The impact of subsidiaries’ internal and external integration on operational performance

Int. J. Production Economics 182 (2016) 73–85

Contents lists available at ScienceDirect

Int. J. Production Economics journal homepage: www.elsevier.com/locate/ijpe

The impact of subsidiaries’ internal and external integration on operational performance Krisztina Demeter a, Levente Szász b,n, Béla-Gergely Rácz c a

Logistics and Supply Chain Management Department, Corvinus University of Budapest, Fővám tér 8, H-1093 Budapest, Hungary Faculty of Economics and Business Administration, Babeş-Bolyai University, Teodor Mihali Str. 58-60, 400 591 Cluj-Napoca, Romania c Faculty of Economics and Business Administration, Babeş-Bolyai University, Teodor Mihali Str. 58-60, 400 591 Cluj-Napoca, Romania b

art ic l e i nf o

a b s t r a c t

Article history: Received 17 August 2015 Received in revised form 12 August 2016 Accepted 15 August 2016 Available online 16 August 2016

Subsidiaries of manufacturing companies operate as members of two distinct networks: the internal manufacturing network of the company and the external network of supply chain partners. Adapting the concept of “dual embeddedness” from international business literature to a manufacturing context, this paper proposes a model explaining the link between internal integration, external integration and operational performance. An international survey containing data on 470 manufacturing subsidiaries is used to operationalize the constructs. Structural equation modeling provides evidence for a full mediation model: external integration mediates the positive impact of internal integration on performance. Based on the results, it is put forward that knowledge generated within the internal manufacturing network can only be converted into subsidiary-level operational performance, if it is shared and recombined with external supply chain partners. The highest performance benefits can only be achieved if both suppliers and customers are involved in this process. A limitation of our approach is that knowledge flows are measured indirectly by assessing the level of integration of a subsidiary in the knowledge flows within the internal network, and the level of integration with suppliers and customers. & 2016 Elsevier B.V. All rights reserved.

Keywords: Manufacturing network Supply chain Knowledge sharing Internal integration External integration Dual embeddedness

1. Introduction Multinational manufacturing companies operate international networks of manufacturing plants dispersed around the globe. Global competition today takes place among these networks. One of the most important source of competitive advantage of these networks is represented by the knowledge that resides within these companies, and that can be transferred between plants and exploited in a more efficient and effective manner than knowledge located outside the network (Kogut and Zander, 1993; Gupta and Govindarajan, 2000; Van Wijk et al., 2008). To realize global competitive advantages multinational companies (MNCs) need to coordinate the intra-firm flow of knowledge, and combine the dispersed knowledge residing at individual units within the network (Kogut and Zander, 1993; Ambos et al., 2006; Ferdows, 2006). While the coordination and transfer of knowledge on the MNC level is an important determinant of competitiveness, the issue on individual subsidiary level is similarly significant. Plant managers n

Corresponding author. E-mail addresses: [email protected] (K. Demeter), [email protected] (L. Szász), [email protected] (B.-G. Rácz). http://dx.doi.org/10.1016/j.ijpe.2016.08.014 0925-5273/& 2016 Elsevier B.V. All rights reserved.

of individual subsidiaries should strive to use the intra-network knowledge to strengthen the competences of their own unit and reach high performances (Tsai, 2001; Van Wijk et al., 2008), thereby contributing to the performance of the whole network (Cheng et al., 2011). The present paper focuses on the transfer of knowledge on the subsidiary level, and investigates how the knowledge residing in the network can be acquired and converted into subsidiary-level operational performance. In this process we primarily concentrate on product and process related knowledge. It is put forward that subsidiaries need high levels of internal integration to be able to acquire such knowledge from their networks, but they also need to integrate in their external network (i.e., supply chain) and share this knowledge with supply chain partners in order to realize operational performance improvement. By pursuing this objective, the paper aims to bring the following contributions to manufacturing network literature. The starting point of the paper is offered by the fact that in practice subsidiaries operate as members of two distinct networks: manufacturing (internal) networks composed of several subsidiaries belonging to the same company, and supply (external) networks identified through information and material flows between different companies that cooperate with each other in a supply chain (Rudberg and Olhager, 2003). While literature offers strong

74

K. Demeter et al. / Int. J. Production Economics 182 (2016) 73–85

support that the integration into both networks can contribute to the success of individual subsidiaries, operations management studies rarely consider the two networks simultaneously. Furthermore, there is little knowledge on how these two networks are interrelated, and whether their interplay contributes to operational performance. International business literature, on the other hand, offers a number of studies that take both internal and external integration into consideration (e.g., Schmid and Schurig, 2003; Figueiredo, 2011; Meyer et al., 2011) without, however, focusing on operations management issues. Thus, this paper adopts the concept of simultaneous internal and external integration from the international business literature, and adapts it to an operations management context: we investigate the role of external integration in converting intra-network knowledge (i.e., internal integration) into operational performance. Additionally, as we primarily concentrate on product and production process related knowledge, a more direct performance measurement is applied by employing subsidiary-level operational performance indicators.

2. Literature review 2.1. Acquiring intra-network knowledge for performance improvement – the role of internal integration Modern multinational companies are viewed as a differentiated network of plants in which knowledge is developed and exploited at multiple units, and can be efficiently transferred to and applied at other network units (Almeida et al., 2002). Various types of knowledge can be found within an international manufacturing network (Wilcox King and Zeithaml, 2003). Gupta and Govindarajan (2000) provide examples for procedural (e.g., product design, production know-how) versus declarative (e.g., financial information) knowledge. Schmid and Schurig (2003) argue that organizational knowledge also varies according to functional activities. Acknowledging that a multitude of intra-firm knowledge types exist, this paper focuses on knowledge related to the product and the production process, i.e., that related to the operations function of the firm (Schmid and Schurig, 2003; Ferdows, 2006). Thus, financial or market-related knowledge, for example, do not constitute the focus of this study. While knowledge is crucial for the competitiveness of the MNC, from a subsidiary perspective knowledge is perceived in a slightly different manner. On one hand, subsidiaries cooperate with other units in the network to share and develop new knowledge in a collaborative manner. On the other hand, however, they also compete for intra-network knowledge (Luo, 2005), as this knowledge can be used to upgrade the competences of the subsidiary, increase performance and thereby secure its future within the network (Birkinshaw, 1996; Fusco and Spring, 2003; Cheng et al., 2011; Feldmann and Olhager, 2013). To be able to absorb intra-network knowledge, a subsidiary has to develop close links with other network units to enable intraorganizational learning characterized by frequent and intense interactions with other subsidiaries (Lane and Lubatkin, 1998; Schmid and Schurig, 2003; Minbaeva, 2007). Thus, a subsidiary needs to be deeply integrated into its manufacturing network, i.e., a more intense participation in knowledge sharing activities among network units is required (Vereecke et al., 2006). There are several elements which enable a higher level of internal integration of a subsidiary (Szász et al., 2016). Information sharing mechanisms represent one of the fundamental components of internal integration, and play an important role in coordinating activities in the network, including the flow of goods between subsidiaries (Rudberg and Olhager, 2003). The existence and richness of possible communication channels has been shown to relate

positively to the amount of knowledge flows that involve a specific subsidiary (Gupta and Govindarajan, 2000). Joint decision-making can also enhance the internal integration of a subsidiary. Participation in decision-making, i.e., cooperation with the management of other subsidiaries and the whole network, increases the ability of the subsidiary to acquire new knowledge from other network units (Colotla et al., 2003; Jansen et al., 2005). Participation in innovation sharing within the network represents another means of being integrated into the knowledge flows within an international manufacturing network. Joint innovation is frequently realized in practice through the exchange of employees within the network, which is essential in transferring tacit knowledge between subsidiaries (Ferdows, 2006; Vereecke et al., 2006). Furthermore, some authors argue that the development of a networklevel performance management system, where targets of subsidiary management are (partially) based on network-level performance indicators fosters internal integration (Gupta and Govindarajan, 2000; Luo, 2005; Greenwood et al., 2010). Thus, subsidiaries that are deeply integrated in their internal networks participate more intensely in knowledge sharing activities within the network. Acquiring intra-network knowledge is beneficial for the subsidiary and represents a key determinant of subsidiary performance (Tsai, 2001; Anh et al., 2006; Rhodes et al., 2008). Mahnke et al. (2005) argue that although a subsidiary might be able to acquire useful knowledge from other network units, it does not necessarily possess the capability to transform and exploit it to its own benefit (Cohen and Levinthal, 1990; Zahra and George, 2002). Nevertheless, they hypothesize and confirm that knowledge transfer has a positive impact on a subsidiary's performance. Thus, in this paper we argue that, on a general level, internal integration has a positive impact on subsidiary-level performance: subsidiaries integrate in their intra-organizational networks to acquire useful knowledge that can directly be used to enhance performance (Monteiro et al., 2008; Van Wijk et al., 2008; Szász et al., 2016). As this paper specifically focuses on the integration in product and process related knowledge flows, we employ operational performance measures instead of general business performance measures found usually in the international business literature (e.g., Tsai, 2001; Mahnke et al., 2005; Monteiro et al., 2008). Performance measures included in this study are related to the dimensions of cost and differentiation performance (Porter, 1985). This approach offers a more immediate and direct assessment of subsidiary performance, better corresponding to the potential benefits of product and process related knowledge. H1. Internal integration in product and process related knowledge flows has a positive impact on a subsidiary's operational performance. To underpin our hypotheses from a practical perspective, for each hypothesis a short business example is also provided. Toyota Motor Corporation is chosen as the multinational company for which company-wide knowledge sharing practices truly offer a competitive edge on the global market (Dyer and Nobeoka, 2000; Chaturvedi and Dutta, 2005). Toyota introduced, for example, the Yokoten System which stores concise reports on successful problem-solving processes implemented at various Toyota plants (Marksberry et al., 2010). All subsidiaries that connect to the system (i.e., are internally integrated in the network-wide system) have access to this knowledge accumulated within the company. Conforming to H1, the access to, and use of, such systems had a powerful impact on subsidiaries’ operational performance in the past, as it “helped to prioritize best practices depending on potential impact [and] advantages” (Chaturvedi and Dutta, 2005, p. 6).

