Journal of World Business: Sergey Morgulis-Yakushev, H. Emre Yildiz, Carl. F. Fey
Journal of World Business: Sergey Morgulis-Yakushev, H. Emre Yildiz, Carl. F. Fey
Journal of World Business: Sergey Morgulis-Yakushev, H. Emre Yildiz, Carl. F. Fey
A R T I C L E I N F O A B S T R A C T
Keywords: Multinationals (MNCs) need to find the balance between developing a globally standardized organizational
Cultural distance culture and having multiple locally- adapted organizational cultures. Past literature embodies the bias that
Cultural differences and similarities differences between MNC units, unless managed, would lead to adverse consequences. To counter this negative
Cultural fit bias, we focus on cultural fit, which is the amount of difference yielding maximum benefit. We argue that
Knowledge transfer
depending on comparison criterion and desired outcome, fit could be achieved by establishing similarities or
Organizational culture
maintaining differences. Using evolutionary economics, we explore knowledge transfer within MNCs and test
our hypotheses on fit using a unique dyadic dataset from 186 MNCs.
☆
The authors would like to thank the Wallander foundation for financial support. The foundation was not involved in the design or conducting the study and only provided financial
support.
⁎
Corresponding author.
E-mail addresses: Sergey.Morgulis-Yakushev@hhs.se (S. Morgulis-Yakushev), Emre.Yildiz@fek.uu.se (H.E. Yildiz).
http://dx.doi.org/10.1016/j.jwb.2017.09.002
Received 21 October 2016; Received in revised form 14 September 2017; Accepted 15 September 2017
1090-9516/ © 2017 Elsevier Inc. All rights reserved.
Please cite this article as: Morgulis-Yakushev, S., Journal of World Business (2017), http://dx.doi.org/10.1016/j.jwb.2017.09.002
S. Morgulis-Yakushev et al. Journal of World Business xxx (xxxx) xxx–xxx
singular index of distance. Using these unified measures, existing re- successful organizational knowledge transfer across different subunits
search often positions two entities vis-à-vis each other, and tests whe- (Fey & Furu, 2008; Kostova & Roth, 2002; Kostova, 1999; Minbaeva,
ther the cultural similarity between these two entities is able to explain Pedersen, Björkman, Fey, & Park, 2003; Szulanski, 1996). While some
success of the firm. However, such an approach cannot completely studies define the success of the transfer process as the extent to which
account for the distinct effect(s) of different cultural dimensions on a the knowledge is replicated by the recipient unit (Nelson & Winter,
given outcome. Thus, despite its methodological convenience, the 1982; Szulanski, 1996), some others extend the scope of transfer suc-
unification of culture/cultural distance under a singular index measure cess by measuring employees’ ownership of, commitment to, and sa-
hides important sources of variation stemming from specific cultural tisfaction with the transferred knowledge (Kostova & Roth, 2002;
dimensions and attributes. We will address this problem by keeping Kostova, 1999; Lervik, 2005). Thus, studies of knowledge flows look
cultural dimensions separate in our analysis, so that we can identify mainly at the amount of knowledge transferred into and/or from sub-
those aspects of organizational culture for which achieving fit requires sidiaries, whereas the studies on transfer success explore the extent to
attaining similarities or maintaining differences. which the incoming knowledge is implemented by the recipient units.
Second, as a result of using index-based measures to calculate cul- In this paper, we regard knowledge transfer as a two-stage process
tural distance, past studies can account only for the degree of difference consisting of both knowledge inflows and knowledge implementation.
without explaining the direction with which these differences affect We also look at both directions of knowledge flow—knowledge inflow
cross-cultural phenomena. Aggregating culture to a unified index and knowledge outflow. In line with our general research design, we
means that cultural distance must be an absolute measure that always take the subsidiary as the main/focal unit of analysis. Accordingly, we
has positive values. As a result, it has to be assumed that the effects of define the degree of outward knowledge flows as the extent to which a
cultural differences are symmetrical, but there is no reason to assume subsidiary sends knowledge to the HQ, whereas the degree of inward
that the effect of differences in organizational culture between two knowledge flows is defined as the extent to which the subsidiary receives
units has symmetrical effects on these units. Our approach in this paper knowledge from its HQ. Thus, outward and inward knowledge flows
relaxes this assumption, which helps us probe into both degree and capture the frequency of bidirectional communication between the
direction of differences. subsidiary and the HQ. Knowledge implementation goes beyond mere
In our attempt to contribute to a more balanced and nuanced un- exposure to new knowledge and refers to the degree to which sub-
derstanding of cultural differences (Stahl et al., 2016), we examine two sidiary employees learn from the HQ and put HQ’s knowledge into
well-established dimensions of organizational culture (i.e. internal in- practice.
tegration and external adaptation), two distinct components of the
knowledge transfer process (i.e. knowledge flows and knowledge im- 2.2. Organizational culture
plementation), and two directions of knowledge flow (i.e., inward
knowledge flow and outward knowledge flow). Theoretically, we build Before proceeding further, it is important to clarify the level at
our reasoning on evolutionary theories of organizational learning which we examine culture. An MNC is normally comprised of a HQ and
(Aldrich & Reuf, 2006; Zollo & Winter, 2002), and argue that for a number of subsidiaries in different countries. The organizational
knowledge to flow between two units different degrees of variation are culture of an MNC subunit might be influenced by the national culture
required. Accordingly, we argue that the direction of these knowledge within which it operates (Hofstede, Neuijen, Ohayv, & Sanders, 1990)
flows depends on which of the two units has a higher level of external because of the need to adapt the organizational culture to the local
adaptation in its organizational culture. We show that knowledge flows environment. Previous research shows that national culture affects the
from the unit with stronger external adaptation (generating higher type of organizational culture that is more effective in a particular
variation) into the unit with weaker external adaptation (generating country (Fey & Denison, 2003). Therefore, national culture to some
more limited variation). Further, we propose that the implementation extent is diffused into the firm’s organizational culture. Diverse national
of knowledge necessitates similar selection mechanisms, and therefore, cultures create pluralities and different organizational sub-cultures
benefits from higher similarity in terms of internal integration. In sum, within MNCs. As a result, there is a need for globally standardizing a
we hypothesize that fit between two interacting units can be achieved company’s organizational in order to create consistency and coherence
by establishing similarities or maintaining differences, depending (a) on within the entire organization. In this regard, the MNC’s home country
the criterion used to compare the two units (i.e., which dimension of culture also influences its (meta) organizational culture, which partly
organizational culture), and (b) the intended outcomes of the interac- gives rise to an inherent tension between standardization and adapta-
tion (i.e., direction and component of knowledge transfer process). tion within an MNC. Balancing this tension is a key challenge for MNC
The paper is organized as follows. In the next section, we define our management and parallels the need to balance the MNCs’ strategic
key terms and concepts. We then develop our hypotheses drawing on and organizational tensions as discussed in the classic work of
evolutionary economics. Next, we explain our general research design, Bartlett & Ghoshal (1988b).
