7 _ s.
‘11,: _
.1; I. \ A ,9
. s. ‘

. w
.u .2. 7/
.14.? . . Ce.»

2
. ”V

V . .

, .
A... I. . . .

. . . . . . , .i/ a”

‘ ~3.1.;

figs...»
.33.... :11. ..

d beam?

V ”Z“
rln

3....
t. a
I.

u.
u. . 5;th . .
I . ,.r A J m
.431; . .
fin _.. firrfan. yd;
. A .7. .
r
I), KP:
. if... d..-
, . .. e. I .
. . . £3.23
mmm. V 3.. -

I“: -
£3.13:

d"

r:-
‘v'li’vpl‘
y

11:4,! x

 

 

 

 

 

 

‘
(0»;-
ur

c . .55

a “.9,

 

. I
I. .5.
x (

ii

p.34. 3,
5e ; i
..

9...». )1

I» 5:.v1ti}
:4.

.4: if. .. I
nit:x....\
7.5:

3:3...

—

Ilia. .\|
:7
5.1 .110

jawxwmkwmmcw

§t

¢

.-

3/4

. .10.

{W-

I
t... 1444
“a

..
3’“?
. .
31-‘

.u

I...

:...:..

fl

.

.

.

a...
L.
:7... .

xi. ....

. f
341:3.)
l...

I
a
In}; a
. f
. k. «3-,.

[LI

; I»,
£34m».
. A . a
.174 an: Fm
. . 4 . . ,
r}. .r. .. 1.
i. I 4;.
3 31;...
.............4.;
32’: A. .v1.
.1 I .

r.)
4!.

rl.

, .9,
.3. \3
:43: .

9.8:.
.1"-

... .ixl?.\1
0).

Fur},
L n
I: if! I

.1.1;Ill;1
(OIL...-

.¢
.£.\ 01

c.
....

I

KM ..
$wa
. . 38.95%.“
#3“ 1.0.5.44 1 .3; O .
:3. A?” u a:

I
awn“.
’ v 4
(99.15:?
. t .

a.
a

.M .
. 3.. “my
bonnfw“... I,

no”
a g

. a...) t 5
» immifi . .2:
alvu -‘

.urfiuwmm

. I ,
. K9.) :55
1‘
I.
:

 

.Wmmn .
and? .

III.

P.

l 6......

i f
..a.‘mrrg.runflntrs.
. ‘ .13.
«4:311.

.r .- r .l 1

é»: (-

. I.I.¢5.V..
tabs}!
#11.

lllmlll'lll Hi ll l'lllllzlllllflllfllllll

312930176

 

 

 

 

 

This is to certify that the

thesis entitled

AN ANALYSIS OF AIRLINE MERGERS: 1951 TO 1972

presented by

David L. Bourke

has been accepted towards fulfillment
of the requirements for

Ph.D. dpgrpp in Finance — Business

A." 5' '

Major professor

Date 5/1/74“

0-7639

 

LIBRARY

8789 , Michiga State

University

 

 

A
P810?" 2098

l

 

 

ABSTRACT
AN ANALYSIS OF AIRLINE MERGERS: 1951 TO 1972
BY

David Laird Bourke

The purpose of the research was to investigate merger activity in
the air transport industry, a regulated sector of the economy. The
research examined mergers from 1951 to 1972 on an ex ante basis.

The research consisted of two principal parts corresponding to two
approaches used to obtain information about airline mergers. One part
examined reasons advanced as to why airlines propose mergers and the
reasons cited by or attributed to the Civil Aeronautics Board (CAB) for
its approval or disapproval of merger applications. The conclusions
for this part of the research were based on an interpretation of an
extensive review of the available literature on airline mergers and on
CAB decisions on mergers. The other main part of the research used a
statistical technique, multiple discriminant analysis, to analyze
airline merger activity. After data were selected to represent
relevant airline characteristics, the technique was used to test
whether a statistical difference between various pairs of groups of
airlines representing their merger activities was accomplished by the
data input. When a difference was found between a pair of groups, the
technique was able to predict the classification of the airlines into
these groups. Significant characteristics of the airlines were

identified for the groups predicted.

 

 

 

 

David Laird Bourke

The significant reasons why airlines merge found in the
literature were (1) to achieve economies of scale, (2) to take advan—
tage of a bargain price, (3) to improve management. (4) to reduce
business and financial risk, (5) to affect tax savings, (6) to grow
rapidly, (7) to improve service and increase demand, (8) to increase
monopoly power and (9) to correct or alleviate financial difficulties.

The board has expected airlines to benefit from mergers because
of (l) economies of scale, (2) the replacement of inefficient manage-
ment, (3) a reduction in business and financial risk, (4) tax savings
and (5) the improvement of service and increases in demand. For the
mergers examined, the CAB required for approval that the price be
reasonable, the protective labor provisions be satisfactory, and the
merger not create a monopoly, jeopardize another carrier, nor cause
significant diversion of traffic.

Multiple discriminant analysis indicated that (l) The airlines
proposing to be the acquiring airlines exhibited better financial,
operating and route characteristics than those airlines not proposing
mergers and those airlines proposing to be acquired. (2) Furthermore,
those airlines not proposing mergers revealed more favorable financial,
operating and route characteristics than those airlines proposing to
be acquired. (3) Both the airlines proposing to be acquired which
were disapproved for merger by the CAB and those airlines proposing to
be the acquiring airlines which were approved by the board showed more
favorable financial, operating and route characteristics than those
airlines proposing to be acquired which were approved. (4) No
differences in the financial, operating and route characteristics of

airlines were found between those airlines proposing to be the

 

 

 

 

David Laird Bourke

acquiring airlines which were disapproved by the CAB and either the
airlines proposing to be the acquiring airlines which were approved by
the board or the airlines proposing to be acquired which were

disapproved by the CAB.

 

 

 

AN ANALYSIS OF AIRLINE MERGERS: 1951 TO 1972
BY

David Laird Bourke

A DISSERTATION

Submitted to
Michigan State University
in partial fulfillment of the requirements

for the degree of

DOCTOR OF PHILOSOPHY

Department of Accounting and Financial Administration

1974

 

 

@Copyright by
DAVID LAIRD BOURKE

1974

Chapter
I

II

VI

TABLE OF CONTENTS

INTRODUCTION . . . . . . . . . . . . . . . . . . . . . .
GENERAL REASONS FOR MERGERS . . . . . . . . . . . . . .
Introduction . . . . . . . . . . . . . . . . . . .
Reasons for Mergers . . . . . . . . . . . .
Quantitative Considerations in the Pricing of
Mergers . . . . . . . . . . . . . . . . . . . . . .
Conclusion . . . . . . . . . . . . . . . . . . . . .

REASONS OF THE AIRLINE INDUSTRY FOR MERGERS AND REASONS
OF THE CAB TO APPROVE OR TO DISAPPROVE MERGERS . . . . .

Introduction . . . . . . . . . . . . .

Airline Reasons for Mergers and Reasons of the CAB
to Approve or to Disapprove Mergers . . . . . . . .

Conclusion . . . . . . . . . . . . . . . . . . . . .

Appendix A: References to Material Which Support
Specific Airline Reasons for Mergers and Reasons
for CAB Approval or Disapproval . . . . . . . . . .

VARIABLE SELECTION AND GROUP PAIRINGS OF AIRLINES USED

IN THE MULTIPLE DISCRIMINANT ANALYSES . . . . . . . . .
Introduction . . . . . . . . . . . . . . . . . . . .
Variable Selection . . . . . . . . . . . . . . . . .
Group Pairings . . . . . . . . . . . . . . . . . . .

MULTIPLE DISCRIMINANT ANALYSIS . . . . . . . . . . . . .
Introduction . . . . . . . . . . . . . . .

The Nature and Use of Multiple Discriminant Analysis.
Multiple Discriminant Analysis Employed in this

Research . . . . . . . . . . . . . . . . .
Results of Nine Group Pairings . . . . . . . . . . .
Summary . . . . . . . . . . . . . . . . . . . . . . .

SUMMARY AND SIGNIFICANCE OF THE RESEARCH .

Appendix I: LEGAL STANDARDS FOR MERGERS .

SELECTED BIBLIOGRAPHY . . . . . . . . . . . . . . . . .

ii

Page

12
l4
l8
18

18
27

4O
4O
40
47
49

49
49

53
66
77
83
89

98

 

 

CHAPTER I

INTRODUCTION

A substantial amount of research on mergers has been done in
recent years but disagreement still exists about both the character-
istics of merging firms and the reasons for mergers. While studies of

the reasons for mergers have dealt with specific industries, the air

 

transport industry has not been examined. The absence of merger
research for the air transport industry is not consistent with (l) the
degree of merger activity and merger discussion in the airline industry;
(2) the significance of this industry; (3) the amount of attention
focused on this industry by the public, government, and the business
community; (4) the efforts of many airlines to merge in the recent
past; and (5) the actual and potential impact of mergers on the struc-
ture of the industry.

The purpose of this research was twofold. One purpose was to try
to identify the reasons why firms merge in the air transport industry
based on available information and to examine the decisions of the
Civil Aeronautics Board (CAB) in approving or disapproving merger
proposals. The second was to use a quantitative technique, multiple
discriminant analysis, to analyze airline merger activity. Discrimi-
nant analysis was used in an attempt to obtain information about
mergers by dividing the airlines into several groups representing
merger activity in order to identify the characteristics of the firms

in these groups. The research evaluated mergers on an ex ante basis.

CHAPTER II

GENERAL REASONS FOR MERGERS

Introduction

This chapter presents the results of a review of the general
literature, made to identify reasons cited for mergers and/or the
factors considered in mergers. Reasons common to many mergers were
selected rather than reasons unique to specific mergers. The findings
from the survey were used to help identify the reasons for mergers in
the air transport industry.

The literature on the reasons for mergers and the factors consid-
ered in mergers is extensive, but not definitive. There is no
universally accepted theory to explain why mergers occur. The
literature is of two types. One type is general in nature; the intent
is to provide broad generalizations rather than to provide detailed
knowledge about specific mergers or mergers in a particular industry.
The findings of this type of approach were suitable for this chapter.
The second type of the literature deals with specific mergers in
detail with little attempt made to provide generalizations. Such
studies resulted in a diversity of conclusions which are less applic—

able to mergers in general.

Reasons for Mergers

Although many reasons for mergers have been advanced in the
literature, it was possible to recognize several common concepts.

2

 

Undoubtedly, each merger has its unique properties, but most also have
properties common to other mergers. The task was to summarize the
reasons for a list useful to this research. Thus, the procedure
followed was to list the common motives acknowledged by the literature.
If wealth maximization is the objective of the firm, the only
reason for merger is to maximize the long-run wealth of stockholders.
The objective of any merger, therefore, should be to increase the
market value of the firm over what it would have been had the merger
not taken place. Hence, a merger would be evaluated as any capital

budgeting project. This idea was supported by Alberts,l Archer and

 

D'Ambrosio,2 Weston and Brigham,3 Schwartz4 and Van Horne.5 other
authors such as McCarthy,6 Butters and Lintner,7 and Mace and
Montgomery8 referred to mergers as an investment decision, but the
objective of maximizing market value was not stated.

Alberts summarized the merger decision as an interaction of the
future dividend stream and its effect on the cost of capital;
therefore, the merger decision ultimately involves share price. He
stated that given the assumption of "intrinsic value" of the exchange
prices in mergers, that is, efficient valuation in merger transactions,
the only possible motive for mergers is synergism due to economies of
scale or to increases in demand brought about by entering a related
industry. Alberts maintained that economies of scale can be achieved
by increasing market share, but that this usually is not feasible,
and that economies of scale can be affected by horizontal mergers,
but this is usually not legal. Therefore, synergistic effects mainly
are due to resulting increases in demand. He also noted that mergers

undertaken in response to increases in demand are less likely to be

 

profitable than internal growth and that mergers to reduce risk are
redundant because investment portfolios presumably already have
accomplished this objective for investors, thus making corporate
diversification unnecessary.

When the assumption of efficient pricing or "intrinsic value" in
mergers is dropped, there arises another motive in addition to
synergisnh-the possibility of achieving gains because of "bargains" or
"bargain prices." Alberts listed four types of bargains: (l) fore—
cast, (2) cost of capital, (3) mismanagement and (4) tax.

Forecast bargains occur because sellers underestimate the future
dividend stream. Cost of capital bargains occur if small firms are
systematically capitalized at a higher rate than large firms. If
larger firms acquire smaller firms, a lower capitalization rate for
the combined entity can achieve higher total value. This cost of
capital difference can be due to the poorer marketability of the
smaller firm's shares in secondary markets and/or its higher cost of
funds in the primary markets.

Mismanagement bargains exist because inept practices can be
changed after a merger. For example, the acquired firm may have excess
liquidity which the acquiring firm could exploit. Tax bargains can
occur because of the institutional and legal advantages of certain
mergers.

Alberts doubted that synergistic effects are common and, therefore,
without the possibility of effecting mergers at bargain prices, there
would be few occasions in which firms would find it worthwhile to grow
in this manner. He continued that since bargain prices are common,

however, mergers are an important means of exploiting investment

 

opportunities. When mergers can be affected at bargain prices, there
are two investment opportunities--synergism and bargains-—that can be
exploited profitably by mergers as well as by internal means. The

synergistic effects are due to economies of scale effected by expanding

 

 

in the same market and to increases in demand effected by expanding in

 

a related market.

fisEconomdes of Scale. This reason was frequently mentioned in the
literature, and it is clearly defined in the economies. Economies of
scale are represented by the declining portion of a long-run average

cost curve, where each point represents the lowest cost for a given

 

output.

