t
6

‘
C

PLACE IN RETURN BOX to remove this cléuut from your mom.
TO AVOID FINES Mum on or baton (MGM. .g. . .

DATE DUE DATE DUE DATE DUE
at ‘5 1993

 

 

 

 

 

143*

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

CZLJ
-| _
ll IE:
1| I I

usu I. An AII'Irmdlvo Adlai/Emu Opponumy summation
W

 

 

 

 

 

 

 

m1

 

 

THE RELATION BETWEEN RELATEDNESS AMONG JOINT VENTURE PARTICIPANTS
AND PARENT FIRMS' MARKET PERFORMANCE

By

Duangporn Chotanakarn

A DISSERTATION

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

DOCTOR OF PHILOSOPHY

Department of Management

1990

ABSTRACT

THE RELATION BETWEEN RELATEDNESS AMONG JOINT VENTURE PARTICIPANTS
AND PARENT FIRMS' MARKET PERFORMANCE

By

Duangporn Chotanakarn

The present research investigates the relationship between two
key variables: relatedness among joint venture participants and parent
firms’ market performance. The basic argument is that there is a
positive relationship between relatedness among joint venture
participants and parent firms' market performance. Relatedness was
defined as similarities between firms in products or markets. Three
types of relatedness among the participants of each joint venture were
examined: relatedness between parents, relatedness between one parent
and child, and relatedness between two parents and child. Firms' market
performance was defined as performance valued by the capital market and
was reflected by stockholder returns. Abnormal stockholder return--the
deviation of observed return from expected return-~was used to capture
changes in firms' market performance.

The entire sample was composed of 209 firms that announced joint
ventures in the Wall Street Journal during 1975-1987. The relationship
between joint venture announcements and abnormal stock returns for the
entire sample was first examined. Announcement period abnormal stock

return for a portfolio of firms was derived, and a t test was performed

to check the significance of this return. The relationships between
each type of relatedness among joint venture participants and parent
firms' abnormal stock returns were analyzed next. Linear regression
techniques were used in these analyses.

Results support the hypothesized relationships between (a) joint
venture announcements and parent firms' abnormal stock returns, (b) one
parent-child relatedness and parent firms' abnormal stock returns, and
(c) two parents-child relatedness and parent firms' abnormal stock
returns. The hypothesized relation between parent-parent relatedness
and parent firms' abnormal stock returns was not supported.
Implications of the empirical results for research and practice were

discussed along with suggestions for future research.

Copyright by
DUANGPORN CHOTANAKARN

1990

DEDICATED TO MY PARENTS

ACKNOWLEDGEMENTS

I wish to thank my committee members: Dr. James E. Skivington,
Dr. Glenn 8. Omura and Dr. Joseph H. Anthony. Dr. Skivington was more
than simply the chairperson of my dissertation committee. From the time
I entered the doctoral program, he has furnished me his willing
contribution, committed integrity, considerate care, compassionate
patience, and valuable time. Having worked closely under his guidance
on this dissertation, I believe that he realizes how invaluable his
support has been to me, personally and professionally.

Dr. Omura gave me his inestimable comments, which contributed
significantly to the preparation of this dissertation. His competence,
intelligence, and friendliness enabled my dissertation to progress
effectively. His constant assistance in this dissertation and in my
other research was greatly appreciated.

Dr. Anthony gave me time and useful suggestions, especially in
the methodology section, and this helped strengthen my dissertation. I
am so grateful for his kindness.

The completion of my doctoral program has been facilitated by
Dr. Michael Moch, my advisor. He was there so often when I needed
assistance. I am indebted to him for his concern for me.

I also wish to express my appreciation to my colleagues and the

faculty and staff in the Department of Management. Their understanding

vi

and friendliness made the working environment less stressful for me.

Finally, I wish to express my profound gratitude to my parents,
Rear Admiral Chalerm and Mrs. Chaveevan Chotanakarn. Their unrelenting
love and support made the road to my doctoral degree bearable and,

beyond that, my accomplishment more worthwhile.

December, 1990 Duangporn Chotanakarn

vii

VOL

v...

.
P;

Hr“

9‘“
it“

LIST OF TABLES .

LIST OF FIGURES.

CHAPTER ONE:

CHAPTER TWO:

CHAPTER THREE:

CHAPTER FOUR:

CHAPTER FIVE:

TABLE OF CONTENTS

INTRODUCTION .

Overview .

Key variables. .
Question of Interest .
Model. .

Premises of the Study.
Significance of the Study.
Scope of the Study. .
Research Method Synopsis .
Order of Presentation.

LITERATURE REVIEW, THEORETICAL FRAMEWORK AND
RESEARCH HYPOTHESES.

Background . .
Literature Review.
Relatedness. .
Market Performance: Stockholder Return .

Theoretical Framework and Research Hypotheses.

METHODOLOGY.

Conceptual Definitions .
Measurement.

Data Sources . .
Sample Selection . .
Methods of Analysis.

EMPIRICAL RESULTS.

Sample .
Tests of Hypotheses.

DISCUSSION, IMPLICATIONS AND CONCLUSIONS .

Discussion .
Implications .

viii

Page

xi

15

15
16
20
24
26

43
43
48
57
58
61
7O

70
75

89

89
98

TABLE OF CONTENTS (continued) Page

Study Limitations/Suggestions for Future Research. 100
Conclusions. . . . . . . . . . . . . . . . . . . . 103

BIBLIOGRAPHY . . . . . . . . . . . . . . . . . . . . . . . . . . . . 105

ix

Iabl

1
:2

Table

10

11

12

LIST OF TABLES

Market Categories and Subcategories.

Information about Business Segments of Parent Firms

Measure of Relatedness between Two Parents .

Joint Venture Sample .

Descriptive Statistics and Correlation Matrix for Variables.

Summary of Abnormal Stock Returns Surrounding the
Announcements of Joint Ventures .

Results from the Regression of Announcement Period Abnormal
Stock Returns on Parent-Child Relatedness .

Results from the Regression of Announcement Period Abnormal
Stock Returns on Parent-Parent Relatedness.

Results from the Regression of Announcement Period Abnormal
Stock Returns on Differences in Parents-Child Relatedness .

Results from the Regression of Announcement Period Abnormal
Stock Returns on Relatedness and Potential Confounding
Variables .

Comparison between Regulated, Nonregulated Subsamples and
Full Sample . . . . . . . . . . . . . . . . . . .

Regression Results for Regulated and Nonregulated Subsamples.

Page
47
52
53
71

73

76

78

79

81

83

85

87

LIST OF FIGURES

Figure Page

1 Examples of Joint Ventures that Differ in the Extent of

Relatedness. . . . . . . . . . . . . . . . . . . . . . . . . 23
2 Higher Parent-Child Relatedness. . . . . . . . . . . . . . . 29
3 Lower Parent-Child Relatedness . . . . . . . . . . . . . . . 32
4 Higher Parent-Parent Relatedness . . . . . . . . . . . . . . 34

xi

CHAPTER ONE

INTRODUCTION

ervi w

How do joint ventures affect firms' performance? One view is
that joint ventures lead to superior performance of parent firms.

Parent firms enjoy superior performance because the joint venture allows
for more efficient utilization of complementary resources (McConnell &
Nantell, 1985). Superior performance also may be derived from the joint
venture's ability to increase the market power of the parent firms
(Pfeffer & Novak, 1976a).

A different view is that joint ventures lower parent firms'
performance (Berg & Friedman, 1981). Parent firms may bear high costs
when the joint venture penetrates into a new, unfamiliar product or
market area. High costs in learning about the new product/market areas
and in competing with established firms may not be adequately covered by
the parent firms' future income streams. This will lead to lower
performance of the parent firms.

This study proposes that joint ventures lead to differences in
parent firms' performance because there are differences in relatedness
among joint venture participants. The "relatedness" construct
originated in diversification studies and has been used to explain the

performance of firms involved in internal diversification or in

2
merger/acquisition. Relatedness has been defined as similarities
between businesses in resources and skills used to manufacture products
or to serve markets (Rumelt, 1974; Salter & Weinhold, 1979).
Determination of relatedness between businesses has been based on
similarities in products or markets (Rumelt, 1974; Palepu, 1985; Capon,
Hulbert, Faley & Martin, 1988). Relatedness was found to be able to
explain differences in firms' performance--both accounting-based
performance (e.g., Rumelt, 1974; Bettis, 1981), including return on
assets, and market-based performance (Singh & Montgomery, 1987; Shelton,
1988), including stockholder return. This study's premise is that
relatedness can be applied to joint ventures; the basic argument in this
study is that relation exists between relatedness among joint venture
participants and parent firms' market performance. The higher the
product/market relatedness between joint venture participants, i.e.,
between parent and child or between parents, the higher the expected
market performance of the parent.

Parent firms that are more related to their child are expected
to be better performers than parent firms less related to their child
for several reasons. One reason is that high relatedness suggests that
the parent firm can share with its child specialized resources and
skills that are used specifically in production or marketing functions,
such as production technology, plants, raw materials, sales force,
distribution channels, and research and development (RSD) skills.
Relatedness also suggests that the parent firm can share with its child
general resources and skills that are not specifically tied to products
or markets, such as financial resources, which can be easily shared

across functions (Kitching, 1967; Lubatkin, 1984; Singh & Montgomery,

3
1987). High relatedness, therefore, may suggest to the capital market
that the parent firm can lower its average cost through sharing both
specialized and general resources with its child.

Conversely, low relatedness between parent and child may suggest
to the capital market that the parent firm can share only general
resources with the child and hence cannot lower its average cost much.
A parent and child that have low relatedness may not be able to share
resources and skills used specifically in production and marketing
functions. For instance, these firms may not be able to share
production technology, plants, sales force, distribution channels, or
R & D skills. Low opportunity to share such specialized resources also
suggests that the parent has to bear high startup costs. Startup costs
include: (a) costs of investments in assets such as production
technology, plants or raw materials (Ansoff, 1965); (b) learning costs,
which occur because management cannot transfer skills in original
product/market areas to new, unrelated product/market areas (Salter &
Weinhold, 1979); and (c) entry costs, such as R & D costs, costs of
switching buyers from the product of established firms to the child
firm's product, or costs of persuading distribution channels to accept
the child firm's product (Porter, 1980). Even though low relatedness
suggests that the parent can share general resources such as financial
resources with its child, the advantages associated with this sharing
tend to be offset by high startup costs. This, then, leads to the
expectation that the performance of parent firms that are more related
to their child will be better than the performance of parent firms less
related to their child.

Another type of relatedness in a joint venture is relatedness

4
between parents. This study expects that the higher the product/market
relatedness between parents, the higher the market performance for these
firms. High product relatedness between parent firms may suggest that
these firms have access to their partners' production technology.
Obtaining more information/knowledge about the production technology
used by their partners, the related parents may be able to improve their
technology. An opportunity to improve production technology and hence
to lower average costs can lead to high market performance for the
related parent firms. Similarly, high market relatedness between parent
firms may lead to high performance for these firms. A joint venture
formed by parent firms operating in similar markets may help strengthen
the market positions of these firms (Harrigan, 1985), and this means
that the firms can longer sustain their profit levels. Higher
relatedness between parent firms, therefore, tends to lead to parents'
higher performance than lower relatedness between parents.

Another type of relatedness that this study examines is
relatedness between parents and child, that is, the relatedness between
one parent and the child compared with the relatedness between the other
parent in the venture and the child. The present study argues that high
relatedness between one parent and child may lead not only to high
market performance for that parent firm, but also to high market
performance for the firm's partner that is unrelated to the child. High
relatedness between one parent and the child suggests that the parent
possesses some resources/skills that allow for the child to penetrate
into related markets at low cost. This means that the partner firm of
this parent can benefit from the parent's relatedness with the child

because the partner does not have to bear high startup costs.

5
Therefore, compared among unrelated parent firms across joint ventures,
the parent firms whose partners are more related to the child are
expected to perform better than the parent firms whose partners are less
related to the child.

The logic underlying this final expectation is consistent with
that of the prior expectations. That is, higher relatedness leads to
higher market performance for the parent firm. This final expectation
only extends the logic to apply to the joint ventures where parent firms
differ in their relatedness with the child. Greater differences between
parents as such creates just like spillover in performance from one
parent to the other parent in the venture.

To sum up, this study expects that, for a joint venture
consisting of two parents, the market performance of a focal parent can
be explained by: relatedness between the focal parent and the child,
relatedness between parents, and differences between the focal parent
and its partner in their relatedness with the child. Relatedness can be
differentiated as more related and less related, based on similarities

between the businesses in products or markets.

Rey—Variables
Relatedness among Joint Venture Participants
Joint venture involves the creation of a new organizational
entity by two or more partner organizations (Bergman, 1962; Boyle,
1968). The partner organizations are called "parents', and the created
joint venture is called ”child" (Harrigan, 1985). Parent and child are
participants in a joint venture. At least three participants, i.e., two

parents and one child, are involved in a joint venture.

6

This study focuses on the capital market reactions to
relatedness differences among joint venture participants when the joint
ventures were first publicly announced. Public announcement of a joint
venture generally provides some information about the businesses of
parent firms and a child firm. This leads to a possibility that (a)
relatedness among joint venture participants can be inferred from the
joint venture announcement and (b) relatedness can be used by the
capital market to determine the value of the joint venture for the

parent firms.

Firms' Market Performance

Market-based performance is concerned with the market valuation
of firms (Salter & Weinhold, 1979; Singh & Montgomery, 1987). An
increase in a firm's market value indicates market expectation of an
increase in the firm's future income stream (Rappaport, 1986). A firm's
increase in market value may be reflected by an increased stock price
(Salter & Weinhold, 1979; Singh & Montgomery, 1987). Market value
created through joint ventures may, therefore, be reflected by an
increase in stock prices approximating the joint venture announcement.
Joint ventures that lead to an increase in the stock prices of parent
firms beyond that required by investors mean value gains to stockholders
of these firms.

Firms' market value was chosen as a dependent variable for
several reasons. One reason is that a firm's market value is relevant
to the concern of management, whose goal is to maximize stockholder
value. An increase in a firm's market value reflects an expectation of

increase in the future income stream and hence an increase in future

7
stockholder value (Woolridge & Snow, 1990). A firm's market value can,
therefore, indicate the market expectation of the firm's long-term
viability.

Second, market value is a typical dependent variable in an
"event study" approach, the approach that has been used popularly in
finance and accounting literature. The approach focuses on capturing‘
market responses to seemingly interesting events. This approach has
been borrowed by several strategic studies and used to examine the
capital market responses to merger/acquisition announcements (e.g.,
Shelton, 1988; Seth, 1990) or joint venture announcements (Woolridge &
Snow, 1990). Because the focus of this study is consistent with the
event study approach, an investigation of a firm's market value hence
seems appropriate.

Finally, the market value of joint ventures is available for
measurement when the ventures are first announced. Early measurement of
the market value of joint ventures may lessen the effects of other
unrelated events. This advantage may not be obtained if the value of
joint ventures is measured after implementation. For instance, it may
take up to two or three years for a joint venture to generate an income
stream. By that time, it may be more difficult to separate the effects
of initiating the joint venture from other events that might have

influenced the firms' performance.

W

Previous studies on joint ventures found a positive relationship
between joint ventures and firms' market performance. McConnell and

Nantell (1985), for example, found that stockholders of parent firms

8
obtained significant value gains approximating the joint venture
announcement. This empirical finding was consistent with that of
Ravichandran (1986). Even though these two prior studies share similar
empirical findings, they seem to have different assumptions. McConnell
and Nantell (1985) presume that the capital market reaction to joint
venture announcements is homogeneous, whereas Ravichandran (1986)
assumes that the capital market reaction to joint venture announcements
is not uniform across all joint ventures. Ravichandran (1986) found
that the market reaction to joint venture announcements varied among
joint ventures in competitive and concentrated industries.

This study also examines the relationships between joint
ventures and firms' market performance. Based on the assumption that
the capital market reaction to joint venture announcements is not
homogeneous across joint ventures, the present study takes a different
position from that of prior studies. This study argues that differences
in the capital market reaction to joint venture announcements depend on
relatedness among joint venture participants. And because the capital
market reaction is assumed to be reflected by the parent firms' market
performance, the question of interest in this study can be expressed as:
"Is there a relationship between relatedness among joint venture
participants and parent firms' market performance?". The purpose of
this study is thus to test empirically the hypotheses relating to this

question of interest.

9
Essie].
The basic model that depicts the focus of investigation in the
present study can be simply shown as:
Relatedness --------- > Market Performance
Relatedness among firms involved in a joint venture is an
independent variable; parent firms' market performance is a dependent
variable. Even though the above model does not explicitly include
potential confounding variables, this study controls for these
variables. Several variables that might have confounding effects
include: parent firm size (McConnell & Nantell, 1985), the
concentration of the child industry (Ravichandran, 1986), the potential
market impact of the parents, and specific regulated industries

(McConnell & Nantell, 1985; Ravichandran, 1986).

W

Several premises in this study can be made explicit. First, not
all joint ventures are homogeneous. Relatedness among participants in
different joint ventures may vary. Relatedness between parent firms,
for example, may differ across joint ventures. In a similar vein,
relatedness between parent and child firms may differ across joint
ventures. Plausible differences in the extent of relatedness across
joint ventures lead to the conjecture of variations in capital market
reaction to joint venture announcements.

Second, firms are viewed as a broad set of resources/skills
(Penrose, 1959; Rubin, 1973). Joint venture is one means that allows
for more effective allocation of resources/skills of parent firms. For

example, the parent firms may utilize more efficiently their excess

10
resources/skills by contributing them to their child.

