Academy of Management

Rivalry between Multinational Enterprises: An Event History Approach
Author(s): Tieying Yu and Albert A. Cannella Jr.

Source: The Academy of Management Journal, Vol. 50, No. 3 (Jun., 2007), pp. 665-686
Published by: Academy of Management
Stable URL: http://www.jstor.org/stable/20159878 .
Accessed: 17/06/2014 04:00
Your use of the JSTOR archive indicates your acceptance of the Terms & Conditions of Use, available at .
http://www.jstor.org/page/info/about/policies/terms.jsp
.
JSTOR is a not-for-profit service that helps scholars, researchers, and students discover, use, and build upon a wide range of
content in a trusted digital archive. We use information technology and tools to increase productivity and facilitate new forms
of scholarship. For more information about JSTOR, please contact support@jstor.org.
Academy of Management is collaborating with JSTOR to digitize, preserve and extend access to The Academy
of Management Journal.
http://www.jstor.org
This content downloaded from 185.44.78.156 on Tue, 17 Jun 2014 04:00:11 AM

All use subject to JSTOR Terms and Conditions
?
Academy ofManagement Journal
2007, Vol. 50,No. 3, 665-686.
RIVALRY BETWEEN MULTINATIONAL ENTERPRISES:

AN EVENT HISTORY APPROACH
TIEYING YU
Boston College
ALBERT A. CANNELLA Jr.

Arizona State University
Our study investigates rivalry between multinational enterprises (MNEs) in host coun
on we show
try markets. Drawing the awareness-motivation-capability perspective,
how the speed of an MNE's response to a rival's attack is influenced by resource
related factors, including distance, government constraints, and subsidiary control,
and by market-related factors, including initiating country importance, location of the
response in the initiating country, and multimarket contact. We provide a new con
ceptualization and empirical approach for studying rivalry between MNEs. Addition
ally, our theory and evidence imply that important constraints on MNE actions in host
country markets go well beyond the constraints faced in domestic competition.
InAugust, 1998, Renault launched a new version of interfirm rivalry. For example, firms tend to re
the Twingo?its best selling mini?in all themajor to actions that are irreversible, to ac
spond faster
European markets. To fend off the tough competi tions taken in strategically important markets, and
tion from the Fiat Seicento . . .Renault surprised the to actions from rivals with resource endowments
market by cutting the price of Twingo by 15 percent.
and market profiles that are similar to their own
-Farhi, 1998 (Chen, 1996; Chen & MacMillan, 1992).
InApril, 2001, Nissan Motor Co. introduced a rede Although important insights have been gained,
the theoretical development and empirical testing
signed Altima, with themost powerful engines of its
of competitive dynamics research have largely been
product category in theUS, to go head to head with
the Toyota Camry. restricted to domestic (i.e., single-country) settings.
Undoubtedly, many drivers of rivalry in domestic
-Rechtin, 2001 to
settings also apply international settings; how
As multinational (MNEs)
enterprisesstrive for ever, the international context contains factors that
global competitive advantage, they must have no ready counterpart in home-country set
constantly
& Zaheer, 1999). As a result, some
respond to their rivals to defend or improve their tings (Kostova
market positions. When such as relatively homogeneous mar
challenged by Renault, assumptions,
how and where will Fiat respond? Similarly, how kets and competitors, consistent and uniform gov
and where will Toyota counter the threat of new ernment regulations, and tight control over organi
model introduction by Nissan? The answers to zations by their top managers, are called into
these questions are important, because how a firm question when one examines rivalry between
acts and reacts with regard to rivals is a vital deter MNEs in host country markets.
minant of its performance (Porter, 1980). Recogniz To begin filling this gap, we study factors that
ing the importance of examining competitive ac affect the response speed of MNEs in host coun
tions and responses, prior competitive dynamics
tries.
Drawing
on the awareness-motivation-capa
research has identified a number of antecedents of bility perspective, we
show how the speed of an
MNE's response to a rivaPs attack is influenced by
resource-related factors, such as distance, govern
ment constraints, and subsidiary control, and by
The authors gratefully acknowledge the insightful
comments of Ming-Jer Chen, Andrew Delios, Lorraine
market-related factors, such as initiating country
Eden, Associate Editor Duane Ireland, Marta Gelet importance, response occurrence in the country of
attack (within-country response), and multimarket
kanycz, Javier Gimeno, Amy Hillman, Richard Lester,
Marvin Lieberman, Mats Lingblad, Metin Sengul, Ken contact. We model the rivalry ofMNE dyads as an
Smith, Anna Stafsudd, three anonymous reviewers, and unfolding and continuous sequence of actions be
participants at the Arizona State University Brown Bag tween the two firms in all the country markets
Series. Any remaining
errors are our own. where both compete. A response, in our study, is
665
Copyrightof theAcademy ofManagement, all rightsreserved.Contentsmay not be copied, emailed, posted to a listserv,or otherwise transmitted
without the copyrightholder's express
written permission.Users may print,download or email articles forindividual use only.

666 Journal
Academy ofManagement June
defined as any action taken by an MNE against a capability framework and then define and explain
given rival that temporally follows an earlier action the concept of response speed.
by that rival (Ferrier, 2001). The response could
occur in the country in which the attack occurred
The Awareness-Motivation-Capability
(the initiating country) or in some other country
Framework
where both rivals compete. Response speed derives
from the time between a rival's initiating action and Drawing on diverse streams of research on organ
the firm's response. We could refer to our depen izational change, learning, and decision making
dent variable as "time between actions" because we (Allison, 1971; Dutton & Jackson, 1987; Kiesler &
cannot be certain that any given action that we Sproull, 1982; Lant, Milliken, & Batra, 1992; Schell
observe is, in fact, a clear-cut "response" to an ing, 1960), competitive dynamics scholars have
earlier action. We decided to retain the "response suggested three drivers of interfirm rivalry: aware
speed" label because, as we describe later, theory ness, motivation, and (Chen, 1996;
capability
predicts that any action, regardless of how long it Smith, Ferrier, & Ndofor, 2001). According to the
has been planned, is likely to be importantly influ awareness-motivation-capability framework, com
enced by the most recent action taken by a petitors will respond to an action only if they are
competitor. aware of it and have the motivation and capability
Using a number of factors not considered in prior to respond. The awareness-motivation-capability
research, our study of framework is grounded in existing theories. For
relatively microcompetitive
actions complements the international business lit example, considering actions and responses from
erature, which has emphasized more macro com an communication
organizational perspective
petitive actions such as market entry and foreign highlights that the decision-making process behind
direct investment (Delios & Henisz, 2000; Graham, competitive actions is an information-seeking and
1990; Knickerbocker, 1973). Moreover, our study information-processing activity (Huber & Daft,
empirically tests the promising awareness-motiva 1987; Smith & Grimm, 1991; Smith, Grimm, & Gan
tion-capability perspective with a data set of com non, 1992). The level of awareness is important
petitive actions and responses undertaken by 13 because it affects the extent to which a firm com
global automakers over seven years. In so doing, we prehends the consequences of its actions and its
extend competitive dynamics research into the rivals' actions within the competitive landscape
global arena. Finally, our study introduces a new (Chen, 1996).
analytical methodology to competitive dynamics A firm can be aware of its rivals and the compet
research, modeling dyad-level competitive interac itive environment without being motivated to re
tion as a series of event histories. The event history spond. Expectancy-valence theory provides themi
approach resolves several
important problems with crolevel underpinnings for the motivation and
modeling response speed and permits us to simul capability dimensions of the awareness-motiva
taneously model response behavior over a range of tion-capability framework. Vroom (1964) argued
host countries. that two basic conditions underlie the proclivity to
act: the subjective reward value (valence) of acting
effectively, and the expectation or perceived prob
ability of earning a reward (expectancy). Thus, the
THEORETICAL BACKGROUND
motivation to respond will be greatest when the
Our study builds on competitive dynamics re potential responder feels that something important
search, which emphasizes the active, energetic, and is at stake (Atkinson, 1964).
purposeful process by which firms interact. Com Finally, even if an attack puts much at stake, the
petitive dynamics research has recently made two responder still must have the capability to respond
important advances. The first is the development of (Vroom, 1964). The resource-based view of the firm
the awareness-motivation-capability framework, highlights the importance of heterogeneous re
which is the theoretical anchor of our study (Chen, sources and capabilities in defending and sustain
1996). The second is a process-based
approach to ing competitive advantage (Barney, 1986; Dierickx
examining competitive interaction
(e.g., Ferrier, & Cool, 1989; Peteraf, 1993). Thus, if the resources
2001). The process-based approach highlights that and capabilities for responding are not available, a
modeling response speed in a sequence of actions firmmay delay its response or not respond at all.
and responses is critical to revealing the pattern of The awareness-motivation-capability framework
competitive interaction. In the remainder of this has been increasingly adopted by strategy research
section, we discuss the theoretical foundations and (e.g., Chen & Su, in press; Ferrier, 2001; Ferrier,
key components of the awareness-motivation Smith, & Grimm, 1999). Our study extends this

2007 Yu and Cannella 667
framework into a global context. Our extended stay ahead of one another. In this race, we believe,
awareness-motivation-capability framework im any action launched against a given rival can (and
that will to a rival's ac likely does) include elements of response to the
plies competitors respond
tion to the extent that they are (1) aware of the rival's most recent action, elements of response to
action and of the country where the action took earlier actions by that rival, elements of indepen
place, (2) motivated to respond to the action and to dent initiative, and even elements of response to
defend or expand their global presence, and (3) actions by that rival. Furthermore, our
anticipated
capable of deploying resources for responding in notion of response fits the global auto industry,
host countries. where high capital intensity and long planning ho
rizons make the link between actions and re
sponses somewhat tenuous (Smith, Ferrier, &
Response Speed Ndofor, 2001).
Porter noted that "finding strategic moves that
will benefit from a delay in retaliation, or making
HYPOTHESIS DEVELOPMENT
moves so as to maximize the delay, are key princi
ples of competitive interaction" (1980: 98). Theo Ourstudy describes several factors that are im
retically, examining response speed is important, portant to rivalry between MNEs, but not consid
because it has vital implications for firm perfor ered in previous research, primarily because past
mance and competitive advantage. Researchers research has been limited to single-country set
have shown that slow-responding rivals often ex tings. To select the factors, we evaluated the central
market share losses or missed from research on MNEs and used the
perience profit op concepts
portunities (Ferrier et al., 1999). awareness-motivation-capability framework to
the lead of Ferrier (2001), our study identify those with the most promise as anteced
Following
takes a process-based approach tomodeling rivalry ents of response speed. To be included, a concept
between MNEs. In this approach, a response is had to impact one ormore dimensions of the frame
defined as any action by an MNE against a given work. We organized our factors into resource
rival that follows an earlier (initiating) action by related and market-related categories. In the re
that rival. The response could occur in the country source-related category, we identified distance,
where the initiating action occurred or in any other government constraints, and subsidiary control.
country or countries where both dyad partners Each of these directly affects the resources needed
compete. Accordingly, response speed derives for an effective response. In the market-related cat
from the time in days between a rival's initiating egory, we identified initiating country importance,
action and an MNE's observed response. Note that response occurrence in the country of attack (with
each action taken by an MNE against a rival creates in-country response), and multimarket contact.
a set of response opportunities?one for each coun Each of these directly relates to the market depen
try where both firms compete. In our conceptual dence and market presence ofMNEs.
ization, an action taken by firmA against rival firm Before we explain in detail how each factor af
B is the starting point for modeling firm B's re fects response speed, it is important to clarify sev
sponse. Firm B's response, which could occur in eral terms. We study MNE dyads, or pairs of rival
any combination or permutation of countries where firms. In our study, the initiator is the MNE that
both compete, is both the response to firm A's took an initial action, and the responder is theMNE
initiating action and the action that initiates the that is expected to respond. Home country refers to
start of a new set of response opportunities (i.e., for an MNE's nation of origin?whether it is initiator or
firmA). Each action is therefore a response (ending responder. Host country refers to any country
one set of response opportunities) and an initiating where an MNE competes except its home country.
action (starting a set of response opportunities). Initiating country refers to a country where a rival's
Like Ferrier's (2001) concepts, our concepts of re attack takes place.
sponse and response speed emphasize accuracy in
identifying the timing and target of action, yet we Resource-Related Factors
do not strongly assert that a given action is a clear
cut and direct "response" to the earlier action by a Much research bolsters the notion that strategies
rival, though it does follow that action sequentially. are constrained by, and dependent on, both the
We believe our notion of response effectively resources available to a firm and factors related to
captures the spirit of the competitive interaction the effective deployment of resources (Barney,
process. Schumpeter (1947) described interfirm ri 1991). The global context in which MNEs operate
valry as a race among competitors to get ahead and features a number of factors that directly affect re