K. Demeter et al. / Int. J. Production Economics 182 (2016) 73–85

2.2. Dual embeddedness – synchronizing internal and external integration Beside internal integration, manufacturing subsidiaries operate also as part of the external supply chain. Operations management literature, however, rarely considers the two types of integration (i.e., internal and external) simultaneously. One exception is a recent paper by Golini et al. (2014). They show that both internal and external integration have a positive impact on performance without considering the possible interplay between the two integration types. Similarly, Blome et al. (2014) provide evidence that both intra-firm and external knowledge transfer have a positive effect on performance. International business literature, however, suggests that there is a connection between the two and designates this relationship as a promising avenue for future research (Easterby-Smith et al., 2008). Song (2014) also proposes that, in order to better understand how subsidiaries transfer intra-network knowledge, future research should consider subsidiaries’ internal and external integration simultaneously. Looking at the interplay of the two integration types he suggests that internal and external integration may actually inhibit the ability of a subsidiary to source knowledge from other network units due to that fact that the two types of integration may create conflicting pressures (Meyer et al., 2011; Yamin and Andersson, 2011). On the contrary, Hallin et al. (2011) find that both internal and external integration can positively enhance the contribution of a received innovation to the subsidiary's performance. Moreover, Figueiredo (2011) argues that “subsidiaries that […] develop knowledge-intensive linkages with specific internal and external actors simultaneously […] achieve much higher levels of […] performance” (p. 435). Similarly, Ferraris (2014) proposes that a subsidiary needs to engage with the external context in which it is embedded (including integration with supply chain partners) in order to recombine external knowledge with the knowledge acquired from within the company, thereby creating subsidiary-level competitive advantages. Ho (2014) also puts forward, without testing it, that subsidiaries should seek for newly created knowledge within the MNC through internal integration, and then recombine it with externally sourced knowledge to adapt it to the recipient subsidiary's local environment. What are the activities that enable knowledge sharing with external supply chain partners? In a critical review of supply chain integration studies, Van der Vaart and van Donk (2008) argue that external integration is comprised of three major dimensions: integration practices (collaboration), interaction patterns (communication) and attitudes (long-term oriented relationships). Therefore, this paper approaches external integration as a construct reflected in

75

the level of information sharing and joint decision making in product or process related issues (Giménez and Ventura, 2005), but also in the efforts to develop a collaborative partnership with suppliers and customers (Van der Vaart and van Donk, 2008). While these activities do not correspond directly to the activities of internal integration, they represent the main methods of external integration that can enable knowledge sharing with supply chain partners. The simultaneous integration of a subsidiary into its external and internal network (see Fig. 1) is termed in the international business literature “dual embeddedness” (Figueiredo, 2011; Meyer et al., 2011). In this paper we adopt the concept of dual embeddedness and transfer it to an operations management context. Dual embeddedness studies in the international business literature focus mainly on how local market knowledge can be acquired from external networks, and disseminated within the MNC through the internal network (e.g., Andersson et al., 2001; Andersson et al., 2002; Foss and Pedersen, 2002; Schmid and Schurig, 2003; Meyer et al., 2011). Some more recent papers propose that the opposite direction should also be considered (Ferraris, 2014; Ho, 2014). As this paper focuses on product and process related knowledge, we also put forward that there is a positive link between internal and external integration, but take this latter, opposite direction. We argue that product and process related knowledge acquired by a subsidiary from within the company cannot be applied in void, and it needs to be shared with external supply chain partners in order to put that knowledge in practice in the specific operational context of the subsidiary. For example, if a subsidiary sources new knowledge from the network related to product modification or an improved production planning process, this knowledge has to be shared with supply chain partners as well in order to transpose the new knowledge into daily practice, i.e., to bridge the knowingdoing gap (Lane et al., 2001; Mahnke et al., 2005). Another argument for “dual embeddedness” is that subsidiaries that have already implemented methods and systems for internal integration might find easier to do the same with external supply chain partners as well compared to subsidiaries that have not yet implemented such internal systems or practices. In line with this argument, several authors find that internal (cross-functional) integration is a precondition for external integration with suppliers and customers (Giménez and Ventura, 2005; Zhao et al., 2011; Horn et al., 2014; Maiga et al., 2015), because “the same capabilities that are necessary to create internal […] ties are likely to be beneficial for the accumulation of [social and structural connections] with external partners” (Horn et al., 2014, p. 57). Thus, we hypothesize that subsidiaries that are deeply integrated in intra-network knowledge flows strive for a higher external integration with supply chain partners as well.

Fig. 1. Integration of a plant in the internal and external network.

76

K. Demeter et al. / Int. J. Production Economics 182 (2016) 73–85

We additionally argue for a specific direction of the relationship between internal and external integration. Knowledge transfer within a network of subsidiaries is viewed as being largely motivated by a specific problem that is faced by a specific unit (e.g., failure to meet performance targets) (Monteiro et al., 2008). To tackle such problems (especially the ones connected to the products or processes of the firm which are the focus of this paper) subsidiaries first try to acquire useful information from within the firm, as transferring knowledge between different legal entities is much more complicated than transferring knowledge between units within the same company (Kogut and Zander, 1993; Inkpen and Tsang, 2005). Internal integration enables the transfer of such intra-company knowledge. Then, however, to be able to use that knowledge in the specific context of the subsidiary (Lane et al., 2001), strong linkages to external partners need to be also developed (Ho, 2014). Thus, we argue that internal integration is an antecedent to external integration. While we acknowledge that knowledge transfer is a dynamic, and generally bidirectional process (e.g., Mudambi and Navarra, 2004) where knowledge sender and receiver roles are frequently interchanged (Monteiro et al., 2008), when focusing on integration mechanisms, internal integration usually precedes external integration. This argument is, again, in line with supply chain management literature, where firm-internal integration is regarded as the primary precondition of external integration with supply chain partners (Giménez and Ventura, 2005; Zhao et al., 2011; Horn et al., 2014; Maiga et al., 2015). H2. Internal integration in product and process related knowledge flows has a positive impact on external integration with supply chain partners. Returning to the Toyota example, several studies suggest that a subsidiary that intends to apply newly acquired intra-network knowledge (e.g., through the Yokoten System) related to lean practices needs to involve its external supply chain partners as well. In line with H2, Shah and Ward (2007) argue that, besides internal practices, both supplier and customer involvement are an integrative part of lean production. Practical examples of Toyota's external integration for sharing its internal knowledge with supply chain partners include supplier associations, the formation of joint problem solving teams with suppliers, inter-firm employee transfers, or the training offered to direct customers (Toyota dealers) by the Toyota University (Dyer and Nobeoka, 2000; Chaturvedi and Dutta, 2005). The strong ties with supply chain partners enable Toyota subsidiaries to both diffuse internal production know-how and source the diverse knowledge that resides within their supply chain partners (Dyer and Nobeoka, 2000). 2.3. The role of external integration in transforming intra-network knowledge into subsidiary performance Intra-network knowledge shared and recombined with the knowledge of supply chain partners has important performance benefits (He et al., 2013; Revilla and Knoppen, 2015). As we