Finally, the capital market is assumed to act efficiently in the
sense that stock prices adjust very rapidly to new information (Fama,
Fisher, Jensen & Roll, 1969). This assumption also implies that stock
prices are indicators of value of resources being allocated. Hence, an
announcement of joint venture will be efficiently processed by the
capital market in such a way that the stock values of the parent firms
reflect how the market values the resource allocation associated with
the joint venture. This assumption of capital market efficiency has

been supported by several studies (e.g., Fama, 1976).

W

The present study is significant in several aspects. First, the
study can provide additional empirical results regarding the effects of
joint ventures on firms' performance. Though joint ventures have become
more popular strategic activities, especially in the recent decade
(Harrigan, 1985), the consequences of joint ventures on firms'
performance are still ambiguous. Few previous studies have examined
relationships between joint ventures and firms' market performance.
Most of these prior studies implicitly assume that joint ventures have
homogeneous effects on firms. This study, however, argues that not all
joint ventures are homogeneous in their effects on firms because the
extent of relatedness among joint venture participants may vary. The
examination of this argument can thus offer insight into the nature of
the relationships between joint ventures and firms' performance.

Second, this study applies the relatedness concept (Rumelt,

1974) in a slightly different way from that typically used in

ll
merger/acquisition studies. With respect to a merger or an acquisition,
only one type of relatedness--relatedness between acquiring and acquired
firms--has been examined in its effects on the firms' performance. With
respect to a joint venture, however, at least three types of relatedness
may affect firms' performance. These relatedness types are:
relatedness between one parent and child, relatedness between parents,
and relatedness between both parents and child. Relatedness between
both parents and child, in particular, has not been examined in prior
studies. The investigation in the present study, therefore, can
contribute to joint venture studies.

In terms of practical significance, the focus of this study on
the relation between relatedness among joint venture participants and
firms' stock returns seems relevant to management, whose goal is to
maximize stockholder value. Maximization of stockholder value is
significant for management for several reasons. One reason is that high
stockholder value is a performance goal that extends beyond the
stockholders' domain to the firm as a whole. For instance, high stock
value may inhibit a proxy battle for control and can enable a firm to
have access to cheap sources of additional capital or equity (Branch &
Gale, 1983; Lubatkin & Shrieves, 1986; Rappaport, 1986). Other reasons
that may explain management's high attention to stockholder returns are:
large stock ownership by management, a tie between management
compensation and stockholder returns, and management's concern for its’

reputation in managing stockholder returns.

12
W

This study focuses on domestic joint ventures that were publicly
announced in the Hgll_§§;gg§_lgg;nal during the period 1975-1987. To be
included in the sample, joint ventures had to meet several criteria.
First, each joint venture was formed by only two parent firms. This
criterion helps control for any confounding effects of varying numbers
of parent firms being involved. Second, all parent firms were U.S.
corporations. This criterion helps control for any international
diversification effects of joint ventures on firms' performance. Third,
at least one of the parent firms in each joint venture had to be traded
on the New York Stock Exchange or American Stock Exchange, and the stock
returns data had to be available in the Center for Research in Security
Prices (CRSP) daily tape 100 days prior to the joint venture
announcement and 20 days following the announcement.

This study determines the relatedness among joint venture
participants based on similarities in products or markets of these
firms. Though it is recognized that the relatedness definition also
includes similarities in technology or R & D skills used, a stream of
research has determined relatedness mainly by product or market
similarities. These studies found that product/market relatedness was
adequate to explain differences in firms' performance (Palepu, 1985;

Capon et a1., 1988).

BMW

An initial sample of joint ventures was composed of joint

ventures that were first publicly announced in the Wall_§t;gg§_§guzngl

during January 1975 to December 1987. The announcements in the Wall

13
Sgrgggdlgggnal were used because they serve as a primary source for
initial sample selection for research in the finance area (Thompson,
Olsen & Dietrich, 1987). The index of joint venture announcements in
the Hg11_§51ggjglggznal was obtained from the joint venture section in
W123 and in the Hall
S;1gg§_lgn:n§1_lnggx. These two sources of data were used to complement
each other.

The approach used to derive relatedness among joint venture
participants, the independent variable in this study, mainly follows the
approach used by Shelton (1988). A continuous measure of relatedness
was derived, based on similarities in products/markets between business
segments of firms. The approach used to estimate parents' abnormal
stock returns, the dependent variable, follows that of McConnell and
Nantell (1985), who employ the residual analysis technique (Fama,
Fisher, Jensen & Roll, 1969).

The methods of analysis employed in this study are: a
correlation analysis, linear regression analyses and multiple regression
analyses. These analyses were used to detect relationships between or
among variables included in this study. The multiple regression
analysis also was used to cdntrol for the effects of potential

confounding variables.

Wan

This dissertation contains five chapters. Chapter One mainly
provides an overview for this study. Chapter Two reviews prior studies
that have examined the relationship between joint ventures and firms'

market performance. This is followed by theoretical arguments on the

14
relation between relatedness among joint venture participants and parent
firms' market performance. The hypotheses based on these arguments are
then proposed. Next, Chapter Three describes the methodology that is
used in the empirical investigation. The chapter presents: conceptual
definitions and measures of dependent, independent and potential
confounding variables; a description of the sampling procedure; and the
statistical methods for hypotheses testing. Chapter Four presents the
results of an empirical investigation. Chapter Five, the final chapter,
discusses the results in light of specific hypotheses, draws some

implications, and makes suggestions for future research.

CHAPTER TWO

LITERATURE REVIEW, THEORETICAL FRAMEWORK AND RESEARCH HYPOTHESES

W

Joint ventures have quite a long history in the United States.
Early use of joint ventures dates back at least to the year 1880, when
railroads used them for large-scale projects. Later, in the early 20th
century, joint ventures were formed to pool risks in shipping, gold
exploration and other undertakings (Harrigan, 1985). The formation of
joint ventures within the United States has blossomed since the mid-
19703 (Pfeffer & Novak, 1976a; Harrigan, 1985). Over the past decade,
more than 7,000 U.S. firms have entered into joint ventures across a
wide variety of industries, such as automobiles, communication services,
computers, electronic components, engines and aerospace, medical
products, pharmaceuticals, videotapes and videodiscs, financial services
and others. Almost all kinds of firms--sing1e-product firms,
multiproduct firms, closely held firms, publicly held firms--have
entered into joint ventures. flgxgg;§_§_figggifiigign§ reported that
billions of investment dollars were spent annually through joint
ventures.

Value creation through joint ventures seems to account for the
popularity of joint ventures. Sources of value creation include, among

others, more efficient utilization of resources and skills due to the

15

16

sharing of these resources and skills (e.g., Kogut, 1988; McConnell &
Nantell, 1985); sharing of costs or risks in large or risky projects
(Pate, 1969; Pfeffer & Novak, 1976a); better access to new technology or
special types of skills (Berg & Friedman, 1981; Kogut, 1988); and
improvement in a firm's competitive positioning or a firm's power over
other organizations in both local and global markets (Harrigan, 1985;
Pfeffer & Novak, 1976a).

Because the present study examines the relationship between
joint ventures and firms' market performance, this chapter begins with a
review of literature that has addressed this issue. Then the
theoretical link between relatedness among joint venture participants
and firms' market performance is developed. Research hypotheses that
are in accordance with this theoretical framework are then proposed.

Potential confounding variables also are identified.

te t e ev ew

Earlier work on joint ventures has focused primarily on two
areas: (a) motivations to form joint ventures (e.g., Pfeffer & Novak,
1976a, 1976b; Rogut, 1988) and (b) effects of joint ventures on
industrial concentration or industrial returns (Berg & Friedman, 1981;
Duncan, 1982). Only a few studies have examined the relationship
between joint ventures and firms' performance. These studies come from
the finance and strategy areas.

Studies in finance lend support to the notion of a relationship
between joint ventures and firms' market performance. These studies
report that stockholders of parent firms obtain positive abnormal stock

returns or value gains around the time of a joint venture announcement.

l7

McConnell and Nantell (1985), for example, found that joint venture
announcements lead to significant value gains to stockholders of parent
firms. Their sample includes 136 domestic joint ventures entered into
by 210 companies over the period 1972-1979. Similarly, Ravichandran
(1986) found that, on average, stockholders of 481 firms entering into
domestic joint ventures during the period 1972-1983 earned significant
value gains around the two-day announcement period. Ravichandran (1986)
also found that the reaction of the capital market to joint venture
announcements was not homogeneous across joint ventures. Firms that
entered into joint ventures in concentrated industries earned
statistically significant value gains, whereas firms that entered into
joint ventures in competitive industries did not earn statistically
significant value gains.

Within the strategy area, the findings by Koh and Sinha (1990)
can be contrasted to those of Ravichandran (1986). Koh and Sinha (1990)
examined the effects of 76 joint ventures announced during 1972-1985 in
the ”information technology sector", comprising electrical and
electronic machinery, communication, computers, data processing,
electronic imaging and others. Their findings suggest that the
profitability of joint ventures to parent firms depends on industrial
concentration. That is, when the industry is not highly concentrated, a
joint venture is profitable to the parent firm because the increase in
the parent's output associated with the venture is high compared with
the reduction in industry price. When the industry is highly
concentrated, however, the joint venture may not be profitable to the
parent firm because the increase in the parent's output associated with

the joint venture may be insufficient to compensate for the reduction in

18
price. Rob and Sinha's findings thus contrast those of Ravichandran.

Another variable besides industrial concentration has been used
to explain variations in firms' performance. Harrigan (1985), for
example, examines the effects of parent firms' asymmetries in
relatedness on the performance of 895 ventures in 23 industries during
the years 1974-1985. Relatedness between parent firms in products,
markets, technology, competitive activities and buyer-seller activities,
and relatedness between parent and venture firms in R & D, production,
marketing activities and buyer-seller activities were considered.
Ventures were defined as all business activities in which parent firms
might cooperate, regardless of their ownership forms and status as
separate organizational entities. Harrigan's (1985) results suggest
that ventures were assessed as more successful when the parent firms
were related to their ventures in products, markets and/or technology
than when the parent firms were unrelated to their ventures.

Previous research also compared the relationship between joint
venture and firms' performance to the relationship between other
strategic investments and firms' performance. The study by Woolridge
and Snow (1990) examines differences between stock market reaction to
joint venture announcements and stock market reaction to other strategic
investment announcements, including research and development projects,
major capital expenditures, and diversification into new products and/or
markets. Their sample includes 767 public announcements of strategic
investments by 248 companies engaging in 102 industries. They found
that the stock market favored announcements regarding joint ventures and
R 5 D investment over those in capital expenditures and product/market

diversification. The two-day announcement period cumulative abnormal

19
returns for joint venture and R & D investment were 0.80% and 1.13%,
respectively, compared with the two-day announcement period abnormal
returns for capital expenditure and product/market diversification of
0.36% and 0.69%, respectively.

Two conclusions can be drawn from the foregoing review of joint
venture research in the finance and strategy areas. First, there are
relationships between joint ventures and firms' performance, and,
second, these relationships are not homogeneous across joint ventures.
Two studies (Koh & Sinha, 1990; Ravichandran, 1986) that focus on the
relationship between joint ventures and parent firms' performance argue
that differences in the concentration of the child industry can explain
these different relationships across joint ventures. One study
(Harrigan, 1985), which focuses on the relationship between joint
ventures and child firms' performance, indicates that perhaps
differences in relatedness among joint venture participants can help

explain different relationships across joint ventures.

e e e o W

The present study employs a slightly different approach from
prior studies. This study argues that the relationship between joint
ventures and parent firms' performance can be explained by the
relatedness among joint venture participants. The present study,
therefore, makes use of the relatedness construct, which is similar to
Harrigan's (1985) approach. However, this study applies relatedness to
explain the performance of the parent firms, rather than the performance
of a child firm, the focus of Harrigan's (1985) study. The present

study also argues that the concentration of the child industry is not a

20

major determinant in explaining the relation between joint ventures and
parent firms' performance, and it treats this concentration variable as
a potential confounding variable, a different approach from that used by
Ravichandran (1986) and Rob and Sinha (1990).

The following section first describes in more detail the
concepts of two key variables: relatedness and market performance.
Then propositions regarding the relationships between these two

variables are presented.

82mg

Relatedness is a construct that has generated a large body of
strategy research. Relatedness refers to similarities between
businesses in resources and skills used to manufacture products or to
serve markets. Rumelt (1974), for example, considered businesses to be
related if they: (a) served similar markets using similar distribution
channels, (b) used similar production technologies, or (c) exploited
similar scientific research. Rumelt (1974) determined similarities in
resources used to manufacture products based on: similarities in
products offered; or relationship between products, which occurs when a
product of one business serves as an input for the production of output
of another business. Rumelt (1974) determined similarities in resources
used to serve markets by comparing similarities in markets served by
businesses. Relatedness defined by Rumelt (1974), therefore, has been
determined by product/market relatedness.

Rumelt's (1974) definition of relatedness is compatible with
that suggested by Salter and Weinhold (1979). The latter classify

diversification strategies by key success factors, which are composed of

21
business experience in product markets and skills in functional
activities such as production, marketing, and R & D. An unrelated
diversifier is defined as "a company pursuing growth in product markets
where the key success factors are unrelated to each other" (Salter &
Weinhold, 1979: 7). In contrast, a related diversifier is "a company
that uses its skills in a specific functional activity or product market
as a basis for diversification” (Salter & Weinhold, 1979: 7). Salter
and Weinhold's (1979) definition of relatedness is compatible with that
of Rumelt because their key success factors, which are used to determine
relatedness between businesses, are composed of the same types of
resources and skills that Rumelt (1974) used to determine relatedness
between businesses. These resources and skills include those
specifically used in production or marketing.

This study makes use of relatedness, the conceptual definition
of which follows that of Rumelt (1974) and of Salter and Weinhold
(1979). Relatedness was chosen for use because it provides a
theoretical ground for explaining differences in firms' performance.
Theoretical notions that may explain differences in firms' performance
may be developed based on the consideration of similarities between
businesses in products, markets, or resources and skills used. Another
reason that relatedness was used is that it can, to some extent, explain
differences in the performance of firms involved in diversification
activities such as mergers or acquisitions (e.g., Singh & Montgomery,
1987; Shelton, 1988). Because joint ventures can be considered one type
of diversification activity, it is likely that the relation between
relatedness and performance can be generalized from merger/acquisition

to joint venture studies.

22
BMW

A joint venture consisting of two parents--say parent A (PA) and
parent B (PB)--and a child (C) can be considered to include three types
of relatedness: relatedness between the parents, relatedness between
one parent and the child, and relatedness between the other parent and
the child. If parent A is of focal interest, parent A is involved in
parent-parent relatedness (Pk-P3), parent-child relatedness (Pk-C) and
parents-child relatedness (Pk-C & Ph-C). Parents-child relatedness
involves a comparison between the focal parent-child relatedness (PA-C)
and the other parent-child relatedness (PB-C). Parents-child
relatedness is considered here because it is speculated that parent B-
child relatedness, by itself, may not help the capital market in
determining the value of the joint venture for parent A unless parent B-
child relatedness was compared with parent A-child relatedness.

Each of the three types of relatedness in which a focal firm may
be involved can be placed on a continuum. The determination of the
extent of relatedness is based on product or market similarities. That
is, firms that are more related have a higher proportion of revenue
coming from similar products or markets. Firms that are less related
have a lower proportion of revenue coming from similar products or
markets.

The three types of relatedness can be represented graphically.
Figure 1 displays examples of four joint ventures that differ in the
extent of relatedness. The figure displays parent-parent relatedness
and parent-child relatedness, along the horizontal and vertical axes,
respectively. Relatedness on each axis is a polarity, i.e., has a low

end and a high end. Joint venture (l) is high in two types of

High

Parent-Child
Relatedness

Low

 

23

 

A4 LOW 34 High a1
(:9 H
(4) (1)

High ow High
C 0
Low Hi h
A3 . . 83 I 9 . 32

(3) (2)
Low Low Low Low
0 O
4
LOW Parent-Parent H39"

Relatedness
Figure 1

Examples oi Joint Ventures that Differ in the Extent of Relatedness

24
relatedness--parent-parent relatedness and parent-child relatedness--but
low in differences between the two parents' relatedness to the child,
that is, low in PM-C and Pal-C difference. Joint venture (2) is high in
parent-parent relatedness but low in both parent-child relatedness and
differences between the two parents' relatedness to the child. Joint
venture (3) is low in all three types of relatedness--parent-parent
relatedness, parent-child relatedness and differences between the two
parents' relatedness to the child. While joint venture (1), (2) and (3)
are low in differences between the two parents' relatedness to the
child, joint venture (4) is high in this type of relatedness because the
difference between PM-C and Pm-C is high. The four joint ventures
shown in the figure are examples of joint ventures that may exist.
Other joint ventures may, however, vary along the continuum of the three

types of relatedness.

W

Market performance is concerned with the market valuation of
specific activities of firms. One indicator of market performance is
stockholder return (Singh & Montgomery, 1987). Market performance or
stockholder return was chosen for use in this study because it is a
performance goal that extends beyond the stockholders' domain (Lubatkin
& Shrieves, 1986). High market valuation or high stockholder return can
inhibit proxy battles for control and can enhance a firm's future
effectiveness by allowing less costly access to additional equity and
debt capital (Branch & Gale, 1983; Lubatkin & Shrieves, 1986; Rappaport,
1986). Rappaport (1986) also suggests that management is induced to pay

more attention to stockholder returns because of reasons such as large

25
stock ownership by management; a tie between stockholder return
performance and management compensation; threats of takeover by other
firms, especially if the stock price is undervalued; and management's
concern for its reputation in managing stockholder returns. High

stockholder return is, therefore, central to the focus of management.