668 Journal
Academy ofManagement June
source availability and/or resource sponse to a new four-cylinder engine developed

deployment but by
are of much less (or no) importance in domestic Honda, General Motors to introduce its
planned
contexts. We chose three factors from the literature own global four-cylinder engine, called the L850,
on MNEs?distance, government constraints, and in the United States and Europe. However, it took
subsidiary control?that are germane to global set GM five years to eventually produce the L850 in
tings and have significant impacts on resources Europe. The engineering director of GM North
needed by MNEs in responding to rivals. Further, America, Otto Willenbockel, noted, "We know how
each of these directly affects one or more of the to design engines, but we don't know how to
design
three dimensions of the awareness-motivation an engine that is suitable for the other group . . .
capability framework. this may be one reason it was not a very quick
. ..
project. Essentially, Europeans want the maxi
mum speed and best cruising feel, while we want
Distance
the best launch torque with the best possible fuel
Ghemawat argued that even in the age of the economy" (Robinson, 1999: 5).
Internet and global communications networks, Home-initiating country distance. Hypothesis 1
"distance still matters" (2001: 137). Focusing on centers on the geographic distance between an
geographic distance, we contend that geographic MNE's home country and a host country where a
distance impedes an MNE's capacity to transfer response may occur. In Hypothesis 2, we focus on
resources and absorb knowledge across countries. the geographic distance between an MNE's home
As a result, the geographic distance between an country and the initiating country where a rival's
MNE's home and a given host country, and be attack has taken place. We argue that when an
tween an MNE's home and an initiating country, attack occurs in a geographically distant country,
will hamper the MNE's response speed. the MNE's awareness of the action as well as its
Home-host country distance. Geographic dis motivation to respond swiftly are hampered. Large
tance between an MNE's home and a given host geographic distance hinders the collection and in
country will constrain the MNE's capability to act terpretation of information about an attack (Ghema
in that country and thereby slow its response. Geo wat, 2001; Ghoshal & Kim, 1986) and makes itmore
graphic distance adds the cost and difficulty of difficult for headquarters managers to recognize the
internal coordination, increasing the resources re attack, understand the intentions behind it, and
quired for responding. As a result, it becomes dif effective response strategies. Furthermore,
develop
ficult for subsidiary managers to transfer their in research on international business, the strategic
knowledge of a local market back to headquarters importance of their home countries forMNEs has
and make swift competitive decisions (Ghemawat, long been recognized (Casson, 1987; Graham,
2001). Moreover, although subsidiaries contribute 1990). This logic suggests that when a rival attacks
significantly to the overall development of an MNE, an MNE in a country that is "close to home," the
they depend on headquarters for critical resources MNE will be more threatened and more motivated
such as capital, innovation capacity, and manage to respond quickly. Therefore:
rial expertise (Ghoshal & Bartlett, 1990; Veugelers,
2. The speed of an MNE's
1995). The resources and capabilities ofMNE head Hypothesis response
are less to in a host country to an action taken by a rival
quarters probably helpful geographically
distant subsidiaries. in any country is negatively associated with the
Furthermore, distance makes
the activity sharing and synergy formation be geographic distance between theMNE's home
tween headquarters and subsidiaries less efficient country and the initiating country (i.e., where
& Ma, For the action was taken).
(Golden 2003; Porter, 1987). example,
Davidson andMcFetridge (1985) found thatMNEs
were more likely to transfer technology to a foreign Government Constraints
subsidiary when the home and receiving countries
were geographically closer. Therefore: Governments can affect the strategic actions and
1. The speed of an MNE's

survival prospects of foreign-based MNEs by creat
Hypothesis response
in a host country to an action taken by a rival ing barriers to their access to critical resources and
increasing their cost of doing business (Delios &
in any country is negatively associated with the
Beamish, 1999; Loree & Guisinger, 1995; North,
geographic distance between theMNE's home
1990; Rugman & Verbeke, 1990). We define govern
country and the host country.
ment constraints as rules and regulations that
Anecdotal evidence supports the argument be hinder MNEs' entry and daily operations. As im
hind Hypothesis 1. For example, in 1994, in re portant players in international markets, govern

ments can constrain the resources and actions of result, an MNE will be more motivated to respond
foreign-based MNEs operating in their domestic in host countries to rivals' attacks. Second, govern
markets (North, 1990) and enhance the resources ment constraints in an MNE's home country also
and actions of home-based MNEs when they com enable it to leverage the advantages it obtains at
pete abroad. home and increase the capability of its subsidiaries
Host government constraints. We propose that to compete overseas. Krugman (1984) and Venables
host government constraints hinder the motivation (1984) showed that home government protection
and capability of an MNE to respond to rivals' lowers domestic firms' average and marginal costs,
attacks. First, government constraints, such as laws helping them "move down the learning curve"
and regulations on
importing exporting and of more quickly than rivals and increasing their capa
goods, attainment of government licenses, taxes, bility to respond. Some authors have argued that
and availability of loans, directly affect an MNE's firms in protected countries tend to lack the skills
or motivation to engage in rivalrous behavior
capability to respond swiftly (Davidson, 1980; Hen
isz & Delios, 2001; Loree & Guisinger, 1995). Con abroad (Marcus, 1990; Rugman & Verbeke, 1990).
sider the Japanese automakers' attempt to enter Eu Although this argument may hold under certain
rope as an example. In 1999, the top three Japanese circumstances, in industries characterized by high
makers, Honda, Nissan, and Toyota, in response to levels of globalization and strategic interdepen
European brands' successful entry into Asia,
were dence (such as auto manufacturing), we believe this
all eager to gain market share in Europe. However, effect is of much less concern (Veugelers, 1995).
for almost a year their performance was disappoint Therefore:
ing. Akira Imai, president of Toyota Motor Europe, 4. The speed of an MNE's
Hypothesis response
complained, "Our imports were hit badly by the 10
in a host country to an action taken by a rival
percent duty; this ismuch higher than the 8 percent in any country is positively associated with the
in Korea or the 2.5 percent in the USA . . .we
level of government constraints in theMNE's
believe it is . . . unfair. . . We
. know we can't
very
home country.
raise market share quickly and dramatically with
out the danger of a trade war" (Kurylko & Catterall, Wenote that governments can also use aggressive
2000: 3). Second, host government constraints cor policy instruments such as export subsidies and
respond with political hazards (that is, unpredict diplomatic negotiations to improve the competitive
able political institutions). Such hazards increase environment for home-based MNEs. However, our
uncertainty for both headquarters and subsidiary study is limited to defensive policies such as con
managers when they try to foresee future invest straints on foreign MNEs' daily operations. Further,
ment conditions (Henisz & Delios, 2001; North, ithas been argued that government protection hurts
1990) and thus reduce their motivation to respond consumer welfare and results in economic distor
quickly. Therefore: tions (Marcus, 1990) and that different government
3. The speed of an MNE's protections have different impacts on foreign direct

Hypothesis response
investment (Ellingsen & W?rneryd, 1999). How
ever, these issues are beyond the scope of our
in any country is negatively associated with the
level of government constraints in the host study.
country.
Subsidiary Control
Home government Building on the
constraints.
awareness-motivation-capability framework, we Subsidiary control refers to the strength of the
develop the notion that a "safe backyard," well link between an MNE's headquarters and its sub
protected by an MNE's home government, in sidiary in a given host country (we assume that
creases the MNE's motivation and capability to re headquarters control is uniformly strong in the
spond in host countries. First, according to multi home country). Conceptually, subsidiary control
market competition research, firms tend to refrain captures the efficiency and effectiveness of infor
from competing aggressively in each others' home mation flows between headquarters and subsidiar
countries because they recognize and respect each ies as well as the extent to which MNE headquar
others' "spheres of influence" (Brander & Krugman, ters can dictate the operating and strategic
1983; Casson, 1987; Graham, 1990; Veugelers, decisions of its foreign-based subsidiaries (Golden
1995). However, home government constraints & Ma, 2003). We propose that strong subsidiary
change the perceived payoff associated with re control increases an MNE's motivation and capa
sponse. A safe backyard reduces uncertainty bility to respond in a given host country.
caused by the fear of rivals' revenge at home. As a Greater control over a subsidiary gives an MNE

670 Academy ofManagement Journal June
headquarters more leverage to affect the subsid larly, previous studies have also shown that the
iary's competitive strategy. The subsidiary will also market environment shapes competitive strategies
be more motivated to consider the integrated global (Miller & Friesen, 1984). In line with these argu
market and respond quickly to a competitor's chal ments, we expect that the response speed of an
lenges in light of the gains to the entire MNE. For MNE is affected by three market-related factors?
example, in August 2001, Ford substantially in initiating country importance, response occurrence
creased its control over its British subsidiaries. As in the country of attack (within-country response),
the president of the Premier Automotive Group, and multimarket contact.
Mike O'Driscoll, put it, "The driving force behind Initiating country importance. Previous re
this is the marketplace. ... search suggests that the degree of threat an action
By centralizing control,
the new organizations will intensify their efforts to poses to a potential responder is a strong predictor
fightwith their rivals in the market place" (Brown of response speed. For instance, Chen and MacMil
& Henry, 2001: 3). Moreover, as we mentioned ear lan (1992) found that the greater a firm's depen
lier, to respond to rivals' attacks, subsidiaries de dence on the markets affected by an attacker's ac
pend on headquarters for some critical resources. tion, the quicker the response. Tsai,
Similarly,
Increased subsidiary control leads to administra MacMillan, and Low (1991) showed that competi
tive mechanisms that facilitate the flow of knowl tors who are highly dependent on a given market
edge, information, and resources between head respond aggressively to the threat of a new entrant
quarters and subsidiaries (Doz & Prahalad, 1981). into that market. In keeping with these arguments,
This flow of knowledge, information, and resources in Hypothesis 6 we propose that the strategic im
increases headquarters' awareness of rivals' attacks portance of an increases an
initiating country
as well as subsidiaries'
capability to respond. For MNE's awareness of an attack and enhances its
example, in October 1998, GM increased its control motivation to respond quickly. First, an MNE is
over many overseas more
subsidiaries. GM hoped the likely to be aware of an action taken in its
new arrangements would cut red tape and speed important markets because such actions will gen
GM vehicles to markets around the world. "This erally be considered more threatening. Although
[new arrangement] allows us to better control the the MNE may initially be uncertain of the implica
global operations and
respond quickly to rivals' tions of an attack, the importance of themarket will
actions, while at the same time leveraging the ex motivate a quick response. Second, in deciding
perience, talent and diversity of our people world how to respond to a rival's attack, a firm must
wide," said Rick Wagoner, president of GM Auto balance the payoff for responding late against the
motive Operations (Kurylko, 1998: 3). Therefore: payoff for responding swiftly. In this decision-mak
5. The speed of an MNE's ing process, the share to be gained or lost by the
Hypothesis response firm is an important determinant of the firm's pay
in host country to an action taken by a rival
a
off. Since firms generally are very sensitive to the
in any country is positively associated with the
revenues and profits derived from key markets, the
MNE's degree of control over its subsidiary in more attackers' moves occur in their key markets,
the host country.
the greater the response payoffs involved, and the
We note that conflicting views exist in the inter more motivated the firms will be to launch speedy
national business literature regarding the implica responses. Therefore:
tions of headquarters control for subsidiaries' inter
Hypothesis 6. The speed of an MNE's response
national activities. Some have argued that high
levels of global integration restrict the autonomy of
subsidiaries and reduce their flexibility and re
& & strategic importance of the initiating country
sponsiveness (Baliga Jaeger, 1984; Gates
(i.e., the country where the action was taken) to
Egelhoff, 1986). Although such reduction may have
theMNE.
a chilling effect on a subsidiary's response speed,
we believe that an effective MNE
competitive strat Within-country response. There are several rea
egy should involve coordinated actions and reac sons to believe that the speed of an MNE's response
tions across subsidiaries enforced by sufficient cor to an attack will be faster in the initiating country
porate control (Prahalad & Doz, 1987; Yip, 1995). (the country where the action occurred). All of the
reasons are linked to motivation. First, although
some research predicts cross-country responses
Market-Related Factors
among multimarket competitors (e.g., Gimeno,
Porter (1980) noted that market conditions 1999), and some initiating actions might well pro
strongly influence the intensity of rivalry. Simi voke responses in countries other than the initi?t