Economies of scale are due to many factors--production, admini—
strative, distribution, sales promotion, research consolidation and
financing economies of scale. For example, Weston and Brigham, Van
Horne, Archer and D'Ambrosio, McKenna,9 McCarthy, Mace and Montgomery,
Lintner,lo Alberts, Erler,ll Johnson,12 Galah13 and Weissl4 supported
the theory that economies of scale are a reason for mergers.

Economies of scale can be achieved either by growth through
merger or by internal growth if a firm is of suboptimal scale. Why
would the firm choose to grow by merger rather than choosing to grow
internally? It may be because a firm cannot achieve its optimal scale
due to structural constraints in its industry; for example, a firm may
not be able to increase its market share because of competition. A
firm may prefer growth by merger to gain economies of scale because a
merger can offer bargain prices which internal growth does not.
Economies of scale can be an important reason for growth either by

internal means or by merger.

é§> Monopoly Power. Mergers can be motivated by the desire to gain
monopoly power by increasing the size of the firm. This motive is
often disguised as an attempt to achieve scale economies. Monopoly
power increases the surviving firm's control over supply, and in
unregulated industries, its control over prices. The Federal Trade
Commission's (FTC) Economic Report 93 Corporate Mergers15 concluded
that many mergers are attempts to achieve monopoly power. Alberts,
Weston and Brigham, Lintner, McKenna and McCarthy also recognized this
notive.

The antitrust laws exist to control the lessening of competition
and increases in monopoly or market power. However, in the air trans—
port industry, if the CAB approves a merger this merger is exempt from
antitrust action. Of course the CAB has not ignored antitrust
considerations in its merger decisions. The CAB's responsibility for
maintaining competition in the industry and the court's interpretation
of the CAB's statutory requirements of this matter are discussed
later. The Securities and Exchange Commission (SEC) requires
disclosure of merger information, but it does not judge the merits of
a particular merger.

Economies of scale have been linked erroneously to mergers whose
purpose is to eliminate excess capacity in an industry. A merger
to achieve the "rationalization" of surplus capacity can be
beneficial to the combined firm, but the true source of the economies
is not due to increased size but to the fact that inefficient firms
and excess capacity have been allowed to exist because of monopolistic
conditions or because excess capacity has been imposed by regulatory

policy in a regulated industry affected by nonopolistic elements. In

 

 

effect, economies can occur because of mergers which increase monopoly
power in an industry already affected by monopolistic elements, not
because of true economies of scale which are the result of efficient
operations in a firm and industry.

(:).Increases in Demand. A merger can be motivated by the desire to
increase the total demand of the combined firm. An increase in demand
can occur because of the complementarity of the firm's products or
services, or because one firm's reputation or brand name favorably
carries over to the combined firm. Alberts, Mace and Montgomery, Van

. l6 .
Horne, Friedman and McKenna supported this reason.

 

Cis Acquiring Mana ement Skills. This reason was widely acknowledged

 

in the literature. It was recognized by Weston and Brigham, Van Horne,
Alberts, McCarthy, Butters and Lintner, Mace and Montgomery, Harvey
and Newgarden,l7 and Hutchinson.18 The objective of a merger may be
to replace mismanagement or to acquire superior management skills.
Also included here is the desire to acquire the skills of research or
technical personnel.
This reason is meaningful only if the transaction price of the

merger is low enough to make the merger attractive to the buyer.

§::£orecast Bargains. Buyers and sellers can estimate the future
earning power of a firm differently. Sellers may underestimate the
future earning power of their firm or buyers may overestimate the
future earning power of the selling firm; in either case, a bargain
price will appear to exist from the buyer's point of view. Gort,19
Alberts, Weston and Brigham, and Mace and Montgomery supported the

motive of bargain prices.

Michael Gort tested various hypotheses of reasons for mergers such
as attempts to reduce competition or to achieve economies of scale. He
concluded that the best explanation fer mergers was that economic
disturbances to the firm, such as growth and technological innovation,
caused valuation problems and dispersions among potential owners and
the actual owners which, in turn, increased the probability of merger.
Those industries experiencing more mergers were characterized by more

economic disturbances.

 

(§E>Ri8k Reduction. Alberts stated that "cost of capital bargains"
can occur if small firms are systematically capitalized at a higher
rate than large firms. Similarly, Lintner stated that mergers are
motivated by a divergence in price-earnings ratios because of size
differences. Business risk and financial risk can be reduced by
merger. The reduction of business risk was cited by Weston and
Brigham, Van Horne, McKenna, McCarthy, Mace and Montgomery, Lintner and
Alberts. A reduction in business risk stabilizes earnings and thereby
lowering borrowing costs and increasing debt capacity. Lewellen
argued that mergers can increase debt capacity even if the combining
firms have optimal capital structures.20 Also, one firm may have

unused debt capacity that the other firm in the merger would like to

utilize.

 

{ESTax Advantages. A merger may take place in order to apply an
existing tax loss which could not otherwise be utilized as an offset
to profits. Some mergers are tax-free if they are deemed to be a
reorganization by the IRS. Many mergers are called nontaxable, but
actually they are tax-postponed. Taxes are deferred on any gains in a

merger transaction involving stock payment until the securities are

sold, whereas in a cash transaction any gain is taxable immediately.
If the gains are deferred long enough, they will be taxed as capital
gains rather than ordinary income.

Alberts stated that mergers occur because of tax bargains.
Bosland found taxes to be important in motivating smaller firms to
merge.21 Butters and Lintner found taxes to be an important reason
for mergers for the acquired firm. The Federal Trade Commission's
Economic Report g§_Corporate Mergers, Harvey and Newgarden, and Van
Horne noted that taxes are a reason for mergers. However, the price

of the merger must be low enough to make the tax advantages worthwhile.

 

Accounting Advantages. A merger may have accounting advantages
over ordinary investments or purchases. The Federal Trade Commission,
in its Economic Report 23 Corporate Mergers, was concerned that many
mergers have taken place because of the significance of accounting
considerations.

The accounting profession has had extensive debate on how to
record mergers. The major issue concerns purchase versus pooling—of—
interests merger accounting. If the merger is transacted by cash
payment, there is no problem since this is a clear purchase. The
debate on which accounting method to use arises when the merger is
transacted by stock payment. The alternative accounting methods are
significant, primarily in their effect on the determination of profits
and EPS of the combined firm. In a "purchase" the excess of the
purchase price over the book value of the assets is allocated to
goodwill; goodwill is not tax deductible and it is amortized against
revenue over future years, usually ten. This amortization depresses

earnings. On the other hand, if a merger is accounted for by the

lO

pooling-of-interests method, the assets are transferred or pooled at
book value regardless of whether or not the purchase price exceeds
book value. Hence, no goodwill arises to depress earnings of the
combined entity. However, purchase accounting is preferred by the
acquirer in the unusual instances of book value exceeding the purchase
price because purchase accounting will allow higher EPS than pooling—
of-interests accounting.

The advantage of the pooling-of—interests method is clear. If a
firm can acquire assets through mergers at prices comparable to those
demanded by the market for the purchase of the assets, then it pays the
firm to account for the assets at less than market value; that is, at
book value in pooling—of-interests in order to report higher earnings.
Of course, this logic assumes that it is in the best interest of the
firm to report the highest possible earnings as early as possible. In
addition, the price of the merger must be deemed a bargain, taking
into consideration the accounting advantages and other benefits, for
the merger to be deemed advantageous.

§S§>Financial Difficulties. A firm may be motivated to merge in order
to alleviate financial problems. The U.S. Senate's Hearings, Economic
Concentration supported this theory.22 It is difficult, however, to
establish this since a firm would be unlikely to admit to such a
motive for reasons of prestige, financing and bargaining power consi—
derations. Moreover, there can exist a considerable latitude of
opinion as to what constitutes financial difficulties and its relative
severity.

<EEIRaEid Growth. This reason represents the idea of growth through

merger for the purpose of rapid growth, without considering the possible

 

11

benefits of mergers such as economies of scale or reducing risk. Growth
by merger can take place much quicker than growth by internal means.
Rapid growth merely to increase the size of the firm may be desired for
prestige, personal satisfaction or enhanced power. Promotional profits
also can motivate growth for the sake of growth; for example, many of
the congolomerate mergers of the 19605 appear to have been motivated

by the desire to build an empire.

Reid concluded that merger active firms were oriented toward sales
growth rather than profitability, whereas more internal growth firms
were more profitable but grew less in terms of sales.23 McCarthy,

Mace and Montgomery, Weston and Brigham, Van Horne and McKenna
supported this reason.

‘Method of Financing Expansion. A merger can be financed at times
when it is not possible to finance internal growth. It may be possible
to use stock or convertible securities in an exchange when it might
be impossible, or at least more costly, to raise cash by selling stock
or by borrowing in order‘to purchase the assets. The use of securities
in mergers reduces cash requirements for the acquisition of assets.
Sellers may be more willing to accept the stock of the acquirer than
would investors in a public distribution. By using securities in
mergers a firm may avoid an upward sloping investor supply curve for
capital. In addition, a stock transaction may be cheaper to the buyer
in a merger because of the tax laws discussed previously, and also
because of the underwriting fees and discounts necessary when the
stock is sold. A stock transaction can permit favorable accounting
treatment. Convertible securities have been popular for merger pay-

ments. Convertible securities can be advantageous if common stock

 

12

prices are temporarily depressed because they do not immediately
depress EPS, because they may be used in mergers when they could not
be sold publicly, and because the dividend payouts of the respective
common shares may be too divergent. Of course, the price of the
merger must be low enough to make this financing alternative for
acquiring assets attractive.

Liquidity. A firm may be acquired because it offers liquid assets
that the acquiring firm wishes to utilize. Alberts, Weston and
Brigham, and Van Horne stated that excess liquidity can be important
in evaluating merger candidates; however, the merger must still be

viewed as a bargain.

Quantitative Considerations in the Pricing of Mergers

 

Although there is a considerable amount of literature on the
specific factors or criteria which determine merger valuations, this
literature does not discuss factors considered in evaluating prospec—
tive mergers; that is, the focus is on valuation problems after the
decision that the merger would be desirable if the price is acceptable
has somehow been decided. Therefore, if the valuation indicates that
the price of the merger is a bargain, then the merger is attractive.
Lynn Dellenbarger, for example, focused on factors determining
valuation and exchange ratios (the most important variables were
earnings per share, market price, dividends and book value per share)
but he did not focus on factors determining the attractiveness of
mergers which would precede such valuation considerations.24

Several books composed of articles by a variety of experts who

participated in acquisitions discuss factors used to evaluate potential

 

 

13

mergers and their varying weights from merger to merger.25 The
articles reveal that financial criteria such as existing return on
investment, expectations of growth and liquidity are important in
evaluating acquisitions.

J. Fred Weston argued that it is necessary to consider factors
which influence the price to be paid, since it is the single most
important consideration.26 The important quantitative factors in
determining the transaction price are the per share values of earnings,
dividends, net current assets, market values, book values and growth
rates, although qualitative factors often are the overriding deter—
minant of merger terms. He stated that the relative importance of
each quantitative factor will vary in mergers and that the qualitative
considerations may suggest that synergistic effects may be present to
warrant paying more for the acquired firm than the quantitative factors
alone would suggest. In discussing the relative importance of the
quantitative factors he stated, "Attempts have been made to determine
statistically the relative weights assigned to each of these factors
in actual merger cases. However, these attempts have been singularly
unsuccessful——in one case, one factor seems to dominate; in another
case some other determinant appears to be the most important. This
absence of consistent patterns among quantitative factors suggests
that qualitative forces are also at work."27 Again, the focus here was
on the valuation problem.

According to Van Horne, important quantitative factors affecting
a merger are earnings per share for both corporations, the effect the
merger has on the earnings per share of the surviving corporation,

future earnings for the combined firm, the market prices per share for

 

14

the two firms, the price—earnings ratios, the size of the respective
firms, working capital per share, book value, estimates of future

cash flows and the effect on the business risk of the firms.28

Conclusion

The literature reviewed expressed many common views. Conclusions
as to the causes of mergers and the factors considered in their evalua-
tion were often the same although individual articles displayed some
degree of diversity in their findings and differences in their
conclusions. Although a merger could have a unique reason most have

reasons common to other mergers; it was possible to recognize common

 

concepts throughout the literature.

FOOTNOTES

1 . . . . . .

William W. Alberts, "The Profitability of Growth by Merger," in
The Corporate Merger, edited by William W. Alberts and Joel E. Segall,
pp. 235—87, (Chicago: University of Chicago Press, 1966).

2Stephen H. Archer and Charles A. D'Ambrosio. Business Finance:
Theory and Management (New York: Macmillan Company, 1966), pp. 391—
400.

3J. Fred Weston and Eugene F. Brigham. Managerial Finance (New
York: Holt, Rinehart and Winston, 1969), pp. 688—92.

 

4Samuel Schwartz, "Merger Analysis as a Capital Budgeting Problem,"
in The Corporate Merger, pp. 139-50.

5James C. Van Horne. Financial Management and Policy (Englewood
Cliffs, N. J.: Prentice—Hall, 1968), pp. 471-73.

6George McCarthy, "Premeditated Merger," in Foundations for
Financial Management, edited by James Van Horne, pp. 565-80 (Homewood,
Ill.: Richard Irwin, 1966).

7Keith Butters and John Lintner, "Tax and Non-tax Motivations for
Mergers," in Elements g£ Financial Administration, edited by John
O'Donnell and Milton Goldberg, pp. 629-51 (Columbus, Ohio: Charles
Merrill Books, 1962).