W

This study focuses on abnormal stockholder return. Abnormal
stock return is the difference between observed and expected returns,
which can be estimated from the market model (Fama, 1976). The market
model basically states that the return for stock 1 at day t is linearly
related to the return on the market portfolio at day t. The regression
coefficients obtained from the market model are used in the residual
analysis (Fama et a1., 1969), in which abnormal returns are estimated.

According to the market model, the return on a market portfolio
reflects market-wide variations affecting all securities in the market,
and the residual or abnormal return reflects other variations.
Therefore, this study examines abnormal return rather than observed
return because it can reflect capital market reaction to variations
including those resulting from firms' specific activities.

Positive abnormal return indicates value gains to stockholders.
Positive abnormal return for parent firms' stockholders may occur if
joint venture announcements lead to an expectation of an increase in

future income streams for the parent firms.

26
e t F w d s
This section brings together two variables discussed earlier--
relatedness and abnormal stock return. Theoretical arguments for the
expected relationship between these two variables across joint ventures
are developed. Hypotheses about the expected relationship also are

proposed.

We

Joint Venture Announcements and Parent Firms' Abnormal Stock Returns

Previous research found consistent evidence of a positive
relationship between joint venture announcements and parent firms' stock
returns (e.g., McConnell & Nantell, 1985; Woolridge & Snow, 1990).

These studies' interpretation of this positive relationship is that the
capital market perceives joint venture announcements, on average, as
value-creating activities for parent firms. That is, joint ventures
are, on average, expected to generate a future income stream for parent
firms and, therefore, the capital market responds positively to the
joint venture announcements.

The first hypothesis in this study replicates empirical findings
from prior studies. The hypothesized positive relationship between
joint venture announcements and abnormal stock returns provides a check
of data set for this study--whether the sampled firms used in this study
provide the relationship that is consistent with previous findings. The
first hypothesis can be stated as:

H1: Joint venture announcements are positively correlated with
parent firms' abnormal stock returns.

27

W

Parent-Child Relatedness and Parent Firms' Abnormal Stock Returns

This study assumes that the relationships between joint venture
announcements and parent firms' stockholder returns are not homogeneous
across joint ventures. It is expected that relatedness among joint
venture participants affects such relationships. Hypothesis Two deals
with the effects of one type of relatedness--re1atedness between parent
and child--on parent firms' stockholder returns.

This study argues that parent firms that are more related to
their child obtain higher abnormal stock returns than parent firms that
are less related to their child. Higher relatedness in products or
markets between parent and child may lead to greater positive responses
from the capital market than lower relatedness. This is because higher
parent-child relatedness may suggest to the capital market that the
parent can share with its child both general resources and specialized
resources, whereas lower parent-child relatedness may suggest that the
parent can share with its child only general resources. The expected
sharing of both general and specialized resources associated with
parent-child relatedness hence suggests lower average costs and a higher
future income stream for the parent that is more related to the child
than for the parent less related to the child.

Specialized and general resources differ in their specific use
for particular products or markets. Specialized resources may be
defined as resources and skills that are used specifically in production
or marketing. Specialized resources include, for example, (a) those

that are used in production such as production technology, plants, raw

28
materials, machinery and tools (Ansoff, 1965); (b) those that are used
in marketing such as distribution channels, sales force, advertising and
sales promotion (Ansoff, 1965); and (c) R & D or management skills that
are used in specific product or market areas. Conversely, general
resources are defined as resources and skills that can be used across
different functions. General resources, therefore, are not tied
specifically to production or marketing functions. An example of
general resources are financial resources used in hiring accountants or
outside consultants.

The association of high relatedness with utilization of
specialized and general resources and the association of low relatedness
with utilization of general resources have been suggested by several
diversification studies. Singh and Montgomery (1987), for example,
argue that relatedness allows for more efficient utilization of both
specialized and general resources, while unrelatedness allows for more
efficient utilization of general resources only. Singh and Montgomery
(1987) then hypothesized that related diversifiers obtained higher value
gains than unrelated diversifiers, and their findings supported this
hypothesis. Other researchers, such as Wrigley (1970), Lubatkin (1984),
Hill and Hoskisson (1987), and Jones and Hill (1988), also raise similar
arguments. They argue that, whereas related diversifiers can share
their inputs, technology, distribution channels, sales forces, financial
resources and others, unrelated diversifiers may share only their
financial resources. Therefore, a body of diversification literature
has proposed the argument that parallels the argument in this study.

The argument that higher parent-child relatedness enables the

sharing of general and specialized resources/skills, which then leads to

29
higher stock returns for parent firms than lower relatedness does, is

illustrated in Figure 2. The following discussion is based on this

 

 

 

 
  
  

 

 

 

 

 

 

 

 

 

 

 

figure.
Market expectation: -
-Producl/markel-relaled
economlee _.
Specialized -Increased market- \
Higher
Parent-child resources shared specific market power abnormal
m°re related General resources i-Low entry costs stock
shared returns
- Economies through
general resources shared
Figure 2
Higher Parent-Child Relatedness
he Relatedness betwee a e t and 1

Higher relatedness in products/markets between a parent firm and
its child may suggest to the capital market that the parent is more
likely able to share with its child specialized resources and skills
used in production or marketing. The parent can, therefore, lower its
average costs or enjoy economies and hence can obtain higher future
income generated through the joint venture. The parent may share with
its related child the resources used specifically in production such as
production technology, plants or machinery. This sharing may be done at
the same level of production or at different levels of production. This
leads to production economies (Lubatkin, 1983; Kogut, 1988; Hennart,
1988) and scheduling economies (Lubatkin, 1983). Production economies
may occur through economies of scale or economies of scope. Economies

of scale occur when a given bundle of resources is used to expand

30
production of a specific product. Economies of scope occur when a given
bundle of resources is used in a joint production of two or more
products. Scheduling economies occur when there is coordination of
resources used at various stages of production (Shepherd, 1979).
Similarly, sharing with the child specialized resources used in
marketing such as sales force or distribution channels enables the
parent to obtain marketing economies (Lubatkin, 1983).

Higher product/market relatedness also suggests a better
opportunity for parent and child to share specialized skills. High
relatedness suggests that the parent can transfer to the child its R & D
knowledge or management skills that are specifically tied to products or
markets. Kogut (1988), for example, suggests that joint ventures enable
firms to transfer "tacit knowledge" (Polanyi, 1967), which is knowledge
that is not easily diffused across the boundaries of firms. A joint
venture enables the parent to transfer knowledge or experience to the
child that operates in similar product/market areas. Cumulative
knowledge or experience may lead to low learning costs for the parent.

Higher product/market relatedness may also suggest an increase
in market power for the parent. Market power increase may occur because
of relatedness in markets or because of lower costs obtained through
relatedness in products or markets. Market relatedness between parent
and child suggests that the parent may extend its power in currently
engaged markets through the formation of the child. Furthermore, an
increase in market-specific market power may be facilitated if
product/market relatedness enables the parent to lower costs and lower
the prices of its products. The firm's ability to exert more influence

on price by definition means that it gains more power in the market. An

31
ability to control price tends to be positively valued by the capital
market because it suggests an increased certainty that a firm will
obtain future income streams (Oliver, 1990).

The parent that forms a joint venture in related markets also
can lower entry costs. This is because the parent tends to possess
specialized resources or skills that enable the child's penetration into
related markets at minimal costs. Furthermore, possessing specialized
resources/skills necessary for success in related markets also suggests
that the parent can erect entry barriers in this related market
(Hopkins, 1987). This may lead the capital market to expect longer
retention of the parent's profits.

Finally, relatedness in products/markets between the parent and
its child also suggests that the parent can share with the child general
resources such as financial resources and hence can obtain financial
economies (Lubatkin, 1984). Financial economies may occur when the
parent can improve its cash management, allocate more efficiently its
investment capital or reduce the cost of debt capital (Paine & Power,

1984).

MW

A parent firm that is less related to its child may not be able
to utilize resources and skills as efficiently as a parent that is more
related to the child. This leads to the expectation that the former
parent can obtain lower abnormal stock returns through a joint venture
than the latter parent can. This notion is illustrated in Figure 3.

As the figure shows, low relatedness in products and markets

suggests that the parent can share with its child only general

 

32
Market expectation;

 

    
 

  

 

 

 
   
 

 

 

 

 

  

 

 

 

 

Economies through
general resources
Parent-child General resources sharing ket $233“
less related "' shared —" £2353?“ mar stock
+ High investments returns
- High learning costs
- High entry costs
Figure 3

Lower Parent-Child Relatedness

resources, not the specialized resources used in production or
marketing. A sharing of general resources may suggest lowered costs for
the parent due to lower bankruptcy risk or increased administrative
efficiencies. In addition to lowered cost obtained through general
resources sharing, the parent that is less related to its child can
increase its market power. This market power increase may arise, for
example, from the increase in absolute size and breadth of the firm.

Low relatedness between parent and child in products/markets
also suggests to the capital market that the parent may not be able to
share with the child its production technology, distribution channels,
or specific R 8 D strength. This means that the parent firm has to bear
high startup costs, including: (a) costs of investing in assets such as
plants or production technology (Ansoff, 1965); (b) learning costs that
occur because management makes a misallocation of resources during the
learning process of a new business (Salter & Weinhold, 1979); and (c)
entry costs, which include large investments of financial resources in
risky or unrecoverable up-front operational activities such as
advertising, costs of switching buyers from competitors' product to the

child firm's product, costs of persuading distribution.channels to

33
accept the child firm's product through price breaks or cooperative
advertising allowances, costs relating to government restrictions such
as air and water pollution standards and product safety and efficiency
regulations, and other types of costs incurred to enable the child firm
to compete with the cost advantages kept proprietary by established
firms (Porter, 1980). The cost advantages enjoyed by established firms
are related to such factors as proprietary product technology, favorable
access to raw materials, favorable locations, and government subsidies
(Porter, 1980).

Parent firms that are less related to their child firm are
expected to obtain lower abnormal stock returns than those more related
to the child firm because the former have to bear the high costs
associated with low parent-child relatedness. These high costs may not
be offset by lowered costs obtained through sharing general resources.
The capital market would then expect the association of higher abnormal
stock returns with higher parent-child relatedness and the association
of lower abnormal stock returns with lower parent-child relatedness.
This notion is indicated in the second hypothesis:

H2: Parent firms that are more related to their child obtain

higher abnormal stock returns than parent firms that are
less related to their child.

W

Parent-Parent Relatedness and Parent Firms' Abnormal Stock Returns

This study argues that parent firms that are more related to
their partners obtain higher abnormal stock returns than parents that

are less related to their partners. The logic for this argument is

34

presented in Figure 4.

 

Market expectation:
- Access to partner's

 

 

 

 

 

 

arent-parenl specialized resources ' E°°"°""” “£2?“ a
more related """- Alliance that operates—H " Strengthened stock
market position eturns
in the same market ..

 

 

 

 

Figure 4
Higher Parent-Parent Relatedness

Product relatedness between parent firms in a venture may lead
to market expectation that these firms can have access to specialized
resources, such as the production technology of their partners. This
access may enable the firms to improve their production technology and
hence lower their costs. Market relatedness between parent firms means
that these firms operate in the same market and forming an alliance may
help strengthen the market position of both. For instance, the parent
firms may join forces to preempt suppliers or customers from integrating
in a manner unfavorable to them and hence can defend or improve their
current market positions (Harrigan, 1985). This preemption enables the
firms to sustain their future income streams. High product or market
relatedness between parent firms can, therefore, lead to high abnormal
stock returns for the parents. In contrast, for parent firms that are
less related, the capital market may expect that these firms cannot
obtain advantages associated with parent-parent relatedness. Low
relatedness between parents, therefore, leads to low abnormal stock
returns for the parent firms. This is the third hypothesis:

H3: Parent firms that are more related to their partners obtain

higher abnormal stock returns than parent firms that are
less related to their partners.

35
W-

Parents-Child Relatedness and Parent Firms' Abnormal Stock Returns

This study argues that high relatedness between one parent and
child may create high abnormal stock returns not only for that parent,
but also for another parent that is less related to the child. A joint
venture in which parent firms greatly differ in their relatedness with
the child--in which one has high relatedness while the other has low
relatedness--may enable the parent with low child relatedness to obtain
high abnormal stock returns. This is because high relatedness between
one parent and the child helps compensate for the potential
disadvantages associated with the low relatedness between the other
parent and the child. The parent firm that is more related to the child
in products or markets is likely able to provide specialized resources
such as production technology, sales force, distribution channels, and
management skills to the child. The parent firm that is much less
related to the child would then benefit from the high relatedness
between its partner and the child. The parent with low child
relatedness may not have to bear high investment costs because its
partner has already invested some assets in the child's market. The
parent with low child relatedness may not have to bear high learning or
high entry costs because its partner already has some knowledge in the
child's market and already has a toehold in the market. Therefore,
among the parent firms that have low relatedness with the child, the
greater the differences between these parents and their partners in
relatedness with the child, the higher the abnormal stock returns that

these parents may obtain. This is the fourth hypothesis:

36
H4: Among the parent firms that are unrelated to their child,
those whose partners are more related to their child obtain
higher abnormal stock returns than those whose partners are
less related to their child.

The rationale underlying this fourth hypothesis is consistent
with the rationale underlying prior hypotheses. That is, higher parent-
child relatedness leads to higher abnormal stock returns for the parent
than lower parent-child relatedness. In a sense, this last hypothesis
can be thought of as an extension of this rationale to a situation in
which high relatedness between one parent and child can compensate, to
some extent, for low relatedness between the child and the other parent
involved in the same joint venture. High relatedness between one parent

and child can result in high abnormal stock returns to the other parent

that is much less related to the child.

W

The present study proposes four hypotheses. Hypothesis One
fecuses on the relationship between joint venture announcements and
abnormal stock returns for the entire sample of parent firms.
Hypothesis Two focuses on the relationship between parent-child
relatedness and parent firms' abnormal stock returns. Hypothesis Three
focuses on the relationship between parent-parent relatedness and parent
firms' abnormal stock returns. Finally, Hypothesis Four focuses on the
relationship between the differences in the two parents' relatedness to
the child and parent firms' abnormal stock returns. The focus of

Hypothesis Four is on the parent firms that are unrelated to the child.

37
W

The relationship between relatedness among joint venture
participants and parent firms' abnormal stock returns may be affected by
several potential confounding variables. Four potential confounding
variables can be identified from prior research and from logical
reasoning. This study controls for the effects of these variables,
including parent firm size (McConnell & Nantell, 1985), industrial
concentration (Ravichandran, 1986; Rob & Sinha, 1990), potential market
impact, and price-regulated industries (McConnell & Nantell, 1985:

Ravichandran, 1986).

W

Evidence revealing the effects of parent firm size on parent
firms' performance has been mixed. Berg and Friedman (1980), for
example, found that the average size of parent firms, as measured by
average sales, was among other variables that had a positive impact on
firms' accounting rate of return. They explained that this was because
large firms were better able to capitalize the technological and other
contributions of small firms. The effect of parent firm size on firms'
performance was also suggested by Harrigan (1988). Harrigan (1988)
found that the size of parent firms' assets added weak predictive power
to the regression model of ventures' perceived success.

Within finance literature, some evidence exists that small firms
earn significantly higher risk-adjusted returns than large firms (Banz,
1981; Reinganum, 1983). One explanation is that investors who had
limited information were uncertain about the true parameters of the

return distribution of small firms. Therefore, only a subset of

38
investors would hold the securities of small firms and this then led to
the higher risk-adjusted returns of these securities than of securities
considered by all investors (Banz, 1978).

In a study on joint ventures by McConnell and Nantell (1985),
these researchers reported that, around the announcement date, parent
firms of smaller size obtained a higher percentage of abnormal returns
(1.10%) than larger firms (0.63%) while obtaining smaller dollar value
gains. That is, when the rates of return were multiplied to the average
market value of outstanding equity for firms, parent firms of smaller
size generated value gains of $4,537,572 compared with $6,650,544 for
larger parent firms. In contrast to McConnell and Nantell's findings,
Ravichandran (1986) did not find significant effects of firm size on
parent firms' performance. Ravichandran (1986) regressed the value
gains on firm size and joint venture project costs and found that
neither of the coefficients was significantly different from zero.
Because of the problematic role of the parent firm size variable on

firms' performance, the present study controls for this variable.

WW

There has been evidence of the effects of the concentration of
the child industry on parent firms' performance. These effects, as
reported by prior studies (Ravichandran, 1986; Rob & Sinha, 1990), were,
however, in different directions. Ravichandran (1986) found that firms
entering into joint ventures in competitive industries did not earn
significant value gains, whereas firms entering into joint ventures in
concentrated industries earned significant value gains. Ravichandran

(1986) explained that this occurred because the capital market did not

39
believe that parent firms could retain the benefits derived from the
joint ventures in competitive industries for a long period because of
low entry barriers. For concentrated industries, however, the
industries signalled to the capital market the ability of parent firms
to retain benefits derived from the joint ventures.

