Journal of International Business Studies (2010) 41, 500–516

& 2010 Academy of International Business All rights reserved 0047-2506

www.jibs.net

Subsidiary expansion/contraction during times

of economic crisis

Chris Changwha Chung1, Abstract

Seung-Hyun Lee2, This paper compares two real options – within-country growth and across-
country operational flexibility – to examine subsidiary expansion/contraction
Paul W Beamish3 and during times of economic crisis. Specifically, we examine how the real options
Takehiko Isobe4 orientations of individual subsidiaries interact with the general characteristics of
multinational enterprise networks. Our main findings are that: (a) economic
1
Korea University Business School, Seoul, crises can be detrimental for subsidiaries with stronger within-country orienta-
Korea; 2School of Management, University tions, and advantageous for those with stronger across-country orientations;
of Texas at Dallas, Richardson, USA; 3Richard and (b) network characteristics are not the sole determinants of subsidiary
Ivey School of Business, University of Western expansion/contraction – what really matters is how the real options orienta-
Ontario, London, Canada; 4Graduate School
tions of individual subsidiaries mesh with the overall characteristics of the
of Business Administration, Keio University,
Yokohama, Japan network they belong to.
Journal of International Business Studies (2010) 41, 500–516.
Correspondence: doi:10.1057/jibs.2009.72
CC Chung, Korea University Business
School, Korea University, Seoul, Korea. Keywords: subsidiary expansion; economic crisis; real options; across-country
E-mail: [email protected] operational flexibility; within-country growth; MNE network characteristics

INTRODUCTION
To maintain competitive advantage, multinational enterprises
(MNEs) must make rapid adjustments to their international invest-
ments in response to fluctuating global market demands and
competition. A major challenge for MNEs is reconfiguring their
value chain activities in a timely fashion to address volatile contin-
gencies in the countries where they operate subsidiaries. The real
options literature emphasizes flexibility for MNEs trying to cope
with heightened uncertainty. Most studies in this area have focused
on the characteristics of multinational networks and their impacts
on the market valuation of firms (Allen & Pantzalis, 1996;
Pantzalis, 2001; Tang & Tikoo, 1999); others have focused on
downside risks for MNEs (Reuer & Leiblein, 2000; Tong & Reuer,
2007). However, when addressing overall MNE network character-
istics, researchers have overlooked the real options orientations of
individual subsidiaries that make up MNE networks. Without
knowing how those orientations mesh with the overall character-
istics of an MNE network, our understanding of real options in a
multinational context will be limited.
MNE subsidiaries can be oriented toward within-country growth
or across-country operational flexibility options, with the first based
Received: 13 November 2007
Revised: 23 May 2009
on a host country’s economic growth potential and the second
Accepted: 18 June 2009 emphasizing operational flexibility among affiliated subsidiaries
Online publication date: 7 January 2010 within a multinational network (Kogut & Kulatilaka, 1994a).
 Subsidiary expansion/contraction at times of economic crisis Chris C Chung et al
501

A subsidiary may emphasize one option over the subsidiary and the overall characteristics of the
other if it perceives local market adaptation and network to which it belongs.
growth to be more important than serving as a Given that within-country and across-country
global export platform, or vice versa. This relative options coexist in a foreign subsidiary, it is essential
perception becomes embedded in subsidiary opera- to establish a boundary condition under which
tions and determines how subsidiaries develop one option has greater value. For this task we will
their real options orientations. Our main argument use the Asian economic crisis that started in 1997
in this paper is that these orientations influence the and persisted for several years. After almost three
realization of MNE flexibility. Examining the decades of strong growth and numerous claims of
corporate characteristics of an MNE is required an impending ‘‘Asian economic miracle,’’ the crisis
because they exercise real options in their subsidi- was a major shock to firms operating in the region
ary portfolios. However, it is important not to (Singh & Yip, 2000). Our analysis uses data for 1519
overlook the essential component that supports subsidiaries of 471 Japanese MNEs in 52 manufac-
MNE flexibility: the individual subsidiaries that turing industries, located in five countries affected
constitute its network. by the crisis between 1997 and 2001.
Since most real options studies focus on MNE
flexibility at the corporate/network level, there is REAL OPTIONS, EXPANSION/CONTRACTION,
a tendency to look at overall levels of aggregated AND CRISIS
subsidiary operations without paying sufficient att- Two types of real options are represented in the
ention to the real options orientations of individual literature: incremental and operational flexibility
subsidiaries. A general assumption is that MNE (Bowman & Hurry, 1993; Sharp, 1991). Incremental
subsidiaries are more or less the same because they options consist of put and call options.2 When
belong to the same parent firm and are therefore a firm sees little potential in an investment, it may
treated as though their individual real options exercise a put option and reduce its exposure,
orientations have little effect on MNE flexibility. and when an opportunity emerges it may exercise
The truth is that each subsidiary has its own real a call option and expand its commitment. Exam-
options orientation based on the path-dependent ples of incremental options include divesting/
development of its strategy. Therefore research acquiring stakes in joint ventures (Chi, 2000; Kogut
needs to consider individual subsidiary orientations 1991) decreasing/increasing investments in exist-
in order to build a more complete understanding of ing facilities or technologies (Coucke, Pennings, &
MNE flexibility.1 Sleuwaegen, 2007; Hurry, 1993). According to the
In a multinational context, formulating real operational flexibility option, firms maintain
options logic based solely on a subsidiary’s real open options for change in preparation for future
options orientation is also inappropriate. For a uncertainties; since uncertainty limits planning
more complete understanding of real options in an effectiveness, operational flexibility is considered
MNE context, we believe it is necessary to investi- complementary to planning (Volberda, 1997). This
gate both the real options orientations of indivi- option can be analyzed as a bundle of interdepen-
dual subsidiaries and how those orientations mesh dent options: two examples are retaining multiple
with overall network characteristics. We examine suppliers in order to cope with future supply fluctu-
interactions between the real options orientations ations (Richardson, 1993), and investing resources
of individual subsidiaries and three network-based in multiple locations so that certain outlets can
characteristics: (a) the relative performance of a focal come to the rescue when one encounters problems
subsidiary vis-à-vis its affiliated subsidiaries in the (McGrath, 1999).
same parent network; (b) the size of the multi- Kogut and Kulatilaka (1994a) applied the strategic
national network to which a focal subsidiary management concepts of incremental vs opera-
belongs; and (c) multinational network redundancy tional flexibility options to the MNE context and
– that is, the overlap between a focal subsidiary identified two real options: ‘‘a within-country option
location and its affiliated subsidiary locations in which, by establishing a grand label or simple
terms of macroeconomic conditions. We argue that knowledge of the market, provides a platform for
the scope of a subsidiary’s operations is not deter- the introduction of new products y [and] an
mined solely by the general characteristics of its across-country option provided by operational flex-
parent MNE; what really matters is the interaction ibility’’ (124–125). The primary difference is that
between the real options orientation of each the first is based on belief in a host country’s

Journal of International Business Studies

 Subsidiary expansion/contraction at times of economic crisis Chris C Chung et al
502