previously argued, to apply intra-network knowledge in an efficient and effective manner, it has to be shared with supply chain partners as well to improve it or adapt it to the specific supply chain context of the subsidiary (Lane et al., 2001). Indeed, supply chain integration studies offer a strong support for the positive link between external (supply chain) integration and performance (e.g., Frohlich and Westbrook, 2001; Swink et al., 2007; Flynn et al., 2010; Schoenherr and Swink, 2012). Furthermore, the ‘arcs of integration’ concept suggests that the highest performance benefit can only be achieved if both suppliers and customers are integrated with the focal subsidiary (Frohlich and Westbrook, 2001), because “synergies might emerge when a firm possesses superior knowledge resources regarding both opportunities and boundary conditions in both demand and supply markets” (Schoenherr and Swink, 2012, p. 101). Following the arguments behind the ‘arcs of integration’, we put forward that if a subsidiary integrates with external supply chain partners on both supplier and customer side, thus enabling to share intra-network knowledge with them, can achieve higher performance improvement. H3. External integration with supply chain partners (customers and suppliers) is positively related to the subsidiary's operational performance. To highlight H3 from Toyota's perspective, Dyer and Nobeoka (2000) argue that the superior performance of Toyota's production system relative to its competitors is largely contributable to its effective knowledge sharing system with suppliers. Furthermore, Shah and Ward (2007) argue that in order to reduce the variability of the production process on a subsidiary level, which is a central objective of lean production, supply and demand variability needs also to be managed, which in turn requires strong supplier and customer integration (Shah and Ward, 2007). Thus, integration with supply chain partners contributes to reducing process variability, which reduces waste, and thereby contributes to subsidiary-level performance improvement. Taking H1, H2 and H3 together, we basically argue that the positive effect of internal integration on a subsidiary's operational performance is mediated by external integration. In concordance with the recommendations of Rungtusanatham et al. (2014) we do not hypothesize a priori for complete or partial mediation, and consequently test all three hypotheses together. The research model applied in this paper is presented in Fig. 2.

3. Research methodology 3.1. Research sample Internal integration, external integration, and performance are measured using the database of the sixth round of the International Manufacturing Strategy Survey (IMSS VI). The survey is administered by an international community of operations management scholars focusing on the links between manufacturing strategies, practices and performance. IMSS VI was carried out in

Fig. 2. Research framework.

K. Demeter et al. / Int. J. Production Economics 182 (2016) 73–85

2013 and early 2014 (www.manufacturingstrategy.net). The data collection process was administered in each country by local coordinators following a centrally coordinated and uniformly applied rigorous procedure. Potential respondents were selected from local official databases based on the following two criteria: manufacturing plants that operate in the assembly industries (ISIC 25– 30) and have at least 50 employees. Where needed, the IMSS questionnaire was translated into the local language using a reliable method (double parallel translation and/or back-translation). Before sending out the questionnaires, local coordinators contacted the plants’ manufacturing/operations manager (or equivalent) to present the project and ask for their approval to send out the questionnaire. Agreeing respondents could fill the questionnaire either on paper, or as a web-survey. Where necessary, a reminder was sent in order to maximize response rate. Returned questionnaires were checked for missing and incorrect data. Questionnaires having less than 70% valid completion rate were returned, wherever possible, to the respondents for clarification and completion, or were discarded. Survey data was collected in a central database where it was further controlled for validity by the central administration team. The required response rate (i.e., the ratio between the number of answers collected and the number of questionnaires sent to potential respondents) for each country was a minimum of 20%. Non-response bias was checked on countrylevel, with no significant differences being found after comparing non-respondents with the rest of the sample in terms of size and industry. Finally, 931 valid questionnaires were collected from manufacturing plants located in a total of 22 countries, reaching an aggregate response rate of 36% for the whole sample. For the purpose of the present study we selected from the IMSS VI database only those manufacturing plants that operate as members of a manufacturing network (i.e., single-plant companies were filtered out from the sample). We further eliminated those cases that had missing data on any of the variables used in the analysis. Thus, the final dataset consists of 470 manufacturing plants located in 22 countries. The distribution of the research sample by country of location is presented in Table 1. The breakdown of the sample by size and industry is presented in Table 2. The origin (i.e., parent/headquarter) of the subsidiaries included in our sample can be traced back to 29 different countries, more than 75% of them belonging to developed country manufacturers. Thus, our sample is biased towards manufacturing companies headquartered in developed countries. 3.2. Measurement model Using AMOS 21.0 software, we employed confirmatory factor analysis (CFA) to develop the constructs measuring internal integration, external integration and operational performance. The exact wording of individual questionnaire items used in this study Table 1 Distribution of the sample by country. Country

Frequency

% of total

Country

Frequency

% of total

Belgium Brazil Canada China Denmark Finland Germany Hungary India Italy Japan (continued…)

17 22 8 44 21 13 5 28 37 25 57

3.6% 4.7% 1.7% 9.4% 4.5% 2.8% 1.1% 6.0% 7.9% 5.3% 12.1%

Malaysia Netherlands Norway Portugal Romania Slovenia Spain Sweden Switzerland Taiwan USA TOTAL

10 28 21 22 15 10 14 20 16 12 25 470

2.1% 6.0% 4.5% 4.7% 3.2% 2.1% 3.0% 4.3% 3.4% 2.6% 5.3% 100.0%

77

Table 2 Distribution of the sample by size and industry. Employees

Frequency % of total

ISIC codea Frequency % of total

Small (o 250) Medium (250– 500) Large ( 4500) TOTAL

156 99

33.2% 21.1%

25 26

140 66

29.8% 14.0%

215 470

45.7% 100%

27 28 29 30 TOTAL

81 103 60 20 470

17.2% 21.9% 12.8% 4.3% 100%

a ISIC 25: Manufacture of fabricated metal products, except machinery and equipment; 26: Manufacture of computer, electronic and optical products; 27: Manufacture of electrical equipment; 28: Manufacture of machinery and equipment not elsewhere classified; 29: Manufacture of motor vehicles, trailers and semi-trailers; 30: Manufacture of other transport equipment.

is presented in Appendix 1, while correlations between individual items are presented in Appendix 2. First, to assess the extent of internal integration (Internal_Int) we asked respondents to indicate on a 5-point Likert scale the extent of effort put in the last 3 years into implementing the five practices presented in the literature review (InfoShare, JointDec, JointInnov, CommTech, NetwPerf) as elements of network integration (1 ¼None, 5¼ High). Questions were specifically targeted at product and process related knowledge, and for each item specific examples were offered to facilitate the common understanding of the questions. Second, external integration (External_Int) was measured as a second-order construct reflected in the two first-order constructs of supplier integration (SuppInt) and customer integration (CustInt). The formation of second-order constructs has to be underpinned by sound theoretical and methodological arguments (Hair et al., 2010). The proposed second-order construct represents the hypothesis that the distinct, but related factors of supplier and customer integration reflect one common underlying factor. The theoretical argument for the formation of a second-order construct to measure external integration is provided by the arcs of integration concept, which implies that manufacturing units need to integrate both their suppliers and customers in order to achieve high performance improvement (Frohlich and Westbrook, 2001; Schoenherr and Swink, 2012). As we primarily investigate its performance impact, external integration is measured as a single, second-order construct. Nevertheless, to cross-validate our results, we have carried out the analyses with separate supplier and customer integration measures as well, as there are some studies that suggest that the two integration constructs might have different performance implications (e.g., Swink et al., 2007; Flynn et al., 2010). Finally, from a methodological point of view, the formation of the second-order external integration construct is supported by the relatively higher correlation coefficients between individual items of supplier and customer integration compared to other correlations (see Appendix 2). In terms of measuring our external integration construct, we used the study of Frohlich and Westbrook (2001) as a starting point which uses items related to three different integration dimensions: knowledge sharing with customers and suppliers, logistics responsiveness, and the use of common resources (Schoenherr and Swink, 2012). As this paper focuses on the integration in product and process related knowledge flows, we used only the items related to the knowledge sharing dimension of supply chain integration. Three main dimensions, i.e., information sharing, collaborative approaches, and joint decision making were selected as potential enablers of external knowledge sharing (Giménez and Ventura, 2005; Van der Vaart and van Donk, 2008). Thus, the two underlying constructs of