Ravichandran's (1986) explanations for the effects of industrial
concentration on parent firms are, however, inconsistent and hence
questionable. While his explanation for no value gains for firms
entering into joint ventures in competitive industries focuses on the
firms that are newcomers, his explanation for value gains for firms
entering into joint ventures in concentrated industries seems to focus
more on the firms once they succeed in entering these industries. He
seems to neglect the fact that the child firm is a newcomer and that
unless it has some strengths that can lower entry barriers in the
concentrated industry, it is not likely to enter this industry easily
and may not earn high profits at all. As Yip (1982) points out, market
concentration indicates the extent to which market leaders will react
against entrants and barriers to entry. Therefore, it is likely that
the capital market considers ease in entry of the child firm into a
concentrated industry along with the child's ability to retain profits
generated.

The study by Koh and Sinha (1990) found contrasting evidence
regarding the directions of the effects of industrial concentration on
firms' performance. Koh and Sinha (1990) limited their sample to
horizontal joint ventures in which a joint venture is a new entrant in
the parents' industry. They found that horizontal joint ventures

increased the value gains of parent firms when the industry was not

40
highly concentrated. Their rationale was that, in such industry, the
increase in the output of a parent firm is sufficient to offset a
reduction in industry price associated with an increase in output. When
industrial concentration is high, however, the increase in output of a
parent may not be sufficient to compensate for a reduction in price in
response to any increase in output.

Though these two prior studies by Ravichandran (1986) and by Koh
and Sinha (1990) report opposite directions of the effects of industrial
concentration, the studies agree on the effects of industrial
concentration on parent firms' stockholder returns. This study,
therefore, controls for the effect of the concentration of a child

firm' 8 industry .

W

There is good reason to believe that the capital market reaction
to joint venture announcements may be affected by the market's
expectation of the parent firms' ability to influence the industry
entered by a joint venture. The more influence that the parent can have
on the child industry, the more likely that the child will be a superior
performer in its industry. For instance, the more relevant
resources/skills that the parent can contribute to the child, the more
likely that the child will be a high performer. Higher performance of
the child suggests higher returns to the parent. The capital market
may, therefore, react more positively to joint ventures in which parents
have greater potential to influence the child's industry than those in
which parents have lower potential influence. This study, therefore,

controls for the effects of potential market impact on parent firms'

41

performance.

WW

Prior joint venture research controlled for the effects of three
price-regulated industries on parent firms' performance. McConnell and
Nantell (1985), for example, excluded from their sample joint ventures
for constructing power generating units, drilling and exploring for oil
and natural gas and constructing oil and gas pipelines. Ravichandran
(1986) investigated joint ventures in regulated utility industries
separately. The three price-regulated industries may influence parent
firms' performance because the capital market may believe that joint
ventures in these industries can earn a rate of return equal to the
risk-adjusted cost of capital only (Ravichandran, 1986). Ravichandran
(1986) reports supporting evidence for this explanation. He found that
joint ventures in regulated utility industries led to insignificant
abnormal returns for parent firms. Joint ventures in oil exploration
industries led to abnormal returns that were, though not statistically
significant, lower than comparison period returns derived from
comparison period methodology (Masulis, 1980; Dann, 1981; Mikkelson,
1981). Because earlier studies found consistent evidence for the
effects of three price-regulated industries, including electric utility,
oil and gas exploration and oil and gas transmission, this study
controls for this variable.

Even though McConnell and Nantell (1985) and Ravichandran (1986)
offer a rationale for the control of effects of three price-regulated
industries, it should not be generalized that other regulated industries

may affect parent firms' performance. In other words, the regulation of

42

an industry may be a necessary but inadequate factor to explain
differences in parent firms' performance in other industries.
Ravichandran (1986), for example, examined joint ventures during the
period 1972-1983. During this period--i.e., in l978--a deregulation of
the oil and gas exploration industry occurred. Ravichandran's (1986)
finding of lower abnormal returns for the oil and gas exploration
industry during the announcement period than during the comparison
period covers both the regulated and deregulated periods of this
industry. Ravichandran's (1986) findings then cannot directly indicate
that the negative market response to the joint ventures in the oil and
gas exploration industry occurred because of the regulation effects.
Furthermore, Ravichandran (1986) also reports that joint ventures in
some industries, such as communication industries before they were
deregulated, did not lead to negative responses from the capital market.

A control for the effects of three regulated industries was
performed in this study not solely because these were regulated
industries, but because of the types of industries. These three
industries may have had past negative average abnormal returns and,
therefore, the capital market did not respond positively to joint
ventures in these industries. Clarke (1980), for example, found that
the utility industries had negative abnormal returns during the period
of examination. However, because the focus of this study is not to
examine the effects of different types of industries on firms, this
study controls for the effects of these three industries by forming a

subsample of joint ventures in these industries.

CHAPTER THREE

METHODOLOGY

The purpose of this study is to investigate the relationship
between joint ventures and parent firms' market performance. More
specifically, the study focuses on the relation between relatedness
among joint venture participants and parent firms' stock returns. This
chapter defines and operationalizes key variables, describes the
sampling procedure, and explains the statistical methods for hypotheses

testing.

mm

W

Relatedness among joint venture participants is an independent
variable for this study. Relatedness (Rumelt, 1974; Salter & Weinhold,
1979) pertains to the relationships between businesses through products
or markets. Businesses are considered related if they offer similar
products, serve similar markets or use similar resources or skills.
Businesses are also considered related if they are engaged in sequential
stages of product processing activities. On the contrary, businesses
are considered unrelated if they have no relationship through products
and markets.

This study determines relatedness based on the concept proposed

43

44
by Rumelt (1974) and by Salter and Weinhold (1979). Relatedness was
determined based on similarities between business segments in products
or markets. Participating firms' businesses were, therefore, considered
related if they met any of the following criteria: (a) offered similar
products, (b) served similar markets, or (c) engaged in different stages
of product processing activities. Participating firms' businesses were
considered unrelated if they did not meet any of these criteria.

Product similarity or product relatedness was determined by
comparing business segments of participating firms. A business segment
is a group of related products that accounts for some portion of the
consolidated revenue of a firm (Amit & Livnat, 1988). For example, a
building material segment consists of groups of related products such as
plywood, lumber, logs, veneers and adhesive. A paper segment may
include related products such as fine paper and milk cartons.
Participating firms' products that belong to the same business segment
are considered related.

Several approaches can be used to determine market similarity or
market relatedness, but perhaps the most fundamental one is through
customer types: consumer vs. industrial markets (Risley, 1972; Stanton,
1981; Udell, 1972). Webster (1978), for example, highlights four
dimensions across which consumer and industrial marketing differ:
functional interdependence, product complexity, buyer-seller
interdependence and buying process complexity. Webster (1978),
comparing consumer and industrial marketing, argues that (a) industrial
marketing success depends more on other functions in the organization;
(b) industrial products are more complex because they involve a variety

of economic, technical and personal relationships between buyer and

45
seller; (c) industrial marketing has a high degree of buyer-seller
dependence that goes beyond the transaction; and (d) industrial
marketing is a more complex buying process because of the firm's
environment.

Empirical evidence exists to support the notion that industrial
and consumer markets do differ. Day, MacMillan and Hambrick (1983a),
for example, reported that consumer buyers had more emotional needs,
were less knowledgeable, were greater in number and were more fragmented
than industrial buyers, and that consumer marketers faced greater short-
term uncertainty. Profiles of firms selling to industrial and consumer
markets also were found to be different across several strategic
dimensions (Day, MacMillan & Hambrick, 1983b). For example, consumer
businesses had a higher fraction of sales represented by advertising and
promotional expenses, purchases and finished goods inventories, whereas
industrial businesses had a higher fraction represented by manufacturing
costs, value added, in-process inventories, receivables, and R & D
expenditures.

Industrial markets may be further subcategorized according to
the following range of industries (Stanton, 1984): (a) agriculture,
forestry and fishing (SIC 01 to 09); (b) mining and quarrying (SIC 10 to
14); (c) construction (SIC 15 to 19); (d) manufacturing (SIC 20 to 39);
(e) transportation, communication and other public utilities (SIC 40 to
49); (f) wholesale and retail trade (SIC 50 to 59); (g) finance,
insurance and real estate (SIC 70 to 89); (h) services (SIC 70 to 89);
and (1) government (SIC 90 to 93). These nine groups of industries are

consistent with nine major divisions of industries listed in the

WW (1987) prepared by the

46

Office of Management and Budget. The two-digit SIC codes for each
division of industries appear in the parentheses after each division.

Business segments of firms would be considered related in
industrial markets only when they shared the same subcategories of these
markets. This criterion was used because consumers in different
subcategories of industrial markets had been shown to be different.
Government markets, for example, are different in procurement processes
from those in the private sector of the industrial markets (Stanton,
1984). The categories and subcategories of markets considered in this
study are shown in Table 1. Table 1 includes two market categories:
consumer and industrial. The industrial market is categorized into nine
groups.

In addition to relatedness based on product or market
similarity, another criterion for determining relatedness between
businesses is that the businesses are engaged in sequential stages of
product processing activities. In other words, businesses are
considered related if the product of one business serves as an input for
the product processing of another business. For example, if firm A's
business is mining bauxite and firm's B business is fabricating
aluminum, these two firms' businesses would be considered related. The
bauxite product of firm A can serve as an input for processing
aluminum, the product of firm B.

One approach to determine whether the product of one business
can be used as an input for the production process of another business
is to use customer types or markets as an indicator. When the customers
or the users of the product of one business are in the same business as

the operator of another business being compared, these two businesses

47

Table 1

Market Categories and subcategories

Consumer market

Industrial market:

(1)
(2)
(3)
(4)

(5)

(6)
(7)
(8)
(9)

agriculture, forestry and fishing;

mining, oil and gas extraction;

construction;

manufacturing: food products, tobacco products,
textile mill products, apparel products, lumber
and wood products, furniture and fixtures, paper
and allied products, printing and publishing
products, chemicals and allied products,
petroleum refining and related industries,
rubber and plastic products, leather and leather
products, stone, clay, glass and concrete
products, primary metal industries, fabricated
metal products, industrial machinery and
computer equipment, electronic and other
electrical equipment, transportation equipment,
measuring and controlling instruments,
miscellaneous manufacturing industries;
transportation, communication, electric, gas and
sanitary services;

wholesale and retail trade;

finance, insurance and real estate;

services;

government

48
are considered related. For instance, if the products of two businesses
being compared are metallurgical coal and steel, respectively, and the
coal buyers are steel producers, these two businesses are related
because the metallurgical coal is an input for the steel production
process.

To summarize, joint venture participants may be related through
either products or markets. This section proposes the criteria that was
used to determine product or market relatedness. This determination was
performed in the development of a measure of relatedness among joint

venture participants.

Wee
Firms' market-based performance is a dependent variable for this

study. Firms' market-based performance can be defined as the

performance of firms as valued by the capital market.

lament

i-v-V..!.-, . :Igegge . ;~.7;_ -.,-.~.— :fliiOs' s ,7 V-, - 'g 0:! 5

One measure of relatedness was developed by Wrigley (1970) and
later elaborated by Rumelt (1974). Rumelt's (1974) determination of
relatedness between diversified and original businesses of a firm was
made mainly through consideration of the proportion of firms' revenues
generated by business segments that were considered related. The
categorical measure of relatedness was then developed based on this
information. This measure of Rumelt (1974) has been commended for its
firm specificity (Montgomery, 1982). That is, the measure conveys

specific information about firms' similar products or markets and the

49
revenue generated by these product/market areas. Even though the
measure was criticized for its dependence on subjective judgment
(Palepu, 1985), research suggests that the interrater reliability and
convergence of this measure with other diversification measures are both
high and acceptable (e.g., Montgomery, 1982).

Whereas a measure of relatedness developed by Rumelt (1974) is
categorical, a measure of ”strategic fit", or relatedness between
acquiring and acquired firms developed by Shelton (1988), is continuous.
Shelton (1988) subjectively determines the relatedness between
businesses through consideration of similarities between products or
markets for each pair of the firms' business segments. She then derives
a continuous measure of relatedness, based on information on the
proportion of revenue generated by related business segments of the
firms. This continuous measure has strength in that there is no
information loss such as that which occurs when using categorical
measures (Cohen & Cohen, 1983).

The present study develops a continuous measure of relatedness.
The procedure for developing the measure follows closely the procedure
used by Shelton (1988). The derivation of this measure, like those of
Rumelt (1974) and Shelton (1988), depends mainly on subjective judgment.
Information on product/market similarities between businesses is used in
the determination of relatedness. This measure of relatedness,
therefore, has strengths in conveying specific information about firms'
products or markets and in utilizing all available data.

The procedure used to develop a continuous measure of
relatedness between each pair of joint venture participants, such as a

pair made up of two parents or a pair of one parent and one child, can

50

be described as follows. First, information on products, markets and
sales breakdown for various business segments of a pair of firms was
collected. Then the collected information on products or markets was
used in the determination of relatedness between each pair of business
segments of these two firms. At this stage, the determination was based
on a dichotomy of relatedness, i.e., related or unrelated. The criteria
for determining relatedness or unrelatedness between business segments
of firms are those proposed earlier in this chapter. Next, sales
revenue generated by each business segment of one firm was multiplied by
sales revenue generated by each business segment of the other firm.
Finally, the products of multiplied revenue for the pairs of related
business segments were added up. The result of this addition is the
continuous measure of relatedness between the two firms of interest.

The measure of relatedness between any pair of firms ranges from
0 to 100%, with 0% of relatedness between any pair of firms indicating
that the firms are unrelated because they have no similar product or
market areas. None of the revenue of these two firms comes from similar
products or markets. On the other hand, 100% of relatedness between any
pair of firms indicates that the firms are totally related because they
are involved in similar products or markets and total sales revenues of
these two firms come from similar products or markets. A measure of
relatedness between different pairs of firms may also vary between these
two extremes, i.e., between 0 and 100%. Firms that have higher levels
of relatedness measure can be interpreted as having greater relatedness
than firms that have lower levels of relatedness measure.

As mentioned earlier, the above procedure to develop the measure

of relatedness could be applied to any pair of firms, such as a pair

51
made up of two parents or a pair of one parent and one child. The
following example describes how the procedure can be followed in the
development of a continuous measure of relatedness between a pair of
parent firms. The two parent firms in the example, American Can Co. and
Pope & Talbot, Inc., first announced their joint venture in the ngl
Street Jogrgal on 3/29/77. To determine the relatedness between these
two parents, the information about their business segments--products
and/or markets and sales revenue--was collected. Table 2 displays the
collected information about the business segments of these two parent
firms. The table consists of two panels: The upper panel contains
information about business segments of American Can Co.; the lower panel
contains the information about business segments of Pope & Talbot, Inc..
The business segments of each parent were arrayed along the column
headings. Then the information on sales breakdown and on products and
markets for each business segment was presented. Other available
information, such as the contributions of the parent firm to the child,
was included at the bottom of each panel.

The collected information about these two parents was then used
to determine relatedness between each pair of business segments of the
parent firms. Outcome of the determination is a measure of relatedness
between two parents, as displayed in Table 3.

Table 3 presents the business segments of Pope & Talbot, Inc.
and of American Can Co. along the horizontal axis and vertical axis,
respectively. The percentages of sales revenues generated by each
business segment of the firms appears in the parentheses. P0pe &
Talbot, Inc. had two business segments: lumber and wood, and land

development, and these two segments generated 96% and 4%, respectively,

52

Table 2

Information about Business Segments of Parent Firms

Parent Firm (1): American Can Co.

Businesses Container & packaging
Products

Sales

breakdown 55% + 15%

Product Metal cans and compo-

examples site containers, plas-
tic tubes and bottles

Markets Industrial: brewing,

and food

Paper products Chemical products

& others
20% 10%

Disposable cups Specialty
and plates, facial chemicals
tissues, paper
napkin
Industrial and Industrial:
household users pharmaceutical,

cosmetic, textile
and plastic

Contributions to the child: American Can Co. manages the entire pulp and

paper complex

Businesses Lumber & wood products Land development
Sales

breakdown 96% 4%

Product Lumber, hardboard, Plywood

examples and veneer

Markets Industrial: Paper products Consumers

Contributions to the child: Pope & Talbot, Inc. supplies long-term wood

fiber requirements to the child.

53

Table 3
Measure oi Relatedness between Two Parents

Pope a Talbot. inc.

 

 

 

 

 

Lumber and wood E Land development
(96%) : (4%)

Packaging Unrelated E Unrelated

(70-1.) (707“ 96%) E (70"X4'5)

American Paper Related 1 Unrelated
Can Go. products 1

(20*) (20V 85%) : (20V4%)

SPOC'E'W Unrelated : Unrelated

ehem'fl'! (1095x9615) i (10%x4'b)
(10%) :

 

 

54
of the firm's total sales revenues. American Can Co. had three business
segments: packaging, paper products, and specialty chemicals, and these
three segments accounted for 70%, 20% and 10%, respectively, of the
firm's sales revenues. The relatedness of each pair of business
segments of these two firms was then considered, based on the proposed
criteria regarding product or market similarities. A pair of segments
consisting of Pope & Talbot's land development segment and American
Can's packaging segment, for example, was considered unrelated because
these two segments were related neither through products nor markets.
0n the other hand, Pope & Talbot's lumber and wood segment and American
Can's paper products segment were considered related because wood fiber
is used as a raw material for the pulp and paper product segment.
Finally, for the business segments that were considered related, the
products of multiplied revenues for these segments were added up.
Because only the lumber and wood segment of Pope & Talbot and the paper
product segment of American Can are related, the relatedness between
these two firms is equal to the products of multiplied revenue generated
by these two segments, or 19.2%.

