THE INFLUENCE OF NATIONAL CULTURE ON THE FORMATION OF MULTI-PARTNER R&D ALLIANCES
By
Charlotte Sophie Kumm
Master’s Thesis MSc BA Strategic Innovation Management
University of Groningen Faculty of Economics and Business
Supervisor: I. (Isabel) Estrada Vaquero Co-assessor: P.M.M. (Pedro) de Faria
18th of June 2015
C.S. Kumm s2527820
c.s.kumm@student.rug.nl
Abstract: This study examines the influence of national culture on a firm’s likelihood to form a multi-partner R&D (MR&D) alliance. To investigate the so far underexplored phenomenon of MR&D alliance formation, the resource-based view and transaction cost economics are used as guiding perspectives. Hypotheses are established based on previous research in line with the two complementary theories. It is proposed that firms are more likely to form MR&D alliances if they come from societies that have low levels of individualism and masculinity and a high level of uncertainty avoidance. An analysis of 124 firms is conducted, including 62 firms that are involved in MR&D alliances and 62 firms that are not, to examine if and in how far the national culture of these firms influences their decision to enter MR&D alliances. Despite clear differences between the two subsamples, none of the hypotheses can be confirmed. Nevertheless, interesting findings emerge. Without including the control variables, a high level of individualism appears to have a negative effect on MR&D alliance formation, whereas a high degree of uncertainty avoidance increases firms’ likelihood to form MR&D alliances. In addition, a U-shaped relationship is found between a firm’s level of individualism and MR&D alliance formation. Until a certain level of individualism, firms are less likely to form MR&D alliances. After that point, however, firms become more likely again to form this type of alliance.
Table of contents
List of figures i
List of tables i
1. Introduction 1
2. Overview of MR&D alliances and national culture 3
2.1 MR&D alliances 3
2.2 National culture 4
3. Hypothesis development 6
3.1 Resource-based view on MR&D alliance formation 6
3.2 Transaction cost economics on MR&D alliance formation 7
3.3 Conceptual model and hypotheses 8
4. Methodology 13
4.1 Sample and data sources 13
4.2 Method of analysis 15 4.3 Measures 15 4.3.1 Dependent variable 15 4.3.2 Independent variables 16 4.3.3 Control variables 16 5. Results 18
5.1 Descriptive and correlation statistics 18
5.1.1 Alliance level 18 5.1.2 Firm level 20 5.2 Hypotheses testing 24 5.3 Post-hoc analyses 27 6. Discussion of findings 29 7. Conclusion 32 7.1 Implications 33 7.1.1 Theoretical implications 33 7.1.2 Practical implications 33
7.2 Limitations and future research 34
References 36
i List of figures
Figure 1: Conceptual model ...8
List of tables Table 1: Summary of cultural dimensions ...6
Table 2: Proposed influence of Hofstede’s cultural dimensions ... 12
Table 3: Comparison of matching samples on the independent variables... 15
Table 4: Summary of variables ... 18
Table 5: Descriptive statistics at alliance level ... 19
Table 6: Cross-tabulation results: Composition * Individualism ... 20
Table 7: Cross-tabulation results: Composition * Masculinity ... 20
Table 8: Cross-tabulation results: Composition * Uncertainty avoidance ... 20
Table 9: Descriptive statistics at firm level ... 22
Table 10: Correlation statistics... 23
Table 11: Logistic regression analysis – simple models ... 25
Table 12: Logistic regression analysis – full models ... 26
Table 13: Curvilinear relationships ... 27
1 1. Introduction
Multi-partner R&D alliances (henceforth MR&D alliances) are voluntarily formed associations between more than two partner firms that jointly execute R&D activities (Lavie et al., 2007). In light of constantly rising competition, entering MR&D alliances represents a powerful strategy for firms to deal with innovation challenges and achieve competitive advantages (Doz and Hamel, 1998; Mothe and Quelin, 2001). Scholars have recognised the increasing relevance of MR&D alliances and research has started to investigate this phenomenon (e.g. Lavie et al., 2007; Li, 2013; Mishra et al., 2015). In today’s interconnected world, firms increasingly form MR&D alliances with partners from different countries to react to shortened product life cycles and rising R&D demands (Goerzen and Beamish, 2005). Both the propensity of forming MR&D alliances and the nature of the cooperation are influenced by the national culture of firms (Steensma et al., 2000). Nevertheless, research remains scarce on why firms with a certain nationality are more prone than others to form MR&D alliances, thereby dealing with more complexity and distinctive dynamics. Hence, it is important to study the impact of national culture on the formation of MR&D alliances. In this research, national culture is conceptualised by three cultural dimensions of Hofstede (1991), namely individualism, masculinity, and uncertainty avoidance.