ing country, the application of the awareness ket contact facilitates mutual awareness because
motivation-capability framework suggests that an firms become familiar with each others' intentions
MNE's motivation to respond will be very strong in and capabilities. Thus, more multimarket contact
the initiating country, because this is where the with a rival implies a greater likelihood that an
signaling value of responding will be maximized initiating action from that rival will be noticed.
(Axelrod, 1984; Clark & Montgomery, 1998). Re Second, multimarket contact captures the extent to
search has shown that a firm's reputation for being which two firms are in direct competition. The
capable and willing to defend its markets lowers interdependence between rivals increases the "va
the expected value of attack for rival firms (Clark & lence" of responding to each others' attacks. Young
Montgomery, 1998; Kreps & Wilson, 1982; Milgrom and colleagues (Young, Smith, Grimm, & Simon,
& Roberts, 1982). Thus, it is strategically important 2000) argued that as a firm's multimarket contact
for a firm to respond in the country where it has with a rival
increases, its assets become increas
been challenged. Furthermore, both theory (e.g., ingly threatened by that rival. As a result, the firm's
Porter, 1980) and evidence (e.g., Ferrier & Lee, motivation to deter the rival from competitive be
2002; Ferrier et al., 1999) suggest that not respond havior also increases. Moreover, a quick response
ing in an initiating market will lead tomarket share signals a strong commitment to defending a market
erosion in that market. Although the cited studies position (Axelrod, 1984; Clark & Montgomery,
were focused on single-country settings, their con 1998). If a rival feels its attack will engender a quick
clusions seem applicable to the global setting as response and the competitive advantage from the
well. Therefore: attack will be short-lived (Camerer, 1991; Weigelt &
Camerer, 1988), it is less likely to attack in the first
in a host country to an action taken by a rival place (Young et al, 2000). Finally, Chen (1996)
argued that if a competitive response to a
in any country is increased when the host challenge
is where the initiating action took requires major organizational restructuring, the re
country rivals are
sponse will be slowed. Since multimarket
place. to we
likely have similar strategic competencies,
Multimarket contact. Multimarket contact oc expect that more multimarket contact will be asso
curs when firms encounter the same rivals inmul ciated with more capability to respond. Therefore:
tiple markets.1 Research in IO (industrial organiza
tion) economics, organization theory, and strategic in a host country to an action taken by a rival
management implies that the behavior of firms in
any single market is affected by the extent towhich contact between
degree of multimarket the
they meet the same competitors in other markets
MNE and its rival.
(Baum & Korn, 1999; Bulow, Geanakoplos, & Klem
perer, 1985; Evans & Kessides, 1994; Gimeno,
1999). Multimarket contact gives a firm the option METHODOLOGY
to respond to actions by a rival not only in the
Data Description
market being challenged, but also in other markets
where they both compete. As a result, multimarket To test our hypotheses, we examined the com
competitors may hesitate to attack in one market for petitive interactions of global automobile manufac
fear of retaliation in other markets (Edwards, 1955). turers from 1995 to 2001. The global automobile
However, how will a firm respond when a mul industry was a good setting for our study for several
timarket competitor challenges it? Drawing on the reasons. First, the industry is known for intense
awareness-motivation-capability framework, we rivalry among an identifiable and manageable set of
believe that the level of multimarket contact an firms (Martin, Swaminathan, & Mitchell, 1998;
MNE has with a given rival will increase its aware Thomas & Weigelt, 2000). In this oligopolistic set
ness of the rival's attack and enhance itsmotivation ting, the of among
degree strategic interdependence
and capability to respond swiftly. First, in an oli the major players is high. As a result, the actions
gopolistic setting like the auto industry, multimar undertaken by one firm are likely to have signifi
cant impacts on other competitors and, accord
ingly, the outcomes of actions hinge on the re
1
Whether a multimarket contact is unintended or pur sponses. Second, distinct (national) markets can be
readily identified, along with the competitors in
poseful is outside the scope of our research. Our study
focuses on realized multimarket contact each market (country). Additionally, country-mar
only (e.g.,
Gimeno, 2002). We thank an anonymous reviewer for ket characteristics around the globe are relatively
bringing this point to our attention. heterogeneous. Third, information about the indus

672 Aca demy ofMan agem en t Journ al June
try and each of the countries involved is widely sified into one of six action types: capacity action,
available. Finally, for firms in the industry, auto major product action, pricing action, marketing ac
mobile manufacture is the major line of business, tion, distribution and service improvement, and
so the potential for confounding influence from minor product action. If one action involved mul
diversification is minimal. tiple categories, we coded it as a single action but
We identified the 13 largest firms in the industry with multiple action types. For instance, GM in
from the 1995-2001 volumes ofWard's Automotive vested $100 million in China to build two new
Yearbook. The firms were DaimlerChrysler, Fiat, plants and introduce a new model. We coded this
Ford, General
Motors, Honda, Hyundai, Mitsub action as one action with two action types: capacity
ishi, Nissan, PSA Peugeot Citroen, Renault, Suzuki, action and
major product action. Finally, we
Toyota, and Volkswagen.2 Between 1995 and 2001, screened the data to remove duplicate reports of
our sample firms accounted for between 76 and 88 competitive actions. Appendix A provides more
percent ofworld motor vehicle production. The 27 detail on the distribution of actions over compa
countries in our sample together represented ap nies, countries, and years. Table 1 gives details on
proximately 99 percent of world motor vehicle the six action categories we identified.
sales during the study period. Survivor bias among When Automotive News reported an action, it
subsidiaries was of minimal concern in our study. typically specified the country or countries where
During our study window, we observed only three the action was undertaken. When an action in
cases inwhich a sample firmwithdrew from a host volved several countries, we noted each. When
country. only a multicountry region was mentioned, we as
sumed the action occurred in all the important
countries of that region. For example, when a new
Identification of Competitive Actions
model was introduced in Western Europe, we
Data on competitive actions were gathered from counted the action as occurring in each of the five
Automotive News. Established in 1925 and deliv key auto markets in Western Europe?the United
ered weekly tomore than 80,000 subscribers in 70 Kingdom, Germany, France, Italy, and Spain. When
countries, Automotive News offers complete and no information was provided regarding the location
detailed reports on global automakers' announce of an action (a very rare event), we used the geo
ments and actions. We used structured content graphic classification of the news article to identify
or countries where it applied. Most of
analysis to identify competitive actions (Jauch, Os the country
born, & Martin, 1980; Miller & Friesen, 1977). First, our actions [77%) involved only one country, and
one author read all the articles published by Auto 97 percent involved fewer than six countries.
motive News in 1995 and generated a list of key To check the overall accuracy and comprehen
words (e.g., "rivalry," "competition," "war") that siveness of Automotive News as a data source for
were likely to indicate competitive events (a full competitive actions, we drew 30
competitive
ac
list of key words is available from the first author tions at random and searched for references to them
upon request). We then searched all Automotive in other major business publications. We found 26
News articles between 1995 and 2001 to identify of the 30 actions (87%) in other publications and
those that mentioned one or more of our sample confirmed the details reported by Automotive News
firms and included at least one keyword in our list. in every case.
This step yielded 6,648 news articles. We then

carefully read each article and identified 2,207 dy
Dependent Variable: Response Speed
ad-level competitive actions. Each action was clas
As we discussed earlier, we used the number of
days between a rival firm's action against a focal
2
The automobile industry has undergone
some con firm and the subsequent response action of the
solidation in recent years. We used an annual measure of focal firm against the same rival to measure re
industry concentration to control for changes in industry sponse speed. The response could occur in the
structure. Our sample firms had quite stable ownership
country inwhich the initiating action occurred (the
structures in the period 1995-2001, with the exception of
initiating country) or in any other host country
DaimlerChrysler, which was created by merger in 1998. where the dyad partners both competed.
Considering thewell-documented difficulty encountered
inmerging the two firms,we treated DaimlerChrysler as
an American firm. In analyses available from the first Variables
Independent
author upon request, we dropped DaimlerChrysler from
the sample. The conclusions from that analysis do not Distance. Geographic distance was measured as
differ from those presented here. the logarithm of the number of kilometers between

TABLE 1
Descriptions and Examples of Competitive Action Types
Action Type Description Example
action Price-related actions such as price In the United States, Ford offered pick-up owners
Pricing
cuts, rebates, and discounts a $500 discount coupon to buy a 1999 Ford
F-150 or F-250 or a 2000 F-150.
Major action Product actions involving Honda launched the Insight, a brand new
product
substantial investment, brand gasoline-electronic two-seat coupe in Europe.
new introductions and
product
key technology breakthroughs
action actions such as GM launched a Hispanic-targeted marketing
Marketing Marketing-related
marketing campaigns, campaign in California.
advertisement investments,
innovation on marketing
methods, and brand
management
Minor action Product redesign, product In the United States, Dodge redesigned the light
product
modification, use of new duty Ram for the 2002 model year.
technology (on minor features),
annual model introduction
action in company's capacity or Toyota decided to build the second factory in
Capacity Changes
output Valenciennes, France, to produce a new
vehicle, the Verso microvan.