78

K. Demeter et al. / Int. J. Production Economics 182 (2016) 73–85

external integration were measured using three questionnaire items that enquired about the effort put in the last 3 years into implementing information sharing, collaborative approaches, and joint decision making practices with suppliers (S_Info, S_Collab, S_Joint) and customers (C_Info, C_Collab, C_Joint). The effort put into implementing these practices was measured on a 5-point Likert-scale ranging from 1 (none) to 5 (high). Third, the IMSS questionnaire also enquired about the extent the manufacturing performance of the plant has changed over the last three years. These indicators were used to develop two operational performance constructs, namely cost (CostPerf) and differentiation (DiffPerf) performance improvement. Differentiation performance was also built as a second-order construct reflected in quality (QualPerf), flexibility (FlexPerf) and delivery (DelPerf) related operational performance. This approach is supported by both theory and statistics. The theoretical argument behind the formation of differentiation as a second-order construct is provided by Porter's (1985) competitive advantage model according to which companies should focus either on cost or on differentiation performance in order to stay ahead of competition. From a methodological perspective the relatively higher correlation coefficients between quality, flexibility and delivery related items (see Appendix 2) also underpin this approach. A final advantage of using a second-order differentiation performance construct is that it provides a more parsimonious model with fewer dependent variables (i.e., differentiation versus quality, flexibility and delivery performance), thus making the results more easily interpretable. Details of the overall CFA are presented in Table 3. The measurement model detailed in Table 3 shows adequate fit to the data (absolute fit indices: χ2 ¼310.62, p¼ .000, df ¼159, χ2/ df¼ 1.954, GFI¼ .939, RMSEA ¼ .045, SRMR¼ .0382; incremental fit indices: NFI ¼.934, CFI¼.966, TLI ¼.960). Reliability, convergent validity and discriminant validity was also established (Hair et al., 2010). CR values of each factor exceed the .7 threshold values,

indicating adequate construct reliability. All factor loadings exceed the minimum .5 value, and except for one loading they are also higher than the ideal .7 threshold. AVE values, computed as the mean variance of the items loading on the same construct also reach the .5 minimum showing adequate levels of convergent validity, i.e., indicating that individual items consistently represent the same construct. Discriminant validity is assessed by comparing the correlation between each pair of constructs to the square root of the AVE measure of the two factors. In each case correlation value is lower than the square root of AVE, indicating that each construct represents a truly different phenomenon than other constructs. Detailed results are presented in Table 4. 3.3. Common method bias The measurement model used in this study combines responses to several single-respondent, self-reported questionnaire items, which means that common method variance (i.e., variance attributable to the measurement method rather than to the constructs these measures represent) is a potential problem (Podsakoff et al., 2003) that could undermine the validity of our analysis. Therefore, we provide some procedural and statistical arguments to show that the possible extent of common method bias is not powerful enough to distort our results. First, the design of the IMSS questionnaire inherently helps to control common method bias. The items used in this paper come from three different sections of the questionnaire (performance measures – section B, p. 4; supplier and customer integration – section C, p. 8; internal integration – section C, p. 9). Although questionnaires were completed by a single respondent in each subsidiary, the relevant questions primarily combined in this study (i.e., integration measures and performance) were well separated from each other, thus reducing respondents’ possible motivation to fill in the gaps based on previous responses.

Table 3 Constructs developed using confirmatory factor analysis. Construct

Item

Main constructs Internal_Int

Item description

InfoShare Information sharing with other plants JointDec Joint decision making with other plants JointInnov Innovation sharing/joint innovation CommTech Use of technology to support communication NetwPerf Network performance management system External_Int SuppInt Supplier-side integration in knowledge flows CustInt Customer-side integration in knowledge flows CostPerf MfgCost Unit manufacturing cost improvement OrdCost Order cost improvement MfgTime Manufacturing lead time improvement DiffPerf Qual Quality performance improvement Flex Flexibility performance improvement Deliv Delivery performance improvement Underlying first-order constructs of External_Int and DiffPerf SuppInt S_Info Information sharing with key suppliers S_Collab Collaborative approaches with key suppliers S_Joint Joint decision making with key suppliers CustInt C_Info Information sharing with key customers C_Collab Collaborative approaches with key customers C_Joint Joint decision making with key customers Qual Qual1 Conformance quality improvement Qual2 Quality and reliability improvement Flex Flex1 Volume flexibility improvement Flex2 Mix flexibility improvement Deliv Del1 Delivery speed improvement Del2 Delivery reliability improvement

Mean

SD

Path loading***

3.24 3.15 3.13 3.31 3.11 3.00 2.66 2.53 2.42 2.82 3.04 2.63 3.24

1.05 1.08 1.07 1.14 1.14 .78 .74 .97 .88 .96 .76 .62 .86

.783 .765 .740 .777 .789 .809 .820 .795 .702 .612 .789 .738 .782

3.27 3.24 3.05 3.07 3.00 3.07 3.10 3.22 3.27 3.19 3.16 3.20

1.00 1.03 1.06 1.10 1.12 1.07 .96 .96 .98 .97 .98 1.01

.769 .856 .793 .836 .828 .744 .835 .855 .827 .719 .817 .907

Absolute fit indices: χ2 ¼ 310.62, p ¼.000, df ¼ 159, χ2/df ¼ 1.954, GFI ¼.939, RMSEA ¼ .045, SRMR¼ .0382. Incremental fit indices: NFI¼ .934, CFI ¼ .966, TLI ¼.960. ***

All path loadings are significant at p ¼ .001.

K. Demeter et al. / Int. J. Production Economics 182 (2016) 73–85

Table 4 Reliability, convergent validity and discriminant validity assessment.

Internal_Int (II) External_Int (EI) DiffPerf (DP) CostPerf (CP)

AVE

CR

II

EI

DP

CP

59.4% 66.3% 59.3% 50.0%

.880 .798 .814 .748

.771 .698 .381 .176

.815 .427 .255

.770 .544

.707

CR – construct reliability, AVE – average variance extracted, square root of AVE – values on the diagonal, remaining values – pairwise correlations between constructs.

Second, the standard data collection procedure, applied consistently in each participating country, was designed to fully protect respondent anonymity. Initial contact and the cover letter used to approach possible respondents highlighted that information collected during the survey will be handled anonymously throughout the research project. Furthermore, the cover page of the questionnaire explicitly states that all responses are handled with absolute confidentiality, and that the names of companies, business units, products and individuals (including the respondent) will not be released. This procedure aims also to make respondents less likely to edit their responses so as to reflect them in a more positive light than in reality. Finally, three different statistical best-practice tools were applied to assess the extent of common method bias in our data. First, in concordance with Podsakoff et al. (2003), Harman's singlefactor test was used in which all variables were loaded into an exploratory factor analysis and the unrotated factor solution was used to determine the number of factors necessary to account for the variance in the variables. The result of the test led to four factors with eigenvalues exceeding the 1.0 threshold value, with the first factor having only a 32.36% contribution to the total variance explained. This indicates that the possible amount of common method variance is not big enough to bias our analysis. To further investigate common method bias, CFA was applied to Harman's single factor model (Zhao et al., 2011). The fit indices were unacceptable (absolute fit indices: χ2 ¼2373.956, p ¼.000, df ¼170, χ2/df ¼13.964, GFI ¼.589, RMSEA ¼.166, SRMR¼.1385; incremental fit indices: NFI ¼.492, CFI¼.509, TLI ¼.451), indicating that the potential common method bias is small. As a third statistical test, we also controlled for the effect of a possible latent common method factor in the original CFA model (Table 3). Following the recommendations of Podsakoff et al. (2003) and Ketokivi and Schroeder (2004), the fit indices of two separate measurement models were estimated and compared: the fit of the original measurement model (trait model), and the fit of a model including a common method factor (method model). The addition of a common factor improved fit only marginally (GFI by .010, RMSEA by .004, NFI by .012, CFI by .009, TLI by .006), with the

79

common method factor accounting for only 2.6% of total variance. Additionally, the path loadings of the main factors of this study were significant and had similar values in both models, suggesting that the factors created are robust in the presence of a common method factor as well (Paulraj et al., 2008; Zhao et al., 2011). Based on the three statistical tests, together with the procedural arguments listed above, we conclude that common method bias is not large enough to distort the results of the analysis.

4. Analysis and findings The relationships presented in Fig. 2 are tested using structural equation modeling (SEM) by converting the measurement model into a structural model and testing it for validity. Size, measured as the natural logarithm of the number of employees within the business unit (LnSize), industry sector (Industry) as a categorical variable, and plant age (PlantAge), measured as the number of years since establishment, are introduced as control variables in our model. We control for the impact of these variables on each dependent or partially dependent variable (External_Int, DiffPerf, CostPerf) in our model. Variance inflation factors (VIF) are computed to assess the extent of multicollinearity in respect of the integration and control variables. All VIF values are well below the usual threshold levels (VIF o4.0), indicating no issues of multicollinearity (O'Brien, 2007). The structural model has an acceptable fit to the data (absolute fit indices: χ2 ¼457.955, p¼ .000, df¼211, χ2/df ¼2.170, GFI¼.924, RMSEA ¼.050, SRMR¼ .0649; incremental fit indices: NFI ¼ .905, CFI¼.946, TLI¼ .935). Results are presented in Fig. 3. For sake of brevity only the main constructs, control variables and their relationships are shown. Individual items forming the main constructs are not included in the figure. SEM results indicate that there is no significant direct relationship between internal integration and operational performance, thus offering no support for H1. Internal integration, on the other hand, is in a strong positive relationship with external integration: subsidiaries that engage intensively in intra-network knowledge sharing are also cooperating and sharing information with supply chain partners intensively. Thus, H2 is supported. Being deeply integrated in the external network brings strong performance benefits on operational level: sharing knowledge with supply chain partners has a positive impact on both cost and differentiation performance. Thus, H3 is also supported. In concordance with Frohlich and Westbrook (2001), and Schoenherr and Swink (2012), who found that in practice supplier and customer integration practices are not always applied simultaneously, we further test our model by splitting up the external integration construct into two distinct constructs of supplier and customer integration. Although fit indices were marginally worse than for the original CFA model, together with reliability and validity measures, they indicated that the measurement

Fig. 3. Structural equation model.

