Subsidiary roles and reverse knowledge transfer: An investigation of the effects of coordination mechanisms

Journal of International Management 17 (2011) 97–113

Contents lists available at ScienceDirect

Journal of International Management

Subsidiary roles and reverse knowledge transfer: An investigation of the effects of coordination mechanisms Larissa Rabbiosi ⁎ Copenhagen Business School, Center for Strategic Management and Globalization, Porcelaenshaven 24, DK-2000 Frederiksberg, Denmark

a r t i c l e

i n f o

Article history: Received 24 September 2009 Received in revised form 8 August 2010 Accepted 17 October 2010 Available online 13 November 2010 Keywords: Reverse knowledge transfer Subsidiary roles Personal and electronic-based coordination mechanisms Subsidiary autonomy

a b s t r a c t In response to the increasing need to balance the pressures of global integration and local responsiveness, foreign subsidiaries must play a prominent role in the creation of knowledge that is valuable to the MNE as a whole. In this context, a key managerial problem relates to the balance between coordination mechanisms and knowledge flows from the subsidiary to the parent company, known as reverse knowledge transfer. It is crucial to understand the interdependencies between subsidiary roles and key coordination mechanisms, such as subsidiary autonomy, personal and electronic-based coordination mechanisms. This paper, therefore, offers new insights into the impact of coordination mechanisms on reverse knowledge transfer. On the basis of a data set consisting of 280 dyads between foreign subsidiaries and their parent companies, two distinctive configurations are found to positively affect the extent of reverse knowledge transfer. The first is the combination of a high degree of subsidiary autonomy and greater use of personal coordination mechanisms, and the second is the combination of low subsidiary autonomy and greater use of electronic-based coordination mechanisms. However, the relevance of these coordination configurations differs for various subsidiary roles. © 2010 Elsevier Inc. All rights reserved.

1. Introduction International management literature highlights the importance of foreign subsidiaries' abilities to create, develop and integrate knowledge through both their internal and external networks (Andersson et al., 2002; Phene and Almeida, 2008; Zanfei, 2000). Given their access to the existing knowledge pool in the local environment (Frost, 1998), foreign subsidiaries play a prominent role in MNE innovation (Subramaniam and Venkatraman, 2001; Venaik et al., 2005; Yamin and Forsgren, 2006) and they can directly enhance the MNE's strategic competitive advantage (Ambos et al., 2006; Bartlett and Ghoshal, 1989; Cantwell, 1995). Given the importance of subsidiaries to the competitiveness of MNEs, a growing number of studies examine how intra-MNE knowledge transfers are likely to have a significant influence on the value these firms can create and appropriate (Gupta and Govindarajan, 2000; Kogut and Zander, 1993; Schulz, 2001; Tsai, 2001). The knowledge flow from foreign subsidiaries to parent companies (i.e., reverse knowledge transfer) is one important process in the intra-MNE knowledge transfer phenomenon that has recently received attention (Eden, 2009; Frost and Zhou, 2005; Gupta and Govindarajan, 2000; Håkanson and Nobel, 2001; Yang et al., 2008). It has been argued that the deployment of coordination mechanisms, such as decentralization of decision-making and communication mechanisms, within the parent company–subsidiary relationship should be seen as a critical antecedent to knowledge exchange (e.g., Gupta and Govindarajan, 1991). While the relevance of subsidiary autonomy and communication mechanisms in knowledge transfer processes within MNEs has been acknowledged (e.g., Schulz, 2001; Tsai, 2001), our understanding of the factors involved is still limited. First, although coordination mechanisms have been found to differ across ⁎ Tel.: + 45 3815 2897; fax: + 45 3815 3035. E-mail address: [email protected]. 1075-4253/$ – see front matter © 2010 Elsevier Inc. All rights reserved. doi:10.1016/j.intman.2010.10.001

98

L. Rabbiosi / Journal of International Management 17 (2011) 97–113

subsidiary roles (Ambos and Schlegelmilch, 2007; Nobel and Birkinshaw, 1998), it is still unclear how these mechanisms affect the level of reverse knowledge transfer for various subsidiary types. Second, coordination mechanisms have typically been treated as independent dimensions (for a notable exception, see Noorderhaven and Harzing, 2009). This paper, therefore, argues that the relationship between subsidiary autonomy and reverse knowledge transfer can be better understood if subsidiary autonomy and communication mechanisms are considered jointly, as autonomy and communication mechanisms represent relevant coordination mechanisms within the parent company–subsidiary dyad. Only Noorderhaven and Harzing's recent study (2009) has analyzed the effect of the interaction between subsidiary autonomy and socialization mechanisms on the extent of reverse knowledge transfer. Although the Noorderhaven and Harzing study extensively increased our understanding of knowledge transfer processes, its findings are not linked to different types of foreign subsidiaries. It is necessary to address this limitation, as subsidiary roles, subsidiary knowledge and coordination mechanisms are closely related (Gupta and Govindarajan, 1991; Harzing and Noorderhaven, 2006; Mudambi et al., 2007; Nobel and Birkinshaw, 1998). Moreover, our understanding of reverse knowledge transfer processes will benefit extensively from consideration of the combined effect of subsidiary autonomy not only with personal coordination mechanisms but also with other mechanisms, such as electronic-based coordination. Accordingly, this paper develops predictions about the nature of the expected relationships between subsidiary autonomy, different communication mechanisms and the levels of reverse knowledge transfer that are contingent on subsidiary roles. As this discussion is limited to the level of reverse knowledge transfer, the unit of analysis is a dyad involving a focal foreign subsidiary and its parent company. The primary concern of this paper is the extent to which parent companies receive and use knowledge from their overseas subsidiaries, where knowledge is defined as “either expertise (e.g., skills and capabilities) or external market data of strategic value” (Gupta and Govindarajan, 1991, p. 773). The paper is organized as follows. The next section outlines the interplay between subsidiary roles and coordination strategies within the parent company–foreign subsidiary dyad. The specification of the hypotheses follows, after which the methodology and empirical results are presented. The paper concludes with a discussion of the findings, and their theoretical and managerial implications. 2. Coordination mechanisms and subsidiary roles Decision-making's decentralization has been the structural coordination mechanism of greatest interest in research on the coordination of parent company–subsidiary relations (Martinez and Jarillo, 1989). In research on multinational organizations, corporate–subsidiary decentralization has largely been operationalized on the dyadic level of the parent company's relationship with a specific subsidiary by measuring the relative control exercised by the parent company on the subsidiary's decision making. In other words, corporate–subsidiary decentralization is a measure of subsidiary autonomy (for an exhaustive review of subsidiary autonomy in an MNE context, see Egelhoff, 1988; Paterson and Brock, 2002; Young and Tavares, 2004). In addition to structural coordination mechanisms, such as the decentralization of decision making, organizations are coordinated through communication mechanisms (Martinez and Jarillo, 1989). Coordination through communication mechanisms usually involves socialization forms, and includes such mechanisms as the participation of subsidiary managers in international task forces and teamwork, the transfer of personnel, the establishment of committees and meetings (Ambos and Schlegelmilch, 2007; Gupta and Govindarajan, 1991; Harzing and Noorderhaven, 2006; Nobel and Birkinshaw, 1998). However, advances in communication capabilities through electronic communication technologies have created new, electronic means of coordination (Fulk and DeSanctis, 1995; Yates and Orlikowski, 1992). Therefore, human-based coordination can be reduced in some parts of the organizational hierarchy, and parent–subsidiary coordination needs can be met by taking advantage of both personal and electronic-based coordination mechanisms. Foreign subsidiaries have been categorized along a number of different dimensions. Some of the subsidiaries within the MNE network may depend on the competence of their parent companies, so that their role is competence exploiting (Cantwell and Mudambi, 2005), either as a local implementer (Gupta and Govindarajan, 1991) or as an assembler (Cantwell, 1987). Others take on a more creative function (Pearce, 1999) as integrated players (Gupta and Govindarajan, 1991) or as centers of excellence (Holm and Pedersen, 2000). As suggested by Ambos and Schlegelmilch (2007), a consensus seems to be emerging that it is useful to distinguish among three types of subsidiary roles: (1) units that adapt products to local market needs, (2) units that exploit the MNEs' technological competencies on a global basis, and (3) units established to augment or create new technological competencies abroad. Previous studies have utilized various typologies of these roles. The conceptualization of foreign units used here is adopted from Ghoshal (1986), and they are subsequently referred to as: (1) implementer subsidiaries, (2) contributor subsidiaries and (3) innovator subsidiaries. Researchers have emphasized the interplay between the decentralization of decision making and subsidiary roles (Ambos and Schlegelmilch, 2007; Bartlett and Ghoshal, 1989; Gupta and Govindarajan, 1991; Martinez and Jarillo, 1989; Nobel and Birkinshaw, 1998). A greater degree of autonomy is often believed to be positively related to subsidiary knowledge creation and development (Ghoshal and Nohria, 1989; Gupta and Govindarajan, 1991; Nohria and Ghoshal, 1994; Venaik et al., 2005). This is based on the idea that independent subsidiaries have strategic mandates (Birkinshaw et al., 1998) that favor local responsiveness (Bartlett and Ghoshal, 1989) and the development of knowledge by tapping into local knowledge bases (Andersson et al., 2002; Cantwell, 1995; Zanfei, 2000). Although the opposite has also been found (Brockhoff and Schmaul, 1996; Frost et al., 2002), management scholars tend to agree that autonomy enhances subsidiaries' abilities to learn from local systems of innovation and allow MNEs to benefit from new knowledge developed by those subsidiaries (Cantwell and Mudambi, 2005; Foss and Pedersen,