8Myles Mace and George Montgomery. Management Problems pg
Corporate Acguisitions (Boston: Harvard University Press, 1962).

9William E. McKenna, "Merger, An Art and a Science," in Readings
i2 Finance, edited by Harold Wolf and Lee Richardson, pp. 349—50, (New
York: Appleton-Century—Crofts, 1966).

10John Lintner, "Expectations, Merger and Equilibrium in Purely

Competitive Securities Markets," American Economic Review 61, no. 2
(May. l97l):109-l4. —"—“"

llRaymond Erler, "Mergers in the Savings and Loan Industry," in

Dissertation Abstracts, August 1969, p. 441A.

15

16

12 . . .
Craig Johnson, "An Economic and Regulatory Ana1y51s of the Merger
and Consolidation Movement in the Motor Trucking Industry," in
Dissertation Abstracts, 1971, p. 61A.

1
3Path El-Bab Galah, "An Analysis of a Decade of Property and
Liability Insurance Company Mergers, 1950—1959," in Dissertation
Abstracts 26, 04, p. 1960, 1964.

14Using the survivor technique, Weiss estimated that from 83-91
percent of the capacity acquired through mergers in the steel, auto,
petroleum, cement, flour and brewing industries between 1929—1958 was
of suboptimal scale. Leonard Weiss, "An Evaluation of Mergers in Six
Industries," Review of Economics and Statistics, May 1965, pp. 172—81.

15Federal Trade Commission. Economic Report pp Corporate Mergers.
Hearings on Antitrust and Monopoly, Committee of the Judiciary, U.S.
Senate, 9lst Congress, lst sess., part 8A, USGPO, 1969.

16 . . .
Milton N. Friedman, “The Research and Development Factor in

Mergers and Acquisitions," Study No. 16, U.S. Senate, Committee on the
Judiciary, Subcommittee on Patents, Trademarks, and Copyrights,
Washington, D.C., 1958.

17John Harvey and Albert Newgarden, Management Guides E9 Mergers

and Acguisitions (New York: John Wiley and Sons, 1969).

18

(New York: Presidents Publishing House, 1968).

19Michael Gort, "An Economic Disturbance Theory of Mergers,"

Quarterly Journal g£_Economics, November 1969, pp. 624—42.

20Wilbur Lewellen, "A Pure Financial Rationale for Conglomerate

Mergers," Journal pf Finance 26 (May 1971):521-37.

21C. C. Bosland, "Has Estate Taxation Induced Recent Mergers?"

National Tax Journal, June 1963, pp. 159-68.

22Hearings, Economic Concentration. Washington, D.C.: U.S.
Senate, Committee on the Judiciary, Subcommittee on Antitrust and
Monopoly, 1964.

23Samuel R. Reid. Mergers, Managers and the Economy (New York:

McGraw—Hill, 1968), pp. 153—264.

 

17

24 . . .
Lynn Dellenbarger, Common Stock Valuation gp_Industr1al Mergers

(Gainesville: University of Florida Press, 1966).

25John Harvey and Albert Newgarden, Management Guides Eg_Mergers

and Acquisitions (New York: John Wiley and Sons, 1969) and G. S.
Hutchinson, ed., The Business pg Acquisitions and Mergers (New York:

Presidents Publishing House, 1968).

26J. Fred Weston and Eugene Brigham, Managerial Finance (New York:
Holt, Rinehard and Winston, 1969), p. 692.

27Ibid., p. 698.

28James Van Horne. Financial Management and Policy (Englewood
Cliffs, N.J.: Prentice Hall, 1967), pp. 473-86.

 

APPENDIX A

CHAPTER IV

VARIABLE SELECTION AND GROUP PAIRINGS OF AIRLINES

USED IN THE MULTIPLE DISCRIMINANT ANALYSES

Introduction

 

This chapter discusses the variables that were used in the
multiple discriminant analyses of merger activity. The twenty-six
variables discussed in this chapter were all used in the multiple
discriminant analyses. This chapter also discusses the group pairings
of airlines that were used in the discriminant analyses.

The qualitative portion of the research identified the reasons
which the airlines have had to merge and the reasons which the CAB has
had to approve or disapprove mergers. There were two objectives of the
qualitative research. The first objective was to examine the reasons
for mergers, and the second objective was to provide a partial guide
for the subjective selection of a set of variables for use in the

quantitative analysis.

Variable Selection

 

The task was to attempt to select variables which could
discriminate between the groups of airlines used in the multiple
discriminant analyses. The selection process was subjective using the

information developed in the earlier chapters as a guide.

40

 

41

The variables were selected on the basis of their ability to
represent important airline characteristics. Variables were chosen to
represent the airlines' profitability, leverage, liquidity, activity
and efficiency, size and capacity, route characteristics, investment
and costs. Variables which have been used in financial research and
which have been supported in the financial literature were examined for
their possible usefulness to this research. A limiting factor was the
availability of data for all the merger cases and the consistency of
data over time.

Preliminary discriminant analyses revealed six variables possessing

 

discriminating ability; they were retained in the second set of
variables for use in the subsequent discriminant analyses. The
discriminant analyses were repeated to further test the hypothesis that
merger activity could be partially represented quantitatively.

The specific variables chosen to represent characteristics of the
airlines are described below.

Profitability. Measures of profitability generalize a firm's
overall effectiveness and the net result of many factors affecting the
firm.

1. Net income/total assets. The rate of return on investment is the
basic regulatory standard used to judge the adequacy of the industry's
and individual carriers' profitability and to determine allowable rate
levels for the industry. It is a summary measure of how effectively a
firm employs its total assets.

2. Earnings before interest and taxes/total assets. In contrast to the

rate of return on investment, this ratio, which is the net operating

42

rate of return on investment, provides a means of comparison between
firms independent of the way the firm is financed.

3. Net income/stockholder's equity. The rate of return on equity is
used as a regulatory standard in conjunction with the rate of return on
investment to determine the adequacy of the airlines' profitability.

It is the basic measure of the return on the owner's investment in a
firm.

4. Earnings per share. This figure is very important to the common
stockholders as a source of dividends and as a determinant of market

price.

 

Leverage. Leverage or debt ratios and variables measure the extent
to which the firm has been financed by debt. They measure the contri—
bution of the owners as compared to the financing provided by
creditors. Creditors look to the equity funds to provide a margin of
safety. Firms with low leverage ratios have less risk, and high debt
ratios increase the difficulty of borrowing. Industry norms are
important in the capital market's appraisal of a company's financial
condition or strength. If a company is materially out of line with the
rest of the industry, it may be penalized in various ways by creditors,
therefore, it is usually important for a company to avoid being
atypical in the investment markets as far as capital structure is
concerned.

5. Total debt/equity. The debt to equity ratio shows creditor funds
in relation to equity funds. From the point of view of the creditor,
the lower the percentage of creditor funds employed, the better.

6. Long-term debt/total capitalization. Total capitalization is long—

term debt plus net worth. This ratio reflects the relative importance

43

of long-term debt in the capital structure. The debt of a firm is most
commonly expressed by this ratio although it is not very adequate in
providing evidence on the real risk of debt--the inability of the firm
to make mandatory cash payments or cash payments important to the long-
run health of the firm.
7. Total debt/total assets. This ratio, generally called the debt
ratio, measures the percentage of funds that have been provided by
creditors. Total debt includes current liabilities and all long—term
debt. The percentage of creditor funds being used to finance the firm
is a pertinent measure of the financial risk the firm has assumed.
Liguidity. Liquidity ratios are used to judge a firm's ability to
meet short—term obligations. The ratios usually deal with cash, near
cash items, assets that will covert to cash in the normal course of
operations and obligations maturing in one year, including the maturing
principal portion of long—term debt.
8. Liquid assets/current liabilities. Probably the most frequently
used liquidity measure is the current ratio or current assets/current
liabilities. The current ratio is the generally accepted measure of
short-term solvency; the higher the ratio, supposedly, the greater the
ability of the firm to pay its short—term obligations. However, all of
the current assets do not possess the same degree of liquidity. There-
fore, the ratio to be used here, liquid assets/current liabilities,
omits less marketable or liquid assets from the current assets such as
material and supplies. This is a type of quick ratio and it is a more
stringent test of liquidity than the current ratio. While the

denominator remains the same, the numerator focuses on cash and near

 

44

cash coverage of the current liabilities. A low ratio indicates a
weak liquidity position.

9. EBIT + depreciation charges/interest charges. The usual interest
coverage ratio used in financial ratio analysis is EBIT/interest
charges which sometimes is classified as a leverage ratio; supposedly,
it measures the extent to which earnings can decline without the
resultant inability to meet annual interest charges. However, since
depreciation is a significant noncash charge, it is more accurate to
add depreciation to EBIT in order to reflect the coverage of interest
charges.

10. Cash inflows/fixed charges. The ability of a firm to service long—
term obligations is more complex than reflected by an interest coverage
ratio. It has been argued that the ability of a firm to service long-
term obligations and its liquidity is related to its cash inflow
coverage of its fixed charges and that the financial risk associated
with leverage should be analyzed on the basis of the firm's ability to
service fixed charges. This analysis is perhaps the best way to
analyze financial risk, but there is some doubt that the external
market for funds analyzes a company in this manner.

The ultimate hazard of running out of cash is the situation where
legal contracts are defaulted, bankruptcy occurs and the firm ceases
operations. Cash inflows were defined as revenues minus an increase
in accounts receivables or plus a decrease in accounts receivables plus
other nonoperating inflows of cash. Fixed charges were defined as
the sum of interest, lease payments and sinking fund requirements.

Activity and Efficiency. Generally, these ratios measure how

 

efficiently and effectively a firm is using its resources.

 

45

11. Overall revenue load factor. The percentage of total revenue
ton-miles (passenger plus nonpassenger) to available ton—miles in
revenue services. This ratio represents the proportion of the overall
capacity that is actually sold and utilized.
12. Revenue passenger load factor. The percentage of revenue
passenger—miles to available seat—miles in revenue passenger service.
This ratio represents the proportion of aircraft seating capacity that
is actually sold and utilized.
l3. Sales/total assets. The turnover ratio reflects the relative
efficiency with which the firm utilizes its resources in order to
generate revenues. Clearly, the smaller the amount of assets
necessary to generate a given level of sales, the better.
14. Utilization of aircraft per day. This is an average annual figure
in hours and minutes. Presumably, the more intensively the aircraft
are utilized, the lower the costs of an airline.

Size and Capacity.
15. Available ton-miles. The aggregate of the products of the air—
craft miles flown on each interairport hop multiplied by the available
aircraft capacity (tons) for that hop. This represents the traffic
carrying capacity offered.
16. Available seat—miles. The aggregate of the products of the air-
craft miles flown on each interairport hop multiplied by the number of
seats available on that hop. This represents the total passenger
carrying capacity offered.

Route Characteristics. They significantly affect operating costs.

 

Overall revenue load factor--see ratio under Activity and Efficiency.

Revenue passenger load factor—-see ratio under Activity and Efficiency.

 

46

17. Overall flight stage length. The average distance covered per
aircraft hop in revenue service, from take-off to landing, including
both passenger/cargo and all cargo aircraft. This figure is derived by
dividing the overall aircraft miles flown in revenue service by the
number of overall aircraft revenue departures performed. Supposedly,
the average length of flight is extremely critical in determining costs.
18. On-line passenger trip length. Average length of a passenger
trip, calculated by dividing the number of revenue passenger—miles in
scheduled service by the number of revenue passenger originations in
scheduled service. It gives the one-way trip length in terms of one
carrier only. Supposedly, if the on—line passenger trip length can be
increased, operating costs will be lower.
19. Number of stations/seat-miles. This is a measure of route density
and it should significantly affect the economies of an airline.
Investment.
20. Flight equipment/total assets. Many mergers are caused because of
a deteriorating financial condition due to a high level of investment
in equipment.
21. Net change in operating property and equipment/cash inflows. The
preceding measure was static while this measure shows the percentage of
cash inflows being used to purchase equipment. If the percentage is
very high, a firm may be headed for, or in, financial trouble since it
may not have enough cash inflows to make payments on other important
items.
geis-
22. Total costs/available ton—miles. This ratio shows the costs of the

airline to its size; hence, it is a measure of economies of scale.

 

47

23. Operating costs/available seat—miles. This is a measure of
operating costs or economies of the airlines.
24. Operating costs/stage length. This measures the effect of stage
length on the costs of operation.

03%-
25. Market price of stock/earnings per share. The literature has
indicated that the P/E ratio affects merger decisions; also, this is a
measure of the value of each dollar of earnings as decided by the
market.
26. Subsidy. The amount of public subsidy to an airline has been very
important to the CAB's rulings on merger cases. The higher the subsidy,

the more inclined the CAB has been to approve a merger.

Group Pairings

There were nine groups of airlines paired for use in the discrimi-
nant analyses. The purpose of the quantitative analysis was to obtain
information about mergers by dividing the airlines into several groups
representing merger activity in an attempt to find a quantitative
difference between the groups. However, some of the group pairings
employed in the discriminant analyses were included with the assumption
that there would be no difference found between these groups. This was
done for two reasons——first, to establish that there was no quantifiable
difference that could be found and, second, to use this information to
better understand other results of the analysis. Discriminant analysis

was used to see if there was a difference between the following groups:

 

48

l. The airlines in a merger application which proposed to be the
acquiring airlines versus the airlines in a merger application which
proposed to be the acquired or the absorbed airlines.

2. Airlines which did not propose mergers versus the airlines which
proposed to be the acquiring airlines in merger applications.

3. Airlines which did not propose mergers versus the airlines which
proposed to be the acquired airlines in merger applications.