economic growth potential and the second empha- in other countries, the operational flexibility of
sizes operational flexibility across multinational global export-focused subsidiaries helps them
networks. MNEs often exercise within-country exploit opportunities generated by fluctuating
options upon learning about a country’s economic economic conditions between countries (Kogut &
potential (Kogut, 1991), and use an existing subsi- Kulatilaka, 1994a; Tang & Tikoo, 1999). For exam-
diary in that country as a sensor for deciding when ple, when one country’s currency sharply depreci-
to increase or decrease a commitment (Reuer & ates, MNE subsidiaries can shift production to take
Tong, 2005). In this scenario, local responsiveness advantage of reduced labor and input costs (Jacque
and chains of incremental options in individual & Vaaler, 2001). MNE networks can be designed so
subsidiaries are central to maintaining the within- that subsidiaries acquire raw materials and other
country growth option (Kulatilaka & Perotti, 1998; inputs from cheaper local markets, produce inter-
Song, 2002; Tong, Reuer, & Peng, 2008). mediate and finished products in low-cost manu-
The across-country option maximizes operational facturing locations, and then redirect the products
flexibility by shifting production and sourcing to more lucrative export markets (Chung, Lu, &
among affiliated subsidiaries according to changes Beamish, 2008; Miller, 1992; Sundaram & Black,
in host-country economies, with the most impor- 1992).
tant factors being integration and interaction with Enacting shifts in value chain activities is more
other subsidiaries within the MNE network. Accord- difficult for subsidiaries oriented toward local
ing to this option, when one subsidiary encounters markets. Rangan (1998) suggests that orientation
difficulties, its problems may be solved through toward local responsiveness inhibits attempts to
interaction with sister subsidiaries in its multi- increase flexibility across multinational networks.
national network (Belderbos & Zou, 2009; Johnson, Further, subsidiaries with strong local market
1995; Kogut & Kulatilaka, 1994a; Reuer & Leiblein, orientations may have fewer interactions with
2000; Roth & Morrison, 1990; Tang & Tikoo, 1999). subsidiaries in other countries, and therefore
MNE flexibility can therefore be conceptualized as experience difficulties when attempting to coordi-
an opportunity cost, in that across-country options nate production across borders. This is consistent
investment may mean giving up a degree of local with the real options logic that firms tend to
responsiveness in exchange for operational net- behave in a path-dependent manner (Kogut &
work flexibility (Rangan, 1998). Kulatilaka, 1994b; Rangan, 1998). Since the focus
of a within-country orientation is host-country
Real Options Orientation and Subsidiary economic potential, abrupt demand reduction
Expansion/Contraction during an economic crisis can make a location
The focus of a subsidiary’s operations has important unattractive for MNE subsidiaries interested in
implications for the path-dependent development taking advantage of that potential. However, the
of its real options orientation (Rangan, 1998). same crisis can support low-cost manufacturing
Kogut and Kulatilaka (1994b) suggest that real expansion if subsidiaries exploit across-country
options investments made a priori reveal their value operational flexibility options via their multina-
during times of uncertainty, and Rangan (1998) tional networks. Based on this background,
maintains that for MNE subsidiaries to be flexible Hypothesis 1 is established as:
across borders during any given period, they must
have been originally developed with across-country Hypothesis 1: During times of economic crisis,
flexibility in mind. Thus a subsidiary with a a subsidiary with a stronger orientation toward
stronger focus on global exports than on local an across-country flexibility option is more likely
markets is likely to develop an across-country flexi- to expand its operations.
bility orientation, and a subsidiary with a stronger
focus on local markets is likely to develop a within- Interaction Effects between Real Options
country growth orientation. Orientation and Network-based Characteristics
The primary implication of an across-country
orientation is that MNE subsidiaries will benefit Relative performance. A poorly performing
from a geographically dispersed network for subsidiary is an indicator that an MNE might
reconfiguring value chain activities during times benefit from restructuring (Haynes, Thompson, &
of uncertainty.3 Since they are more likely to have Wright, 2003). Retaining such a subsidiary during
previously interacted with network subsidiaries a crisis period can only add to existing problems,

Journal of International Business Studies

 Subsidiary expansion/contraction at times of economic crisis Chris C Chung et al
503

which is why MNE executives demand that every (Birkinshaw et al., 2000; Tang & Tikoo, 1999), thus
subsidiary demonstrate its worth during times of providing more avenues for addressing perfor-
adverse environmental change (Birkinshaw & mance problems. Based on this rationale, we
Hood, 1997; Feinberg & Keane, 2006). Therefore, propose:
during times of economic crisis, MNEs tend to
reexamine past internationalization efforts, beginn- Hypothesis 2: During times of economic crisis,
ing with subsidiaries experiencing performance the lower the performance of a subsidiary relative
problems (Rugman, 1979). During restructuring, to other subsidiaries in the same network, the
an MNE is likely to focus initially on subsidiaries in more likely it is that a subsidiary with a stronger
crisis-stricken countries, but performance evalua- focus on the across-country option will expand
tion usually entails comparisons of poorly perform- its operations.
ing subsidiaries in crisis-stricken countries with all
other subsidiaries in the same network. Subsidiaries
in crisis-stricken countries may not be the worst
performers – those with across-country orientations Network size. An important difference between the
may be better performers due to manufacturing within- and across-country options is tied to a
cost advantages. subsidiary’s independence from (or interdepen-
For subsidiaries with within-country orientations, dence with) other subsidiaries in the same
decreased local market demand can exacerbate network (Kogut & Kulatilaka, 1994a). Subsidiaries
existing performance problems and encourage with stronger within-country orientations are more
contraction. Contraction is made easier when independent: therefore network size is not as
subsidiaries with this orientation are isolated, since important when they must decide to expand or
the impacts are less likely to affect others in the contract their operations. In other words, subsi-
same network (Birkinshaw, Holm, Thilenius, & diaries oriented toward the within-country option
Arvidsson, 2000). Compared with subsidiaries with are less likely to benefit from the across-country
across-country orientations, those with within- operational flexibility associated with large subsi-
country orientations are more likely to work in diary networks (Lee & Makhija, 2009a; Monteiro
isolation because their focus is on local adaptation, et al., 2008).
resulting in reduced compatibility (Johnson, 1995; In contrast, subsidiaries with stronger across-
Roth & Morrison, 1990). Accordingly, sudden country orientations are more dependent on other
demands for global coordination may not be readily network subsidiaries, since any shift in value chain
achievable – if attempted, considerable disagree- activities must involve at least two parties (Allen &
ment and disharmony may arise (Chung & Beamish, Pantzalis, 1996). Having subsidiaries in multiple
2005a; Monteiro, Arvidsson, & Birkinshaw, 2008), countries enhances flexibility by allowing subsidi-
exacerbating performance problems for within- aries to coordinate production in response to
country-oriented subsidiaries. environmental change. Benefits from operational
Poor performance is not as problematic for flexibility are especially important when host-
subsidiaries that have an across-country orienta- country operating environments become hostile
tion, since cost advantages maintain or increase the (Pantzalis, Simkins, & Laux, 2001; Tang & Tikoo,
potential to benefit from operational flexibility 1999). Greater breadth in subsidiary dispersion
across a network during times of economic crisis, provides greater operational flexibility to subsidi-
thanks to exchange rate depreciation, lower factor aries in trouble (Allen & Pantzalis, 1996), and
costs, and other favorable trade conditions (Jacque membership in a large network can provide oppor-
& Vaaler, 2001; Tambunan, 2000). For subsidiaries tunities for redirecting intermediate and finished
with across-country orientations, past interactions goods to more lucrative export markets (Lee &
with other subsidiaries in the same network may Makhija, 2009b). Based on this rationale, we
result in advantageous production shifts (Kogut & propose:
Kulatilaka, 1994a; Rangan, 1998). Especially com-
pared with isolated subsidiaries, the operating Hypothesis 3: During times of economic crisis,
scopes of well-connected subsidiaries oriented the larger the multinational network, the more
toward across-country flexibility are not limited likely it is that a subsidiary oriented toward the
to specific subsidiaries, but are linked to those of across-country flexibility option will expand its
multiple subsidiaries located in different countries operations.

Journal of International Business Studies

 Subsidiary expansion/contraction at times of economic crisis Chris C Chung et al
504