L. Rabbiosi / Journal of International Management 17 (2011) 97–113

99

2002; Nobel and Birkinshaw, 1998; Taggart and Hood, 1999). Accordingly, relative to implementer subsidiaries, contributor and innovator subsidiaries will experience higher levels of decentralization in decision making. Moreover, the degree of autonomy will be highest for innovator subsidiaries (Ambos and Schlegelmilch, 2007; Harzing and Noorderhaven, 2006; Nobel and Birkinshaw, 1998). Similarly, different scholars have investigated the level of personal coordination mechanisms for different subsidiary roles. Given the many definitions of subsidiary roles (for an exhaustive review, see Cantwell and Mudambi, 2005), the findings are not clear-cut. However, using some degree of generalization, it can be asserted that, relative to implementer subsidiaries, both contributor and innovator subsidiaries will experience coordination through personal mechanisms. This type of coordination should be highest for innovator subsidiaries (Ambos and Schlegelmilch, 2007; Harzing and Noorderhaven, 2006; Nobel and Birkinshaw, 1998). Furthermore, no studies appear to have considered coordination by electronic-based mechanisms in conjunction with different subsidiary roles. 3. Enhancing the levels of reverse knowledge transfer: hypothesis development Recent, but well-established, literature analyzes reverse knowledge transfer — the transfer of knowledge from foreign subsidiaries to their parent companies (Ambos et al., 2006; Frost, 1998; Frost and Zhou, 2005; Gupta and Govindarajan, 2000; Håkanson and Nobel, 2001; Yang et al., 2008). This recent trend is in line with the broader recognition that foreign subsidiaries can serve as sources of innovations (Birkinshaw et al., 1998; Pearce and Papanastassiou, 1999) that can be transferred to and used by parent companies, thereby contributing to the creation of firms' competitive advantages (Ambos et al., 2006). Parent companies may benefit from the utilization of knowledge originating in a foreign subsidiary for different reasons (Eden, 2009). They can use subsidiary knowledge to better coordinate a global strategy; to improve the development of new products, technologies or services (Ambos et al., 2006); or to control and monitor subsidiaries' power (Yamin and Forsgren, 2006). Parent companies may also play a role in channeling knowledge to the appropriate MNE unit, thereby orchestrating knowledge transfer processes in their own network (Criscuolo and Narula, 2007; Phene and Almeida, 2008). Although boundaries can be blurred, knowledge in innovator subsidiaries can be distinguished from knowledge in contributor or implementer subsidiaries. Innovator subsidiaries engage in the development of new products and technologies, they discover new tastes or business practices, and they are the “company's innovative spark plugs” (Bartlett and Ghoshal, 1989, p. 153). The creation and development of new knowledge by innovator subsidiaries are often based on intensive information and knowledge exchange with local organizations, which leads to more context-specific knowledge (Andersson et al., 2002). Therefore, the competencies and knowledge of innovator subsidiaries are likely to have a high degree of tacitness and limited overlap with the parent company's existing knowledge. Conversely, contributor subsidiaries – similar to Kuemmerle's (1999) home base exploiting units – engage in improvements of products and services, often expressly for foreign markets, and they can offer technological enhancements and provide backup for local and global manufacturing units. There is a certain extent of overlap between the capabilities and knowledge of contributor subsidiaries, and the capabilities and knowledge of their parent companies (Cantwell and Mudambi, 2005). Finally, implementer subsidiaries usually execute corporate strategy. They do not engage in extensive knowledge creation and seldom possess knowledge relevant for the rest of the MNE (Ambos et al., 2006). Strategic management research on the factors affecting reverse knowledge transfer has grown dramatically in the last decade. One important issue investigated in previous studies is the effect of coordination mechanisms, such as the decentralization of decision making, personal coordination mechanisms and electronic-based coordination mechanisms, on knowledge transfer processes (Ambos and Ambos, 2009; Gupta and Govindarajan, 2000; Håkanson and Nobel, 2001; Noorderhaven and Harzing, 2009; Schulz, 2001; Tsai, 2001). The majority of the literature has focused on the impact of formal and informal personal coordination relationships on knowledge transfer (Björkman et al., 2004; Gupta and Govindarajan, 2000; Schulz, 2001; Singh, 2005; Tsai, 2002). Attention has also been paid to the role played by electronic-based coordination mechanisms in knowledge transfer processes, although this aspect has received less emphasis (Ambos and Ambos, 2009; Persaud, 2005). Although these studies offer an excellent starting point, they are limited in that they fail to consider the fact that the diversity of subsidiary roles implies significant differences in subsidiary knowledge (Bartlett and Ghoshal, 1989) and in parent–subsidiary coordination mechanisms (Ambos and Schlegelmilch, 2007; Nobel and Birkinshaw, 1998). In fact, the effects of subsidiary characteristics and transfer mechanisms on reverse knowledge transfer are expected to depend on the nature of the subsidiary mandate (Mudambi et al., 2007). Accordingly, the effect of different communication coordination mechanisms within the parent–subsidiary relationship on reverse knowledge transfer will vary for different subsidiary roles. Furthermore, the literature has failed to present unambiguous evidence of the direct effect of subsidiary autonomy on the extent of reverse knowledge transfer: positive effects, negative effects and no effects have all been found (Gammelgaard et al., 2004; Ghoshal et al., 1994; Noorderhaven and Harzing, 2009; Schulz, 2001). Different degrees of autonomy explain the magnitude and scope of knowledge creation expected from the foreign subsidiary (Gupta and Govindarajan, 1991). Therefore, different degrees of subsidiary autonomy are closely related to subsidiary roles (Harzing and Noorderhaven, 2006). Moreover, within parent company–foreign subsidiary dyadic relationships, the design of coordination strategies that utilize different combinations of coordination instruments is expected to depend on the subsidiary role (Ambos and Schlegelmilch, 2007; Nobel and Birkinshaw, 1998). Accordingly, this paper suggests that the effect of subsidiary autonomy on reverse knowledge transfer cannot be considered independently from the use of different communication mechanisms or separately from subsidiary roles. The next section develops a theoretical contribution on the contingent effect of subsidiary roles on the relationship between the level of reverse knowledge transfer and the use of different coordination mechanisms.

100

L. Rabbiosi / Journal of International Management 17 (2011) 97–113

3.1. Communication mechanisms and reverse knowledge transfer: the effect of subsidiary roles Face-to-face coordination mechanisms can facilitate multiple contacts within the MNE network. These contacts can be used to collect and transfer knowledge across units. The direct impact of coordination through personal-based communication mechanisms on knowledge transfer appears to be quite univocal: extensive personnel transfers and meetings are mechanisms for creating a “verbal network”. A verbal network is a system for transmitting knowledge and information complementary to or even substitutive for written systems (Edstrom and Galbraith, 1977). Using patent citation data, Singh (2005) finds a greater probability of knowledge flows within a firm when inventors have close interpersonal ties. Indeed, researchers have suggested that the successful development of relations between individuals from two distinct units of an MNE facilitates the transfer of knowledge within the MNE (Bartlett and Ghoshal, 1989; Kostova, 1999). Accordingly, Nohria and Ghoshal (1997) suggest that an MNE is a conglomeration of interpersonal ties that hold the differentiated and geographically dispersed units together. Coordination through the interpersonal network makes knowledge exchange between interconnected units possible (Björkman et al., 2004; Ghoshal et al., 1994; Gupta and Govindarajan, 2000; Hansen, 1999; Tsai, 2001). Coordination mechanisms based on electronic media are designed to support and augment organizational knowledge management. They should complement and enhance the communication ability of the firm (Alavi and Leidner, 2001). Consensus also exists on the role played by electronic-based coordination mechanisms in the international transfer of knowledge (Almeida et al., 2002; Andersen and Foss, 2005). Electronic communication systems, from electronic mail to more sophisticated systems, enable communication among managers of different units and the sharing of different types of documents (DeSanctis and Fulk, 1999; Niederman, 2005). Recent literature shows that electronic-based mechanisms reduce knowledge heterogeneity and promote exploitation learning (Kane and Alavi, 2007), such as incremental learning focused on refinement, and the reuse of existing knowledge. In sum, this discussion suggests the following baseline hypotheses: Hypothesis 1a. Personal coordination mechanisms positively affect the level of reverse knowledge transfer. Hypothesis 1b. Electronic-based coordination mechanisms positively affect the level of reverse knowledge transfer. Confirmation of these hypotheses would provide a basic setting against which we can hypothesize and assess the contingent effects of subsidiary roles. The competencies and knowledge of innovator subsidiaries have limited overlap with the extant parent company's knowledge. In contrast, the capabilities and knowledge of contributor subsidiaries partially overlap with those of the parent company (Cantwell and Mudambi, 2005). Therefore, the knowledge held by contributor subsidiaries does not need to be re-contextualized in order to be transferred (Makino and Delios, 1996), and parent companies have a higher capacity to absorb, apply and integrate it into the organization (Yang et al., 2008). Recent research on the search for innovation emphasizes that the mobility of individuals not only creates an opportunity to transfer expertise but also to facilitate the interpretation of knowledge in a new context (Rosenkopf and Almeida 2003). From a cognitive perspective, when subsidiary knowledge becomes more complex and tacit, personal coordination mechanisms offer a set of richer information transmission channels that have more capacity to transfer complex and tacit knowledge (Daft and Lengel 1986; Gupta and Govindarajan 1991). In contrast, electronic tools are effective for transfers of standard data, and well-understood messages and information, while their limitations are evident when less codified knowledge has to be transferred (Daft and Lengel, 1986; Pedersen et al., 2003). Indeed, the transfer of knowledge through electronic-based coordination mechanisms requires that firms develop “knowledge objects” (Hansen et al., 1999) — knowledge that can be viewed as an object that can be stored and manipulated. Such knowledge is mostly collected in databases, reports or handbooks through a “people-to-documents” process, and it can be accessed and used relatively easily by anyone in the company (see, for example, Hansen et al., 1999; Zollo and Winter, 2002). Under these conditions, reverse knowledge transfer from contributor subsidiaries can be expected to benefit from the use of electronic-based coordination mechanisms. The effective transfer of innovator subsidiaries' knowledge through electronic-based mechanisms should also be possible in many cases, but only after the application of a specific codification strategy that involves the transformation of tacit knowledge into explicit knowledge (Schulz and Jobe, 2001). However, given the cost of codifying a subsidiary's knowledge that has limited overlap with the parent company's knowledge, parent companies are likely to abstain from such codification and rely more on personal coordination mechanisms (Mudambi et al., 2007). Accordingly, we arrive at the following hypotheses: Hypothesis 2a. The positive impact of personal coordination mechanisms on the level of reverse knowledge transfer will be highest for innovator subsidiaries. Hypothesis 2b. The positive impact of electronic-based coordination mechanisms on the level of reverse knowledge transfer will be highest for contributor subsidiaries. As implementer subsidiaries rarely possess knowledge relevant for the rest of the MNE, they seldom engage in reverse knowledge transfer (Ambos et al., 2006). Therefore, a specific hypothesis on reverse knowledge transfers from these subsidiaries is not included here. The same argumentation holds for the hypotheses that follow.