in distribution in company's mounted an ambitious to set up its
Improvement Changes Chrysler push
and after-sales service distribution systems and after dealer network outside Bangkok.
sales service
the capital cities of an MNE's home country and the same measure (albeit for different countries) to
country of interest. We measured both home-host gauge host government constraints and home gov
country distance, and home-initiating country dis ernment constraints.
tance. We also gauged cultural distance, using four Subsidiary control. An MNE can enter a foreign
dimensions of the Hofstede index (Hofstede, 1980). market in a number of ways: via a wholly owned
In analyses available from the first author upon subsidiary, joint venture, strategic alliance, tech
request, we reached the same conclusions with cul nology license, manufacturing collaboration, or
tural distance as we report for geographic distance. other form. These forms of entry are quite hetero
However, given the concerns raised about the Hof geneous in terms of the degree of control corporate
stede index (e.g., McSweeney, 2002), we report can exert over subsidiaries in host
headquarters
only geographic distance results here. markets (Harrigan, 1985). According to Gatignon
Government constraints. The degrees of home and Anderson, "Higher operational control results
and host government constraints were assessed from having a greater ownership in a foreign ven
with the Executive Opinion Survey conducted each ture" (1988: a proxy for subsidiary control,
3). As
year by the World Economic Forum. This survey corporate ownership has been found to influence
collects information on the perceptions of 4,000 top the types of reports and mandates exchanged be
executives about the competitive characteristics of tween MNE headquarters and subsidiaries
59 countries. Three variables from the survey were (Egelhoff, 1988) as well as the motivation of sub
selected. Average responses to the statements, "Ad sidiary managers in their dealings with headquar
ministrative procedures are not an important obsta ters and other (Martinez & Jarillo,
subsidiaries
cle to starting a new business," "Government regu 1989). To capture subsidiary control, we used the
lations do not impose a heavy burden on business percentage of a subsidiary's equity held by itsMNE
competitiveness," and "Irregular payments con headquarters (Curhan, Davidson, & Suri, 1977; De
nected with import and export permits, business lios & Henisz, 2000). Who Owns Whom: The Direc
licenses, exchange controls, tax assessments, police tory of Corporate Affiliation was our source for
protection or loan applications are not common" equity ownership data.
were reverse-coded to reflect MNE managers' per Initiating country importance and within-coun
ceptions of restrictions on their business. We then try response. We gauged initiating country impor
factor-analyzed the three variables and created a tance as the proportion of a responding MNE's
measure =
composite [a .85). Note thatwe used the sales that occurred in the initiating country. To

capture whether or not a response was a within market concentration, measured as the percentage
country response, we included a dummy variable, of a country's total sales represented by the four
within-country response, indicating whether or not largest competitors in that country. As a final char
the focal host country was also the initiating acteristic of the initiating country, we included a
country. dummy variable, initiator's home, coded 1 if the
Multimarket contact. To
capture the degree of initiating country was also the initiator's home
multimarket contact between rivals, we chose the country and 0 otherwise.
dyadic measure developed by Baum and Korn To control for the characteristics of a responder,
(1999). This measure takes into account the strate we included responder size, measured as the log of
gic significance (centrality) of each country the two the firm's world production (number of vehicles) in
firms compete in (the proportion of each firm's a given year. Research has shown that firms of
total sales represented by each country). It is im varying sizes may possess different levels of re
portant to consider the sales centrality of the mar sources and have different propensities for taking
kets where two competitors meet because, as Baum action. We also controlled for responder interna
and Korn pointed out, "A smaller number of con tional experience, because firms with more inter
tacts in high-centrality markets . . . can national experience can better deal with the liabil
yield higher
contact than a larger number of con
multimarket ity of foreignness and respond more quickly
tacts in low-centrality markets" (1996: 262). (Anand & Khanna, 2000; Delios & Beamish, 2001;
We used the measure below to capture the level Erramilli, 1991; Li, 1995). We measured a firm's
of multimarket contact between two firms, i and /, international experience using the transnationality
at time t (for all XMit[Dimt X > 1, otherwise = index developed
Djmt] by United Nation's World Invest
0): ment Report, which is the average of three ratios:
foreign sales to total sales, foreign assets to total
x [Dimtx x [(Dimtx
^Mit[Cimt Djmt)] + ^Mjt[Cjmt Djmt)] assets, and foreign employment to total
Mit +
Mjt employment.
We included several measures to control for the
where m denotes a given market in the set ofmar
characteristicsof an initiating action. Prior studies
kets Mit or served by either or both firms i and ;
Mjt have found that competitors are more motivated
at time t. Cimt and are the centralities (strategic
Cjmt and better able to respond quickly to tactical ac
importance) of market m to firms i and ; at time t
tions than to strategic ones because strategic ac
As noted earlier, in including strategic impor tions generally involve larger commitments of re
tance?centrality?we recognized thatmultimarket
sources and planning (Chen & MacMillan, 1992;
contact increases when two firms meet in impor
Smith, Grimm, Gannon, & Chen, 1991). We split
tantmarkets, relative to the contact between similar
our six action types into two groups: major moves
dyads thatmeet in smaller and less important mar
kets. Dimt and are indicator variables set equal (capacity actions and major product actions) and
Djmt minor moves
to 1 if firms i and / are active inmarket m at time t (pricing actions, marketing actions,
distribution and service improvements, and minor
and 0 otherwise.
product actions) and we created a dummy variable,
major initiating move, coded 1 for a major move
Control Variables
and 0 otherwise. Similarly, we controlled for the
magnitude of an initiating move using the total
As noted
earlier, prior research has established number of countries involved in an initiator's at
that competitive rivalry can be importantly influ tack with the variable number of initiating coun
enced by the characteristics of (1) the host country, tries. Finally, we controlled for the complexity of
(2) the initiating country, (3) the responder, (4) the the initial move, since firms tend to respond more
initiating action, and (5) the initiator (Smith, slowly to more complex actions. As noted earlier,
Grimm, & Gannon, 1992). To ensure a fully speci some actions involved more than one action cate
fied model, we included measures of these charac gory. Complexity of initiating move was the total
teristics as controls. number of action categories involved in an initia
To control for host country characteristics, we tor's action.
included host GDP growth, measured as the annual To control for the characteristics of an initiating
percentage growth rate of a country's gross domes firm, we included a variable for the geographic
tic product (GDP), as reported in theWorld Devel distance between the focal firm and the initiating
opment Indicator database. Studies suggest that firm, dyad partner distance. Competitors that are
firms compete less aggressively in high-growth closer geographically will find it easier to under
countries (Porter, 1980). We also controlled for host stand each other's strategic intentions, and they are

likely to be perceived similarly by customers as likelihood of response (Box-Steffensmeier & Jones,

well. As a result, actions initiated by a geographi 2004; Turna & Hannan, 1984).
as a Given the merits of the event history approach,
cally close rival will generally be perceived
we used continuous-time event history analysis to
greater threat.
model the speed of global automakers' competitive
responses in all countries where two rivals (an
Analytical Methodology MNE dyad) competed. Our analyses derived from
Prior studies in competitive dynamics have mod Cox proportional hazards regression models (Cox,
eled both response likelihood and response speed 1972) of the followingform:
(Chen, Smith, & Grimm, 1992). The analytical =
of research raised some signif Ht) h0(r) x exp{S?ic X [Xik[t)]},
methodologies prior
icant concerns in our context. For example, when where h0(r) is the baseline hazard function, and
modeling the time between an action and a re
Xik[t) is the value of the ?th covariate (independent
sponse, a researcher might be inclined to use ordi
variable) for firm i at time t.A key benefit of the Cox
nary least squares (OLS) regression, with response model is that no assumptions are made as to the
time as the dependent variable. However, this ap distribution of the baseline hazard function. For
proach is fraught with problems. First, because ob examples of management research using the Cox
servation windows are finite, no response might be see Katila and Shane
approach, (2006) and Spell
observed during the study period?a situation and Blum (2005).
known as right censoring. If this is the case, how In our data set, the unit of analysis was firm
are nonresponders coded? Deleting them would the
competitor-country (firm dyad-country), and
bias the response time downward. Coding them as failure event (in event history terminology) was the
the number of days observed without a response occurrence of a competitive response. As described
would also bias the estimates downward, as they
earlier, a firm is at risk of responding to its rival (its
are coded as responding when they did not. Fur dyad partner) when that rival acts against it in one
ther, in an OLS regression of response time, there is or more countries where both compete. Starting on
no viable way to specify the model's disturbance the date of the first action taken by either dyad
term; that is, one cannot be assured that the error
partner after January 1, 1995, we noted the number
term will be unbiased (Box-Steffensmeier & Jones, of days that elapsed until either (1) the responding
2004; Turna & Hannan, 1984). firm responded in any country or countries where
Event history analysis resolves these problems by both or (2) the same initiator took an
competed
modeling not the response time, but the hazard other action against the responding firm in one or
rate: the likelihood that a response will be observed more countries where both competed (with no re
at time t, given that no response occurred prior to
sponse to the first action).3 In both cases, we noted
time t. Because they model a conditional likeli the time in days between the two actions. Then, for
hood, event history methods are not biased by right
each initiating action, we created a "set" of obser
censoring. In our methodology, each action taken
vations, one for each country where both dyad part
by one firm against another established an oppor ners competed. The response time was the same for
tunity to respond in every host country in which each observation, but the outcome was not. We
both rival firms competed. However, as we will
coded observations conforming to case 1 above as
describe, most often a response occurred in only
("failures," because the modeled out
"responses"
one of these countries, and data from the rest were come was observed) for all of the countries where
right-censored at the time of the response. There the response took place, and as "censored" (be
fore, a method that was not biased by this right cause the modeled outcome was not observed by
censoring was critical. Another key advantage of the end of the waiting period) for all other obser
modeling a hazard function was that, given a haz
vations in the set (other countries where both rivals
ard function, we could directly calculate a response
speed (the average number of days until response,
for example). A significant increase in the hazard 3
Because we we any interac
function was directly interpreted as a significant analyzed dyads, ignored
tions between dyad members and other firms outside a
increase in response speed. Also, given a fixed time
dyad. For example, ifToyota attacked Ford, we wait for
interval starting with an initiating action, we could either Ford to respond or Toyota to attack Ford again. In
also convert the hazard rate directly into the re themeantime, Ford may attack and be attacked by other
sponse likelihood for that time interval, as well as firmsbesides Toyota, but these are only modeled in the
interpret the significance of the change in hazard cases inwhich Ford ismatched in dyads with these other
directly as the significance of the change in the firms.