2
governance structure (Killing, 1983; Li et al., 2011). Relationships between multiple partners are not as clearly drawn as between two partners, which relates to the concept of reciprocity. If firms do not cooperate as agreed upon, partners may not recognise this (Ekeh, 1974; Li, 2013; Zeng and Chen, 2003). Hence, the interaction across multiple partners results in distinctive dynamics (Lavie et al., 2007) and it is unclear whether the findings of previous studies on dyadic alliances can be applied to MR&D alliances (Li et al., 2011).
To address the lack of research concerning the impact of national culture on MR&D alliance formation, this study investigates the following research question: How does a firm’s national culture influence its likelihood to form a MR&D alliance? Answering this question provides valuable input and knowledge for current theory, managers as well as policy makers. In times of an increasing number of international MR&D alliances it is essential to understand why firms from certain countries are more likely than others to take on the challenges and dive into the complicated relationship with multiple partners. To guide this study, the resource-based view (RBV) and the transaction cost economics (TCE) are taken as guiding perspectives since they constitute the main theoretical framework in the alliance literature (Beamish and Kachra, 2004).
3
uncertainty avoidance increases firms’ likelihood to form MR&D alliances. In addition, a U-shaped relationship is found between a firm’s level of individualism and MR&D alliance formation. With a certain level of individualism, firms become more likely again to form this type of cooperative agreement.
The remainder of this paper is structured as follows: First, a literature review provides background information on MR&D alliances and national culture. Second, hypotheses are developed in line with the RBV and TCE. Third, the methods used in this study are described and research findings documented and further discussed. Finally, conclusions are drawn, implications established, and limitations as well as suggestions for future research provided.
2. Overview of MR&D alliances and national culture 2.1 MR&D alliances
4
Despite the growing popularity of MR&D alliances in recent years most research still focuses on firms entering dyadic alliances (e.g. Nielsen, 2003; Rothaermel and Boeker, 2008; Yu et al., 2013). Studies concerning which firms enter MR&D alliances and why remain scarce (Lavie et al., 2007, Li et al., 2011). Some emerging research has started to investigate firms’ motivations and incentives to form MR&D alliances (e.g. Li, 2013; Yin and Wu, 2003). Mitchell et al. (2002) studied which type of alliance firms typically form to combine different types of resources and which governance mechanism they use. The authors included other functions next to R&D and only involved the influence of geographic location instead of national culture. Thorgren et al. (2012) investigated how small firms can benefit from entering MR&D alliances but they did not include the influence of national culture on firms’ propensity to form this type of cooperative agreement. Li (2013) examined the influence of market uncertainty on the likelihood of new ventures to form MR&D alliances and found an inverted U-shaped relationship between these two concepts. Even though this study is one of the only ones that investigates the formation of MR&D alliances, it also neglects the impact of national culture. The research at hand hence differs from previous studies and extends them by incorporating the influence of national culture on the formation of MR&D alliances.
2.2 National culture
5
confirmed by Koghut and Singh (1988), who state that there are several studies that provide theoretical and empirical support about the relationship between a firm’s home country and its entry mode (e.g. Puxty, 1979).
National culture is an important determinant of choosing a cooperation agreement (as for instance MR&D alliances) since it influences managers’ perceptions concerning the possible costs and uncertainty involved in entering a cooperation (Kogut and Singh, 1988). The cultural context, in which a firm operates, affects the management processes as well as managerial decision-making (Hofstede, 1994; Steenkamp and Geyskens, 2012). Hence, whether a firm decides to enter a MR&D alliance is largely determined by its cultural background since “national-cultural priorities will encourage the activation of organizational choices that are in line with these priorities and conducive in maintaining them, while organizational choices that run counter to these cultural priorities are discouraged” (Steenkamp and Geyskens, 2012, pp. 255-256). Country patterns might exist on the likelihood of firms to engage in one type of entry mode instead of the other (Kogut and Singh, 1988).