L. Rabbiosi / Journal of International Management 17 (2011) 97–113

101

3.2. Subsidiary autonomy, communication mechanisms and subsidiary roles: the impact on reverse knowledge transfer Greater autonomy motivates subsidiary managers to take initiatives and innovate (Ghoshal and Bartlett, 1988; Venaik et al., 2005). To do so, they often take advantage of close relationships with local suppliers, customers and research institutions (Andersson and Forsgren, 2000; Andersson et al., 2002). Therefore, changes in the degree of subsidiary autonomy are expected to affect subsidiary knowledge. This paper argues that when subsidiary autonomy is high, the amount of reverse knowledge transfer will rise given greater use of personal coordination mechanisms, while a high use of electronic-based coordination mechanisms will be detrimental. However, these relationships are contingent on the actual role of the subsidiary. Personal coordination mechanisms are expected to play a prominent role when uncertainty increases and hierarchical means of coordination are likely to fail (Nobel and Birkinshaw, 1998). When subsidiary autonomy is high, forms of autonomy-control tension can manifest in the MNE (Asakawa, 1996, 2001) thereby limiting reverse knowledge transfer. Therefore, when a high level of autonomy is granted to the subsidiary, the hierarchical coordination, connections and dependency between the parent company and the subsidiary may decrease (Noorderhaven and Harzing, 2009), and reciprocal trust may be reduced, which in turn will diminish the transfer of knowledge to the other units of the MNE (Forsgren et al., 2000; Gammelgaard et al., 2004). Nevertheless, higher levels of personal coordination will stimulate the convergence of interests, and the sharing of beliefs, common values and norms among managers of different units (Tsai and Ghoshal, 1998), thereby reducing parent–subsidiary tension (Nobel and Birkinshaw, 1998). These arguments suggest that the amount of reverse knowledge transfer will increase when parent companies use personal coordination mechanisms extensively with those subsidiaries that have more autonomy. However, differences exist among subsidiary roles. Given their network centrality and resource endowments, innovator subsidiaries usually possess greater power within the MNE (Ambos and Schlegelmilch, 2007; Bouquet and Birkinshaw, 2008). Personal coordination mechanisms are, therefore, used extensively as a control strategy to counter the shift of power to the foreign unit (Medcof, 2001). In other words, a combination of decision-making decentralization with personal coordination mechanisms is often necessary in the management of innovator subsidiaries. Therefore, an increase in the level of an innovator subsidiary's autonomy will not modify the actual coordination structure between these subsidiaries and their parents, and the emphasis on the use of personal coordination mechanisms will change only marginally. Accordingly, no additional effects on the level of reverse knowledge transfer should be observed when parent companies use personal coordination mechanism extensively with innovator subsidiaries with high degrees of autonomy. On the other hand, changes in the level of autonomy held by contributor subsidiaries should modify the scope of the activities performed by these units. Specifically, when contributor subsidiaries receive or gain greater autonomy, they can evolve and reach a new status. Connections with the external research and business communities become stronger and more power is gained within the MNE (Andersson et al., 2007; Bouquet and Birkinshaw, 2008; Harzing and Noorderhaven, 2006). The unit can also aspire to a more creative role with new product development responsibilities and a more international scope (Birkinshaw et al., 1998). Increasing levels of power make an unfavorable power balance between the contributor subsidiary and its parent company more likely (Asakawa, 1996, 2001). According to our general arguments, parent companies will satisfy their need to reduce this autonomy–control tension by applying more social control. In other words, when the degree of autonomy of contributor subsidiaries increases, parent companies are likely to use more personal coordination mechanisms as a means of modifying and sustaining the parent–subsidiary relationship. Given the effect of personal coordination mechanisms on knowledge transfer, the following hypothesis is formulated: Hypothesis 3. For contributor subsidiaries, the higher the degree of autonomy held by the focal subsidiary, the greater the impact of personal coordination mechanisms on the level of reverse knowledge transfer. The literature suggests that low levels of subsidiary autonomy reflect specific characteristics of the subsidiaries, such as a low level of R&D activities, or greater involvement in the adaptation of manufacturing technology and marketing decisions for the local environment (Taggart, 1997). Accordingly, in these cases, subsidiary knowledge is likely to partially overlap with extant knowledge of the parent company. In other words, low degrees of subsidiary autonomy are more likely when subsidiaries are expected to face environments with relatively low complexity (Ghoshal and Nohria, 1989), and when subsidiary knowledge is mainly explicit or codifiable in a cost-efficient manner (Schulz and Jobe, 2001). Under these conditions, the transfer of knowledge from subsidiaries to parent companies through electronic-based mechanisms is expected to be efficient. Therefore, a positive impact on the level of reverse knowledge transfer should be evident when a low degree of subsidiary autonomy is combined with greater use of electronic-based coordination mechanisms within the parent–subsidiary relationship. In other words, the level of reverse knowledge transfer from highly autonomous foreign subsidiaries that use electronic-based coordination mechanisms extensively with their parent companies will be lower. This negative effect on reverse knowledge transfer should be stronger for innovator subsidiaries than for contributor subsidiaries. In fact, organizations that experience only minor environmental turbulence have been shown to benefit from electronic-based coordination mechanisms (Kane and Alavi, 2007). However, organizations generally abstain from codifying their knowledge if the costs of the process exceed the benefits. High degrees of subsidiary autonomy are more likely in subsidiaries that generally deal with tacit knowledge and face a more turbulent environment (Cantwell and Mudambi, 2005; Gupta and Govindarajan, 1991). However, as explained above, under these circumstances the subsidiary's knowledge might be more context-specific and different from that of the parent. Accordingly, in order to use electronic-based coordination mechanisms, knowledge must be processed and codified prior to the transfer (Schulz and Jobe, 2001) or knowledge flows will be subject to transmission losses (Daft and Lengel, 1986; Shannon and Weaver, 1998;

102

L. Rabbiosi / Journal of International Management 17 (2011) 97–113

Szulanski, 1996). Such codification involves considerable costs of creating and maintaining repositories of organizational knowledge (Hansen et al., 1999; Schulz and Jobe, 2001). To ensure that the codification process works efficiently, organizations have to invest in the creation of an integrated electronic repository, institute comparable formats, and design a customized, common taxonomy (Weiss et al., 2004). Only if standardization, communalities and compatible information systems are widespread and maintained among the units for some time will MNEs be able to achieve efficient knowledge transfer when using electronic-based communication. However, the management information systems literature suggests that a decentralized MNE will be more likely to pursue independent information system operations in each subsidiary (Jarvenpaa and Ives, 1993), and that local managers with a high degree of autonomy in decision making, such as managers of innovator subsidiaries, may perceive a high level of corporate standards as intrusive (Karimi and Konsynski, 1991). Accordingly, we propose: Hypothesis 4. For innovator subsidiaries, the higher the degree of autonomy of the focal subsidiary, the lower the impact of electronic-based coordination mechanisms on the level of reverse knowledge transfer. 4. Method 4.1. Sample and data collection Sample selection was guided by the following criteria. First, the sample frame was generated from the Reprint database (Mariotti and Mutinelli, 2005), which contains census data on the foreign activities of Italian firms as of the beginning of 2004. Second, the sample frame was limited to all Italian MNEs with more than 50 employees operating in manufacturing industries, with at least one majority-owned subsidiary located in advanced countries and involved in “primary upstream activities”, such as R&D and manufacturing (Gupta and Govindarajan, 2000). On this basis, the final sample frame consisted of 358 Italian MNEs, of which 84 were studied through on-site face-to-face structured interviews with the parent companies' top managers (response rate of approximately 24%). Five of these MNEs also served as sites for the pre-testing and refinement of the questionnaire. The dyad of a focal foreign subsidiary and its parent company is used as the unit of analysis. Data collection took place from December 2004 to July 2005 and involved six researchers. The top manager of each parent company was contacted by phone and asked for a face-to-face interview. At the same time, a personalized letter with a description of the project and assurances regarding the confidentiality of collected data was sent to each manager. For each of the 84 MNEs, data regarding the dyadic relationships of the parent company with each of its majority-owned foreign subsidiaries was collected, regardless of the subsidiaries' locations. In general, the interviews took from one to three hours, although for parent companies with more than five subsidiaries (about 15% of the sample), the interviews were typically longer and sometimes spanned over two days. The final dataset consisted of 290 parent company-foreign subsidiary dyads. To assess non-response bias, a test of whether responding MNEs differ from non-responding MNEs with respect to size (class of number of employees) and sector, as defined by Pavitt (1984, 1990), was undertaken. Four technological trajectories were identified: supplier-dominated, specialized supplier, science based and scale intensive. The tests indicate that science-based (p b 0.1) and specialized-supplier sectors (p b 0.05) are overrepresented in the sample, while supplier-dominated sectors (p b 0.001) are underrepresented (see Table 1). Given the low representativeness of MNEs in the supplier-dominated sectors, we removed the related 10 observations from the sample. 280 parent company-foreign subsidiary dyads are, therefore, included in the empirical study. Accordingly, the generalization of the results to supplier-dominated industries must be made with the necessary caution. The use of perceptual measures introduces the risk of common method bias. The survey was therefore designed using a set of procedural remedies to reduce the potential for common method bias. First, during the in depth face-to-face interviews, the measurement of the dependent and the explanatory variables was separated by introducing a time lag (Podsakoff et al., 2003). Second, different scale endpoints and formats were used for the different variables in order to reduce “the possibility that some of the covariation observed among the constructs examined may be the result of the consistency in the scale properties rather than the content of the items” (Podsakoff et al., 2003, p. 884). The Harman's single-factor test was performed on items included in the

Table 1 Sample representativeness.

Industry

Size

⁎ p b .10. ⁎⁎ p b .05. ⁎⁎⁎ p b .01.

Science based Specialized suppliers Scale intensive Supplier dominated 50–249 250–499 500–5000 N 5000

Sample frame

Non-respondent

Respondent

P-value (χ2 test)

44 65 163 86 98 81 145 34

29 42 125 78 80 66 102 26

15 23 38 8 18 15 43 8

0.0757 ⁎ 0.0122 ⁎⁎ 0.9765 0.0003 ⁎⁎⁎ 0.1624 0.2325 0.0225 ⁎⁎ 0.9924

L. Rabbiosi / Journal of International Management 17 (2011) 97–113

103

Table 2 Validity response tests. Observations a Dependent variable Level of reverse knowledge transfer Independent variables Definition of R&D projects, planning, etc. Introduction of new technologies Changes in products/services Hiring and firing of subsidiary workforce Teamwork Temporary transfer of managers Temporary transfer of professionals Exchange of documents, blueprints, dbs, etc. Internet tools a b

Kruskal–Wallis test

62

1.250

62 62 62 62 59 55 55 57 56

1.395 0.009 0.012 4.822 1.988 26.988 23.554 0.015 1.621

P-value b 0.264 0.238 0.923 0.912 0.028 ⁎⁎ 0.159 0.000 ⁎⁎⁎ 0.000 ⁎⁎⁎ 0.904 0.203

Number of parent company-foreign subsidiary dyads available. Rejecting the null hypothesis means that the two distributions – the answers from the parent companies and the answers from the subsidiaries – are different. ⁎ p b .10. ⁎⁎ p b .05. ⁎⁎⁎ p b .01.

econometric model to assess common method bias in the sample (Harman, 1967; Podsakoff and Organ, 1986). The unrotated factor analysis reveals ten factors with eigenvalues greater than one, the first of which (eigenvalue = 3.57) explains 14.30% of the variance. This result indicates that common method bias is not a serious problem. To rule out possible one-respondent bias, the dependent and independent variables based on perceptual data were subjected to validity response tests. In fact, for 62 dyads, relevant information was collected not only from the parent companies' manager but also from the subsidiaries' top manager. On this basis, a Kruskal–Wallis equality of populations rank test, which tests whether two (or more) samples are from the same population (Brett et al., 1995; Downey et al., 1975), was run. In general, the validity tests confirm that the responses from managers at parent companies and at subsidiaries regarding our dependent and independent variables are not significantly different (see Table 2) with three exceptions. Two of them concern the items “temporary transfer of managers” and “temporary transfer of professionals”, which are used to operationalize the concept of personal coordination mechanisms. The parent companies perceive a greater number of personnel transfers with their subsidiaries than the foreign subsidiaries perceive. However, this is to be expected. Although personnel transfers may be one-way or two-way (from the parent company to the subsidiary and vice-versa), the interviews indicate that such transfers are generally unidirectional from the parent to the subsidiary. In addition, as the diffusion of this coordination practice in the MNE is usually encouraged and formalized by the parent company, it is not surprising that the foreign subsidiaries report fewer personnel transfers than their parent companies. The last exception concerns the item “hiring and firing of subsidiary workforce”, which is used in the operationalization of the concept of subsidiary autonomy. In this case, the answers from the parent company managers average 3.19, while the answers from the subsidiaries' top managers average 3.66. The difference might be explained by the wording of the item, which does not specify the type of the workforce to be considered. Therefore, subsidiary managers are likely to have considered all possible employees categories (from top managers to blue collar workers), while parent managers may have focused more on managerial positions. 4.2. Measures 4.2.1. Reverse knowledge transfer The dependent variable of this study, reverse knowledge transfer (RKT), is operationalized as the degree to which the foreign subsidiary's knowledge is used, as perceived by the parent company (Minbaeva et al., 2003). Indeed, “[t]he key element in knowledge transfer is not the underlying (original) knowledge, but rather the extent to which the receiver acquires potentially useful knowledge and utilizes this knowledge in own operations” (Minbaeva et al., 2003, p. 587). In this respect, the transfer of technology, know-how, skills and capabilities (i.e., expertise) from the foreign subsidiary to its parent company are considered. Using open questions, informants were first asked to provide descriptions of the subsidiary's expertise and knowledge of products, technologies and primary activities (Schulz, 2001). Based on their answers, other questions were posed that focused on whether the identified subsidiary's knowledge pertaining to the above three domains was transferred to and used by the parent company. Examples of these questions include: Has the subsidiary's technology that you described been transferred to the parent company?1 Has this technology been used by the parent company? Has the technological know-how related to the described technology been used by the parent company? Have the specific subsidiary's competencies that you described been

1 During the interview, the question referred specifically to the technology that was mentioned during the previous open questions. Examples of these technologies and related know-how include: technological know-how for the production of surfaces of catalytic stoves; packaging technologies; insulation technologies; technologies for closure devices in aluminium; technological know-how for the production of thick tubes with wide diameters; technological and product development know-how for carbon clutches used for racing cars; software and technological know-how for angiography and quantitative data analysis.