Network redundancy. Decreased manufacturing RESEARCH METHOD

costs can make subsidiaries candidates for produc-
tion shifts, but the potential is lower when subsi- Data and Sample
diaries in the same network experience similar cost Foreign subsidiary data were derived from the Trend
decreases – in other words, when considerable Survey of Overseas Business Activities, published
macroeconomic overlap exists in multiple locations annually by the Japanese Ministry of Economics,
(Fisch, 2008). This fits with Kogut’s (1989) descri- Trade and Industry (METI). METI gathers informa-
ption of multinationality as a bundle of options tion on the business activities of foreign subsidi-
whose value grows as differences across options aries for all Japanese corporations except those in
increase (Girotra, Terwiesch, & Ulrich, 2007; real estate, finance, and insurance. METI distributes
McGrath, 1999). When MNE subsidiaries in diff- copies of the Trend Survey (one for the parent firm
erent countries have little overlap, the resulting and one for each of its foreign subsidiaries), and
diversity gives the greatest advantages to subsi- in some cases conducts supplementary interviews
diaries oriented toward the across-country option – by phone or other means. The survey purpose is
in other words, the combination of a strong focus to support industrial and trade policy decisions:
on the across-country option and dispersal across therefore the data are organized according to
multiple countries with diverse macroeconomic industry type. Compiled data are published by the
conditions provides super-additive synergies during Enterprise Statistics Office of the Research and
times of uncertainty (Tanriverdi & Venkatraman, Statistics Department and the Trade and Invest-
2005). In such scenarios, joint value becomes ment Facilitation Division of the Trade and Eco-
greater than the sum of the two combinations nomic Cooperation Bureau.
(Davis & Thomas, 1993). For the empirical context we used the 1997 Asian
However, when macroeconomic conditions are economic crisis. Our sample consists of subsidiaries
similar in multiple countries where an MNE in the five countries most affected by the crisis:
operates subsidiaries, the value of across-country Thailand, Indonesia, South Korea, Malaysia, and
flexibility is lower than where a broader range of the Philippines (UNCTAD, 2000). Our observation
macroeconomic diversity exists. In environments period was from 1997 to 2001, a sufficient time
marked by similar macroeconomic conditions, the frame for capturing the real impacts of the crisis, and
option values of individual subsidiaries are only for mitigating a potentially serious right-censoring
partly additive, and become sub-additive or redun- problem. Our sample was limited to manufacturing
dant when their option characteristics overlap with subsidiaries owing to the likelihood that the manu-
those of other affiliates. Accordingly, a subsidiary facturing, service, and trade industries may not
will have lower across-country option value if the have reacted to the crisis in the same manner. We
environmental changes it faces resemble those followed the established real options research
faced by sister subsidiaries. When strong correla- practice of focusing on manufacturing subsidiaries
tions exist among macroeconomic changes in in order to reduce the potential for compounding
countries with MNE subsidiaries, those oriented across different industry types (Allen & Pantzalis,
toward the across-country option will have less 1996; Belderbos & Sleuwaegen, 2005). The final
room for production shifts.4 Our final hypothesis sample consisted of 1519 manufacturing subsidi-
reflects the idea that greater overlap in the poten- aries of 471 Japanese corporations in 52 industries,
tial value of across-country options among sub- operating in five Asian countries between 1997
sidiaries in an MNE network reduces the ability and 2001.
of across-country-oriented subsidiaries to exploit
country differences by shifting sourcing and pro- Variables
duction to subsidiaries with more favorable cost
structures: Dependent variable. The dependent variable for the
four hypotheses was a change-based continuous
Hypothesis 4: During times of economic crisis, measure of subsidiary expansion/contraction from
the greater the degree of redundancy in macro- year t1 to year t.5 We used two measures to
economic conditions among subsidiary locations operationalize subsidiary expansion/contraction:
in a multinational network, the less likely it is percentage change in subsidiary sales, and percen-
that an across-country option-oriented subsidiary tage change in subsidiary employees. To control
will expand its operations. for size bias, percentages were used instead of

Journal of International Business Studies

 Subsidiary expansion/contraction at times of economic crisis Chris C Chung et al
505

unit changes. Any analysis of subsidiary expansion/ 25.0

contraction without considering continuation/

Percentage of parent firm-year observations (%)

termination will create a selectivity problem.
During the observation period, 187 of the 1519 20.0
manufacturing subsidiaries in the sample were
terminated. As we will explain in the statistical
analysis section, a two-stage procedure was used 15.0
in response to the selectivity problem arising from
subsidiary continuation/termination (Heckman,
1979; Wooldridge, 1995). 10.0

Independent variables. Given that our dependent

variable is a change-based measure, to avoid reverse 5.0
causality problems we investigated the effects of
independent variables at time t1 on the change-
based dependent variable from time t1 to time t. 0.0
Our main independent variable was the real 0-0.1 0.1-0.2 0.2-0.3 0.3-0.4 0.4-0.5 0.5-0.6 0.6-0.71
Within-parent firm variation (standard deviation)
options orientation of each subsidiary, reflecting
the relative importance of the across-country Figure 1 Distribution of within-parent firm variation on
flexibility and within-country growth options subsidiaries’ export ratios.
(Hypothesis 1). We used the ratio of export vs Notes: Percentage of parent firm-year observations in each
local sales at the subsidiary level, since it is an within-parent firm variation category; N¼27,837 parent firm-
year observations worldwide.
efficient proxy for the relative importance of either
option embedded in each subsidiary. For the sake
of consistency, we used the export ratio with all
empirical models. regions, we computed an average return on sales
As explained above, we believe it is inappropriate (RoS) for all other subsidiaries in the same network,
to assume that the real options orientations of then subtracted the RoS for a subsidiary of interest
subsidiaries are automatically identical because from the average RoS for all affiliated subsidiaries.
they belong to the same parent firm. We therefore This reverse operationalization of poor relative per-
examined variation in subsidiaries belonging to the formance allows for a more intuitive interpretation,
same firm in terms of export ratios by calculating since our theoretical arguments focus on the rela-
within-firm variation across subsidiaries for a given tively poor performance of individual subsidiaries.
year, and found that some firms have a standard The higher the value of this variable, the lower the
deviation as high as 0.71. As shown in Figure 1, the relative performance of a subsidiary compared with
majority of parent firms had large within-firm its affiliates in the same parent network.
variation in subsidiary export ratios, with many The second interaction we examined was with
implementing subsidiary-specific strategies empha- multinational network size (Hypothesis 3) – that is,
sizing either export or local sales. For example, in the number of countries in which affiliated man-
1998 one firm had 19 subsidiaries worldwide and ufacturing subsidiaries are located in a given year.
a within-firm standard deviation of 0.38. Of the 19 Given that country risks are homogeneous for
subsidiaries, two were located in a country affected subsidiaries located in the same host country, this
by the 1997 Asian crisis; although operating in is an adequate measure from a real options theory
the same country they exhibited different orienta- perspective. A second possible measure of network
tions – an export ratio of 0.07 in one and 0.55 in size is the number of affiliated manufacturing
the other.6 subsidiaries in a given year. These variables are
Hypothesis 2 addressed the interaction between highly correlated (0.90); we chose the number of
the main independent variable and the relative countries in the network because we believe it is
performance of a focal subsidiary in the context of more relevant to real options theory. The number
other affiliated subsidiaries in the same parent of subsidiaries was used as a robustness check.
network. Since a focal subsidiary’s performance The third interaction effect was with the averaged
must be compared with subsidiaries owned by the sum of exchange rate correlations between one sub-
same parent firm but located in other countries or sidiary and other manufacturing subsidiaries in the

Journal of International Business Studies

 Subsidiary expansion/contraction at times of economic crisis Chris C Chung et al
506

same network but in other countries or regions. (Hannan & Freeman, 1984) during times of eco-
Following the example of Belderbos and Zou (2009), nomic crisis, using parent firm sales and subsidiary
we calculated annual correlations of monthly real sales as the size measures. We incorporated two
exchange rates between a subsidiary’s host country types of parent firm experience – international and
and other countries with subsidiaries from the same local – to control for the ability of MNEs to manage
parent firm in a given year. We then averaged the foreign subsidiaries in multinational networks and
sums of annual correlations of real exchange rates host countries, respectively. We measured parent
of a subsidiary’s host country and other countries firm international experience as the number of
associated with the network. We extended two years with foreign operations, and local experience
empirical aspects of Belderbos and Zou’s network as the number of years operating in a specific host
correlation measure: (a) while they computed country. Acknowledging a liability associated with
multinational network correlations for nine Asian newness (that is, older subsidiaries may be more
countries, we incorporated all countries in the efficient than younger ones because they have
world, resulting in a more complete measure of more cumulative experience, more established
multinational network correlations; (b) while they organizational routines, a more experienced work-
used static network correlations measured in 1995 force, etc.; Stinchcombe (1965)), we incorporated
only, we used a more dynamic network correlation subsidiary age to control for the experience factor
variable reflecting year-to-year changes in exchange at the subsidiary level. To control for other
rates and network configurations. unobserved effects, we used dummies for calendar
year, country, industry, and parent firm. For
Control variables. Our real options logic was based manufacturing industry dummies, we used the
on competitive devaluation in crisis-stricken coun- Japanese equivalent of a two-digit Standard Indus-
tries: therefore we controlled for structural try Classification code. In our empirical models we
devaluation by operationalizing it as change in included dummies for 5 years, five countries, 52
the institutional development of crisis-stricken manufacturing industries, and 471 parent firms.
countries (Chung & Beamish, 2005b). A negative
change was perceived as indicating structural Statistical Analysis
devaluation and a positive change as indicating If subsidiary continuation/termination is not taken
structural enhancement, with different magnitudes into consideration, any analysis of subsidiary
representing different change values. We used expansion/contraction may encounter selectivity
Chan et al. (2008) institutional development score problems. We used a two-stage procedure in res-
to compute change in institutional development ponse to this potential problem (Heckman, 1979).
from year t1 to year t in individual crisis-stricken Since our analysis was based on panel data, we used
countries. a panel data extension of this procedure as
According to the real options literature, different described in Wooldridge (1995). In the first-stage
subsidiary orientations determine accompanying selection model we estimated a probit of subsidiary
options during times of uncertainty (Kogut & continuation/termination on xi for each t (i.e.,
Kulatilaka, 1994b; Rangan, 1998). However, a sub- period-specific estimation) and calculated the
sidiary may respond to an economic crisis by inverse Mills ratio lit for all i and t. For the first-
dramatically altering its export ratio. To control stage model we used a set of independent variables
for this alternative explanation, we incorporated similar to those used in the second-stage model, but
export ratio change for subsidiaries in crisis-stricken with a different number of variables to minimize
countries. Another growth option is change in the the identification problem (Sartori, 2003). We
number of subsidiaries in each host country, but excluded the interaction terms and added a new
this occurs at the parent firm rather than subsidiary variable: whether a focal subsidiary is the parent
level. Since our focus was on the subsidiary level, firm’s only subsidiary in a host country in a given
we controlled for this growth option possibility. To year – that is, a country platform subsidiary.7 In
control for what takes place outside crisis-stricken the main second-stage model we ran a pooled
countries, we also included change in the number linear regression of subsidiary expansion/contrac-
of subsidiaries in the rest of the world. tion for continuing subsidiaries with the selectivity
We incorporated the effects of parent firm and correction term lit, and corrected the asymptotic
subsidiary size to control for the liability of small- variance of b for general heteroskedasticity and
ness (Stinchcombe, 1965) and structural inertia serial correlation.8