In this study, national culture is conceptualised by Hofstede‘s (1991) three dimensions of culture, namely individualism, masculinity, and uncertainty avoidance1. Individualism is the degree to which individual interests or values are stressed in a society (Shenkar and Zeira, 1992). In individualistic societies people are expected to look after themselves and their immediate families. They are not very group oriented and enjoy being independent (Steensma et al., 2000). Masculinity refers to how aggressive and competitive people are. Masculine societies tend to be more assertive and less modest than feminine societies (Steensma et al.,
1
6
2000). Uncertainty avoidance is the “extent to which the members of a culture feel threatened by uncertain or unknown situations" (Hofstede, 1991, p. 113). People from countries scoring high on this dimension see change as something dangerous and feel comfortable with structure and security. Table 1 summarises the different dimensions.
Table 1: Summary of cultural dimensions
Dimension Description
Individualism The extent to which people emphasise individual interests and aim at being independent from groups (Hofstede, 1991).
Masculinity The degree to which people are assertive and competitive compared to being more modest (Steensma et al., 2000).
Uncertainty avoidance The extent to which people feel threatened by uncertainty or unknown situations (Hofstede, 2001).
3. Hypothesis development
The resource-based view (RBV) and transaction cost economics (TCE) are used as theoretical foundations to examine the influence of national culture on MR&D alliance formation. They have been widely used in previous research to explain why firms engage in cooperative partnerships (e.g. Goerzen and Beamish, 2005; Gulati, 1995; Parkhe, 1991) and constitute the main theoretical framework in the alliance formation literature (Beamish and Kachra, 2004). The two theories provide complementary arguments concerning the benefits and costs of entering MR&D alliances. Whereas arguments for the benefits of MR&D alliances are usually grounded in the RBV, arguments concerning the costs involved in entering a MR&D alliance are typically based on the TCE (Li, 2013). As Beamish and Kachra (2004) note, “it is an empirical as much as a theoretical question whether the potential benefits of better resources offset the costs of managing [the] complex organization form” of the MR&D alliance (p. 109).
3.1 Resource-based view on MR&D alliance formation
7
competencies and competitive advantages (Das and Teng, 2000; Lin and Darnall, 2015). It is especially valuable if different resources are needed and if these cannot be efficiently obtained through, for instance, market exchange or acquisitions (Das and Teng, 2000). Firms might also strive to force industry-wide changes and need a larger number of alliance partners for their resources to be more visible and stronger (Lin and Darnall, 2015). By forming MR&D alliances firms can spread their innovation risks across the multiple firms and thereby enhance their chance of survival (Li, 2013). Another motive is to increase organisational learning (Lin and Darnall, 2015). By collectively investigating the market’s trends and emerging technologies, multiple alliance partners can change business practices and initiate new technology generations (Lin and Darnall, 2015).
Considering the influence of national culture it can be argued that some nationalities might be more prone than others to combine resources with multiple partner firms to create value and sustained competitive advantages (Steensma et al., 2000). Some cultural backgrounds might result in firms feeling more threatened by uncertainty and therefore might choose to enter MR&D alliances to improve their market position (Stevens and Dykes, 2013).
3.2 Transaction cost economics on MR&D alliance formation
8 Control Variables - Firm size - Firm age - Founding year - Industry - Alliance experience - Type alliance experience - Public status
- R&D intensity
- Type alliance in same year
Uncertainty avoidance H1: – H2: – H3: +/- Masculinity Individualism MR&D alliance formation
behavioural uncertainty (Steenkamp and Geyskens, 2012). In MR&D alliances the latter dimension is especially present. With two partners, a reciprocal relationship exists in which one partner knows what the other one does and does not do (Ekeh, 1974). In MR&D alliances, however, firms are less sure about partners’ activities and hence behavioural uncertainty is present, leading to increased transaction costs.
Concerning the influence of national culture economists highlight the rational and calculative background of firms’ choices to enter alliances (Steenkamp andGeyskens, 2012). Repeated interactions lead to higher levels of trust and firms might be more likely to form MR&D alliances despite the originally higher transaction costs. The perceptions of what constitutes high transaction costs as well as the fear of opportunistic behaviour might vary per country (Steensma et al., 2000). Some firms might perceive transaction costs as lower than others when entering MR&D alliances.