104

L. Rabbiosi / Journal of International Management 17 (2011) 97–113

transferred to and used by the parent company?2 After this conversation, respondents were asked to aggregate the subsidiary's knowledge across the identified domains, and indicate of the acquired subsidiary knowledge the extent to which the parent company had used (“null” = 0; “low” = 1; “medium” = 2; “high” = 3). Based on this answer, the dependent variable reverse knowledge transfer is given a value ranging from 0 (no transfer at all) to 3 (high transfer). Therefore, transfers were assessed from the receiving unit's perspective, i.e., the parent company. In line with Lord and Ranft (2000, p. 582), “this was done primarily because to try to measure knowledge transfer from the sender's perspective is inherently problematic — e.g., knowledge that is ‘sent’ is not always ‘received’ (Szulanski, 1996)”. Furthermore, it is crucial to note that parent–subsidiary relationships in which reverse knowledge transfer did not occur were also considered (i.e., the dependent variable assumes the value of zero). Removing or ignoring these cases would introduce serious problems of selection bias (Heckman, 1979). 4.2.2. Subsidiary autonomy The measure of subsidiary autonomy is essentially based on questions originally developed by Ghoshal and Nohria (1989). Specifically, we know at which MNE level each of the following strategic decisions are taken: (i) definition of R&D projects, planning, resources, etc.; (ii) introduction of new technologies; (iii) changes in products/services; and (iv) hiring and firing of the subsidiary workforce. The following scale was used: (1) “the parent company decides alone”, (2) “the parent company decides but considers subsidiary inputs”, (3) “both the parent company and subsidiary have roughly equal influence on decisions”, (4) “the subsidiary decides, but considers parent company suggestions”, and (5) “the subsidiary decides alone” (Ghoshal et al., 1994; Ghoshal and Nohria, 1989). The final measure of subsidiary autonomy is the average of responses to the four items (Cronbach's alpha = 0.74). 4.2.3. Personal and electronic-based coordination mechanisms Respondents were asked to indicate how frequently the following communication mechanisms were used to coordinate relations between the parent company and the focal subsidiary: (i) teamwork involving people from both the foreign subsidiary and the parent company, (ii) temporary (short-term) transfers of managers within the parent–subsidiary dyad, (iii) temporary (short-term) transfers of scientific and technical staff (researchers, engineers, etc.) within the parent–subsidiary dyad, (iv) internet-based instruments, such as forums, newsletters, e-mails, instant messages, etc., and (v) the exchange of documents, such as handbooks, blueprints and databases. Respondents were asked to assess the intensity of the use of these coordination mechanisms on a seven-point Likert scale ranging from “used rarely” to “used very often”. With regards to “short-term” transfers, respondents were asked to consider the movement of people other than those visiting for only one or a few days, or expatriates. The variable personal coordination is the average of responses to the first three items (Cronbach's alpha = 0.72) and captures parent-subsidiary coordination based on face-to-face interactions (Ambos and Schlegelmilch, 2007; Edstrom and Galbraith, 1977; Gupta and Govindarajan, 2000). Conversely, the average of responses to the last two items (Cronbach's alpha = 0.60) measures the variable electronic-based coordination (Haas and Hansen, 2005; Pedersen et al., 2003). 4.2.4. Interaction effects To capture the combined effect of using a specific coordination configuration, this paper follows previous work (e.g., Birkinshaw et al., 2002) in the construction of interaction terms between each of the two communication mechanisms and the subsidiary autonomy variable. Specifically, the following variables are added to the model: personal coordination × subsidiary autonomy and electronic-based coordination × subsidiary autonomy. 4.2.5. Subsidiary roles Previous studies have utilized a number of different approaches to operationalize subsidiary typologies. The implementation of this variable is adopted from Ghoshal (1986), and a distinction is made among “implementer subsidiaries”, “contributor subsidiaries” and “innovator subsidiaries”. Following Nobel and Birkinshaw (1998), and Ambos and Schlegelmilch (2007), a relatively simple heuristic based on the nature of the subsidiary's activities is also applied. Specifically, respondents were asked to indicate whether the focal foreign subsidiary was devoted to activities aimed at the creation of new products and/or new technologies (“capability-augmenting activities”), or to activities directed towards product or process improvements (“capabilityexploiting activities”). Those subsidiaries that are neither capability-augmenting nor capability-exploiting are called implementers. Those that are capability-exploiting but not capability-augmenting are contributors, while those that are capability-exploiting and capability-augmenting are labeled innovators. Accordingly, what is measured is not the formal mandate given by the parent company to the focal subsidiary but rather the realization of that mandate by the subsidiary. It acts as a proxy of the actual influence that subsidiaries have on value creation in the MNE.

2 These competencies could refer to product development expertise, such as product development know-how related to: cooking oil retrieval; glue for ceramic tiles and stone materials; admixtures for concrete; and rubber packing and sheathing. They can also refer to more general knowledge and skills, such as knowledge about procedures in clinical research for cardiovascular aspirators; quality management techniques and lean production practices; and marketing strategies.

L. Rabbiosi / Journal of International Management 17 (2011) 97–113

105

Table 3 Cross-tabulation: level of reverse knowledge transfer and subsidiary roles. Level of reverse knowledge transfer

0 1 2 3 Total

Subsidiary roles Innovator

Contributor

Implementer

Frequency (%)

Frequency (%)

Frequency (%)

51 (58.62) 9 (10.34) 14 (16.09) 13 (14.94) 87 (100)

23 2 10 11 46

121 (80.13) 19 (12.58) 9 (5.96) 2 (1.32) 151 (100)

(50.00) (4.35) (21.74) (23.91) (100)

4.2.6. Control variables Other likely predictors of reverse knowledge transfer are controlled for to avoid spurious effects. In particular, the following variables are operationalized. - Entry mode. Although acquisitions and joint ventures have traditionally been viewed as a common way for MNEs to access local competencies and skills (Kogut and Zander, 1993; Lane et al., 2001), empirical studies have found that the level of technology transfer from subsidiaries to parent companies is higher for greenfield subsidiaries than for acquisitions (Frost, 1998; Zhou, 2002). Therefore, in order to capture the effects of the entry mode on the levels of reverse knowledge transfer, the dummy variable greenfield is added to the model. - Parent-subsidiary distance. The literature indicates that the cost of knowledge transfer increases as the distance between home and host country increases (Ambos and Ambos, 2009; Yamin and Forsgren, 2006). Accordingly, it is necessary to control for the cultural distance in the parent-subsidiary dyad. The variable cultural distance is measured using Kogut and Singh's (1988) cultural distance index (others have also adopted this measure. See, for instance, Ambos et al., 2006; Håkanson and Nobel, 2001). - Subsidiary location. As the level and nature of knowledge available at the subsidiary can be affected by the subsidiary's location, the dummy variables Western Europe and North America are added to the empirical specification. The former captures whether the subsidiary is located in a west European country, excluding Portugal and Greece. The latter equals one when the subsidiary is located in the US or Canada. - Type of industry. As the type of industry can affect knowledge characteristics and coordination strategies in parent company– subsidiary relations, it is crucial to control for industry-specific effects. Following the taxonomy developed by Pavitt (1984), the dummy variable high-tech industries equals one if the subsidiary operates in either “science-based” or “specialized suppliers” sectors, with the benchmark being subsidiaries operating in scale-intensive industries. - Foreign investment age. This variable is the difference between 2005 (the year when the interviews were conducted) and the year when the subsidiary became a part of the Italian MNE. - Subsidiary size. Subsidiary size is measured using two variables (Taggart, 1997). The first, subsidiary sales, measures the annual sales (millions of EUR) of the subsidiary in 2004. The second, relative size, is the difference between the natural logarithm of the number of employees in the subsidiary and the natural logarithm of the number of employees in the parent company in 2004. - Activity similarity. As previous research indicates, absorptive capacity can play a role in explaining reverse knowledge transfer (Ambos et al., 2006; Gupta and Govindarajan, 2000). In order to capture knowledge familiarity between senders and receivers, the dummy variable activity similarity is used. It equals one when the parent company and foreign subsidiary produce similar products, or are involved in similar research and development activities. - Type of knowledge. Reverse knowledge transfer and related coordination mechanisms may depend on the characteristics of the knowledge transferred. For this reason, the dummy variable codified knowledge is added to the model. This variable captures whether the knowledge transferred from the subsidiary, and used by the parent, has been classified as codified knowledge by the receiver.3 5. Results and discussion In line with the theoretical expectation that implementer subsidiaries rarely possess knowledge relevant to the rest of the MNE and seldom engage in reverse knowledge transfer (Ambos et al., 2006), we find that, compared to innovator (42%) and contributor (47%) subsidiaries, only 20% of implementer subsidiaries were involved in reverse knowledge transfers. In particular, Table 3 shows that about 93% of implementer subsidiaries either have transferred low levels of knowledge to their parent or did not engage in reverse knowledge transfer at all. We acknowledge that this finding does not exclude the possibility that implementer subsidiaries can learn from their local contexts and business relationships and, therefore, can also transfer useful knowledge back to their parent companies. On the other hand, our data indicates that the size of the phenomenon is relative small. 3 To control for the type of knowledge, we are making the assumption that the variable codified knowledge equals zero also if the parent company perceives that there is no transfer of knowledge from the subsidiary (i.e., the level of reverse knowledge transfer equals zero).

106

Table 4 Descriptive statistics: full sample. Mean

S.d.

(1)

(2)

(1) Reverse knowledge 0.61 1.01 transfer (2) Innovator 0.31 0.46 0.17 ⁎⁎⁎ (3) Contributor 0.16 0.37 0.25 ⁎⁎⁎ − 0.29 ⁎⁎⁎ (4) Cultural distance 1.16 0.92 − 0.07 − 0.18 ⁎⁎⁎ (5) Western Europe 0.56 0.50 0.09 0.11 ⁎ (6) North America 0.15 0.36 0.04 0.22 ⁎⁎⁎ (7) High-tech industries 0.41 0.49 0.01 0.14 ⁎⁎ (8) Greenfield 0.32 0.47 0.05 − 0.13 ⁎⁎ (9) FDI age 9.70 6.98 0.00 − 0.05 (10) Relative size − 1.73 1.19 0.14 ⁎⁎ 0.14 ⁎⁎ (11) Subsidiary sales 0.05 0.15 0.10 − 0.03 (12) Activity similarity 0.88 0.32 0.00 − 0.25 ⁎⁎⁎ (13) Codified knowledge 0.92 0.27 − 0.36 ⁎⁎⁎ − 0.17 ⁎⁎⁎ 2.50 0.78 0.06 0.27 ⁎⁎⁎ (14) Subsidiary autonomy a 3.21 1.80 0.29 ⁎⁎⁎ − 0.20 ⁎⁎⁎ (15) Personal coordination a 4.53 2.12 0.20 ⁎⁎⁎ − 0.01 (16) Electronic-based coordination a

(3)

− 0.02 − 0.05 0.02 0.01 0.10 ⁎ 0.05 0.06 0.12 ⁎⁎ 0.13 ⁎⁎ − 0.01 − 0.07

(4)

− 0.36 ⁎⁎⁎ − 0.27 ⁎⁎⁎ − 0.01 0.09 − 0.24 ⁎⁎⁎ − 0.09 − 0.18 ⁎⁎⁎ 0.11 ⁎ 0.13 ⁎⁎ − 0.14 ⁎⁎

(5)

(6)

(7)

(8)

(9)

(10)

(11)

(12)

(13)

(14)

(15)