Journal of International Business Studies

 Subsidiary expansion/contraction at times of economic crisis Chris C Chung et al
507

Given the study objective, we used random variables: biserial correlations between dichoto-
effects as our estimation in the second-stage model. mous and interval variables, and tetrachoric corre-
Our focus was on between-subsidiary variation (i.e., lations between two dichotomous variables. We
differential effects of real options orientations conducted variance inflation factor and tolerance
between subsidiaries) rather than within-subsidiary level tests, and found no evidence of a multi-
variation (i.e., the changing effects of real options collinearity problem. The sales percentage-change
orientation in a subsidiary over time). Note also variable mean was 0.24, with a standard deviation
that the fixed-effects model does not allow the of 0.95. The employee percentage-change variable
inclusion of variables that do not vary over time – mean was 0.12, with a standard deviation of 0.64.
that is, the country, industry, and parent-firm The correlation between the sales and employee
dummies used to control for unobserved hetero- change variables was 0.41. This positive correlation
geneity. indicates that subsidiary sales and employee num-
bers generally changed in the same direction, with
RESULTS sales changing to a greater degree. The export
Descriptive statistics and correlation matrices for ratio change (control variable) mean was 0.00, with
the variables are presented in Table 1. Pearson a standard deviation of 0.16, suggesting that
correlations were calculated between two interval subsidiaries may not be capable of dramatically

Table 1 Descriptive statistics and correlations

Mean s.d. 1 2 3 4 5 6 7 8

1 Sales percentage change 0.24 0.95 1.00

2 Employment percentage change 0.12 0.64 0.41 1.00
3 Export ratio 0.26 0.35 0.27 0.06 1.00
4 Relative performance 0.03 0.71 0.09 0.01 0.03 1.00
5 Network size 8.06 7.04 0.01 0.01 0.05 0.02 1.00
6 Network redundancy 2.37 2.90 0.02 0.01 0.06 0.02 0.52 1.00
7 Interaction b/w rel. performance and export ratio 0.04 0.35 0.62 0.03 0.45 0.16 0.00 0.01 1.00
8 Interaction b/w network size and export ratio 2.15 9.71 0.38 0.03 0.66 0.03 0.19 0.08 0.37 1.00
9 Interaction b/w network redundancy and 0.20 1.20 0.09 0.02 0.50 0.01 0.14 0.21 0.11 0.62
export ratio
10 Change in institutional environment 0.08 0.23 0.01 0.01 0.00 0.01 0.01 0.22 0.01 0.01
11 Parent firm sales (log) 11.56 3.62 0.01 0.00 0.04 0.00 0.49 0.33 0.00 0.08
12 Subsidiary sales (log) 6.99 3.18 0.02 0.01 0.02 0.05 0.16 0.12 0.02 0.03
13 Parent firm international experience 15.67 10.42 0.00 0.01 0.01 0.02 0.38 0.24 0.01 0.10
14 Parent firm local experience 13.27 9.60 0.03 0.00 0.07 0.03 0.26 0.17 0.01 0.03
15 Subsidiary age 12.42 9.06 0.04 0.02 0.09 0.06 0.14 0.12 0.02 0.05
16 Change in subsidiary export ratio 0.00 0.16 0.29 0.05 0.70 0.03 0.01 0.02 0.36 0.48
17 Change in no. of subsidiaries in each host country 0.04 0.41 0.00 0.01 0.01 0.01 0.10 0.05 0.01 0.07
18 Change in no. of subsidiaries outside crisis countries 0.37 1.49 0.02 0.01 0.00 0.02 0.18 0.06 0.00 0.07

9 10 11 12 13 14 15 16 17 18

9 Interaction b/w network redundancy and 1.00

export ratio
10 Change in institutional Environment 0.06 1.00
11 Parent firm sales (log) 0.07 0.00 1.00
12 Subsidiary sales (log) 0.02 0.00 0.14 1.00
13 Parent firm international experience 0.08 0.01 0.15 0.12 1.00
14 Parent firm local experience 0.03 0.01 0.18 0.14 0.44 1.00
15 Subsidiary age 0.04 0.00 0.09 0.15 0.33 0.58 1.00
16 Change in subsidiary export ratio 0.28 0.05 0.00 0.01 0.02 0.02 0.04 1.00
17 Change in no. of subsidiaries in each host country 0.04 0.04 0.07 0.01 0.01 0.06 0.02 0.01 1.00
18 Change in no. of subsidiaries outside crisis countries 0.03 0.04 0.09 0.03 0.02 0.04 0.01 0.04 0.32 1.00
Note: Correlations with 5 year dummies, five country dummies, 52 industry dummies, and 471 parent firm dummies are not shown because of space
constraints.

Journal of International Business Studies

 Subsidiary expansion/contraction at times of economic crisis Chris C Chung et al
508

changing their export ratios owing to cost and during times of economic crisis. Conversely, this
capability issues associated with switching from suggests that when a local market collapses during
a local to export orientation (Rangan, 1998). times of economic crisis, then the stronger the
Regarding results estimated from the first-stage within-country orientation of a subsidiary, the
selection model, we found that during times of greater the likelihood of it experiencing decreasing
economic crisis a subsidiary was more likely to sales. Hypothesis 1 is therefore supported.
continue operations if it had a higher export ratio, The negative coefficient of the relative poor
was the only subsidiary of a parent firm in a host performance variable in Model 1 indicates that,
country, or had a larger amount of total sales. The during times of economic crisis, poorly performing
main effects of relative performance and network subsidiaries generally face reduced sales, but the
size indicate that a subsidiary exhibiting stronger main effect was not statistically significant. Its
relative performance and larger network size was interaction effect with the real options orientation
more likely to continue operating, but the results of the subsidiary is worth noting. In Model 2,
were not statistically significant. Moreover, accord- the positive interaction between relative poor
ing to the main effect of network redundancy, a performance and export ratio (b¼0.228; po0.01)
subsidiary was less likely to continue operating if its indicates that even though poorly performing sub-
location was highly correlated with those of other sidiaries contract their operations in general, some
subsidiary locations in the same network in terms subsidiaries expand if they have greater focus on
of macroeconomic conditions – but again, this the across-country operational flexibility option,
result was not statistically significant. A panel data thus supporting Hypothesis 2.
procedure based on Wooldridge’s (1995) two-stage In Model 3 the positive coefficient of the inter-
model generated period-specific probit estimations. action term between network size and export ratio
The five first-stage estimations used to compute the (b¼0.675; po0.01) indicates that during times of
selectivity correction term for each year between crisis, the larger the size of a multinational network,
1997 and 2001 are reported in the Appendix. the greater the likelihood of a subsidiary with an
Estimates for the second-stage models with across-country orientation experiencing an incre-
corrections for selectivity are shown in Table 2 ase in sales, thus supporting Hypothesis 3. How-
(percentage change of subsidiary sales) and Table 3 ever, the main effect of network size alone did not
(percentage change of subsidiary employees). In significantly increase subsidiary sales in Model 1.
Table 1 we first included all main effects variables Combined, these results indicate that membership
(export ratio, relative performance, network size, in a large multinational network does not necessa-
and network redundancy) in addition to various rily help subsidiaries in crisis-stricken countries
control variables (Model 1). Next, we introduced increase sales. Instead, a more important factor is
each interaction term separately before creating a how a large multinational network meshes with the
full model: the interaction between relative perfor- real options orientations of individual subsidiaries:
mance and export ratio in Model 2, the interaction if the orientation is local, membership in a large
between network size and export ratio in Model 3, multinational network is not as useful as when the
and the interaction between network redundancy orientation is toward across-country flexibility. This
and export ratio in Model 4. Model 5 is the full explains our observation of a significant and
model containing all three interaction terms. The positive interaction effect between network size
same method was applied in Table 3 (Models 6–10). and export ratio in Model 2, and a non-significant
To minimize the potential for multicollinearity we effect of network size alone in Model 1.9
centered all variables used to create the interaction In addition to the two positive interaction effects
terms. of the across-country option embedded in indivi-
We used the ratio of export vs local sales at the dual subsidiaries, we also investigated that option’s
subsidiary level to operationalize the relative negative interaction effect when combined with
importance of across-country flexibility versus network redundancy. During times of economic
within-country growth options embedded in each crisis, across-country-oriented subsidiaries can make
subsidiary. The positive coefficient of the export better use of a network that is dispersed across
ratio variable in Model 1 (b¼1.121; po0.01) multiple countries. However, if affiliated subsidi-
indicates that the stronger the subsidiary’s across- aries in the same network experience similar
country flexibility orientation, the greater the production cost decreases, the production shift
likelihood of the subsidiary increasing its sales may not be as large as when greater variety in