3.3 Conceptual model and hypotheses
In this section hypotheses are formulated which propose relationships between firms’ orientations on Hofstede’s cultural dimensions and their likelihood to form MR&D alliances instead of any other relational mode of entry (such as dyadic alliance, joint venture or licensing). The RBV and TCE are taken as guiding perspectives. Figure 1 visualises the proposed relationships and Table 2 summarises the arguments made.
9
Individualism. Societies with a high level of individualism aim at being independent from relational connections (Steensma et al., 2000). In line with the RBV, it can be argued that firms from individualistic societies form MR&D alliances only to gain access to partners’ resources, which then allows them to be more independent afterwards. Since opportunities and rewards stem from individual performance that may discourage collaboration and sharing of new information (Zahra et al., 2004), it is unlikely that firms with an individualistic background form MR&D alliances. Firms scoring high on individualism act as though they are defined as an entity, which consists of a single firm (Wagner, 1995). They hence look for the necessary resources within the firm boundaries and aim at developing them on their own. Contrary, more collectivistic firms act as though they are defined as an entity consisting of several partners and thus forming a group (Wagner, 1995). Firms from these cultures believe that the best solutions can be found through collaborative effort (Zahra et al., 2004). Hence, firms that come from an individualistic society are less likely to engage in alliances with multiple partners. According to Steensma et al. (2000), these firms are more attracted by arm’s length and flexible arrangements in which their independence is not threatened and in which they can keep their resources internally.
10
due to a lack of trust since no repeated interactions occur (Gulati, 1995). Based on these relatively high transaction costs, it is expected that firms from individualistic societies are not likely to enter MR&D alliances. Following the arguments in line with both the RBV and TCE, it is hypothesised that:
H1: The higher a country’s level of individualism, the less likely the firms from that
country are to engage in MR&D alliances.
Masculinity. Societies with a more masculine background are competitive and perceive the world in terms of winners and losers (Marino et al., 2002). Contrary, feminine societies are more in favour of cooperation and mutual gains – hence win-win situations are possible (Steensma et al., 2000). In line with the RBV, it can be argued that firms with a masculine background would rather not enter MR&D alliances to access complementary resources since this might not enable them to be superior and dominant over their partners. It is essential that firms stand out from the group instead of being average and blending in (Steensma et al., 2000). Since forming MR&D alliances reduces competition (Steensma et al., 2000), firms would have fewer opportunities to present themselves as clear winners in the marketplace and are hence less likely to form cooperative agreements. The resources that a firm owns are crucial and entering MR&D alliances might be perceived as weak since it shows that the firm misses resources and cannot be successful alone.
11
firms from masculine societies appear to rather choose hierarchical entrance modes in which centralisation of authority is present (Steenkamp and Geyskens, 2012). It is hypothesised that:
H2: The higher a country’s level of masculinity, the less likely the firms from that
country are to engage in MR&D alliances.
Uncertainty avoidance. Societies vary in their degree to which they feel comfortable or threatened by uncertainty. In line with the RBV it can be argued that firms from societies that score high on uncertainty avoidance are more likely to form MR&D alliances to achieve their strategic goals as well as share the risks of possible failure (Marino et al., 2002). This is especially the case in cooperative partnerships that involve the joint execution of R&D since more resources and higher risks are involved when innovating (Marino et al., 2002). Firms with a high level of uncertainty avoidance form MR&D alliances to access the complementary resources needed and do not refrain from cooperative strategies (Steensma et al., 2000). To avoid ambiguity and risks as well as establish security for the present and the future (Stevens and Dykes, 2013), these firms are more inclined to form MR&D alliances to ensure access to additional resources needed to better react to changes in the environment. In order to minimise the potential threat of uncertainty in the future, these firms prefer to enter MR&D alliances to reduce the risks of not achieving their goals due to market hazards and changing resource availabilities (Marino et al., 2002). Hence, it is hypothesised that:
H3a: The higher a country’s level of uncertainty avoidance, the more likely the firms from that country are to engage in MR&D alliances to secure access to complementary resources.
