a slid ing scale,

 

wherein factor loadings of |.7| are extremely important, |.5 or more

are important, and factor loadings of 1.3I may be significant. Cooley

'I)

2
and Iohnes have used a cut off of; l.35|.6

I‘ach factor is arranged in a row vector, and each of the financial
ratios in that row vector represent the loading of that ratio on that
factor. Therefore, the relation between any of the variables and any of

the factors can be observed. F‘ach row vector has the relationship that

:11

correlation coefficient. In Table 5-4 F92 = am because it is in the

N
jz: (aj i)2 = 1 where each a-- has an analogous interpretation to the

first factor and it is the second variable; 1791 's loading = .05327 and
was not included in this excerpt, that is its coefficient was too small .

Given that 792 has a loading of .99227, we know that (£19227)2 =
.96485, and 96.5 percent of 192's total variance is explained in factor
1.

Table 5-4- serves another purpose. It shows that factor1 includes
nine of the thirty—seven original ratios. In spite of their high
loadings, those nine ratios account for only approximately twenty
percent of the variance, (see Table 5-3) . "be second factor, primarily
a set of four lever%e ratios, accounts for only an additional thirteen
percent of the variance. A factor analysis will not successfully reduce
the variable dimension in this particular application, no matter which
cut off method is used. Using Fair's suggested cut off guide
necessitates six factors, but six factors only reduce the dimension to
thirty from the thirty-seven ratios. Using T’aiser's cut off, twelve
factors, we explain eighty-four percent of the variance, but doing so
requires thirty-five of the thirty-seven original ratios, using loadings
3 I..5|

C. Analysis of Stevens' Factor Analysis Procedures

 

Stevens concluded he could reduce to ten ratios, the original

83

list of twenty. However, in observing his rotated factor matrix,7
several ratios with factor loadings in excess of |.7| were omitted. T'e
still had fifteen financial ratios with factor loadinqsz |.5| after his
first five rotations.

Though the cutoff process is quite subjective, he may have
disregarded important financial ratios. .In his first factor, his factor
loadings in excess of_>_ [.5| were ‘98 (.919), long term debt to market
value of equity; T915 (.962), interest to cash plus marketable
securities; T“P31 (.950), net working capital to sales; and T910 (.020),
long term debt to total assets. T-’rcm this list he omitted T98, and
thereby discarded a ratio with a loading of .819, which most analysts
would consider to be extremely important.

Stevens discarded his entire second factor in spite of its high
factor loading scores. "There the key ratios were T92 (.890), gross
profits to sales; I93 (.946), income before taxes to sales; and 7194
(.899), net incone after taxes. Instead, from the third factor he
selected two of three profitability ratios though third factors are less
significant than are second factors. In the third factor, the key
ratios were T"R1 (.895), earnings before interest and taxes to total
assets; F95 (.823) earnings before interest and taxes to sales; and F96
(.969), net income after taxes to net stockholders' equity. Of those
three, he omitted F95 (.923) presunably because it had a lower factor
loading. While these 1c 2d ings are high, they are still less important
than the omitted variables from the second factor.

Ebr activity ratios he selected 799 (.897), long term debt to
stockholders equity; T914 (.974), sales to quick current assets; and

3913 (.851 ), cost of goods sold to inventory. ‘-'ost analysts consider

811

789, long term debt to equity to be a leverage ratio. 1‘t'ore puzzling
though is his decision to incorporate 71714, which did not appear until
his ninth factor; his cut off was six rotations.

His liquidity ratios were net income to total assets
(profitability ratio) and sales total assets (activity ratio). r"he two
previously defined liquidity ratios, from his table3.1 on gave 78 of his
dissertation were TF3241 (.880-tenth factor) , net working capital to total
assets and W31 (.050 — first factor) net worldnq capital to sales, which
he included in his list of leverage ratios.

”able 5-5 is a reproduction from page fifty two of Stevens'
dissertation. It is his summary of factor analysis; the percent of

variance by factor .

r"able 5-5

Stevens' Factor - Percent Variance Table

factor Percent variance Cumulative Percent
1 21.79 21.7
2 17.40 39.20
3 13.28 52.49
4 7.27 59.77
5 5.61 65.38
6 5.31 70.69
7 5.23 75.92
8 5.21 81.14
9 5.18 86.32
10 4.87 91.19
(11-19 anitted)
20 0.00 99.99

The purpose of Table 5-5 is to compare Stevens' Tactor Analysis
produced percent of variance results with this researcher's, which are
shown in Table 5-3(a) . Clearly it now takes more factors and ratios to

use factor analysis in a financial ratio analysis of merger candidates.

85
Additionally, it now requires more factor dimensions and therefore more
ratios to attain the level of cunulative variance explanations obtained
by Stevens. '1‘o illustrate, r.T‘able 5-6 is a listing of factor loadings by
factor for the first seven factors on my data. It was decided not to
list all twelve factors, though seven only accounts for sixty—eight

percent of the variance as revealed in Table 5—3( a) .

86

mvw.

hmh.
mrh.
mmm. **Nmm.
Nwm. **mcm.
wow.
emm.
0mm.l
mmm.
mam.
mom.
Pea.
mmm.

e 888 s .8me m 888 e .8988 n 829... m Spawn...

 

AFIPV o w<H

888 mm 838
assesses has. $888 888
$8. Poe? magma

sum 286

F

set
so:
cs...
as...

as.
Nmo.

.Upufip

mm
Fm
mm
mm
em
mm

m—

ON

or
a
a
9.
ON

mmfi

3 Em

87

$8838 8 883.88 8.85
8.8 as 85“ Em .88 8888 8.8 8 82 was Ea as:

.88. 98 8.8.3.5.988
8888 8.888 meats. .SE .8.“ 885

was cogmmoxm cm. “96%“ Egan mam. m. mmcflcmoa povomw
So.
.08.
PE.
Em.
met.
CE.
*ECV.
$8.
mmm.

$88388 mum 035

be
we
@—

mm
#1
NF
hm

88

"‘able 5-6 shows that with only seven factors (yet only sixty-eight
percent of the variance explained), there still are thirty ratios. r“o
further reduce the nunber of ratios would be unwarranted given their
high factor loadings. The problem also exists for LAG 1 and IAG 2 factor
analyses (see 'T‘ables 5—7 and 5-8 for an illustration). T?ecause of these
problems, it was decided to abandon factor analysis as a possible

dimension reduction technique in this specific application.

89

8 8.

mammo.
mummm.
088m».
ommmw.
- spams.
meson.
emmmc.
memom.
memm.
mmmrm.

b 8080mm 0 povofiw m povoma e povowm m aovofiw

 

Amlpv F m<H

8”? one. am 838
Hmwocmam pow manauon popes?
same m.orov wwpmwm

bum manna

omewm.l ma

scape.
mambo.

C

m

b

m
mmomm. m
8

w
Newbm. m
P

m 88.88 . 80.8% 88m

9O

ewmmb.
Odwmh.

hmmqm.

.m. mo msam> oesfiomom cm 28:9 mmmH mmcficmoa
888a 8: a8 8.1% 88 .8 .919 838 8888

N&¢Om.

emmew.
memm.

$88808 sum 828

FOmmh.

"meom*

:88...
M88...
Nmmm®.l

rm
mm
Fm

om
rm
mm
mm
8N
mm

91

.m. cmsp pm3oH mum: mmzam> mpSHomLm mmo:3 chHcmoH

“Sufi E; Sm mmumm .mm .mm .a Eng 6 83.8 Eofidfi*

mmm—w.
mmmmm.
NQme.
vm¢bw.
Mama.
Frmmo.
mvvmo.
mvvvd.
mmfifim.
Ovmvm. _
mmwmw.
wmvmb.
wmmmo.
hmVON.
momma.
Oqum.
waCh.
dvmmo.
mmm¢m.

b 883 m SSS m “Sufi q “88,” m 85mm m #8th

 

Ami: m 93

8330a 88% an
83mm Ragga you mwcfl E8 “Bump
$3 99.9 $me

mlm mach:

oomhm.
mmm¢m.
mmhwm.
wOOmo.

hm
mm
Fw
Om.
*mm
hm
mm
mm
*Vm
*mm
ON
0F
#3
3
*OF

PNK‘O'U'LOKOP‘CD

mm

F bpofl 03mm

92

There are two userl discussions remaining; however: 1. To
compare factor characteristics from my 197 '8 data test with those
obtained by Stevens, and 2. To contrast factor analysis results with

“TDA acquired results.

D.

Factor:

1

93

Distad—Stevens Comparisons of Results Usingg‘actor Analysis

Table 5—9

1. Stevens' Factor Characteristics Compared
toDistad's Factor Characteristics

LAG o (t—1)

SmEVETTS (1960's data) DlsmAD (1970's data)

leverage

Profitabil ity
Profitab il ity
Profitahil ity
Activity

Ac tiv ity

leverage

P/E‘ Patio

Profitability

Liquidity

F’arket Price to Cash Per Share
leverage

Profitab il ity
leverage

Liquid ity
Activity
Liquid ity
Profitab il ity

P/E and Dividerd Payout

914

”able 5-10

2. Final PTA and Factor Analysis Vey Patios
Compared, (From Tables 5-1 and 5-6.)

IAG O (t—1)
Standardized
Discriminant
Ratio Function Factor Rotation Tactor loading
VR1 +.36O 6th .76s
5 -.251 1st .090
7 -.707 6th .877
8 —.277 2nd .555
9 -.181 3nd -.959
14 -.158 5th .497
15 -.220 Ifid not appear
16 -.942 7th .?39
17 +.862 7th .007
18 -.428 6th .877
21 -.662 Did not appear
25 -. 254 4th/5th .522/ .588
27 -.130 4th .787
36 -.373 Did not appear
57 . 276 5th . 558

The key Observation from mable 5—9 is that leverage and
profitability’ ratios are the most important financial variables.
Activity ratios do not appear to be of import in the t-1 models. P/F
ratios and dividend payout percents were of comparable levels of
hnportance in spite of the time gap between the two studies, but are of
greater import than indicated by Stevens.

'T‘he second discussion is to contrast results derived from a factor
analysis with those derived from a.discriminant analysisrnodel. Table
5-10 uses 1970's data tests performed by Distad and are unrelated to
Stevens' earlier study.

From the preceding analysis we see that while "DA is of only

limited significance in dimension reduction, it is clearly superior to

95

factor analysis in this application. factor analysis was unable to
significantly reduce the nunber of variables. Of perhaps as much
importance is the extent of contradicting results of important variables
selected via discriminant functions as Opposed to the variables selected
using a factor analysis, as illustrated in “able 5—10 on the preceding
page.

E. Replication of the Stevens liodels

 

After studying merger activity in the Sixties, Stevens cited
the need for a follow-up study in a subsequent time period to determine
if the financial ratios which reflect the attractiveness of a target

vary over time.8

r:"he purpose of this section is to review his study and
to provide the follow—up study.

He conducted two separate tests in his dissertation. “be first
discriminant function was an ex post test of forty mergers. From an
original twenty ratios, Steven's reduced to four, the nunber of ratios in
his final model. His discriminant function was:

where the coefficients have been standardized and,

("327) X1 - Net lncome/"btal Assets

(W12) x2

Sal es/ Total Assets

(7331) x3 T‘Tet Vbrlcing capital/sales

(mo) x4 = Long Term Liabilities/“Votal Assets, and
Xi’ i = 1-4 are ranked in ascending order of imlcortance.g

Stevens' classification matrix was:10

96

"‘able 5-11

”DA: Classification Vatrix. Observed Groups

Actual Predicted
Acquired Ton-acquired
< «s > .4! < . )
Acquired (85) 33 (X11) (17) '7 (X12) = 40
lion-Acquired (52) _2_1_ (X21) (48) 12 (X22) = 9.2
54 2.6. .89.

His overall classification success was 65¢.

The Stevens replication conducted here was not as successful.
Those results are in Table 5-12.

Three separate classifications were conducted, using financial
data three, two, and one year prior to the merger. Overall
classification scores were 54.4, 53.4, and 52.1”: respectively, all

significantly below his 65%.

97

Table 5-12
Distad's Replication of Stevens' Coefficients

(Equation 5-1 )

t—5 t—2 t—1 Stevens
Observed Groups
(LAG 2) (IAG 1) (IAG 0) (t—1)
Ratio PR7 -1.004 .599 .52? X1 = .005
W10 - .835 .805 .667 X4 = .351
FR12 - .258 -.158 «299 X2 = .934
Overall
Classification
Success Rate: 54.4407 53.36“? 52.05"? 65%

One of the four most critical differences in the Stevens
comparison is the lack of uniformity in the signs of the coefficients
over time. A second difference is in the coefficients of the Stevens
model. Why are the signs of the coefficients different in sign and
magnitude? For example; consider sales to total assets, X2 (TR12) . As
Stevens indicates:

This ratio is an indication of how efficiently the assets of

the firm are being employed with respect to the generation

of sales. ’Ihe higher the ratio value, the greater the

efficiency of the assets in producing sales. "he higher the

ratio value, the greater the efficiency of the assets in

producing sales. If acquired firms were systematically

inefficient, with excess liquidity and under utilized
assets, ratio X2 would be lowgr for the acquired group than

for the non-acquired group.

It is puzzling to encounter the sign and magnitude of the
coefficient for his sales to total assets ratio (+ .954) . A researcher
would hypothesize an absolute value somewhere near zero if the ratio was

of limited significance, and would certainly hypothesize a. negative sign

if a weak ratio was significant. Ratios in my updated Stevens test were

98
negative for all three years, cross—sectionally, prior to the merger,
and their absolute values were not extremely large, thereby confoming
more to a hypothesized coefficient.

The third critical difference is the difference in the
classification ratios' overall ability to properly categorize the two
samples. Stevens produced a 657‘ classificaion rate, but using his
ratios 1 was unable to repeat his success using data from another time
period. Fbr a comparison, refer again to Table 5—12.

The fourth critical difference is in misclassification costs.
(he such type of misclassification appears in Stevens' classification
matrix in rfable 5—11, and are summarized in Table 5-14.

Tlisclassification errors in this matrix format are separated into
two type of errors (X12 and X21). A Type I error, rejecting a correct
hypothesis, is the incidence of acquired firms being misclassified as
non-acquired firms (X12) . Stevens, using two samples of 40 firms each,
correctly classified ‘35??? of the acquired firms, (55/40). Hence,
Stevens' model is successful in this respect with only a 17% failure
rate. P’fy attempt to replicate his model was not as successful. Refer
to Table 5-14 for a comparison.

(he economic effect of a Type I error is that arbitrageurs are
deprived of an opportunity to achieve the risk adjusted superior returns
which accrue to stockholders.

A second effect of a Type I error is that managers do not perceive
their firms to be likely candidates and are unable to improve the
Operations of the firm sufficiently to either maximize shareholder
returns or to thwart a takeover attempt. It is difficult to assess the

impact of this type of error because one aspect of it is the incidence of

99

layoffs of both management and labor. Perhaps offsetting this aspect is
that both managenent and labor may be shareholders, and they may be
extrenely well compensated given a takeover attenpt, especially if the
target is the benefactor of a bidding war between two or more potential
acquirers.

A third economic effect of Type I errors is that firms possibly
have been acquired unwisely or that the acquirer was motivated for
reasons which are not anbod ied in the four financial ratios employed by
Stevens. Again, the literature is replete with the incidence of unwise
acquisitions occurring in the Sixties, measured ex-post.

A fourth effect of Type I errors is that firms seeking to acquire
other firms for diversification benefits conclude, erroneously, that the
target will not becone an acquisition and acquirers forego any of the
resulting benefits.

A Type II error is the second form of misclassification error.
Such classification errors occur when a model suggests acceptance of a
false hypothesis. In this application that event exists when the model
incorrectly categorizes a merged firm as not likely to be merged, (X21) .

’lhe major Type II misclassification cost occurs to arbitrageurs.
They incur opportunity costs after omitting funds to the purchase of
shares of firms which were not subsequently acquired. Additionally, if
the misclassified firms demonstrate the hypothesized below average
performance, the arbitrageur is exposed to additional losses in the
market place, resulting from falling share prices.

Here, Stevens' classification success with non—acquired firms was
much lower. His model failed to categorize less than half of the 40 non—

acquired sample, 19./40 = 49-5”), or ()(22/X2 ). And as a result, his

100

Type II error rate here was large, 21 of the 40 (520??) firms not acquired
were incorrectly classified as merger target candidates.

Stevens suggested that though these firms may be candidates for
acquisition, the market has not yet perceived them as such.1'5 Put, the
probability that the market has not yet realized the attractiveness of
the target as an acquisition is reduced given the results of my update of
Stevens' test. Type II errors are shown to exist three years prior to a
merger, and very little reduction of that form of misclassification is
observed over time, according to my replications. T--"y results appear in
Table 5-14.