80

K. Demeter et al. / Int. J. Production Economics 182 (2016) 73–85

model with separate supplier and customer integration constructs can still be applied. Using these constructs the SEM model showed a narrowly acceptable fit to the data with some fit indices being just below the usually acceptable thresholds. Nevertheless, based on the theoretical argument presented above (Frohlich and Westbrook, 2001; Schoenherr and Swink, 2012) we carried out the structural analysis. Numerical results are presented in Appendix 3. These results indicate that by separating supplier and customer integration, the strong relationship between internal and external integration persists, while the separate impact of supplier and customer integration on operational performance becomes much weaker (and significant only in one instance). The weakening relationship between external integration and performance is also accompanied by one of the direct path loadings between internal integration and performance becoming significant. These findings actually offer further support for the arcs of integration concept. Schoenherr and Swink (2012) argue that “firms today realize significant improvements in performance even from small integration steps [on either supplier or customer side]” (p. 107), which is in line with the single weak, but significant performance impact found in our SEM model with separate supplier and customer integration constructs. Nevertheless, “closer integration with both customers and suppliers yields enhanced performance over more myopic […] approaches” (Schoenherr and Swink, 2012, p. 110), which supports that our initial model with an aggregated external integration construct is essentially more realistic. Thus, we return to our initial model (Fig. 3), and further test the robustness of our results by investigating the same structural model on different subgroups of the research sample. The formation of subgroups is based on a further important control variable that concerns knowledge flows within manufacturing networks. Several authors suggest that intra-network knowledge flows are related to the flow of goods within the same network (Foss and Pedersen, 2002; Vereecke et al., 2006; Prajogo and Olhager, 2012). The IMSS questionnaire enquired about the exact percentage of input and output goods that are provided by/to internal network partners (i.e., different plants within the same company), the rest of goods being exchanged with external network partners (i.e., supply chain partners). Based on these two percentage values a hierarchical cluster analysis with Ward's method has been carried out. Results of the analysis clearly indicated that the two-cluster solution is the most appropriate. Consequently, using k-means clustering, the sample has been split into two subgroups: (1) HighEmbed group which has a high percentage of goods exchanged with internal network partners both on input and output side, and (2) LowEmbed group having only

limited exchange of goods with internal network partners. The characteristics of these two groups, as well as SEM results for each subgroup is presented in Table 5. Following Feldmann and Olhager (2008), a further group of subsidiaries (ZeroEmbed) was created within the LowEmbed group, consisting of network units that have no physical exchange with other plants within the network, i.e., physically isolated plants whose purchasing and distribution operations involve exclusively external supply chain partners. Running the analysis on different subsamples brings further evidence for the relationships found: in all three clusters internal integration has a positive effect on external integration (H2 supported), but is not directly related to performance (H1 not supported), while external integration has a significantly positive impact on at least one of the two performance indicators (H3 at least partially supported). Related to H3, it can be seen that the HighEmbed group increases only its cost performance, while the LowEmbed and ZeroEmbed groups manage to improve their differentiation performance. Thus, subsidiaries that are more embedded in the flow of goods within their network seem to focus more on improving cost, even in their relations with supply chain partners. This suggests that by having a stronger company-level control on the flow of goods, economies of scale and an optimal production allocation can be realized leading to an improvement in cost efficiency (Rudberg and Olhager, 2003). On the other hand, subsidiaries only loosely embedded into the physical flows within their networks use external integration to improve their differentiation performance. Having only limited possibilities to optimize production and reach economies of scale (Rudberg and Olhager, 2003), they base their competitive advantage on differentiation related factors. Taken together, the results suggest that the beneficial effect of internal integration on both performance indicators is mediated by external integration, i.e., internal integration has only an indirect effect on performance. The indirect effect is computed as the product term between the direct effect of internal integration on external integration and the direct effect of external integration on performance. To test whether the indirect (mediated) effects are significant, we used the bootstrapping method (Rungtusanatham et al., 2014) by generating 2000 samples to estimate indirect effects with 95% bias-corrected confidence intervals. Results are presented in Table 6. Results summarized in Table 6 show that internal integration in itself has a direct positive impact on both performance indicators, but when the mediator variable (external integration) is introduced in the model, the positive direct effect diminishes and becomes insignificant. Thus, the beneficial effect of internal

Table 5 SEM results for different clusters of subsidiaries. Clusters Clusters based on the extent of embeddedness in the flow of goods within the internal manufacturing network Within-cluster results

HighEmbed

LowEmbed

ZeroEmbed

No. of cases % of input sourced from the network (mean, st. dev.) % of output sold to the network (mean, st. dev.) SEM results (path loadings) Internal_Int – External_Int External_Int – CostPerf External_Int – DiffPerf Internal_Int – CostPerf Internal_Int – DiffPerf

N¼ 172 59.2% (.277)

N¼ 269 12.2% (.134)

N¼ 51 0% (.000)

60.51% (.295)

9.1% (.119)

0% (.000)

nnn

.677 .427nn .215  .142 .198

Significant relationship on the np=.05, nnp=.01, nnnp=.001 level. The effect of control variables was accounted for in each of the three clusters.

nnn

.742 .161 .468nn .068 .060

.661nnn .093 .599nn .038 .060

K. Demeter et al. / Int. J. Production Economics 182 (2016) 73–85

Table 6 Bootstrapping results for testing the mediation effect of external integration.

Direct effect without mediator Direct effect with mediator Indirect (mediation) effect Conclusion

Internal_Int – CostPerf Path loading (p-value)

Internal_Int – DiffPerf Path loading (p-value)

.178 (.002)

.380 (.000)

 .102 (.288)

.079 (.393)

.698  .380 ¼.265 (.008)

.698  .411¼ .287 (.002)

Significant full mediation Significant full mediation

integration on operational performance is fully mediated by external integration.

5. Discussion and conclusion 5.1. Summary and implications The primary objective of this paper was to investigate how intra-network knowledge transfer can be used to enhance subsidiary-level operational performance. In answering this question we took into consideration that subsidiaries operate as members of two distinct networks: manufacturing (internal) networks composed of several subsidiaries with the same owners, and supply (external) networks identified through cooperation between different companies (Rudberg and Olhager, 2003). The central finding of this paper indicates that external integration with supply chain partners plays a key mediating role in transforming internal integration into performance. The analyses bring the following contributions to the literature. First, our paper addressed the role of internal and external integration simultaneously, an approach that has rarely been considered in operations management literature. We adopted the concept of “dual embeddedness” of a subsidiary from the international business literature (Figueiredo, 2011; Meyer et al., 2011), and transferred it to an operations management context. Our results suggest that subsidiaries that are deeply integrated into the product and process related knowledge flows of the intra-firm network also develop integrated links with their supply chain partners, which is a novel finding in the operations management literature. We propose that this dual integration is needed to (1) acquire useful knowledge from other subsidiaries within the internal network, and (2) share intra-network knowledge with external supply chain partners. Second, our analysis offered strong statistical support for a full mediation model, showing that the positive performance effect of internal integration is fully mediated by external integration with supply chain partners. This result suggests that managing intranetwork links is only successful if the subsidiary manages its external linkages with supply chain partners as well. We offer two possible, non-exclusive explanations on why external integration of a subsidiary fully mediates the performance impact of internal integration. First, synchronizing operations with supply chain partners can inherently contribute to the swift and even flow of materials within a subsidiary, thereby increasing its cost efficiency (Schmenner and Swink, 1998; Shah and Ward, 2007). Similarly, close integration with supply chain partners can, in itself, lead to higher differentiation performance in terms of quality, delivery and flexibility (Frohlich and Westbrook, 2001). Second, the full mediation model suggests that supply chain integration is a critical element for a subsidiary to exploit (i.e., turn into performance) knowledge generated within the international manufacturing network (Zahra and George, 2002). Literature argues that suppliers