4. Airlines which did not propose mergers versus airlines which
proposed mergers.

5. The airlines which proposed to be the acquiring airlines and were

 

approved by the CAB versus the airlines which proposed to be the
acquiring airlines and were disapproved by the CAB.

6. The airlines which proposed to be acquired and were approved by
the CAB versus the airlines which proposed to be acquired and were
disapproved by the CAB.

7. The airlines which proposed to be the acquiring airlines and were
disapproved by the board versus the airlines which proposed to be the
acquired airlines and were disapproved by the board.

8. The airlines which proposed to be the acquiring airlines and were
approved by the CAB versus the airlines which proposed to be the
acquired airlines and were approved by the CAB.

9. The mergers approved by the CAB versus the mergers disapproved by

the CAB.

CHAPTER V

MULTIPLE DISCRIMINANT ANALYSIS

Introduction

The variables selected in the preceding chapter were used in
multiple discriminant analysis (MDA) of nine pairs of groups of
airlines representing merger activities. The purpose was to quantify
differences between groups.

In this chapter a background discussion of the nature and use of
MDA is given and references are supplied for the reader who wishes to
pursue the subject in greater depth. Then, the mathematical formation
and computational procedure for using MDA are discussed. A test for
the assumption of equality of group centroids is given and the
classificational procedure employed is explained. Next, MDA is
discussed in relation to multicollinearity and factor analysis.

Some preliminary MDA results are discussed in brief, followed by
a discussion of how these results were improved. A subsequent set of
multiple discriminant analyses are then discussed in detail. How the
data were gathered is explained, and the results of the nine pairs of

groups of discriminant analyses are given.

The Nature and Use of Multiple Discriminant Analysis

 

Multiple discriminant analysis (MDA) is a multivariate statistical

technique used to classify an observation into one of several a priori

49

 

 

50

groupings dependent upon the observation's individual characteristics
or set of independent variables. The technique is used to classify and
make predictions in situations where the dependent variable is non—
metric or qualitative. MDA is an alternative to multiple regression
when the dependent variable is qualitative.

MDA classifies a set of observations into one of two or more
mutually exclusive groups. This classification is done by maximizing
the ratio of among-groups variance-covariances to within groups
variance-covariances of the independent set of variables for each

observation. That is, the technique gives the greatest separation

 

among groups with a given set of independent variables. In addition,
the discriminant analysis discloses which variables account for the
largest portion of intergroup differences.

MDA was first used in 1935 by M. M. Bernard to classify Egyptian
skulls.l In 1936 R. A. Fisher used the technique to classify
biological specimens.2 In the earlier years of its use it was applied
in the biological and behavior sciences for classification problems.

Discriminant analysis also has been used successfully in the
business areas of research. The technique has been used for consumer
credit evaluation by D. D. Durand,3 installment loan evaluation by
H. Myers and E. W. Forgy,4 in the marketing "revolution" as discussed
by J. Sheth,5 and by J. E. Walter to classify firms using price/
earnings ratios.6 Durand used the method for differentiating between
good and bad loans in consumer installment lending, and in his
discussion, gave a basic discussion of the theory of MDA. Durand

noted that the method weighs several credit factors by their relative

51

importance, allows for interrelationships between factors (inter-
correlations or multicollinearity) and allows for maximum group
differences.

Edward Altman investigated the predictive ability for bankruptcy
of a set of economic and financial ratios employing MDA.7 An initial
sample of sixty—six manufacturing firms were used to develop a
discriminant model to differentiate between companies in two groups—-
bankrupt and nonbankrupt. The results obtained from the initial
sample, and the reliability of the discriminant model as a predictive
and classifying technique, were impressive. Altman pointed out that
the methodology of studies assessing bankruptcy potential were
essentially univariate analyses of ratios, and therefore, questionable.
On the other hand, MDA selected those ratios which were most important
in detecting bankruptcy potential and determined their relative
importance in a multivariate analysis. Twenty-two ratios were chosen
for the MDA on the basis of their popularity in the literature and
their potential relevancy. The best discriminant function contained
five ratios and the greater a firm's bankruptcy potential, the lower
its discriminant score. In essence, according to a firm's profile, it
was assigned to the group it most closely resembled, based on the
relative proximity of the firm's score to the various group means.
Altman pointed out that the method was explanatory and predictive.

In a later article, he proposed the use of MDA for predicting
bankruptcies in the railroad industry by using again financial ratios.8
He explained that the first study suffered from the potentially serious
drawback of the heterogeneous nature of the sample, whereas in the

railroad industry, the firms were a homogeneous group. Because of this

52

homogeneity, it was easier to be more confident that a new firm to be
classified would have similar characteristics that were present in the
group of firms used to construct the model. Also, he noted that due to
ICC reporting requirements, available data were comprehensive, and
generally speaking, comparable and uniform. He stated that the primary
advantage of MDA in dealing with classification problems is its ability
to analyze the entire variable profile of the observations simultan—
eously with their interrelationships. A univariate study, on the

other hand, only can consider the measurements used for group assign-

ments one at a time.

 

Adelman and Morris used discriminant analysis for selecting under-
developed countries with immediate development potential.9 The results
showed that 97 percent of the discriminable variance between three
groups was represented by a function of four variables selected from
the original twenty-nine.

A review of several studies using MDA was done by William
Cochran.lo He examined the performance of the linear discriminant
function in twelve examples in the literature. He pointed out that a
common use of the linear discriminant function is to classify specimens
into one of two populations, with the objective being to minimize the
probability of misclassification. A comprehensive discussion of the
classification problem under various degrees of knowledge about the
populations can be found in a work by T. W. Anderson.11
M. M. Tatsouaka, D. G. Morrison and J. G. Bryan gave mathematical
derivations of the technique for the general case of n groups.12

Frank, Massy and Morrison discussed the nature and causes of

sample bias leading to an overstatement of predictive power of

53

discriminant analysis. Two validation procedures were presented since

statistical theory provides no means of adjusting for bias.13
Discriminant analysis assumes a multivariate normal distribution

of all populations; it also assumes the groups differ in the average

value of the measurements or mean vectors.

Multiple Discriminant Analysis Employed in This Research

 

If the MDA was successful, it would provide two main results in
this research. One, it would provide for the separation of groups,

and therefore it would be able to classify firms with an accuracy

 

greater than chance; moreover, since preclassification data were used,
the model would possess predictive ability in its classifications.
Second, it would identify those variables that aid most in classifica—
tion or possess the most discriminating power in a multivariate
context.

The Discriminant Model. As only two groups were used at a time

 

in this research, the discussion of the mathematics of the discriminant
model will be confined to the two group case. For two groups, the
discriminant function is in the form:
= . . . +
z lel + v2x2 + ann (l)

where v v . . . , vn are the discriminant coefficients and x x

1' 2' l, 2'
. . . , xn are the independent variables. The number of discriminant
functions is equal to the smaller of g—l or n, where g is the number
of groups and, n, as above, is the number of independent variables.
For two groups, there is only one discriminant function. The

discriminant function transforms individual variable values to a single

discriminant score, or Z value, which is then used to classify the

54

object. The MDA computes the discriminant coefficients, vj, while the
independent variables, xj, are the actual values and j = l, 2, . . . n.
The discriminant analysis used in this research determined, by an
analysis of variance, the linear combinations of variables on the
airlines which best discriminated between the groups. The discriminant
functions selected successively those variables which accounted for
most of the variance between group means, given the other variables
already included. Variables were added successively to the
discriminant functions until all the variance between group means was

exhausted, or until some predetermined point was reached where only a

 

certain percentage of the variance was left unaccounted for.

Discriminant analysis, then, proceeds by first choosing one of
the variables which has the greatest variance between its mean value
for each group relative to its variance within the groups. It is in
this sense the variable "closest" to the one on which the original
grouping was based. The next step selects a second variable that,
when combined with the first in a linear combination, forms an
artificial variable explaining, as before, as much as possible, the
variance between group means. The procedure stops when all variables
have been included or a certain portion of the total variance has been
explained. When formed in this manner, the discriminant function
differentiates as much as possible between the groups.

The problem is to maximize, 1, the ratio of the among-groups sum
of squares to the within-groups sum of squares, so that the among-
groups differences will be large relative to within-group scatter.
The purpose is to seek the best discriminant function of the measure-

ment vector x, where x is a deviation from the grand centroid,

55

xki = xki ' m' (2)
where in is the dependent vector variable for the ith subject in the
kth sample, k = l, 2, . . . 9, where g is the number of populations
under study; m is the grand centroid, or vector of total sample means.
The desired function is

y = v'x. (3)

The among-groups sum of squares on the function will be the
following quadratic from the matrix of among-groups sums of squares and
crossproducts, A:

v‘Av (4)

9
where A = E Nk(mk—m) (mk—m)' (5)
k=1

and Ink is the centroid for sample k; N is the number of subjects in
all groups. Similarly, the within—groups sum of squares on the function
is the following quadratic from the matrix of pooled within—groups sums
of squares and crossproducts of deviation scores,
1

v WV 9 Nk (6)
where W = kil iii(in—mk)(xki-mk)'. (7)
The task for the calculus of partial differentiation with respect to
the unknown weights in v is to maximize the ratio of these two
quadratic forms,

A v'Av

=-—— x
v'Wv MR , (8)

subject to the restriction v'v = l. The maximum value 1 and the
associated vector of weights, v, are shown by the differential calculus
to be the largest eigenvalue and its eigenvector of the equation

(w‘lA — Anv = o, (9)

 

56

where I is the identity matrix. The solution of (9) is lengthy and
will not be shown here, but it yields the vector, v, in (3) or vi, 1 =
1' 2' o o o n, in (l).

Equality of Group Centroids. Discriminant analysis assumes that

 

the groups are centered at different locations in the measurement space
spanned by the vector variable. The equality of the population
centroids, H1, is tested by Wilks's Lambda. Lambda is a ratio test

statistic defined, in the two group case, as

_ 1
A‘1+A (10)

 

ALI-'94. <11)
lTl

or

where T = A + w.

The null hypothesis is uk = u for k = l, 2, . . . g where the best
estimator of the common populations centroid, u, is m, the grand
centroid. The test of H1 may be viewed as a test of the discriminating
power of the measurement battery for the grouping criterion. The
general utility of the test statistic, A , depends on the availability

. . . . 2
of transforms of it that distribute approx1mately as X and F. Rao's

F approximation was used; the F—ratio is

l -/\) ( N-p-l)’ (12)
A p

where p is the number of variables and the degrees of freedom for the
numerator; (N—p-l) represents the degrees of freedom for the
denominator and N is the total number of subjects in all groups. The
Lambda test of the null hypothesis of the equality of mean vectors

assumes that the 9 group dispersion matrices are based on samples of g

57

multivariate normal populations with the same dispersion; this
assumption is analogous to that of the homogeneity of variance in the
univariate F—ratio test of equality of means. As in the univariate
case, this assumption is very rarely tested in a multivariate context
on the basis that the test of the equality of the mean vectors is
fairly robust under departures from its assumptions, and also because
computer programs which test the equality of the dispersions or
variance—covariance matrices are rare and extremely complicated.

Classification Procedure. The observations were classified

 

according to their Z scores. In equation (1) the Z score was calculated

 

for each observation by substituting the values of the independent
variables, Xi' i = l, 2, . . . n, into the equation; the vi, i = l, 2,
. . . n, are the calculated discriminant coefficients which remain

the same for each observation. The classification boundary for the

two group case is simple and is calculated by

l 1 2 2 3 3 n n = Zcritical. (13)

An observation with a score above Z critical was classified in one
group while an observation with a Z score below Z critical was
classified in the second group. The 2i are the grand means of the
independent variables.

MDA, Factor Analysis and Multicollinearity. The discriminant

 

model may be interpreted as a special type of factor analysis that
extracts orthogonal factors of the measurement battery for the specific
task of displaying and capitalizing upon differences among criterion
groups. The model derives the components which best separate the

groups. The basic input for factor analysis is the correlation matrix.

58

It can be shown that discriminant analysis is based on the correlation
matrix. The purpose of factor analysis is to reduce the amount of
data without great loss of information by means of reducing the amount
of multicollinearity or correlation among the variables; this is
accomplished by producing artificial or imaginary groups of variables.
The purpose of discriminant analysis is to account for the maximum
variance between groups in order to discriminate maximally among
groups. Discriminant analysis has better chance of success when it
uses more variable input. It does not matter whether the variables in
the discriminant analysis are highly intercorrelated in the sense of
producing group separation. Of course, more group separation is
possible if the variables are not highly correlated with each other if
we are speaking of n variables highly correlated and an equal n number
of variables not so intercorrelated. If, however, the n variables are
reduced by factor analysis to p variables, p<n, the discriminant
analysis is adversely affected in its task of providing group separa-
tion because there has been some loss of information of the original
data because of the use of factor analysis. It is not possible to
justify this loss of information on the grounds that the multi—
collinearity of the variables can be reduced since the discriminant
analysis is capable of utilizing all of the information whether the
information is highly correlated or not. Therefore, if the purpose
of the discriminant analysis in a research is to provide maximum
group separation in order to classify firms as correctly as possible,
then the loss of information by factor analysis is incorrect.

MDA will select successively those variables which most explain

the remaining variance between the groups. It is in this fashion that

59

MDA considers the multicollinearity of the variables. For example,
the first six of twenty variables could account for 90 percent of the
variance among groups, and therefore, for 90 percent of the discrimina-
ting ability of the function, although it will employ the use of all
twenty variables in order to account for 100 percent of the variance
among groups which was provided by the data input. If one purpose of
the discriminant analysis is to reveal a few number of variables which
are capable of discriminating among groups to the extent that it is
acceptable to reduce the discriminating ability of the entire test
battery, then it is possible to do this 23225 the maximum group
separation has been accomplished by using the entire set of data.