The effect of Type II errors in Stevens' mode " . . . raised some
doubt with respect to the usefulness of the model in an Operational
sense."14

***************

Recause of the larger percent of Type II errors in his model,
Stevens reformulated his data into "natural" groups of firms that were
merged and those that were non-acquisitions. He referred to the latter
as: " . . . eligible firms not yet acquired."15

That reclassification of his data produced the following
discriminant function. Again, he used the same four ratios, which
unfortunately does not tell us if the discriminant function for his
"natural groups" contains the "best," or optimal, set of ratios.

The new discriminant function is:

"natural": 2 = .531 x,, - .573 x2 + .002 x3 - .619 x);g
(Tquation 5-2) , compared with his

"original": Z = .005 X1, + .954 X2 + .064 X2 + .552 X4 (F‘quation

5-1).

10].

Again, X1 = Net Income/Total Assets (RR?)
:2 = Sales/Total Assets (.1212)
X3 = Net tbrking Capital/Sales (“R/’1)

X4 = long Term liabilities/"otal Assets (Two)

Using Fquation 5-2 provides results such that the coefficients for
X2 and X4 are nearer their hypothesized values. The coefficient for
sales to total assets, X2 (W12) , should be less than average because the
firm is not prOperly utilizing its assets. The result provided from
Iliuation 5-2 is much closer to the coefficient produced in my
replication (in Table 5-12) . rT'he coefficient of long term liabilities
to total assets, (X4, W10), should reflect lower than average debt
levels as a percent of capitalization.

The new classification matrix (Table 5-15) indicates an
improvement in overall success. Though classification success fell to
6394 from 657’ (see Tables 5-11, 5-12), there was a reduction in the extent

of Type I and II errors.

102

r“able 5-1 3

fievens' ITew Classification of Tv'erger Groups17

Actual ‘Tew Groupings Predict
I‘actcr 1 Factor 2
(Likely to be (Not Likely to be
Acquired) Acquired)

(675) f? (‘4)

Acquired ( 67) 26 ( 3'5) 15 30.
Non-Acquired ( 41) 1: ( 59) §_2_ fl
ii 15 16.

Overall classification success is: 65% = 26/76 + 22/76

Type II errors (41¢) were much greater for Stevens as those levels
occurring in my tests, shown in Table 5-14, below.

Type I error levels are much greater in my replications, using his
ratios, on new data.

***************

103

Tabl 8 5-1 4

Conparisons of "‘ype I and II I‘rr'or Levels

Stevens

Initial Study Few Classification

(Table 5-11) (Table 5-13)
TYPE I 1'7?! 320’
Thor (1) (7/40) (1.3/39
Pym II 5274 417?:
Frror (9??) (21/40) (15/37)

by 9.15033
t—3 t-2 t-1

(IAG 2) (IAG 1) (LAG 0)
Type I 8.194 92??! 917’.

Error (%) (189/254) (218/235) (214/235)

Type II 15?“ 795 8’.”

Error (%) (42/275) (152/271) (20/253)

1011
Stevens then ran a third test, generating a new discriminant

function, but not restricting himself to the ratios derived fron his
original test. Tran that third test the following discrim inant function
was calculated:

Z = .052 19+ .163 X2 -1- .079 X3 - .95? X4 - .276 Y5 (unation 5-
'2’)
where: X1 = Dividend Payout (W16)

X2 = Net Income/Total Assets (“3.7)

X3 = Net Working Capital/Total Assets (W24)

x4 = Sales/Total Assets (r912)

- Long rem Debt/Total Assets“? (“910)

>4
U'I
I

It is not especially useful to compare the coefficients of this
function with his two earlier versions. Pecause of the inclusion of the
fifth ratio, the magnitude of the other four coefficients will be
altered. However, only net income to total assets and sales to total
assets of the original four ratios renain in the final discriminant
function. Stevens failed to explain the different set of ratios in this
model. Furthermore Stevens changed his best leverage ratio to long term
debt to total assets from his original long term liabilities/total
assets. His distinction between debt and liabilities is:

The only items included in long term debt were actual

10% term instrument such as bonds and notes. long term

liabilities include in addition, various accounting egtries

such as deferred compensation, deferred taxes, etc. ,

No explanation is given for the change, for the effect of the
change, or why the new ratio was not employed in the original version of

the model. The sign of the coefficient is negative, indicating below

average leverage, which is consistent with his natural grouping

105

coe ffic ient .

“y efforts to replicate his third model were less successful.

Use of those five ratios produced a model which was significant at
the .0763 level (92.34%). In comparing coefficients:

Stevens': Z = .052 X1 + .163 X2 + .079 X5 - .953 X4 - .236 X5
(Equation 5-3)

Distad's Replication of Stevens:

(IAG 0): Z = —.538 X1 — .417 x2 + .361 KB - .025 X4 - .665 X5
(Fmation 5-4)

Again, clearly there are differences in magnitude and signs of the
coefficients.

Stevens indicated an overall classification rate as follows:

106

‘9-

El
E1

rmpmwc

 

 

was wcs 0e
as am Ame J A: static;
mum mam Ass V mm Aoav swaps;
A s v A v V
conpoalcow mommy...
mflflflm “Mfla

schemesasmm m.smpmss

swa soaps as mass

soaks H mass

sac aUmpSoo< scameWwwmmmHo

mco>mpm

 

 

ms mm ms
.mw .Iw A smv .ma A «av s>asssssp<usoz
a. N Am; 9. Ass 838.52
s Asv s Asv
m>aposssp<.soa_ m>ssosspp<
assessmss mmmmmq

figsfifisfififlmgtfi

 

Hmwo; cmPMHSELo%mm..wco>opm

 

w 0.. ,mam EO..®,. wcm>® 2 AOL mQOHL w“... COH wOH Hmmm
F + k4 0 U C. o FF 0 o o

m Tm w Home

107

Clearly, my replication of Stevens' reformulated model was
unsuccessful in identifying likely merger candidates, identifying less
than ten percent of firms which were merged. It also produced a low
total classification score using his model,contrary to the results he
obtained from his test of mergers in the 19605.

IAG 1 and IAG 2 classification rates in the replications produced
a 54¢ overall classification success in both time periods.

In a subsequent article, Stevens introduced a fourth discriminant
function from the same time period.20 In this model,

Z = -.0'53 X1 + .987 X2 + .108 X3 + .111 X4 (Fquation 5—5)
arranged in ascending order of importance where:

X1 = Net Working Capital to Total Assets (r324)

X2 = Sales/Total Assets (r812)
X3 = Sales/ Total Assets (T15)
X4 = Long Term Liabilities/motel Assets (H210)

Only X2 and X4 appear in the discriminant functions generated in
the earlier model versions of his dissertation. In comparing the

coefficients of X2 and X4 fran the article and the dissertation we find:

108

"‘able 5-16

Stevens' Coefficient Comparison

Original TIatural Ti nal Article

 

“"odel "odel l'odel "odel
Ratio (W: 5-1) (170‘: 5—2) (7101 5.5) (’5le 5—5)
X2: Sales/Wotal Assets (32.12) .954 —.575 —.955 .98?
X4: long "‘erm rebt (Liabs)/ .552 -.619 —.256 -.111

Total Assets (2910)

Stevens' models tell us that sales to total assets ratios for
firms being acquired are important, but there is a sharp contradiction
considering coefficient signs. Only in fievens‘ natural (Equation 5-2)
and final (Fquation 5-4) models are the coefficients as expected:
target firms should be expected to be less efficient than those firms not
considered to be as desirable for acquisition. It is possible that
different ratios in the final discriminant function would account for
coefficient magnitude variations, but there does not appear to be an
explanation for the differing signs.

Furthermore, there does not appear to be any reconciling the
positive coefficient calculated for long term liabilities (or debt) to
total assets. In his equation 5-3, the coefficient using long, term debt
rather than long term liabilities was not materially different from his
coefficient in equation 5—5. Again, I am unable to reconcile his
positive coefficient (+5552) in equation 5-1.

The classification matrix from the model appearing in his article

is:

109

rTable 5-1'7

Stevens Vodel (Tquation 5-5)

 

 

 

 

 

A ctual Pred icted
(Pegged [a T:fon—Verged
Verged (PS ) 34 (15 ) 6 = 40
‘Ton-"eraed (45 ) E (55 ) ?_? fig
52 2‘3 = 90
Distad's Replication (LAG 0)
A ctual Pred icted
I‘lerged Non—”arsed
( "’ ) .fi‘ ( "5 )
"eraed (7) 16 (9'5 ) 215: 229
Non-"erged ( 6) fl (“4) 27g 22:
31 451: 482
Stevens Qistad
Classification Accuracy 70¢
Type I Error 154
Type II Error 457'?

110

To summarize Stevens' endeavors, he derived four different models;
the first of which (Equation 5—1), using return on assets (FY37) , total
asset turnover (£0.12), net working capital as a percent of sales (W31),
and long term debt as a percent of total assets (W10) produced an
overall classification success rate of 657’. This first model would have
produced considerably better results if the Type II error level, 52
percent error rate, could be reduced. “‘Iy three replications of his
first model were less successful producing success rates in the 52-54
percent range.

The extent of the Type II error levels in Stevens' first model
necessitated a second model, (Fquation 5—2). His overall classi-
fication success dropped slightly to 63:33 from 6557:. and his Type II error
level rate fell to 4170 from 52%. r‘z‘he ratios used in his second test were
those from his first model, but the regrouping by " natural" rather than
by actual groups produced discriminant coefficients with conflicting
signs and sharp differences in the magnitude of their absolute values.

Stevens' third model (Equation 5-3) was another discriminant
function of the same data; the only differences were the inclusion of an
additional ratio, dividend payout (I816) and a redefinition of long term
debt. In this third model long term debt disregarded various deferred
liabilities including taxes and compensation. ”his third model
demonstrated much higher classification success (92,71) and dramatically
lower Type I (5:19) and Type II (13“?) error rates. F’y efforts to replicate
this third model were not as successful. Y-‘y replication produced a 53”
classification rate, and Type I (9%) and Type II (7’5) error rates. his
model, when applied to my sample in a different time span could not

isolate mergers, classifying only 23 of 235 firms which had already been

lll

acquired. Please refer to Table 5—15 for a more extensive discussion.

Stevens' final model (T’quation 5-5) was constructed for a
published article [125]. Using this model Stevens obtained a "Or-f
classification success. He also attained Type I (1‘5“) and Type II (4555)
error rates. My attempt to replicate his model were less successful
generally with the exception of a lower Type II error rate(6"’-). F'y
replication produced an overall classification rate of 5351’» and a high
Type I error rate (93¢) . In other words, his model classified only 16 of
229 firms which had already been acquired in my sample. A more detailed
discussion of his model and my replication appear in Table 5-17.

It is important to emphasize that any conclusions reached as to
the efficacy of Stevens' models in subsequent time periods is reduced to
the limitations of my sample and replications. l'y replications
indicated that his ratios, their coefficients, and his models did not
demonstrate significant ex post classification success in any instance.
Obviously it is necessary that his models be tested again in subsequent
time periods, as well as mine, to ascertain their significance
econanically and statistically. T'ore importantly, his models were also
unable to predict mergers, ex ante, occurring in a subsequent time span.
A more detailed discussion of that deficiency is provided in Chapter
VII.

F. Analysis of the Simkowitz—T-"onroe "odel

 

At approximately the same time as the Stevens [155] study,
Simkowitz and It'onroe (Sl') completed a study of firms likely to be
acquired by conglcmerates [118]. Pecause the study was restricted to
conglomerate acquirers it has received less attention here than that

given the Stevens Study.

112

Hewever, it needs to be considered because the S” study used a
dimension reduced version of an T"DA model which resulted in a set of
"key" ratios unlike those derived in Stevens' reduced set of variables
obtained using a factor analysis. In a subsequent article, Stevens
wondered if the resulting dissimilar set of ratios was attributable to
the differing dimension reduction techniques.

S“ concluded that four financial ratios were most important.
Listed in descending order of importance, they are: 1. low price to
earning ratios, 2.‘lowwlividend payout rates, 3. lowygrowth.rates in
equity, and 4. lower dollar levels of sales than their acquirer's
sales levels.

One of the chief differences between the SN and Stevens studies
was in the importance of price to earnings per share ratios (p/e). ST
concluded that lower than average p/e ratios were the most important
variable in classifying a firm likely to be acquired by a conglonerate in
the 19603. That is very consistent in the merger literature of that
time as amotive for mergers as well as a characteristic of acquisitions.
Stevens concluded otherwise in his test. His test was not confined to
conglomerates, and nowhere in his four<iifferent discriminant functions
was he able to statistically justify the inclusion of p/e's in his
models.14

In my studies, p/e' 8 (W17) failed to enter the final discriminant
fUnction in IAG 2,'but was -.2165 in IAG 1,and +.9622 in IAG 0.

looking at those standardized coefficients, we see that p/e's are
not hnportant three years prior to a firm.being acquired. Pewever, two
years in advance, p/e's become important, and because of the negative

sign, we know that acquired firms have lower p/e's than non acquired

113

firms. Finally, one year prior to a merger, we see that acquired firms
have p/e's higher than nonacquired firms. One plausible explanation for
a change in the sign of coefficients here is the market has identified
the potential acquisitions and forced up the price of the common stock in
anticipation of a subsequent tender premiun by a potential acquirer.
Another possible explanation for a sharp increase in the p/e's of
acquisitions is a decrease in earnings per share in the year preceding a
merger, with no accanpanying decreases in their stock prices.

Comparing the above to the results obtained in a factor analysis,
the factor loading for p/e's is high (.8328) in the IAG 2 model , but does
not appear until the eleventh rotation. For IAG 1, again, the loading
is high (.7296) but does not appear until the seventh factor, which is
also the case for the IAG 0 factor loading of IAG 0 (.9096). IAG 0 and
IAG 1 loadings are illustrated in Tables 5-6 and 5—7.

I conclude that the strongest case for p/e's comes from
discriminant analysis, because p/e's have limited importance using a
factor analysis. 'Ihe latter can be verified by checking rTable 5-10 to
see how little a reduction of variance is obtained through a seventh
factor.

S" 8 second ratio was dividend payouts (T5216), which was also
Stevens' second most important ratio. They hypothesized that dividend
payouts would be lower for acquired firms than for non acquired groups.
Stevens' results rejected his hypothesis: the sign of the coefficient
was positive, indicating that acquired firms in his study had higher
dividend payouts (.052), as in equation 5-3.

T‘y findings indicate that payouts are not important three years

prior to an acquisition, of only moderately importance (.3098) two years

11”

prior, and very important one year prior to a firm's being acquired
(—.9418). ’T‘he IAG 1 standardized coefficient should be considered more
of an ex post classification. 'Ihe key discriminant coefficient one year
prior to the merger (IAG O) is large and negative indicating acquired
firms do have lower dividend payouts than do non acquired firms, which is
consistent with their hypotheses. In a factor analysis, dividend
payouts do not become significant until the seventh factor, at the
earliest, again connoting little significance. 8"8 third ratio, growth
rates in common equity, was hypothesized to be lower for acquired firms,
which neither Stevens nor I tested. However, I did test return on net
stockholders' equity (7R6) . A discriminant coefficient with a negative
signiLigfi be similar to the Si“ hypothesis. Using V’I‘A, FPS appeared in a
final discriminant function only in IAG 1, with a standardimd
coefficient of .2767 which is not significantly large, but does not
possess a negative sign. Pecause factor analysis results were
inconclusive as well, I conclude that rates of return on stockholders'
equity are not important in identifying potential acquisitions.

'f‘he fourth financial variable, dollar level of sales was not
tested here, but observation of the data in mable 4-3 illustrates large
differences in sales levels. Clearly, acquired firms are smaller when
measured by sales.

G. Conclusion

 

F’y test supports two of the four ratios used by $7. “'erged
firms in my test had lower dividend payout ratios, and had lower dollar
levels of sales. Put I was unable to produce meaningful return on equity
ratio results. Because he tested growth rates in return on equity,

rather thzm return on equity, my test results may not contradict his

115

hypothesis.
Furthermore, the review of actor analysis in this section
suggests that it does produce the differing set of key ratios as

hypothesi. red by Stevens .

12.

13

I C

14.

1C

I O
'

16.
17.
18.

19.

116

I"ootnotes

Chapter V

William W. Cooley and Paul P. Iohnes, '"ultivariate Data Analysis
(“few York: John Wiley and Sons, 1971): 145.

 

Joseph 77.1-Iair et al, ”ultivariate Data Analysis With Peadings
(Tulsa: Petroleum Publishing Company, 1979): 2:22.