81

and customers possess knowledge stocks that can be recombined with intra-network knowledge to tailor them to the subsidiary's specific supply chain context, which in turn contributes to the subsidiary's performance (Andersson et al., 2001, 2002; Schmid and Schurig, 2003; Ho, 2014). External knowledge stocks, however, can only be exploited by the focal subsidiary, if it highly integrates with external partners (Bessant et al., 2003). Thus, external integration can help to put intra-network knowledge into practice, and to bridge the knowing-doing gap within subsidiaries (Lane et al., 2001; Mahnke et al., 2005). Third, we found that the mediating role of external supply chain integration is strong only if it is implemented on both customer and supplier side. This finding offers further support to the arcs of integration concept (Frohlich and Westbrook, 2001; Schoenherr and Swink, 2012), but from an intra-network knowledge sharing perspective. We argue that intra-network knowledge has weaker performance effects if the subsidiary solely cooperates with either suppliers or customers, while the impact on performance becomes much stronger if both suppliers and customers are integrated. Fourth, our results offer a more comprehensive view on why companies such as Toyota are able to achieve high operational performance. By managing both internal and external networks consciously, their subsidiaries are able to exploit and put into practice internal knowledge by a close integration with their supply chain partners (Dyer and Nobeoka, 2000). This integration is beneficial not only for the focal subsidiary, but for supply chain partners as well, as they can enhance their own performance (e.g., Modi and Mabert, 2007; He et al., 2013), which in turn can reinforce the performance improvement of the subsidiary. For example, if the subsidiary receives components faster and more reliably from its suppliers, it can more easily improve its own delivery performance. 5.2. Limitations and further research As in case of most survey-based studies, our paper addressed the issue of knowledge transfer and its impact on performance on an aggregate level, without having the possibility to offer information on the details of this process. An important limitation related to this approach is that we were only able to measure knowledge flows indirectly by assessing 1) the level of integration of a subsidiary into the product and process related knowledge flows within the internal network, and 2) the supply chain integration with suppliers and customers. While the two integration constructs do reflect the effort of a subsidiary put into enabling knowledge exchange with internal and external network partners, future studies could offer a more detailed understanding by explicitly focusing on the transfer of knowledge per se. For example, it would be worth investigating how different integration practices enable and facilitate the transfer of product and process related knowledge within and across firm boundaries, and whether these practices depend on the type of knowledge transferred (Ferdows, 2006). Similarly, we could not control whether and how the knowledge acquired from the intra-firm network is transmitted to supply chain partners via the integration mechanisms (information sharing, joint decision making, and collaborative approaches) included in our study. Literature argues that the intra-network knowledge acquired by a subsidiary needs to be assimilated (understood) and transformed (internalized) before it can be exploited (turned into performance) (Zahra and George, 2002). Thus, case studies should investigate how intra-network knowledge is assimilated, transformed, and exploited by subsidiaries, and how integration with suppliers and customers facilitates this process. Another limitation of this study is that it did not consider the possible impact of contingency factors on the relationships established in this study. International business literature argues that

82

K. Demeter et al. / Int. J. Production Economics 182 (2016) 73–85

there are several factors – internal and external to the subsidiary – that can have a significant impact on knowledge transfer. In this study we used two integration constructs that act as enablers for knowledge sharing, but whether knowledge sharing is successfully orchestrated is also dependent on other factors. Here we provide a list of prevalent contingency factors from the literature (characteristics of the knowledge transferred, particularities of the parties involved in the transfer, and the relationship with other units and the broader external context) that have been suggested to significantly influence knowledge sharing (Argote et al., 2003). The investigation of these factors in relation to our model could represent an important direction for future research. While this paper offers solid evidence for the role of internal and external integration in improving subsidiary performance, the inclusion of internal and external contingency variables in our model, such as the ones discussed in more detail below, has the potential to enrich our understanding of the contextual conditions under which integration and knowledge sharing practices can significantly contribute to the performance of manufacturing subsidiaries. First, characteristics of knowledge, especially tacitness (i.e., noncodifiability), has been widely argued to represent an important barrier to knowledge transfer. For example, Zander and Kogut (1995) established that the codifiability and teachability of knowledge has a significant impact on the speed of knowledge transfer. Ferdows (2006) argues that depending on the tacitness of knowledge, different transfer mechanisms might be appropriate. Other characteristics, such as the complexity of knowledge (Minbaeva, 2007), the specificity of knowledge to a certain functional area or context (Schmid and Schurig, 2003), or the speed of change of the production know-how (Ferdows, 2006) might have similar effects. In light of these findings it would be worth investigating how knowledge sharing within the internal and external network, as well as the interaction between the two networks, is affected by the characteristics of the knowledge transferred. Second, this paper handled the focal subsidiary and its internal and external network partners as “black boxes”. Nevertheless, characteristics of these units might have a significant impact on knowledge sharing. For example, literature argues that the receiving unit's absorptive capacity (Cohen and Levinthal, 1990; Zahra and George, 2002) has a significant influence on the amount and performance impact of the knowledge transferred. Minbaeva et al. (2003)

operationalize absorptive capacity as the interaction of the ability and motivation of the focal unit's employees to take in new knowledge. In light of our study, it would represent a promising avenue for future research to investigate how the levels of absorptive capacity at the focal subsidiary and its supply chain partners affect the way intranetwork knowledge is transformed into operational performance. Since both the subsidiary and its supply chain partners have to understand and internalize the same knowledge, the levels of relative absorptive capacity (Lane and Lubatkin, 1998) between the parties involved could represent another decisive factor in exploiting the knowledge for improving the subsidiary's operational performance. Other characteristics of the units, such as the willingness of the sender and receiver to share knowledge (Minbaeva, 2007) or the strategic role of the focal subsidiary (Feldmann and Olhager, 2013) can also have an impact on knowledge sharing with members of the internal network and the external supply chain. Third, the relationship of the focal subsidiary with other network units as well as with the broader external context represents another set of factors that can have a significant impact on our model. Frost et al. (2002), for example, find that frequent interaction with other intra-network units and parent-subsidiary relationships influence how knowledge is accumulated at subsidiaries. Another fundamental issue for multinational subsidiaries is the way they can balance their adaptation to the international and local contexts. The integrationresponsiveness framework (Prahalad and Doz, 1987; Bartlett and Ghoshal, 1989) is meant to describe MNC strategies to simultaneously deal with the forces of globalization and localization, which determines the role subsidiaries play within the internal and external networks. Thus, it would be worth investigating whether the relationships among internal integration, external integration and performance differ according to the various strategies of the integration-responsiveness framework.

Acknowledgments This research was supported by the Hungarian Research Fund (OTKA-112745) and SCOPES Joint Research Project IZ73Z0_152505 of the Swiss National Science Foundation.

Appendix 1 – Questionnaire items G7. Indicate the effort put in the last 3 years into implementing action programs related to manufacturing network integration: Effort in the last 3 years None Improve information sharing for the coordination of the flow of goods between your plant and other plants of the network (e.g. through exchange information on inventories, deliveries, production plants, etc.) Improve joint decision making to define production plans and allocate production in collaboration with other plants in the network (e.g. through shared procedures, shared forecasts) Improve innovation sharing/joint innovation with other plants (through knowledge dissemination and exchange of employees inside the network) Improve the use of technology to support communication with other plants of the network (e.g. ERP integration, shared databases, social networks) Developing a comprehensive network performance management system (e.g. based on cost, quality, speed, flexibility, innovation, service level)

High

1 2 3 4 5 1 2 3 4 5 1 2 3 4 5 1 2 3 4 5 1 2 3 4 5

K. Demeter et al. / Int. J. Production Economics 182 (2016) 73–85

83

SC6. Indicate the effort put in the last 3 years into implementing action programs related to external integration: Effort in the last 3 years None Sharing information with key suppliers (about sales forecast, production plans, order tracking and tracing, delivery status, stock level) Developing collaborative approaches with key suppliers (e.g. supplier development, risk/revenue sharing, long-term agreements) Joint decision making with key suppliers (about product design/modifications, process design/modifications, quality improvement and cost control) Sharing information with key customers (about sales forecast, production plans, order tracking and tracing, delivery status, stock level) Developing collaborative approaches with key customers (e.g. risk/revenue sharing, long-term agreements) Joint decision making with key customers (about product design/modifications, process design/modifications, quality improvement and cost control)

High

1 2 3 4 5 1 2 3 4 5 1 2 3 4 5 1 2 3 4 5 1 2 3 4 5 1 2 3 4 5

B6. How has your manufacturing performance changed over the last three years? Compared with three years ago the indicator has

Conformance quality Product quality and reliability Volume flexibility Mix flexibility Delivery speed Delivery reliability

Unit manufacturing cost Ordering costs Manufacturing lead time

Decreased (  5% or worse)

Stayed about the same (  5%/ þ5%)

Slightly increased (þ5–þ 15%)

Increased (þ 15– Strongly increased 25%) (þ25% or better)

1

2

3

4

5

1

2

3

4

5

1 2 3 1 2 3 1 2 3 1 2 3 Compared with three years ago the indicator has Increased (þ5% or Stayed about the same Slightly decreased worse) (þ 5%/  5%) (  5/  15%) 1 2 3

4 4 4 4

5 5 5 5

Decreased (  15/  25%) 4

Strongly decreased (  25% or more) 5

1 1

4 4

5 5

2 2

3 3

Appendix 2 – Correlation table of individual items used in the analysis See Table A2 here.