A measure of the relatedness between a parent and a child can be
developed following the same procedure. In the example of a joint
venture between American Can Co. & POpe & Talbot, Inc., the child was
established to own and operate American Can's pulp mill in Oregon.
Because this pulp mill provides raw materials for the paper products
segment of American Can Co. and this business segment generates 20% of
sales revenues for the company, the relatedness measure between American
Can Co. and the child is hence 20% multiplied by 100%, or 20%. The

paper mill that the child would operate would obtain wood fiber from the

55
other parent, Pope & Talbot, Inc. Because the lumber and wood products
segment of Pope & Talbot accounts for 96% of the company's total sales,
the relatedness measure between Pope & Talbot, Inc., and the child firm

is 96% multiplied by 100%, or 96%.

y-a‘. - . ;-_: "!‘S in‘;‘ . 9 V‘! ,e 'a -a_ ' “am:

To sum up, the measure of relatedness between two parent firms
or between a parent and a child firm is a continuous measure. This
relatedness measure's strength is its focus on firms' specific aspects,
such as products and markets for each business segment and the
percentage of revenue generated by these business segments. A summation
of the percentages of revenue generated from the firms' related business

segments is a measure of relatedness between the firms.

hammer:

Within strategy and finance literature, one popular indicator of
a firm's market-based performance is stock returns (Lubatkin & Shrieves,
1986). Stockholder returns have been used to indicate the effects of
events such as merger announcements (e.g., Mandelker, 1974; Chatterjee,
1986; Singh & Montgomery, 1987) and joint venture announcements
(McConnell & Nantell, 1986; Ravichandran, 1986). Because this study
examines the effects of joint venture announcements on the capital
market, stockholder returns are thus a relevant measure of firms'

performance.

56

MW

Parent firm size, one potential confounding variable, was
measured in this study by the market value of a firm's outstanding stock
at year end prior to the year of the joint venture announcement. This
measure was selected because of its relevance to the study's purpose of
examining the effects of joint ventures on stockholder returns. This
measure also is consistent with the measure used in several prior

studies (e.g., McConnell & Nantell, 1985).

MW

Industrial concentration, another potential confounding
variable, is represented by the percentage of total industry sales
contributed by the largest few firms, ranked in order of market share
(Scherer, 1980). The most commonly used indicator of industrial
concentration is perhaps the four-firm concentration ratio. The four-
firm concentration ratio is the percentage of total industry sales
originated by the leading four firms. The ratios, which are published
in the Census 9f Mgnufgctgrgs, are prepared by the U.S. Department of
Commerce, Bureau of the Census. Because a number of studies, including
joint venture studies, have used the four-firm concentration ratio, this
study used the same ratio to indicate the concentration level of a child

industry.

MW
Potential market impact is another potential confounding

variable. One indicator of the impact that the parent firm may have on

the child industry is the size of the parent firm relative to the size

57
of the child industry. The size of the child industry has been
frequently represented by industry sales, so, to be compared on the same
basis, the parent firm size is represented by its sales revenues. The
potential market impact of parent firms in a venture is, therefore,
measured by the proportion of the combined sales of the parent firms to
the sales of the child industry. Potential market impact that the joint
venture between Du Pont Co. and National Distiller & Chemical Co.
announced 10/08/75, for example, may have on synthetic gas industry, the
child industry, is equal to the combined sales of Du Pont Co. and
National Distiller & Chemical Co. divided by the sales of the child
industry. The sales revenue of Du Pont Co., of National Distiller &
Chemical Co., and of the child industry for the year 1974 were $6910.1,
$1430.l, and $679.3 million, respectively. The value of potential

market impact is, therefore, 12.28 or $(6910.10+l430.1)/679.3.

We.

The information necessary for the determination of relatedness
among joint venture participants, the independent variable, was obtained
from several sources. These sources included Mggdy's Indusrrial Manual,
Compustat tape, 10-K reports, annual reports, and joint venture
announcements in the flgll_§rrggr_lggrn§1. These sources supplement each
other in providing information about firms' products, markets and sales
revenues. For each joint venture, information on parent firms'
products, markets and sales revenues from Mggng§_1ndu§§rial_Manual or
from annual reports one year prior to the year of joint venture
announcement was used. This is likely to be the most recent information

that the capital market could use to make a judgment about relatedness

58
between parent firms. Information about the child firm's business came
mainly from the joint venture announcement in the Egll Srreet Journal.
The data on parent firms' observed stock returns, which were
used to derive abnormal stock returns, the dependent variable, were
obtained from the Center for Research in Security Prices (CRSP) daily

tapes. These tapes are also maintained at Michigan State University.

W211

An initial sample of joint ventures was composed of joint
ventures that were first publicly announced in the Hgll_§rrg§r_iggrggl
over the period beginning January 1975 and ending December 1987. The
announcements in the Wall §rregr Jgurggl were used as a source for
primary sample selection for several reasons. One reason was that both
the New York Stock Exchange (NYSE) and the American Stock Exchange (ASE)
require member firms to transmit "any release of information that could
reasonably be expected to have impact on the market for the company's
securities” to Dow Jones & Company, and, presumably, most of these
releases subsequently appear in the Hall_§£reet_leurnal. a Dow Jones
publication. Another reason was that the Wall Street Journal provides
daily information so that the date of the announcement event could be
specified precisely. Furthermore, the Wall Street Journal has long
served as a source for primary sample selection for research in the
finance area (Thompson, Olsen & Dietrich, 1987).

The index of joint venture announcements in the Egll Streer

Jggrngl appears in two sources, i.e., the joint venture section of the

Special tabulation column in W
Qgrngrgrg_§h§nge§ and the joint venture section in Wall Srrggr Jgurnal

59
IDQEK- These two sources of index on joint venture announcements,
though overlapping to some extent, were not identical and were used to
complement each other. The Ereeleeere Ennk Q SQQEE lneex ef Corporate
Chengee abstracts articles from over 750 financial publications and
trade magazines, including the Hell.§£rse£_lesrnal. and thus provides a
more comprehensive index of joint venture announcements than the Well
W. The W. on the other hand.
provides more descriptive information about the joint venture
announcements than the Eredleeere EEDE fi Seort Inger ef Cornerate
W-

A thirteen-year period, from 1975 to 1987, was chosen to
increase the probability that statistical power can be achieved. The
statistical power, as expressed by l-fi, involves the probability of
avoiding type II errors. Type II errors or 3 occurs when the null
hypothesis is not rejected, even though it is false. A thirteen-year
period increases the probability that the sample size approaches a
desirable level and statistical power can be achieved. Furthermore, the
period from 1975 to 1987 covers relatively recent announcements of joint
ventures and choosing that period enables examination of recent
responses of the capital market to such announcements.

To be included in the final sample, a joint venture has to meet
several criteria. First, each joint venture must have only two parent
firms at the time of the announcement. This criterion was specified to
control for variations in the number of parent firms that were involved
in a joint venture. Second, both parent firms in each joint venture
must be U.S. corporations. This criterion helps avoid possible

contaminating effects of international diversification associated with

60
international joint ventures. Third, information from the Well Street
Jenrnnl_lngex for a 40-day period surrounding each joint venture
announcement--i.e., 20 days before and 20 days after the announcement
date--contains no other joint venture or acquisition announcements that
can have impact on the stock price. This criterion was specified to
lessen the effects of these events on parent firms' stock prices.
Finally, the announcements of joint ventures in the Wall Street Jeurnal
have not been preceded by other publications. This criterion helps
ensure that the joint venture announcements in the Well_Srreer_lenrnel
triggered the initial responses of the capital market to the
announcements.

From each sampled joint venture, only one parent firm was
selected. These selected firms became the sampled firms in this study.
The rationale for selecting only one parent firm from a sampled joint
venture is to avoid a potential problem of violating the assumption of
cross-sectional independence of returns. This assumption may be
violated if the stockholder returns of parent firms engaging in the same
joint venture are simultaneously considered. The criteria for selecting
one parent from each sampled joint venture are that: the firm has to be
traded on the NYSE or the ASE, and the firm's stock return data must be
available on the CRSP tape for a period of 100 days prior to the joint
venture announcement and 20 days following the announcement. For a
joint venture consisting of two parent firms, both of whose stock return
data are available on the CRSP tape, one parent firm would be randomly
selected from the joint venture.

The screening procedure yielded a sample of 209 announced joint

ventures entered by 209 firms during the period 1975-1987. This study

61
does not claim that the sampled joint ventures are representative of the
total joint venture population in the United States, which may be
considered to be composed of all joint ventures formed in the United
States by all domestic parents, by all foreign parents, or by domestic
and foreign parents in all types of industries and over a number of
years. The sampled joint ventures are, however, considered
representative of the target population, which is composed of domestic
joint ventures consisting of two parents. The sampled joint ventures
can well represent the target population because the sample size is

sufficiently large and the joint ventures formed in all types of

industries over a number of years are included in the sample.

 

The primary methodology used to derive abnormal stock returns,
the dependent variable, was a residual analysis based on the market
model (Fama, 1976). The technique was pioneered by Fama et a1. (1969)
and was described in Fama (1976) and Brown and Warner (1980). The
market model predicts each security's return, given the market return,
which is a control for exogenous events, and given the firm's historical
relationship to the market. The regression equation of the market model

may be written as:

Rn. ' at + fliRmt + ‘it. (1)
where R“,- individual firm's daily return for stock 1 on day t
finn" market portfolio's daily return or value-weighted
aggregation of all risky securities available for
investment

a1 - intercept term for stock 1

62
B1 .. historical market relationship for stock i
e1,- disturbance term in the model for security i at time t
- this is normally distributed with mean 0 and
variance oz, i.e., en ._ N (0,02).

According to the market model, each security's period t return

is expressed as a linear function of the contemporaneous market return

6R“) plus an error term. The model controls for market-wide variations
through market return, Rut. .Any variations due to factors not present
in the market portfolio will be captured in the disturbance term, 51e-
The regression coefficients from the market model in equation (1) were
estimated over a maximum of 100 and a minimum of 20 trading days prior
to each joint venture announcement date, represented by day 0. The
relative days, defined as the days relative to the announcement date,
that were included in the analysis ran from day ~100 to day +20.

The regression coefficients in equation (1), which were
estimated during day -100 to day -21, were used in a prediction of Run
with resulting prediction errors or residuals for the test period,
running from day -20 to day +20. The prediction error, which indicates

the abnormal return to stock 1 at time t, can be expressed as:

Q

ARID - Rita ' (a1 + BiRmt) t - -20...+20 (2)
where ARtb - abnormal returns or residuals for stock 1 at day t
at, 51" estimated values of coefficient in equation (1)

R1,, & Rnt- observed returns for stock 1 and market portfolio,
respectively, at day t relative to the announcement
date (e.g., t - -20...+20).

The abnormal return was calculated for each day during the test

period. Daily returns were used, rather than monthly returns, because

of the added precision in measuring the effects of joint venture

63
announcement on stock returns. The use of daily data permits
researchers to take advantage of information about the specific day on
which an announcement takes place (Brown & Warner, 1980). Moreover, the
use of daily returns minimizes the likelihood that the returns would be
influenced by events that are not related to the announcement event

(Jensen & Ruback, 1983).

0- v: 0! . 5910-! ‘M‘lt '- od i' aqua t- as . - tir- am-le

To test Hypothesis One, the announcement period abnormal stock
returns for the entire sample of firms have to be computed. Hypothesis
One states that there is a positive relationship between joint venture
announcements and parent firms' abnormal stock returns. This hypothesis
would be rejected if the announcement period abnormal return for the
portfolio of sampled firms was significantly different from zero. The
announcement period covers two days, i.e., one day prior to the joint
venture announcement date in the Well Srreet Jonrnal (day -l) and the
announcement date (day 0) .

The two-day announcement period, which has been popularly used
in the finance literature, was chosen because there is no reason to
believe that the accurate date is in fact day -l for some sample firms
(Dodd, 1980). Some of these reported announcements were released to the
press during trading hours the day before, and it was expected that any
information in the announcements was reflected in the market prices of
that day (day -l). Other announcements were, however, released after
the close of trading on the day before publication in the Well Streer
Jenrnel, and any information in these announcements was reflected in the

market prices of day 0. Therefore, the market reaction to the joint

64
venture announcement seems to be more accurately represented by the
announcement period cumulative abnormal returns over days -1 and 0.
The two-day cumulative average abnormal return (CAR) for the
entire sample of firms can be derived from an aggregation of abnormal
returns of all of the firms to a portfolio over days -1 and day 0. The

cumulative average abnormal return can be expressed as:

CAR - Z AH;t (3)
to
where ARpg- average abnormal returns for the portfolio at day t
to - first relative day in the CAR calculation (day -l)
T - days through which the CAR is calculated (day 0).

The average abnormal returns for the portfolio at day t were obtained by

taking an equal-weighted portfolio:

Npt
N 3'1
pt
where ARpg- average abnormal returns for the portfolio at day t
ARjg- average abnormal returns for security j at day t
Npt - number of securities on the portfolio.

A one-tailed one-sample t test was used to test the significance
of the announcement period abnormal return. Under the assumption that
the capital market reaction to new information is immediate and
unbiased, the significantly positive value of announcement period return
for the portfolio of sample firms can be interpreted as the significance

of positive market reaction to the joint venture announcements.

 

C a o l s s

The interrelationships of all variables included in the present
study can be analyzed using a correlation analysis. These variables
include: announcement period abnormal returns, parent-child
relatedness, parent-parent relatedness, industrial concentration,
potential market impact, and firm size. The Pearson correlation
coefficient, a statistical summary of the degree and direction of
relationship between two variables, can be derived from a correlation
analysis. The minimum and maximum values of the correlation coefficient
are 0 and 1.0, respectively. The closer the absolute value of
correlation coefficients to 1.0, the greater the relationships between
the variables. These relationships may be either positive or negative.
A positive value suggests that the variables change in the same
direction, whereas a negative value suggests that the variables change
in different directions.

The relationships between independent and dependent variables
were further analyzed through the regression method. The regression
analysis is a data analytic system that yields measures of the magnitude
of the relationship of an independent variable to the dependent variable
(Cohen & Cohen, 1983). The regression analysis was chosen because it
has built-in effect size measures that are unit-free and easily
understood and communicated (Cohen & Cohen, 1983). The measures of
effect size of the magnitude of the phenomenon being studied include,
for example, regression coefficients and proportion of variance. These

effect size measures come with their significance test value and,

66
therefore, no confusion between the two issues of "whether" and "how
much" need arise (Cohen & Cohen, 1983). That is, the measures of effect
size focus on how much relationship the variables have, and the
significance test focuses on whether the relationships between variables

are statistically significant.

Linear_Bsszs§eien_fleth2d

Because a linear regression method focuses on testing the
relationship between one independent and one dependent variable, this
method was used to test Hypothesis Two, Hypothesis Three and Hypothesis
Four. Hypothesis Two focuses on the relationship between parent-child
relatedness and parent firms' abnormal stock returns. The regression

equation that can be used to test Hypothesis Two can be expressed as:

AR - a + b1 PCR (5)
where AR - abnormal stock returns to a sample firm
a - intercept of a regression line
b1 ,. regression coefficient for relatedness between parent

and child (PCR).
Hypothesis Two would be supported if b1 was significantly positive,
suggesting a positive relationship between parent-child relatedness and
parent firms' abnormal stock returns.

Hypothesis Three focuses on the relationship between parent-
parent relatedness and parent firms' abnormal stock returns. The
regression equation used to test Hypothesis Three can be expressed as:

AR - a + b2 PPR (6)
where AR - abnormal stock returns to a sample firm

a - intercept of a regression line

67

b2 - regression coefficient for relatedness between parents
(PPR).

Hypothesis Three would be supported if b2 was significantly positive,
suggesting a positive relationship between parent-parent relatedness and
parent firms' abnormal stock returns.

Hypothesis Four focuses on the relationship between parents-
child relatedness and abnormal returns to a group of parent firms
unrelated to their child firm. A subsample of parent firms that have
zero value of relatedness with their child firm was formed. Differences
between the firms in the subsample and their partner in the venture in
relatedness with the child were then calculated. The regression

equation that was used to test Hypothesis Four can then be expressed as:

ARL - a + bu DPC (7)
where ARL'- abnormal stock returns to a firm in the subsample
a - intercept of a regression line
bLl .. regression coefficient for differences between parents

in their relatedness with the child (DPC).
DPC represents the difference between two parent firms in the same joint
venture in their relatedness with the child. A high value for DPC
suggests that parent firms in the same venture greatly differ in their
relatedness with the child. On the contrary, a low value for DPC
suggests that parents in the same joint venture do not differ much in
their relatedness with the child. Hypothesis Four would be supported if
bu was significantly positive. Positive bu suggests that the higher
the differences in parent-child relatedness between parent firms in the
same joint venture, the higher the abnormal stock returns the parent

firms in the subsample would obtain.