Journal of International Business Studies

 Subsidiary expansion/contraction at times of economic crisis
Table 2 Percentage change in subsidiary sales

Model 1 Model 2 Model 3 Model 4 Model 5 (full model)

b s.e. b s.e. b s.e. b s.e. b s.e.

Independent variables
Export ratio (Hypothesis 1) 1.121*** (0.122) 1.119*** (0.120) 1.319*** (0.118) 1.072*** (0.122 1.212*** (0.112)
Relative performance 0.003 (0.006) 0.037*** (0.007) 0.004 (0.006) 0.004 (0.006) 0.027*** (0.006)
Network size 0.100 (0.363) 0.103 (0.358) 0.110 (0.349) 0.123 (0.362) 0.226 (0.331)
Network redundancy 0.009 (0.132) 0.040 (0.130) 0.051 (0.127) 0.072 (0.132) 0.160 (0.121)
Interaction b/w relative performance and export ratio 0.228*** (0.019) 0.173*** (0.017)
(Hypothesis 2)
Interaction b/w network size and export ratio 0.675*** (0.033) 1.087*** (0.039)
(Hypothesis 3)
Interaction b/w network redundancy and 0.538*** (0.109) 1.437*** (0.121)
export ratio (Hypothesis 4)

Control variables
Selectivity correction (lit) 3.309** (1.382) 3.663*** (1.362) 3.789*** (1.327) 3.258** (1.378) 4.119*** (1.259)
Change in institutional environment 1.843 (1.238) 1.788 (1.220) 1.327 (1.190) 1.872 (1.235) 1.102 (1.129)

Chris C Chung et al
Parent firm sales (log) 0.092 (0.119) 0.109 (0.117) 0.043 (0.114) 0.088 (0.118) 0.009 (0.108)
Subsidiary sales (log) 0.259*** (0.097) 0.299*** (0.095) 0.242*** (0.093) 0.249*** (0.096) 0.218** (0.088)
Parent firm international experience 0.008 (0.084) 0.005 (0.082) 0.025 (0.080) 0.010 (0.083) 0.039 (0.076)
Parent firm local experience 0.07 (0.048) 0.076 (0.047) 0.101** (0.046) 0.061 (0.047) 0.085** (0.043)
Subsidiary age 0.072* (0.043) 0.055 (0.042) 0.021 (0.041) 0.081* (0.043) 0.022 (0.039)
Change in subsidiary export ratio 2.930*** (0.341) 2.985*** (0.336) 3.008*** (0.328) 2.615*** (0.347) 1.665*** (0.319)
Change in no of subsidiaries in each 0.377 (0.585) 0.322 (0.577) 0.963* (0.563) 0.326 (0.584) 1.045* (0.534)
host country
Change in no. of subsidiaries outside 0.036 (0.111) 0.001 (0.110) 0.002 (0.107) 0.042 (0.111) 0.016 (0.101)
crisis countries
Year dummies Included Included Included Included Included
Journal of International Business Studies

Country dummies Included Included Included Included Included

Industry dummies Included Included Included Included Included
Parent firm dummies Included Included Included Included Included
Constant 0.210 (9.835) 0.169 (9.689) 0.280 (9.447) 0.452 (9.811) 0.487 (8.964)
Wald statistics 1256.58*** 1443.93*** 1771.79*** 1387.01*** 2510.09***
Note: Results are based on 1519 subsidiaries of 471 Japanese MNEs, operating in 52 manufacturing industries of five crisis-stricken countries between 1997 and 2001.
*po0.10; **po0.05; *** po0.01; all from two-tailed tests.

509
Journal of International Business Studies

510
Subsidiary expansion/contraction at times of economic crisis
Table 3 Percentage change in subsidiary employees

Model 6 Model 7 Model 8 Model 9 Model 10

(full model)

b s.e. b s.e. b s.e. b s.e. b s.e.

Independent variables
Export ratio (Hypothesis 1) 0.364*** (0.134) 0.137 (0.152) 0.319** (0.138) 0.204 (0.138) 0.107 (0.154)
Relative performance 0.097 (0.144) 0.150 (0.144) 0.096 (0.144) 0.111 (0.143) 0.148 (0.144)
Network size 0.032 (0.024) 0.046* (0.024) 0.047* (0.024) 0.050** (0.024) 0.051** (0.024)
Network redundancy 0.071 (0.049) 0.069 (0.049) 0.072 (0.049) 0.091* (0.049) 0.091* (0.049)
Interaction b/w relative performance and export ratio 0.183*** (0.057) 0.113** (0.055)
(Hypothesis 2)
Interaction b/w network size and export ratio (Hypothesis 3) 0.022** (0.011) 0.032* (0.019)
Interaction b/w network redundancy and export ratio 0.205*** (0.044) 0.244*** (0.058)
(Hypothesis 4)

Control variables
Selectivity correction (lit) 0.769* (0.401) 0.795* (0.481) 0.773* (0.451) 0.767* (0.435) 0.779* (0.423)

Chris C Chung et al
Change in institutional environment 0.298 (0.486) 0.300 (0.486) 0.287 (0.486) 0.291 (0.485) 0.310 (0.485)
Parent firm sales (log) 0.021 (0.034) 0.02 (0.034) 0.02 (0.034) 0.021 (0.034) 0.022 (0.034)
Subsidiary sales (log) 0.007 (0.035) 0.005 (0.035) 0.006 (0.035) 0.007 (0.035) 0.008 (0.035)
Parent firm international experience 0.006 (0.012) 0.005 (0.012) 0.005 (0.012) 0.005 (0.012) 0.006 (0.012)
Parent firm local experience 0.016 (0.016) 0.016 (0.016) 0.017 (0.016) 0.020 (0.016) 0.019 (0.016)
Subsidiary age 0.021 (0.016) 0.022 (0.016) 0.022 (0.016) 0.025 (0.016) 0.024 (0.016)
Change in subsidiary export ratio 0.140 (0.138) 0.166 (0.138) 0.141 (0.138) 0.061 (0.139) 0.062 (0.140)
Change in no. of subsidiaries in each host country 0.090 (0.254) 0.090 (0.254) 0.108 (0.254) 0.113 (0.254) 0.086 (0.254)
Change in no. of subsidiaries outside crisis countries 0.003 (0.044) 0.003 (0.044) 0.004 (0.044) 0.004 (0.044) 0.003 (0.044)
Year dummies Included Included Included Included Included
Country dummies Included Included Included Included Included
Industry dummies Included Included Included Included Included
Parent firm dummies Included Included Included Included Included
Constant 0.153 (0.852) 0.228 (0.851) 0.143 (0.851) 0.074 (0.850) 0.122 (0.850)
Wald statistics 493.28*** 582.94*** 501.40*** 601.88*** 731.95***
Note: Results are based on 1519 subsidiaries of 471 Japanese MNEs, operating in 52 manufacturing industries of five crisis-stricken countries between 1997 and 2001.
*po0.10; **po0.05; ***po0.01; all from two-tailed tests.
 Subsidiary expansion/contraction at times of economic crisis Chris C Chung et al
511