− 0.48 ⁎⁎⁎ − 0.11 ⁎ − 0.23 ⁎⁎⁎

0.12 ⁎⁎ 0.10 ⁎ 0.09 − 0.06 0.21 ⁎⁎⁎ 0.07 0.25 ⁎⁎⁎ 0.06 0.03 − 0.22 ⁎⁎⁎ − 0.16 ⁎⁎⁎ 0.00 0.02 0.30 ⁎⁎⁎ 0.16 ⁎⁎⁎ − 0.01 −0.18 ⁎⁎⁎ − 0.17 ⁎⁎⁎ − 0.06 − 0.11 ⁎ − 0.07 0.02 0.03 0.03 0.08 − 0.08 − 0.09 0.01 0.15 ⁎⁎ − 0.03 − 0.11 ⁎ − 0.09 0.00 0.19 ⁎⁎⁎ 0.04 0.03 0.13 ⁎⁎ − 0.08 − 0.08

0.10 ⁎

0.14 ⁎⁎

− 0.05

− 0.13 ⁎⁎

0.02

0.04

− 0.04

0.00

− 0.19 ⁎⁎⁎ 0.23 ⁎⁎⁎

0.00 0.21 ⁎⁎⁎

− 0.11 ⁎ 0.08

0.13 ⁎⁎ − 0.10 ⁎

0.10 ⁎ − 0.01

0.09 − 0.32 ⁎⁎⁎ − 0.05 0.26 ⁎⁎⁎ − 0.01 − 0.31 ⁎⁎⁎ − 0.12 ⁎⁎

− 0.01

0.02

0.23 ⁎⁎⁎

a The variable has been centered around its mean value in order to avoid high correlations between interaction terms and interacting variables (Haas and Hansen, 2005; Smith and Sasaki, 1979). The table lists the means and standard deviations of these variables prior to the centering. ⁎ p b .10. ⁎⁎ p b .05. ⁎⁎⁎ p b .01.

L. Rabbiosi / Journal of International Management 17 (2011) 97–113

Variable

L. Rabbiosi / Journal of International Management 17 (2011) 97–113

107

A summary of the descriptive statistics and correlations for all variables in this study is presented in Table 4. No variables exhibit distribution or correlation problems, and the pair correlations are in line with the literature. The correlation of 0.27 between the variables innovator and subsidiary autonomy confirms that subsidiaries with more strategic roles are likely to have high levels of autonomy (Nobel and Birkinshaw, 1998). Interestingly, the correlation between the degree of subsidiary autonomy and the use of personal coordination mechanisms is negative (−0.31). This finding echoes recent research (Ambos and Reitsperger, 2004; Ambos and Schlegelmilch, 2007; Asakawa, 1996, 2001; Nobel and Birkinshaw, 1998). It suggests that because the use of personal coordination mechanisms can be viewed as a tool for subtle control (Edstrom and Galbraith, 1977), parent companies may refrain from establishing overly strong personal coordination systems with more autonomous subsidiaries (i.e., subsidiaries that are more likely to have a mandate to search for new knowledge). However, the pair-wise correlation nature of this finding does not allow us to make concrete conclusions about the entirety of the relationship between personal coordination mechanisms and subsidiary autonomy. Indeed, other research has indicated that parent companies need to move closer to the more autonomous subsidiaries that are expected to search for new knowledge and augment the MNE knowledge base — such as centers of excellence, international creators or strategic leaders (Andersson et al., 2007; Bartlett and Ghoshal, 1989). Ordered probit models are estimated given the ordinal nature of the dependent variable. The results from the econometric estimations are reported in Tables 5–7. The dyadic approach raises an issue of possible non-independence among the observations (Greene, 2000). Therefore, using the Stata's cluster option, a robust variance estimate is obtained that adjusts for within-cluster correlation (Williams, 2000). This controls for the possibility that observations (i.e., dyads) belonging to the same MNE might not be independent. Finally, due to the non-linearity characteristics of ordered probit regression models, the estimated probabilities of the interaction terms depend on the values of the other covariates, making their interpretation difficult (Norton et al., 2004). Accordingly, an estimation of the interaction effects is undertaken using OLS, although OLS is an inefficient estimator in this case (Greene, 2000). The results of the OLS models, in which the interaction terms personal coordination × subsidiary autonomy and electronic-based coordination × subsidiary autonomy have been added to the econometric specifications, are presented in the Appendix. Hoetker's (2007) recommendation was followed to test the differences among subsidiaries roles. Specifically, when the moderating variable is a dummy – as are the variables capturing subsidiary roles – the cross-group differences can be analyzed by splitting the sample and examining the effects in two different groups. Therefore, the appropriate way to test the hypotheses is to split the sample into the relevant sub-samples of innovator subsidiaries and contributor subsidiaries. The first model in each series (Models 1, 4 and 7 in Tables 5–7, respectively) shows the hypotheses tested on the full sample, which includes subsidiaries that are innovators, contributors or implementers. The second model in each series (Models 2, 5 and 8 in Tables 5–7, respectively) illustrates the hypotheses tested on innovator subsidiaries. Finally, the third model in each series (Models 3, 6 and 9 in Tables 5–7, respectively) shows the test of the hypotheses on contributor subsidiaries. In Model 1, the variable personal coordination shows a positive and significant coefficient (p b 0.01), indicating that coordination through personnel movement supports a higher level of reverse knowledge transfer. Extensive use of face-to-face coordination mechanisms maximizes the level of interaction among parent and subsidiary employees and offers the most opportunities to transfer knowledge through the movement of individuals with special expertise. Extensive use of electronic-based coordination

Table 5 Impact of coordination mechanism on the level of reverse knowledge transfer.

Innovator Contributor Cultural distance Western Europe North America High-tech industries Greenfield FDI age Relative size Subsidiary sales Activity similarity Codified knowledge Subsidiary autonomy Personal coordination Electronic-based coordination Log-pseudolikelihood Wald χ2 McFadden's Adjusted Pseudo-R2 Observations

Full sample

Innovator

Contributor

Model 1

Model 2

Model 3

0.13 (0.23) 1.29 (0.69) ⁎ 1.09 (0.85) 0.47 (0.39) 0.23 (0.43) − 0.01 (0.03) 0.09 (0.12) 0.50 (0.40) 0.00 (0.39) − 1.22 (0.35) ⁎⁎⁎ 0.07 (0.14) 0.27 (0.07) ⁎⁎⁎

0.01 (0.28) 0.31 (0.52) − 0.38 (0.79) − 0.84 (0.87) 0.24 (0.41) − 0.02 (0.03) 0.08 (0.23) − 3.37 (1.50) ⁎⁎ –a − 0.08 (0.54) − 0.81 (0.55) − 0.06 (0.25) 0.46 (0.20) ⁎⁎

0.96 (0.27) ⁎⁎⁎ 1.22 (0.47) ⁎⁎⁎ 0.14 (0.14) 0.64 (0.34) ⁎ 0.48 (0.43) − 0.04 (0.30) 0.52 (0.22) ⁎⁎ 0.00 (0.02) 0.16 (0.07) ⁎⁎ − 0.34 (0.62) 0.20 (0.34) − 1.39 (0.22) ⁎⁎⁎ 0.17 (0.15) 0.28 (0.09) ⁎⁎⁎ 0.12 (0.07) ⁎ − 205.95 158.57 ⁎⁎⁎ 0.22 280

− 0.02 (0.09) − 81.04 54.78 ⁎⁎⁎ 0.16 86

Ordered probit regressions. In brackets are robust standard errors corrected for heteroscedasticity and cluster-correlated data. a Dropped due to collinearity. ⁎ p b .10. ⁎⁎ p b .05. ⁎⁎⁎ p b .01 (two-tailed tests applied).

− 41.81 31.36 ⁎⁎⁎ 0.17 44

108

L. Rabbiosi / Journal of International Management 17 (2011) 97–113

Table 6 Joint impact of degree of autonomy and personal coordination mechanisms on the level of reverse knowledge transfer.

Innovator Contributor Cultural distance Western Europe North America High-tech industries Greenfield FDI age Relative size Subsidiary sales Activity similarity Codified knowledge Subsidiary autonomy Personal coordination Electronic-based coordination Personal coordination × subsidiary autonomy Log-pseudolikelihood Wald χ2 McFadden's Adjusted Pseudo-R2 Observations

Full sample

Innovator

Contributor

Model 4

Model 5

Model 6

0.14 (0.24) 1.29 (0.69) ⁎ 1.10 (0.86) 0.47 (0.39) 0.24 (0.44) − 0.01 (0.03) 0.10 (0.13) 0.49 (0.41) − 0.01 (0.41) − 1.22 (0.35) ⁎⁎⁎ 0.07 (0.14) 0.26 (0.07) ⁎⁎⁎

0.19 (0.31) 1.26 (0.78) 2.03 (0.99) ⁎⁎ − 0.70 (0.73) − 0.12 (0.46) − 0.03 (0.04) 0.18 (0.28) 0.13 (2.18) −a 0.58 (0.78) − 6.19 (4.03) 1.41 (0.48) ⁎⁎⁎ 1.69 (0.87) ⁎ 5.13 (1.84) ⁎⁎⁎

0.93 (0.27) ⁎⁎⁎ 1.19 (0.46) ⁎⁎⁎ 0.17 (0.14) 0.68 (0.34) ⁎⁎ 0.56 (0.42) − 0.09 (0.29) 0.57 (0.23) ⁎⁎ 0.00 (0.01) 0.19 (0.08) ⁎⁎ − 0.41 (0.63) − 0.01 (0.40) − 1.46 (0.21) ⁎⁎⁎ 0.28 (0.16) ⁎ 0.29 (0.09) ⁎⁎⁎ 0.12 (0.07) ⁎ 0.19 (0.10) ⁎ − 202.80 168.08 ⁎⁎⁎ 0.23 280

− 0.02 (0.09) 0.01 (0.08) − 81.04 54.78 ⁎⁎⁎ 0.16 86

− 23.88 51.62 ⁎⁎⁎ 0.53 b 44

Ordered probit regressions. In brackets are robust standard errors corrected for heteroscedasticity and cluster-correlated data. a Dropped due to collinearity. b Estimation corrected for heteroscedasticity; cluster-correlated data correction not applicable. ⁎ p b .10. ⁎⁎ p b .05. ⁎⁎⁎ p b .01 (two-tailed tests applied).

mechanisms also enhances the levels of reverse knowledge transfer, as indicated by the coefficient of electronic-based coordination, which is positive and significant at p b 0.1. These results support the baseline Hypotheses (1a and 1b) and they are in line with the previous literature on reverse knowledge transfer that points to both types of coordination mechanisms as effective channels for transferring knowledge. Notably, the estimations also reveal that contributor and innovator subsidiaries are more likely to transfer knowledge to their parent companies than implementer subsidiaries (the benchmark), which is in line with the existing theory. Table 7 Joint impact of degree of autonomy and electronic-based coordination mechanisms on the level of reverse knowledge transfer.