12
opportunistic behaviour. The inherently risky and structurally volatile nature of MR&D alliances further enhances the relational uncertainty perceived by partner firms (Marino et al., 2002). Firms scoring high on uncertainty avoidance would fear that the goals of partners change and might not be in line with their own goals anymore (Marino et al., 2002). Additional contracts would need to be set up to prevent partners’ opportunistic behaviour and to decrease the uncertainty about not fulfilling the goals of the partnership (Marino et al., 2002). These contracts and extra monitoring activities would lead to higher transaction costs, which are perceived as negative in line with the TCE (Steesma et al., 2000). Thus, firms from these societies appear unlikely to enter MR&D alliances and instead would try to internalise activities. Following these arguments it is hypothesised that:
H3b: The higher a country’s level of uncertainty avoidance, the less likely the firms from that country are to engage in MR&D alliances due to the fear of partners’ opportunistic behaviour as well as increased costs.
Table 2: Proposed influence of Hofstede’s cultural dimensions
Dimension RBV Arguments TCE Arguments
Individualism - Firms from individualistic societies desire independence and would form MR&D alliances only to gain access to resources which eventually would enable them to be more independent from any group.
- Firms from individualistic societies are inherently guided by self-interest and might feel more threatened by relational risks potentially occurring within MR&D alliances than firms from collectivistic societies. Masculinity - Firms with masculine backgrounds do
not enter MR&D alliances to access complementary resources since this might not enable them to be superior and dominant over their partners.
- Since firms from masculine societies adhere to a zero-sum perspective they are rather unlikely to perceive MR&D alliances as win-win situations and do not enter this type of collaboration. Uncertainty
avoidance
+ Firms need and aim at accessing additional resources if they are from societies with a high level of uncertainty avoidance to secure the access to these complementary resources.
13 4. Methodology
4.1 Sample and data sources
Information on MR&D alliance formation is collected from the Securities Data Corporation (SDC) Database. This database is suitable since it provides data on many different sectors as well as various agreement types such as joint ventures, strategic alliances, R&D, marketing and manufacturing agreements (Schilling, 2009).
Initially, information is collected from the SDC Database on an alliance level. In the years 2000 until 2010, 347 MR&D alliances were formed. From these 347 MR&D alliances, a list of firms is compiled that meet the following criteria. First, firms operate in high-tech industries, in which MR&D alliances are especially prevalent, as identified by the AeA, an association of high-tech firms representing the technology industry (Li, 2013). Based on the AeA’s categorisation of high-technology manufacturing, high-technology services as well as software and computer-related services, firms belonging to these categories and having the according SIC codes (357, 365, 366, 367, 381, 382, 384, and 386; 481, 482, 484, and 489; or 737) are retained in the sample (Li, 2013). Second, to ensure a complete dataset, firms for which the SDC database does not provide the number of employees at the time of alliance formation, and which cannot be obtained via the Orbis database, are excluded from the dataset. Third, firms have a public status of being either private or public to ensure that they constitute an individually owned and operated entity (Li, 2013). 136 firms meet the described criteria and the time period reduces to 2000 until 2009 since no firms that match the criteria were formed in 2010.
14
listed several times in the dataset for one focal year, they are only retained once to avoid duplication and biased results. Here, 83 firms remain. Subsequently, if firms formed several MR&D alliances in the sample time frame of 2000 until 2009, they are considered only once with their first activity and additional activities in subsequent years are excluded. This leads to the year 2008 being excluded from the dataset since no MR&D alliances were formed in this year by a firm that fulfils all mentioned criteria. The ultimate sample of firms entering MR&D alliances consists of 62 firms.
Finally, the sample is matched with a similar sample of firms not involved in MR&D alliances. The selection criteria used above are also applied to establish this sample and thus the firms (1) have an R&D agreement flag; (2) formed cooperative agreements in the period between 2000 and 2009; (3) operate in high tech industries with the above-mentioned SIC codes; (4) their number of employees at the time of forming the agreement is provided in the SDC or Orbis database; and (5) they have either a private or public status. Moreover, the sample falls within the same minimum and maximum numbers of age and size as the sample of firms that formed MR&D alliances and firms that established a cooperative agreement in the year 2008 are excluded from the dataset to match the main sample. Additionally, firms are only retained once in the dataset with their first activity and subsequent activities are excluded. From the remaining firms a random sample of 62 firms is generated to establish a total dataset with an equal distribution of firms involved/not involved in MR&D alliances.