macroeconomic conditions exist throughout the structural devaluation between 1997 and 2001,
network. The negative coefficient for the interac- results from a multivariate analysis indicate that
tion between network redundancy and export the impact of devaluation on subsidiary expansion/
ratio in Model 4 (b¼0.538; po0.01) confirms that contraction was not statistically significant. It is
the stronger the correlation of macroeconomic possible that the positive effect of competitive
conditions with other subsidiaries in the same devaluation offsets the negative effect of structural
parent network, the less likely it is that an across- devaluation. This explanation is consistent with
country oriented subsidiary will expand during the core real options premise that successful firms
times of economic crisis. Model 4 therefore sup- capitalize on uncertainty rather than run away
ports Hypothesis 4. These results did not change from it (Amram & Kulatilaka, 1999; Garud, Kumar-
substantially when we added all three interaction aswamy, & Nayyar, 1998).11
terms to the full model (Model 5).
We also estimated the second-stage model using DISCUSSION AND CONCLUSION
percentage changes of subsidiary employees as a This paper makes several contributions to the real
dependent variable (Table 3). In Model 6, we found options literature. In MNE contexts, real options
that the stronger the orientation of a subsidiary to studies originally focused on the impact of multi-
global export, the greater the likelihood of that nationality on firm values at the corporate level. As
subsidiary increasing the number of employees the field developed, researchers started using a
(b¼0.364; po0.01). We also found a positive multinational network perspective when focusing
interaction effect between relative poor perfor- on actual configurations of international invest-
mance and export ratio in Model 7 (b¼0.183; ments. Given the emphasis on a multinational
po0.01), a positive interaction effect between net- network perspective, real options researchers have
work size and export ratio in Model 8 (b¼0.022; primarily examined overall MNE network charac-
po0.05), and a negative interaction effect between teristics at the expense of addressing the real
network redundancy and export ratio in Model 9 options orientations of individual subsidiaries that
(b¼ 0.205; po0.01). These results did not change constitute MNE networks – the focus of our
in the full Model 10, thus supporting all four research. We also examined how the real options
hypotheses. While the results for the percentage orientations of individual subsidiaries mesh with
change of subsidiary employees are consistent with the overall characteristics of their MNE networks.
the results for the percentage change of subsidiary Our results show that general MNE characteristics
sales, the employment results are weaker than the are not the sole determinants of subsidiary expan-
sales results in terms of both the significance level sion/contraction – the real options orientation of
and coefficient size. We speculate that change in individual subsidiaries is an equally important
employees may be less evident than change in sales factor.
because people can do overtime or work less hours Another shortcoming in the real options litera-
during times of uncertainty. Change in employees ture is that researchers have tended to investigate
is also a more sticky issue, because it is harder to either within-country or across-country options
hire or fire people compared with increasing or individually rather than simultaneously. According
decreasing sales.10 to our comparison of the two, the same economic
The control variable results shown in Tables 2 and crisis can spell disaster for subsidiaries with a strong
3 are worth noting. The selectivity correction term within-country focus while creating opportunities
coefficient is positive and significant in all second- for those with a strong across-country focus. Our
stage models, reflecting a covariance of error terms results indicate that subsidiaries with a stronger
in the first-stage (continuation/termination) and across-country focus are more likely to expand than
second-stage (expansion/contraction) equations, contract operations during times of economic
and underscoring the importance of correcting crisis, since they are positioned to extract benefits
the selectivity bias in our main empirical models. from production and sourcing shifts within their
Given that our real options logic is based on networks (Kogut & Kulatilaka, 1994a).
competitive devaluation during times of economic According to our results, subsidiaries oriented
crisis, we also controlled for the structural deval- toward the across-country option are more likely to
uation of crisis-stricken countries. Although the benefit from real options investments, especially
Table 1 descriptive statistics show that the five when host countries suffer severe damage during
crisis-stricken countries in our sample experienced times of economic crisis. In other words, MNE

Journal of International Business Studies

 Subsidiary expansion/contraction at times of economic crisis Chris C Chung et al
512

success requires having subsidiaries in different In terms of managerial implications, our results
host countries and knowing how to use them suggest that during times of economic crisis,
during times of crisis. As Rangan (1998: 220) notes, subsidiaries with a stronger focus on across-county
‘‘[to] be flexible in the current period, MNEs need to operational flexibility are better positioned to take
have planned and invested accordingly in previous advantage of their multinational networks. Aware-
periods.’’ However, benefits tied to an across- ness of how past investment decisions and stra-
country orientation are not cost-free: investing in tegic orientations affect specific subsidiaries during
that option means giving up the benefits of local crises can provide managers with a priori insight.
responsiveness when a country’s economy starts to Although operational flexibility is normally con-
recover. Instead of refuting a within-country per- sidered an adaptive and reactive response to envi-
spective outright, our findings clarify the boundary ronmental change, MNE subsidiaries may use their
condition under which one option or the other has operational flexibility to proactively redefine mar-
greater value. ket uncertainties. The across-country perspective
Whereas previous real options studies on MNE of real options theory helps firms battle with
flexibility have looked at the general effects of (McGrath, 1997), capitalize on (Amram & Kulatilaka,
multinational networks, we examined network- 1999; Garud et al., 1998), and even befriend the
specific conditional effects in relation to the rela- uncertainty that is a constant factor in today’s
tive importance of within-country versus across- global economy.
country options embedded in each subsidiary. We Our data sources emphasize large and established
found that during times of economic crisis, the Japanese firms rather than small or entrepreneurial
inferior/superior performance of a subsidiary in ones: therefore our sample may have biased our
comparison to its affiliated subsidiaries in the same results toward across-country operational flexibil-
parent network is not the sole determinant of ity. Future researchers may be interested in exam-
subsidiary contraction/expansion. When a local ining how small and/or entrepreneurial firms
market collapses, what really matters is a subsidi- manage their foreign subsidiaries, and how these
ary’s real options orientation, which determines its firms’ decisions regarding subsidiary expansion and
ability to take advantage of changing conditions. contraction during times of economic crisis differ
Similarly, even though multinational network size from those of their larger counterparts. They may
by itself may not affect expansion/contraction also be interested in comparing our results with
decisions for individual subsidiaries, the interaction those for non-crisis periods. Given country-specific
between network size and a subsidiary’s across- differences, there is also a need to examine other
country flexibility orientation does matter. This host countries.
may explain, at least in part, the non-significant or
mixed results from previous examinations of the
ACKNOWLEDGEMENTS
general characteristics of multinational networks
The authors are grateful to Department Editor Sea-Jin
(e.g., Allen & Pantzalis, 1996; Rangan, 1998; Reuer
Chang and three anonymous reviewers for excellent
& Leiblein, 2000; Tang & Tikoo, 1999).
comments and guidance. The authors are also grateful
We also examined a negative interaction effect in
to the Japanese Ministry of Economics, Trade and
the form of network redundancy, and found that
Industry for data support.
subsidiaries oriented toward the across-country
option are less likely to expand if their environ-
mental changes resemble those faced by other NOTES
1
affiliates in the same network. Accordingly, even Some researchers have examined the implications
if MNE subsidiaries are capable of shifting sourcing of real options for subsidiary growth and divestment
and production from one location to another, they using individual subsidiaries as the unit of analysis
are less likely to do so if fewer benefits are to be (e.g., Belderbos & Zou, 2007, 2009), but their focus
gained from exploiting country differences. This is has mostly been on environmental factors and not on
consistent with what real options theory suggests: the real options orientations of individual subsidiaries.
2
MNEs positioned to take advantage of environ- The holder of a put option has the potential (but
mental changes do not have to do so when con- not the obligation) to sell a particular asset at a given
ditions are not favorable, and only those MNEs price – known as the exercise or strike price. A call
ready to take advantage of such changes can do so option confers the privilege of purchasing an asset at a
when opportunities arise. specified exercise price.