Appendix 3 – SEM model with separate supplier and customer integration constructs First, a similar measurement model has been built to that presented in Table 3, but the second-order external integration construct has been dropped from the model. Thereby, supplier integration (SuppInt) and customer integration (CustInt) can be used as main constructs in the analysis. Fit indices, although somewhat lower than in the case of the original model, still suggested that the measurement model can be accepted (absolute fit indices: χ2 ¼309.376, p ¼.000, df ¼157, χ2/df¼ 1.971, GFI ¼.939, RMSEA ¼ .045, SRMR¼.0379; incremental fit indices: NFI ¼.934, CFI ¼.966, TLI¼ .959). Additionally, all reliability and validity analyses have been performed and suggested no concerns with the modified measurement model. Thus, we proceeded to test the structural model. For sake of brevity, the SEM model shows only the main constructs without the control variables (the impact of the three control variables on the CustInt, SuppInt, CostPerf, DiffPerf constructs has been accounted for) and the individual items loaded on the constructs. While most of the fit indices of the SEM model (Fig. 4) show adequate values (absolute fit indices: χ2 ¼ 525.98, p ¼.000, df¼207, χ2/ df¼ 2.541, GFI¼.915, RMSEA¼ .057, SRMR¼ .0784; incremental fit indices: NFI ¼.890, CFI¼ .930, TLI¼.914), it has to be also noted that some of them are only hardly acceptable (e.g., SRMR close to .08, NFI o.90). Modification indices (Hair et al., 2010) suggest that model fit could substantially be improved by adding a covariance between the error terms of the customer and supplier integration constructs,

84

K. Demeter et al. / Int. J. Production Economics 182 (2016) 73–85

Table A2

InfoShare (1) JointDec (2) JointInnov (3) CommTech (4) NetwPerf (5) S_Info (6) S_Collab (7) S_Joint (8) C_Info (9) C_Collab (10) C_Joint (11) Qual1 (12) Qual2 (13) Flex1 (14) Flex2 (15) Del1 (16) Del2 (17) MfgCost (18) OrdCost (19) MfgTime (20) * **

(1)

(2)

(3)

(4)

(5)

(6)

(7)

(8)

(9)

(10)

(11)

(12)

(13)

(14)

(15)

(16)

(17)

(18)

(19)

(20)

– .663** .537** .635** .590** .313** .325** .304** .390** .347** .251** .223** .206** .152** .165** .164** .192** .106* .094* .184**

– .593** .530** .576** .353** .391** .343** .386** .379** .329** .131** .217** .209** .175** .159** .139** ,072 ,065 .097*

– .568** .592** .402** .401** .368** .354** .372** .313** .148** .211** .232** .217** .180** .191** .121** .114* .122**

– .656** .348** .325** .298** .435** .326** .313** .240** .217** .228** .196** .151** .195** ,088 .124** .138**

– .365** .404** .368** .411** .347** .324** .223** .235** .184** .163** .152** .192** ,061 ,083 ,086

– .666** .591** .445** .410** .388** .205** .195** .140** .151** .153** .159** .123** .152** .129**

– .683** .418** .464** .420** .186** .249** .193** .164** .187** .196** .127** .176** .112*

– .424** .488** .478** .152** .224** .132** .158** .126** .148** .133** .160** .102*

– .696** .620** .217** .197** .162** .168** .182** .206** ,082 ,074 ,077

– .610** .157** .192** .147** .162** .160** .200** .126** .153** .104*

– .186** .197** .230** .237** .187** .212** .129** .143** .117*

– .714** .418** .360** .412** .464** .239** .226** .319**

– .395** .339** .427** .520** .196** .223** .267**

– .595** .364** .420** .311** .194** .313**

– .385** .350** .206** .169** .264**

– .741** .250** .246** .423**

– .264** .246** .387**

– .587** .475**

– .385**

–

Pearson correlation coefficients significant at p¼ .05 level. Pearson correlation coefficients significant at p ¼ .01 level.

Fig. 4. SEM results with distinct customer and supplier integration measures.

which further indicates that the initial SEM model with an aggregate, second-order external integration construct is more correct. In summary, results of the SEM model with two separate external integration constructs indicate that: 1. There is a strong positive link between internal integration and both dimensions of external supply chain integration; 2. The link between external integration (customer and supplier integration) and performance becomes much weaker (and in three of the four cases insignificant); 3. Due to the weakening effect of the two external integration dimensions on performance, the direct link between internal integration and differentiation performance becomes significant.

References Almeida, P., Song, J., Grant, R.M., 2002. Are firms superior to alliances and markets? An empirical test of cross-border knowledge building. Organ. Sci. 13 (2), 147–161. Ambos, T.C., Ambos, B., Schlegelmilch, B.B., 2006. Learning from foreign subsidiaries: an empirical investigation of headquarters' benefits from reverse knowledge transfers. Int. Bus. Rev. 15 (3), 294–312. Andersson, U., Forsgren, M., Holm, U., 2001. Subsidiary embeddedness and competence development in MNCs – a multi-level analysis. Organ. Stud. 22 (6), 1013–1034. Andersson, U., Forsgren, M., Holm, U., 2002. The strategic impact of external networks: subsidiary performance and competence development in the multinational corporation. Strateg. Manag. J. 23 (11), 979–996. Anh, P.T.T., Baughn, C.C., Hang, N.T.M., Neupert, K.E., 2006. Knowledge acquisition from foreign parents in international joint ventures: an empirical study in Vietnam. Int. Bus. Rev. 15 (5), 463–487. Argote, L., McEvily, B., Reagans, R., 2003. Managing knowledge in organizations: an integrative framework and review of emerging themes. Manag. Sci. 49 (4), 571–582. Bartlett, C.A., Ghoshal, S., 1989. Managing across borders: the transnational solution. Harvard Business School Press,, Boston.

Bessant, J., Kaplinsky, R., Lamming, R., 2003. Putting supply chain learning into practice. Int. J. Oper. Prod. Manag. 23 (2), 167–184. Birkinshaw, J., 1996. How multinational subsidiary mandates are gained and lost. J. Int. Bus. Stud. 27 (3), 467–495. Blome, C., Schoenherr, T., Eckstein, D., 2014. The impact of knowledge transfer and complexity on supply chain flexibility: a knowledge-based view. Int. J. Prod. Econ. 147, 307–316. Chaturvedi, R., Dutta, S., 2005. Knowledge Management Practices at Toyota Motors. ICMR Center for Management Research, Hyderabad,, India, Case study no. 905031-1, (distributed by The Case Center). Cheng, Y., Farooq, S., Johansen, J., 2011. Manufacturing network evolution: a manufacturing plant perspective. Int. J. Oper. Prod. Manag. 31 (12), 1311–1331. Cohen, W.M., Levinthal, D.A., 1990. Absorptive capacity: a new perspective on learning and innovation. Adm. Sci. Q. 35 (1), 128–152. Colotla, I., Shi, Y., Gregory, M.J., 2003. Operation and performance of international manufacturing networks. Int. J. Oper. Prod. Manag. 23 (10), 1184–1206. Dyer, J.H., Nobeoka, K., 2000. Creating and managing a high-performance knowledge-sharing network: the Toyota case. Strateg. Manag. J. 21 (3), 345–367.