data collection procedure and methodological details. We will then Although some studies consider the role of organizational cultural
present our empirical analyses and results, and conclude by positioning differences (e.g. Fey & Beamish, 2001; Pothukuchi, Damanpour, Choi,
them against previous research. Chen, & Park, 2002; Weber, Shenkar, & Raveh, 1996), extant IB litera-
ture focuses primarily on national cultural differences. As such, despite
2. Key terms and definitions our focus on organizational cultural differences in this paper, we draw
on work focusing on culture at the national level. Further, while it is
2.1. Knowledge transfer beyond the immediate scope of this paper, our balanced interest in both
positive and negative aspects of cultural similarities/differences is re-
The transfer of knowledge-based resources across different subunits levant to studies exploring the role of national culture differences and
has been a popular topic for IB studies. There is a stream of research would be a useful topic for future empirical study.
that looks mainly at subsidiaries and examines the direction (e.g. in- We adopt Schein’s (1992, p. 12) popular definition of organizational
ward vs. outward, vertical vs. horizontal) and amount of the knowledge culture as the “pattern of shared basic assumptions that the group learned as
flows within MNCs (Ghoshal & Bartlett, 1988; Gupta & Govindarajan, it solved its problems of external adaptation and internal integration.” We
1991, 2000; Harzing & Noorderhaven, 2006; Minbaeva, 2007; decided to adopt this definition, since (1) Schein’s work is widely used (it
Monteiro, Arvidsson, & Birkinshaw, 2008). Focusing specifically on the has over 35,000 citations); (2) many of the major models of organizational
effectiveness of knowledge transfer into subsidiaries, there is another culture are built upon the tension between external adaption and internal
stream of work which is concerned primarily with the drivers of integration (e.g., Denison & Mishra, 1995; O'Reilly, Chatman, & Caldwell,
2
S. Morgulis-Yakushev et al. Journal of World Business xxx (xxxx) xxx–xxx
1991; Quinn & Rohrbaugh, 1983); and (3) this definition of organizational environment and increase their chances of survival (Burgelman, 1990).
culture is consistent with evolutionary theory which we use to develop our An important advantage of applying an evolutionary perspective in
hypotheses. In this paper we focus on those practices and routines that re- organization studies is that it enables the development of a set of clear,
flect MNC units’ degree of external adaptation and internal integration. theoretically founded and testable predictions (Dosi & Marengo, 2007).
External adaptation is the degree to which an organization is re- The mechanisms of variation, selection and retention lay at the heart of
sponsive to its external environment and the changing conditions, de- this perspective, and are used to explain how organizations engage with
mands and pressures it entails. Thus, organizations with adaptable their environment and identify, develop, use and replicate routines and
cultures are able to sense and seize opportunities in their external en- practices as a result of this engagement (Nelson & Winter, 2002). In-
vironment, are willing to take risks associated with constant learning tegrating these mechanisms within the framework of knowledge
and improvement, and have sets of values that encourage and support transfer in a MNC context is useful for theory development for several
change (Calori & Sarnin, 1991). In other words, external adaptation reasons.
entails normative support for organizational members to take risks re- First, variation, selection and retention mechanisms are likely to
lated to learning from and adapting to their environment significantly differ across diverse organizational environments.
(Denison & Mishra, 1995). It stands to reason that external adaptation is Therefore, HQs and subsidiaries are likely to face different forces and
conducive to engaging in ‘institutionalized experimentation’, which opportunities emerging from these mechanisms due to operating in
endorses playfulness, inquisitiveness and curiosity. External adaptation different environments. The evolutionary perspective can provide a
is especially important in dynamic and fast changing contexts where the parsimonious explanation for the complex processes through which
ability to monitor, adapt to, and learn from the environment is a pre- these organizations generate varying degrees/types of routines and
requisite to avoid the risks of obsolescence and loss of competitive edge capabilities, which they then can exchange with each other. Second,
(Wu, 2010). there are inherently asymmetric dependence relationships within
While external adaptation determines the types of action and MNCs, which are complex given their dual dependence on their local/
practices needed for the organization to survive in its environment, external and MNC-wide organizational/internal environments (Ambos,
internal integration defines the courses of action that could facilitate Andersson, & Birkinshaw, 2010). As a result, understanding the me-
coordinated interaction among different parts/units within an organi- chanisms of variation, selection and retention might explain how or-
zation. Consequently, organizations with strong internal integration are ganizational culture of the HQ and subsidiaries can shape the way this
able to maintain internal coherence and reach agreement whenever dual dependence is managed via different types and degrees of resource
differences of opinion arise (Denison, 1990). When basic assumptions exchange. The evolutionary perspective can shed light on the condi-
and core values are widely shared within an organization, interaction tions under which subsidiaries are able to learn from their immediate
between different groups of individuals within an organization becomes environment (i.e. host country) or are dependent mostly on the
more efficient, productive, and smooth (Meglino & Ravlin, 1998). As knowledge provided by their corporate HQ. In sum, by considering
noted by Denison, Haaland, and Goelzer (2004:p. 100), “this type of organizational cultural dimensions as a manifestation of the MNC units’
consistency is a powerful source of stability and internal integration responses to evolutionary mechanisms, we can develop a set of theo-
that results from a common mindset and a high degree of conformity.” retical predications about how (1) exploratory learning in subsidiaries
Similarly, Tsui et al. (2006) argues that internal integration explains the might exceed (fall short of) that in the HQ and might result in outward
extent to which organizations are able to establish a common vision and (inward) knowledge flows, and (2) the relative characteristics of the HQ
shared values among members and/or subunits, which eventually will and the subsidiary’s selection mechanisms jointly determine the like-
foster strong identification with and commitment to the organization. lihood of knowledge transfer success (i.e. incoming knowledge being
It is important to note that external adaptation and internal in- implemented by the subsidiary). Thus, the evolutionary perspective is
tegration can often be at odds creating a tension that needs to be ba- useful and well suited to theorize about the knowledge transfer pro-
lanced. Thus, the values and assumptions required for external adap- cesses.