In sum, the maximum separation and classification ability of MDA
is accomplished by the use of all the data input. Subsequently, it
is possible to reduce the number of variables so that a much smaller
set can explain group differences, although some loss of discriminating
power of the model results.

Preliminary Results of the Multiple Discriminant Analyses. MDA

 

was applied to the nine group pairings which were given in the
preceding chapter and which were used for the second multiple
discriminant analyses.

MDA is based on group separation provided by the group means of
the variables. Firms are then classified on the resemblance of their
individual profile of variables to the groups'. It is quite
possible than an individual observation will be misclassified although
the variable averages provide good group discrimination. The real
test of discriminant analysis is its ability to classify observations

correctly. If many firms are fairly atypical when compared to the

 

60

group centroids, then MDA will unfortunately misclassify a high
percentage of firms even though group separation is provided.
Apparently, this phenomenon operated with the first set of MDA because
of the disappointing classificational ability of the discriminant
functions. The classificational power of the discriminant functions,
although significant, were all much greater than chance (50 percent)

and ranged from 64 percent to 91 percent. This classification test

for the various discriminant functions was performed on the observa-
tions which constructed the functions. Therefore, a more extreme test

of the classificational power of the functions, by using a replication

 

sample or a sample of new firms, could be expected to lower the
percentage of correct classifications, but was not used when the
decision to try to improve the discriminating power of the functions
was made.

Six out of thirteen variables were retained from the first set of
MDA and used in the second set of MDA because of their relative
importance in the various functions in providing discriminating power.
The other variables were examined and it was decided not to include
them in the second set of multiple discriminant analyses because of
the difficulty of determining their meaning in providing group
differences. Other new variables were added in an attempt to provide
more discriminating power of the functions.14

The first computer program applied to the first set of MDA could
not be adapted in order to reduce the functions to a lesser number of
variables, thus accounting for a certain percentage of the available
discriminating variance between groups. That is, it was possible to

determine the relative importance of the variables in the discriminant

61

functions as compared to the other variables, but it was not possible
to determine that five or six variables, for example, accounted for
over 90 percent of the variance between groups. Therefore, the
functions could not determine an adequate reduced model which would
account for most of the variance with fewer variables. The identifi—
cation of these fewer variables was expected to clarify and simplify
group differences for a qualitative explanation of the quantitative
differences between groups.

It should be noted that the average values for the ratios and the
variables of two yearly figures were introduced to the results of the
first set of multiple discriminant analyses in an attempt to improve
the results. This approach caused the classificational ability of the
functions to decline, although not markedly.

Finally, a second set of MDA were performed on the nine group
pairings in an attempt to improve the classificational ability of the
functions and facilitate the interpretation of the group differences
as displayed by the functions.

Data Collection. The data available from the CAB regarding
individual firms are standardized, and therefore, comparable. The
data also are very comprehensive, so that the problem was to select
a limited number of variables. Data were available on all firms
requiring operating authority from the CAB regardless of size;
therefore, the data were not biased by firms of large size only (a
common problem in much research).

Since the firms in the air transport industry are relatively
homogeneous, the prospective results of the analyses were improved.

The heterogeneous nature of many firms in samples is a common

 

62

difficulty in business research. Also, only horizonal mergers were
considered in the hope that this would improve the results of the
analyses.

The data collected for each observation were premerger data. This
was done in order to establish the predictive ability of the classifica—
tion powers of the functions. It also corresponds to the premerger
identification of the causes of mergers in the qualitative portion of
the research.

Data were collected on firms from 1951 through 1971. During that

time, there were several significant accounting changes detailed by

 

the CAB. These changes were taken into account and some adjustments
to the data were made. This time period was necessary in order to
obtain samples of adequate size, although some samples were fairly
small. 1951 was chosen as the earliest date for two reasons. It was
felt that twenty-one years was too long a test period, however, this
was necessary in order to obtain the desired samples. But more
important was the fact that comparable data on all variables is
provided from 1950 to 1971.

The data for the various groups of firms were collected in the
following manner. In any given year, t-l, an airline will potentially
belong to one of the groups in the year t. The measurements of the
set of variables to be used for each airline was for the year t—l,
where t is the year of classification. As was reported above, an
average of t—l and t-2 data was used in the first set of MDA in order
to improve group separation, however, this had the opposite effect.
The purpose of this approach was to construct predictive models by

using data prior to an airline's group classification.

63

One severe data problem was that over time a firm would belong to
different groups, thus affecting the discriminating ability of the
functions. The solution to this problem was to provide as much
temporal separation as possible without further significantly reducing
the samples. The decision was to separate all firms in a group by a
minimum of six years. That is, if a firm were in one group in the
year t, it would not appear in another group for at least t + 6 years
and it also was not in any other group for at least t—6 years.

The groups were of different size but every airline had the same

set of quantitative variables. Data are symbolized by Xi. where i

Jk’

 

refers to a group, j refers to the observation or airline and k refers
to a variable. k equals 26 for all i and j. There were ten unique
groups or i and the j varied from group to group. The number of
Iobservations available for each group was divided between an original
sample which constructed the discriminant functions and a replication
sample used to test the ability of the functions to classify new
firms not used in the construction of the functions. The sizes of
each group of firms is given below along with the division of the
available sample between an original (0) and a replication sample (R).

1. Proposed acquirers 48; original (0): 31; replication (R): 17.

2. Proposed acquireds 48; O: 31; R: 17.

3. Nonproposing 76; O: 50; R: 26.

4. Proposed airlines 76; O: 50; R: 26.

5. Proposed acquirers approved 34; O: 22; R: 12.

6. Proposed acquirers disapproved l4; 0: 8; R: 6.

7. Proposed acquireds approved 34; O: 22; R: 12.

8. Proposed acquireds disapproved 14; O: 8; R: 6.

 

 

64

9. Airlines approved for mergers 68; O: 46; R: 22.

10. Airlines disapproved for mergers 28; O: 18; R: 10.

The sample of nonproposing airlines was chosen by a random method
from the available airlines in this group. It was necessary that these
firms satisfy the temporal separation from other groups by plus and
minus six years.

Since extreme values for any variable adversely could have affected
the results, the ratios and variables which consistently displayed
extreme values during the collection process were excluded.

Results Reported For Nine Multiple Discriminant Analyses. There

 

 

were nine discriminant analyses corresponding to the nine group
pairings. For each of the nine MDA, several factors will be reported.
1. Test of group means. The significance of a will be given if the
hypothesis that the group means are equal was rejected. The
hypothesis was rejected if a was equal to or less than .05. A
rejection of the hypothesis indicated that there was a large
probability that a real difference existed between the population
centroids or mean vectors, and that this difference was not attribut—
able to chance. The acceptance of the hypothesis indicated that there
was no real difference between the group centroids and, therefore,
there was no basis to proceed with the analysis since the functions
produced would not have been able to locate group differences. How
the acceptances and rejections relate among one another will be
explained after all the MDA have been discussed.

2. Discriminant functions usingiall variables. Each discriminant

 

analysis was performed using all of the data input. (This procedure

was explained earlier.) Basically, this approach allowed for the most

 

65

discriminating power of the functions to be produced. In other words,
this afforded the model the greatest chance to classify firms
correctly. The relative importance of all variables could have been
reported from the discriminant functions, but this description of the
significance of the variables was deferred for the reduced model

where a much easier interpretation of the variables was possible.

The classificational ability of the functions was tested in two ways.
First, the number of correct and incorrect classifications and their
corresponding percentages are given for the original sample of firms
whose characteristics constructed the model. Second, a more severe
test of the classificational powers of the functions was employed by
using a replication sample. Both tests were based on the number of
"hits and misses." The second test of the discriminating power was
more reliable than calculating the hits and misses of the samples used
to form the predictors.

3. Reduced model. After the most discriminating power of the techni—
que was permitted by using all the variables, it was useful to
summarize or simplify the functions in order to clarify interpretation
of those variables that were important in determining group
differences. Each variable in the discriminant analysis was ranked on
the basis of the decision rule that it was the variable of the
remaining variables which offered the greatest variance between groups,
relative to its variance within groups. This procedure was followed
until a certain portion of the variance between groups was accounted
for which represented a certain portion of the discriminating power of
the model, or until a maximum number of variables had been included,

whichever came first. For this research, variables were included until

 

 

66

90 percent of the discriminating variance was accounted for or until
nine variables had been included. The relative importance of the
variables is given by removing the scale factors from the variables.
The classificational ability of the reduced model was tested using the
same original and replication samples used to test the classificational
ability of the full model. This was done to see how much of the
discriminating power of the full models was lost by the reduced models

which were composed of much fewer variables.

Results of Nine Group Pairings

 

1) Proposed Acquirers — Proposed Acguireds.

 

1. Test of means. a = .01, so Hl was rejected.
2. Full model classifications. Classification matrices are given

for the original sample and the replication sample.

Original 87% Replication 85%

a b T a b T

a 26 5 31 a 13 4 17
b 3 28 31_ b 1 16 ll
62 34

a = proposed acquirers b = proposed acquireds T = total

The original sample interpretation is: of the thirty—one
airlines in group a (those airlines which proposed to be the
acquiring airlines) in year t or the year of classification,
twenty-six were classified correctly in group a and five were
classified incorrectly in group b (those airlines proposing to be
the acquired airlines) by the discriminant model constructed from
t-l data on each airline. Of the thirty-one airlines in group

b in year t, the year of classification, three incorrectly were

 

67

classified in group a and twenty-eight were classified correctly
in group I) by the discriminant model constructed from t—l data
on each airline. That is, data were taken on each of the
thirty-one airlines in groups a and b in year t—l to test the
classificational or predictive ability of the discriminant
function for year t or the year when the actual classification
of each airline was a fact. The variable profiles based on t—l
data of the airlines in group a, relative to the variable pro-
files or the composite quantitative picture for airlines in group
b based on t—l data, produced a discriminant function that
correctly classified twenty—six of the thirty—one airlines into
group a and twenty—eight of the thirty—one airlines into group
b, for the year t or for the year following the data profiles.
For example, American Airlines proposed to acquire Eastern Air
Lines in 1962. Based on 1961 data, the discriminant function
would predict the classification of American and Eastern into
groups a and b respectively for 1962. Thirty other examples were
used for the full model constructed from the original sample of
thirty—one such pairings. The overall classification accuracy
for the full model was 87 percent for the original sample of
thirty-one airline pairings that constructed the discriminant
function.

The interpretation of the results of the replication sample
is very similar. The replication sample consisted of airlines
that were not included in the original sample which constructed
the full model. For example, American and Eastern would not be

in the replication sample if they were in the original sample.

 

68

This was done as a more extreme test of the classification ability
of the full model. 0f the seventeen airlines in group a,
thirteen were classified correctly in group a and four were
classified incorrectly in group b. Of the seventeen airlines in
group b, one was classified incorrectly in group a and sixteen
were classified correctly in group b. The overall classifica-
tional accuracy for the replication sample of seventeen pairings
was 85 percent.

Reduced model. Eight variables accounted for 95 percent of the
discriminating variance between groups. Information on the
function and its variables is given in four columns. The first
column gives the class of variable to which the variable belongs
(see chapter on variable selection); the second column, the
specific variable; the third column, the scaled importance of
each variable; and the fourth column, the rank of importance of

the variable in the function. The numbers immediately preceding

Scaled

Class Variable Importance Rank
1. Profitability l-net income/total assets 34 l
2. Activity and 12—revenue passenger load 32 2

route factor
3. Investment 20—f1ight equipment/total assets 27 3
4. Leverage 5-tota1 debt/equity 25 4
5. Costs 23—operating costs/available 20 5
seat miles

6. Size l6—available seat—miles 14 6
7. Liquidity 10-cash inflows/fixed charges 8 7
8. Route charac— l7-overall flight stage length 3 8

teristics

 

69

the specific variable in the second column refer to the number of
the variable as listed in the chapter on variable selection. The
interpretation of the variables in the function is as follows:

An airline is mostly likely to be classified in group b (proposed
acquireds), if relative to airlines in group a (proposed
acquirers), it has a low net income/total asset percentage, a low
revenue passenger load factor, a high flight equipment/total
assets percentage, a high total debt/equity ratio, high operating
costs per available seat—miles, low cash inflows/fixed charges
ratio, a short overall flight stage length and if it has fewer
available seat-miles.

For example Eastern Air Lines would be classified in group b
by the discriminant model if it had enough of the above mentioned
quantitative characteristics for airlines in group b relative
to the airlines in group a. On the other hand, American Airlines
would be classified in group a if it did not possess a variable
profile similar enough to the variable profile for group b and,
therefore, it had a profile more closely resembling airlines in
group a.

The classification matrices for the reduced model are:

Original 84% Replication 76%

a b T a b T

a 24 7 31 a 12 5 17
b 3 28 21 b 3 14 ll
62 34

a = proposed acquirers b = proposed acquireds

The full model employed all twenty-six variables in order to

establish a variable profile for each airline for its predictive

 

 

 

7O

purpose. The reduced model for this pairing used only eight of
the twenty—six variables. This reduced function was then tested
for its predictive ability from eight variables The original
sample of thirty-one pairings was the same as the full model.

The replication sample consisted of the same seventeen pairings
used to test the full model. The interpretation of the classifi—
cational matrices for the reduced model is the same as for the
full model (see above explanation).

. . . . 15
2) NonpropQSing Airlines - Proposed Acguirers.

1. Test of Means. a = .05 and the null hypothesis that the group

 

centroids were equal was rejected.