Innovator Contributor Cultural distance Western Europe North America High-tech industries Greenfield FDI age Relative size Subsidiary sales Activity similarity Codified knowledge Subsidiary autonomy Personal coordination Electronic-based coordination Electronic-based × subsidiary autonomy Log-pseudolikelihood Wald χ2 McFadden's Adjusted Pseudo-R2 Observations

Full sample

Innovator

Contributor

Model 7

Model 8

Model 9

0.14 (0.24) 1.30 (0.70) ⁎ 1.02 (0.88) 0.57 (0.39) 0.23 (0.43) − 0.01 (0.03) 0.08 (0.13) 0.61 (0.40) 0.04 (0.43) − 1.13 (0.37) ⁎⁎⁎ 0.02 (0.14) 0.24 (0.07) ⁎⁎⁎ 0.00 (0.08) − 0.13 (0.08) ⁎

0.04 (0.29) 0.29 (0.52) − 0.38 (0.79) − 0.90 (0.91) 0.24 (0.42) − 0.01 (0.04) 0.09 (0.23) − 3.18 (1.56) ⁎⁎ −a − 0.01 (0.55) − 0.90 (0.48) ⁎ − 0.06 (0.25) 0.49 (0.23) ⁎⁎

0.93 (0.27) ⁎⁎⁎ 1.26 (0.48) ⁎⁎⁎ 0.12 (0.15) 0.64 (0.35) ⁎ 0.44 (0.44) − 0.02 (0.30) 0.49 (0.22) ⁎⁎ 0.00 (0.02) 0.14 (0.07) ⁎ − 0.27 (0.60) 0.23 (0.36) − 1.34 (0.22) ⁎⁎⁎ 0.16 (0.15) 0.27 (0.09) ⁎⁎⁎ 0.13 (0.06) ⁎⁎ − 0.11 (0.07) − 204.17 149.41 ⁎⁎⁎ 0.23 280

− 79.37 71.25 ⁎⁎⁎ 0.17 86

Ordered probit regressions. In brackets are robust standard errors corrected for heteroscedasticity and cluster-correlated data. a Dropped due to collinearity. ⁎ p b .10. ⁎⁎ p b .05. ⁎⁎⁎ p b .01 (two-tailed tests applied).

0.15 (0.28) − 41.63 41.92 ⁎⁎⁎ 0.17 44

L. Rabbiosi / Journal of International Management 17 (2011) 97–113

109

The coefficients of the variables innovator and contributor are positive and statistically significant. With respect to the other control variables, as suggested by the positive and significant coefficient of the variable relative size, subsidiaries bigger than their parent companies are likely to be involved in higher levels of reverse knowledge transfer. Subsidiaries located in Western Europe are more likely to have higher levels of reverse knowledge transfer. The same is true for subsidiaries that have been established through greenfield investments. Finally, as suggested by the negative and significant coefficient of the variable codified knowledge, our data indicate that the parent's perception of the level of reverse knowledge transfer is lower when codified knowledge is involved. The estimates in Models 2 and 3 confirm Hypotheses 2a and 2b. Specifically, the coefficient of personal coordination is positive and significant (p b 0.01) in Model 2 but it is not significant in Model 3. Conversely, the coefficient of electronic-based coordination is positive and significant (p b 0.05) in Model 3 but it is not statistically significant on any conventional level in Model 2. As subsidiary knowledge is expected to differ depending on the subsidiary role, the level of reverse knowledge transfer will benefit from parent companies implementing coordination mechanisms that are appropriate for the subsidiary's actual role. Accordingly, the impact of personal coordination mechanisms on the level of reverse knowledge transfer is stronger for innovator subsidiaries, while the impact of electronic-based coordination mechanisms is stronger for contributor subsidiaries. Considering personal coordination mechanisms, we know – as the literature suggests – that they are costly to maintain (Daft and Lengel, 1986; Pedersen et al., 2003) as a result of travel costs, time limits, and differences in cultures and languages. Therefore, their extensive use should be justified. For example, the effective management of knowledge transfers from innovator subsidiaries, which are expected to be more innovative and to own context-specific knowledge (Cantwell and Mudambi, 2005; Gupta and Govindarajan, 1991), requires the use of richer communication mechanisms (Mudambi et al., 2007). Parent companies would be willing to “pay” because of the expected positive influence on the costs and the benefit trade-off arising from such transfer processes. At the same time, managers face several difficulties in transferring context-specific and tacit knowledge across countries through electronic-based coordination mechanisms (Pedersen et al., 2003). Therefore, significant investments in information technologies to support this type of communication may not be justified if the people-to-documents process cannot be pursued in an economical way or if too much knowledge will be lost during the transfer (e.g., from innovator subsidiaries). However, this codification process should be efficient in the case of contributor subsidiaries. Of the control variables, location effects are relevant for innovator subsidiaries. Specifically, innovator subsidiaries located in more developed countries, such as those in Western Europe, are more likely to transfer knowledge to their parent companies. Also the coefficient of the variable codified knowledge is significant (p b 0.01) and the negative sign indicates that reverse knowledge transfer from innovator subsidiaries is less likely to involve codified knowledge. It is interesting to observe that the same variable is not significant in the estimation for the contributor subsidiary sub-sample. This result is in line with our expectation that there is a certain overlap between knowledge of contributor subsidiaries and knowledge of their parent companies; such overlapping is expected to facilitate codification strategies and knowledge transfer through electronic media. Turning to Hypotheses 3 and 4, the coefficient of the interaction term personal coordination × subsidiary autonomy is not significant in the case of the innovator subsidiaries sub-sample (Model 5), and it is positive and significant (p b 0.01) in the subsample of contributor subsidiaries (Model 6).4 These findings are in line with our theoretical predictions and therefore support Hypothesis 3. The coefficient of the interaction term electronic-based coordination × subsidiary autonomy is negative and significant at p b 0.1 in the estimation for the sub-sample of innovator subsidiaries (Model 8) and it is not significant in the case of the contributor subsidiaries sub-sample (Model 9). These results support Hypothesis 4. Driven by their own initiative or by decisions made by the parent company's managers, subsidiaries can change their roles within the MNE (Birkinshaw and Hood, 1998; Birkinshaw et al., 1998; Frost et al., 2002). When this occurs, one is likely to observe, for instance, a contributor subsidiary with a relative higher degree of autonomy than its peers; an innovator subsidiary in which electronic-based coordination mechanisms are used more than the average; an innovator subsidiary with a lower degree of autonomy than to its peers; etc. These results signal that in these “situations of disequilibrium”, modifications of the coordination strategy will affect the amount of reverse knowledge transfer. Specifically, a positive effect will be evident if more personal coordination is used with contributor subsidiaries that have higher degrees of autonomy. In other words, if the degree of autonomy of a contributor subsidiary is above the average level of autonomy, an increase in the use of personal coordination mechanisms will have a positive marginal benefit on the level of reverse knowledge transfer, while the effect will be negative if the degree of autonomy of a contributor subsidiary is below the average level of autonomy. This result provides evidence of the bargaining process between parent companies and subsidiaries highlighted in the literature (e.g., Andersson et al., 2007; Asakawa, 2001; Mudambi and Navarra, 2004). Higher degrees of autonomy have been linked to a subsidiary's better access to resources and to greater knowledge creation in the subsidiary (Andersson et al., 2002; Gupta and Govindarajan, 1991; Venaik et al., 2005), and access to critical resources has been viewed as an important source of subsidiary power (Andersson et al., 2007). Under these conditions, the tension between the parent company and the focal subsidiary increases (Asakawa, 2001), and a new bargaining process arises between the two units (Mudambi and Navarra, 2004). The erosion of the parent company's control will be mitigated through the use of more personal coordination mechanisms, which ultimately has a positive effect on the level of reverse knowledge transfer. The negative effect of the interaction of electronic-based coordination mechanisms with subsidiary autonomy can be said to provide support for the information processing theory advanced by Egelhoff (1988). The theory suggests that “information 4 As we have subtracted the means of the variables before creating the interaction term, the main effects in Models 4–9 can be interpreted as the effect of a variable at the average observed score of the other variable (Finney et al., 1984).

110

L. Rabbiosi / Journal of International Management 17 (2011) 97–113

processing” works better when centralization dominates and that electronic media may be considered as carriers of “information”. Centralization helps to reduce costs associated with redundancies, incompatibility, and a lack of common understanding and standards, costs that are expected to decrease the effectiveness of electronic-based coordination mechanisms. In addition, when subsidiary autonomy is high, differences between parent and subsidiary “knowledge structures”, which may include knowledge domain, terminology, interpretation of phenomena and social context, are expected to increase, giving rise to misunderstandings between senders and receivers that limit codification strategies and the impact of electronic coordination media on the level of reverse knowledge transfer. This negative effect is apparent in innovator subsidiaries where more emphasis on electronic-based coordination mechanisms can be perceived as a parent company's interference in how things are done, and in the shifting of the subsidiary's time and attention toward codification processes. 6. Conclusion This paper seeks to contribute to the understanding of the impact of coordination strategies between parent companies and their subsidiaries on the levels of reverse knowledge transfer. The key contribution to the strategic management literature is the identification and examination of the direct and the indirect effects of subsidiary autonomy on reverse knowledge transfer, and the effects of personal and electronic-based coordination mechanisms on reverse knowledge transfer. Moreover, given the relevance of subsidiary roles, the analysis is developed in a conditional sense — it is conditional on the actual responsibilities of the subsidiaries. Both subsidiary autonomy and different communication mechanisms have been found to play a role in knowledge transfer processes within MNEs, but they have typically been treated as independent dimensions. The research presented here shows that the effects of different coordination strategies on reverse knowledge transfers differ systematically for subsidiaries with different roles. These findings have potentially important managerial implications. The analyses suggest that coordination mechanisms within the parent company-foreign subsidiary dyadic relationship can be designed to influence the amount of reverse knowledge transfer. We should expect changes in the coordination mechanisms to be driven by both parent companies and their foreign subsidiaries. For instance, when engagement in reverse knowledge transfer plays a crucial role in defining the importance of a subsidiary within the MNE, the subsidiary will have an interest in influencing the design of the coordination mechanisms in a way so that reverse knowledge transfer can be maximized. In line with previous findings from the intra-MNE context, the empirical analyses show that both personal and electronic-based coordination mechanisms can have a positive effect on the extent of reverse knowledge transfer. However, the choice of mechanisms and the decision on how to combine them cannot be independent from the actual role and autonomy level of the subsidiary. Our findings indicate that reverse knowledge transfer can be effectively increased by using more personal coordination mechanisms with innovator subsidiaries, while the use of additional personal coordination mechanisms does not influence the levels of reverse knowledge transfer for contributor subsidiaries. However, when contributor subsidiaries have a degree of autonomy that differs from the average (because they gain or formally receive it), a modification of the coordination strategy can affect the level of reverse knowledge transfer. For instance, it would be beneficial to increase the use of personal coordination mechanisms with contributor subsidiaries that have higher degrees of autonomy, while a similar action would have little effect when used with innovator subsidiaries with relatively high degrees of autonomy. In contrast, electronic-based coordination mechanisms are more closely linked to reverse knowledge transfer when they are used with contributor subsidiaries. Information technology provides new electronic-based means of coordination that enable MNEs to achieve reverse knowledge transfer while avoiding the high cost of personal coordination. However, a greater use of electronic coordination can effectively enhance reverse knowledge transfer only when used with contributor subsidiaries or with innovator subsidiaries that are evolving, such as innovator subsidiaries in which the degree of autonomy is below average. It is effective to use more electronic-based coordination mechanisms because the subsidiary role has a strong influence on the characteristics of the knowledge that can be transferred. Thus, our findings support the general consensus that both personal and electronic-based coordination mechanisms are appropriate in enhancing intra-MNE knowledge transfer. However, our evidence also contributes to the discussion of under which conditions electronic-based coordination or face-to-face interaction is the most effective mechanism (e.g. Ambos and Ambos, 2009; Noorderhaven and Harzing, 2009). The limitations of this study should also be noted. The context of the study – reverse knowledge transfer within Italian MNEs – was specific, which naturally imposes limits on the generalizability of these results to other national samples. Nevertheless, the findings suggest that examinations of the interdependence among coordination dimensions that could influence the transfer of knowledge from foreign subsidiaries to their parent companies are worthwhile. Furthermore, because supplier-dominated industries (such as many parts of the textile, furniture and paper industries) are not represented in our sample, these findings cannot be extended to the entire group of manufacturing industries. In addition, as perceptual data from a single respondent on the parent company level is used, there may still be some common-method bias, despite the validity checks on the subsidiary level and the inclusion of procedural remedies in the survey design. In terms of the measures, the following limitations apply. First, the dependent variable only captures the level of reverse knowledge transfer — it is not able to reflect the value created by the transfer at the receiving unit. Therefore, there is an implicit assumption that all knowledge transfers, regardless of how they occur, are equally valuable. Testing of the hypotheses in a way that takes the benefits of such transfers into account could highlight further relevant theoretical and managerial implications. Moreover, our dependent variable does not measure the objective level of reverse knowledge transfer but captures the parent company's perception of the extent to which subsidiary knowledge has been used by the parent company. This perceived amount