tation might be different from those required for maximizing internal There are three evolutionary processes (i.e., variation, selection, and
integration and consistency. This inherent tradeoff has important im- retention), which can explain why and how organizations need new
plications for understanding the role of organizational culture in inter- routines to adapt better to their environment (Ziman, 2000). Variation
organizational knowledge transfer, and underscores the importance of refers to changes in organizational forms and existing routines and
separate examination of different cultural dimensions. Since external competencies. Variation can be intentional (i.e., when individuals or
adaptation and internal integration are distinct dimensions that gen- organizations purposefully act to find new alternatives to problems), or
erate different outcomes (i.e. flexibility/adaptability vs. consistency/ blind (i.e., when change occurs in the absence of formal plans or de-
uniformity), it would be erroneous to cluster these two dimensions liberate actions). The second evolutionary process is selection, which
under the single construct of ‘organizational culture’. In the next sec- refers to selective elimination of certain kinds of variation. Selection
tion, we develop separate hypotheses concerning the role of external can be based on factors external (e.g., market conditions, competitive
adaptation and internal integration in facilitating different knowledge pressures, institutionalized norms, etc.) or internal (e.g., administrative
transfer outcomes. heritage, dominant logics, corporate culture, etc.) to the organization.
Lastly, retention is the process through which selected variations are
3. Theory and hypothesis development preserved, replicated and reproduced in order to benefit from new al-
ternatives over time (Aldrich & Reuf, 2006).
In this paper, we adopt the evolutionary perspective on organiza- Our model focuses in particular on variation and selection me-
tions to guide our theoretical reasoning (Nelson & Winter, 1982; chanisms.1 The starting point of our theoretical model is that firms need
Zollo & Winter,2002). With intellectual roots in biology, evolutionary to upgrade their practices in order to deal more effectively with con-
theory was developed to explain the necessary and sufficient conditions stant changes to the technological and competitive conditions and to
under which incremental changes occur in systems with scarce re- avoid the risks of obsolescence and loss of competitive advantage. In
sources (Campbell, 1969). The evolutionary perspective provides a
useful analytical tool to examine the mechanisms with which organi-
zations search and process information, and the selection rules they 1
Due to the cross-sectional nature of our empirical design, retention is beyond the
employ when adopting practices from their environment (Dosi, 1982). scope of the present investigation. However, an integrated examination of all three
At the most fundamental level, the evolutionary perspective argues that evolutionary mechanisms would be a fruitful direction for future research using long-
organizations need to evolve in order to better adapt to (fit with) their itudinal designs.
3
S. Morgulis-Yakushev et al. Journal of World Business xxx (xxxx) xxx–xxx
this context, organizations with different emphasis on external adap- H1a. For any given level of HQE, there is an inverted U-shaped relationship
tation and internal integration react to internal or external stimuli between DE and inward knowledge flows into the subsidiary from the HQ.
differently. Firms with a stronger emphasis on external adaptation will
H1b. For any given level of HQE, the highest degree of inward knowledge
be more likely to pay attention to changing external stimuli (e.g.
flow into the subsidiary will be achieved when DE is positive (i.e.,
competitors’ behavior, changes to the institutional landscape, emerging
SE < HQE).
technological trends, etc.). This, in turn, will contribute to the gen-
eration of higher variation. In this generative variation stage of the Our first set of hypotheses are concerned primarily with inward
knowledge evolution cycle, those organizations that emphasize external flows of knowledge from the HQ to the subsidiary. An analogous – yet
adaptation are likely to generate more diverse ideas to handle existing reverse – line of reasoning can also be applied to develop our hy-
and/or new problems, and use values/beliefs related to internal in- potheses about outward knowledge flows from the subsidiary to the
tegration when selecting among this increased variation. HQ. Accordingly, we expect higher levels of outward knowledge flows
Our model is bilateral in the sense that it specifies how the relative (i.e., knowledge flow from subsidiary to HQ) when the level of SE
characteristics of the HQ’s and subsidiary’s (S) organizational cultural generates higher level of variation at the subsidiary and HQE is high
dimensions affect patterns of knowledge transfer between these two enough to induce a sufficient level of HQ willingness to learn from its
entities. Accordingly, for external adaption, we focus on the HQ’s score subsidiary. Unlike inward knowledge flows where optimum fit is
for external adaptation (HQE), the subsidiary’s score for external achieved for moderately positive values of DE, the highest level of
adaptation (SE) and the difference (DE) between HQE and SE, where outward knowledge flows is observed when the subsidiary has stronger
DE = 0 when SE = HQE, DE (i.e., HQE-SE) is positive when HQE > SE external adaptation focus in its culture compared to HQ (i.e., when
and DE is negative when HQE < SE. As noted in the previous section, SE > HQE, hence DE is negative and skewed to left with respect to the
external adaptation determines the degree to which organizational SE = HQE line). In other words, we expect the optimum point for
culture encourages and supports responsiveness to changing conditions, knowledge outflows (i.e., the maximum point on the inverted-U curve)
demands and pressures external to the organization. Thus, we expect to will be achieved on the left hand side of the perfect similarity line.
see a greater degree of variation in units where the external adaption Taken together with our preceding hypotheses, we claim that different
dimension of organizational culture is strong. This argument is corro- types of knowledge flows will be optimized at different signs of distance. In
borated by earlier empirical studies that report positive effects of other words, whether or not cultural differences generate cultural fit
market orientation (e.g., Hurley & Hult, 1998), learning orientation depends on the specific outcome intended within the cross-cultural
(e.g., Calantone, Cavusgil, & Zhao, 2002; Hult, Hurley, & Knight, 2004) interaction. Accordingly, we hypothesize that:
and openness to and receptivity to new ideas (Laursen & Salter, 2006)
H2a. For any given level of HQE, there is an inverted U-shaped relationship
on the innovativeness of the firm.
between DE and outward knowledge flows from subsidiary into HQ.