676 Academy Journal
ofManagement June
competed). Case 2 was coded as "censored" for all also analyzed within-country response (i.e., re
observations in the set. Finally, there could be a sponse behavior limited to a country of initiation)
third case?the observation period ending before and reached conclusions that parallel those re
the focal firm responded. In these cases, all obser ported here for the hypotheses that could be tested
vations in the set were right-censored at December in a within-country setting.
31, 2001, and time was coded as the number of days
between the initiating action and that date. Appen
dix B provides an example that will help to clarify
RESULTS
this somewhat complex procedure.
After all observations were coded, we deleted all Table 2, providing means, standard deviations,
home country observations, for several reasons. and correlations for all variables used in our anal
First, our theory is very specific to host countries, yses, suggests no critically collinear variables exist
and it does not apply to home countries directly. in our data (e.g., r > .8 [Kennedy, 2003]). Further,
Second, many of our independent variables only using OLS regression with number of days to a
make sense in host markets (e.g., home-host coun response as the dependent variable, we calculated
try distance, subsidiary control, host government variance inflation factors (VIFs) for all the indepen
constraints). Finally, firms are likely to exhibit dent variables in our analyses. All VIFs were
somewhat different competitive behavior in their within the 1.03-1.30 range. Finally, in analyses not
home countries (Chen, Yu, Michel, & Cannella, reported here, we entered our key independent
2006), a circumstance that, although theoretically variables one at a time in separate models and
interesting, is outside the scope of our study. Note found the same signs we found in our event history
that although we do not model home country re analyses (see Table 3) and very similar significance
sponses, our sampling design accounts for home levels.
country actions. Each and every home country ac Table3 presents the results of the event history
tion initiates a set of host country response oppor analyses. Model 1 provides a baseline Cox model
tunities. Additionally, responses in a home country with only control variables. Model 2 adds our in
terminate all host country response sets by censor dependent variables. All coefficients reported in
ing all nonresponding host country observations on Table 3 represent odds ratios. These are interpreted
the date of response. Therefore, home country ac as the proportional change in hazard rate for a
tions and responses are accounted for in the sam one-unit increase in the independent variable. For
pling design, though not modeled as dependent example, an odds ratio of 2 indicates that a one-unit
variables. Finally, in analyses not reported here, we increase in the independent variable doubles the
TABLE 2
Descriptive Statistics21
Variables Mean s.d. 10 11 12 13 14 15 16
1. Host GDP growth 2.83 3.00

2. Host market concentration 0.66 0.15 -.03
3. Initiator's home 0.03 0.18 -.01 .11
4. Responder sizeb 15.27 0.53 -.01 .03 -.03
5. Responder international 41.16 12.27 -.01 .02 .02 -.50
experience
6. Major initiating move 0.45 0.50 .01 .00 .02 -.08 .03
7. Number of initiating countries 1.64 1.92 .02 -.02 .02 -.08 .03 .09
8. Complexity of initiating move 1.12 0.36 .01 -.01 .00 -.05 .03 .11 -.02
9. Dyad partner distance 5.39 4.36 .01 -.01 .15 -.15 .16 .11 .11 .03
10. Home-host country distance 8.62 0.87 .06 .16 .05 .16 -.13 .10 -.15 .02 .14
11. Home-initiating country distance 5.80 4.10 .03-.02 .06-.26 .28 .14 .17 .04 .40 -.03
12. Host government constraints 5.02 0.28 .01 .04 .06 .08 -.06 .01 .02 .02 .01 .11 .05
13. Home government constraints 5.09 0.30 .04 -.01 .02 -.54 .53 .01 -.01 .03 .13 .05 .40 -.02
14. Initiating country importance 31.28 27.44 .05 .02 -.06 .20 -.17 -.29 -.29 .02 .35 .21 .73 .02 .18
15. Within-country response 0.07 0.25 .00 .01 .30 -.07 .04 .05 .40 -.01 .09 -.09 .17 -.01 .04 .20
16. Subsidiary control 92.10 18.82 .05 -.23 -.05 -.01 .00 .02 .02 .00 .02 -.12 .04 -.02 .13 .01 .00
17. Multimarket contact 0.57 0.16 .06 .03 -.06 .30 -.14 .16 -.20 .00 .39 .30 .33 -.02 .22 .52 .13 .02
a
n = 30,652. r's > |.0l| are significant at p < .05.
b
Logarithm of units sold.

TABLE 3 sponse speed. Results in model 2 support this hy

Results of Cox Event History Analysis of Global The odds ratio of .96 < .05) for home
pothesis. [p
Automakers' Reactions41 initiating country distance indicates that a one-unit
increase in geographic distance decreases the haz
Variables Model 1 Model 2
ard rate by 3.7 percent.
Hypothesis 3 states that the level of host govern
Control
Host GDP 0.98* 0.97* ment constraints on foreign-based MNEs is nega
growth
Host market concentration 1.06 1.11
tively related to response speed in that host coun
Initiator's home 4.20*** 1.45*** The evidence in Table 3 this
try. supports
sizeb 1.53*** 1.56***
Responder
hypothesis as well, with an odds ratio for host
Responder international 0.99** 0.99***
experience
government constraints of .69 [p < .01), indicating
that a one-unit increase in host government con
Major initiating move 0.97 0.98
Number of initiating 1.10*** 0.96*** straints reduces the hazard rate of response in that
countries
country by 30.7 percent, also significantly decreas
Complexity of initiating 0.75** 0.80*
move ing response speed.
distance 0.95*** 0.98* Hypothesis 4, stating a positive relationship be
Dyad partner
tween an MNE's home government constraints and
Independent its response speed in host countries, is very
Home-host distance 0.91*
strongly supported, as shown inmodel 2 of Table 3.
Home-initiating country 0.96*
distance
The odds ratio of 1.60 [p < .001) for home govern
Host constraints 0.69** ment constraints indicates that for each one-unit
government
Home government 1.60** increase in home government constraints, the haz
constraints ard rate of response (in host countries) increases by
control 1.00*
nearly 60 percent, thus also indicating a highly
Subsidiary
Initiating country 1.00
< .001) increase in response
importance
significant [p speed in
response 11.87*** all host countries.
Within-country
Multimarket contact 10.73***
Hypothesis 5 states that as subsidiary control
increases, response speed also increases. The evi
Log-likelihood -7,271.58 -6,766.59
412.57***
dence in model 2 of Table 3, however, does not
Chi-square 1,422.56***
support this prediction. The odds ratio of 1.00
a
The hazard rate of response is the dependent variable. Re (.997; p < .05) suggests that for each unit increase
values are odds ratios?the hazard rate of a one-unit
ported in headquarters control, the hazard rate of response
increase in an variable divided by the hazard rate
actually decreases by 0.3 percent, thereby decreas
independent
of a baseline level of the independent variable. Odds ratios are
interpreted as the
proportional change in hazard rate from a ing response speed. Thus, although the odds ratio
one-unit increase in the independent variable; values of 1 indi for subsidiary control is significant, its sign is op
cate no change. Values lower than 1 indicate that increases in
posite that hypothesized.
the independent variable decrease the hazard rate (decrease
Hypothesis 6 states that as the importance of an
response speed), and those greater than 1 indicate that increases
increases, the hazard rate of re
in the independent variable increase the hazard rate initiating country
(increase
sponse increases in all host countries. The evi
response speed).
b
of units sold. dence from model 2 of Table 3 does not support
Logarithm
* <
p .05 this hypothesis. The odds ratio of 1.00 (.997, n.s.)
**
p < .01 indicates no association, for our sample, between
< .001
***p the strategic importance of an initiating country
and response speed. We conducted two additional
hazard. Below, we use the evidence from model 2 analyses before drawing conclusions about Hy
of Table 3 to test our hypotheses. pothesis 6. First, we ran models with only control
Hypothesis 1 states that home-host geographic variables and the strategic importance of an initiat
distance decreases response speed. The evidence ing country?no other independent variables?and
from model 2 of Table 3 supports this hypothesis. achieved the same insignificant result (nonsignifi
The odds ratio for home-host geographic distance cant). Second, we included home country re
is .91 [p < .05), indicating that a one-unit increase sponses in several additional models, since the
in home-host geographic distance reduces the haz home country is likely to be very strategically im
ard rate of response by 8.63 percent (significantly portant. However, the effects of strategic impor
decreasing response speed). In Hypothesis 2, we tance were still insignificant. Therefore, Hypothe
state a negative relationship between home and sis 6 is not supported.
initiating country geographic distance and re Hypothesis 7, stating that response speed is fast

est in an initiating country, is very strongly sup Kogut & Singh, 1988) and foreign direct investment
ported. The odds ratio of 11.87 [p < .001) inmodel (FDI; Graham, 1990; Knickerbocker, 1973). More
2 indicates that when a host country is the country over, our study introduces the event history meth
of initiation, the hazard rate is over 11 times greater odology to competitive dynamics research. Our
than it is in host countries where the initiating specific sampling design, treating each competitive
action did not occur. This is very strong evidence move as both a response to a previous action by a
that within-country responses are very important. rival and an initiator of that rival's next response,
Finally, Hypothesis 8 states that the degree of and treating each action as an initiator of response
multimarket contact an MNE has with a given rival opportunities over a wide range of markets, paves
is positively related to its response speed. The odds the way formore sophisticated analyses than ours.
ratio formultimarket contact inmodel 2 of Table 3 Because so much competitive dynamics research
< .001) indicates that for each unit in
(10.73, p predicts response likelihood and/or response time,
crease in multimarket contact, the hazard of re we believe that the event history methodology is
sponse increases about ten times.4 broadly applicable (e.g., Morita, Lee, & Mowday,
With respect to the effects of our control vari 1989) and holds great promise for future research
ables, we can see (again, inmodel 2, Table 3) that, ers. Finally, we extended the awareness-motiva
all else being equal, response speeds are slower in tion-capability framework to a global arena. In do
host countries with higher GDP growth (h.r. = 0.97, ing so, we show that this framework is useful for
=
p < .05), faster in an initiator's home market (h.r. examining strategic organizational behavior in a
4.20, p < .001), faster for larger responding firms variety of contexts.
= < and slower for firms with
(h.r. 1.56, p .001),
=
more international experience (h.r. 0.99, p <
are slowed when an Implications
.001). Additionally, responses
action involves more countries = The theoretical of this research are
initiating (h.r. implications
0.96, p < .001), when the initiating move is more manifold. First, we found that both the distance
= < and when the two between a home country and a host country and the
complex (h.r. 0.80, p .05),
rivals are more geographically distant from each distance between a home country and an initiating
other (h.r. = 0.98, p < .05). Aside from the evidence country dampen MNEs' response speed. Despite
about international experience, which was ex extensive research in the international business lit
pected to be positive but turned out to be negative, erature on the distance between countries, rela
all of the evidence from our control variables is tively little effort has been made to examine the
aligned with prior evidence and theory in compet impact of distance on interfirm rivalry (Chen &
itive dynamics research. Stucker, 1997; Ma, 1998). Our finding corroborates
the argument Ghemawat made when he wrote that,
although "much has been made of the death of
DISCUSSION
distance in recent years . . .when it comes to busi
Anchored in competitive dynamics research, our ness, that is not only an incorrect assumption, it is
study examined the response speed ofMNEs using a dangerous one" (2001:138). The operational com
a set of resource-related and market-related factors plexity geographic distance causes places a pre
derived from previous MNE research. Focusing on mium on responsiveness and agility, making it
the interactive and dynamic nature of global com hard for cross-border competitors to react swiftly to
petition, our study addresses how the complexity rivals' attacks in host countries. Ghemawat (2001)
and diversity of host country markets impact ri suggested that, to comprehensively measure the
valry among MNEs with factors that have no clear distance between countries, researchers need to fo
in domestic settings. Through investigat cus on four basic dimensions: the cultural, admin
parallels
ing a wide range of MNE competitive actions, our istrative, geographic, and economic. We were able
approach adds to traditional research in the inter to analyze and develop clear predictions for only
national business literature, inwhich scholars have one of these, and we cannot say specifically how
devoted attention largely to topics such as market the other three might influence response speed. As
entry (Delios & Henisz, 2000; Kim & Hwang, 1992; Ghemawat noted, different forms of distance have
different influences. Thus, more work is needed to
unpack different types of distance and their
4
Of course, the index we used to gauge multimarket impacts.
contact from 0 to 1, so our statement about Beyond cultural distance, in the international
only ranges
the effects of a one-unit change may seem a bit business literature the significance of government
misleading. regulations in global business settings has long