Journal of International Business Studies

 Subsidiary expansion/contraction at times of economic crisis Chris C Chung et al
513

3
Our focus is on two sources of uncertainty that experience, parent firm local experience, subsidiary
reflect economic conditions in a crisis-stricken country: age, change in subsidiary export ratio, change in the
changes in demand (Bell & Campa, 1997; Brouthers, number of subsidiaries in each host country, change in
Brouthers, & Werner, 2008; Campa, 1994) and changes the number of subsidiaries outside crisis countries, and
in factor prices due to fluctuations in exchange rates country dummies.
8
and input prices (Kogut & Chang, 1996; Pantzalis See Wooldridge (1995) for formal equations and
et al., 2001; Miller & Reuer, 1998; Rangan, 1998). detailed descriptions.
9
Both are at the center of the real options literature in We also operationalized network size by counting
MNE contexts. the number of subsidiaries under the same parent firm:
4
This sub-additive logic has been studied in the this alternative option did not affect our results.
10
contexts of alliances (Vassolo, Anand, & Folta, 2004), We also investigated the percentage change of
R&D projects (Girotra et al., 2007), and a regional subsidiary capitalization using the same independent
subsidiary network (Belderbos & Zou, 2009). variables of interest. While the directions of the coeffi-
5
Since the dependent variable emerges from a cients were generally consistent with those reported
change-based measure of subsidiary expansion/con- here, the results were not statistically significant. We
traction, we used the Trend Survey data from 1996 speculate that subsidiary capitalization may not be
to 2001. easily adjusted according to changes in external
6
The Trend Survey is conducted under the authority environments such as economic crises.
11
of Article 4 of the Statistical Reports Coordination Act We also ran fixed-effects models as a robustness
in Japan, therefore the names of the participating check. Since the fixed-effects model prevents the
corporations and subsidiaries are confidential. inclusion of variables that do not vary across time,
7
The following variables were included in the first- we could not add the time-invariant indicators for
stage model: country platform subsidiary, export ratio, countries, industries, or parent firms to the fixed-
relative performance, network size, network redun- effects model. Results for the fixed-effects models
dancy, change in institutional environment, parent were generally consistent with the random-effects
firm sales, subsidiary sales, parent firm international results.

REFERENCES
Allen, L., & Pantzalis, C. 1996. Valuation of the operating Campa, J. 1994. Multinational investment under uncertainty in
flexibility of multinational corporations. Journal of International the chemical processing industries. Journal of International
Business Studies, 27(4): 633–653. Business Studies, 25(3): 557–578.
Amram, M., & Kulatilaka, N. 1999. Real options: Managing Chan, C. M., Isobe, T., & Makino, S. 2008. Which country
strategic investment in an uncertain world. Boston: Harvard matters? Institutional development and foreign affiliate per-
Business School Press. formance. Strategic Management Journal, 29(11): 1179–1205.
Belderbos, R., & Sleuwaegen, L. 2005. Competitive drivers and Chi, T. 2000. Option to acquire or divest a joint venture.
international plant configuration strategies: A product-level Strategic Management Journal, 21(6): 665–687.
test. Strategic Management Journal, 26(6): 577–593. Chung, C. C., & Beamish, P. W. 2005a. Investment
Belderbos, R., & Zou, J. 2007. On the growth of foreign affiliates: mode strategy and expatriate strategy during times of
Multinational plant networks, joint ventures, and flexibility. economic crisis. Journal of International Management, 11(3):
Journal of International Business Studies, 38(7): 1095–1112. 331–355.
Belderbos, R., & Zou, J. 2009. Real options and foreign affiliate Chung, C. C., & Beamish, P. W. 2005b. The impact of
divestments: A portfolio perspective. Journal of International institutional reforms on characteristics and survival of foreign
Business Studies, 40(4): 600–620. subsidiaries in emerging economies. Journal of Management
Bell, G., & Campa, J. 1997. Irreversible investments and volatile Studies, 42(1): 35–62.
markets: A study of the chemical processing industry. Review of Chung, C. C., Lu, J., & Beamish, P. W. 2008. Multinational
Economics and Statistics, 79(1): 79–87. networks during times of economic crisis versus stability.
Birkinshaw, J., & Hood, N. 1997. An empirical study of deve- Management International Review, 48(3): 279–295.
lopment processes in foreign-owned subsidiaries in Canada and Coucke, K., Pennings, E., & Sleuwaegen, L. 2007. Employee
Scotland. Management International Review, 37(4): 339–364. layoff under different modes of restructuring: Exit, downsizing
Birkinshaw, J., Holm, U., Thilenius, N., & Arvidsson, N. 2000. or relocation. Industrial and Corporate Change, 16(2): 161–182.
Consequences of perception gaps in the headquarters–subsidiary Davis, R., & Thomas, L. 1993. Direct estimation of synergy:
relationship. International Business Review, 9(3): 321–344. A new approach to the diversity-performance debate. Man-
Bowman, E., & Hurry, D. 1993. Strategy through the option agement Science, 39(11): 1334–1346.
lens: An integrated view of resource investments and the Garud, R., Kumaraswamy, A., & Nayyar, 1998. Real options or
incremental-choice process. Academy of Management Review, fool’s gold? Perspective makes the difference. Academy of
18(4): 760–782. Management Review, 23(2): 212–214.
Brouthers, K., Brouthers, B., & Werner, S. 2008. Real options, Girotra, K., Terwiesch, C., & Ulrich, K. 2007. Valuing R&D
international entry mode choice and performance. Journal of projects in a portfolio: Evidence from the pharmaceutical
Management Studies, 45(5): 936–960. industry. Management Science, 53(9): 1452–1466.

Journal of International Business Studies

 Subsidiary expansion/contraction at times of economic crisis Chris C Chung et al
514