Based on the above reasoning, we argue that inward knowledge
flows to the subsidiary will be limited if DE attains negative values (e.g., H2b. For any given level of HQE, the highest degree of outward knowledge
when SE is high compared to HQE). This is because, negative levels of flow from the subsidiary will be achieved when DE is negative (i.e.,
DE mean that the degree of external adaptation is stronger in the sub- SE > HQE).
sidiary compared to the HQ, which increases the subsidiary’s ability to
Whereas external adaptation affects the degree of variation to which
generate the necessary variation on its own and reduces its dependence
the organization is exposed, internal integration determines the selec-
on the HQ for new ideas and solutions. On the other hand, when DE
tion processes by determining what will be adopted from the pool of
attains highly positive values (i.e., when SE is very low compared to
alternative ideas and solutions (Dosi & Marengo, 1994). Put differently,
HQE), inward knowledge flows from the HQ to the subsidiary will also
external adaption fosters increased variation and helps organizations to
be limited. This is because; a subsidiary with low levels of external
expand their opportunities for learning while internal integration
adaptation would not have the norms and values that are conducive to
functions as a ‘filter’ which determines what should be considered
learning from external sources including the HQ (Tsui, Wang, & Xin,
technically and normatively appropriate, acceptable and useable from
2006). This means that higher levels of inward knowledge flows can be
the new external knowledge made available by other units
achieved when (a) HQE is sufficiently high so that HQ has a strong
(Aldrich & Reuf, 2006). If two units do not use the same selection me-
ability to generate higher variation and become a source of novel
chanisms, one (i.e., recipient) cannot benefit effectively from the
knowledge, and (b) SE is above a certain threshold so that the subsidiary
knowledge stock of the other (i.e., sender). In the case of dyadic
has the need and the ability to engage in external knowledge search.
knowledge transfer between HQ and subsidiary, this reasoning suggests
Thus, we expect knowledge inflows to be highest when the HQ and
that effective inter-organizational learning will be higher when both
subsidiary are moderately different from each other in terms of the
sender and recipient units use similar rules, procedures, policies, logics
external adaptation dimension of their organizational culture.
and schemas (Argote, McEvily, & Reagans., 2003). This is predicated
In sum, we postulate that the extent of inward knowledge flows
upon the idea that inter-organizational similarities in terms of internal
would be minimum at extremely low (i.e., subsidiary is self-sufficient
working principles create relevant and overlapping assumptions and
and has little or no need for HQ’s knowledge) and high (i.e., subsidiary
values that guide action and interpretation in both sending and re-
is inert and has a weak culture to support learning from HQ) values of
cipient units. This is consistent with earlier findings that a high level of
DE. Further, we expect inward knowledge flows to be highest when the
overlap between source and recipient units’ knowledge bases is con-
sending unit (i.e., HQ) has a slightly higher level of external adaptation,
ducive to relative absorptive capacity and mutual learning
and therefore is able to generate a higher degree of variation compared
(Lane & Lubatkin, 1998).
to the recipient (i.e., subsidiary). The recipient unit, on the other hand,
In line with the above intuition, we argue that the implementation
should have a certain degree of external adaptation such that it has both
of knowledge in the subsidiary will be higher for low degrees of dif-
the need for and interest in learning from the sending unit. This sug-
ference (DI = HQI − SI) between the internal integration scores of the
gests that an optimum level of knowledge inflows will happen when DE
subsidiary (SI) and the HQ (HQI). That is, when SI is very different from
takes on a moderate non-zero (positive) value. In other words, we ex-
HQI, we will observe either a high level of rigidity and inertia (as in the
pect that the point at which knowledge inflows into a given subsidiary
case of strong SI and highly negative DI) or lack of clear guidelines and
are maximized when DE is moderately skewed to the right of perfect
internal rules (as in the case of weak SI and highly positive DI). Hence,
similarity (DE = 0) line.
we argue that subsidiaries will more likely implement the knowledge
Based on the above, we hypothesize that:
transferred from the HQ if all parties have similar levels of internal
4
S. Morgulis-Yakushev et al. Journal of World Business xxx (xxxx) xxx–xxx
integration. Furthermore, we expect a maximum level of knowledge from each subsidiary (a member of the executive team and a subsidiary
implementation if subsidiary and HQ use the same filters and action employee). This resulted in four sets of responses for each HQ-subsidiary
logics. In other words, an optimum degree of knowledge implementa- pair. This multi-respondent strategy enabled us to avoid, ex ante, potential
tion will be achieved if SI and HQI are identical. Put differently, unlike biases resulting from common rater effects (Podsakoff, MacKenzie,
outward and inward knowledge flows where the best fit requires a Lee, & Podsakoff, 2003). In addition, we took a number of post-hoc statis-
certain level and type of differences between sending and recipient tical measures, which we explain before presenting our findings.
units, knowledge implementation will be maximized if the two entities In this study the level of measurement (individual) differed from the
have the same levels of internal integration. Thus, we propose the fol- level of analysis (firm) for the organizational culture dimensions
lowing hypotheses: (Rousseau, 1985). Since we are interested in the firm level measure-
ments of organizational culture, data collected from multiple re-
H3a. For any given level of HQI, there is an inverted U-shaped relationship
spondents were aggregated to the firm level. Thus, each firm’s scores
between DI and the extent of the subsidiary’s implementation of knowledge
were based on the mean of the responses provided by two raters within
coming from the HQ.
the firm. Three complementary indicators, RWG, ICC(1), and ICC(2) are
H3b. For any given level of HQI, the highest level of knowledge usually used to assess the appropriateness of the aggregation of in-
implementation in the subsidiary will be achieved when DI converges to dividual level measures to the firm level (Bliese, 2000; Seibert,
zero (i.e., HQI = SI). Silver, & Randolph, 2004; Vogus & Sutcliffe, 2007; Zohar & Luria,
2004). Within-group inter-rater agreement (RWG) measures the degree
Our hypotheses concerning the curvilinear effects of DE and DI and
to which individual responses within a group are interchangeable. The
the different conditions under which optimum outcomes for knowledge
RWG values for the organizational culture dimensions reached the
flows (inward and outward) and knowledge implementation are de-
conventionally acceptable level of 0.70 (George, 1990), suggesting
picted in Fig. 1.
good within-group agreement and justifying the aggregation of in-
dividual responses to firm level. For further justification of the ag-
4. Methodology gregation, ICC(1) statistics (i.e., inter-rater reliability index), and IC-
C(2) values (i.e., t reliability of group means) were used (Chen & Bliese,
4.1. Sample 2002; Liao & Rupp, 2005). For all variables, F values for ICC(1) statistics
were significant, and the ICC(2) statistics were higher than 0.75, which
To construct our sample, we used the Bureau van Dijk research meets the usual suggested guidelines (Klein et al., 2000). These results
database to obtain lists of firms in Belgium, Denmark, Finland, confirm that multiple measures for each subsidiary and HQ could be
Germany, Netherlands, Norway, Sweden, Switzerland, and the United meaningfully aggregated to the corresponding firm level.