 

Cooley and lohnes, p. 145.

Donald Iee Stevens, "A ’“ultivariate Analysis of I‘inancial
Characteristics of Acquired Firms In Industrial r'ergers," (Ph.D.
dissertation, P’ichigan State University, 1972): 59, 59.
Hair et al, p. 234.

Cooley and Iohnes, p. 1653.

Stevens, "A F'ultivariate Analysis," p. 53.

Ibid, p. 116.

Ibid, pp. 70-72.

Ibid, p. '76.

Ibid.

Ibid, p. 70.

Ibid, p. 78.

Ibid, p. 76.

Ibid, p. 78.

Ibid, p. 86.

Ibid.

Ibid. pp- 91-93.

Ibid, p. 93.

117

20. Donald Iee Stevens, "Tinancial Characteristics of T'erged T‘ir'ns, A
1"Yultivariate l‘nalysis," Journal of l‘inancial and Quantitative

Analysis 8 (”arch, 19773): 154.

 

21. Ibid, p. 156.

Chapter VI

Validation of T’DA

In i'erger Analysis

A. Discriminant Analysis Results Compared to Univariate

Results

Empirical studies using TDA generally provide a comparison of
key variables isolated using ”DA with the univariate F test variable
scores. 'Ihe difference between the two forms of analysis is the "DA
method analyzes variables simultaneously whereas the F tests are a
sequential (univariate) form of analysis. Authors expect better
results using 1FDA because the simultaneous analysis is expected to
provide a better profile.

'T‘able 6-1 illustrates those ratios which appeared in the final FDA
model, their standardized discriminant coefficients, and then in
comparison, each ratio's ‘5‘ test score and its resulting level of
significance.

Data for this table is from SPSS using a FAQ V VDA dimension

reduction, and Statistic 6 of the Subprogram DISCRD'I‘A‘TT.

118

1 l 9
'I‘abl e 6-1

Distad's 1970's Data r."est: Univariate ’I‘ests of

Significance on "DA Selected Vey Dinancial Patios

6-1 (a)
LAG 2, (t—B)
(From Table 5-1)
Ratio Discriminant I“ Score Significance
Score

r95 .5659 2.529 (.1125)
4 —.6514 1.770 (.1840)
7 .5625 4.427 (.0559)
10.5008 1.026 (.9117)
12 .5091 .006 (.9401)
15 .5527 .500 (.5845)
21 .7074 50. 5.40 E .0000)
25 .2628 7.799 .0054)
26 .2060 1.891 (.1697)
29 .2975 1.651 (.2022)
52 —.2090 4.718 (.0504)
,5 -.4958 10.16 (.0015)
34 -.2689 2.997 (.0841)

120

There were some contradictions in the LAG 2 comparison. T"ost
conspicuous are FR12 (sales to total assets) and T313.15 (interest expenxs
to cash and marketable securities) which have marginal importance in an
FDA context but virtually no significance in a univariate context.
Other ratios which have poor univariate 7.“ tests also tend to have low
discriminant scores. r"hese include T7R10 (long term debt to total
assets), 26 (current ratio), and 29 (current assets to sales). One
surprising result is that while T‘P-A (net income to sales) has a large
discriminant score it is only significant at the (0.1840) level.
Finally, p/e ratio (“1117, 0.0301 significance), cash flow to total
assets (F1119 0.0363) and cash to current liabilities (F828, 0.0171) all
were highly significant in a univariate analysis, but in a multivariate
test failed to enter the final discriminant function.

Table 6—1 (b) and 6—1(c) are LAG 1 and LAG 0 Versions of 'T‘able 6.

Table 6-1(b)

IAG1,(tQ)
Discriminant

Patio Score 1“ Score Significance

FR1 .4202 1.122 ( 2901)
6 .2767 1.448 (.2295)
9 .4571 .576 (.4481)
15 .5284 5.667 ( .0561)
16 .5098 1.537 ( 2157)
17 -.2165 .151 (.6978)
21 .8061 55.4 (m M)
22 .1656 .276 ( 5994)
26 .2145 1.572 (.2421)
52 -.5466 1.262 (.2618)
33 -.5583 .084 (.7721)
54 .2810 .5549 (5516)
55 —.1884 2.658 (.1050)
56 .2110 1.017 (. 5156)

121

The most striking observation in IAG 1 comparisons is that only
one variable T’RZ1, total assets, was clearly significant. Interest
expenses to cash and marketable securities (”815, 0.0561) would be
categorized as marginal). The remaining ratios, though identified by

”DA had no univariate significance.

T‘able 6-1(c)
LAG 0, (t—1)
Patio Discriminant I” Score Significance
Score
F‘R1 .5600 2.000 (.1580)
5 -.2514 1.592 (.2587)
7 -.7066 2.5 5 (.0891)
8 -. 2768 4.250 ( .0405)
9 -.1812 .900 (.3453)
14 —.1579 .009 (.9251)
15 -.2196 9.512 (.0022)
16 —.9418 2.698 (.1012)
17 .8622 .020 (.8887)
18 -.4277 2.710 (.1004)
21 -. 6620 57. 24 ( .0000)
27 -.1297 .421 (.5168)
56 —. 5727 2.595 (.1224)
57 .2762 2.494 (.1150)

The IAG 0 model is more complex. long term debt to total assets,
(1.5210, 0.0249) significance) and total liabilities to total assets,
(F1111, 0.0429) failed to enter the final discriminant function.
Conversely p/e ratios (IP17) had a large discriminant coefficient
(.8622) but virtually no univariate significance (0.8897). Only total
assets (P5121) can be categorized as having both a high discriminant
function and a high F score (-.662 and .0000).

The other ratios fall provide mixed results in terms of

discriminant functions and '17 scores. For example, net income to total

122
assets and. dividend percent payout (IR? and. 16) have high discriminant

13

scores, but marginal scores. Refer to table 6—1(c); P score
significance levels are 0.0891 and 0.1012 respectively. Conversely,
long term debt to market value equity, and interest expenses to cash and
marketable securities, (T8 and 15) , have low discriminant coefficients
(~.276’~3 and -.2196) but very good F scores; (0.0403 and 0.0022
respectively) .

B. Conclusion

 

To summarize, LAG 0 has few clear contrasts or agreements and

many ambiguities.
Recapitulating over all three models, univariate P tests are more
restrictive, isolating fewer key ratios, thereby producing a model with
lower classification success than attainable using a reduced dimension

discriminant function. See Table 6—2 for an illustration.

Table 6—2
Vey Patios Selected Using

Univariate I” rVests Significant To .05

IAG 2 LAG 1 LAG 0

FR? 731115 FEB
17 21 1O
19 11
21 15
23 20
2‘3 21
32

33

Chapter VII

Implications and Uses

A. The Prediction of ”ergers

 

Earlier in this test a reference was made to Joy and
Tollefson's contention that there is a sizeable difference in results

between ex post classification and ex ante prediction:l

An ex post
classification MDA based model "fits" a profile of variables, and
weights those variables according to their importance in the maximal
separation of two or more groups. Some of the variables may be
coincidental, yet "fitted" and as a result be spurious in a predictive
model unless that coincidental relationship between variables continues
in the future.

It is important therefore to subject all the models tested in this
paper to an ex ante predictive model, because it is the most important
test of these models.

A sample of fifty firms, acquired in 1979, were classified by the
discriminant functions generated by Stevens and by me. It is important
to emphasize that the 1979 data submt was never used in the
determination of any of the previously discussed discriminant models; to
do so would impart an upward bias in the classification success in the ex

ante prediction test and invalidate the results. rFable 7-1 illustrates

ex ante prediction rates.

123

1. Stevens' ”odels

 

12“

".‘able 7-1

B: Ante Prediction V'odels

7-1 (a)

Stevens' Original VOdel: (VQN 5-1).

rm,1o,12,31. LAG 2 5 of 39 mergers
LAG 1 2 of '58 mergers
LAG O 1 of 37 mergers
where m7 =
10 =
12 = sales to total assets
31 =

124

594

74
/.l J

net incone to total assets

long term debt to total assets

net working capital to sales

Table 7-1 (b)

Stevens' Reformulated ‘r‘"ode1: EG‘I 5-3).

m7,10,12,16,24 LAG 2 7 of 40 mergers
IAG 1 1 of 39 mergers
LAG O 1 of 38
where m7 =
10 =
12 = sales to total assets
16 =

of net incane

24

net working capital to

total assets

18‘5

34

1%
/VI

net income to total assets

long term debt to total assets

cash dividends as a percent

125

“able'7-1 (0)

Stevens' Article “"odel: (TC‘L‘I 5-5)

 

FR5,10,12,24 LAG 2 6 of 39 mergers = 15“
IAG O of 39 mergers = 05
LAG O O of 37 mergers = 04
where TBS = net operating income to sales
10 = long term debt to total assets
12 = sales to total assets
24 = net working capital to total
assets
7—1 (d)
2. Distad's Vodels
Reduced IAG-2 15 of 39 mergers = 39“
Dimension
(from.fiable IAG 1 8 of 38 mergers = 21%
5-1)
LAG O 9 of 34 mergers = 26%
7—1 (e)
Vethod Direct IAG 2 9 of 38 mergers = 24“
(from Table
4-3) LAG 1 9 of 36 mergers = 254

IAG O 8 of 33 mergers = 241

126

The important contrast is that Stevens' models demonstrate
significantly lower ex ante classification success than the more recent
discriminant functions, which is consistent with the high level of type
I error rates found in my 1970' 8 data replication of his model, as shown
in Table 5-14.

Stevens' FRS, earnings before interest and taxes divided by sales
had consistently lower discriminant scores than another profitability
ratio, F37, his second ratio and one of my most important ratios, net
income dividend by total assets.

Stevens' third ratio, W10, a leverage ratio, long term debt
divided by total assets, was not important in my tests, though this was
Stevens' most important ratio. Two other leverage ratios, me, long
term debt divided by the market value of common stockholders' equity and
FR9, long term debt divided by book value net stockholders' equity, were
of only marginal significance. I found F‘R15 to be a very important
ratio, (interest expenses divided by cash and marketable securities).
Stevens categorized W15 as "other," but I categorized it as a leverage
ratio because the extent of leverage measured by debt to assets or equity
is not as important as the ability to pay annual interest eXpenses. In
that context I consider F‘R15 to be an excellent leverage ratio.

Stevens' fourth ratio, sales divided by total assets (TR12)
appeared once in the LAG 2 model. The only asset efficiency ratio of
significance was F1237, sales divided by net fixed assets. The
significance of IRE? may confirm my earlier hypothesis that acquired
firms tend to have higher fixed asset turnover ratios (F???) because
they are using depreciated fixed assets which are worth more to an

acquirer than their book value, or improperly valued natural resources.

127

Nevertheless, I837 did not become significant until one year prior to an
acquisition, LAG 0.

Stevens fifth ratio, IR16, dividend payout, was not an important
ratio in this study, particularly in LAG 0 model.

FR24, net working capital divided by total assets, a liquidity
ratio, never entered any of my discriminant functions.

The last ratio used by Stevens was 3131, net working capital to
sales. That ratio apparently no longer is considered to be important.
It had low discriminant scores in my original model and failed to appear
in any of my dimension reduced final models. Again, assuming that ratio
is classified as a liquidity ratio, it contributed nothing to the
prediction of mergers.

Low returns on assets and low dividend payout rates remain
important in the selection of merger candidates. After those two
ratios, very little comparability exists over time. leverage does not
appear to be as important as the ability to pay interest exIenses.
Total asset turnover's importance has diminished as the focus now
appears to be on the value of fixed assets. Also, unless compared to
market price (3333-35), liquidity ratios seem to have diminished in
importance. One possible explanation for the diminished success of
Stevens' ratios is that they have contributed to the efficiency of the
merger process; therefore the ratios are no longer useful.

***************

Pecause Stevens examined financial ratios of acquired firms one
year prior to their acquisitions, he could not have known that he might
have obtained greater success using different ratios three years prior

to a firm's acquisition. Oily additional follow-up studies replicating

128
both Stevens' and my models can verify that possibility.

(“he ability of these three models to identify merged firms from
that subset represents only a portion of a complete ex ante prediction
endeavor. The models must also be able to correctly identify non-merged
firms in the same time span, 1979. Iurthermore, it is inapprOpriate to
sample an equal number of non—merged firms because such a composition of
samples is not representative of the level of merger activity in any
year. In any given year, it is not possible to predict precisely the
percent of all corporations acquired in the United States.
Additionally, such a percent would lack validity for at least two
reasons: 1. it would be distorted by the large number of firms too
small to be tested in this study, and 2. because all acquired cases
studied here were sufficiently large to warrant inclusion in the
Compustat Industrial File, this test addresses acquisition size in
dollars, rather than percent of acquisitions. However, some attempt to
imbalance the two subsemples must be undertaken to acknowledge the large
difference, even in years of extreme merger activity.

A randam sample was extracted from the Compustat Industrial File,
to serve as the control group, a list of firms still not acquired. This
random sample was 843 cases, which was reduced to 823 cases to avoid the
problem of firms with dissimilar accounting definitions. Also, because
the SPSS program will not accept any company with missing accounting
variables, there were finally 684, 653, and 553 cases in the control
group providing accountirg data for the years 1976, 1977, and 1978
respectively, (LAG 2 through IAG 0). Because this part of the test is
not cross-sectional in nature, IAG 2 does equal 1976, three years prior

to the acquisition, and so on for 1977 = LAG 1 and 1978 = LAG O.

129

Again, it is not possible to predict percent levels of merger
activity in any year, but I decided that 5.7% (39/694); 5.9.”- (33/6575);
and 6.151." (34/553) would be representative.

She only model used in this portion of the test was the Distad
Reduced Dimension model because of its comparatively better predicting
results in classifying acquired firms, as reported in "able 7-1(d).
Other models appearing in “able 7-1 may better classify non-merged cases
offsetting lower merger identification results, but in light of the
results it is unlikely for such an event to occur.

Because neither the merged nor nonemerged cases classified here
were used in the determination of the discriminant models, and because
they were selected randomly, I am assuming these cases are largely
unbiased and are valid samples.

The following three matrices provide the results of the total

classification success of each of the three discriminant functions.

130

”‘able 7-2

I): Ante Classification Results,

Distad's Reduced Dimension TTodel

 

 

7—2(a)
LAG 2, 1976
Actual Predicted
l"erged Von—merged
# (ii) if (7’)
merged 15 (2.1) 24 (3.3) 39
Non—merged 60 (8.3) 624 (96.3) _6__Q_1

ri‘ype I Error Rate (24/39) = 61.5“:

Type II Error Rate (60/684) = 8.W

rl‘otal Classification Success: 98.47:
F'aximum Chance Success: 94.65”»; t = -7.37

Proportional Chance Success: 89.34; t = —1.24

723

131

 

 

7—2 (h)
IAG 1, 1977
Actual Pred icted
l’erged “Ton-merged
x/ (a i (ci)
T‘erged a (1.2) 30 (4.3)
Non-merged 178 (25.8) 475 (68.7)

”yea I Frror Rate (30/38) = 78.9%

”V109 II Error Rate (178/653) = 27,394
Overall Classification Success: 69.9%
l‘aximun Chance Success: 94.59“; t = -28.36

Proportional Chance Success: 89.8%; t = -16.96

0\
gm
I KN

132
7—2(c)

LAG o, 1978

 

Agtual Predicted
Verged Yon—merged
# C“) # (3)
Verged 9 (1.5) 25 (4.3)
lknemerged 140 (23.9) 413 (70.3)

Type I Frror Rate (25/34) = 73.5%

Type II Error Rate (140/553) = 25.3%
Overall Classification Success: 71.94
”aximum Chance Shccess: 94.2%; t = -23.11

Proportional Chance Success: 89.1%; t = -13,37

\J'I

\PJ

 

Ii

133

Results of'ex ante classification.are disappointing because of“the
very low t statistics, indicating that these three models cannot predict
mergers.

However, these results are important for several reasons:

1. These endeavors reflect the results of the first attempt to
predict mergers. Much of what has been previously written in the
literature must be subjected to closer scrutiny because those tests were
ex post classifications, and their conclusions were presumed to be val id
in an ex ante application, which was Clearly not so based on this first
test. Acquisition candidates do not appear to always be obvious, and
each case may be unique.

2. ”.hese results suggest that it is not possible to predict
mergers using publicly available information. Or if it is possible, a
model must use a set of financial ratios different than the thirty seven
used here, which seems unlikely.

3. The classification rates and t scores reveal the "best" model
for predicting mergers is the LAG 2, use of information publicly
available three years prior to a.firm‘s subsequent acquisition. Such.a
finding, if confirmed by subsequent studies should be a significant
contribution .