68

W

A multiple regression analysis computes the relationships of one
dependent variable with more than one independent variable. This method
was used to control for potential confounding variables, which include
parent firm size, industrial concentration, potential market impact and
specific price-regulated industries. The procedure was that these
potential confounding variables were entered into the regression
equation, and then the relatedness variables were entered. Potential
confounding variables were entered into the equation before relatedness
variables so that any of their relationships with the abnormal stock
returns can be identified easily. The order of entry of these potential
confounding variables was not prespecified because there was no strong
theoretical basis that could be used to determine the relative
importance of these variables in explaining variance in abnormal stock
returns. Various orders of entry were attempted and differences in
results were examined. The multiple regression equation can then be

expressed as:

where AR - abnormal stock returns to a sample firm

a - intercept of a regression line

b1 - regression coefficient for potential market impact
(MKTIMP)

b2 - regression coefficient for industrial concentration
(CONC)

b3 - regression coefficient for parent firm size (SIZE)

b, - regression coefficient for parent-child relatedness
(PCR)

b5 - regression coefficient for parent-parent relatedness

(PPR)

69
The confounding effects of potential market impact, industrial
concentration and parent firm size would be indicated by the significant
‘values of b1, b2 and b3, respectively.

Finally, the effects of three price-regulated industries,
including electric utility, oil and gas exploration, and oil and gas
pipeline transmission, were controlled. This was performed by dividing
the entire sample of firms into two groups, i.e., regulated and non-
regulated groups. Parent firms in the regulated group are those that
participated in joint ventures in the three regulated industries.

Parent firms in the non-regulated group, on the other hand, participated
in joint ventures in other industries besides these three regulated
industries. The announcement period abnormal returns to the firms in
each group were regressed on the relatedness and the potential
confounding variables. The results obtained from the regression

equations were then compared.

CHAPTER FOUR

EMPIRICAL RESULTS

The present study hypothesizes that there is a positive
relationship between relatedness among joint venture participants and
parent firms' abnormal stock returns. This chapter presents the results
of an empirical investigation of the hypotheses proposed in chapter
three. The chapter first describes the characteristics of the sample

firms and then presents the statistical results of hypotheses testing.

5.213212

Wise

The screening procedure described earlier in Chapter Three
yielded a sample of 209 firms that entered into 209 joint ventures
announced during the period 1975-1987. The number of firms and of joint
ventures are identical because only one parent firm that met the
criteria specified in Chapter Three was randomly selected from each
joint venture. The distributions of the sample of joint ventures by
year and by industry type are shown in Table 4.

Table 4 consists of two panels. Panel A tabulates the joint
ventures by year of first announcement in the Well Srreer Journal.
During the thirteen-year period of study--1975-1987--the distribution of

the joint ventures is skewed toward more recent years than earlier

70

71
Table 4
Joint Venture Sample

A. Distribution of Joint Ventures by Year

YeaLszLEirat __r_fNumbe o
Announeenent Jolnr Ventures
1975 13
1976 8
1977 10
1978 5
1979 4
1980 6
1981 7
1982 20
1983 20
1984 29
1985 17
1986 31
1987 12
209

B. Distribution of Joint Ventures by Industry

1W1 Mine
gelnr Venrures
Communications (48) 33
Oil & gas extraction (13) 24
Electric & gas services (49) ‘ 23
Electronic & other electric equipment (36) 17
Industrial machinery & equipment (35) 15
Chemical & allied products (28) 13
Real estate (65) 7
Motion pictures (78) 7
Metal mining (10) 5
Rubber & plastic products (30) 5
Fabricated metal products (35) 5
Transportation equipment (37) 5
Instruments & related products (38) 5
Coal mining (12) 4
Construction (15,16) 4
Hotels (70) 3
Food products (20) 2
Printing & publishing (27) 2
Pipelines (46) 2
General merchandise stores (53) 2
Qatari 2i
Full Sample 209

"Others" contain joint ventures in paper products, furniture,
insurance, health services, and other industries.

72
years. That is, whereas around 25% of the sampled firms entered into
joint ventures during 1975-1981, approximately 75% of the sampled firms
entered into joint ventures during 1982-1987. These figures suggest a
growing trend of joint ventures over time.

Panel B of Table 4 presents the joint ventures distributed by
industry, of which two-digit SIC code is in the parenthesis. The
industries containing the largest number of joint ventures are
communications, followed by oil and gas extraction, electric gas and
services, electronic equipment, and industrial machinery and equipment.
Joint ventures in these five industries represent around 54% of the

entire sample.

|°j . v- ; ~ g,. , 7 p. - ; .,- . v; 7;.

Table 5 presents means, standard deviations, and a correlation
matrix of six variables. The entire sample of firms' mean cumulative
abnormal return during the announcement period is equal to 0.516%. The
positive value of this abnormal stock return suggests that the parent
firms' stockholders obtain value gains approximating the announcement
event. The mean values of two measures of relatedness--parent-child and
parent-parent relatedness--are 58.34% and 36.70%, respectively. These
figures indicate that the sampled parent firms were, on average, more
related to their child than to their partners. The joint ventures
entered by these sampled firms were in the industries whose level of
concentration averaged 39.42%. The child industries had sales revenues
that were, on average, less than the sales revenues of the parent firms.
This is indicated by the value of 3.11 for potential market impact,

which is the proportion of the parent firms' sales revenues relative to

73

Table 5

Descriptive Statistics and Correlation Matrix for Variables

(1) Announcement
period CAR .516% 3.30 1.00

2) Parent-child
relatedness 58.34% 37.18 .336""1.00

(3) Parent-parent
relatedness 36.70% 39.50 .055 .525""1.00

(4) Concentration39.42% 18.31 .089 -.058 -.10 1.00

(5) Potential
market impact 3.11 14.95 -.177* .095 .113 .l96+1.00

(6) Size ($M) 3103.67 6879.51 -.017 .006 -.060 .050 .172*l.00

74
the child industry's sales revenue. Finally, the average parent firm
size, as indicated by the average market value of the firms' outstanding
stocks at year end prior to the joint venture announcement, is $3103.67
million. This relatively large size of the firms comes about probably
because the sample includes several of the largest companies listed on
the two major stock markets, including International Business Machines
and General Electric.

The correlation matrix shows that the announcement period
abnormal stock return correlates significantly with parent-child
relatedness (r-.336). The positive value of the correlation coefficient
indicates that the announcement period abnormal return increases with an
increase in parent-child relatedness. The announcement period abnormal
return also correlates significantly with potential market impact (r-
-.177). The negative value of the correlation coefficient indicates
that the announcement period abnormal return decreases when the parent
firms had larger proportion of sales revenues relative to those of the
child industry. The correlation coefficient between the two independent
variables, i.e., parent-child and parent-parent relatedness, is
significantly positive (r-.525). This correlation value suggests the
potential problem of multicollinearity between the two independent
variables. Finally, among the potential confounding variables,
potential market impact correlates significantly with parent firm size
(r-.172) and industrial concentration (r-.l96). These figures indicate
that the parent firms with large potential market impact tend to have
large market value of outstanding stocks and form joint ventures in more

concentrated industries, respectively.

75

W
. 1 :9 - £910 1 :ue - aq- itt°qu= . te .s; o e Entire
MW

Hypothesis One focuses on the relationship between joint venture
announcements and abnormal stock returns for the entire sample of parent
firms. This hypothesis states that: Joint venture announcements are
positively correlated with parent firms' abnormal stock returns. The
examination of this hypothesis provides a check of whether the set of
firms for this study exhibit positive abnormal stock returns; a finding
reported in prior research (e.g., McConnell & Nantell, 1985; Koh &
Sinha, 1990; Woolridge & Snow, 1990).

Estimated changes in the entire parent firms' abnormal stock
returns associated with the joint venture announcements are presented in
Table 6. Column (1) identifies the day relative to the event day zero,
the joint venture announcement day; Column (2) presents the cumulative
average abnormal stock returns over every two-day interval during the
test period; Column (3) presents the percentage of firms with positive
two-day abnormal stock returns; and Column (4) presents the cumulative
average abnormal returns during the test period.

Column (2) shows that the two-day announcement period (day -l,0)
cumulative average abnormal return for the entire sample of firms is
0.516%. A one-tailed t test of this abnormal return was performed. The
t test yielded a statistic of 2.26, suggesting that the parent firms'
abnormal stock return was significantly different from zero at p<.05
level of significance. As the market model (Fama, 1976) suggests, a
significant deviation of abnormal stock return from zero around any

announcement event could be caused by market adjustment to the flow

76

Table 6

Summary of Abnormal Stock Returns Surrounding
the Announcements of Joint Ventures"

(1) (2) (3) (4)

Day Two-day Percent positive Test period
cumulative average two-day average cumulative average
abnormal return abnormal return abnormal return

-20 0 155% 49.8% 0.155%
-18 -0 050 48.3 0.105
-16 0 045 50.7 0.150
-14 -0 002 46.4 0.148
-12 0 431 51.7 0.579
-10 0 099 54.6 0.678

-8 0 033 48.8 0.711

-6 0 116 48.3 0.827

-4 -0 031 50.2 0.796

-2 -0 059 36.8 0.737

0 0 516% 57.0% 1 253%

2 0 064 45.9 1.317

4 -0 195 45.9 1.122

6 0 058 52.2 1.180

8 -0 052 47.4 1.128

10 0 166 45.0 1.294

12 0 067 48.3 1.361

14 0 178 49.8 1.539

16 0 096 50.7 1.635

18 0 083 49.3 1.718

20 -0 258 45.9 1.460

77

of announced information that is specific to the firm. Based on this
perspective, the significant deviation of the abnormal stock return from
zero around the joint venture announcement date suggests that the
capital market adjusted to this announcement. In other words, the joint
venture announcements had effects on the parent firms' abnormal stock
returns, and these effects were positive. Therefore, Hypothesis One was
supported.

This finding is not due to a few outlier observations. None of
the observations have standardized residuals greater than +3 or -3.
Additionally, 119, or 57%, of the two-day CARs during the announcement
period are positive. Column (4) shows that between day -20 to day 0 and
day -2, CARs grew from 0.155 to 1.253% and to 1.317%, respectively.
Between day 0 and day 10, CARs declined a mere 0.247%, suggesting
moderately persistent effects of the joint venture announcements during
this period.

Empirical support for Hypothesis One suggests that the firms
sampled in this research share some characteristics with firms sampled
in previous research. Such similarity strengthens the possibility that

the findings of this study can be generalized to other joint ventures

not included in this study's sample.

 

A linear regression method was used to test Hypothesis Two,
which states that: Parent firms that are more related to their child
obtain higher abnormal stock returns than parent firms that are less

related to their child. The two-day announcement period abnormal stock

78
returns for 209 parent firms entering into joint ventures during the
period 1975-1987 were regressed on parent-child relatedness. Table 7
displays the regression results, including F-value, regression
coefficient, R? and adjusted R2, for parent-child relatedness variable.
Because the F-value for the equation is significant, this
suggests a linear relationship between parent-child relatedness and
parent firms' abnormal stock returns. The regression coefficient for
parent-child relatedness is .336, and is statistically significant at
p<.001 level, suggesting that the second hypothesis is strongly
supported. This .336 value can be interpreted that, as parent-child
relatedness increases by 1%, the abnormal stock return for the parent
firm increases by 0.336%. The proportion of variance in abnormal stock
returns explained by parent-child relatedness is equal to .113, the
value of R2. The value of adjusted Rf, which provides a more realistic
estimate of the population value than R2, is equal to .109.
Table 7

Results from the Regression of Announcement Period Abnormal Returns
on Parent-Child Relatedness')

Variable Regression Coefficient T-statistics
Parent-child relatedness . 336"" 5 . 085
Constant -1.223"'' -3.035
R2 .113
Adjusted R2 .109
F 25.85”"

H 209

" Regression equation (5) from Chapter Three: AR - -l.223"+.336'" PCR
“ p<.01. *” p <.001.

 

A linear regression method was used to examine Hypothesis Three,
which states that: Parent firms that are more related to their partners
obtain higher abnormal stock returns than parent firms that are less
related to their partners. Table 8 displays the regression results,
including F-value, regression coefficient, R? and adjusted R2, for
parent-parent relatedness variable.

Because the F-value for the equation is not statistically
significant, this suggests no linear relationship between parent-parent
relatedness and parent firms' abnormal stock returns. The regression
coefficient for parent-parent relatedness also is not statistically
significant and this suggests insignificant relationship between parent-
parent relatedness and parent firms' abnormal stock returns. Hypothesis
Three was, therefore, not supported. The value of R?, which indicates
the proportion of variance in abnormal stock returns explained by
parent-parent relatedness, is approximately zero. Similarly, the value
of adjusted R? is approximately zero.

Table 8

Results from the Regression of Announcement Period Abnormal Stock
Returns on Parent-Parent Relatedness‘)

Variable Regression Coefficient T-statistics
Parent-parent Relatedness .055 .754
Constant .339 1.04
R2 .003
Adjusted R? - .002
F .568
N 209

" Regression equation (6): AR - .339+.055 PPR

80

- d t s ent Firms' Abnormal
Ret

Hypothesis Four states that: Among parent firms that are
unrelated to their child, the parent firms whose partners are more
related to their child obtain higher abnormal stock returns than the
parent firms whose partners are less related to the child. A subsample
of parent firms that have zero value of relatedness with their child was
formed. This subsample contains 22 parent firms. Differences between
the parent firms in the subsample and their partner in relatedness with
the child were then calculated. Next, the abnormal stock returns to the
parent firms in the subsample were regressed on the differences between
parent firms in their relatedness with the child. Table 9 displays the
regression results, including F-value, regression coefficient, R2 and
adjusted R2.

Because the F-value for the regression equation is significant,
this suggests that the differences between parent firms in their
relatedness with the child and the abnormal stock returns to the parent
firms in the subsample has a linear relationship. The regression
coefficient for differences in parents-child relatedness is
significantly positive at p<.10 level of significance, suggesting that
Hypothesis Four is moderately supported. The value .425 of this
regression coefficient can be interpreted that as the difference between
the parent firms in their relatedness with the child increases by 1%,
the abnormal stock return to the parents that are unrelated to their
child increases by .425%. The value of R2, or the proportion of
variance in parent firms' abnormal stock returns explained by

differences in parents-child relatedness, is .181. The value of

81
adjusted R2 is .13.
Table 9

Results from the Regression of Announcement Period Abnormal Stock
Returns on Differences in Parents-Child Relatedness"

Variables Regression coefficient T-statistics

Differences in

Parents-Child Relatedness .425+ 1.88
Constant -5.a30‘ -2.18
R2 .181
Adjusted R2 .130

F 3.534+
n 22

‘oiRegression equation (7): AR. - -5.43' + .425+ DPC
+ t L
p<.10. p<.05.

l o P ent a C 0 din V bl s
This study controls for the effects of four variables:

potential market impact, the concentration of the child industry, parent
firm size, and price-regulated industries. The approach used to control
for the first three variables was to enter these variables into the
regression equation that has parent firms' abnormal stock returns as a
dependent variable. There was no prespecified order of entry for these
three variables because no strong theory regarding the relative
importance of these variables for firms' performance could be
identified. When the regressions were computed specifying various other
orders of entry of the three variables, however, no appreciable
difference in results was found.

Parent-child relatedness was entered next into the regression
equation so that changes in R? and adjusted R? could be determined.

Parent-parent relatedness was not included in the equation because it

82
has high multicollinearity with parent-child relatedness, and it has
insignificant relationship with the abnormal stock returns. The
regression results are presented in Table 10.

Table 10 arrays the values of regression coefficients, R2, R2
change and adjusted R2, as the variables were entered cumulatively into
a regression equation. The relationships between parent-child
relatedness and parent firms' abnormal stock returns do not appear to be
significantly affected by the child industry's concentration or parent
firm size because these two variables do not have statistically
significant relationships with parent firms' abnormal stock returns.

The relationships between parent-child relatedness and parent firms'
abnormal stock returns appear to be moderately affected by potential
market impact, which has a significant relationship with abnormal stock
return at p<.10 level of significance. Potential market impact can,
however, explain slight variance in parent firms' abnormal stock returns
because the value of adjusted R2, when only potential market impact was
in the equation, was .013.

The procedure used to control for the effects of the three
price-regulated industries on parent firms' abnormal stock returns began
with the division of the entire sample of 209 firms into two subsamples:
regulated and non-regulated. The regulated subsample consists of 33
parent firms that entered into joint ventures in the three regulated
industries, i.e., electric utility, oil and gas exploration and oil and
gas pipeline transmission; the non-regulated subsample consists of 176
parent firms that entered into other industries. The descriptive

statistics for the regulated and non-regulated subsample are displayed

83

Table 10

Results from the Regression of Announcement Period Abnormal Stock
Returns on Relatedness and Potential Confounding Variables‘)

Variables Regression T-sta- R? R2 Adjusted

coefficients tistics change R2

Potential market impact -.234+ -l.740 .032 .032 .013

Industrial concentration .146 1.109 .048 .016 .011

Parent firm size .005 .037 .048 .000 -.009

Parent-child relatedness .437' 2.893 .181 .133 .116

Constant -2.20+ -l.806

F 2 . 768“

N 209

" Regression equation (8): AR - -2.20* -.234* MKTIMP +.l46 CONCEN
+.005 SIZE +.437* PCR

p<.01. *" p<.05.

t

84
in Table 11.

Table 11 consists of two panels. Panel A arrays the descriptive
statistics of abnormal stock return, relatedness, and three potential
confounding variables, for the two subsamples and for the full sample.
Panel B displays the results of two-sample t tests that were used to
examine differences in the groups on the variables included in Panel
A of the table.