Kingdom with wholly owned subsidiaries (WOS) in Russia. We ex-
cluded very young (i.e. < 4 years) firms and very small (i.e. < 20 em- 4.3. Measures
ployees) firms in order to obtain a meaningful unit of analysis.2 Finally,
for each home country, 90 firms were randomly selected from among 4.3.1. Knowledge flow and knowledge implementation
the remaining cases. Our measures for outward and inward knowledge flows were adapted
We used two different sets of questionnaires separately addressed to from Gupta and Govindarajan (2000). We asked the subsidiary re-
HQ and subsidiaries (see below for further details). The questionnaires spondents to indicate the degree to which they had sent and received
were administered in January–March 2012 in the form of an electronic knowledge to/from the parent company HQ over the preceding three
survey sent to the parent firms’ offices. Two reminders were sent to non- years related to the following items: management systems and prac-
responders at two-week intervals in order to increase the response rate. tices, manufacturing know-how, marketing know-how, distribution
Due to difficulties related to data collection and achieving a sufficient know-how, and R & D know-how. For each of these items, subsidiary
response rate from Russian companies, we made follow-up phone calls respondents were asked to evaluate on a seven-point scale (1 = not at
to ask subsidiary managers to complete the questionnaires. In total, we all, 7 = to a great extent) the degree to which the subsidiary was in-
received 372 completed questionnaires (186 responses from HQ and volved in outward and inward knowledge transfer flows with the HQ.
186 responses from subsidiaries) a response rate of 23% (ranging from Cronbach alphas for our outward and inward knowledge transfer
16% in the Netherlands to 38% in Finland). On average, the sub- measures were 0.83 and 0.79 respectively.
sidiaries in our final sample are 7 years old, employ 430 people, and The measure for knowledge implementation was adopted from Lyles
operate in 7 countries. The number of firms from each home country, and Salk (1996). Using a seven-point scale (1 = not at all to 7 = to a
the number of firms fulfilling sample selection criteria, and the eventual great extent), subsidiary respondents were asked to rate to what extent
number of sampled firms are shown in Table 1. they had learned from the parent corporation over the previous three
years, in the following dimensions: management systems and practices,
4.2. Questionnaires manufacturing know-how, marketing know-how, distribution know-
how, and R & D know-how. Cronbach’s alpha for our knowledge im-
Unlike the majority of knowledge transfer studies in the IB literature plementation construct was 0.81.
(see Michailova & Mustaffa, 2012; Minbaeva, 2007), we collected data from
both parent company HQ and subsidiary via two questionnaires: one filled 4.3.2. Organizational culture
by representatives of HQ (Q1), and one filled by representatives of the Organizational culture was measured by focusing on two main di-
Russian subsidiaries (Q2). Q1 was used to collect data on the organizational mensions: internal integration and external adaptation. This measure is
culture of HQs while items in Q2 covered the subsidiary’s organizational guided theoretically by the definition of organizational culture as a
culture, knowledge transfer outcomes (inward and outward flows, knowl- system of shared assumptions developed in the organization over time
edge implementation), and control variables. We collected data from two to solve problems of internal integration and external adaptation
informants in each HQ (two employee representatives) and two informants (Schein, 1985). Thus, our basic assumption here is that, as a latent
construct, organizational culture entails largely tacit assumptions and
2
beliefs which can be observed/measured only by looking at how its
We assume that age is an important selection criterion since subsidiaries need to be
established long enough to experience different knowledge transfer episodes. Similarly,
different dimensions are reflected in the use of specific organizational
we assume that small subsidiaries are too peripheral to be targets or sources of knowledge policies and routines (Ravasi & Schultz, 2006). Below, we discuss how
transfer. we decided how to measure organizational culture.
5
S. Morgulis-Yakushev et al. Journal of World Business xxx (xxxx) xxx–xxx
6
S. Morgulis-Yakushev et al. Journal of World Business xxx (xxxx) xxx–xxx
Table 1 dimension (internal and external) in our model, which was helpful to
Sample characteristics. keep our questionnaire at a manageable length. For the internal in-
tegration dimension, we examined team orientation, employee devel-
Location of Number or Meeting Randomly Responded Response rate
parent firm firms in the selected opment, coordination, rules orientation, and centralization dimensions
population criteriaa (Alpha values for HQ and subsidiary were 0.74 and 0.79, respectively).
For external adaptation we considered market orientation, organiza-
Belgium 230 96 90 16 18%
tional learning, goal orientation, future orientation, and change or-
Denmark 387 128 90 18 20%
Finland 1193 471 90 34 38% ientation dimensions (Alpha values were 0.79 for HQ and 0.87 for
Germany 2089 749 90 27 30% subsidiary).
Netherlands 1277 511 90 14 16% To verify the two-factor structure of organizational culture, we used
Norway 219 87 87 16 18% exploratory factor analysis (EFA) to the 10 organizational culture items.
Sweden 583 214 90 17 19%
The results yielded a two-factor solution (only two factors with eigen-
Switzerland 866 679 90 17 19%
United 1892 1398 90 27 30% values greater than one) where Factor 1 (Internal Integration) explains
Kingdom 38.2% of all variation, Factor 2 (External Adaptation) explains 31.6% of
Total 807 188 23% variation. Taken together, these results suggest that the indicators ac-
a
count for a large proportion of the variance (69.8%) in the hypothesized
The criteria are that the subsidiary should be at least four years old and have a: at
construct, and provide strong support for the validity of the measure.
least 20 employees and older than 4 years.