4. This model has attempted to approximate the likelihood of
merger in 1979, using a 5?; sample of merged firms.

Results from this study suggest that additional studies may be
warranted because of this lange variation in size of the two samples.
The LAG 2 model had an 88.4% overall classification rate, but this is
distorted by the much larger sample of nonamenged firms. The Type II

error rate was less than nine percent (60/694), but because the

1311
non-merged firms were 94.64 of the total test, it was impossible to
exceed that maximum chance likelihood. To obtain a one tailed t test t
score greater than 1.96 ( o: .005, no d.f.) required a 96.37--
classification success. If the model had been able to predict all of
the merged firms, the model still would have produced an unacceptable
level of prediction because of a nine percent Type II error rate.

"-‘he classification matrix below is that which is required to

obtain an a: .005, (t =1.96, co d.f.):

'T‘able 7—3
Hypothetical Classification "atrix, LAG 2
Assuming a 94.6‘?’ E"aximum Chance Rate, and

a 100,4 Success Pate Predicting F‘erged Firms

Actual Pred ic ted

We rged Non-merged
P'erged 39 O
Non-merged 29 655

Such a classification result is necessary to produce the 606/723
overall classification rate significant at .005. mhose results require
amodel producing no 'T‘ype I error rates and only a (29/684) 4.25% "‘ype II
error rate. T.‘se of the prOportional chance model provides less
restrictive conditions, but is still a rigorous test when the sample
sizes are so unequal. If the model again had perfect merger
classification success (39/39), the number of non—merged firms

misclassified as merged ("Type II error) would have to decline only by 2

135

to 58 to produce a mddel significant at a: .005, with so d.f. at t =
1.96.

Finally, if it were possible to obtain amerged sample ten times as
large as the above results, 150 merged, 240 misclassifications

totalling 3% cases, the following results would be reported:

7-4

Hypothetical Enlarged Sample
Classification TTatrix, IAG 2

 

Actual Predicted
f’v'erged Non-merged
T'erged 150 240 = 390
Non-merged 60 624 = 634
107/1

Total Classification Success = 72.1%

T'"faximun Chance Classification Success = 63.773; t = 5.71

Finally, equal sample sizes with proportional results for merged
firms classification produces a larger difference in significance
canpared to maximum chance, t = 10.98.

Obviously, sample sizes affect the conclusions of these results
greatly and must be dealt with carefully in subsequent studies.

B. Conclusions:

 

Results of this study indicate that even with the use of ”DA on
financial ratios, it is not possible to predict mergers, thus perhaps
precluding any opportunities to isolate firms which will ultimately

offer risk adjusted excess returns.

136

The next steps to be taken in future studies are to replicate the
study using larger samples of both merged and non-merged firms,
maintaining approximately a 3—7 percent sample of merged firms. (”he
other study ought to be an analysis of'the individual firms to ascertain
if the premiums garnered from pred icting 15 of 59 mergers were
sufficient to offset any losses from an investment in the 60 of the 684
non—merged firms mistakenly classified as being merger candidates, and
also to ascertain if those sixty firms were sUbsequently acquired.

Finally, it must be emphasized that this study is an initial
inquiry. While preliminary results strongly suggest that it is not
possible to predictrnergers it will be necessary to replicate this test
for confirmation. Fest significantly, and surprisingly, the evidence
suggests that the greatest likelihood of obtaining a working model
should address the data available three years prior to a firm's
acquisition.

The next chapter, the conclusion.ccmments further upon.the ratios,

their coefficients, and their charges in magnitude and sign over time.

Ibotnotes

Chapter VII

1. Vaurice 0. Joy and John O. Tollefson, "On the Financial Applications
of Discriminant Analysis," Journal of Financial and Quantitative
Analysis 10 (December, 1975): 727.

 

137

Chapter VI II

Conclusions and IValuations,

limitations and Inture Pesearch Cfuggestions

A. Summary of the Results

 

This study has been conducted to determine whether financial
ratios can predict mergers.

A stepwise T'T‘A. model was employed in an attempt to reduce to amore
manageable level, the nunber of ratios from the original thirty seven.
The final models produced thirteen to fifteen ratios. Ffforts to
further reduce the dimensions of the model produced models with lower
classification results.

r“he discriminant functions used to classify firms in the samples
were significant at the .001 level to .0001 level and beyond, in ex post
tests, but lacked significance in the ex ante tests.

Considerable care was taken to ensure that the necessary
statistical precautions were employed in this application. Such
precautions include split sample procedures and a hold out group used
for an ex post classification. More importantly, those models were then
used to conduct an ex ante prediction of mergers in a subsequent time
period. This is the first ex ante predictive mergers test known to me.
As presumed , results were lower for the ex ante test than for the ex post

test .

138

139

The test was thorough, using data from two Compustat data files.
Two hundred thirty five acquired firms were tested with three hundred
and four non-acquired firms. It is presumed that thew sample sizes are
sufficiently large to make the study robust to the violation of the equal
variance—covariance requirement asserted by Fisenbeiss [52]. TTo one
has yet detennined what are "large" sample sizes , only that large sample
sizes will strengthen the design of any FDA designed experiment.

""ergers tested occurred in the period 1970 through 1976. ”he ex
ante test was applied to a set of fifty mergers occurring in 19.79 and
also to a set of eight hundred forty five randomly selected non-acquired
firms in 1979.

The fifteen ratios selected by the step wise "DA process, ranked
by type of ratio, were:
Profitability Ratios

 

1. net operating income as a percent of total assets

2. net operating income as a percent of sales

5. net income after taxes as a percent of total assets

4. cash flow to book value tangible stockholders' equity

5. tax loss carry forward per share to market price per share
of common equity

Liquidity Ratios

 

6. quick current assets as a percent of total assets
7. quick current assets as a. percent of current liabilities

Leverage Ratios

 

8. long term debt as a percent of market value stockholders'

equity

1110

9. long term debt as a percent of book value tangible stock-
holders' equity

10. interest expenses as a percent of cash and marketable
securities

Activity Ratios

 

11. sales to quick current assets
12. sales to net fixed assets

it"arket Ratios

 

13. dividends paid as a.percent of earnings
14. price to earnings per share ratio
9.21.221;
15. total assets - not a ratio, a stratification measure

In spite of changing financial characteristics and a myriad of
motives fbr merger, some of’the hypothesized financial characteristics
of acquisitions were confirmed. In Chapter 1 of this report, my
hypotheses, which were consistent with the existing literature, were
that an acquired firm tended to have excess liquidity, unutilized debt
capacity, favorable earnings, and be priced in the markets at a level
below the company's book value per share.

Acquired firms were profitable, but interestingly only operating
profits were favorable. Net income and cash flows ratios were less than
those of nonacquired firms, suggesting extraordinary or nonrecurring
economic or accounting events depressing after tax profits.

T’erged firms tended to be less levered and less liquid than

nonacquired firms, but the differences were not major. One possible

1141

exception to this was the large negative coefficients of TR33, market
price to cash per share in the IAG 2 and IAG 1 models.

Activity ratios indicated that sales to quick:current assets were
lower for acquired firms. “Tore interestingly, sales as a percent of net
fixed assets were greater fbr acquired firms. The possibility there is
the firm's assets have been depreciated more fully than fer nonacquired
firmS'Which suggests their market value for assets is greater than their
bookrvalues.

Acquired firms were not priced by the markets at levels below
their book values. A "market to book" ratio failed to enter the final
discriminant fhnction. One possible explanation for thisrmight be that
speculators may have bid up the price of the equity previously in
anticipation of a takeover. Chat is consistent with the literature
reviewed in Chapter II. Further support fbr that expflanation is
provided by the higher than average P/E ratios shown by acquisitions one
year prior to a.merger.

Analyzing market value as a percent of book value in models three
years and then two years prior to a merger, market is lower than book
three years prior. Then tm years until amerger, the ratio is no longer
significant suggesting that stock prices appreciate as much as two years
prior to a merger, Which again is consistent with the literature.

The important results from this study are that financial ratios
may yet be used to predict mergers, but it takes many more ratios to
accomplish that objective than previously indicated. Mthermore,
success rates are not as good as reported in previous tests fer several
reasons incluling:

1. previous tests never attempted to predict mergers,

1112

2. econanic, financial and. accounting conditions have changed
introducing the possibility of more motives for mergers.

'5. the presence of arbitrageurs may affect market ratios and
force managenent to develop strategies more rapidly to
fend off potential acquirers, which will reduce the
comparability of ratios significant in 1960s mergers to
those useful for predicting mergers in the 1970s,.

4. the absence of any qualitative criteria.

The remaining portion of this chapter elaborates upon the above

conclusions and others of less significance.

P. Prediction of Vergers

 

Financial ratios can predict mergers, but the thirty-nine
percent success rate produced using a dimension reduced model on
financial ratios existing three years prior to the merger is a
disappointing result. Suggestions for subsequent studies include
testing other financial ratios not used here, obtaining larger test
samples, and perhaps most interestingly, ascertaining whether stock
prices began appreciating prior to the ratio model's prediction of a.
merger. Another interesting test is to rank the percent stock price
gains of the identified candidates. large premiuns would enhance the
predictive value of this model from the perspective of the arbitrageur.
Simkowitz and F’onroe, Eelkaoui, or Stevens did not use ex ante
predictive models so it is not possible to comment on their usefulness in
that context. It is now lmown that my replications of Stevens' models
in this instance, were unable to identify merger candidates either in an
ex post classification, (see Tables 5—14, 5-15, and 5-17) or in an ex

ante prediction, (see Table 7-1), when replicated on mergers data from

lLl3
the 197 's.

Such a dramatic decline in the success of the Stevens models from
results obtained in these replications necessitates an inquiry into any
economic events transpiring since his 1960's study which might account
for this presumed decrease in efficiency. There are several reasons for
the reduced applicability of his models. His last model, equation 5-5,
which is presumed to be his most successful model, included these key
ratios:

(”Equation 5-5) a. Profitability: T‘RS; earnings before interest

and taxes to sales

b. Leverage: F‘R10, long term liabilities to

total assets

0. Activity: 13812; sales to total assets

d. Liquidity: T1324; net working capital to

total assets

a. Profitability

 

Currently other profitability ratios appear to be more
significant than TBS. Those ratios include “‘21, earnings before
interest and taxes to total assets and T7, net income to total assets.
Explanations for the current importance of F91 and TV? are two fold:
first, total assets appears to be more important than sales as a criteria
of percent profitability. "he large positive coefficients of T‘R‘l
suggest that acquired firms have larger operating income, or more

D

likely, have lower book value total assets which supports the recently

11114
advocated cheaper to buy than build theory. ’T‘he second explanation is
derived more from the negative coefficient “R7 which illustrates that
acquired firms have less net income and fewer total assets than do non-
acquired firms. qhe difficult part of the explanation is reconciling
the higher than average operating income with the lower than average net
income after taxes, (‘37).

Some accounting or economic event(s) reduces higher than average
levels of net Operating income to lower than average after—tax profits.
("ax loss carryforwards are one possible explanation. W36, the ratio of
tax loss carryforward per share divided by the market price of the common
stock, produces a +.21101 coefficient in IAG 1 which indicates a higher
than average ratio. In IAG O, the ratio's coefficient falls to a
-.37273, suggesting a dramatic stock price increase between IAG- 1 and
LAG— 0. It can be concluded that tax loss carryforwards are only one
possible explanation. No tests for other nonrecurring accounting
events were tested.

F. Leverge

Two other leverage ratios currently appear to be more
important than F‘R10, long term liabilities to total assets. They are
1739, long term debt to the book value of tangible common stockholders'
equity and T1‘R15, interest expense as a percent of cash plus marketable
securities. F‘R10's reduced reliability is probably due to the declining
comparability of book values to market values of both debt and equity.
T’arket values of debt are likely to be smaller than book values because
of rising interest rates throxghout the 1970's. "There is considerable
evidence to suggest that market values of assets in some industries are

greater than their book values. Because these variations would not

1115

offset each other, TF10, a ratio using both book value long term debt to
book value total assets would be of diminished economic significance.

73215 suggests that the extent of a debt to assets ratio is not as
important as the ability to have sufficient liquidity to meet interest
expense requirements. Because the coefficients of mg and 3215 change
over time it is necessary to explain then. 3‘89 has coefficients of .457
and -.181 in the LAG 1 and IAG- 0 models. 'The .457 coefficient suggests a.
higher than average debt to tangible equity ratio than firms not likely
to be acquired. Pecause the numerator, long term debt, is not likely to
change rapidly, it appears that the subsequent decline in the LAG 0 model
to -.181 is attributable to an increase in the denominator, the book
value of the equity. However, because of the small coefficient, the IAG
0 ratio has reduced significance.

171215 has coefficients of .333, .528 and -.220 (LAG 2 through IAG 0
respectively), suggesting a deteriorating ability to fund interest
expenses, until IAG 1 when the acquired firm increased its cash, retired
debt or a combination of both. the cash accumulation possibility may be
the result of a. takeover defense.

0. Activity

Activity ratios no longer appear to have much significance,
with one very important exception, fixed assets turnover, a ratio not
tested by Stevens because it was not important in the rising stock market
characteristic through the 1960's. What ratio is sales divided by the
net book value of fixed assets. 'Pnat ratio is justly criticized because
it fails to separate firms with new fixed assets Operating at high levels
of efficiency from firms operating less efficiently using older more

heavily depreciated assets. In a multivariate context firms whose

1116

market value of fixed assets is greater than their book value should be
isolated as potential acquisitions assuming the cheaper to buy than
build philosophy is val id.

In this test, it was presumed that the buy versus build contention
was a motive for mergers. Acquisitions were hymthesized to possess
higher fixed asset turnovers because either: 1. their fixed assets
possess replacement values which are greater than are their book values,
or 2. included in the accounting designation fixed assets are physical
assets, including natural resources, whose market values are greater or
at some future date can be expected to be greater than are the current
book values as stated on the potential acquisition's balance sheet.

Fixed assets turnover, I837, when using "DA was found to be of some
importance, thomgh not significantly so until the year immediately
preceding a takeover, when its coefficient was +.276 (from Table 5-1).
Though “9.37 is a contributor in the isolation of merger candidates it
does not appear in the reduced dimension models until LAG 0. It is not
known whether that information would be of any use to either financial
managers or to arbitrageurs because of its time lag and coefficient
size.

d. Liquidity

Liquidity ratios are no longer useful in the selection of
potential acquisitions. The ten liquidity ratios tested here appeared
infrequently in the reduced dimension discriminant models, their
coefficients' absolute values tended to be low, and their signs
alternated over time, suggesting little if any economic significance.
There was only one exception to the above: IP29, current assets to

sales, had a +.29725 in IAG 2. Pecause of the marginal magnitude of the

1117

coefficient and because it did not reappear in the IAG 1 or LAG 0 models,
it may lack significance.

Perhaps the most logical explanation for a sharp decline in the
importance of liquidity ratios is attributable to the changes in the
domestic economy since the 1260's. Fecause of greater than average
inflation in tte 1970's, cash or near cash assets tended to diminish in
value relative to physical assets such as inventories or fixed assets
during that time span. Variable FR26, current ratios, which includes
inventories in the current assets calculation was generally more
significant than was the acid test ratio, 1177327, which does not include
inventories in its determination. In spite of inventories being
incluied, current ratios were not significant. 'i’heir coefficients were
.206 and .214 in LAG 2 and IAG 1, and the ratio did not appear at all in
the reduced dimension IAG 0 model.

Another explanation for the decline in the importance of liquidity.r
is that in the 1960's firms acquired other firms because of the latter' 3
liquidity which assisted the acquirer' a financing of the merger. "hose
acquisitions tended to be financed by equity or debt. Such a motive for
merger or such financing arrangements are no longer prevalent; currently
a more common scenario finds the acquirer borrowing, being supported by
venture capital from insurance companies, or being financed by foreign
capital to purchase a firm with incorrectly valued long term assets.
Firms no longer need to acquire other firms for their excess liquidity.
Acquiring firms either possess cash to purchase other firms or can
readily acquire it.1

While new discriminant functions generated from merger data of the

1970's produce better results than that obtained from replications of

1118

1960's models, the models produced lower than anticipated results. Ex
post classifications, though statistically significant, showed overall
classifications success of 68.16, 62.78, and 69.71, respectively, IAG 2

through IAG O. 'Z‘hese results appear in Table 8-1.