2. Full model classifications.

Original 73% Replication 65%

a b T - a b T

a 36 14 50 a 17 9 26
b 8 23 .31 b 6 11 ll
81 43

a = nonproposing airlines b = proposed acquirers

3. Reduced model. Nine variables accounted for 83 percent of the

discriminating variance.

Scaled
Class Variable Importance Rank
1. Route charac- l7—overa11 flight stage length 33 l
teristics
2. Investment 21—net change in operating 30 2

property and equipment/
cash flows

3. Profitability l—net income/total assets 28 3

71
4. Costs 22—total costs/available ton- 27 4
miles
5. Leverage 7-total debt/total assets 22 5
6. Other 25-market price of stock/EPS 21 6
7. Route 19—no. of stations/seat—miles 14 7
8. Profitability 4-earnings per share 6 8
9. Liquidity 8—liquid assets/current 5 9
liabilities

Firms in group a (nonproposing airlines), relative to firms in

group b (proposed acquirers), have short overall flight stage

 

lengths, a greater positive net change in operating property and
equipment as a proportion of their cash flows, a lower net income
to total assets percentage, higher total costs per available ton-
mile, a higher total debt to total asset ratio, lower price-
earnings ratios, more stations per seat-mile, less earnings per
share and less liquid assets as a proportion of current
liabilities.

The classification matrices for the reduced model are:

Original 63% Replication 60%

a b T a b T

a 31 19 50 a 16 10 26
b 11 20 21 b 7 10 ll
81 43

a = nonproposing airlines b = proposed acquirers

72

3) Nonproposing Airlines — Proposed Acquireds.l6

 

1. Test of means. a =

.05 and the null hypothesis was rejected.

2. Full model classifications.
Original 81% Replication 74%
a b T a T
a 42 8 50 a 21 5 26
b 7 24 21 b ‘ 6 11 11
81 43

a = nonpropos

3. Reduced model. Eight variables accounted for 92 percent of the

ing airlines b = proposed acquireds

discriminating variance.

Class

1. Leverage

2. Profitability

3. Investment

4. Route charac—
teristics

5. Other

6. Costs

7. Profitability

(1)

Activity and
route

 

Scaled
Variable Importance Rank
6—1Qng-term debt/total 26 l
capitalization
2—EBIT/total assets 22 2
21-net change in operating prop— 20 3
erty and equipment/cash flows
l7—overall flight stage length 16 4
25—market price of stock/EPS ll 5
22—total costs/available ton— 10 6
miles
4—earnings per share 8 7
ll—overall revenue load factor 5 8

The airlines in group a (nonproposing airlines), relative to

airlines in group b (proposed acquireds), as a group have less

long-term debt to total capitalization, more earnings before

interest and taxes as a percentage of total assets, less of a

positive net change in operating property and equipment as a

proportion of cash flows, longer overall flight stage lengths,

73

higher price-earnings ratios, lower total costs per available
ton-mile, greater earnings per share and a higher overall revenue
load factor.

The classification matrices are for the reduced model:

Original 73% Replication 70%
a b T a b T
a 39 ll 50 a 19 7 26
b 11 20 g; b 6 11 ll ’
81 43

a = nonproposing airlines b = proposed acquireds

4) Nonproposing Airlines — Proposed Airlines.

 

1. Test of means. The null hypothesis was accepted (no difference

 

was found between the groups). The proposed airlines in this
pairing (4) consisted of one—half proposed acquirers and one-half
proposed acquireds. It was found that the nonproposing airlines
in pairing (2) were not as strong a group as the proposed
acquirers, while in pairing (3) above the nonproposing airlines
were stronger than the proposed acquireds. Therefore, this group
pairing (4) was composed of two other group pairings (2 and 3)
whose quantitative differences conflicted when combined to form
the pairing (4).

5) Proposed Acquirers Approved - Proposed Acquirers Disapproved.

 

1. Test of means. The null hypothesis was accepted (no difference
was found between the two groups). However, this finding proved
helpful. The group, mergers approved, consisted of acquirers and
acquired airlines as did the group, mergers disapproved (group
pairing (9) below). Since it was shown that there was no
difference between proposed acquirers approved and proposed

acquirers disapproved (5), it also was likely to be shown there

74

was no difference between mergers approved and mergers disapproved
(9 below) since both these latter groups contained acquirers.
This could be true even if a difference was found between
proposed acquired airlines approved and proposed acquired air~
lines disapproved ((6) below). That is, the group pairing,
mergers approved and mergers disapproved ((9) below), was composed
of the group pairings of (5) and (6). Since no difference was
found in (5), the possibility of a group difference in (9) was
unlikely even though a group difference was found in (6).

6) Prgposed Acquireds Approved - Proposed Acquireds Disapproved.17

1. Test of means. a = .01 and the null hypothesis was rejected.

2. Full model classifications.

 

Original 93% Replication 89%
a b T a b T
a 20 2 22 a 10 2 12
b 0 8 _J§ b 0 6 _§
30 18

a = proposed acquireds approved
b proposed acquireds disapproved

3. Reduced model. Seven variables accounted for 91 percent of the

discriminating variance.

Scaled

Class Variable Importance Rank

1. Leverage 6-long-term debt/total 38 1
capitalization

2. Route charac- l7-overall flight stage length 35 2

teristics
3. Profitability l-net income/total assets 30 3
4. Costs 23-operating costs/available 20 4

seat-miles

75

5. Size lS-available ton-miles ll 5

6. Other 26—subsidy 8 6

7. Activity and ll-overall revenue load factor 6 7
route

The interpretation of the variables in the discriminant function
is: as a group, the airlines in a (the proposed acquireds
approved), relative to the airlines in b (the proposed acquireds
disapproved), displayed a higher long—term debt/total capitaliza-
tion ratio, a shorter overall flight stage length, a lower return
on investment, higher operating costs per available seat-mile,
fewer available ton-miles, more subsidy and lower overall

revenue load factors.

The classificational ability of the reduced model follows.

 

Original 93% Replication 83%
a b T a b T
a 20 2 22 a 10 2 12
b O 8 ._8 b l 5 '_§
30 18
a = proposed acquireds approved
b = proposed acquireds disapproved

7) Proposed Acquirers Disapproved - Proposed Acquireds Disapproved.

1.

Test of means. The null hypothesis was accepted (no group
difference was found). Notice that the classificational ability
of the function for the pair of groups, proposed acquirers
approved-proposed acquireds approved (8 below), was greater

for both the full model and the reduced model than with the
function for the pair of groups, proposed acquirers-proposed
acquireds (1 above). This was because the pair of groups
showing no difference (7) here, proposed acquirers disapproved-

proposed acquireds disapproved (7), were removed from the

 

 

76

proposed acquirers—proposed acquireds group pairing (1) to form
the group pairing (8), proposed acquirers approved-proposed
acquireds approved. That is, pairing (1) was composed of pairings
(7) and (8). No difference was found in pairing (7) even though
a difference was found in pairing (1). Therefore, pairing (8)

was the reason for the difference in pairing (1), although
pairing (7) reduced the discriminating power of pairing (1).

8) Proposed Acquirers Approved - Proposed Acquireds Approved.18

 

1. Test of means. a = .01, thus the null hypothesis was rejected.

2. Full model classifications.

Original 91% Replication 83%

a b T a b T

a 19 3 22 a 10 2 12
b 1 21 22 b 2 10 lg
44 24

a = proposed acquirers approved
b = proposed acquireds approved

3. Reduced model. Six variables accounted for 94 percent of the

discriminating variance.

Scaled
Class Variable Importance Rank
1. Profitability l-net income/total assets 15 l
2. Leverage 5—tota1 debt/equity l3 2
3. Investment 20—flight equipment/total 9 3
assets
4. Other 26—subsidy 5 4
5. Profitability 3-net income/stockholder's 2 5
equity
6. Size lS-available ton—miles l 6

Airlines in the group b (proposed acquireds approved),relative to

the airlines in the group a (proposed acquirers approved),

 

 

77

displayed a lower rate of return on investment, a higher total
debt to equity ratio, more flight equipment as a percentage of
total assets, more subsidy, a lower return on stockholder's
equity and fewer available ton-miles.

The classification matrices for the reduced model are:

 

Original 86% Replication 79%

a b T a b T

a 18 4 22 a 9 3 12
b 2 20 .22 b 2 10 .13
44 24

a = proposed acquirers approved
b proposed acquireds approved

9) Mergers Approved (Airlines Approved for Mergers) - Mergers
Disapproved (Airlines Disapproved for Merggrs).

 

1. Test of means. The null hypothesis was accepted (no difference
was found between the groups). There was no difference found
between proposed acquirers approved and proposed acquirers
disapproved (5) although a difference was found between the
proposed acquired airlines approved and the proposed acquired
airlines disapproved (6). Since the pairing, mergers approved
and mergers disapproved (9), consisted of the total of these two
pairings (5) and (6) no difference was found between the groups,

mergers approved and mergers disapproved.

Summary

The twenty-six variables which were selected in the preceding
chapter were used in this chapter as the basic input to the multiple
discriminant analyses (MDA). MDA used group differences to classify
observations with an accuracy greater than chance and to identify

variables which were significant in detecting group differences. MDA

is "successful" if the hypothesis that there is no difference between
groups is rejected.

The variables which were significant in the reduced functions
varied from function to function as did their rank. This does not mean
the results were unstable. The variables changed from discriminant
function to discriminant function, but this occurred gply because nine
MDA were performed on nine pairings. If only one group pairing had
been employed in the research instead of nine, there would have been
only 923 reduced function. The approach employed first provided for

the most discriminating power of the functions between groups by using

 

the full function containing all variables, and secondly, the most
discriminating variables of the full function were identified by the
reduced functions.

If only one discriminant analysis had been performed on one group,
only one unique reduced function would have been produced. Since
multiple discriminant analyses were performed, other reduced functions
were produced which also were unique for their group. One way to
minimize the changes in variables from reduced function to reduced
function would have been to reduce the original set of variables avail—
able for the full model; this approach was considered but was rejected
because this would have reduced the classificational ability of the
full discriminant functions, and that it was possible to reduce
subsequently the full model for within group interpretation of
variables.

The relative importance of the variables in the functions in a

multivariate context was given; the relative importance of the variables

79

in a univariate context of disclosing group differences was not given,
since no information would have been provided by doing 50.

Various sources of sampling and nonsampling errors were discussed
and the methods used to limit their influences were given. Potential
sources of error must be considered when evaluating the results. Also,
the assumption of the equality of group dispersions (variances -
covariances) was not tested, as is the usual procedure in business
research. This was done mainly since the test of the equality of means
is fairly robust under departures from equality of dispersions; that

is, the assumption that the group centroids were different would not

 

be accepted easily if the dispersions were not equal.

The interrelationships between several of the groups were dis-
cussed and, hence, will not be repeated here.

The results were encouraging. The full functions correctly
classified firms of the original samples for 73, 81, 87, 91 and 93
percent of the observations. For the replication samples, the full
models correctly classified 65, 74, 83, 85 and 89 percent of the firms
in five group pairings. In general, then, the results of the full
models were better with the original samples. The results of the
reduced models generally were not as good as the full model results,
as was expected. For the original samples, the reduced functions
correctly classified 63, 73, 84, 86 and 93 percent of the firms. For
the replication samples, the reduced functions correctly classified
60, 70, 76, 79 and 83 percent of the firms. Again, the results with
the original samples were better than with the replication samples.

A rough ordering of the classificational accuracy of the MDA on

th

(D

various group—pairings is: (1) proposed acquireds approved—proposed

8O

acquireds disapproved (pairing 6); (2) proposed acquirers approved—
proposed acquireds approved (pairing 8); (3) proposed acquirers-
proposed acquireds (pairing 1); (4) nonproposing airlines—proposed
acquireds (pairing 3); and (5) nonproposing airlines—proposed acquirers
(pairing 2). This ranking is not exact because it refers to the
classificational ability of the full and the reduced functions for
both the original and replication samples for each group pairing,
hence, there is some overlap in the classificational accuracy among
groups.

The discriminant functions corresponding to these pairs of groups

 

were predictive in their classificational abilities because of the
data input which was used to construct the functions. The data were
collected for the year prior to any group classification for every
observation; if an airline was known to be in a group in the year t,

t—l data were used for its variable profile.

FOOTNOTES

1 . . .
M. M. Bernard, "The Secular Variations of Skull Characters in Four
Series of Egyptian Skulls," Annals pf Eugenics 6 (1935):352-71.

2R. A. Fisher, "The Use of Multiple Measurements in Taxonomic
Problems," Annals g: Eugenics 6 (1936):179—88.

3 . . . .

D. D. Durand, "Risk Elements in Consumer Installment FinanCing,"
Studies 33 Consumer Installment Financing (New York: National
Bureau of Economic Research), 1941.

 

4H. Myers and E. W. Forgy, "Development of Numerical Credit
Evaluation Systems," Journal 9: American Statistical Association 50
(September l963):797-806.

5Jagdish Sheth, "The Multivariate Revolution in Marketing Research,"
Journal pg Marketing, Vol. 35, January 1971.

6J. E. Walter, "A Discriminant Function for Earnings-Price Ratios
of Large Industrial Corporations," Review 2E Economics and Statistics,
February 1959, pp. 44—52.

7Edward Altman, "Financial Ratios, Discriminant Analysis and the
Prediction of Corporate Bankruptcy," Journal pg Finance 23 (September
l968):589-609.

8Edward Altman, "Railroad Bankruptcy Propensity," Journal 95

Finance, May 1971, pp. 333-45.