We found that the factor loadings were all above 0.4. Correlations be-
tween our two factors ranged from 0.01 to 0.21, which indicates the
absence of, cross-correlation between factors. Results for the EFA are
The macro dimensions of organizational culture, external adapta- summarized in Table 2. We also conducted a confirmatory factor ana-
tion and internal integration are reflected in most of the well-known lysis, which confirmed that a three or four factor solution was not
models of organizational culture (e.g. Denison & Mishra, 1995; O'Reilly significantly better than a two-factor solution.
et al., 1991; Quinn & Rohrbaugh, 1983). In this study we use dimen- We conducted a non-parametric analysis to test whether sub-sam-
sions/items from the widely used Denison organizational culture survey ples from different MNC home countries originated from the same
(Denison, Haaland, & Goelzer, 2003) to measure organizational culture. distribution. We conducted this test to check whether there is a statis-
The Denison organizational culture model is comprised of 12 dimen- tical need to include the HQ’s national culture as a control variable in
sions, six related to external elements and the other six related to in- the subsequent analyses. A series of Kruskal–Wallis tests with multiple
ternal elements. In addition to grouping the dimensions according to comparisons (Siegel & Castellan, 1988) were performed to determine
internal vs. external focus, the Denison model also differentiates be- whether there were differences in the MNC home country means for the
tween dimensions related to flexibility or stability. This yields a 2 × 2 Internal Integration and External Adaptation indices across the nine
matrix with four quadrants, each related to three sub-dimensions. subsamples representing different HQ locations. Using MNC home
Specifically, the Denison model first divides all 12 sub-dimensions into country as the grouping variable, the Kruskal–Wallis tests showed that
two groups – internal vs. external focus of organizational culture (6 there were no significant differences in the subgroup means of our or-
dimensions each). Next, the sub-dimensions in each of these two groups ganizational cultural dimensions (see Table 3). This shows that the
are further subdivided into a further two sub-groups – flexible vs. organizational culture measures in our data already incorporate var-
stable. However, in our study, for simplicity we do not use the flexible iation originating from home country-level variations. Thus, the hy-
vs. stable focus, the rationale for which will be explained in more detail pothesized relationships among the variables are free from any sys-
in the following three paragraphs. Thus, we performed only the first tematic variation due to different home country HQ locations.
step in the model, dividing the dimensions into two groups—internal Therefore, we decided not to include home country dummies in our
vs. external focus. Considering that Schein’s (1992) definition of orga- regression models to control for national culture.
nizational culture makes explicit reference to these two dimensions,
and since this two-dimensional model is strongly supported by our 4.3.3. Controls
factor analysis (our exploratory factor analysis of the organizational Since host country experience may affect knowledge transfer out-
culture items in the Russian subsidiaries produced only two factors with comes (Birkinshaw, Nobel, & Ridderstråle 2002; Hansen, 2002), we
eigenvalues greater than one), we deduce that omitting the second step controlled for the number of years each MNC has been operating in
of Denison model is not an over-simplification.
We conducted a pilot study to determine how to best measure or- Table 2
ganizational culture in the Russian context since one side of the mea- Exploratory factor analysis results for items of organizational culture.
surement in this study is in Russia (Morgulis-Yakushev, 2015). The pilot
study consisted of asking 200 managers from different firms in Russia to Items Factor 1 Factor 2
list the five words which best describe their organizational culture. The Team Orientation 0.89 −0.03
data was then broken into two sub-groups of 100 firms and the words Empowerment 0.82 0.18
suggested by the respondents were sorted into groups of related words Employee Development 0.93 0.07
by two research assistants resulting in five groups of similar words re- Agreement 0.78 0.21
Coordination 0.77 −0.04
lated to internal and five groups related to external. The results were
Customer Orientation −0.04 0.92
similar for the two sub-samples. The 10 resulting groups closely related Learning Orientation 0.03 0.87
to 10 of the 12 sub-dimensions of organizational culture suggested by Openness to Change −0.13 0.74
Denison—five for external and five for internal. Thus, we decided to use Goal Orientation 0.12 0.76
already tested items/sub-dimensions from the Denison Organizational Vision Orientation −0.01 0.71
Culture Survey. We only used the above-mentioned 10 sub-dimensions Eigenvalue 6.34 4.92
in our study (instead of all 12 sub-dimensions from the Denison model) Percent variance explained 38.2 31.6
Cumulative percent variance explained 38.2 69.8
since it is difficult to measure something that does not have any cor-
respondence in a country. Note: Extraction method: Principal Component Analysis. Rotation method: Varimax.
One seven-point likert-type indicator represented each sub-dimen- a
Marked loadings are higher than 0.40.
sion of organizational culture. Thus, we had five indicators per Bold value denotes the loading values that are higher than 0.40.
7
S. Morgulis-Yakushev et al. Journal of World Business xxx (xxxx) xxx–xxx
Table 3 tests for Q2, our first model showed good fit with the data (RMSEA 0.071,
Kruskal–Wallis tests using MNC home country as the grouping variable. SRMR 0.053, CFI 0.956). Compared to this, the second model with single
CMF showed a poorer fit (RMSEA 0.148, SRMR 0.177, CFI 0.902). Although
Variable Statistics p-value Difference
our third model for Q2 also fitted well with data (RMSEA 0.068, SRMR
Internal Integration in HQ 7.275 0.507 no 0.051, CFI 0.955), the improvement again was not significant (p = 0.21).
External Adaptation in HQ 6.262 0.618 no Collectively, these results confirm that neither group of respondents used a
common method when responding to the questionnaires. Therefore, we can
Notes: N = 186.
safely to conclude that our data are not confounded by common method bias.