1119
’T‘able <3—1
Final Classification T’atrices

Table 8—1 (a)

 

 

IAG 2
CLASSIFICATION RESULTS - IAG 2
ITO. O? URFDICWD GROUP “-"VWFF'FIIP
ACTUAL GROUP CASES Group 1 Group 2
GROUP 1 220 121 9’?
55.0 a", 45.0 ‘1
GROUP 2 270 57 213.
21.1 5% 78.9 F”
GROUP 5 59 15 24
58.5 °’ 61.5 if

FCRCHIT OF GROUPZD CASES COFPJCTLY CIAF’SIFEZD - 68.16, Ix Post.

150

Table 8-1 (b)

 

IAG 1
CLASSE‘ICA’TION RT'EULTS - LAG 1
YO. OP PREDICT?!) GROUP T-‘TVPP‘RSTHIP
AchAL GROUP CASRS Group 1 Group 2
GROUP 1 226 10? 119
47.5 91 52-7 9’
GROUP 2 265 65 200
24.0 :1» 76.0 0"
GROUP 5 58 8 50
21.1 91 78.9 $1

Emmi-WT OF GROUPYZD CASES CORRTIITLY CLA?SFIED — 62.78, 3 Post.

Table 8-1 (0)

 

IAG o
CLASSIFICATION R‘TBULPS - IAG O
NO. or PREDICTPD GROUP wreamsmp
ACTUAL GROUP CASES Group 1 Group 2
GROUP 1 218 122 96
56.0 ~77 44.0 "’ WP? I
ERROR
GROUP 2 244 44 200
18.0 51-. 82.0 6’
GROUP 5 54 9 25
26.5 5’1 73 s 0’

PERCENT" 07‘ GROUPED CASFS CORRECmLY CLASSIF‘DID - 69.70, 3 Post.

151

Pk ante prediction successes from the nemrmodels are considerably
lower than hypothesized. The prediction test fbr the dimension reduced
model was superior to any other model, but also produced lower than
anticipated results. Those results appear in Tables 7-1 and 7-2.

The<iifferences in results between ex post classifications and ex
ante predictions, as hypothesized by' Joy" and TollefsonQ are of
considerable magnitude.

The dimension reduced discriminant models, LAG 2 - LAG O,
illustrated in Table15-1, produced overall classification success rates
of 68, 65, and 705, respectively (see Table 5-2a, b, and c) and did so
with very high levels of statistical significance.

Vance? claimed to be able to predict seventeen of twenty one
mergers in the late 19608, using published financial statements.
Hewever his criteria contradict the evidence produced here using arnwch
larger time span and sampfle size. Pore finportantly, vance never tested
his hypothesis using an ex ante prediction test.

anroved classification results from the new models demonstrate
that the 1960's motives are even less valid. ’Ihe new models will
predict more successfully if the Type I error rates can be reduced.

As illustrated in Table 8-1, Type I error levels for my new models
are 45, 52.7 and 44.5 respectively, LAG 2 through IAG O.

Type I error rates are attributable to at least two sources: 1.
The model is using ratios which do not fully embody all motives for
mergers. It is possible that the thirtyhseven ratios used at the onset
do not incorporate all of the quantitative motives for mergers, but it is
more prdbable that nonquantitative motives for mergers have a

considerable influence on the likelihood of acquisition. 2. Votives

152

for mergers include unique company by company circumstances either of a
quantitative or qualitative nature, which may supercede any general set
of quantitative ratio criteria.

Type II error rates are low in this test. rThis indicates that the
model is rarely incorrect once it has classified a firm as a likely
acquisition.

In essence, the model is successful in that it does predict
mergers; however it fails to select enough of the firms which ultimately
are acquired to improve its statistical significance.

What can be generalized from this study regarding the
characteristics of the financial ratios studied?

Ratios that seem to have significantly large coefficients and
appear often in all three models (LAG 2 - LAG C) are:

7131 : EBIT/ Total Assets (m1)

“P7 : Rate of Return on Assets (PR7)

IR15: Interest / cash and marketable securities (FR15)

FR16: Dividend Payouts, Percent of P.P.S. (7R16)

r917: P/E Ratios (r217)

FR56: Price / "ax Loss Garry Forward Per Share (T1336)

Sumarizing the above ratios, we find higher than average
operating profits, lower than average earnings after taxes and
extraordinary itens, declining dividend payouts, increasingly larger
p/ e ratios suggesting declining net income as well as price appreciation
and slightly higher than average leverage, (but only in the context of
interest expense as a percent of cash and near cash (items) , to the chief

determinants in the prediction of mergers.

153

C. Confirmation of the Ratio Hypotheses

 

In Chapter I a set of hypotheses was formulated based upon a
review of the literature. Those hypotheses were that an acquisition
would be: a. less levered, b. have greater liquidity, c. possess
greater earnings prospects, and d. would be priced lower relative to its
book value per share than are firms not likely to be acquired.

Conclusions based upon this study are not entirely in accord with
the hypothesized results.

a. I conclude that acquired firms are not less or more levered
than non-acquired firms, leverage simply is not a discriminating
criteria, except for IR15, which reflects a firm's ability to pay its
interst expenses; the extent of debt may be irrelevant, what is
important is the firm's ability to finance its indebtedness.

Acquired firms do tend to be more heavily levered, but it is
misleading to accept that statement without mrther investigation. "he
only leverage ratios providing significant results were book value long
term debt to market value common stockholders' equity, PR8, and book
value long term debt to tangible book stockholders' equity, PR9. It
would be useful to test for market value long term debt to market value
common equity in that market value debt is almost invariably less than
the book value of that debt. Using market values for both debt and
equity might suggest that acquired firms are not more heavily levered.
One other ratio which Stevens categorized as "other" could be classified
as a liquidity ratio, but I elected to treat it as a leverage ratio ; it is
interest to cash and marketable securities, 717.15. Results show that
book value debt to book value tangible common stockholders' equity, IR”),

was insignificant as were long term debt (IP10), or total liabilities

15“
(PR11), to total assets.

b. It appears that firms have average liquidity; it is incorrect
to hypothesize either illiquidity (H‘ excessive liquidity. The
condition of a finn's liquidity shmply was not a determining factor in
the selection of candidates fOr acquisitions in the 1970's.

c. Acquired firms do possess above average profitability as
originally'hypothesized, however, the above average profitability, 3R1,
is operating profits as a percent of'total assets. Profits after taxes
as a percent of total assets, FR7, produces lower than average ratios
than fbr the non—acquired firms.

Again, the dichotomy between above average operating
profitability to the below average after tax profitalei y suggests one
or more of several nonrecurring accounting events Which have made the
firm.more Tusceptible to a sUbsequent takeover.

Of the seven profitability ratios tested, only the two described
above were important.

An analyst fbcusing on earnings of earnings per share growth, and
disregarding operating income would have incorrectly hypothesized that
acquisitions possess lower than average profitability. Use of any other
profitability' ratios ‘would, have ‘led an. analyst to conclude that
profitability was inconsequential in isolating candidates.

d. Assets worth more than their stated book value suggests that a
firm's book values are understated relative to their market value,
assuning that the market pricing mechanisn is functioning and all
information is available.

Conversely an alternate possibility is that the market pricing

mechanism is not functioning and asset values and market values are in

155
disequilibrium.

Acquired firms do appreciate in value after announcement of their
being merged, however the greatest price appreciation occurs months
prior to that announcement date. 'Ihe coefficients of IR52, market value
to book value are -.20991 and -.54566, in LAG 2 and LAG 1 respectively.
'Ihe LAG 1 coefficient of -.54656 tells us that the market price of the
acquired firm has deteriorated even further than in LAG 2 relative to the
book value of the firm. The imbalance between market value of the firm
per share to book value per share of the firm is eliminated in the LAG 0
model which indicates that market price to book value is no longer of any
value in isolating potential merger candidates. In essence, at some
point in time twelve to twenty-four months prior to a merger the stock
price reacted to the imbalance but not sufficiently so to prevent a
merger. 'Ihe greatest returns to the arbitrageur came one to two years
prior to the merger, which is consistent with most of the empirical
findings reviewed in Chapter II.

In an attempt to further analyze the relationship of market price
per share to book value per share, I then separated total assets into
certain subsets, which I considered to be attributes. I hypothesized
that firms not only had greater liquidity, but also hypothesized that
certain attributes could not be identified simply by looking at book
value per share. I also felt these attributes, W55, cash per share;
IR54, net working capital per share; and W5, a restricted net working
capital per share (current assets less inventories and less current
liabilities, per share) ought to be examined in the context of their
value relative to their market prices per share. While the conventional

liquidity ratios appeared to be of limited significance, one liquidity

 

 

156

relative to market price of equity is of greater import. PR55, the
market price of the equity to cash per share does have significantly
lower than average coefficients, -.E"381 and -.55‘Q26 LAG 2 and LAG 1
respectively, which indicates that acquired firms have greater liquidity
than do their non-acquired counterparts or that other influences have
simultaneously reduced market prices to create these significantly
negative coefficients.

FR54, market price per share to net working capital per share has
coefficients of‘-.26886 in IAG 2, + .29098 in.IAG 1, but does not appear
in the LAG 0 model. ’Ihe low magnitude of the coefficients suggests they
appear to be of little importance. TR55 only appeared in the LAG 1
model; its coefficient sign was appropriate, but itsrmagnitude suggests
that it is not very valuable in the selection process (-.18959).

Conclud ing the discussion on market price of equity to book value
of assets, it appears there clearly is a disequilibrium between them
which is eliminated at some time twelve to twenty-four months prior to a
firm's acquisition.

There is also evidence that acquired firms either: 1. Uavexmore
cash than non-acquired firms relative to their equity market price per
share, or 2. that these acquired firms have lower market prices per
share because of other circumstances which dominate the cash effect.
Because of: 1. large negative coefficients fOr T”R55,market price to
cash per share, 2. The reduced extent of interest expenses as a
percent of cash and mar etable securities, and 5. The clearly
documented price appreciation of takeover candidates preceding the
announcement of’their acquisition, the evidence supports the higher

level of cash position of the acquisitions.

 

157

D. F‘“A.TIS”ICAL SIG’II‘T‘ICA‘ICT“, OT“ PENIS

 

Considerable research has been written citing financial
misapplications of WA in recent years [52,57,70,99]. A considerable
effort was made to comply with suggestions offered in those studies.
Tests were conducted for equal variance-covariance matrices, and split
sample hold out validation techniques were employed. Various chance
classification methods were used upon the validation sample and. the ex
post tests all produced results significant at tie .05 level or better,
(see Tables 4—-2a, b and c).

This test then employed the first ex ante merger prediction model,
(to the best of the author's knowledge) but produced lower than expected
results.

**************

E. DI F'R‘ISION REDUCTION

 

Table 7-1 illustrates that it is possible to reduce to a lower
nunber, the thirty-seven ratios used at the onset. Unfortunately
attempts to reduce the models to four or five ratios were unsuccessful .
Several attenpts not reported in this paper using snaller sets of ratios
were msuccessful.

Why is it no longer possible to produce the four or five ratio
models derived by Stevens or by Simkowitz and "‘onroe‘? There are several
possible explanations. Stevens may have had methodological problems
first using factor analysis and then using a discriminant model on
variables extracted, perhaps erroneously, from that factor analysis.
Stevens also may have rejected sane important ratios. S" appear to have
avoided those types of errors yet obtained a five ratio model.

Another possible explanation is offered: ”The entire process of

 

 

158

mergers appears to be much more complex now because there are many more
motives for acquisition. Simultaneously, because of bidding wars,
arbitrageurs have become more competitive.4 They may have successfully
discounted many of the quantitative measures, and may now be focusing on
non-quantitative measures. If the quantitative measures have not been
discounted, it is still necessary to use more ratios than those used in
this study to reduce the "‘ype I error levels observed.

1 . Factor Analysis

 

Table 5-7 perhaps best illustrates a surprising result, that
factor analysis as unable to successfully reduce the thirty seven ratio
dimension initially introduced. Seven factors only eXplained sixty
eight percent of the variance but need thirty ratios to accomplish that.
A greater nunber of factors is theoretically justified, but that
increased the nunber of required ratios to thirty five of the original
thirty seven ratios.

Factor analysis also produced results which conflicted with "DA.
Table 5-10 contrasts important financial ratios selected by use of "'DA
and those ratios identified by use of factor analysis. "ost
importantly, key T-"DA ratios were generally insignificant until the
fourth throth seventh factors which illustrates an obvious disparity in
results.

The difference in best sets of ratios derived from the two
alternating statistical procedures should help reconcile the
differences in "test" sets of ratios obtained by Stevens and ST.

2. Sequential Univariate Analysis

 

Tables 6—1 (a-c) and 6-2 illustrate two important considerations in

the decision to use EDA rather than F tests performed sequentially in an

159
analysis of variance. First, there were numerous contrasting results,
key ratios selected using “DA had low univariate T." levels of
significance, and 2., univariate l." test results tended to restrict the
number of ratios used. Indeed, Table 6-2 indicates that at = .05 the
number of ratios selected sequentially by F scores was eight, t_wq, and
six ratios, IAG 2 — LAG 0 respectively.

***************

1“. Ratio Selection Over Time

 

Because of the dynamics of the economy, industry and company
characteristics, firms' values change over time, and presumably those
changes are reflected in cwapany financial statements. Ultimately any
financial ratio analysis must consider those changes.

Previous mergers studies have focused on the accounting data
available immediately prior to a firm's acquisition. Conclusions
formed from those ratios are likely to be incorrect, especially ratios
using stock prices. Stock prices appreciate months, perhaps two to
three years prior to its acquisition. “inancial managers may
restructure the firm's financial characteristics months to years in
advance in an attempt to thwart would-be acquirers. Countering this,
acquirers are trying to accelerate the merger process reducing the time
available to potential acquisitions to defend themselves by increasing
the premiums. Read Welles [129] for an extended. discussion of
arbitrageurs. Read Rotbart [114] for a discussion of a failure to
defend against acquisition. 'f'he implication of the above is these
effects may have reduced the comparability of ratios useful in isolating
candidates in tte 1960s.

Perhaps the best illustration of the dynamics of the

160
pre—acquisition process is to study the ratio coefficients in 'T‘able 5-1.
Those ratios generally reflect financial strength, but progressing
across time, cross sectionally LAG 2 to LAG 0, you observe a
deterioration of the ratios. Py LAG O practically all the ratios have
negative coefficients which in most cases connotes below average
performance levels.

G. Limitations

 

Limitations in this test include the lack of ratios compatible
to certain industries inclining financial institutions and gas or
electric utilities. Therefore results from this test may be
inapplicable to those types of industries.

Of a more important nature, it has been established in this test
that ratios probably are unable to predict mergers, except perhaps the
LAG 2 model, especially if the sample sizes reflect the probability of
acquisition in general. Also, the financial ratios used to predict
mergers are different than those ratios used in tests conducted in the
1960s. At this time it is not certain that future economic or legal
circumstances will be able to improve the value of these models.

H. Future Research Suggestions

 

One of the motives for this test was a response to suggestions
from Stevens and 1"”onroe and Simkowitz to update their studies. Clearly
someone should ultimately replicate my results for purposes of
comparison.

A test for industry effects needs to be completed. Such a test
was planned at the onset of this test, but rejected because sample si res
were too small to produce statistically significant results.

A third necessary test is to evaluate the predictive ability of

161

this model over time. Can nonquantitative factors improve the ability
of these models to predict mergers, and if so which factors? Also, a
test to observe when stock prices begin increasing for potential
acquisitions is warranted. Finally, the ex ante prediction rates raise
the possibility of two sorts of acquisitions, some become candidates
three years prior, but others do not become vulnerable until one year
prior. A test ought to investigate that possibility.

What have we learned? we know that financial ratios may not be
able to predict ex ante. Such a test has never been done before. It is
difficult to assess the magnitude of such a result without investigating
premiums offered existing shareholders and when prices of acquired
firms' stock began to increase. Pecause of the LAG 2 model results,
there is some evidence that a.financial ratio based model may serve as a
predictor of an upcoming financial event.

Pecause the IAG 2 model produced the highest ex ante prediction
success, and because market price sensitive ratios did not appear to
begin changing until the LAG-1 model it still may'be possible to Obtain
by hedging, risk adjusted excess profits, whiCh in this amqflication
would be an exception to the semi—strong fbrm of the efficient markets
hypothesis; that there is no publicly available financial information
which can assist an investor pursuing those excess profits. Hewever,
results in the aggregate suggest the likelihood of this is remote.

What we have learned needed to be empirically tested . "-"otives for
mergers have changed since the 1960s. Higher interest rates and greater
economic uncertainty have increased the demand for physical assets and
the market value of debt has decreased. During periods of rising prices

physical assets appear likely to appreciate in value based on the 1970s

162 '
experience.