our research setting. Auto manufacture is very cap

been recognized. Yet despite extensive study of the
effects of government regulations on MNEs' entry ital-intensive, and subsidiaries endure relatively
modes, location choices, and internationalization constrained discretion. For example, most country
processes (e.g., Delios & Beamish, 1999; Gomes level subsidiaries cannot design and manufacture
Casseres, 1990; Loree & Guisinger, 1995; Rugman & cars on their own but must work with MNE head
Verbeke, 1998), most current research is silent on quarters to do so. In contrast, subsidiaries in indus
how government constraints affect the day-to-day tries like fast food can be much more creative and
competitive behavior ofMNEs. By showing the sig independent in action. For instance, KFC's Japa
nificant impact of government constraints on the nese subsidiary was very creative and experimen
response speed of MNEs in host countries, our tal, tryingmenu items and store designs that had no
study begins to fill this gap. Further, we point to ready parallels in U.S. operations (Bartlett & Ran
ward the potential for nonmarket strategies in gan, 1992). Future studies of subsidiary control
global competition (Baron, 1995). Boddewyn and should address the level of discretion afforded a
Brewer (1994) contended that Porter's (1980) para subsidiary by the nature of its industry.
digm of three generic strategies is incomplete. Regarding the role of market-related factors, our
When firms compete globally, they may find that finding of initiating country importance is intrigu
nonmarket forces control many opportunities. Ac ing. Although we predicted that the strategic im
cordingly, firms should strive tomake the best use portance of an initiating country would have a pos
of government policy?Boddewyn and Brewer's itive effect on response speed, the coefficient we
"fourth generic strategy"?to assist them in com reported was insignificant. As we noted, that result
peting more effectively. Our study bolsters the idea was not due to multicollinearity, because we ran
that government policy can be a source of strength models including only control variables and strate
we en the same results. This
forMNEs operating in global contexts, and gic importance and achieved
courage future researchers to integrate nonmarket insignificant finding has several possible explana
strategies into their analyses of competition. tions. First, all of our companies competed inmany
With respect to subsidiary control, in the inter countries. Our dyad country counts ranged from 12
national business literature there has long been a to 24. When competitors meet in so many markets,
debate over its impact on subsidiary activity. Some their behavior might diverge somewhat from what
scholars argue that a high level of subsidiary con mainstream theory would predict. Perhaps it is
trol restricts the flexibility and autonomy of sub more important to simply respond to each action,
sidiaries and impedes their responsiveness to local so the centrality of themarket where an action was
market environments (Baliga & Jaeger, 1984; Gates taken is less important. Second, it is possible that
& Egelhoff, 1986). On the other hand, theoretical the magnitude of response was greater following
arguments and empirical evidence show that effi attacks in strategically important markets, though
cient subsidiary control facilitates global integra our event history methodology does not readily
tion and intersubsidiary coordination across differ lend itself to a study of response strength. However,
ent country markets (Ghoshal, 1987; Golden & Ma, stronger (more forceful) responses would be consis
2003; Prahalad & Doz, 1987; Yip, 1995). In our tent with prior research and theory (Chen & Mac
study, emphasizing the important role headquar Millan, 1992; Tsai et al., 1991). Third, as we
ters can play in providing subsidiaries with critical pointed out earlier, the auto industry is very capi
resources and information, we propose a positive tal-intensive, and planning and implementing a re
relationship between subsidiary control and re sponse can take time. This fact could render the
sponse speed. The results did not support our pre strategic importance of an initiating country less
diction. One possible explanation could be that important than itwould be in other settings. Future
there was not enough variance in our subsidiary research is needed to fully explain the implications
ownership data. Most of the countries included in of initiating country strategic importance for com
our study were developed countries where govern petitive response.
ment restrictions on foreign ownership were rela We found strong support for our predictions
tively weak. Further, our sampled automakers about within-country response. As expected, MNEs
seemed to favor majority ownership, as 82 percent tend to respond quickly in the locations where they
of their subsidiaries were majority-owned, versus are attacked. This finding reinforces the notion of
18 percent that were minority owned. With this competitive signaling (Clark & Montgomery, 1998;
low variation in ownership structure, ours should Simmel, 1950). A forceful response in a country of
not be interpreted as a definitive test of the subsid attack sends a strong signal to rivals, helping to
iary control hypothesis. Additionally, some impor keep them from future attacks. Additionally, much
tant factors (and possibly limitations) arise from competitive dynamics research suggests that at

680 Academy ofManagemen t Journal June
tackers gain the upper hand when no response is strong home government protection reduces do
forthcoming (Chen & MacMillan, 1992; Ferrier et mestic firms' motivation to expand and compete
al., 1999). This idea suggests a strong motive for abroad (Marcus, 1990; Rugman & Verbeke, 1990).
MNEs to respond in a country of attack, even Although we do not think this is a serious concern
though they have the capability to respond in other in the highly globalized auto industry, itmay be of
concern in other industry settings and
markets.
thereby war
Finally, our findings about multimarket contact rants consideration in future research. Fourth, our
add to the theory of multimarket competition. In study considered rival dyads (pairs of competitors)
the strategic management and industrial organiza to be the appropriate level of analysis. Other re
tion economics literatures, a number of studies searchers have argued strongly that the dyad is the
have examined the relationship between multi appropriate level of analysis (e.g., Baum & Korn,
market contact and the intensity of rivalry (Evans & 1996; Chen, 1996), but global markets naturally
Kessides, 1994; Gimeno, 1999). Relatively little ef consist ofmultiple rivals, and our choice tomodel
fort, however, has been devoted to studying the dyads is therefore a limiting one. For example,
actual moves and countermoves ofmultimarket ri there may be constellations of competitors that at
vals. Chen (1996) proposed that although a firm tack and respond at the alliance level, rather than
with greater multimarket contact is less likely to the firm level (e.g., Das & Teng, 2002). Or a single
initiate an attack, it ismore likely tomove quickly firmmay aim some attacks or focus some responses
to counter a multimarket rival's move (i.e., to re toward several competitors simultaneously. In
spond when attacked). Our study is one of very few deed, we strongly encourage future investigators of
that have delved into this issue (for an exception, global rivalry to explicitly consider competitive
see Young et al. [2000]). constellations as well as other ways that single
In addition to theoretical implications, important attacks could be simultaneously directed toward
implications for multinational managers also multiple rivals. Finally, our study only examined
emerge from our study. It reinforces the notion of the main effects of some resource-related and mar
competitor analysis whereby a manager must ket-related variables. However, we do suspect some
know, understand, and predict the competitive ac interaction effects may exist between these vari
tions of rivals (Porter, 1980). In today's business ables. All of these limitations leave potentially
context, more and more firms do business interna fruitful avenues for future research.
tionally. Given the increasing price that these firms To conclude, our study, building on competitive
have to "pay to play" in diverse markets, the theo dynamics research, examined factors predicting the
retical framework and empirical findings of our speed of MNEs' competitive responses in host
study can multinational managers develop a countries. In presenting our framework as a first
help
sound understanding of their own and their rivals' step in this new direction, we hope to improve
resources and capabilities and increase their readi understanding of firm competitive behaviors in the
ness to engage in competitive activities. global context.
Limitations REFERENCES
its contributions, our study has several
Despite Allison, G. T. 1971. Essence of decision: Explaining the
limitations. First, our findings are based on firms in Cuban missile crisis. Boston: Little, Brown.
a single industry over a specific seven-year period.
Anand, B. N? & Khanna, T. 2000. Do firms learn to create
For this reason, the results may reflect some factors value? The case of alliances. Strategic Management
specific to the industry or period under study. Fu Journal, 21: 295-315.
ture replication of our research in other settings
Atkinson, J.W. 1964. An introduction to motivation.
will help to address this concern. Second, our
Princeton, NJ: Van Nostrand.
study only involved competitive actions that could
be identified via structured content analysis. We Axelrod, R. 1984. The evolution of cooperation. New
York: Basic Books.
deemed Automotive News a reliable source for
B. R., &
studying competitive actions of global automakers. Baliga, Jaeger, A. M. 1984. Multinational corpo
rations: Control and issues.
Nevertheless, and despite the extensive checks we systems delegation Jour
or un nal of International Rusiness Studies, 15(2): 25-40.

made, Automotive News may have missed
some competitive actions. Third, Barney, J. B. 1986. Organizational culture: Can it be a
derrepresented
source
because data availability and the scope of our study of sustained competitive advantage? Acad
11: 656-665.
were both limited, we were not able to fully address emy of Management Review,
several issues. For instance, some have argued that Barney, J. B. 1991. Firm resources and sustained compet

itive advantage. Journal ofManagement, 17(1): 99 Cross-border competitive aggressiveness among

120. MNCs: Market and resource considerations (Work
Baron, D. P. 1995. Market and non ing paper). Charlottesville: University of Virginia.
Integrated strategy:
market components. California Management Re Clark, B. H., & Montgomery, D. B. 1998. Deterrence, rep
view, 37(2): 47-65. utations and competitive cognition. Management
Science, 44(1): 62-82.
Bartlett, C. A., & Rangan, S. 1992. Kentucky Fried
Chicken (Japan) Limited (Case no. 9-387-043). Bos Cox, D. R. 1972. Regression models and life-tables. Jour
ton: Harvard Business School. nal of the Royal Statistical Society, 34: 187-200.
W. & Suri, R. 1977.
Baum, J.A. C, & Korn, H. J. 1996. Competitive dynamics Curhan, J. P., Davidson, H., Tracing
of interfirmrivalry.Academy themultinationals: A sourcebook on US-based en
ofManagement Jour
nal, 39: 255-291. terprises. Cambridge, MA: Ballinger.
Baum, J.A. C, & Korn, H. J. 1999. Dynamics of dyadic Das, T. K, & Teng, B.-S. 2002. Alliance constellations: A
social
competitive interaction. Strategic Management exchange perspective. Academy of Manage
20: 251-278. ment Review, 27: 445-456.
Journal,
T. Davidson, W. H. 1980. The location of foreign direct

Boddewyn, J. J., & Brewer, L. 1994. International busi
ness political behavior: New theoretical directions. investment characteristics and ex
activity: Country
Academy ofManagement Review, 19: 119-143. perience effects. Journal of International Rusiness
Studies, 11(2): 9-22.
Box-Steffensmeier, J.,& Jones, B. S. 2004. Event history
W. & McFetridge, D. G. 1985. charac
modeling: A guide for social scientists. Cambridge,
Davidson, H., Key
U.K.: Press. teristics in the choice of international technology
Cambridge University
transfermode. Journal of International Rusiness
Brander, J., & P. 1983. A
Krugman, "reciprocal dumping" Studies, 16(2): 5-21.
model of international trade. Journal of Interna
Delios, A., & Beamish, P. W. 1999. strategy of
tional Economics, 15: 313-321. Ownership
Japanese firms: Transactional, institutional, and ex
Brown, P., & Henry, J. 2001. Ford merges British PAG influences.
perience Strategic Management Jour
Automotive News, 13: 3.
operations. August nal, 20: 915-933.
Bulow, J. I., Geanakoplos, J. D., & P. D. 1985.