K. Demeter et al. / Int. J. Production Economics 182 (2016) 73–85

Easterby-Smith, M., Lyles, M.A., Tsang, E.W.K., 2008. Inter-organizational knowledge transfer: current themes and future prospects. J. Manag. Stud. 45 (4), 677–690. Feldmann, A., Olhager, J., 2008. Internal and external suppliers in manufacturing networks – an empirical analysis. Oper. Manag. Res. 1 (2), 141–149. Feldmann, A., Olhager, J., 2013. Plant roles: site competence bundles and their relationships with site location factors and performance. Int. J. Oper. Prod. Manag. 33 (6), 722–744. Ferdows, K., 2006. Transfer of changing production know-how. Prod. Oper. Manag. 15 (1), 1–9. Ferraris, A., 2014. Rethinking the literature on ’multiple embeddedness’ and subsidiary-specific advantages. Multinatl. Bus. Rev. 22 (1), 15–33. Figueiredo, P.N., 2011. The role of dual embeddedness in the innovative performance of MNE subsidiaries: evidence from Brazil. J. Manag. Stud. 48 (2), 417–440. Flynn, B.B., Huo, B., Zhao, X., 2010. The impact of supply chain integration on performance: a contingency and configuration approach. J. Oper. Manag. 28 (1), 58–71. Foss, N.J., Pedersen, T., 2002. Transferring knowledge in MNCs: the role of sources of subsidiary knowledge and organizational context. J. Int. Manag. 8 (1), 49–67. Frohlich, M.T., Westbrook, R., 2001. Arcs of integration: an international study of supply chain strategies. J. Oper. Manag. 19 (2), 185–200. Frost, T.S., Birkinshaw, J.M., Ensign, P.C., 2002. Centers of excellence in multinational corporations. Strateg. Manag. J. 23 (11), 997–1018. Fusco, J.P., Spring, M., 2003. Flexibility versus robust networks: the case of the Brazilian automotive sector. Integr. Manuf. Syst. 14 (1), 26–35. Giménez, C., Ventura, E., 2005. Logistics-production, logistics-marketing and external integration: their impact on performance. Int. J. Oper. Prod. Manag. 25 (1), 20–38. Golini, R., Deflorin, P., Scherrer-Rathje, M., 2014. Enhancing operational performance in production subsidiaries: balancing autonomy, leveraging embeddedness. In: Proceedings of the EurOMA Conference, Palermo. Greenwood, R., Morris, T., Fairclough, S., Boussebaa, M., 2010. The organizational design of transnational professional service firms. Organ. Dyn. 39 (2), 173–183. Gupta, A.K., Govindarajan, V., 2000. Knowledge flows within multinational corporations. Strateg. Manag. J. 21 (4), 473–496. Hair, J.F., Black, W.C., Babin, B.J., Anderson, R.E., 2010. Multivariate Data Analysis. Pearson Prentice Hall, Upper Saddle River, NJ. Hallin, C., Holm, U., Sharma, D.D., 2011. Embeddedness of innovation receivers in the multinational corporation: effects on business performance. Int. Bus. Rev. 20 (3), 362–373. He, Q., Ghobadian, A., Gallear, D., 2013. Knowledge acquisition in supply chain partnerships: the role of power. Int. J. Prod. Econ. 141 (2), 605–618. Ho, Y.C., 2014. Multilateral knowledge transfer and multiple embeddedness. Multinatl. Bus. Rev. 22 (2), 155–175. Horn, P., Scheffler, P., Schiele, H., 2014. Internal integration as a pre-condition for external integration in global sourcing: a social capital perspective. Int. J. Prod. Econ. 153, 54–65. Inkpen, A.C., Tsang, E., 2005. Social capital, networks, and knowledge transfer. Acad. Manag. Rev. 30 (1), 146–165. Jansen, J.J.P., Van den Bosch, F.A.J., Volberda, H.W., 2005. Managing potential and realized absorptive capacity: how do organizational antecedents matter? Acad. Manag. J. 48 (6), 999–1015. Kogut, B., Zander, U., 1993. Knowledge of the firm and the evolutionary theory of the multinational corporation. J. Int. Bus. Stud. 24 (4), 625–645. Lane, P.J., Lubatkin, M., 1998. Relative absorptive capacity and interorganizational learning. Strateg. Manag. J. 19 (5), 461–477. Lane, P.J., Salk, J.E., Lyles, M.A., 2001. Absorptive capacity, learning, and performance in international joint ventures. Strateg. Manag. J. 22 (12), 1139–1161. Luo, Y., 2005. Toward coopetition within a multinational enterprise: a perspective from foreign subsidiaries. J. World Bus. 40 (1), 71–90. Mahnke, V., Pedersen, T., Venzin, M., 2005. The impact of knowledge management on MNC subsidiary performance: the role of absorptive capacity. Manag. Int. Rev. 45 (special issue), 101–119. Maiga, A.S., Nilsson, A., Ax, C., 2015. Relationships between internal and external information systems integration, cost and quality performance,and firm profitability. Int. J. Prod. Econ. 169, 422–434. Marksberry, P., Badurdeen, F., Gregory, B., Kreafle, K., 2010. Management directed kaizen: Toyota's Jishuken process for management development. J. Manuf. Technol. Manag. 21 (6), 670–685. Meyer, K.E., Mudambi, R., Narula, R., 2011. Multinational enterprises and local contexts: the opportunities and challenges of multiple embeddedness. J. Manag. Stud. 48 (2), 235–252. Minbaeva, D., Pedersen, T., Björkman, I., Fey, C.F., Park, H.J., 2003. MNC knowledge transfer, subsidiary absorptive capacity, and HRM. J. Int. Bus. Stud. 34 (6),

85

586–599. Minbaeva, D., 2007. Knowledge transfer in multinational corporations. Manag. Int. Rev. 47 (4), 567–593. Modi, S.B., Mabert, V.A., 2007. Supplier development: Improving supplier performance through knowledge transfer. J. Oper. Manag. 25 (1), 42–64. Monteiro, L.F., Arvidsson, N., Birkinshaw, J., 2008. Knowledge flows within multinational corporations: explaining subsidiary isolation and its performance implications. Organ. Sci. 19 (1), 90–107. Mudambi, R., Navarra, P., 2004. Is knowledge power? Knowledge flows, subsidiary power and rent-seeking within MNCs. J. Int. Bus. Stud. 35, 385–406. O'Brien, R.M., 2007. A caution regarding rules of thumb for variance inflation factors. Qual. Quant. 41 (5), 673–690. Paulraj, A., Lado, A.A., Chen, I.J., 2008. Inter-organizational communication as a relational competency: antecedents and performance outcomes in collaborative buyer-supplier relationships. J. Oper. Manag. 26 (1), 45–64. Podsakoff, P.M., MacKenzie, S.B., Lee, J.Y., Podsakoff, N.P., 2003. Common method biases in behavioral research: a critical review of the literature and recommended remedies. Journal of Applied Psychology 88 (5), 879–903. Porter, M.E., 1985. Competitive Advantage: Creating and Sustaining Superior Performance. The Free Press, New York. Prahalad, C.K., Doz, Y., 1987. The multinational mission: balancing local demands and global vision. Free Press, New York. Prajogo, D., Olhager, J., 2012. Supply chain integration and performance: the effects of long-term relationships, information technology and sharing, and logistics integration. Int. J. Prod. Econ. 135 (1), 514–522. Revilla, E., Knoppen, D., 2015. Building knowledge integration in buyer-supplier relationships: the critical role of strategic supply management and trust. Int. J. Oper. Prod. Manag. 36 (10), 1408–1436. Rhodes, J., Lok, P., Hung, R.Y., Fang, S.C., 2008. An integrative model of organizational learning and social capital on effective knowledge transfer and perceived organizational performance. J. Workplace Learn. 20 (4), 245–258. Rudberg, M., Olhager, J., 2003. Manufacturing networks and supply chains: an operations strategy perspective. Omega 31 (1), 29–39. Rungtusanatham, M., Miller, J.W., Boyer, K.K., 2014. Theorizing, testing, and concluding for mediation in SCM research: tutorial and procedural recommendations. J. Oper. Manag. 32 (3), 99–113. Schmenner, R.W., Swink, M.L., 1998. On theory in operations management. J. Oper. Manag. 17 (1), 97–113. Schmid, S., Schurig, A., 2003. The development of critical capabilities in foreign subsidiaries: disentangling the role of the subsidiary's business network. Int. Bus. Rev. 12 (6), 755–782. Schoenherr, T., Swink, M., 2012. Revisiting the arcs of integration: cross-validations and extensions. J. Oper. Manag. 30 (1), 99–115. Shah, R., Ward, P.T., 2007. Defining and developing measures of lean production. J. Oper. Manag. 25 (4), 785–805. Song, J., 2014. Subsidiary absorptive capacity and knowledge transfer within multinational corporations. J. Int. Bus. Stud. 45 (1), 73–84. Swink, M., Narasimhan, R., Wang, C., 2007. Managing beyond the factory walls: effects of four types of strategic integration on manufacturing plant performance. J. Oper. Manag. 25 (1), 148–164. Szász, L., Scherrer-Rathje, M., Deflorin, P., 2016. Benefits of internal manufacturing network integration: the moderating effect of country context. Int. J. Oper. Prod. Manag. 36 (7), 757–780. Tsai, W., 2001. Knowledge transfer in intraorganizational networks: effects of network position and absorptive capacity on business unit innovation and performance. Acad. Manag. J. 44 (5), 996–1004. Van der Vaart, T., van Donk, D.P., 2008. A critical review of survey-based research in supply chain integration. Int. J. Prod. Econ. 111 (1), 42–55. Van Wijk, R., Jansen, J.J.P., Lyles, M.A., 2008. Inter- and intra-organizational knowledge transfer: a meta-analytic review and assessment of its antecedents and consequences. J. Manag. Stud. 45 (4), 830–853. Vereecke, A., Van Dierdonck, R., De Meyer, A., 2006. A typology of plants in global manufacturing networks. Manag. Sci. 52 (11), 1737–1750. Wilcox King, A., Zeithaml, C.P., 2003. Measuring organizational knowledge: a conceptual and methodological framework. Strateg. Manag. J. 24 (8), 763–772. Yamin, M., Andersson, U., 2011. Subsidiary importance in the MNC: what role does internal embeddedness play? Int. Bus. Rev. 20 (2), 151–162. Zahra, S.A., George, G., 2002. Absorptive capacity: a review, reconceptualization, and extension. Acad. Manag. Rev. 27 (2), 185–203. Zander, U., Kogut, B., 1995. Knowledge and the speed of the transfer and imitation of organizational capabilities: An empirical test. Organization Science 6 (1), 76–92. Zhao, X., Huo, B., Selen, W., Yeung, J.H.Y., 2011. The impact of internal integration and relationship commitment on external integration. J. Oper. Manag. 29 (1), 17–32.