Can we predict in a different environment? Perhaps, if inflation
will subside, natural resources type assets and depreciable fixed assets
will diminish in importance, interest rates will fall reduc ing financial
risk, and managers of firms will believe it is cheaper to build than it
is to buy. Subsequent tests should provide differing sets of key
ratios.

It is important to realize that no model will ever explain
rationally all of the motivations for merger activity. 'T‘here are
several reasons for this. first, managers will not always act in a
manner which maximizes shareholder returns. 'T‘his was discussed
extensively in Chapter II, but reiterating here Duffett contends that
most managers place themselves by their sales level ranking on the
Fortune 500.5 He also suspects many of those managers do not . . . know
where his corporation places on the list which Fortune just as
faithfully compiles ranking the same 5m corporations by profitab il ity.6

Finally acquirers may be over Optimistic in their projections of
an acquisition's mture cash inflows or of their ability to more
efficiently manage another firm. "'ost of the evidence brought forth in
Chapter II indicates that firms pay excessive sums for their
acquisitions. Euffett likens this to the "princess and the toad" story.

T“any managements apparently were overexposed in impression

able childhood years to the story in which the imprisoned

handsome prince is released from a toad's body by a kiss

from a beautiful princess. Consequently they are certain

their managerial kis will do wonders for the profitability

of Company T(arget).

. . . investors can always buy toads at the going price for

toads. If investors instead bankroll princesses who wish

to pay double for the right to kiss the toad, those kisses

had better pack some real dynamite. We've observed many
kisses, but few miracles. Fievertheless, many managerial

163

princesses remain serenely confident about the future
potency of their kisses - even after theig corporate
backyards are knee—deep in unresponsive toads.“

n4

Tibotnotes

Chapter VIII

"Asset Ped eployment - Werything Is T“or S‘ale Vow," T‘usiness ‘.=7eek, 24.
August 1981, p. 68.

 

T"aurice 0. Joy and John O. Tollefson, "01 the Financial Applications
of Discriminant Analysis," Journal of Financial and Quantitative

 

Analysis 10 (December, 1975): 727.

Jack O. Vance, "Is Your Company a "‘ake—over Target?" Harvard
Pusiness Review, Fay-June, 1969, p. 94.

 

A humorous , but important article assessing the current state of the
arbitrage "game" is in

Chris Welles, "Inside America's Arbitrage Game," Institutional

 

Investor, International Iliition, August, 1981, pp. 214-216.

Warren E. Tiif‘fett, Annual Report (Omaha: Perkshire-Eathaway, Inc.,
[1981]). p- 4-

Ibid.

 

Ibid.

Ibid.

“4

10.

11.

12.

PBLIOG‘RAIHY

'T‘PX'.‘ S

Alberts, W. W. and Segall, J. P. "be Corporate 1"erger. Chicano:
University of Chicago Press, 1966.

 

Cooley, William W. and Iohnes, Paul R. Y'ultivariate Data.
Analysis. New York: John Wiley 9r. Sons, 1971.

 

Copeland, Thomas “. and Weston, J. Pred. Pinancial 'T‘heory and
Corporate Policy. P'enlo Park: Addison—i=lesley, 197T

 

 

Dean, Joel and Smith, Winfield. The Relationships Between
Profitability and Size. Edited by William W. Alberts and Joel P.
Segall. rPhe Cogporate Merger. Chic%o: University of Chicago
Press, 1966.

 

 

 

Durand, David D. "Risk Elements in Consumer Installment
Financing." Studies in Consumer Installment Financing.
New York: National Bureau of Economic Research, 1941.

 

Finn, Jeremy. A General 1"odel For Multivariate
AnalEis. Pew York: Holt, Rinehart and Winston, 1974.

 

Green, Paul P. Analyzing T'O’iultivariate Data. Hinsdale, Illinois:
Dryden Press, 1978.

 

Hair, Joseph ‘3. Jr.; Anderson, R. 13.; r"atham, R. L.; and
Grablowsky, B. J. T-‘ultivariate Data Analysis With
Readings. Zulsa: Petroleum Publishing Company, 1979.

 

Haley, Charles W. and Schall, Iawrence D. "‘he Theory
of Financial Decisions 2nd ed. Tiew York: ”cGraw-Hill Rook

 

 

Company, 1979.

Harman, Harry H. "odern T"actor Analysis. Chicago: University
of Chicago Press, 1967.

 

Farris, Richard. J. A Primer of Multivariate Statistics.
New York: Academic Press, 1975.

 

K]. ecka, William R. "Discriminant Analysis," Chapter 23.
Statistical Package for the Social Sciences, 2nd ed. Pd ited by
Norman H. Nie, C. Hadley Hull, Jean G. Jenkins, T-V'arin Steinbrenner
ani Dale H. Pent. New York: VcGraw-Hill, 1975.

165

13.

14.

15.

16.

17.

18.

21.

22.

23.

24.

166

Lev, Paruch. Pinancial Statement Analysis: A T'Eew
Approach. Biglewood Cliffs, Pew Jersey: Prentice-Hall, 1974.

 

1P’orrison, Donald P. “'ultivariate Statistical ”ethods. TTew York:
“'bGraw-Hill, 1976.

 

Rao, C. R. Linear Statistical Inference and Its
Applications. New York: John Wiley 9r, Sons, 196“.

 

 

Reid, &3muel Richardson, fergers,;‘ana_,:ers, and the Pconomy. i‘ew
York: F’cGraw-Hill Book Company, 1968.

 

Roberts, D. R. Pxecutive Compensation. T:E‘ew York: 'Ihe Free Press
of Glencoe, 1959.

 

Rubinstein, Park E. A Ween-Variance Synthesis of Corporate
Pinancial rL‘heory. Edited by Stewart C. Pyers. l‘odern
DeveIOpments in Corporate Pinancial r"heory. Wew York: Praeger
Publishers, 1975.

 

 

 

Scherer, P. “v". Industrial Parket Structure and Rconomic
Performance. Chicago: Rand T‘ci’ally College Publishing Company,
1970.

 

 

Singh, Aj it. Takeovers; (“heir Relevance to the Stock
T’arket and the Theory ofthe Pirm. Cambridge: Cambridge

 

 

University Press , 1 971 .

rI‘atsuoka, Maurice H. "ultivariate Analysis. New York:
John Wiley :9: Sons, 1971.

 

Weston, J. PTed. Determination of Share Exchange Ratios in

 

Vergers. Edited by William W. Alberts and Joel P. Segall. ”he

Corporate Perger. Chicago: University of Chicago Press, 107,196.

 

WPORmS -— PTNLISEFTD

Puffett, Warren P. Annual Report. Omaha: Perkshire Hathamy,
Inc., [1981].

 

ART IC IFS

"Addirg Foreign Stocks Por a Safer ”ix." Rusiness ‘rfeek
February 27, 1978.

 

"A Financial Empire In the ”along." Rusiness Week,
"arch 9, 1981, p. 87.

 

Altman, Iliward I. "T‘inancial Patios, Discriminant Analysis
and the Prediction of Corporate Panhuptcy." Journal of Tninance
23 (September, 1968): 599-609.

 

‘éf

27.

D.)
Q)

31.

33.

34.

37.

38.

40.

167

, and Iisenbeiss, Robert A. "Financial Appli—
cations of Discriminant Analysis: A Clarification."

Journal of Financial and Quantitative Analysis 13 (“‘arch, 1978):
185-195.

Amihud, Yakov and Lev, Baruch. "Risk Reduction As A ‘fanagerial

Votive for Conglomerate Pergers." The Bell Journal of Economics
12 (Autumn, 1981): 605-617.

 

 

 

" Asset Redeployment: Pverything Is For Sale Now." Business
Week, (August 24, 1981): 68-74.

Paker, H. Went; Filler, Thomas 0.; and Ramsparger, Prian J. "An
Inside look at Corporate Pergers and Acquisitions." EST] TOpics 29
(Winter, 1981): 49-57-

 

Parnard, M. 1‘7. "'Ihe Secular Variations of Scull Characteristics in
Pbur Series of ngptian Slmlls." Annals of Eugenics VI (1935):
352-371.

Baumol, William J. "(h the rTheory of the Expansion of the Pirm."
American Economic Review 52 (December, 1962): 1078-1087.

 

 

Beaver, William H. "Financial Ratios As Predictors of Pailure."
Journal of Accounting Researdm 4 (Supplement, 1966): 71 -111.

 

Belkaom’, Ahmed. "Financial Ratios as Predictors of Canadian
Takeovers." Journal of Business Finance 8: Accounting 5
(Spring, 1978): 93-107.

 

Peman, Lewis. "Phat We learned Prom "he Great Perger Frenzy."
Fortune, April, 1973, PP. 70—73, 144, 148-150.

Pettner, Jill. "Pbreign Stocks Catch On With Small Investors;
Cains Are Bigger Lure Than Diversification." Wall Street Journal,
20 April 1981, p. 42.

 

Block, fianley. "She T‘r‘erger Impact on Stock Price Povements."
PSU Business quics 17 (Spring, 1969): 7-12.

Bradley, Michael. "Interfirm Tender Offers and the T‘Tarket
for Corporate Contro ." Journal of Business 40 (October, 1980):
345-376.

Brick, ebhn R. and Thompson, Howard E. "ihe Economic Life of an
Investment and the Appropriate Discount Rate." Journal of
Pinancial and Quantitative Analysis XIII (December, 1978): 831-
846.

Carberry, James. "F’any REITs Stage Comeback, Aided Ry an
Attraction of Foreign Investors." Wall Street Journal, 6 August
1979, p. 6.

 

 

 

 

41.

42.

43.

45.

46.

470

48.

49.

53.

54.

168

Carley,William ”. "Airline—Service Breakdown Persists long After
Pan Am, National T'erge." Wall Street Journal,
10 ”arch 1981, p. 25.

 

Chappell, Henry W., Jr. and Cheng, David C. "prectations, Tobin' s
q, and Investment: A Note." Journal of Finance XXXVII (T'arch,
1982): 231-236.

Conn, R. L. and Pielsen, J. P. "An Tmpirical Test of the Iarson—
Gonedes Ikchange Ratio Determination Podel." Journal of Pinance
moon (June, 1977): 749-760.

 

 

Dodd, Peter. "Commpany Takeovers and the Australian Equity
Market." Australian Journal of Panagement 1 (October, 1976):
15-35-

 

. "Merger Pr0posals, A’anagement Discretion, and

 

Stockholder Wealth." Journal of Financial Economics 8 (F’arch,
193)): 105437.

and Ruback, Richard. "Tender Offers and Stockholder

 

 

Returns, An Empirical Analysis." Journal of Financial Economics
5 (December, 1977): 351-373.

"DuPont's Costly Bet on Conoco." Business Week, July 20, 1981,
pp. 52-54.

 

 

mrand, David. Risk Elements in Consumer Installment Lending,
Stuiies in Consumer Installment Financing, No. 8, National Bureau
of Economic Research, Inc., New York, 1941.

 

Dyson, R. T. "Mergers Are Not Growth." Business Week, August 10,
1981, p. 98.

Pdmister, Robert 0. "An Empirical Test of Financial Ratio Analysis
for Small Business Failure Prediction." Journal of Financial and
Quantitative Analysis 7 (March, 1972): 1477-1493.

 

 

 

Fhrbar, A. F. "Corporate Takeovers Are Here to Stay." Portune,
Pay 8, 1978, pp. 91-100.

Eisenbeis, Robert A. "Pitfalls In the Application of Discriminant
Analysis In Business, Finance, ani Pconomics." Journal of Finance
'52 (June, 1977): 875-900.

 

,Problems in Applying Discriminant Analysis in Credit

 

Scoring Models." Pederal ReserveRulletin, January, 1978, pp. 14-

28.

Ellert, J. C. "A’ergers, Antitrust Law thorcement, and Stockholder
Returns." Journal of Finance XXXI (Pay, 1976): 715-732.

1?-

55.

56.

57.

61.

62.

63.

64.

65.

66.

67.

68.

69.

Journal '58 ("arch-April, 1982): 24—26.

Research 2 (August, 1965): 250—8.

169

1lerguson, Robert and Poyicin, Philip. "Pulling Rabbits Oit of Eats
in the Oil Iusiness—and Elsewhere." P‘inancial Analysts

 

Pisher, R. A. "The Use of ”ultiple :‘"easurements in Taxanomic
Problems." Annals of Eugenics VII (1936-1937): 179-188.

 

Prank, Ronald E.; T-‘Tassy, William F.; and Iforrison, Donald G. "Bias
ln i-‘fultiple Discriminant Analysis." Journal of .Warketing

 

Gort, T'ichael. "An Pbononic Disturbance Theory of “'ergers." The
Quarterly Journal of Economics LXXXIII (lTovember, 1969): 624-642.

 

"’Ihe Great Takeover Binge." Business Week, Ifovember 14, 1977,
pp. 176-184.

 

"Growing From Within Pay Pay Off faster." Pusiness Week, September
17, 1966, pp. 44-46.

 

Haley, Charles W. and Schall, lawrence D. "Problems With the
Concept of the Cost of Capital." Journal of Financial and.
Quantitative Analysis XIII (December, 1978): 847-870.

 

 

Halpern, Pad. J. "Empirical Estimates of the Amount and
Distribution Cains to Companies in I'ergers." Journal of Business
46 (October. 1975): 554-575.

 

I-Iaughen, Robert A. and Iangetieg, Terence C. "An Empirical Test
for Synergism in T'erger." Journal of Pinanc: 30 (September,
1975): 10034014.

 

Higgins, Robert C. and Schall, lawrence D. "Corporate Panh‘uptcy
and Conglomerate F'erger." Journal of Finance 30 (larch, 1975):

Hogarty,flhomas F. "'Ihe Profitability of Corporate I‘ergers."
Journal of Business XLIII (July, 1970): 317-327.

 

 

Honeycutt, T. Crawford. "’Ihe T-Zicroeconomic Consequences of
Corporate i'fergers: A Comment." Journal of Business 48 (April,
1975): 267—274.

 

Bong, Hai; Iu’aplan, Robert S.; and i‘v'andelker, Gershon. "Pooling vs.
Purchase: The Effects of Accounting for Wergers on Stock Prices."
The Accounting Review Llll (January, 1978): 31-47.

 

Horrigan, James O. "The Determination of Long-Term Credit
Standing With Financial Patios." Journal of Accounting Research
4 (Supplement, 1966): 44-62.

 

"Ira. Harris: Chicago's fig Dealmaker." Pusiness Week, June 25,

 

‘E'?

70.

71.

72.

74.

75-

77.

LJSO

170

Joy, 0. Pvaurice and rT‘ollefson, John C. "On the Zinancial
Applications of Discriminant Analysis." Journal of Rinancial and
Quantitative Analysis 10 (December, 1975): 723-739.

 

 

. "Some Clarifying Garments (h Discriminant
fl ‘0 O I C O
Analysst. Journal of I: manual .8: Quantitative Amalgam 13

(Z'arch, 1978): 197-200.

 

T'Vaiser, H. 17'. "The ‘hrimax Criterion For Analytic Rotation In
Factor Analysis." Psychometrila 25 (September, 1953): 137—200.

 

Tlamerschen, David R. "A theory 0f Conglanerate "erxzers:
Comment." Quarterly Journal of Pconomics LIXDIII
(November, 1970): 665-673-

 

?Teown, Arthur and Pinkerton, John I. "Larger Announcements and
Insider Trading Activity: An Empirical Investigation." Journal
of Finance new: (September, 1991 ): 855-869. “'—'"

 

I’Cohers, Theodor and Conn, Robert L. "’Ihe Effect of Yerger
Announcements on Snare Prices of Acquiring Iirms." Review of
Business 8; Economic Research XII (Fall, 1976): 55—62.

 

Iangetieg, ri‘erence. "An Application of a fires-Factor Performance
Index to Veasure Stockholder Gains Pram 7"erger." Journal of
Financial Economics 6 (December, 1978): 365-585.

 

 

Iarson, K. D. ani Gonedes, N. J. "Insiness Combinations: An
Mohange Ratio Determination iodel." The Accounting, Review 44
(October, 1969): 720—728.,

 

Ienzer, Robert. "my High Sell Higher." Parrons, 7 “Tarch, 1977,
pp. 11, 26.

Iev, Paruch 8: T’andelker, Gershon. "The T-v'icro-economic
Consequences of Corporate T-‘ergersJ' Journal of ~Eusiness 45
(January, 1972): 85-104.

 

. "Rejoinder to Reid 19: I‘bneycutt." Journal of

 

 

Business WApril, 1975): 281-2.

Levy, Heim and Sarnat, Marshall. "Diversification, Portfolio
Analysis and the Uneasy Case for Conglomerate F'ergers." Journal
of Finance 25 (September, 1970): 795-802.

 

. "International Diversification of Investment
Portfolios." American Economic Review LX (September, 1970):
668—675.