L. Rabbiosi / Journal of International Management 17 (2011) 97–113

111

of reverse knowledge transfer can also be inflated by the extent of subsidiary knowledge that has been acquired but not used by the parent company. Second, this study operationalizes personal coordination as the use of international teamwork, the transfer of managers, and the transfer of scientists and technical employees. Other studies have used different combinations that include such mechanisms as international training programs, informal communications and liaison personnel. The measure used in this paper could be improved by the addition of other forms of socialization coordination. Despite this drawback, the variable used here captures a significant aspect of coordination mechanisms — the ability to create a verbal information network (Edstrom and Galbraith, 1977). Similar shortcomings hold for the operationalization of electronic-based coordination mechanisms. For example, research in information systems views electronic communities of practice as important coordination mechanisms for transferring and sharing knowledge. Moreover, our distinction of personal and electronic-based coordination is rather coarse. Other more finegrained dimensions of classification that are of equal significance and relevance can be developed and used to provide more insights about the appropriate mechanisms for reverse knowledge flows when subsidiaries with different degrees of autonomy are involved. The use of cross-sectional data means that only the association between the relevant variables could be tested. Such data fails to directly capture dynamic relationships in terms of how the effect of coordination strategies on reverse knowledge transfer changes as knowledge varies and subsidiaries evolve over time. Finally, this study analyzes a specific case of reverse knowledge transfer from a foreign subsidiary to a parent company. However, a more comprehensive understanding of the hypothesized relationships could be gained by analyzing all of the possible reverse knowledge transfer processes among different entities of an MNE, such as knowledge transfer among peer subsidiaries. Acknowledgements I thank Ulf Andersson, Nicolai Foss, Keld Laursen and Grazia D. Santangelo for their comments and suggestions on a prior version of this paper. I also thank the participants of the 2008 DRUID Conference and the Academy of Management Annual Meeting. I am grateful to three anonymous reviewers for their helpful suggestions. Appendix A

Table 1a Results of OLS regressions for the level reverse knowledge transfer. Full sample

Constant Innovator Contributor Cultural distance Western Europe North America High-tech industries Greenfield FDI age Relative size Subsidiary sales Activity similarity Codified knowledge Subsidiary autonomy Personal coordination Electronic-based coordination Personal coordination × Subsidiary autonomy Electronic-based × Subsidiary autonomy Observations R-squared

Innovator subsidiaries

Contributor subsidiaries

Model a1

Model a2

Model a3

Model a4

Model a5

Model a6

0.11 (0.37) 0.44 (0.17) ⁎⁎ 0.72 (0.40) ⁎ 0.03 (0.08) 0.24 (0.20) 0.05 (0.26) 0.02 (0.17) 0.22 (0.14) 0.00 (0.01) 0.04 (0.05) − 0.11 (0.47) − 0.05 (0.23) − 1.14 (0.16) ⁎⁎⁎ 0.18 (0.11) ⁎ 0.16 (0.06) ⁎⁎⁎ 0.05 (0.03) ⁎

0.22 (0.35) 0.46 (0.16) ⁎⁎⁎ 0.75 (0.41) ⁎ 0.02 (0.08) 0.25 (0.20) 0.04 (0.27) 0.04 (0.17) 0.22 (0.13) 0.00 (0.01) 0.03 (0.05) − 0.08 (0.46) 0.01 (0.20) − 1.10 (0.16) ⁎⁎⁎

0.27 (0.76)

0.10 (0.80)

0.31 (1.20)

1.67 (1.46)

0.08 (0.19) 0.95 (0.39) ⁎⁎ 0.78 (0.53) 0.43 (0.30) 0.15 (0.31) − 0.01 (0.03) 0.07 (0.11) 0.48 (0.38) − 0.04 (0.35) − 1.06 (0.33) ⁎⁎⁎ 0.02 (0.12) 0.23 (0.06) ⁎⁎⁎

0.11 (0.19) 1.01 (0.40) ⁎⁎ 0.79 (0.56) 0.49 (0.30) 0.14 (0.31) − 0.01 (0.03) 0.06 (0.11) 0.59 (0.36) − 0.05 (0.35) − 0.94 (0.34) ⁎⁎⁎ − 0.01 (0.11) 0.19 (0.06) ⁎⁎⁎

− 0.03 (0.08) − 0.01 (0.06)

− 0.01 (0.07)

0.15 (0.27) 0.23 (0.44) 0.15 (0.67) − 1.04 (0.60) 0.32 (0.39) − 0.01 (0.03) 0.08 (0.22) − 2.08 (1.36) −a − 0.28 (0.71) 0.10 (0.44) 0.08 (0.21) 0.23 (0.16) 0.75 (0.24) ⁎⁎⁎

0.08 (0.31) 0.29 (0.51) − 0.25 (0.80) − 0.88 (1.01) 0.22 (0.45) 0.00 (0.04) 0.09 (0.26) − 2.79 (1.65) −a − 0.03 (0.62) − 0.68 (0.45) − 0.05 (0.27) 0.37 (0.21) ⁎

0.10 (0.08) 0.16 (0.06) ⁎⁎⁎ 0.04 (0.03)

0.07 (0.05)

280 0.35

− 0.05 (0.03) 280 0.35

86 0.32

− 0.10 (0.06) 86 0.35

44 0.52

0.13 (0.27) 44 0.31

In brackets are robust standard errors corrected for heteroscedasticity and cluster-correlated data. a Dropped due to collinearity. ⁎ p b .10. ⁎⁎ p b .05. ⁎⁎⁎ p b .01 (two-tailed tests applied).

References Alavi, M., Leidner, D.E., 2001. Knowledge management and knowledge management systems: conceptual foundations and research issues. MIS Quarterly 25, 107–136.

112

L. Rabbiosi / Journal of International Management 17 (2011) 97–113

Almeida, P., Song, J., Grant, R.M., 2002. Are firms superior to alliances and markets? An empirical test of cross-border knowledge building. Organization Science 13, 147–161. Ambos, T.C., Ambos, B., 2009. The impact of distance on knowledge transfer effectiveness in multinational corporations. Journal of International Management 15, 1–14. Ambos, B., Reitsperger, W.D., 2004. Offshore center of excellence: social control and success. Management International Review 44, 51–65. Ambos, B., Schlegelmilch, B.B., 2007. Innovation and control in the multinational firm: a comparison of political and contingency approaches. Strategic Management Journal 28, 473–486. Ambos, T.C., Ambos, B., Schlegelmilch, B.B., 2006. Learning from foreign subsidiaries: an empirical investigation of headquarters' benefit from reverse knowledge transfers. International Business Review 15, 294–312. Andersen, T.J., Foss, N.J., 2005. Strategic opportunity and economic performance in multinational enterprises: the role and effects of information and communication technology. Journal of International Management 11, 293–310. Andersson, U., Forsgren, M., 2000. In search of centre of excellence: network embeddedness and subsidiary roles in multinational corporations. Management International Review 40, 329–350. Andersson, U., Forsgren, M., Holm, U., 2002. The strategic impact of external networks: subsidiary performance and competence development in the multinational corporation. Strategic Management Journal 23, 979–996. Andersson, U., Forsgren, M., Holm, U., 2007. Balancing subsidiary influence in the federative MNC: a business network view. Journal of International Business Studies 38, 802–818. Asakawa, K., 1996. External-internal linkages and overseas autonomy-control tension: the management dilemma of the Japanese R&D in Europe. IEEE Transactions on Engineering Management 42, 24–32. Asakawa, K., 2001. Organizational tension in international R&D management: the case of Japanese firms. Research Policy 30, 735–757. Bartlett, C.A., Ghoshal, S., 1989. Managing Across Borders: The Transnational Solution. Harvard Business School Press, Boston. Birkinshaw, J., Hood, N., 1998. Multinational subsidiary evolution: capability and charter change in foreign-owned subsidiary companies. Academy of Management Review 23, 773–795. Birkinshaw, J., Hood, N., Jonsson, S., 1998. Building firm-specific advantages in multinational corporations: the role of subsidiary initiative. Strategic Management Journal 19, 221–241. Birkinshaw, J., Nobel, R., Ridderstrale, J., 2002. Knowledge as a contingency variable: do the characteristics of knowledge predict organization structure? Organization Science 13, 274–289. Björkman, I., Barner-Rasmussen, W., Li, L., 2004. Managing knowledge transfer in MNCs: the impact of headquarters control mechanisms. Journal of International Business Studies 35, 443–455. Bouquet, C., Birkinshaw, J., 2008. Weight versus voice: how foreign subsidiaries gain attention from corporate headquarters. Academy of Management Journal 51, 577–601. Brett, J.F., Cron, W.L., Slocum, J.W.J., 1995. Economic dependency on work: a moderator of the relationship between organizational commitment and performance. Academy of Management Journal 38, 261–271. Brockhoff, K.K., Schmaul, B., 1996. Organization, autonomy, and success of internationally dispersed R&D facilities. IEEE Transactions on Engineering Management 43, 33–40. Cantwell, J.A., 1987. The reorganisation of European industries after integration: selected evidence on the role of transnational enterprise activities. Journal of Common Market Studies 26, 127–151. Cantwell, J.A., 1995. The globalisation of technology: what remains of the product cycle model? Cambridge Journal of Economics 19, 155–174. Cantwell, J.A., Mudambi, R., 2005. MNE competence-creating subsidiary mandates. Strategic Management Journal 26, 1109–1128. Criscuolo, P., Narula, R., 2007. Using multi-hub structures for international R&D: organisational inertia and the challenges of implementation. Management International Review 47, 639–660. Daft, R.L., Lengel, R.H., 1986. Organizational information requirements, media richness and structural design. Management Science 32, 554–571. DeSanctis, G., Fulk, J., 1999. Shaping Organization form: Communication, Connection, and Community. SAGE Publications, Inc., Newbury Park, CA. Downey, K.H., Hellriegel, D., Slocum, J.W.J., 1975. Environmental uncertainty: the construct and its application. Administrative Science Quarterly 20, 613–629. Eden, L., 2009. Letter from editor-in-chief: reverse knowledge transfers, culture clashes and going international. Journal of International Business Studies 40, 177–180. Edstrom, A., Galbraith, J.R., 1977. Transfer of managers as a coordination and control strategy in multinational organizations. Administrative Science Quarterly 22, 248–263. Egelhoff, W.G., 1988. Organizing the Multinational Enterprise: An Information-Processing Perspective. Ballinger, Cambridge, MA. Finney, J.W., Mitchell, R.E., Cronkite, R.C., Moos, R.H., 1984. Methodological issues in estimating main and interactive effects: examples from coping/social support and stress field. Journal of Health and Social Behavior 25, 85–98. Forsgren, M., Johanson, J., Sharma, D., 2000. Development of MNC Centres of Excellence. In: Holm, U., Pedersen, T. (Eds.), The emergence and impact of MNC centres of excellence. MacMillan, Basingstoke, pp. 45–67. Foss, N.J., Pedersen, T., 2002. Transferring knowledge in MNCs: the role of sources of subsidiary knowledge and organization context. Journal of International Management 8, 49–67. Frost, T., 1998. The geographic sources of innovation in the multinational enterprise: U.S. subsidiaries and host country spillovers 1980–1990, Massachusetts Institute of Technology, Unpublished Ph.D. Dissertation. Frost, T.S., Zhou, C., 2005. R&D co-practice and ‘reverse’ knowledge integration in multinational firms. Journal of International Business Studies 36, 676–687. Frost, T.S., Birkinshaw, J.M., Ensign, P.C., 2002. Centers of excellence in multinational corporations. Strategic Management Journal 23, 997–1018. Fulk, J., DeSanctis, G., 1995. Electronic communication and changing organizational forms. Organization Science 6, 337–349. Gammelgaard, J., Holm, U., Pedersen, T., 2004. The Dilemmas of MNC Subsidiary Knowledge Transfer. In: Mahnke, V., Pedersen, T. (Eds.), Knowledge flows, governance and the multinational enterprise. Palgrave Macmillan, New York. Ghoshal, S., 1986. The innovative multinational: a differentiated network of organizational roles and management processes, Harvard Business School, Unpublished Ph.D. Dissertation. Ghoshal, S., Bartlett, C.A., 1988. Creation, adoption, and diffusion of innovations by subsidiaries of multinational corporations. Journal of International Business Studies 19, 365–388. Ghoshal, S., Nohria, N., 1989. Internal differentiation within the multinational corporation. Strategic Management Journal 10, 323–337. Ghoshal, S., Korine, H., Szulanski, G., 1994. Interunit communication in multinational corporations. Management Science 40, 96–110. Greene, W.H., 2000. Econometric Analysis. Prentice-Hall, Inc., Upple Saddle River, New Jersey. Gupta, A.K., Govindarajan, V., 1991. Knowledge flows and the structure of control within multinational corporations. Academy of Management Review 16, 768–792. Gupta, A.K., Govindarajan, V., 2000. Knowledge flows within multinational corporations. Strategic Management Journal 21, 473–496. Haas, M.R., Hansen, M.T., 2005. When using knowledge can hurt performance: the value of organizational capabilities in a management consulting company. Strategic Management Journal 26, 1–24. Håkanson, L., Nobel, R., 2001. Organization characteristics and reverse technology transfer. Management International Review 41, 392–420. Hansen, M.T., 1999. The search-transfer problem: the role of weak ties in sharing knowledge across organisational subunits. Administrative Science Quarterly 44, 82–111. Hansen, M.T., Nohria, N., Tierney, T., 1999. What's your strategy for managing knowledge? Harvard Business Review 77, 106–116. Harman, H.H., 1967. Modern Factor Analysis. University of Chicago Press, Chicago, IL. Harzing, A.-W., Noorderhaven, N., 2006. Knowledge flows in MNCs: an empirical test and extension of Gupta and Govindarajan's typology of subsidiary roles. International Business Review 15, 195–214. Heckman, J., 1979. Sample selection bias as a specification error. Econometrica 47, 153–161.