The t-test results indicate that the abnormal stock returns for
the three groups are not significantly different. At the approximately
similar levels of announcement period abnormal stock returns, however,
the regulated subsample has higher level of parent-child and parent-
parent relatedness than the non-regulated subsample or the full sample,
and these differences are statistically significant. These results
hence suggest that, on average, high relatedness between parent and
child or between parents does not lead to as high positive abnormal
stock returns to the firms entering into joint ventures in the three
regulated industries as those to the firms entering into joint ventures
in other industries.

The next stage used in controlling for the effects of the three
regulated industries was to run the regression analyses for the
regulated and non-regulated subsamples. Announcement period abnormal
stock returns to the parent firms in these two subsamples were regressed

on relatedness and potential confounding variables. The results are

85

Table 11
Comparison between Regulated, Non-regulated Subsamples and Full Sample

(A) Descriptive Statistics

Regulated Non-regulated
Variables subsample subsample Full sample
(ll-'33) (3-176) (ll-209)
WM
Two-day announcement
period abnormal return
Means .49 .52 .52
SD 2.15 3.48 3.30
e v a s
Parent-child relatedness
Means 82.35 54.28 58.34
SD 31.29 36.62 37.18
Parent-parent relatedness
Means 59.27 35.06 38.70
SD 46.30 37.15 39.50
a Va a e
Potential market impact
Means 8.72 1.86 3.11
SD 34.26 3.87 14.95
Concentration
Means N.A. 39.42 39.42
SD N.A. 18.31 18.31
Parent firm size
Means 1312.95 3103.67 3103.76
SD 1930.81 6879.51 6879.51

(B) T-statistics for Differences in Variables

Differences between Regulated and Regulated and Non-regulated
non-regulated full sample and full sample

Two-day announcement period

abnormal stock return .06 -.05 .01
Parent- child relatedness -4 . 64'" 4 . 02'" - 1 . l3
Parent-parent relatedness -2.801". 2.39. -.94
Potential market impact -.92 -2.l7* -2.23*
Concentration N.A. N.A. .00
Parent firm size 2.75“. -2.75” .00

N.A. -> Information for the sample not available
" p<.05. "" p<.01. *” p<.001.

86
presented in Table 12. Table 12 consists of two panels, i.e.,Panel A
and B, which present the regression results for the regulated and non-
regulated subsamples, respectively. The regression coefficient for
parent-child relatedness is -.087 for the regulated subsample, compared
with .484 for the non-regulated subsample, or .437 for the full sample.
These figures suggest that the regulated industries mitigated the
relationship between parent-child relatedness and parent firms' abnormal
stock returns. This is indicated by the increase in: the regression
coefficient for parent-child relatedness from .437 to .484, R2 from .181
to .197, and adjusted R? from .116 to .133, after the regulated
subsample was excluded from the full sample,.

The three price-regulated industries also affect potential
market impact. In other words, interaction exists between price-
regulated industries and potential market impact. The regression
coefficient for potential market impact, which was statistically
significant for the full sample, became statistically insignificant
after the regulated subsample was excluded from the full sample.

In summary, only one potential confounding variable--specific
regulated industries--had confounding effects on the relationships
between relatedness among joint venture participants and parent firms'
abnormal stock returns. Regulated industries also interact with
potential market impact, that is, they lead to the significant effects

of potential market impact when their effects are not controlled for.

87

Table 12
Regression Results for Regulated and Nonregulated Subsamples

A. Regression Results for Regulated subsample“)

Variables Regression T-statistics R2 R? Adjusted

Coefficients change R2

Potential market impact -.536 —l.785 .289 .289 .225

Industrial concentration N.A. N.A. N.A. N.A. N.A.

Parent firm size .150 .509 .312 .023 .175

Parent-child relatedness -.087 -.260 .320 .008 .093

Constant 1.062 .603

P 1.411

n 33

N.A. -> Information not available
‘” Regression equation (8): AR - 1.062 -.536 MKTIMP +.15 SIZE -.087 PCR

Variables Regression T-statistics R2 R2 .Adjusted

Coefficients change R2

Potential Market Impact -.l64 -l.219 .022 .022 .003

Industrial concentration .130 1.008 .035 .013 -.002

Parent Firm Size -.019 -.148 .035 .000 -.021

Parent-Child Relatedness .4848“ 3.292 .197 .162 .133

(.437)' (2.893) (.181) (.133) (.166)

Constant -2.1l3+ -1.717

P 3.063"

n 176

flote Figures in parentheses are for the full sample
'” Regression equation (8): AR - -2.133+ -.l64 MKTIMP +.l3 CONCEN
-.019 SIZE +.484f' PCR

+ p<.10. ' p<.01.

88

Man
This chapter presents the results of testing the following

hypotheses:

H1: Joint venture announcements are positively correlated with parent
firms' abnormal stock returns. This hypothesis was supported.

H2: Parent firms that are more related to their child obtain higher
abnormal stock returns than parent firms that are less related to
their child. This hypothesis was strongly supported.

H3: Parent firms that are more related to their partners obtain higher
abnormal stock returns than parent firms that are less related to
their partners. This hypothesis was not supported.

H4: Among parent firms that are unrelated to their child, the parent

firms whose partners are more related to their child obtain higher
abnormal stock returns than the parents whose partners are less
related to the child. This hypothesis was moderately supported.

CHAPTER FIVE

DISCUSSION, IMPLICATIONS AND CONCLUSIONS

The purpose of this study is to examine the relationship between
joint venture announcements and parent firms' abnormal stock returns.
This study assumes that relationships across joint ventures are not
homogeneous and hypothesizes that relatedness among joint venture
participants affects parent firms' stock returns. This chapter first
discusses the empirical findings and then provides their implications.
Limitations of this study, suggestions for future research along with

conclusions also are presented.

W
We
Hypothesis One posits a positive relationship between joint
venture announcements and parent firms' abnormal stock returns.
Empirical support in this study of this hypothesis is consistent with
the findings from prior studies (e.g., McConnell & Nantell, 1985;
Woolridge & Snow, 1990). These findings share the suggestion that joint
ventures are value-creating activities for stockholders of parent firms.
That is, the capital market perceives that investment decisions, such as
joint ventures, enhance a firm's ability to generate future cash flow

(Rappaport, 1986). The capital market, therefore, responds to joint

89

90
venture announcements by showing higher market valuation for the parent
firms approximating the joint venture announcement event.

Another interpretation of the capital market's positive reaction
to joint venture announcements is that the market expects joint ventures
to generate competitive advantages for parent firms. For instance, the
capital market might perceive that a joint venture enables a parent firm
to become a low-cost producer, whose strength is low cost, or a
differentiator, whose products/services are perceived as unique by the
buyers (Porter, 1980, 1985). Similarly, the capital market might
perceive that a joint venture enables a firm to erect entry barriers to
new entrants and hence could sustain the firm's competitive advantages.

The finding that the CARs for the parent firms are still high
over twenty-day period after the announcement date suggests moderately
persistent effects of the joint venture announcements during the period.
However, because this study does not examine the CARs beyond the twenty-
day period after the announcement date, the study cannot indicate the
period that the effects of the joint venture announcements would last.
One prior joint venture study (McConnell & Nantell, 1985), however,
reports that the announcement effects are evident 60 trading days after
the announcement date, and beyond this point there is no further
systematic valuation effect.

The mitigating effects of joint venture announcements may not
conflict with management's long-term goal of stockholder value
maximization. This is because the initial announcement of a joint
venture can be considered as one event in a series of moves constituting
joint venture activities. The declining effects of the joint venture

announcements may occur because the capital market anticipates a further

91
move in a series of joint venture activities such as a starting
operation by a joint venture firm, after which the persistent effects of

the joint ventures on stockholder value may occur.

W

The finding of the significant, positive relationships between
parent-child relatedness and parent firms' abnormal stock returns lends
support to Hypothesis Two. This finding suggests that investors in the
stock market, on average, form expectations about future income streams
generated by joint ventures based on similarities between parent and
child firms in their product/market areas. Such positive relationships
occur for joint venture participants in various types of industries.
Exceptions include joint ventures in oil and gas exploration, oil and
gas transmission and electric utility industries, where no such positive
relationship was found.

The confirmation of Hypothesis Two can be interpreted as a
justification for the theory underlying this hypothesis. The theory
proposes that the capital market perceives the association between high
parent-child relatedness and distinctive advantages, which are not
associated with low parent-child relatedness. The capital market might
perceive that high parent-child relatedness enables a share between
parent and child firms in specialized resources/skills used in
production or marketing, and this leads to the distinctive advantages
that cannot be obtained if parent and child firms are less related.
These distinctive advantages are, for example, lowered average costs in
production or marketing functions, which, in turn, implies higher future

income streams for the parent firms.

92

A corollary of the theory suggests that the stock market
perceives the association between low parent-child relatedness and
distinctive disadvantages. The stock market might perceive that low
parent-child relatedness leads to high costs that are unlikely to occur
if parent and child firms are highly related. These costs include, for
example, investment costs in plants or production technology, learning
costs and entry costs.

The relation between parent-child relatedness and parent firms'
abnormal stock returns diminishes in regulated industries. The capital
market does not react positively to joint venture announcements in oil
and gas exploration, oil and gas transmission and electric utility
industries probably because the market does not perceive specific
advantages associated with high relatedness or specific disadvantages
associated with low relatedness. High parent-child relatedness in these
industries may not suggest parent firms' ability to control factors that
lead to distinctive advantages for the firms. These factors include,
for example, investment costs and product price. High parent-child
relatedness in regulated industries may not suggest low investment costs
for the parent firms. The parent firms that are 100% related to their
child and are in the regulated subsample have varied investment costs,
ranging from $4 million to $1.3 billion, and have the average cost of
$203 million. High parent-child relatedness in regulated industries
also does not suggest that parent firms can control product prices
because the prices are determined by regulatory agencies. High parent-
child relatedness in regulated industries, therefore, does not lead to
market expectation of parent firms' low investment costs or parent

firms' power to control market price.

93
W

Hypothesis Three was not supported because there was no
statistically significant relationship between parent-parent relatedness
and parent firms' abnormal stock returns. The interpretation of this
finding is that higher similarities between parent firms in their
product/market areas do not suggest greater advantages than lower
similarities. High parent-parent relatedness suggests that parent firms
possess similar specialized resources because they operate in similar
product/market areas. A joint venture between firms that are highly
related may enable the parent firm to supplement its specialized
resources with the resources of its partner. This may lead to better
utilization of resources, and hence lowered average costs for the parent
firm.

In a similar vein, low parent-parent relatedness may suggest to
the capital market that the parent firms can better utilize their
resources. Low parent-parent relatedness suggests that parent firms
possess dissimilar specialized resources because they operate in
different product/market areas. A joint venture between firms that have
low relatedness may enable the parent firm to complement its resources
with the partner's resources. The complement of resources may enable
the parent firm to better utilize its resources, and hence lower its
average costs.

Examples of the joint ventures that are consistent with the
above discussion can be provided. Two sampled joint ventures will be
discussed: a joint venture between Mitel Corp. and American Satellite
Company, announced 12/14/82; and a joint venture between MCA and IBM,

announced 9/6/79. The sampled parent firms from these two joint

94
ventures are Mitel and MCA. Mitel's products were telecommunication
equipment and integrated circuits used in telecommunication industries,
the industries in which American Satellite operated. Mitel and American
Satellite were, therefore, 100% related because Mitel's products serve
as input for the industries where American Satellite was in. On the
other hand, MCA and IBM were 0% related because they operated in
different product/market areas and did not have buying-selling
relationships. MCA's business concern was video disk, whereas IBM's
business was in the field of information handling systems.

The announcement period abnormal returns that Mitel and MCA
obtain are 0.041% and 0.046%, respectively. Because these returns are
not statistically different, this suggests that a firm that is 0%
related to its partner can obtain as high abnormal stock return as a
firm that is 100% related to its partner. The joint venture between
Mitel and American Satellite to provide a telecommunication carrier to
businesses may suggest a great opportunity for these two firms to
supplement their resources with those of the partner. As the news in
the Hgll_§§;gg§_lgu;ngl stated: "...The [telecommunication] carrier will
combine American Satellite's satellite transmission capability with
Mitel's telecommunication switching technology." On the other hand, the
joint venture between MCA and IBM to manufacture and market disks for
consumer and industrial markets may suggest a great opportunity for
these two firms to supplement their resources with those of the partner.
The joint venture announcement in the Hall Street Journal states that
IBM would contribute patents and technology from its own exploration of
video disks, plus an undisclosed amount of cash. IBM's contributed

resources could, therefore, complement MCA's resources. IBM also could

95
benefit from this joint venture because it could expand industrial and
educational markets for video disks. The examples indicate that a joint
venture between related parent firms or between unrelated parent firms
suggests better utilization of resources, the advantage of which carries
similar weight for high or low parent-parent relatedness. This then
explains the finding of no positive relationship between parent-parent
relatedness and parent firms' abnormal stock returns.

The finding of significant correlation (r-.525) between the
independent variables--parent-child relatedness and parent-parent
relatedness--suggests that each independent variable is at least partly
carrying information about the parent firms' abnormal stock returns that
is also supplied by the other. High parent-child relatedness may carry
the notion that parent and child firms can share specialized and general
resources, which, in turn, leads to lowered average costs. Similarly,
high parent-parent relatedness may carry the notion that parent firms
can share specialized resources, which, in turn, leads to lowered
average costs. The notion of lowered average costs carried by high
parent-child relatedness or high parent-parent relatedness is thus
redundant.

In spite of the partially redundant information, parent-child
relatedness or parent-parent relatedness also carries the information
that is not redundant. High parent-child relatedness carries the notion
that parent and child firms can share specialized resources, the
advantage that may not be obtained by low parent-child relatedness.

High parent-parent relatedness, on the other hand, carries the notion
that the parent firms can obtain supplementary resources. The capital

market may perceive that this advantage carries similar weight to the

96
advantage associated with the acquisition of complementary resources,
which tends to occur when there is low parent-parent relatedness. Based
on this logic, the capital market tends to rely on parent-child
relatedness when it determines the value of joint ventures because the
additional information carried by parent-child relatedness can indicate
the distinctive advantages associated with high relatedness, but not
with low relatedness. On the contrary, the additional information
carried by parent-parent relatedness cannot indicate the distinctive
advantages associated with either high or low parent-parent relatedness.
This explanation is consistent with the findings of significant
relationship between parent-child relatedness and parent firms' abnormal
stock returns, but nonsignificant relationship between parent-parent
relatedness and parent firms' abnormal stock returns.

A conclusion is that parent-child relatedness is more relevant
in explaining variance in parent firms' abnormal stock returns than
parent-parent relatedness. Parent-parent relatedness is more like a
peripheral variable because it partly carries redundant information with
parent-child relatedness, and the additional information that it carries
cannot indicate differences in advantages associated with high or low

parent-parent relatedness.

W
The finding of the statistically significant, positive

relationship between differences in parents-child relatedness and
abnormal stock returns for parent firms that are unrelated to their
child lends support to Hypothesis Four. This finding suggests that

relatedness between one parent and child can provide advantages that

97
compensate for disadvantages associated with unrelatedness between the
other parent and child. This is an important finding which adds to
theoretical notions concerning relatedness. That is, the capital market
does not always react negatively to unrelatedness between parent and
child. If there is a combination of parent-child relatedness and
parent-child unrelatedness, the disadvantages associated with parent-
child unrelatedness can be mitigated.

An example of the sampled joint venture that is consistent with
the postulation of Hypothesis Four can be provided. In the joint
venture between Gillette Co. and Williamson, Ariedge & Richards, Inc.,
announced 3/30/83, one parent firm was related to the child while
another parent firm was not. Gillette was a producer of razors, blades,
grooming aids and writing instruments. Williamson, Ariedge & Richards,
Inc., was a closely held oil and gas exploration and development
concern. These two companies entered into an oil and gas exploration
joint venture with the investment value at about $3.8 million. Gillette
and the joint venture child were 0% related because they operated in
different business segments. Williamson and the child, on the other
hand, was 100% related because they were engaged in the same business
segment. The capital market reacted positively to Gillette's decision
because it might perceive that though Gillette was not related to the
child, its partner or Williamson was. Williamson had specific
resources/skills necessary for oil and gas exploration and this could
compensate for Gillette's lack of these resources. On the other hand,
from Williamson's point of view, the firm would still benefit from this
joint venture because Gillette could provide general resources such as

financial resources required for the venture.

98
Wm

The empirical findings in this study lead to several
implications. One implication is that not all joint ventures are
homogeneous in their effects on stock returns. Relatedness between
parent and child might differ across joint ventures and this leads to
different reactions from the capital market to joint venture
announcements. Joint ventures, therefore, should not all be perceived
as similar. A more fine-grained approach that considers different
aspects of joint ventures should be used when conducting a study on
joint ventures.

Another implication is that it is possible to generalize the
concept of relatedness from merger/acquisition studies to joint venture
studies. The theoretical notion regarding specific advantages
associated with relatedness proposed by several diversification studies
(e.g., Hill & Hoskisson, 1987; Singh 6 Montgomery, 1987) can be applied
to joint venture studies. This theoretical argument can explain the
empirical findings of the relationships between relatedness and firms'
performance in a number of diversification studies and this joint
venture study. That is, higher relatedness accounts for better
performance of firms. And because of the compatibility in findings
between joint venture and merger/acquisition studies, this implies that,
to some extent, investors in the capital market used similar logic to
determine the value of joint ventures and mergers/acquisitions.