 

Iewellen, Wilbur G. "A Pure financial Rationale for the

84 .

86.

87.

92.

95-

95-

96.

171

Conglomerate ”erger." Journal of Rinance XXVI (Pay, 1971 ): 521-
557-

 

, and Huntsman, Elaine. "l”anagerial Pay and

 

Corporate Performance." American Economic Review LX (September,
1970): 710—720.

 

Lintner, John. "Ikpectations, Mergers and Pquilibrium in Purely
Competitive Securities Markets." American Rconomic Review LXI
(May, 1971): 101—111.

 

Iogue, Dennis E. and T‘Iaert, Emilippe A. "A 'Iheory of Conglomerate
Pr'ergers: Comment and Extension." Quarterly Journal of Economics
IXXXVII (November, 1970): 663-667.

 

Vandelker, Gershon. "Risk 8: Return: The Case of the 7"erging
Firms." Journal of Financial Economics 1 (December, 1974): 303-
555-

 

..“rianne, Henry G. "l‘v'ergers and the "iarket for Corporate Control."
Journal of Political Economy LXXIII (April, 1965): 110-120.

 

"i‘r’arket Commentary: Mergers and Acquisitions - How to Buy An
American Company." Euromoney, November, 1976, p. 141.

Marks, Sidney and Dunn, CZLiver Jean. "Discriminant Functions When
Covariance I‘atrices Are Une ual." Journal of the American
Statistical Association 69 C(June, 1974): 555—559.

McGuire, Joseph 11.; Chiu, John S. Y.; and ELbing, Alvas 0.
"Executive Incomes, %les, and Profits." American Economic
Review 52 (September, 1962): 753—761.

 

 

 

Nelicher, Ronald W. and Barter, 'Ihomas R. "Abstract: Stock Price
Movements of Firms Engaging in large Acquisitions." Journal of
Financial 8:. Quantitative Analysis 7 (Perch, 1972): 1469-1475.

 

 

"Merger Leaks Abound, Causing I’any Stocks to Rise Before the Pact -
Moves to Keep E ans Secret Often Pail; But Proving Vh‘ongdoing Is
Difficult - Using Relatives, Striss Ranks." Wall Street Journal,
12 July 1978, p. 1.

 

m’erjos, Anna. "Takeover Targets." Rarron's, 15 ”ay1978. pp. 9,
12. 14. 52. 34. 55-

:‘fetz, Tim and Joseph, Raymond A. "C'arborundun Stake May Be Bonanza
For Broker Who Foresaw Bidding 11’ ." Wall Street Journal,
28 November 1977, p. 20.

 

I’eyerson, Adam. "Merger it"ania and the High Takeover Premiums."
Wall Street Journal, 20 July 1981, p. 14.

100.

101.

102.

105.

104.

105.

106.

107.

108.

109.

110.

172

F'Tonroe, Robert J. "Comment: Financial Characteristics of T‘erged
Finns: A Multivariate Analysis." Journal of Financial and
Quantitative Analysis 8 ("arch, 1975): 163—165.

 

 

"The 1’orning After - Post of the Stockholders Who Accepted Those
Fancy rBender Offers Now Have Reason to Regret iheir Taste."
Forbes, February 1, 1970. pp. 15-16.

Norrison, Donald G. "On the Interpretation of Discriminant
Analysis." Journal of Marketing Research 6 (Way, 1969): 156—
163.

 

“:"ossin, J. "-1 uilibrium in A Capital Asset "arket."
Econometric: 34 October, 1966): 768-83.

 

Mueller, Dennis C. "A Theory of Conglomerate Mergers."
Quarterly Journal of Economics LXXXIII (November, 1969):
643-659.

 

. "A theory of Conglomerate T'ergers: Re y."
Quarterly Journal of Economics LXXICIV (November, 1970 :
674-679.

Myers, H. and Porgy, E. w. "DeveIOpment of Numerical Credit
Evaluation fistems." Journal of the American Statistical
Association 50 (September, 1963): 797-806.

 

 

 

 

Myers, Stewart C. "Procedures for Capital Budgetirg Under
Uncertainty." Industrial P'angement Review 9 (Spring, 1968): 1-
19.

Neter, John. "Discussion of Financial Ratios As Predictors of
Failure." Journal of Accounting Research 4 (Supplement, 1966):
112—118.

 

Nielsen, James P. and ifelicher, Ronald W. "A Financial Analysis
of Acquisition and rv'lerger Premiums." Journal of Financial and
Quantitative Analysis 8 (F-‘arch, 1973): 139-148.

 

 

O'Connor, Melvin C. "(h the Usefulness of Financial Ratios to
Investors in Common Stock." The Accounting Review XLVIII
(April. 1973): 339-352-

 

Patton, Arch. "Deterioration in rPop ercutive Pay." Harvard
Business Review 43 (November-December, 1965): 106-118.

Pekar, Peter Paul. "Pergers and Acquisitions: Improving the
Odds." Cash Flow1 (may, 1980): 47-53.

Retzinger, rIhomas Jr . "Troubled Couplings: To Win A Bidding Var
Doesn't Insure Success of Verged Companies." Wall Street
Journal, September 1981, pp. 1, 12.

 

Vi-

111.

114.

115.

116.

117.

118.

119.

120.

121.

122.

123.

124.

173

Pinches, George, and .‘~"ingo, I’ent A. "A ’bltivariate Analysis of
Industrial Bond Ratings." Journal of Pinance 28 (T'arch, 1973):
1.180

 

"The Profit Potential In Spotting Takeovers," Pusiness Week, 24
Octobe 1977, pp. 100-102.

 

Reid, Muel Richardson. ""he Vicroeconomic Consequences of
Corporate F’ergers: Comment." Journal of Business 48 (April,

 

Rotbart, Dean. "J”arathon Oil ’I‘Xecutives Sharply Criticized Ibr
Wot Building Petter Rakeover Defenses." Wall Street Journal, 21
August 1979.

Rout, Iawrence, "Tbying American: Weak Dollar, Stocks Spur
Foreigners to Seek Acquisitions in the U.S." Wall Street
Journal, 21 August 1979, pp. 1, 13.

 

 

Scott, Ilton. "01 the Financial Applications of Discriminant
Analysis: A Comment." Journal of Financial and Quantitative
Analysis 13 (I’arch, 1978): 201-5.

 

Shae, F’aria. "Wining Companies Become Targets of Oil Industry's
Drive to Diversify." Wall Street Journal, 11 ‘Tarch 1981, p. 25.

 

Simkowitz, T’ichael and Monroe, Robert J. "A Discriminant
Analysis Function For Conglomerate Targets." Southern Journal of
Business 6 (November, 1971): 1-16.

 

Singh, Ajit. "Takeovers, Economic Natural Selection, and The
Theory of the Firm: Bridence from the Post-war United Tfingdom
Experience." The Economic Journal 85 (September, 1975): 497-
515.

 

Smith, C. A. B. "Some Examples of Discrimination." Annals of

Eggenics 13 (1946-1947): 272-282.

Smith, Veith V. "Comment: A Financial Analysis of Acquisition
and I"'erger Premiums." Journal of Financial and Quantitative

Analysis VIII (larch, 1973): 159-162.

Solnik, Bruno H. "Why Not Diversify Internationally Rather 'I’han
Domestically‘?" Financial Analysts Journal 30 (July/ August,
1974): 48—54.

Stevens, Donald L. "Financial Characteristics of TergedPinms, A
T“ultivariate Analysis." Journal of Financial and Quantitative
Analysis 8 ("arch, 1973): 149-158.

 

 

Tobin, James and Brainard, William C. "Asset T‘iwkets and the Cost
of Capital." Cowles Foundation Rapier No. 440. .T-Iemr feven:
Cowles Ibundation For Research in Economics at Yale University.

 

127.

128.

129.

r

130.

132.

133‘.

1711
1977, pp. 255-262.

"'Ihe rTrend r“oward Buying Quiet Companies." Pusiness Week, 22
Novembe‘ 1976, pp. 112, 116.

 

Vance, Jack O. " Is Your Company A Rake-Over Target?" Uarvard
TBusiness Review (Hay-June, 1969): 93-98.

 

Von Purstenberg, George ‘2; “’alkiel, Burton G.; and Watson, Harry
S. "Ikpectations, Robin's o, and Industry Investment." Journal
of Rinance K‘DIIV (Pay, 1979): 549—561 .

 

Walter, James IT. "A Discriminant Tunction for Iarninfls Price
Ratios of large Industrial Corporations." Review of Roonomics
and Statistics XLI (February, 1959): 44-52.

 

 

Welles, Chris. "Inside America's Arbitrage Game." Institutional
Investor, International I‘dition, August, 1981, pp. 213-231.

 

Weston, J. Pred. and Rice, Rimard "7. "Discussion." Journal
of Rinanoe Ail-XI (Pay, 1976): 743-747.

 

Zonana, Victor. "Planning a Ilant: How A. H. Robins Co. Wade
Decision to Build A Chemical T‘acility." Wall Street Journal, 22
October 1975, p. 1.

 

Zucker, Seymour. "'Ihe Q—Ratio: Fuel ~br the Verger Vania."
Business Week. 24 August 1981, p. 30.

 

CG'PUCFR ACCISSRD DATA PASRS
Compustat. Standard and Poors Corp., Prglewood, Colorado.
IBIPUBLISTIPD ’I’AWRIAL
Gabhart, David R. L. "An Inquiry In Construction of An Accounting
Predictor of luture T‘unicipal Financial Insolvency." Ph.D.
dissertation, I“'Tichigan State University, 1977.
Stevens, Donald L. "A E‘ultivariate Analysis of financial

Characteristics of Acquired Rims In Industrial T'ergers." R1.D.
dissertation, Ti’ichigan State University, 1972.

.1.

APPENDIX

Subfile l, Merged Group

 

Acquired Firms

 

ROLEUH CORP
QQRF

EAL

S
2.—
ID.
NR

PTSA
RCOR
0. cN.
C.LI

R299.

 

0..
Til

w

.9me0
9.2:;
«H
SRSMU

 

ITL
TNL.
RUIH
AC”.

OXC JUD—
Uu..Mr... ”CEMC.
n 552! 1.1.1.

LLL

EII LLL-
,AAAA AAA,

CO

1:40

 

0N STEEL MFE
SPECIALTY

NUH

ALLIS
ALUPI
AHCC

inns

JDIE Cf

\
1

11K

.

 

 

USTRIAL CORP

 

INC

 

503

 

.u C

.m .b

.2. ON.

«LI

3 C

U .MN

JL 8b

C 05
L r... C
CH.. ON
recur-N I

.31.” .Al:
50 CPLC
CRH T. AC
vino. KJS
11¢.”QIR)
5H?” .tn...
3 T: “:34
DR. .HIH
RAF. HGN(
O_t0_ hUU
FB. In. 3
K KATIQHR
LbC IRSK
TAIU RULY
688 63829

 

L
I
U

.f

NS
.U.“
Irv
TI
LAT
.tRJ.
.COO
TLP
SPXS
XED
IE N
“I
UTA
NRIA
A. EUG
VAC
F. LNN
“.93
CCCC

 

0
C
3

C C S
C I 5
NJ R 39...:
1001 omKaK

CC BOP.
SH E CK
.U U’LP o E
NUIERAN
IN C1CL
UJYC RA
Nd(R ETC
UO UOCEI
33.... .RNPN
KR; Rafti
A “AuthcflHH
CCCCCCCC

 

 

 

175

6
7
l

P
on.
0
C

V...
LS
ED
VN

RP

JAII
Con A

CNN.
Y. . C.
353.:
NKKE
KER-r.

.EAAF
.Hlffiu

Cavern

CU lNC

UNE
S CORP

SI
ENT

0

O
.ti
L
CDC
NINS
NBA
.IR A
. LI
_tSAry
FUNNY-3H
ElNF.

LLLH
.UnUOQ
.DCCC

 

LLFUUHH

D.

R

0

C
S G
R P N
.t R .1
N 0 .l
G C S
I A
S L C
.t EU any
0 E.CC.A.N
«J: r AUDI
9+ SC R
0 P IE3.)
CSRLR.U D
TOATANN
TNCTCAA
Saw hERI
ETEEL HT
TLNNtRTE
.UUVIL .CNN
f..\.STIKFrLC
HNNNUPRR
COOCUCOCV
CCCQCCCC

 

 

CHHUNICATIONS 1N0
MT_A>SJC
JLCUN

C
G
R

HT

.UTLH

ACAH
Katrtnu
ORRU
CCCC

 

 

D.
R
O
C

S
E S
I C
PR I
RU N
0r. _U
Aha in R
RCFJP T.
LCNRR C
LBIAUPE
.UA PCRLL
hit)? nBtI
DANG Q
SRN v. L
IEPLNIIPLN
.IT. CIIEC
RTJVKRNS
UUHEIOYA
CCUDDODE

 

 

Pnbcc

.HSAE

EEFF

 

is ELEC no

.0
NSA. BCEC
.BD( Y£UE
LN gkkLS
RIRATA
O ERA

VCRIELNRH
BYRNREO
LRAEELL
FFfiGGGG

 

E ,

 

FOG..UG .

T SCOTT SUPERHARKETS

HOUND CCHPUTER
ERY STORE PROUUCTS CO

DON Imps INC

«luv-“ban.

 

ICAN LOKP

 

- C
.UDF.

INC

N

CED
NIN.
.lSI

CCIY
Nour-
.L US

I.RE
LK J
ASAA
NM

RALR
AEEE
HHfiH

 

 

.CGT

 

A__

PRUUUCTS CORP
-CL

L CURP

A
m?

b CC
FCCC
PERI
BC?

H
F

F
"LCIR

O
E

N
A
E
I

CR.

YY.N.
HHIL

 

 

177

 

 

 

 

 

 

 

 

   

m ..
P i 0 0 C.
R Z M C C D. R
0 T. Q . C S K S P
C I j H C NTT O C Q x
R S C N INC C C PM E
N C TCTY T I P C EU RIC
.c p: .N NNCS .u N .R NJMO r.n7o C N N
K H I.EIUB. H E H O IREC N RN CR1 I
C T M 01. D U C ECRC E OI A A C
I P G WSOL. R R L SK PM H CO R 3 T N
H Y R F N CdR SE A A C t ECU I P LOA E N f
C O O 00 I IP+ NU G N R PIANO I LICCINOU O
T C C S PIA ALI U RRPAUS U U ICU DC
0 S U NcaLSRHN 6P IT 5000 LOE G d)NNL M R
E R c... EL 0 ANSLDOET RR TNS DICCTFTFN .t PJAARRRLsuc.
ISCCPE C L PH EINCRT AC TCOEIR Rt AER IC NI C IOEIF
RDCNRK A R R LEI BI UCCRNCI EPCLGLREOCLNO HJI ATTCCS
UFN ICC P A OPORCIN N H G CAI RNHRC FIOdFIRIT .51RR AN AL
HC ID CO 3 H.CIC0 SCSNR. SL H EAATO NHOE AI U EFTVINRV
Y RP BPCPC H LHCCASF IX RTHLCC C AIINAPAO KYHMSE ABAC
HLKEDIKRRRR. NS+ 06 ITTERNOOHET IR IOTLRR R C RRA ALCE.DN
LCEFTOEOEON AR HR .TNSNOU VTNiNEdDQLCS UUnKbE 0R EEICL I
AUNSSLKCACIAKEHGASYSO I CSAAOOUIN RTLSESUURRCOR YEHHH FOEE
DTAGBKC V GET AIBICEC INAMPCICLUAAAhESSEAYRE 81 151 CKIE
NNRNNII VON AAHCHBGNNNEDG”HNRRSUCIDNDLSSOHLNI HHR RICNNRP
EEEIILN TVA EEEEEEIOOOOPAAAAAAAAACCEEIIIIOOOOU EEOA OTAAAAE.
KKKKKKKLLLL LLL LLLLLLLHHHMHHMHMMHHMMHHHMMMHHH NNN.NOPPPPPJ

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

O
C
A
. R P U
L .. .L R C
C C” 6 C C P C 8
W N .. M C N E SR N C R
I C P“ U 3 I L UOP I h C
G N Rm L C D a NCR I C
NS I 0Q NPN 0 A m CE I O S P S
S C , NN CC + IRI C C NV CC I H E PL R
N N .1090 Y _ ha I Ir... NH .u D C RA E
00 I HIRN A SC. 0 MC... ’RU .l AIM .r. a, C R 0N T.
SC TOION HI 0 P A EN 5 SO M30 H0 C U CC C EC
S RPEAC CE ERS 03R LC‘ T: E DH S as UC N L PCN. I N .41 J
)LLE RNC S C STIYSS NDONII 3 0R NRSLKH R I O RSQCT SI CH C

PPAEROA CIIROGA ESC AONCYCCNOU RUECQ IIDLCOPIfS. P Q CHINA 0 TI.