Klemperer, Delios, A., & Beamish, P. W. 2001. Survival and profit
Multimarket oligopoly: Strategic substitutes and ability: The roles of experience and intangible assets
complements. Journal of Political Economy, 93: in foreign subsidiary performance. Academy of
488-511.
Management Journal, 44: 1028-1038.
Camerer, C. F. 1991. Does strategy research need game Delios, A., & Henisz, W. J. 2000. firms' invest
Japanese
theory? Strategic Management Journal, 12 (winter ment strategies in emerging economies. Academy of
special issue): 137-152. Management Journal, 43: 305-323.
Casson, M. 1987. The firm and themarket: Studies on Dierickx, I., & Cool, K. 1989. Asset stock accumulation
multinational enterprise and the scope of thefirm. and of competitive Man
sustainability advantage.
Cambridge, MA: MIT Press. agement Science, 35: 1504-1511.
Chen, M.-J. 1996. Competitor analysis and interfirm ri Y. & Prahalad, C. K. 1981. influ
Doz, L., Headquarters
Toward a theoretical ence and control in MNCs. Sloan
valry: integration. Academy of strategic Manage
21: 100-134. ment Review,
Management Review, 23(fall): 15-29.
Chen, & MacMillan, I. C. 1992. and
M.-J., Nonresponse Dutton, J. E., & Jackson, S. E. 1987. Categorizing strategic
delayed response to competitive moves: The roles of issues: Links to organizational action. Academy of
competitor dependence and action irreversibility. Management Review, 12: 76-90.
Academy ofManagement Journal, 35: 539-570. C. D. 1955. as a source of

Edwards, Conglomerate bigness
Chen, M.-J., Smith, K. G, & Grimm, C. M. 1992. Action Rusiness concentration and
power, price policy:
characteristics as
predictors of competitive
re 331-352. Princeton, NJ: Princeton University Press.
sponses. Science, 38: 439-455.
Management
Egelhoff,W. G. 1988. Strategy and structure inmultina
Chen, M.-J., & Stucker, K. 1997. Multinational manage tional corporations: A revision of the Stopford and
ment and multimarket rivalry: Toward a theoreti Wells model. Strategic Management Journal, 9:
cal development of global competition. Paper pre 1-14.
sented at the annual meeting of the Academy of
Boston.
Ellingsen, T., &W?rneryd, K. 1999. Foreign direct invest
Management, ment and the political economy of protection. Inter
Chen, M.-J., & Su, K.-H. In press.
Competitive tension: national Economic Review, 40(2): 357-379.
The awareness-motivation-capability perspective. Erramilli, M. K. 1991. The experience factor in foreign
Academy ofManagement Journal. market entry behavior of service firms. Journal of
Chen, M.-J., Yu, T., Michel, J.G, & Cannella, A. A. 2006. International Rusiness Studies, 22: 479-501.

682 Journal June
Academy ofManagement
Evans, W. N., & Kessides, I. N. 1994. Living by the ative game. International Trade Journal, 4: 259
"golden rule": Multimarket contact in theUS airline 277.
industry. Quarterly Journal of Economics, 109:

Harrigan, K. R. 1985. Strategies for intrafirmtransfersand
341-366.
outside sourcing. Academy ofManagement Jour
Farhi, S. 1998. Renault to lower price of face-lifted nal, 28: 914-925.
Twingo. Automotive News 17: 7.

Europe, August Henisz, W. J., & Delios, A. 2001. Uncertainty, imitation,
Ferrier, W. J. 2001. Navigating the competitive land and plant location: Japanese multinational corpora
scape: The drivers and consequences of competitive tions, 1990-1996. Administrative Science Quar
aggressiveness. Academy of Management Journal, 46: 443-475.
terly,
44: 858-877.
Hofstede, G. 1980. Culture's consequences, interna
Ferrier, W. J., & Lee, H. 2002. Strategic aggressiveness, tional differences in work-related values. Beverly
variation, and surprise: How the sequential pattern Hills, CA.: Sage.
of competitive rivalry influences stock market re & Daft,
Huber, G P., R. L. 1987. The information envi
turns. Journal ofManagerial Issues, 14: 162-180. ronment of organizations. In F. M. Jablin, L. L. Put
Ferrier, W. J., Smith, K. G, & Grimm, C. M. 1999. The role nam, K. H. Roberts, & L. W. Porter (Eds.), Handbook
of competitive action in market share erosion and communication: 130-164. New
of organizational
industry dethronement: A study of industry leaders bury Park, CA: Sage.
and challengers. Academy ofManagement Journal, Jauch, L. R, Osborn, R. N., & Martin, T. N. 1980. Struc
42: 372-388.
tured content analysis of cases: A complementary
Gates, S. R, & Egelhoff, W. G 1986. Centralization in method for organizational research. Academy of
5: 517-526.
headquarters-subsidiary relationships. Journal of Management Review,
International Rusiness Studies, 17(2): 71-92.
Katila, R., & Shane, S. 2006. When does lack of resources
Gatignon, H., & Anderson, E. 1988. The multinational make new firms innovative? Academy ofManage
corporation's degree of control over foreign subsid ment Journal, 48: 814-829.
iaries: An test of a transaction cost
Kennedy, P. 2003. A guide to econometrics
empirical expla (5th ed.).
nation. and
of Law, Economics,
Cambridge, MA: MIT Press.
Journal Organiza
tion, 4: 305-336.
Kiesler, S. B., & Sproull, L. 1982. Managerial response to
Ghemawat, P. 2001. Distance still matters?The hard re
environments: on
changing Perspectives problem
ality of global expansion. Harvard Rusiness Re
sensing from social cognition. Administrative Sci
view, 79(8): 137. ence 27: 548-570.
Quarterly,
Ghoshal, S. 1987. Global strategy:An organizing frame Kim, W. C, & Hwang, P. 1992. Global strategy and mul
work. Journal, 8: 425-440.
Strategic Management tinationals' choice.
entry mode Journal of Interna
Ghoshal, S., & Bartlett, C. A. 1990. The multinational tional Rusiness Studies, 23(1): 29-53.
as an network.
corporation interorganizational Knickerbocker, F. T. 1973. Oligopolistic reaction and
Review, 15: 603-625.
Academy of Management multinational enterprise (Unpublished doctoral
Ghoshal, S., & Kim, S. K. 1986. Building effective intel dissertation). Cambridge, MA: Harvard University.
for Sloan
Kogut, B., & Singh, H. 1988. The effect of national culture
ligence systems competitive advantage.
Management Review, 28(1): 49-58.
on the choice of entry mode. Journal of Interna
Gimeno, J. 1999. Reciprocal threats in multimarket ri tional Rusiness Studies, 19: 411-432.
valry: Staking out of influence' in the U.S.

'spheres Kostova, T., & Zaheer, S. 1999. Organizational legitimacy
airline industry. Strategic Management Journal, under conditions of complexity :The case of themul
20: 101-128.
tinational enterprise. Academy of Management Re
Gimeno, J. 2002. The performance effects of unintended view, 24: 64-81.

and purposive multimarket contact. Managerial D. M., & Wilson, R. 1982. and
Kreps, Reputation imper
and Decision Economics, 23: 209-224.
fect information. Journal of Economic Theory, 27:
Golden, B. R, & Ma, H. 2003. Mutual forbearance: The 253-279.
role of intra-firm integration and rewards. Academy P. R. 1984. as promo
Krugman, Import protection export
ofManagement Review, 28: 480-495.
tion: International competition in the presence of
Gomes-Casseres, B. 1990. Firm ownership preferences oligopoly and economies of scale. InH. Kierzkowski
and host government restrictions: An integrated ap (Ed.), Monopolistic competition and international
proach. Journal of International Rusiness Studies, trade: 180-193. Oxford, U.K.: Clarendon Press.
21(1): 1-22.
Kurylko, D. T. 1998. GM reshapes itsworld in the image
Graham, E. M. 1990. Exchange of threats between multi of Ford: Wagoner to head all operations; Hughes' old
nationals firms as an
infinitely repeated noncooper job eliminated. Automotive News, October 12: 3.

Kurylko, D. T., & Catterall, M. 2000. Humbled Japanese Porter, M. E. 1987. From competitive advantage
to cor
get serious. Automotive News Europe, January 3:3. porate strategy.Harvard Rusiness Review, 65(3):
43-59.
Lant, T. K, Milliken, F. J., & Batra, B. 1992. The role of
managerial learning and interpretation in strategic Prahalad, C. K, & Doz, Y. L. 1987. The multinational
persistence and reorientation: An empirical explora mission: Ralancing local demands and global vi
tion. Strategic Journal, 13: 585-608. sion. New York: Free Press.
Management
Li, J. 1995. Foreign entry and survival: Effects of strategic Rechtin, M. 2001. Camry: Back to nonbasics. Automotive
choices on in international markets. News, 20: 3.
performance August
Strategic Management Journal, 16: 333-351. Robinson, A. 1999. Global GM effortcreates new engine.
Loree, D. W., & Guisinger, S. E. 1995. Policy and non Automotive News Europe, August 2: 5.
policy determinants of US equity foreign direct in
vestment. Journal of International Rusiness Stud Rugman, A. M., & Verbeke, A. 1990. Global corporate
26: 281-299.
strategy and trade policy. London: Routledge.
ies,
Rugman, A. M., & Verbeke, A. 1998. strategy
Corporate
Ma, H. 1998. Mutual forbearance in international busi
and international environmental policy. Journal of
ness. Journal of International Management, 4(2): International Rusiness Studies, 29: 819-833.
129-147.
A. A. 1990. Business demand for regulation: An

Schelling, T. C. 1960. A strategy of conflict. Boston:
Marcus,
Harvard University Press.
exploration of the Stigler hypothesis. In L. E. Preston
1947.
(Ed.), Government regulation and business re Schumpeter, J. A. Capitalism, socialism, and de
sponse: Research issues and empirical studies: mocracy (2d ed.). New York, London: Harper & Row.
33-54. Greenwich, CT: JAI Press. Simmel, G 1950. Superordination and subordination
Martin, X., Swaminathan, A., & Mitchell, W. 1998. and degrees of domination and freedom. In K. Wolff
Orga
nizational evolution in the interorganizational envi (Ed.), The sociology of Georg Simmel: 268-303.
ronment: Incentives and constraints on international New York: Free Press.
Administrative Science Quar

expansion strategy. Smith, K. A., Ferrier, W. J., & Ndofor, H. 2001. Compet
terly, 43: 566-601. itive dynamics research: Critique and future direc
tions. In M. A. Hitt, E. R. & J. R. Harrison
Martinez, J. I., & Jarillo, J. C. 1989. The evolution of Freeman,
research on coordination mechanisms in multina (Eds.), The Rlackwell handbook of strategic man
tional corporations. Journal of International Rusi agement: 315-361. Oxford, U.K.: Blackwell.
ness Studies, 20: 489-515. & Grimm,

Smith, K. G, C. M. 1991. A communication
B. 2002. Hofstede's model of national cul information model of competitive response timing.
McSweeney,
tural differences and their consequences: A triumph Journal of Management, 17: 5-23.
of faith?A failure of analysis. Human Relations, 55: K. G, C. M., & Gannon, M.

Smith, Grimm, J. 1992. Dy
89-118.
namics of competitive strategy. Newbury Park, CA:
Milgrom, P. R., & Roberts, J. 1982. Pr?dation, reputation, Sage.
and entry deterrence. Journal of Economic Theory, K. G, C. M., M.
Smith, Grimm, Gannon, J., & Chen, M.-J.
27: 280-312. com
1991. Organizational information processing,
Miller, D., & Friesen, P. H. 1977. in petitive responses, and performance in the U.S. do
Strategy making
context: Ten mestic airline
empirical archetypes. Journal of Man industry. Academy of Management
Studies, 14: 251-260. Journal, 34: 60-85.
agement
Miller, D., & Friesen, P. H. 1984. Organizations: A quan Spell, C. S., & Blum, T. C. 2005. Adoption ofworkplace
tum view. New York: Prentice-Hall. substance abuse prevention programs: Strategic
choice and institutional perspectives. Academy of
Morita, J.G, Lee, T. W., & R. T. 1989. Intro
Mowday,
Management Journal, 48: 1125-1142.
survival to organizational research
ducing analysis
ers: A selected application
to turnover research. Thomas, L., & Weigelt, K. 2000. Product location choice
Journal of Applied Psychology, 74: 280-292. and firm capabilities: Evidence from theUS automo
North, D. 1990. Institutions, institutional change, and bile industry. Strategic Management Journal, 21:
897-909.
economic performance. Cambridge, U.K.: Cam
bridge University Press. Tsai, W. M.-H., MacMillan, I. C, & Low, M. B. 1991.
1993.
Effects of strategy and environment on corporate
Peteraf, M. A. The cornerstones of competitive
venture success in industrial markets.
A resource-based view. Man Journal of
advantage: Strategic
Rusiness 6: 9-28.
14: 179-191. Venturing,
agement Journal,
Tuma, N. B., & Hannan, M. T. 1984. Social
Porter, M. E. 1980. Competitive strategy: Techniques dynamics:
Models and methods. New York: Academic Press.
for analyzing industry and competitors. New York:
Harper & Row. Venables, A. J. 1984. Multiple equilibria in the theory of

international trade with monopolistically competi rate strategy: A review of recent theory and applica
tive commodities. Journal of International Econom tions. Strategic Management Journal, 9: 443-454.
ics, 16(1/2): 103-121.