15
Table 3: Comparison of matching samples on the independent variables
Mean (overall sample n=124)
Mean (sample of firms involved in MR&D
alliance n=62)
Mean (sample of firms not involved in MR&D partner alliance n=62) Two-tailed t-test of sample differences Individualism 71,15 64,95 77,34 2,675** Masculinity 63,48 64,50 62,45 -.681 Uncertainty avoidance 57,68 62,18 53,18 -2,597* * p<0.05; ** P<0.01. 4.2 Method of analysis
Testing the hypotheses requires modelling the probability that a firm enters a MR&D alliance among a range of potential other agreement types. A binomial logistic regression model is applied, using the statistical software SPSS. A logistic regression analysis is suitable for this study since the dependent variable is of dichotomous nature consisting of two possible options, namely ‘1’ if a firm forms a MR&D alliance in a focal year and ‘0’ if it enters any other type of cooperative R&D agreement (Gulati, 1999). The model is estimated as follows:
Probability (formation of MR&D alliance = 1)
= β0 + β1 Individualism + β2 Masculinity + β3 Uncertainty avoidance + β4 Firm size + β5 Firm age + β6 Founding year + β7 Industry
+ β8 Alliance experience + β9 Type alliance experience + β10 Public status + β11 R&D intensity + β12 Type alliance in same year + εi
where εi is the random error for each observation. 4.3 Measures
4.3.1 Dependent variable
16
alliances include those formed with any partner, be it a previous or new partner. The unit of analysis is the individual firm and whether or not it forms a MR&D alliance.
4.3.2 Independent variables
Hofstede’s (1980, 2001) measures are utilised to develop the dimensions of individualism (IND), masculinity (MASC), and uncertainty avoidance (UA)2 (Steensma et al., 2000; Stevens and Dykens, 2013). The numerical values calculated by Hofstede (1980, 2001) are taken as the measures for each of the three national culture values. The use of these measures is appropriate since they involve 53 countries from both industrialised and developing countries, they were drawn from factor analysis, and researchers have supported the validity of the measures by correlating them with the indices of other researchers (Steensma et al., 2000). Hofstede’s measures are standardised and range from approximately 0 to 100. The 19 countries involved in this study exhibit a great variance across the three dimensions (see Appendix A.1).
4.3.3 Control variables
Several control variables at both firm and alliance level are included in this research to control for other possible factors influencing MR&D alliance formation.
Firm size and age. It is important to control for the size and age of firms entering MR&D alliances (Li, 2013). They can have an impact since the larger and older the firms are, the more resources they typically have and the easier they can access additional resources (Li, 2013). Firm size is measured by using the firms’ number of employees. Firm age is calculated by detracting the year of foundation from the year of alliance formation (Lin et al., 2012). Both firm size and age are used to indirectly control for firms’ organisational culture.
2
17
Founding year. Another important control variable is the founding year of the alliance. Due to external factors such as economic crises the founding year can influence firms’ propensity to enter MR&D alliances (Park et al., 2002).
Industry. Firms’ industries also need to be controlled for (Steensma et al., 2000). A dummy variable is created in which firms that operate in high-technology manufacturing industries (SIC codes: 357, 365, 366, 367, 381, 382, 384, and 386) are coded with a ‘1’ and those operating in high-technology service industries (SIC codes: 481, 482, 484, 489, and 737) are coded with a ‘0’ (Li, 2013).
Alliance experience. This control variable relates to the number of previous alliances that a firm had in the years prior to MR&D alliance formation. It is measured by counting a focal firm’s number of alliances in the five years prior to the year of alliance formation.
Type alliance experience. To also control for the type of firms’ alliance experience (Park et al., 2002) in the previous five years, a dummy variable is added concerning whether firms had experiences with multi-partner alliances (1) or not (0).
Public status. A control variable is included concerning the possible influence of a firm being public and hence listed in the stock market or private on its choice to form MR&D alliances (Reuer and Lahiri, 2014). Public firms could have had more access to needed resources compared to private firms (Stevens and Dykes, 2013). A dummy variable is created with firms being public (1) or private (0).
R&D intensity. It is essential to control for the R&D intensity of firms, which is measured by their amount of R&D expenses in US$ as provided in the Orbis database. Since the Orbis database usually only goes back to 2005, the available data on R&D expenses on the closest year of alliance formation is used.