9Irma Adelman and Cynthia Morris, "Performance Criteria for

Evaluating Economic Development Potential: An Operational Approach,"
Quarterly Journal pf Economics 82 (May 1968):260-80.

 

loWilliam Cochran, "On the Performance of the Linear Discriminant

Function," Technometrics, May 1964, pp. 179—90.

11T. W. Anderson. Ag Introduction 39 Multivariate Statistical
Analysis, (New York: John Wiley and Sons, 1958).

 

81

82

12 . . . . .
M. M. Tatsouaka and D. V. Tiedeman, "Discriminant AnalySis,“

Review pg Educational Research. (Washington, D.C.: American
Educational Research Association, 1954).

J. G. Bryan, "The Generalized Discriminant Function: Mathemati-
cal Foundation and Computional Routine," Harvard Educational Review,
no. 2, pp. 90-96.

D. G. Morrison, "On the Interpretation of Discriminant Analysis,"
Journal pg Marketing Research 6 (May 1969):156-63.

13Frank, Ronald; Massy, William and Morrison, Donald. "Bias in
Multiple Discriminant Analysis," Journal g: Marketing Research,
August 1965, pp. 250-58.

 

14All the variables used were discussed in the preceding chapter.

15For a detailed explanation of what the results mean, see
pairing (1).

 

16 . . . .
For an explanation on how to interpret the results see pairing

(1).

17See pairing (1) for an explanation on how to interpret the

results reported.

18See pairing (1) for a detailed explanation of the reported

results.

CHAPTER VI

SUMMARY AND SIGNIFICANCE OF THE RESEARCH

This research was organized into two parts. One part examined the
reasons for the airlines to merge and the reasons for the CAB in
approving or disapproving a merger application. The second half of
the study used a statistical technique, multiple discriminant analysis,

to quantitatively identify airline group differences concerning

 

merger activity. The two methods of research dealt with the same goal
of divulging merger information about airlines. The first approach
was qualitative in nature; it examined the qualitative reasons of the
airlines to merge and the reasons of the CAB to approve or to
disapprove such merger proposals. The second approach was quantitative
in nature; statistically it examined the group differences of the
airlines concerning their merger activity.

Reasons of the airlines to merge have been the desire to achieve
economies of scale, to improve management, to take advantage of a
bargain price, to reduce business and financial risk, to affect tax
savings, to grow rapidly, to improve service and increase demand, to
increase monopoly power and to correct or relieve operating weaknesses
and financial difficulties.

The benefits that the board expected from mergers have been
economies of scale, replacing inefficient management, reduction in

business and financial risk, tax savings, improved service and increases

83

 

 

84

in demand. Typically, the board has been responsive to these expected
benefits of mergers only when it has considered a merger desirable in
order to improve the operating characteristics and the financial
condition of a carrier. The board has been reluctant to reduce
significantly competition by mergers. In each merger that the CAB has
approved, it has found the price to be reasonable and the labor
protective provisions satisfactory. In all but two approved mergers,
the board has found that the merger would not create a monopoly,
jeopardize another carrier or cause significant diversion of traffic
from other carriers.

Multiple discriminant analysis was employed to detect group
differences, to classify firms and to identify variables important
in uncovering group differences. Discriminant analysis was used to
detect any differences between the following groups based upon the
variable input, and, if so, what variables detected this difference
between the groups.1
1) The airlines which proposed to be the acquired or the absorbed
airlines relative to the airlines which proposed to be the acquiring
airlines in merger applications exhibited unfavorable financial, opera—
ting and route characteristics and were of a smaller size.
2) The airlines which did not propose mergers relative to the air-
lines which proposed to be the acquiring airlines in merger applications
exhibited unfavorable financial, operating and route characteristics.
3) The airlines which did not propose mergers relative to the airlines
which proposed to be the acquired airlines in merger applications
exhibited more favorable financial, operating and route character-

istics.

 

 

85

4) No difference was found between those airlines which did not
propose mergers and the airlines which did.

5) No difference was found between the airlines which proposed to be
the acquiring airlines and were approved by the CAB and the airlines
which proposed to be the acquiring airlines and were disapproved by the
CAB.

6) The airlines which proposed to be acquired and were approved by the
CAB relative to the airlines which proposed to be acquired and were
disapproved by the CAB exhibited unfavorable financial, operating and
route characteristics, were smaller in size and were receiving more
subsidy from the government.

7) No difference was found between the airlines which proposed to be
the acquiring airlines and were disapproved by the board and the air-
lines which proposed to be the acquired airlines and were disapproved
by the board.

8) The airlines which proposed to be the acquiring airlines and were
approved by the CAB relative to the airlines which proposed to be the
acquired airlines and were approved by the CAB exhibited more favorable
financial and operating characteristics, were larger in size and were
receiving less subsidy from the government.

9) No difference was found between the airlines approved for mergers
by the CAB and the airlines which were disapproved.

The quantitative results, therefore, indicated several con-
clusions. The airlines proposing to be the acquiring airlines
exhibited stronger financial, operating and route characteristics than
those airlines not proposing mergers and those airlines proposing to

be absorbed. In turn, the nonproposing airlines showed more favorable

 

86

quantitative characteristics than the airlines proposing to be acquired.
The CAB has approved most mergers when the acquired airline has shown
unfavorable financial, operating and route characteristics, when they
were of smaller size and when they had been receiving more subsidy from
the government relative to the airlines proposing to be acquired it
has disapproved for merger. In most mergers the CAB has approved the
acquiring airline has exhibited a stronger quantitative profile than
the acquired airline. On the other hand, the board has shown no
tendency to distinguish between the characteristics of the acquiring
airlines approved for mergers and those disapproved. Also, the
airlines proposing to be acquired which were disapproved by the board
were mostly not different than the airlines proposing to be the
acquiring airlines and were disapproved.

The main limitation of the qualitative approach was that it was
difficult to establish basic relationships in a complex situation. It
was necessary to interpret the available pertinent information to
identify the reasons for mergers. However, there were very few merger
cases where the reasons of the airlines and the CAB had to be inferred
from very limited information. The main limitation of the quantitative
approach was due to the necessity of attempting to reduce complicated
merger motives and CAB action to quantitative surrogates.

Mergers have a great actual and potential impact on the structure
of an industry. In order that antitrust and regulatory statutes can
be applied judiciously it is first necessary to have a sound under—
standing of the motives which bring two firms to a merger decision. In
addition, in order to be able to review the desirability of regulatory

policies on mergers, it is necessary to have a basic knowledge of the

 

87

reasons why the regulators make their decisions on merger proposals.
The structure of an industry is fundamental in explaining the conduct
and performance of an industry. Therefore, it seemed appropriate to
enrich the knowledge on a major determinant of structure, namely
mergers. The research was specific to the air transport industry

for several reasons--author interest, the significance of the industry
in the economy and the hope that a restricted scope would provide
understandable and specifically applicable results.

The air transport industry is an excellent area for merger
research or for that matter, other economic research. Documents,
reports and data are readily available. The firms in the industry
are very homogeneous; data is comprehensive and comparable. It should
be possible and, indeed, useful to investigate other aspects of mergers
in this industry, such as a capital budgeting approach to the merger
decision or a post-merger evaluation of the effects of mergers on the

conduct and performance of the individual firms and the industry.

 

 

 

 

 

FOOTNOTES

1The listing of the group-pairings is identical to their order for

the results of chapter five. For a more detailed explanation of the
results see chapter five.

88

 

 

 

 

APPENDIX I

 

The Federal Aviation Act of 19581

APPENDIX I

LEGAL STANDARDS FOR MERGERS

2 . 3 . . 4
Board to approve or disapprove any merger between air carriers.

Section

5
mergers.

408 of this act specifically confers the board's power over

approved by order of the board

for two or more air carriers, or for any air carrier and
any other common carrier or any person engaged in any
other phase of aeronautics, to consolidate or merge their
properties, or any part thereof, into one person for the
ownership, management, or operation of the properties
theretofore in separate ownerships.

The board also pointed out that

Section

jurisdiction over merger transactions may also be asserted
under section 408(a) (l) which, among other things,
requires Board approval for the purchase of a substantial
part of the properties of one carrier by another and

under section 408(a) (5) which, among other things,
requires Board approval for any air carrier to acquire
control of another in any manner whatsoever.

408(b) provides that:

Any person seeking approval of a consolidation, merger,
purchase, lease, operating contract, or acquisition of
control, specified in subsection (a) of this section,
shall present an application to the Board, and
thereupon the Board shall notify the persons involved
in the consolidation, merger, purchase, lease,
operating contract, or acquisition of control, and
other persons known to have a substantial interest in
the proceeding, of the time and place of a public
hearing. Unless, after such hearing, the Board finds
that the consolidation, merger, purchase, lease,
operating contract, or acquisition of control will

not be consistent with the public interest or that the

89

authorizes the Civil Aeronautics

 

Section 408(a) (1) provides that it shall be unlawful unless

 

 

90

conditions of this section will not be fulfilled, it
shall by order, approve such consolidation, merger,
purchase, lease, operating contract, or acquisition
of control, upon such terms and conditions as it
shall find to be just and reasonable and with such
modifications as it may prescribe: Provided, that
the Board shall not approve any consolidation,
merger, purchase, lease, operating contract, or
acquisition of control which would result in creating
a monopoly or monopolies and thereby restrain
competition or jeopardize another air carrier not a
party to the consolidation, merger, purchase, lease,
operating contract, or acquisition of control.8

Under section 408(b), then, an application for a consolidation or
merger is to be approved, unless it is found that the proposed consoli-
dation or merger will not be consistent with the public interest, or
that it will result in creating a monopoly and thereby restrain
competition or jeopardize another air carrier not a party to the
. . 9 . . . .
consolidation or merger. Section 408(b) also reqUires the filing
of an application by parties seeking relief from the restrictions of
section 408(a), and a subsequent hearing upon such an application.
Section 408(b) directs the board to determine if a proposed merger
is inconsistent with the public interest. The board has held on many
occasions that the standards for determining the public interest are
. . 10 . .
enumerated in section 102. In the first merger case under regulation
the Civil Aeronautics Authority pointed out that:
"Public Interest" as thus used in the act is not a mere
general reference to public welfare, but has a direct
relation to definite statutory objectives. Thus,
section 2 of the act directs the authority to consider
certain specific objectives as being in the public
interest. 1

Section 102 states:
In the exercise and performance of its powers and duties
under this act, the Board shall consider the following,
among other things, as being in the public interest,

and in accordance with the public convenience and
necessity:

 

91

(a) The encouragement and development of an air
transportation system properly adapted to the present
and future needs of the foreign and domestic commerce
of the United States, of the Postal Service, and of
the national defense;

(b) The regulation of air transportation in such
manner as to recognize and preserve the inherent ad—
vantages of, assure the highest degree of safety in,
and foster sound economic conditions in, such trans-
portation, and to improve the relations between, and
coordinate transportation by, air carriers;

(c) The promotion of adequate, economical, and
efficient service by air carriers at reasonable charges,
without unjust discriminations, undue preferences or
advantages, or unfair or destructive competitive
practices;

(d) Competition to the extent necessary to assure the
sound development of an air—transportation system
properly adapted to the needs of the foreign and domestic
commerce of the United States, of the Postal Service,
and of the national defense;

(e) The promotion of safety in air commerce; and

(f) The promotion, encouragement, and development of
civil aeronautics.

 

In each case the evidence is examined in light of the criteria given by

the act in Section 102. In the Southwest - West Coast Merger Case it

 

was stated that

the Board has frequently held in proceedings of this
nature, the public interest standard referred to in
section 408 of the Act must be interpreted in the light
of the statutory objectives laid down in section 2.
Therefore, to be consistent with the public interest
the proposed merger must be a step ngard effectuating
the aims and objectives of the Act.

In the same case the board merely added:

The statutory purposes of the Act are well known and

fully understood and need not be restated in detail

here. Suffice it to say that the underlying objective

of these purposes is the sound development of an adequate,
economical, and efficient air transportation system,
properly adopted to the present and future needs of the
foreign and domestic commerce of the United States, of the
Postal Service, and of the national defense.1

Section 401(h)15 of the FAA provides that "no certificate may be

transferred unless such transfer is approved by the board as being

 

 

 

 

92

consistent with the public interest."16 Section 401(h) has applied to
all consolidation and merger applications since all applications have
included the request to transfer the certificate of public convenience
and necessity of the acquired airline to the surviving airline.

Section 408(b) requires only the negative finding that the
proposed merger will not be inconsistent with the public interest for
the board to approve a merger. Approval of the transfer of a
certificate of public convenience and necessity under section 401(h)
requires as affirmative finding that the proposed transfer will be
consistent with the public interest. Thus the public interest is the
primary consideration under each of the sections. In connection with
the negative versus affirmative required finding the board said:

Under our interpretation of these sections, in accordance
with well-established rules of construction, the
distinction between the affirmative and the negative
findings is an unimportant one for the purposes of this
decision since the public interest is the primary
consideration of either of them.

If the proposed merger involves an air carrier which is authorized
to engage in foreign air transportation, it is also subject to the

. . l8 . l9 . .
approval of the preSident under section 801. Section 414 immunizes
an airline merger approved by the board from prosecution under the
2
antitrust laws. 0

The broad and general statutory principles in effect allow the
board to exercise great discretion in approving or disapproving mergers.
In addition, section 408(b) allows the board to attach such terms and
conditions to a proposed merger as it finds just and reasonable.