Yip, G. S. 1995. Total global strategy: Managing for
Veugelers, R. 1995. Strategic incentives formultinational worldwide competitive advantage: Englewood
and Decision Economics,
operations. Managerial Cliffs, NJ: Prentice-Hall.
16(1): 47-57.
1964. Work K. G, C. M., & Simon, D. 2000.
Vroom, V. H. and motivation. New York: Young, G, Smith, Grimm,
Multimarket contact and resource dissimilarity: A
Wiley.
competitive dynamics perspective. Journal ofMan
K, & Camerer, C. F. 1988. and corpo
Weigelt, Reputation 26: 1217-1236.
agement,
APPENDIX A TABLE A3
Frequency Counts of Actions Counts of Actions
Frequency by Country
TABLE Al
Country Actions Percent
Frequency Counts of Actions by Year
Argentina 14 0.63
Year Actions Percent Austria 93 4.21
Belgium 50 2.27
1995 343 15.54
Brazil 73 3.31
1996 671 30.40
Canada 6 0.27
1997 242 10.97
China 35 1.59
1998 267 12.10
France 214 9.70
1999 329 14.91
214 9.70
Germany
2000 191 8.65
India 41 1.86
2001 164 7.43
Indonesia 18 0.82
Ireland 45 2.04
Italy 162 7.34
TABLE A2 Japan 73 3.31
Counts of Actions Malaysia 18 0.82

Frequency by Company
Mexico 13 0.59
Netherlands 49 2.22
Actions Percent
Company
Philipines 15 0.68
Poland 38 1.72
89 4.03
DaimlerChrysler
Portugal 47 2.13
Fiat 168 7.61
30 1.36
Russia
Ford 358 16.22
Korea 24 1.09
South
General Motors 480 21.75
113 5.12
Spain
Honda 145 6.57
Sweden 43 1.95
44 1.99
Hyundai Switzerland 59 2.67
Mitsubishi 87 3.94
Thailand 54 2.45
Nissan 96 4.35
United Kingdom 186 8.43
Peugeot 133 6.03
404 18.31
United States
Renault 142 6.43
Subaru 29 1.31
Toyota 153 6.93
Volkswagen 283 12.82

APPENDIX B those for Germany and Canada are coded as responses (a

Data Structure Example response was observed) and those for South Korea,
France, and theUnited States are coded as censored (no
For this example, we assume two companies and
response was observed by the end of the observation
(Ford
Toyota) and six countries (France, Germany, South Ko period).

rea, Canada, theUnited States, and Japan). Table Bl lists
Toyota's response now becomes the initiating action
the initiators, and countries involved in each
responders, for the second set of response opportunities. Here, Ford
of four actions taken by these rivals against each other is the target, and can respond in any of its five host
over a 100-day interval.0 Table B2 lists the event history
countries. At day 70 (35 days later) Ford responds against
observations by the actions and responses of
generated in Canada, France, and the ob
Toyota Japan. Therefore,
the dyad over the 100-day window. servations for Canada, and are coded as
France, Japan
and those for Germany and South Korea are
TABLE Bl responses,
coded as censored.
Action Sequence
Ford's response becomes the initiating attack for the
Initiator Date Countries Involved
third set of response opportunities (with Toyota as the
Target
Observations are created for each of Toyota's
responder).
Ford
Toyota 5 France, Japan
fivehost countries. At day 85 (15 days later), Ford attacks
Toyota Ford 35 Germany, Canada Toyota again. Therefore, each of Toyota's five observa
Ford
Toyota 70 Canada, France tions are censored at 15 days?no was observed
response
by the target firm in any country, and the initiator at
Ford Toyota 85 South Korea
tacked again.
Each initiating action generates five opportunities fora Finally, Ford's attack on day 85 sets up the final set of
five response opportunities for Toyota. Each of these is
response?one for each host country inwhich the dyad
censored at 15 days, because the observation ended
competes. Note that Japan is the home country forToyota period
at that point, and no response was observed from Toyota.
and the United States is the home country for Ford.
Although we do not model responses in home countries, TABLE B2
actions taken in the home country are included as initi Event History Observations
ating actions.
The first action, taken by Ford against Toyota, starts Response
Start End or Host
the sequence on the 5th day of the observation period. Initiating
Initiator Responder Day Day Censoring Country Countries
Ford is the initiator, and Toyota the target.The initiating
action was taken in France and Japan. This sets up the Ford Toyota 5 35 C France France,
Japan
first five observations, with Toyota as the responder, Ford Toyota 5 35 R Germany France,
Japan
Ford 5 35 C S. Korea France,
Ford as the initiator, and Toyota able to respond in every Toyota
5 35 R
Japan
Ford Toyota Canada France,
Japan
country except its home country (Japan). Figure Bl (next Ford Toyota 5 35 C U.S. France,
Japan
page) illustrates this sequence. On day 35 (30 days after Toyota Ford 35 70 R France Germany, Canada
Ford 35 70 R Germany, Canada
the attack) Toyota responded against Ford in Germany Toyota Japan
Toyota Ford 35 70 C Germany Germany, Canada
and Canada. Therefore, in the first set of observations, Ford 35 70 C S. Korea Germany, Canada
Toyota
Toyota Ford 35 70 R Canada Germany, Canada
a
We would have used a real example, but all were far Ford 70 85 C France France
Toyota Canada,
more than this one because so many countries Ford 70 85 C Canada, France
complex Toyota Germany
were involved of 12), so many actions Ford 70 85 C S. Korea Canada, France
(a minimum and Toyota
Ford 70 85 C Canada Canada, France
were taken by dyad partners against each other. Toyota
Ford Toyota 70 85 C U.S. Canada, France
Note that we use
"target" instead of "responder" in Ford 85 100 C France S. Korea
Toyota
we are actions. Ford 85 100 C S. Korea
Table Bl, because describing The target is Toyota Germany
not every action Ford Toyota 85 100 C S. Korea S. Korea
clear for each action; however, because
Ford Toyota 85 100 C Canada S. Korea
led to a response, we concluded that would
"responder" Ford Toyota 85 100 C U.S. S. Korea
have been less clear in this setting.
Tieying Yu (yuti@bc.edu) is an assistant professor at Bos Albert A. Cannella Jr. [albert.cannella@asu.edu) is the
ton College. She received her Ph.D. from Texas A & M Hahnco Companies Professor of Strategic Management and
University. Prior toher academic career, she worked in the the chair of theManagement Department at Arizona State
banking industry in Shanghai. Her research interests focus University. He received his Ph.D. fromColumbia University.
on global strategyand competition, competitive dynamics, His research interests focus on executives, entrepreneurship,
and competitor analysis. and competitive dynamics.

FIGURE Bl
Exemplary Action-Response Sequence
Responding
Country
Ford attacks Toyota in France and Japan on
5; Toyota in Germany and
Canada day responds
on
(R-host; responding)
Canada day 35.
Germany
(R-host; responding
South Korea
(R-host)
initiating) I^^^L ; Hl^^"\
France
(R-host; initiating)
United States
(R-host; initiator
home)
Responding
Country
Toyota attacks Ford on day 35 in Canada
and Germany; Ford in France,
responds
and Canada on 70.
Japan day
France
(R-host; responding)
United States
a
"R" indicates "responder," the company whose response is awaited in an interaction. The two columns of circles represent all the
countries that both the initiator and the responder compete in. Black circles (always in the left-hand column of circles) indicate the
countries in which the initiator attacked the target (responder). The right-hand column
of circles has three colors of circles: white means
that no response occurred in that country; light gray means that a response did take place in that country; dark gray means that the country
was the responder's home country, and so that responder-country combination was deleted from the analysis (as discussed earlier, no
home country responses were modeled).


Academy of Management

Uploaded by

Document Information

Original Title

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

Academy of Management

Uploaded by

Copyright:

Available Formats

Rivalry between Multinational Enterprises: An Event History Approach

Author(s): Tieying Yu and Albert A. Cannella Jr.

This content downloaded from 185.44.78.156 on Tue, 17 Jun 2014 04:00:11 AM

RIVALRY BETWEEN MULTINATIONAL ENTERPRISES:

ALBERT A. CANNELLA Jr.

This content downloaded from 185.44.78.156 on Tue, 17 Jun 2014 04:00:11 AM

This content downloaded from 185.44.78.156 on Tue, 17 Jun 2014 04:00:11 AM

This content downloaded from 185.44.78.156 on Tue, 17 Jun 2014 04:00:11 AM

source availability and/or resource sponse to a new four-cylinder engine developed

1. The speed of an MNE's

This content downloaded from 185.44.78.156 on Tue, 17 Jun 2014 04:00:11 AM

3. The speed of an MNE's protections have different impacts on foreign direct

This content downloaded from 185.44.78.156 on Tue, 17 Jun 2014 04:00:11 AM

This content downloaded from 185.44.78.156 on Tue, 17 Jun 2014 04:00:11 AM

This content downloaded from 185.44.78.156 on Tue, 17 Jun 2014 04:00:11 AM

This step yielded 6,648 news articles. We then

This content downloaded from 185.44.78.156 on Tue, 17 Jun 2014 04:00:11 AM

Action Type Description Example

methods, and brand

vehicle, the Verso microvan.

This content downloaded from 185.44.78.156 on Tue, 17 Jun 2014 04:00:11 AM

This content downloaded from 185.44.78.156 on Tue, 17 Jun 2014 04:00:11 AM

likely to be perceived similarly by customers as likelihood of response (Box-Steffensmeier & Jones,

This content downloaded from 185.44.78.156 on Tue, 17 Jun 2014 04:00:11 AM

Variables Mean s.d. 10 11 12 13 14 15 16

1. Host GDP growth 2.83 3.00

This content downloaded from 185.44.78.156 on Tue, 17 Jun 2014 04:00:11 AM

TABLE 3 sponse speed. Results in model 2 support this hy

This content downloaded from 185.44.78.156 on Tue, 17 Jun 2014 04:00:11 AM

This content downloaded from 185.44.78.156 on Tue, 17 Jun 2014 04:00:11 AM

our research setting. Auto manufacture is very cap

This content downloaded from 185.44.78.156 on Tue, 17 Jun 2014 04:00:11 AM

or un nal of International Rusiness Studies, 15(2): 25-40.

This content downloaded from 185.44.78.156 on Tue, 17 Jun 2014 04:00:11 AM

itive advantage. Journal ofManagement, 17(1): 99 Cross-border competitive aggressiveness among

T. Davidson, W. H. 1980. The location of foreign direct

Bulow, J. I., Geanakoplos, J. D., & P. D. 1985.