Hannan, M. T., & Freeman., J. 1984. Structural inertia and Pantzalis, C., Simkins, B., & Laux, P. 2001. Operational hedges
organizational change. American Sociological Review, 49(2): and the foreign exchange exposure of US multinational
149–164. corporations. Journal of International Business Studies, 32(4):
Haynes, M., Thompson, S., & Wright, M. 2003. The determi- 793–812.
nants of corporate divestment: Evidence from a panel of UK Rangan, S. 1998. Do multinationals operate flexibly? Theory
firms. Journal of Economic Behavior and Organization, 52(1): and evidence. Journal of International Business Studies, 29(2):
147–166. 217–237.
Heckman, J. 1979. Sample selection bias as a specification error. Reuer, J., & Leiblein, M. 2000. Downside risk implications of
Econometrica, 47(1): 153–161. multinationality and international joint ventures. Academy of
Hurry, D. 1993. Restructuring in the global economy: The Management Journal, 43(2): 203–214.
consequences of strategic linkages between Japanese and US Reuer, J., & Tong, T. W. 2005. Real options in international joint
firms. Strategic Management Journal, 14(Summer Special ventures. Journal of Management, 31(3): 403–423.
Issue): 69–82. Richardson, J. 1993. Parallel sourcing and supplier performance
Feinberg, S., & Keane, M. 2006. Accounting for the growth of in the Japanese automobile industry. Strategic Management
MNC-based trade using a structural model of US MNCs. Journal, 14(5): 339–350.
American Economic Review, 96(5): 1515–1558. Roth, K., & Morrison, A. 1990. An empirical analysis of the
Fisch, J. 2008. Investment in new foreign subsidiaries under integration-responsiveness framework in global industries.
receding perception of uncertainty. Journal of international Journal of International Business Studies, 21(4): 541–564.
Business Studies, 39(3): 370–386. Rugman, A. M. 1979. International diversification and the
Jacque, L., & Vaaler, P. 2001. The international control multinational enterprise. Lexington, MA: Lexington
conundrum with exchange risk: An EVA framework. Journal Books.
of International Business Studies, 32(4): 813–832. Sartori, A. E. 2003. An estimator for some binary-outcome
Johnson Jr, J. 1995. An empirical analysis of the integration- selection models without exclusion restrictions. Political
responsiveness framework: US construction equipment indus- Analysis, 11(2): 111–138.
try firms in global competition. Journal of International Business Sharp, D. J. 1991. Uncovering the hidden value in high-risk
Studies, 26(3): 621–635. investments. Sloan Management Review, 32(2): 69–74.
Kogut, B. 1989. A note on global strategies. Strategic Manage- Singh, K., & Yip, G. 2000. Strategic lessons from the Asian crisis.
ment Journal, 10(4): 17–31. Long Range Planning, 33(5): 706–729.
Kogut, B. 1991. Joint ventures and the option to expand and Song, J. 2002. Firm capabilities and technology ladders:
acquire. Management Science, 37(1): 19–33. Sequential foreign direct investments of Japanese electronics
Kogut, B., & Chang, S.-J. 1996. Platform investments and volatile firms in East Asia. Strategic Management Journal, 23(3):
exchange rates: Direct investment in the US by Japanese 191–210.
electronic companies. Review of Economics and Statistics, 78(2): Stinchcombe, A. L. 1965. Social structure and organizations. In
221–231. J. G. March (Ed.), Handbook of organizations. Chicago, IL: Rand
Kogut, B., & Kulatilaka, N. 1994a. Operating flexibility, global McNally.
manufacturing, and the option value of a multinational Sundaram, A. K., & Black, J. S. 1992. The environment and
network. Management Science, 40(1): 123–139. internal organization of multinational enterprises. Academy of
Kogut, B., & Kulatilaka, N. 1994b. Options thinking and Management Review, 17(4): 729–757.
platform investments: Investing in opportunity. California Tambunan, T. 2000. Development of small-scale industries during
Management Review, 36(2): 52–71. the new order government in Indonesia. Oxon, UK: Gower
Kulatilaka, N., & Perotti, E. C. 1998. Strategic growth options. Publishing Limited.
Management Science, 40(8): 744–758. Tang, C., & Tikoo, S. 1999. Operational flexibility and market
Lee, S., & Makhija, M. 2009a. The effect of domestic uncertainty valuation of earnings. Strategic Management Journal, 20(8):
on the real options value of international investments. Journal 749–761.
of International Business Studies, 40(3): 405–420. Tanriverdi, H., & Venkatraman, N. 2005. Knowledge relatedness
Lee, S., & Makhija, M. 2009b. Flexibility in internationalization: and the performance of multibusiness firms. Strategic Manage-
Is it valuable during an economic crisis? Strategic Management ment Journal, 26(2): 97–119.
Journal, 30(5): 537–555. Tong, T. W., & Reuer, J. J. 2007. Real options in multi-
McGrath, R. G. 1997. A real options logic for initiating national corporations: Organizational challenges and risk
technology positioning investments. Academy of Management implications. Journal of International Business Studies, 38(2):
Review, 22(4): 974–996. 215–230.
McGrath, R. G. 1999. Falling forward: Real options reasoning Tong, T. W., Reuer, J. J., & Peng, M. W. 2008. International joint
and entrepreneurial failure. Academy of Management Review, ventures and the value of growth options. Academy of
24(1): 13–30. Management Journal, 51(5): 1014–1029.
Miller, K. D. 1992. A framework for integrated risk management UNCTAD. 2000. World investment report 2000: Transnational
in international business. Journal of International Business corporations, market structure and competition policy. New
Studies, 23(2): 311–331. York: United Nations.
Miller, K. D., & Reuer, J. J. 1998. Firm strategy and economic Vassolo, R. S., Anand, J., & Folta, T. B. 2004. Non-additivity in
exposure to foreign exchange rate movements. Journal of portfolios of exploration activities: A real options-based
International Business Studies, 29(3): 493–513. analysis of equity alliances in biotechnology. Strategic Manage-
Monteiro, L., Arvidsson, N., & Birkinshaw, J. 2008. Knowledge ment Journal, 25(11): 1045–1061.
flows within multinational corporations: Explaining subsidiary Volberda, H. 1997. Building flexible organizations for
isolation and its performance implications. Organization fast-moving markets. Long Range Planning, 30(2):
Science, 19(1): 90–107. 169–183.
Pantzalis, A. 2001. Does location matter? An empirical analysis Wooldridge, J. 1995. Selection corrections for panel data models
of geographic scope and MNE market valuation. Journal of under conditional mean independence assumptions. Journal of
International Business Studies, 32(1): 133–155. Econometrics, 68(1): 115–132.

Journal of International Business Studies

 Subsidiary expansion/contraction at times of economic crisis
APPENDIX
See Table A1.

Table A1 Period-specific first-stage selection results

Dependent variable (Continuation ¼ 1; Termination ¼ 0) 1997 1998 1999 2000 2001

b s.e. b s.e. b s.e. b s.e. b s.e.

Independent variable
Export ratio 1.345 (2.101) 1.616** (0.805) 0.402** (0.201) 3.434** (1.722) 0.606** (0.250)
Country platform 1.235* (0.739) 0.740 (0.652) 0.413* (0.238) 0.305* (0.179) 0.117 (0.095)
Relative (poor) performance 0.026 (0.075) 0.014 (0.116) 0.037 (0.031) 0.005 (0.036) 0.003 (0.032)
Network size 0.051 (0.444) 0.043 (0.276) 0.110 (0.091) 0.077 (0.096) 0.031 (0.052)
Network redundancy 0.019 (0.103) 0.002 (0.001) 0.001 (0.001) 0.002 (0.005) 0.001 (0.003)
Change in institutional environment 1.968 (3.310) 0.611 (1.459) 0.294 (0.374) 0.041 (1.416) 0.326 (0.431)

Chris C Chung et al
Parent firm sales (log) 0.037 (0.213) 0.081 (0.079) 0.119* (0.069) 0.006 (0.048) 0.145*** (0.052)
Subsidiary sales (log) 0.299*** (0.097) 0.271*** (0.064) 0.253*** (0.051) 0.240*** (0.049) 0.132*** (0.027)
Parent firm international experience 0.016 (0.056) 0.022 (0.026) 0.003 (0.012) 0.003 (0.021) 0.014 (0.009)
Parent firm local experience 0.082 (0.091) 0.037 (0.031) 0.005 (0.014) 0.042 (0.027) 0.002 (0.011)
Subsidiary age 0.027 (0.050) 0.007 (0.030) 0.018* (0.010) 0.012 (0.023) 0.004 (0.010)
Change in subsidiary export ratio 0.477 (5.302) 0.862 (3.712) 0.410 (0.327) 2.132 (3.225) 0.002 (0.091)
Change in no. of subsidiaries in each host country 0.052 (0.877) 0.502 (0.793) 0.176 (0.232) 0.136 (0.509) 0.111 (0.246)
Change in no. of subsidiaries outside crisis countries 0.079 (0.130) 0.079 (0.164) 0.011 (0.054) 0.005 (0.079) 0.006 (0.039)
Country dummies Included Included Included Included Included
Constant 0.836 (2.918) 2.990 (1.876) 3.746** (1.874) 2.439** (1.135) 2.339*** (0.726)
Number of manufacturing subsidiaries 943 1104 1174 1214 1254
Journal of International Business Studies

LR chi-square 181.68*** 253.55*** 119.12*** 149.14*** 163.5***

Note: Parent firm and manufacturing industry dummies are not included, since several parent firms and industries in the sample had only one subsidiary in crisis-stricken countries in a given year;
those parent and industry dummies would have been perfectly correlated with subsidiary continuation/termination.
*po0.10; **po0.05; ***po0.01, all from two-tailed tests.

515
 Subsidiary expansion/contraction at times of economic crisis Chris C Chung et al
516

ABOUT THE AUTHORS Paul W Beamish holds the Canada Research Chair
Chris Changwha Chung is an assistant professor of in International Business at the Ivey Business
International Business & Strategy at the Korea School, University of Western Ontario. At Ivey, he
University Business School, Korea University. He serves as director for the Engaging Emerging
received his doctorate in international business and Markets Research Centre. He is a past Editor-in-
strategic management from the Ivey Business School, Chief of JIBS, and a Fellow of the AIB. E-mail:
Canada. Born in Korea, he is a Canadian citizen. His [email protected].
research interests include real options, international
joint venture evolution, and foreign subsidiary Takehiko Isobe is a professor of Graduate School
management. E-mail: [email protected]. of Business Administration at Keio University
of Japan. His research interests include the rela-
Seung-Hyun Lee is an associate professor of tionship between cross-country variations in
Strategy & International Business at the University institutions and behaviors and performance of
of Texas at Dallas. He received his PhD in interna- multinational enterprises. He has previously pub-
tional business and strategic management at the lished in journals such as Academy of Management
Ohio State University. His research interests include Journal, Journal of International Business Studies, and
real options theory and institutional theory. He is a Strategic Management Journal. E-mail: [email protected]
Korean citizen and US permanent resident. E-mail: .ac.jp
[email protected].

Accepted by Sea-Jin Chang, Area Editor, 18 June 2009. This paper has been with the authors for four revisions.

Journal of International Business Studies