According to a Senate Commerce Committee report, the statute was

. 2 . .
intended by Congress to "favor mergers, where pOSSible." l Airline

 

 

 

93

mergers appear to be judged by the policies of the antitrust laws only
to the extent that they are assumed to be embodied in the public

interest criteria of section 102.22 In McLean Trucking Co. v. United

 

23 . . .
States, the Supreme Court upheld the antitrust immunity of the
Interstate Commerce Act, which contains a provision comparable to
section 414 of the FAA. The Court held that the commission was not
bound to apply the antitrust laws rigorously to a merger of motor
carriers, and, rather, the commission in each case was to give "strict
regard" to the antitrust laws and then to strike an appropriate
balance between competition and the other objectives of the act. The
board stated in the Local Cartage Agreement Case that:
Where an agreement has among its significant aspects
elements which are plainly repugnant to established
antitrust principles, approval should not be granted
unless there is a clear showing that the agreement is
required by a serious transportation need or in order
to secure important public benefits.

Although the board has never disapproved a merger on the explicit
grounds that it would create a monopoly it has stated that a merger
would not be consistent with the public interest if it would give the
merger airline control of air transportation greatly in excess of that
possessed by competitors, or would endanger a reasonably balanced

25

system of air transportation in every section of the country. In

American Airlines, Inc., Acquisition of Control of Mid-Continent

 

Airlines, Inc.26 the board again stressed the competitive factor in
determining the public interest.27 It was concerned with the competi—
tive advantage inherent in sheer size and geographical scope of
operations and also the potential attendant economic injuries to rival

carriers.

 

94

All decisions by the board are subject to judicial review except
mergers involving a foreign air carrier or foreign routes subject to
approval by the president.28 The Supreme Court has held in Chicago So.
Airlines, Inc. v. Waterman S. S. Corp.29 that presidential action under
section 801 is not subject to review by the courts. Any other order
by the board is reviewable by either the federal court of appeals for
the circuit where the petitioner resides or has his principal place of
business or the Court of Appeals for the District of Columbia.30
Section 1006(e) directs that board findings shall be conclusive "if
supported by substantial evidence."31 However, the courts have often
stated that the board must not act arbitrarily.32 The question on
appeal is whether there is substantial evidence to support findings
of fact and whether the inferences drawn are reasonable. Reviewing
courts, however, have been reluctant to substitute their judgment for
that of the agency.

Other influences on the board's actions, although indirect,
include congressional pressure, the Justice Department's role as an
intervener and additional interveners appearing before the board. But
as an independent agency, the board is delegated by Congress to exer-
cise considerable discretion in deciding merger cases; this power has

been upheld by the courts. Moreover, the board is empowered to attach

terms and conditions to an approved merger as it sees fit.

 

 

 

 

FOOTNOTES

1The Civil Aeronautics Act of 1938 [52 Stat. 973 (1938); 49 U.S.C.
1301-1542 (1938)] was the predecessor statute governing mergers and was
adopted, except for minor changes, as the Federal Aviation Act of 1958
[72 Stat. 731; 49 U.S.C. 1301 — 1542 (Supp. 1958); hereafter referred
to and cited as FAA]. The sections of the FAA of 1958 concerning
mergers are identical to the sections of the Civil Aeronautics Act of
1938 except section 2 of 1938 is section 102 of 1958 and section
401(i) of 1938 is section 401(h) of 1958.

2 . . . , . . .
Originally the Civ1l Aeronautics Authority until 1940.

3 , . . . .
The term merger will include consolidation herein.

4Air Carrier is defined to mean "any citizen of the United States

who undertakes, whether directly or indirectly or by a lease or any
other arrangement, to engage in air transportation" [FAA 101 (3),
49 U.S.C. 1301 (3) (1958)].

5FAA 408(a), (b), 49 U.S.C. 1378(a), (b) (1958).

672 Stat. 767 (1958), 49 U.S.C. 1378(a) (1958).

714 CAB 356, 360 (1951).

852 Stat. 1001 (1938); 49 U.S.C. 1378(b) (Supp. 1958).

9The board has interpreted monopoly as a particular degree of

control of air transportation rather than a general tendency to stifle
competition and control prices. See 1 CAA 739 (1940). It has been
suggested that the board has read the monopoly proviso out of the
statute. See the following comment: "Merger and Monopoly in Domestic
Aviation," Columbia Egg Review 62 (May 1962):851—83.

1072 Stat. 740 (1958), 49 U.S.C. 1302 (1958). Section 102 of the
FAA of 1958 was section 2 of the Civil Aeronautics Act of 1938; 52
Stat. 980 (1938); 49 U.S.C. 1302.

95

 

 

 

 

96

111 CAA 739, 741 (1940). For other references to the criteria
of public interest provided by section 102, see 2 CAB l, 4 (1940); 2
CAB 503, 506 (1940); 4 CAB 654, 656 (1944); 4 CAB 680, 681 (1944);
7 CAB 365, 371-72 (1946).

1272 Stat. 740 (1958), 49 U.S.C. 1302 (1958).

1314 CAB 356, 361 (1951).

l4Ibid., note 4.

15Section 401(i) in the 1938 act is section 401(h) in FAA (1958).

16FAA 401(h) (1958); 49 U.S.C. 1371(h) (1958).

174 CAB 680, 681 (1944). This concept was affirmed in 16 CAB 647,
685-86 (1952).

18FAA 801, 49 U.S.C. 1461 (1958).

1972 Stat. 770 (1958); 49 U.S.C. 1384 (1958).

20The Justice Department has intervened in proceedings although
the board considers them and any other intervener with a position to
state. Also the Justice Department has appealed rulings of the board
to the courts. The Supreme Court has generally held that the regula-
tory agency rather than the Justice Department has the final word.

21U. S. Senate, Report No. 445, 87th Cong., lst sess., 701 (1961).

22 . . .
G. E. Hale and R. D. Hale, "Competition or Control: Air

Carriers," University pg Pennsylvania Egg Review 109 (January 1961):3ll-
60; G. E. Hale and R. D. Hale, "Competition or Control VI: Application
of Antitrust Laws to Regulated Industries," ibid. 111 (November 1962):
46—59; Louis B. Schwartz, "Legal Restriction of Competition in the
Regulated Industries: An Abdication of Judicial Responsibility,"
Harvard Egg Review 67 (January 1954):436-75.

 

23321 U.S. 67 (1944).

2415 CAB 850, 852 (1952).

251 CAA 739, 749—50 (1940).

267 CAB 365 (1946).

 

 

 

 

 

97

27Ibid., 378-79.

28FAA 1006(e), 49 U.S.C. 1486(a) (1958).

29333 U.S. 103 (1948).

3OFAA 1006(b), 49 U.S.C. 1486(b) (1958).

3J'FAA 1006(e), 49 U.S.C. 1486(e) (1958).

32North Central Airlines v. CAB, 265 F. 2d 581, 584-85 (D.C. Cir.),

cert. denied, 360 U.S. 903 (1959); Consolidated Flower Shipment v. CAB,
213 F. 2d 814 (9th Cir. 1954).

 

 

SELECTED BIBLIOGRAPHY

 

SELECTED BIBLIOGRAPHY

Books

Alberts, William W. and Segall, Joel E. (ed) The Corpgrate Merger,
Chicago: University of Chicago Press, 1966.

Anderson, T. w. Ag Introduction pg Multivariate Statistical Analysis,
New York: John Wiley and Sons, 1958.

 

Archer, Stephen H. and D'Ambrosio, Charles A. Business Finance:
Theory and Management, New York: Macmillan Company, 1966.

Caves, Richard. Air Transport and its Regglators, Cambridge, Mass.:
Harvard University Press, 1962.

Frederick, John. Commercial Air Transpgrtation, Homewood, Ill.:
Richard Irwin, 1955, 4th edition.

Harvey, John and Newgarden, Albert. Management Guides pg Mergers and
Acguisitions, New York: John Wiley and Sons, 1969.

Hutchinson, G. S. (ed.) The Business gf_Acguisitions and Mergers,
New York: Presidents Publishing House, 1968.

Mace, Myles and Montgomery, George. Management Problems g: Corporate
Acguisition, Boston: Harvard University Press, 1962.

Reid, Samuel R. Mergers, Managers and the Economy, New York: McGraw—
Hill, 1968.

straszheim, Mahlon. The International Airline Industgy, Washington,
D.C.: The Brookings Institution, 1969.

Van Horne, James C. Financial Management and Policy, Englewood Cliffs:
Prentice-Hall, 1968.

Weston, J. Fred and Brigham, Eugene F. Managerial Finance, New York:
Holt, Rinehart and Winston, 1969.

Wheatcroft, Stephen. Air Transport Poligy, London: Michael Joseph,
1964.

98

 

99

Articles and Periodicals

 

Adelman, Irma and Morris, Cynthia. "Performance Criteria for
Evaluating Economic Development Potential: An Operational Approach,"
Quarterly Journal pg Economics, LXXXII (May 1968) 260—280.

Alberts, William W. "The Profitability of Growth by Merger," in The
Corporate Merger, edited by William W. Alberts and Joel E. Segall,
Chicago: University of Chicago Press, 1966, 235—287.

Altman, Edward. "Financial Ratios, Discriminant Analysis and the
Prediction of Corporate Bankruptcy," Journal pg Finance, XXIII
(September 1968) 589—609.

Altman, Edward. "Railroad Bankruptcy Propensity," Journal pg Finance
(May 1971) 333-345.

Bryan, J. G. "The Generalized Discriminant Function: Mathematical
Foundation and Computational Routine," Harvard Educational Review,
No. 2 (Spring 1951) 90-96.

 

Butters, Keith and Lintner, John. "Tax and Non-tax Motivations for
Mergers," in Elements 9: Financial Administration, edited by
John O'Donnell and Milton Goldberg, Columbus, Ohio: Charles
Merrill Books, 1962, 629-651.

Cochran, William. "On the Performance of the Linear Discriminant
Function," Technometrics (May 1964) 179—190.

Durand, D. D. "Risk Elements in Consumer Installment Financing,"
Studies ip Consumer Installment Financing, New York: National
Bureau of Economic Research, 1941.

 

Federal Trade Commission. Economic Repgrt pp Corporate Mergers,
Hearings on Antitrust and Monopoly, Committee of the Judiciary,
US Senate, 9lst Congress, lst sess., part 8A, UPSGO, 1969.

Frank, Ronald; Massy, William and Morrison, Donald. "Bias in Multiple
Discriminant Analysis," Journal pg Marketing Research (August
1965) 250—258.

Gort, Michael. "An Economic Disturbance Theory of Mergers," Quarterly
Journal g£ Economics (November 1969) 624-642.

Koontz, H. D. "Economic and Managerial Factors Underlying Subsidy
Needs of Domestic Trunk Line Air Carriers," Journal g: Air Law
and Commerce 18, No. 2 (Spring 1951) 127—156.

Lewellen, Wilbur. "A Pure Financial Rationale for Conglomerate Mergers,"
Journal pg Finance, XXVI (May 1971) 521-537.

 

100

Lintner, John. “Expectations, Merger and Equilibrium in Purely
Competitive Securities Markets," American Economic Review LXI,
No. 2 (May 1971) 109-114.

Marion, David. "The American-Eastern Application: Crucial Test of
CAB Merger Policy," University gfi Pennsylvania Law Review
(December 1962) 199.

McCarthy, George D. "Premeditated Merger," in Foundations for
Financial Management, edited by James Van Horne, Homewood, 111.:
Richard Irwin, 1966, 565-580.

McKenna, William E. "Merger, an Art and a Science," in Readings £5
Finance, edited by Harold Wolf and Lee Richardson, New York:
Appleton-Century-Crofts, 1966, 349-350.

Morrison, D. G. “On the Interpretation of Discriminant Analysis,"
Journal pg Marketing Research VI (May 1969) 156-163.

Myers, H. and Forgy, E. W. "Development of Numerical Credit Evaluation
Systems," Journal g: American Statistical Association, L (Septem—
ber 1963) 797-806.

 

Proctor, Jesse W. and Duncan, Julius. "A Regression Analysis of
Airline Costs," Journal g£_Air Law and Commerce, XXI (Summer 1954)
282—292.

Schwartz, Samuel. "Merger Analysis as a Capital Budgeting Problem,"
in The Corporate Merger, 139—150.

Walter, J. E. "A Discriminant Function for Earnings-Price Ratios of
Large Industrial Corporations," Review g: Economics and Statistics,
(February 1959) 44-52.

Weiss, Leonard. “An Evaluation of Mergers in Six Industries," Review
g£ Economics and Statistics (May 1965) 172-181.

 

 

 

 

 

 

 

 

 

 

 

 

.. . :73... . .
.t \...,

MICHIC
1
1
!

6..., 53¢ 31.1.... .
.\x.. $32.2...1
1 .. L1. 1

xi! rw \. 3
“$34.... 3‘ A. y...
. : . 3an ha. ,

.. in...»

tap?!
... . .1...

1..

r... _
5.2.1.53»: .155...
,{su\.1. EX}!!!
1.3.. .N wits
5.x... E
‘1‘: .. . 41...... :
51734.. .,. ,

1... x. .

, , . , 55...»?
. .yv...—....

...

z r: .
.....»...:.5.:.
J... :7.

7....
,...t.

.. .
:53?“
. .‘ \tv'r
.~ m...» ewv... u.“ 3%;
.. .. 2:...
112.1,... 1. .
. 1.... ..
,.:..:
.322...
. .
. .

t. _. ..
£334...

a

i. . 2..
. S f
. ,. T... . .
Laura: . , . nu. manta ~41“
. l2: _ 3. :5 a... 2.: =1
. . . 13:9. ..... an...“
I , . v .r... War: . z o.
. , . w x
. . . . w. w
.. . . . . . _ . , .
. . . .. , , . , . ......~........... .
,....r...... .
.. . L. , . I .

L.

 

.
.r. .
24.11....