Past studies (e.g., Fey & Beamish, 2001; Pothukuchi et al., 2002) have
Russia. Furthermore, we controlled for MNCs’ international experience
used unitary/aggregate scores for organizational culture difference. Instead of
by looking at the number of the MNC’s subsidiaries in different coun-
running our estimations based on such conventional index-based measures, in
tries. Since larger firms may have more slack resources for knowledge
this paper we used appropriate statistical techniques to identify how specific
creation and transfer, we controlled also for subsidiary size using the
organizational culture dimensions of HQ and subsidiaries fit together and
natural log of the number of employees in the subsidiary. The percen-
optimize knowledge transfer between these two units. To be more specific,
tage of a subsidiary’s activities in manufacturing was included as an
without reducing the difference between two entities (e.g., S and HQ) to a
industry control.
single value (distance between S and HQ, which is D), our technique derives
Table 4 provides means, standard deviations, correlations, and in-
the effects of differences by keeping cultural scores of these two entities as
ternal consistency coefficients (Cronbach's alpha) of the measures used
separate parameters while making the estimations. This way, we were able to
in our models. The size of the correlation coefficients suggests that
derive our estimations and test our hypotheses without any loss of di-
multicollinearity was not a problem.
mensionality. For this purpose, we first ran traditional regression with two
independent variables, their interactions and second order variants (X, Y, X2,
5. Findings Y2, XY). This was complemented by additional transformations that allowed
us to test hypotheses not regarding the variables from the equation (X and Y),
5.1. Analysis but regarding the difference between these two variables (D = X − Y). Si-
milar techniques have been used in past exploratory studies (e.g., Cafri, van
Although using multiple respondents was an effective way to alleviate den Berg, & Brannick, 2010; Harris, Anseel, & Lievens, 2008; Shanock, Baran,
potential problems associated with common rater effects, we also used ad- Gentry, Pattison, & Heggestad, 2010), but not in IB or management research
ditional post-hoc statistical diagnostics to ensure that our data were not or for confirmatory purposes. Here we extend this technique for confirmatory
confounded by common method bias. We conducted several steps to control purposes to predict cultural fit, i.e., when relative cultural characteristics of
and to account for common method bias. First, we conducted Harman’s one- two entities are combined in such a way that the dependent variable under
factor tests on all of the items consecutively for Q1 (questionnaire answered investigation achieves its maximum possible value. This way we identified
by respondents from HQ) and Q2 (questionnaire answered by respondents the conditions under which cultural differences and similarities might max-
from Russian subsidiaries). We ran different diagnostic tests for different re- imize specific outcomes of interest. We suggest further that this technique
spondent types (Podsakoff & Organ, 1986). Our test for Q1 extracted two would be useful for future IB studies which aim to study other organizational
distinct factors that accounted for 72.3% of the total variance, with the first phenomena where fit is an issue.
factor explaining 41.6%. Our test for Q2, extracted three distinct factors that We run different analyses for each of the three dependent variables
accounted for 81.5% of the total variance, with the first factor explaining (Knowledge Outward Flow, Knowledge Inward Flow, and Knowledge
39.1%. This suggests that no single factor accounted for a large proportion of Implementation) to test the hypotheses about the conditions under which the
the variance in our data. Second, using confirmatory factor analysis, we optimum organizational culture fit between HQ and subsidiary is achieved.
compared three measurement models: (1) a model where each item is loaded As we explain further below, we performed analysis in two steps. The first
onto its theoretically assigned latent variable, (2) a model loading where all step is a traditional second-order regression model. The second step is a
items are linked to a common method factor (CMF), and (3) a combined transformation of the coefficients of X and Y to D = X − Y, and the formal
model where items are loaded to both their theoretically relevant latent test of our hypotheses. This second step is done in order to estimate fit
variables and to the CMF. Our first model for Q1 showed a good fit with the without reducing the cultural values/profiles of two entities to a singular/
data (RMSEA 0.062, SRMR 0.058, CFI 0.961). The second model for Q1 with unified distance measure. In our analysis, the unique aspect is that instead of
a single CMF showed a worse fit (RMSEA 0.128, SRMR 0.183, CFI 0.931) using differences scores (e.g., DE = HQE-SE) like most extant approaches use,
compared to the first model. Combining the first and second models, the third we use original culture levels (HQE and SE). Proceeding this way, we do not
model for Q1 also showed a relatively good fit (RMSEA 0.060, SRMR 0.053, lose information about the levels (i.e., about HQE and SE), and thus do not
CFI 0.961) but the improvement in the fit indices was not significant, and consider all HQ-S pairings identical as long as they produce to the same DE.
accounted for only a small portion of variance (p = 0.17). Turning to our
Table 4
Descriptive statistics, correlations, and reliabilities.
Notes: N = 186.
Bold value denotes the loading values that are higher than 0.40.
*
p < 0.05.
**
p < 0.01.
***
p < 0.001.
8
S. Morgulis-Yakushev et al. Journal of World Business xxx (xxxx) xxx–xxx
3
External Adaptation in Headquarter (HQE) 2.355 As an alternative to our main hypotheses, it is also possible to consider a model where
External Adaptation in Subsidiary (SE) 3.572 the relationship between the degree of DE and knowledge flows is moderated by SE (for
(HQE)2 2.268 H1b) and HQE (for H2b). That is; the degree of DE represents the extent to which there is a
HQE x SE 3.080 knowledge gap between two units, and thereby the opportunity to realize knowledge
(SE)2 2.150 transfer. Yet, realization of this opportunity hinges on the extent to which recipient unit
Internal Integration in Headquarter (HQI) 2.544 (i.e., subsidiary in case of inward knowledge flows and headquarters in the context of
Internal Integration in Subsidiary (SI) 3.725 outward knowledge flows) is oriented for learning from its external environment. To
(HQI)2 3.101 check if our results remain robust in this alternative model, we ran additional analyses to
HQI x SI 2.193 account for moderation effects. Our additional findings are in line with those we report in
(SI)2 1.745 the paper. For the sake of brevity, these results are not reported here but they are
available from the first author upon request. We would also like to thank one of the
Notes: N = 186. anonymous reviewers for encouraging us to conduct these additional tests.
9
S. Morgulis-Yakushev et al. Journal of World Business xxx (xxxx) xxx–xxx
Table 7
Confidence Intervals for Maximum Points.
DEmax
for the inward knowledge flow into 0.871** 0.692 1.048
subsidiary coming from HQ
DEmax
for the outward knowledge flow from −1.784*** −1.978 −1.590
subsidiary to HQ
DImax
for the subsidiary’s implementation of 1.448 −0.301 3.197
knowledge coming from HQ
Notes: Significance levels are based on confidence intervals constructed from coefficients
from 10,000 bootstrap samples, using the percentile method to determine critical values.
*
p < 0.05.
**
p < 0.01.
***
p < 0.001.