L. Rabbiosi / Journal of International Management 17 (2011) 97–113

113

Hoetker, G., 2007. The use of logit and probit models in strategic management research: critical issues. Strategic Management Journal 28, 331–343. Holm, U., Pedersen, T., 2000. The Emergence and Impact of MNC Centres of Excellence. MacMillan, Basingstoke. Jarvenpaa, S.L., Ives, B., 1993. Organizing for global competition: the fit of information technology. Decision Sciences 24, 547–580. Kane, G.C., Alavi, M., 2007. Information technology and organizational learning: an investigation of exploration and exploitation processes. Organization Science 18, 796–812. Karimi, J., Konsynski, B.R., 1991. Globalization and information management strategies. Journal of Management Information System 7, 7–26. Kogut, B., Singh, H., 1988. The effect of national culture on the choice of entry mode. Journal of International Business Studies 19, 411–432. Kogut, B., Zander, U., 1993. Knowledge of the firm and the evolutionary-theory of the multinational-corporation. Journal of International Business Studies 24, 625–645. Kostova, T., 1999. Transnational transfer of strategic organisational practices: a contextual perspective. Academy of Management Review 24, 308–324. Kuemmerle, W.T.J., 1999. The drivers of foreign direct investment into research and development: an empirical investigation. Journal of International Business Studies 30, 1–24. Lane, P.J., Salk, J.E., Lyles, M.A., 2001. Absorptive capacity, learning, and performance in international joint ventures. Strategic Management Journal 22, 1139–1161. Lord, M.D., Ranft, A.L., 2000. Organizational learning about new international markets: exploring the internal transfer of local market knowledge. Journal of International Business Studies 31, 573–589. Makino, S., Delios, A., 1996. Local knowledge transfer and performance implications for alliance formation in Asia. Journal of International Business Studies 27, 905–927. Mariotti, S., Mutinelli, M., 2005. Italia multinazionale 2004. Le partecipazioni italiane all'estero e estere in Italia, Rubbettino Editore, Soveria Mannelli. Martinez, J.I., Jarillo, C.J., 1989. The evolution of research on co-ordination mechanisms in multinational corporations. Journal of International Business Studies 3, 489–514. Medcof, J.W., 2001. Resource-based strategy and managerial power in networks of internationally dispersed technology units. Strategic Management Journal 22, 999–1012. Minbaeva, D.B., Pedersen, T., Björkman, I., Fey, C., Park, H., 2003. MNC knowledge transfer, subsidiary absorptive capacity, and HRM. Journal of International Business Studies 34, 586–599. Mudambi, R., Navarra, P., 2004. Is knowledge power? Knowledge flows, subsidiary power and rent-seeking within MNCs. Journal of International Business Studies 35, 385–406. Mudambi, R., Piscitello, L., Rabbiosi, L., 2007. Mandates and Mechanisms: Reverse Knowledge Transfer in MNEs. Paper presented at the Academy of Management Annual Meeting, August 3–8, Philadelphia, USA. Niederman, F., 2005. International business and MIS approaches to multinational organizational research: the cases of knowledge transfer and IT workforce outsourcing. Journal of International Management 11, 187–200. Nobel, R., Birkinshaw, J., 1998. Innovation in multinational corporations: control and communication patterns in international R&D operations. Strategic Management Journal 19, 479–496. Nohria, N., Ghoshal, S., 1994. Differentiated fit and shared values: alternatives for managing headquarters-subsidiary relations. Strategic Management Journal 15, 491–502. Nohria, N., Ghoshal, S., 1997. The Differentiated Network: Organizing Multinational Corporations for Value Creation. Jossey-Bass Publishers, San Francisco, CA. Noorderhaven, N., Harzing, A.-W., 2009. Knowledge-sharing and social interaction within MNEs. Journal of International Business Studies 40, 719–741. Norton, E.C., Wang, H., Ai, C., 2004. Computing interaction effects and standard errors in logit and probit models. The Stata Journal 4, 154–167. Paterson, S.L., Brock, D.M., 2002. The development of subsidiary-management research: review and theoretical analysis. International Business Review 11, 139–163. Pavitt, K.L.R., 1984. Sectoral patterns of technical change: towards a taxonomy and a theory. Research Policy 13, 343–373. Pavitt, K.L.R., 1990. What we know about the strategic management of technology. California Management Review 32, 17–26. Pearce, R.D., 1999. The evolution of technology in multinational enterprises: the role of creative subsidiaries. International Business Review 8, 125–148. Pearce, R., Papanastassiou, M., 1999. Overseas R&D and the strategic evolution of MNEs: evidence from laboratories in the UK. Research Policy 28, 23–41. Pedersen, T., Petersen, B., Sharma, D.D., 2003. Knowledge transfer performance of multinational companies. Management International Review 43, 69–90. Persaud, A., 2005. Enhancing synergistic innovative capability in multinational corporations: an empirical investigation. The Journal of Product Innovation Management 22, 412–429. Phene, A., Almeida, P., 2008. Innovation in multinational subsidiaries: the role of knowledge assimilation and subsidiary capabilities. Journal of International Business Studies 39, 901–919. Podsakoff, P.M., Organ, D.W., 1986. Self-reports in organizational research: problems and prospects. Journal of Management 12, 531–544. Podsakoff, P.M., MacKenzie, S.B., Podsakoff, N.P., 2003. Common method bias in behavioral research: a critical review of the literature and recommended remedies. Journal of Applied Psychology 88, 879–903. Rosenkopf, L., Almeida, P., 2003. Overcoming local search through alliances and mobility. Management Science 49, 751–766. Schulz, M., 2001. The uncertain relevance of newness: organizational learning and knowledge flows. Academy of Management Journal 44, 661–681. Schulz, M., Jobe, L.A., 2001. Codification and tacitness as knowledge management strategies. An empirical exploration. The Journal of High Technology Management Research 12, 139–165. Shannon, C.E., Weaver, W., 1998. The Mathematical Theory of Communication. University of Illinois Press, Urbana and Chicago, IL. Singh, J., 2005. Collaborative networks as determinants of knowledge diffusion patterns. Management Science 51, 756–770. Smith, K.W., Sasaki, M.S., 1979. Decreasing multicollinearity. Sociological Methods & Research 8, 35–56. Subramaniam, M., Venkatraman, V.N., 2001. Determinants of transnational new product development capability: testing the influence of transferring and deploying tacit knowledge. Strategic Management Journal 22, 359–378. Szulanski, G., 1996. Exploring internal stickiness: impediments to the transfer of best practice within the firm. Strategic Management Journal 17, 27–43. Taggart, J.H., 1997. Autonomy and procedural justice: a framework for evaluating subsidiary strategy. Journal of International Business Studies 28, 51–76. Taggart, J., Hood, N., 1999. Determinants of autonomy in multinational corporation subsidiaries. European Management Journal 17, 226–236. Tsai, W., 2001. Knowledge transfer in interorganizational networks: effects of network position and absorptive capacity on business unit innovation and performance. Academy of Management Journal 44, 996–1004. Tsai, W., 2002. Social structure of “coopetition” within a multiunit organization: coordination, competition, and intraorganizational knowledge sharing. Organization Science 13, 179–190. Tsai, W., Ghoshal, S., 1998. Social capital and value creation: the role of intrafirm networks. Academy of Management Journal 41, 462–476. Venaik, S., Midgley, D.F., Devinney, T.M., 2005. Dual paths to performance: the impact of global pressures on MNC subsidiary conduct and performance. Journal of International Business Studies 36, 655–675. Weiss, L.M., Capozzi, M.M., Prusak, L., 2004. Learning from the internet giants. MIT Sloan Management Review 45, 79–84. Williams, R., 2000. A note on robust variance estimation for cluster-correlated data. Biometrics 56, 645–646. Yamin, M., Forsgren, M., 2006. Hymer's analysis of multinational organization: power retention and the demise of the federative MNE. International Business Review 15, 166–179. Yang, Q., Mudambi, R., Meyer, K., 2008. Conventional and reverse knowledge flows in multinational corporations. Journal of Management 34, 882–902. Yates, J., Orlikowski, W.J., 1992. Genres of organizational communication: a structurational approach to studying communication and media. The Academy of Management Review 17, 299–326. Young, S., Tavares, A.T., 2004. Centralization and autonomy: back to the future. International Business Review 13, 215–237. Zanfei, A., 2000. Transnational firms and changing organisation of innovative activities. Cambridge Journal of Economics 24, 515–554. Zhou, C., 2002. Transnational flows of knowledge in multinational corporations: R&D co-practice as an integrating force, The University of Western Ontario, Faculty of Graduate Studies, Unpublished Ph.D. Dissertation. Zollo, M., Winter, S.G., 2002. Deliberate learning and the evolution of dynamic capabilities. Organization Science 13, 339–351.