R NCCC.CTLECFR IYEHNONLC E ECNHR I ANCAHRS C SRSPPCRTIN NS FLO D
CObC.T NBPCH RSLRISIPFHDHHN CILSPC IKPICNLN\ 000 R CUR I0 REC A
hC(ICSL DEUAF TOT EI A A IHSHI RIF .CILI NCNLOELuLE NCOL OE
I TN C LCPP LLCSN AYHRUKCKCC OYCSSSDDC A I IRCLTCITPPRINCEHCH

RSAILO 0R ALE UAO EOERODCUON +X ICAEHRRR HHO R 0 YAEUNRRO I T R
HEMN NT GARLHEVRDRMNL NAUNUNOO EYI BHAAAAAA C EAEPUUNRBIOOSR RRRRE
GMA N0 .N EEAHEfiNTARNIVURAIAHT NLLCK HIDDURNNN LIISCC CCUEOOEOEH
LYAEC C ICUGIKSKIEGUETIUENINRL,CPL CATRNhhAIlR FSF 18.15" CClhhhT
OLTLCCE ECDEVCCC FNTHORAAEAEEU HHEIILUAAAAAAEE MXH FRSSIAALLEICGLRA
LCUYAEE.HIIIIOOUHAAACCCEEHHHHH IIK.LOOPTTIIIII.EEIRUU..RTCLAEEIAAAE
PPPPRRR.. RRRRRRRRSSS.SSSSSSSJSSS SSS. SSSJSSSQSSS ITTJTIUJUUVJVVVJHuHN

I
o
o
w

 

 

 

 

 

 

WESTERN NUCLEAR INC
HHEzLABRAICR cox?
HHITERCROSS sroaes INC
W

HILSON CERTIFItU Foqos In;
NILSCN-SINCLAIR co

HILSON spourrnu COOUS co
YQNKERS RACENAJ Luc- _
YOC-HOO CHCCCLAT: agveaaua
YOUNGSTCHN STEEL seas co
NATIONAL INDS

 

yo

180

APPENDIX II

Subfile 2, Control Group,

Non-acquired Firms

 

 

.U S SUGAR CORP

CORP

NLHC
Inn 1
RAF
X‘sLl
ADCC
MEEA
NFHP

 

INC

065 CORP
URGES + CHERICAL

L.’ CORP

LF
DO
SO
R.

U
5th
39.

E
(

XELR
EHUU
IPGd

_

 

lhr

 

 

DI E PC
TOPLR S RN
L R0 E OI
RORN S C CC 5
LOCI AOR N SE
LI E G U IS C
IRLP JOE E R
REI + +OSVET U
RPOS ‘ BtENA 0
EU N LNRCIB SPP
MSLI LI P ERR
TS I0ALNA+ R00
.thA C RAIN CC
AAIBN E.SA G K

II. OVYRTILN

SDUIT CRRtFAItANE
ALFBAAHA JXER .vEEC FIT

NNLEUES- IICAM “8N
RAF?! «UNE NHLE 253.5.
«UCOGflI‘K UCGRSNSC

 

LUERS CUR}T
CORP

BUI
TRON
CRP

vLCC

A
E

.KILR
€an

NNU
EYL
53F

 

E INC
INC

RIG
0 0N
OOPI
C"

N§AC
C Sun
IH‘I
RC S
UAE

BBCN
IHAO
LCLT.
LSLR
AIAC
HFHN

 

NG CO INC

 

S

E 0
I C
SR
”Ho-00-
NSC o
AU J
RON 0
85C“
11‘
D T

TIILnN

N253...»
UVCH

 

IRRI.

ETA-I. t

 

 

 

C B
N
I L
C
S .
G P S
N 0 R
I H E
L 03 L
L C DL
I Y CTI
U MORD NLT
Tu CAN I S
0 GA EI
.80 pcnunb HFD
o INS CE
H LI RPSENC
oCLRDERUKAN
JNAUEHCBSMI
TI PTRC.UR
0 AA R CA
RARNMFOECFI
ERAOA CSN.E
KCUTCYNUINL
CAESLNEtLhB
LNALANRHRCC
HCEAHA‘UNARE
SCSRAFCACBH

BRANDS INC

181
NC

 

 

 

 

 

 

 

 

,
_
P
C R
- S 0 0
E C C P C
I N RCC C

R L I ONN NP R
T A C OS CII IR AP E
S 0 0 3 N N PCRA DVD. 0 R B
C C C C U 0 PI D. R F. CCRRSS SC L0 3
C N S N C 0 PI R SR ONH NOALL 0 L CC U
N I D .b Cf... f m N RT IUSEO R 3915 )ECEAA C C «RI 0 . D. R

C I 0 N N I DA TEIC E E1 .6 LCC NS CP P C OH R
N 0 R I I.) YE c CN IR PPRTL H... LCIIO+PI ID. R TU 0 f

I 03 R P N E DINNS Q. PSRIK AR ETB C RAUHTC R HI 0 E LE C
TELO N S PONAITPEEPRETSRP PCNAU NN CNE DOGU ES C C L E
S LILF H .IPLRRBNG RRUTROHSUAR CIPP IHEI H + oCUU HS 9. I LG I: R
RP R F C PRLC CACOROCANOCCUCLC D - P CSHUCCIU RC CICPRL V PIR H I
ARSTHAPT SOI CERLAPCCIC HDNCC RT. Yh RCJUEN «HI SCCRC SNROI RPRCTCIPT
GODS IRE CHU P ES S E NI. AFNTO OCIOHADMLGN PNNSIOCO E ENJR
ICLUXSCR L A th DANERLI Y PUACIR C HCLLREIIG IA C SC LPIICE
C U0 .0... ALSNC cOAGROCCTOA LL RORCC 93M GOAI LNN MHC KA C A CC
OAN R S NLG IT RN00 IIUNNLR OHC IR A NADNHH EKIUNS ARRR S LIOOR N
KRI .OK DEN LIE ICHEPNCOAEE C RELE TEA EEHPYLHLSOUTLEA. EYLRRC. SA
LaR “CL CISIprPCp; LSDCSUHCMHIP. W8 TEhFP Aha-5L c L..NTIR.“LR.\~CTI — "RILEELCA II
YLU YYT. ISRAGUUNNTAIASMTCCM LFPCHR CHRLLHINLAIEHVARCXNAAILHPSX CL
AUOA AUN AUD. NLCCUUTJHAOKEII POATIA ROAILEECIEMTCEUILEISHIOQHEEAEL
BCCF CHILNB SACHJMOSBCMCNRTK AFPSSH ABHNANRILSSSJRCEADKA CHUIIT.. AA

 

 

 

 

 

 

 

 

 

 

 

182

L

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

0 A C
C P N
S R I c S
E C N E
R o C S I _ N
I C SE C S I P
50.5 X .b C C L AIR N FL N CR 0
S 0R0 § E N N N C 00 Ner CID. I .30 C
P E 0N0 P PM I I I N RC 0T5 NPR N. NC
R I CIC R LR N 0 I T‘ ISU IE0 0 N P Y
00R 0 EOE I S 3 C mm “W0 EC R O R PSST G
P CCT SHHN C ECO F E E L CC NVN SN C I O RUCN O
R S LUCI T E I I b A I RLI EIE CA S C OBIE L
0 GSU ALRL L SLO RNCR R . S 5 IGN CL I C In 0
0 CP NSDCIDUHCE ER ANT TSS R LS E SPRNI I C S TEP N
C R IANNRUBCNE NEE CHIS SK. E ME WLNRTER 6H EOCPANNIPHS
DEC KLJIELSUIT RTCPER U LRL CV CSP HAO S A C F L RCIRTIGURCI
NCTC CC I TI SCESAR ALDPOO< NI R RC UNMPCN MITPCP TOARAQOEC
0 I 0 RNAYTLE NT OAHANRNH SIN I+O E 00 RNIC TNRN GECDPMECTU
SRRECHYfihMNI KLISCEC. TICI TC U I CCEAE hICC ABAICIWFN . CO
0! OC ATO EP+AA EID I+E C YTNX M4 CWMN ITRC DION C MGOLODLXEOC
R EH RHSDNH. LNKHL 0L MN SEEIEW RmmET 0 SA.CR HNN S A?O.EfiECR
AYDSNCK HAGOSRCEHBSCCNAOOILRwUu.EbdAS T SUOPEOOTIORC SHPRUITRAPO
HEI CBC, ICECSEITILCSSCBICRNRLL TMNA HS EEBEIRCACEEI K .IELICR Cr
CRRS CONULLPITTRRPBRthTSRAnCIM.3aLRW Y TMIPRXMRNRVNENC NEFGMCBNETG
IAI NROOULHLNACOEUAOAECACITTWH;PE§CENEWAOUOOEAAICCOHIH OFFIETAE
RBSU AB SVAABINNNRTHCHKAEN357 I KIK YCORPRCMCCDIABIMCSACMSTI
.

 

 

 

CONP OF AMERICA

ECTRIC CO

183

S

D

N

I

D

E

L

L

0

RC

TN

NI P

C C R

CR N O
OUIOCO

NMC C C

A. L H

CHGL AN

ITAIEEA

RLT. VGEF

EAYGNNP

MEACIUQ

AHMSSSI

fl

 

PR
RC
0AM
C.E
LT
COS
NNY
RIS
T
CCN
LIO
ERI
TT
00A
PIEC
RDLI
OAEN
CR U
AH
TETM
STIC
ETHC
TES
.YUH
€3.50
RFTN
.AAY
LMLTHJM

RAL STGNRL-CORP
CO

‘15 CORP

E
K
h
G

SON (E.F.)

URHO

 

 

P
RC
FCNP

m 0
XTC
EE
XPNS

 

RCIR.

 

 

P

R

U

C

S

C

I

N

0?

RR

W0 5

C T

E 7 NC
LRD EI

EET HRO
ILPUTC
{F R C
100 EC
TCC LF
C EM.J
EYH
REC RN

SE OI

LET BL
TEHbLH

CAOOUMVT NLE AC
AAACEIPSDBE

Ex} iNC-CL a

 

'4E3TCN CCRPfi

 

FDI INC

CORP
fiExECUTIVE INDUSTRIES

SHITH (5.0.)
TRH INC

 

UPMAN CORP

GR

ELL INTL

CORP

ROCKH

 

C
30 N
“C I
RC 3
TI E
bFS I
UIC R P
UTIPPT R
PNNSRRS c
RIEYCOU c
C IHCCD
COOP NCK
55.x INC
LA AE II
USRRR Y H
KEETU EKS
.LhHCQ— LN.“
INSEA Lfiu
TJFSH T56

 

 

 

 

 

 

 

N
0
H
H LP S
55 N U 0
N 3 TL 0 I C
SI EP 5 I H H
C E (P5 M NR+T TB E O S
N INRE I0 “NOT E0
,7 R901 P CLCLL CICS FNIC
TECR RSNR tL [T Y ROT
S SH TPUYIIT E NSGS ESIC
E UTCSRC AH dOUJN CHILI
I D IUD 5 GS CHCIRC DIR
RONOFDCS EDIE? MUTE EIT
TCISIN NHNLE SLCASHN UC
S DIED IR? EEC ADC—H E
UEDYfl LR LCF RTE CPS TSL
DRNT ECT N 50 L80 IGLEE
NAA DY R R NRE AYO AT
[ER INC? ICI ETNOhE EAN
PB Rd A ARA CT RE HTA
OSNSURNNd A ISDCBHN:WSI
. A0, . ACY FEE 60 a I
;MMWSSu%SLTBRNIhT.TET MFA
SfiLNSGLALLAE CCI FLS HLd
AHIAIICEtEEMUAON ALO 0U”

(HSH HRGLYDSE-PRU.TA3 CGH

181!

 

0
LC
HA
ERS
TUA
SIC
YA
SNN
0
SCI
nEN
GTUC
A N
ACNI
IIY
LL
CKR
USCC
LNCC
OCRI
CCBN

 

 

 

3

CE

R COCI
HC IC IR
PE R EIE
LOTCCLH
*ECCIIIC
E VIP
fEL RU
CE E C
631 SDI
IAV+ NR
LGR CAT
ES ILC
CSSAOLKE
II C BCL
ROC UOE
TNIE PR
CILL N
ELBL IOI
LLUI R S
EIPV UEN
. $5060
3531 ShC
NflkU.SfiS
COCO IQI
III HON

NSO
TCC

5N5
ROI

TCE
GEM
HL

BEL

NGI

MIR
NLE
cRM

EH
RIO
53C

 

ERA
HIT“

ANC‘

 

0
cC
C
N #0
C I N C 0
N II SN C
I S T EI
EEUS R 1
R R CR 03 o
.t CO <5§u.t u
L NT HC I .
LOIS ET R F
IC I0) TF‘
H pU RR. SR
.OOL IBS UOH
K TA P o DCI
CHSV RL N,R
IU. NE( ITO
HHAR CS . RH

CC.ELSSC AfiL
flIVP RCCNOHC
HEAUCflLCLC 0
CKSS CGGKSKN

 

NFIF
AAHO
.Vlh..\fK
RNNC
CBUH
BHST

 

 

 

C

N

I

N

C CPE

C NRK

ICC

SS CI

RES 5 H

EXT C EC

HO NtR

IIKNCCIHUD

OSR NN TE

CR 8 IIS II

NBRH .NR
I J RCSOBRFN

0.5 IEH U

«3

AURANIS INC

 

INNS
RUG STORES NH

E
REC
E

SRERVURS FS SR5
SCNA 30

SIS

INNZfl<fi E
TCYCKL

US

 

Ih kKIRhSC
R5?NCVU.IPY
HRBE EH Rfifi
ICCKRLC FKP

INC

P
P R
R C
C C

CC
PO P NE
RC RNII
0 00 4
Ct CIPT
P SR3
L)RCCO

5006806

 

NPCR KR
PG. CLNI

 

RIJT TAP
CT(SH£BN
CL U N E
NNR S
I AT TIT
EG SSRS
NLRS.HRR
«any rxnnr;

GCHU FHF

 

 

A
K I
PB N
R A
CLP L PP
CTRLT)RR C
ACAaooa N
LNCN CCC I
N P. W P
CINS RNLE SR
NFRE C(TB EU
I ER C A RC
LHAN PM? A
SAIH GR EPHN
KFUS DOALRSI
NTOKKLCN 0C3
ANSN H I CNN
BE QU NLV A0
C+8 LCC ICC
C NIIRES S
ILS“ ANREUEI
TANG NUCSRCH
NNEI SIR
AOZN IIHCEEI
LII SSTNVMS
DTTMRRRUNEMR
IflICIIICELCI
HNCD FFSICCF.

 

ERATION

 

185

S
C N

P N I

R I E
0 n!! F
C GCNRI

0 om.tEL
LC DVD"
A TIAS
IT SLC E
CN n$§CFE
nu .LUACY
.AHCOS .0
NTNNA§0L
ISIYC CP
FE E E M

VRR .fFSCC
TNE IN
(418’ ELI?!
E BoF N
“NECILCE
AWILAEM
NC R INN
EIEEANIR
CRNThCBE
LEITTLMV
OHAIECCC
TAPNACCG

 

 

INDEPENCENT LIFE * ACCIUENT
LAMAR LIFE coap
mfi + ACCIDENT

PRovzogpt LIFE

 

 

 

P

U

0

R

G

L

A Y 0

N o C

0 N CP P
I? O Ro R
TRP +0 0
AOR CR C
NCCK) E

R CC.HK L
EL ABPN T
STARL LA NPP
FNIEBKCB IRR
RICH NO P CO
0 NRfANQRLCC
C55 RALCC
ANNNFRL RAN
ECILOI.E ENC
FIFOC

IR CRLACEATI
LEANRLN RNLR
SMNI‘AAUHAEO
UACLCHC ACCH

 

 

 

 

S
R E H
OT RS C P
IT TV N A
SL N AP R
ERVSI NU G
V NR EC E
N+IOY TR L
I TT NC E
G SLCI T
SYT ERNAS
ETHAV INR T
IL TN+ 0 C
RAN I 25 SI
TEE GTz$HRV

SR5 LTALEET

U AML VTS
SATAN NSI
NCU.OL* To
fICRIA A 3

RI TUN R N
LET UTATEEA
LHCRTUHNTNC

LHPVOZEAE INC—UII

FKN.TSICPLR
ANN.SSL MoE
HAOUNAERONM
sac INSPCOA

 

"I7fijiflfifl'flfll‘ifliflfifl'lmfl'flmflr