Fig. 3. The Relationship Between DE and Outward Knowledge Flow coming From the
Subsidiary into Headquarters.
that are difficult for other firms to imitate. MNCs are large and complex
entities, which operate in multiple locations and cultural/institutional
contexts (Zellmer-Bruhn & Gibson, 2006). This multiplicity can be both
a blessing and a curse for the MNC. However, the IB literature assumes
that cultural differences are prone to risks and problems (Stahl et al.,
2016). Furthermore, most work in IB assumes that all dimensions of
culture difference will have the same effect. However, there is no
reason to assume that this is the case. In this paper, we theoretically
motivate and empirically show that different dimensions of organiza-
tional culture create the optimum fit when there are different degrees of
dissimilarity between the HQ and its subsidiaries. We also showed that
the optimal amount of similarity varied depending on the stage or di-
rection of knowledge transfer (outcome variable) one considered. We
used evolutionary theory to develop hypotheses about how the relative
cultural characteristics of these two entities affect patterns of inward
and outward knowledge flows and the degree to which knowledge
Fig. 2. The Relationship Between DE and Inward Knowledge Flow into the Subsidiary transfer leads to effective learning in subsidiaries. In contrast to earlier
Coming from Headquarters. research, we do not consider cultural differences as an aggregate index-
based measure, but instead showed that different dimension of orga-
To test H3b, we need to check whether or not KI is highest when nizational culture behave differently. In addition, for each particular
HQI = SI. In order to see if that is the case, we need to examine the FC knowledge transfer outcome, our hypotheses made specific predictions
value presented in Table 6b, which contains F-ratios for the test of about the optimum degree and direction of distance between HQs and
constraints imposed by the squared difference score. Since FC (0.397) is subsidiaries.
not significant, we can conclude that a perfect fit for KI is achieved We used a unique dataset of matched pairs of HQs located in Europe
when DI equals zero. In other words, our results show that the degree of and subsidiaries established in Russia. Data were collected from both
KI is maximized when HQI and SI are identical (see Fig. 4). Therefore, HQs and subsidiaries. We used this dyadic dataset to test our hy-
we find support for H3b. Appendix B of Supplementary material pro- potheses. Our results reveal some interesting patterns. First, we found
vides the calculations and estimation procedures used to test H1b, H2b that extreme degrees of distance between HQs and subsidiaries are
and H3b in detail. detrimental to all knowledge transfer outcomes (HQ-to-subsidiary in-
flows, subsidiary-to-HQ outflows and knowledge implementation at the
subsidiary) considered in this study. Thus, in line with past theoretical
6. Conclusion and discussion
predictions (Björkman, Stahl, & Vaara, 2007), we show that moderate
degrees of organizational cultural differences between the HQ and a
Barney (1986) argues that organizational culture can be a basis for
subsidiary are more conducive to knowledge transfer. However, our
firms’ competitive advantage because it generates intangible resources
10
S. Morgulis-Yakushev et al. Journal of World Business xxx (xxxx) xxx–xxx
11
S. Morgulis-Yakushev et al. Journal of World Business xxx (xxxx) xxx–xxx
12
S. Morgulis-Yakushev et al. Journal of World Business xxx (xxxx) xxx–xxx
Seibert, S. E., Silver, S. R., & Randolph, W. A. (2004). Taking empowerment to the next 345–376.
level: A multiple-level model of empowerment, performance, and satisfaction. Van Wijk, R., Jansen, J. J., & Lyles, M. A. (2008). Inter-and intra-organizational knowl-
Academy of Management Journal, 47(3), 332–349. edge transfer: A meta-analytic review and assessment of its antecedents and con-
Shanock, L. R., Baran, B. E., Gentry, W. A., Pattison, S. C., & Heggestad, E. D. (2010). sequences. Journal of Management Studies, 45(4), 830–853.
Polynomial regression with response surface analysis: A powerful approach for ex- Vogus, T. J., & Sutcliffe, K. M. (2007). The safety organizing scale: Development and
amining moderation and overcoming limitations of difference scores. Journal of validation of a behavioral measure of safety culture in hospital nursing units. Medical
Business and Psychology, 25(4), 543–554. Care, 45(1), 46–54.
Shenkar, O. (2001). Cultural distance revisited: Towards a more rigorous con- Weber, Y., Shenkar, O., & Raveh, A. (1996). National and corporate cultural fit in mer-
ceptualization and measurement of cultural differences. Journal of International gers/acquisitions: An exploratory study. Management Science, 42(8), 1215–1227.
Business Studies, 32(3), 519–535. Wu, L.-Y. (2010). Applicability of the resource-based and dynamic capability views under
Stahl, G. K., & Tung, R. L. (2014). Towards a more balanced treatment of culture in environmental volatility. Journal of Business Research, 63(1), 27–31.
international business studies: The need for positive cross-cultural scholarship. Yildiz, H. E., & Fey, C. F. (2016). Are the extent and effect of psychic distance perceptions
Forthcoming in Journal of International Business Studies. http://dx.doi.org/10.1057/ symmetrical in cross-border M & As? Evidence from a two-country study. Journal of
jibs.2014.68. International Business Studies, 47(7), 830–857.
Stahl, G. K., Tung, R. L., Kostova, T., & Zellmer-Bruhn, M. (2016). Widening the lens: Zellmer-Bruhn, M., & Gibson, C. (2006). Multinational organization context: Implications
Rethinking distance, diversity, and foreignness in international business research for team learning and performance. Academy of Management Journal, 49(3), 501–518.
through positive organizational scholarship. Journal of International Business Studies, Ziman, J. (2000). Technological innovation as an evolutionary process. Cambridge, UK:
47(6), 621–630. Cambridge University Press.
Szulanski, G. (1996). Exploring internal stickiness: Impediments to the transfer of best Zohar, D., & Luria, G. (2004). Climate as a social-cognitive construction of supervisory
practice within the firm. Strategic Management Journal, 17, 27–43. safety practices: Scripts as proxy of behavior patterns. Journal of Applied Psychology,
Tinsley, H. E. A. (2000). The congruence myth: An analysis of the efficacy of the person- 89(2), 322–333.
environment fit model. Journal of Vocational Behavior, 56, 147–179. Zollo, M., & Winter, S. G. (2002). Deliberate learning and the evolution of dynamic
Tsui, A. S., Wang, H., & Xin, K. R. (2006). Organizational culture in China: An analysis of capabilities. Organization Science, 13(3), 339–351.
culture dimensions and culture types. Management and Organization Review, 2(3),
13