The findings from this study also have implications for
practitioners. One implication for management is that managers may
consider a joint venture formation as one approach that can help them

increase the stockholder value. A firm with a market value of $3

99
billion, for example, can increase its market value by $16.05 million
over the two-day announcement period, of which the cumulative average
abnormal return is 0.517%. This increase in stock market value
associated with the joint venture announcements also was found to be
greater than the increase in stock market value associated with other
strategic investment decisions. Woolridge and Snow (1990), for example,
found that joint venture announcements led to a greater increase in
stock market value than other investment announcements, such as internal
product/market diversification or capital expenditure, including plant
modernization or capacity expansion. Therefore, a joint venture
formation seems to be one viable approach that enables management to
increase stockholder value.

Another implication for management is that there is more than
one approach for management to increase the firm's stockholder value
associated with a joint venture formation. Forming a joint venture in
related product or market areas is just one approach. Another approach
is that management forms a joint venture in unrelated product or market
areas with a partner firm that is related in product or market areas to
the child firm. This approach can be beneficial for both parent firms.
The parent firm that is unrelated to the child firm can fulfill its goal
of diversifying business into different product or market areas through
the child's business. This parent firm also can lessen its
disadvantages associated with unrelatedness to the child because its
partner tends to have specific resources/skills required by the child.
The parent firm that is related to the child also can benefit from this
joint venture because its partner may possess resources/skills that it

is lacking.

100

This study also has an implication for stockholders.
Stockholders of firms that announce joint ventures may obtain high value
gains if they immediately sell their stocks, either on the announcement
day or within several days after the announcement day. This study found
that the cumulative average abnormal returns for the parent firms peaked
during the two days after the announcement day and remained moderately
high over twenty-day period after the announcement date. Therefore,
stockholders who need to obtain value gains from joint venture

announcements are suggested to act immediately after the announcements.

t i t es s o ture Rese rch

Several limitations in this study provide issues for future
research. First, the dependent variable in this study is abnormal stock
return, which is market-based performance because it reflects the
performance of firms evaluated by investors in the capital market.
Other performance measures may be used in future research to verify the
relationship between relatedness among joint venture participants and
parent firms' abnormal stock returns. Examples include accounting-based
measures such as return on investment, net profits or sales growth.

Second, the measure of relatedness in this study was developed
based on Shelton's (1988) approach of using one measure to indicate
product and/or market relatedness. Firms may be related either through
products or markets. Product relatedness and market relatedness were,
therefore, given equal weights in the calculation of measure of
relatedness in this study. However, it is possible that the stock
market does not assign equal weights to product relatedness or market

relatedness. Future research may, therefore, measure product

101
relatedness and market relatedness separately so that the relationship
of each with parent firms' performance can be determined. Prior studies
on acquisitions such as Capon et al. (1988) and Palepu's (1985) may be
helpful about this issue. This is because the former study develops a
measure of market relatedness, whereas the latter study develops a
measure of product relatedness.

In addition to the classification of relatedness by products or
markets, relatedness may be classified by horizontal or vertical
relationships among firms. Firms that have horizontal relationships are
those that compete in the same product/market areas. Firms that have
vertical relationships are engaged in the same product lines and have
buying-selling relationships. Because studies on a relationship between
horizontal joint venture and firms' performance are more prevalent
(e.g., Duncan, 1982, Koh & Sinha, 1990), future research may, therefore,
focus on a relationship between vertical joint venture and firms'
performance. Additionally, a comparison between the effects of
horizontal and vertical joint ventures on accounting-based performance
or market-based performance may be investigated.

Third, this study samples joint ventures, each of which consists
of only two parents. A joint venture consisting of more than two
parents may, however, be sampled and examined because the number of
parent firms and their linkage may have some effects on firms'
performance. Rob and Sinha (1990), for example, argue that a horizontal
joint venture consisting of a greater number of parents is less
profitable than a venture consisting of a fewer number of parents. -Koh
and Sinha (1990) argue that when market concentration is high, a joint

venture with many parents leads to a small increase in the output of

102
each parent and this output increase may not be sufficient to compensate
for a dropping in price in response to any increase in the output. Each
parent in the venture may, therefore, not gain much because of the joint
venture formation. To examine the effects of the number of parents
firms on the firms' performance, a theoretical notion concerning
network (e.g., Thorelli, 1986) or interorganizational linkage (e.g.,
Sterns, Hoffman & Heide, 1987) may be helpful.

Fourth, this study focuses on the effects of joint venture
announcements on parent firms' abnormal stock returns. A joint venture
announcement may be considered I"intended" strategy (Mintzberg, 1978)
because it reflects management's intention to implement a joint venture
decision. However, it may be possible that the intended strategy
becomes unrealized or is not implemented. Therefore, future research
may focus on the effects of joint ventures after they are implemented or
become "realized" strategy (Mintzberg, 1978). It would be interesting
to determine whether the announcement of joint venture completion leads
to a persistent increase in the stockholder value of parent firms.
Furthermore, the relation between relatedness among joint venture
participants and actual performance of a joint venture child can be
another focus of future research.

Finally, with respect to the joint ventures in the three
regulated industries, where the relationship between relatedness and
stock market responses was not as expected, other predicting variables
may be examined. For example, the relationships between investment
costs in joint ventures or environmental variables, such as price/cost
structure in regulated industries or demand instability (Hofer, 1975),

and parent firms' abnormal stock returns may be examined.

103
W

Over the past decade, joint ventures have been increasing in
popularity. One reason for this popularity may be the expected increase
in the value of parent firms. Most of prior studies assume that joint
ventures have similar, positive effects on firms. The present study,
however, has a different assumption from prior studies. This study
assumes that not all joint ventures are homogeneous and then
hypothesizes that relatedness among joint venture participants affects
parent firms' market performance. The findings in this study lend
support to this hypothesized notion. That is, relatedness between
parent and child firms could explain variance in parent firms' abnormal
stock returns approximating the joint venture announcement event. The
more the relatedness between parent and child firms was, the higher the
value gains the stockholders of parent firms obtained.

The findings in this study offer insight into relatedness
concept. That is, for a group of firms participating in a joint
venture, it is insufficient to predict the performance of one parent
firm based on only its relatedness to a child firm. Relatedness between
the other parent firm and a child has to be considered simultaneously
with the relatedness between the former parent and the child. This is
because the finding reveals that, from the perception of the capital
market, relatedness between one parent firm and a child generates just
like spillover in value gains to the other parent firm that is unrelated
to the child.

The findings from this study also have several implications for
practitioners. This study implies that management have several avenues

to increase stockholder value through a formation of joint venture.

104
Management may increase stockholder value by forming a joint venture in
related product or market areas. Alternatively, if management intends
to diversify a firm's businesses into different product or market areas
through a joint venture, it can still increase the stockholder value by
selecting a partner firm that is related to the joint venture child.
Management, however, should also be aware that these suggested avenues
may not be functional in some exceptional industries, such as oil and
gas exploration or electric utility industries. For these particular
industries, management are suggested to form their expectations about
the effects of joint ventures based on other factors besides

relatedness.

BIBLIOGRAPHY

Ansoff, H. I. (1965). Qtrpgrgtt_§trgtggyr New Yerk: McGraw-Hill.

Amit, R., & Livnat, J. (1988). Diversification and the risk-return
tradeoff- WM. 3.1. 154-166.

Ban2. R. W. (1978). WWW
gmnirlt§l_§n§lygl§r Unpublished doctoral dissertation, University
of Chicago.

Banz, R. W. (1981). The relationship between return and market value of
common stocks. WW. 2. 3-18.

Berg, 8. V., & Friedman, P. (1981). Impacts of domestic joint ventures
on industrial rates of return: A pooled cross-section analysis,

1964-1975. MW. 51. 293-298.

Bergman, M. (1962). Corporate joint venture under the antitrust laws.

Wales. .31: 712-734-

Bettis, R. A. (1981). Performance differences in related and unrelated

diversified firms. Wei. 2. 379-393.

Boyle, S. E. (1968). An estimate of the number and size distribution of

domestic Joint subsidiaries. W. i.
81-92.

Branch, 8., & Gale, B. T. (1983). Linking corporate stock performance to
strategy formulation. W. a. 51-58.

Brown, S. J., & Warner, J. B. (1980). Measuring security price perfor-
mance. WW. 8. 205-228.

Buzzell, R. D., & Gale, B. T. (1987). Ihg_£lfl§_prlntlplg§r New York:

Free Press.

Capon, N., Hulbert, J. M., Farley, J. U., & Martin, L. E. (1988).
Corporate diversity and economic performance: The impact of market

specialization. WW1. 2. 61-74.

Chatterjee, S. (1986). Types of synergy and economic value: The impact
of acquisitions on merging and rival firms. Strategic Managemegt
m: 1: 119-139-

105

106

Clarke, R. G. (1980). The effect of fuel adjustment clauses on the
systematic risk and market values of electric utilities. Jgtrgal

finance. 2. 347-387.

Cohen. J.. & Cohen. P. (1983). MW
211W Hillsdale. NJ: Lawrence

Erlbaum.

Dann, L. Y. (1981). Common stock repurchases: An analysis of returns to

bondholders and stockholders. W. 2.

113-138.
Day, D. L., MacMillan, I. C., 6 Hambrick, D. C. (1983a). An_gmplrlt§l
“ O l' L!’ '9 ' ‘ ' '-‘ 9' "Q 01‘-11‘-duusrlal
nrggttt_htrlnt§§t§ (Working paper). Columbia Graduate School of
business.

Day, D. L., MacMillan, I. C., & Hambrick, D. C. (1983b). Market share
e c S' e o d u a businesses
(Working paper). Columbia Graduate School of business.

Dodd, P. (1980). Merger proposals, management discretion and stockholder
wealth. Wm. 8.. 105-137.

Duncan, J. L. (1982). Impacts of new entry and horizontal joint ventures

on industrial rates of return. Egvleg 9f Eggngmltg gnd Statistics,
339-342.

Fama, E. F. (1976). Egtnggtlgn§_gf_fln§ntgr New York: Basic Books.

Fama, E. F., Fisher, L., Jensen, M., 8 R011, R. (1969). The adjustment

of stock prices to new information. lnterggtlonal Ecgggmic Review,
February, l-21.

Harrigan. K. R. (1985). WWW Lexington. MA:
Lexington Books.

Harrigan, K. R. (1988). Strategic alliances and partner asymmetries. In

I"- J. Contractor & 1’. Lorenzo (Eds). WW1:

1n;2:na£12ng1_hg§1ng§§ (pp. 205-226). Lexington, MA: Lexington
Books.

Hennart, J. F. (1988). A transaction cost theory of equity joint

ventures. WW. 2. 361-374.

Hill. C. W., & Hoskisson, R. E. (1987). Strategy and structure in the
multiproduct firm. AW. 2. 331-341.

Hofer, C. (1975). Toward a contingency theory of business strategy.

WW. 5. 784-810.

Hopkins, H. D. (1987). Acquisition strategy and the market position of
acquiring firms. WW1. 2. 535-547.

107

Jensen, M. J., 6 Ruback, R. (1983). The market for corporate control:

The scientific evidence. MW. 11. 5-50.

Jones, G. R., 6 Hill, C. W. (1988). Transaction cost analysis of

strategy-structure choice. WW. 2. 159-
172.

Kitching, J. (1967). Why do mergers miscarry? ngxgrd_§g§12g§§_fitxlgg,
45, 84-101.

Kogut, B. (1988). Joint ventures: Theoretical and empirical

perSPectives. WM. 2. 319-332.

Koh, J., 6 Sinha, D. K. (1990). A model of horizontal joint ventures. In

mismanagement (pp. 27-31).

Lubatkin, M. (1983). Merger and the performance of the acquiring firm.

AW. 8. 218-225.
Lubatkin. M. (1984). WWW

fgr_mtrggr_§yngrgy (Working paper). University of Connecticut.

Lubatkin, M., 6 Shrieves, R. E. (1986). Towards reconciliation of market
performance measures to strategic management research. Atgggm1_2§

Wigs. 11. 497-512.

Mandelker, G. (1974). Risk and return: The case of merging firms.
MW. 1. 303-335.

Masulis, R. (1980). The effects of capital structure changes on security
prices: A study of exchange offers. lgtrngl_gfi_filn§ntltl
W9 £9 139-178-

McConnell, J. J., 6 Nantell, T. J. (1985). Corporate combinations and
common stock returns: The case of joint ventures. lgtrggl_gf

finance. 2.. 519-536.
WNW McLean. VA:

Mergers 6 Acquisitions.

Mikkelson, W. H. (1981). Convertible calls and security returns. Journal

2W. 1].. 155-181.

Mintzberg, H. (1978). Patterns in strategy formation. Mgnagtmgnt
fitlgntg, 934-948.

Montgomery, C. A. (1982). The measurement of firm diversification: Some

new empirical evidence. AW. 22. 299-
307.

M2QgyL§_lngt§trlgl_u§ntgl§r New York: Moody's Investor Service.

108

Oliver, C. (1990). Determinants of interorganizational relationships:

Integration and future direction. Atgggmx_g£_n§ntgtmtgt_fitxltg, 2,
241-265.

Paine, F. T., 6 Power, D. J. (1984). Merger strategy: An examination of
Drucker's five rules for successful acquisitions. fitrgttgit

W. 2. 99-110.

Palepu, K. (1985). Diversification strategy, profit performance, and the

entropy measures. W51. 6. 239-255.

Pate, J. L. (1969). Joint venture activity, 1960-1968. Eggngm1t_fitxltg,
Federal Research Bank of Cleveland, 16-23.

Penrose. E. T. (1959). MW Oxford:

Blackwell.

Pfeffer, J., 6 Novak, P. (1976a). Joint ventures and interorganizational
interdependence. W11. 2.1. 398-418.

Pfeffer, J., 6 Novak, P. (1976b). Patterns of joint venture activity:
Implication for anti-trust research. Antltrg§t_figllttln, 21, 315-
319.

Polanyi, M. (1967). Th2_£acit_dimens12ni New York: Doubleday.

Porter, M. E. (1980). figmggtltlx§_§tr§ttgy. New York: Free Press.

Porter, M. E. (1985). ngpttltlyt_tg2§ntggg. New York: Free Press.

MW Cleveland. 011:
Predicasts.

Rappaport. A. (1986). WWW

hurlnt§§_ptr£grm§nttr New York: Free Press.
Ravichandran. R. (1986). WWW

Unpublished doctoral dissertation, University of Iowa.

Reinganum, M. R. (1983). The anomalous stock market behavior of small
firms in January: Empirical tests for the tax loss selling

effects. W. 12. 89-103.

Risley, G. (1972). ugggrn_lngt§trltl_m§rktt1ng, New York: McGraw-Hill.

Rubin, P. H. (1973). The expansion of firms. lgtrngl_2£_figlltlt§l
Eggngmy, July-Aug, 936-949.

Rumelt. R. P. (1974. 1986). WWW

ptrfgrggnttr Boston: Division of research, Harvard Business
School.
Salter, M. S., 6 Weinhold, W. A. (1979). Dizersification_£hrough

ttgtlgltlgnr New York: Free Press.

109

Scherer. F. M. (1980). lndnstrial_narket_atruntnre_and_222n2mic
ptrfgrmtntgL Boston: Houghton Mifflin.

Seth, A. (1990). Value creation in acquisitions: A reexamination of

acquisition: Empirical evidence. 2tzateais_uanasement_lournal. 11.
99-115.

Shelton, L. M. (1988). Strategic business fits and corporate

acquisition: Empirical evidence. 2tratesic_nanasement_22urnal. 2.
279-287.

Shepherd. W. G. (1979). Ihe_e22n2mics_2f_industrial_2rsanizationl
Englewood Cliffs, NJ: Prentice-Hall.

Singh, H., 6 Montgomery, C. A. (1987). Corporate acquisition strategies

and economic performances. 2tratesic_uananzementalgurnal. 2. 377-
386.

Stanton, W. J. (1984). Etnggmtntal§_gf_m§rktt1ng. New York: McGraw-Hill.

Stearns, T. M., Hoffman A. H., 6 Heide, J. B. (1987). Performance of
commercial television stations as an outcome of
interorganizational linkages and environmental conditions. Attgtmy

ofluanasement_lournal. l. 71-90.

Thompson, R. 8., Olsen, C., 6 Dietrich, J. R. (1987). Attributes of news
about firms: An analysis of firm-specific news reported in the

Wall Street Journal Index. l2urna1.2f_82222nting_Eesearch. 25.
245-274.

Thorelli, H. B. (1986). Networks: Between markets and hierarchies.

2tratesis.flanszementalournal. 1. 37-51.
U.S. Department of Commerce. (1982, 1987). gtn§t§_tf_m§ntfigtttrgr§r

Washington, DC: Bureau of the Census.

U.S. Office of Management and Budget (1987). Standard_lndt§tr1§l

tlggglfltgtlgn_m§ngglr Washington, DC: U.S. Government Printing
Office.

Udell. J. G. (1972). 2n2cessfnl_marketins_stratezxi Madison. WI: Miner
Publishers.

Hall_§trggttlgurngl. New York: Dow Jones.
Eall_§treet_lournal_1ndex. New York: Dow Jones.

Webster, F. E. (1978). Management science in industrial marketing.

Journal_2f_flarketins. £2. 21-27.

Woolridge, J. R., 6 Snow, C. C. (1990). Stock market reaction to

strategic investment decisions. 2tratesic_uanasement_lournal. 2.
353-363.

llO

Wrigley. L. (1970). Won. Unpublished

doctoral dissertation, Harvard University.

Yip, G. S. (1982). Diversification entry: Internal development versus

acquisition. WW. 2. 331-345.