18
between firms that only form MR&D or dyadic alliances versus firms that enter both dyadic and MR&D alliances. A final control variable examines whether firms are only involved in MR&D alliances in a focal year (1) or also in other agreement types (0). Table 4 provides a summary of the variables used in this research.
Table 4: Summary of variables
Variable Variable name Measure Source
Dependent Formation A firm’s choice to form either a MR&D alliance (1) or any other type of cooperative R&D agreement (0)
SDC database Explanatory IND Numerical values calculated by Hofstede for a country’s
dimension of individualism
The Hofstede Centre MASC Numerical values calculated by Hofstede for a country’s
dimension of masculinity
The Hofstede Centre UA Numerical values calculated by Hofstede for a country’s
dimension of uncertainty avoidance
The Hofstede Centre Control Firm size A firm’s number of employees SDC database
Firm age A firm’s years of existence at the point of alliance formation
Orbis database Founding year The year in which the alliance was formed SDC database Industry A dummy variable: If firms operate in high-technology
manufacturing (1) or other high-tech industries (0)
SDC database Alliance
experience
The number of a firm’s alliances in the five years prior to alliance formation
SDC database Type alliance
experience A dummy variable: If firms have experiences with multi-partner alliances (1) or not (0)
SDC database Public status A dummy variable: If firms are public (1) or private (0) SDC database R&D intensity A firm’s R&D expenses in the most recent year provided Orbis database Type alliance in
same year
A dummy variable: If firms are only involved in MR&D alliances (1) or also formed other agreements (0)
SDC database
5. Results
5.1 Descriptive and correlation statistics 5.1.1 Alliance level
19
alliances that involve cross-border participants, the cultural diversity ranges from 0.24 to 0.753. 13.6% of alliances have a low level of cultural diversity, 50% have a moderate level and 36.4% have a high level.
Table 5: Descriptive statistics at alliance level
Variable N Mean4 S.D. Min Max
Composition 27 18,5%(0), 81,5%(1) .396 0 1
Number of partners 27 3,33 .832 3 7
Number of nationalities 27 2,19 .834 1 4
Cultural diversity 22 .5132 .14160 0.24 0.75
Cultural diversity categories 22 13,6%(1), 50,0%(2), 36,4% (3) .685 1 3
To explore the composition of the alliances further, cross-tabulation analyses are done and the results presented in tables 6, 7 and 85. Looking at alliance composition from a focal firm’s perspective, it is examined with whom the firms from the 27 MR&D alliances have allied. Table 6 reveals that 88.9% of firms enter MR&D alliances with cross-border participants if the level of IND is low. Table 7 shows that if the level of MASC is low or high, 100% of the firms choose to enter MR&D alliances with cross-border participants instead of firms from their own nationality. Table 8 reveals that if the level of UA is high, 100% of firms enter MR&D alliances with firms from other countries instead of their own. Hence, firms enter MR&D alliances with cross-border participants if they have low levels of IND, low or high levels of MASC, and high levels of UA. Tables 6, 7 and 8 show that if the levels of IND, MASC, and UA are moderate (25%, 23.3%, and 33.3%), firms enter MR&D alliances with firms from their own country.
3
The cultural diversity within MR&D alliances is measured using the Blau Diversity Index, which is computed as 1-∑p2, p being the proportion of countries in each MR&D alliance (Zoogah et al., 2011). A value approaching 1 indicates heterogeneity whereas 0 stands for homogeneity (Blau, 1977). The calculated values are divided into three equally large categories of low (0.24-0.40), moderate (0.41-0.58), and high (0.59-0.75).
4
For dummy variables the frequencies instead of the means are provided.
5
20
Table 6: Cross-tabulation results: Composition * Individualism
Individualism
Total Low Moderate High
Composition No cross-border participants Count % within IND 1 11,1% 1 25,0% 5 23,8% 7 20,6% Cross-border participants Count % within IND 8 88,9% 3 75,0% 16 76,2% 27 79,4% Total Count % within IND 9 100,0% 4 100,0% 21 100,0% 34 100,0%
Table 7: Cross-tabulation results: Composition * Masculinity
Masculinity
Total Low Moderate High
Composition No cross-border participants Count % within MASC 0 0,0% 7 23,3% 0 0,0% 7 20,6% Cross-border participants Count % within MASC 2 100,0% 23 76,7% 2 100,0% 27 79,4% Total Count % within MASC 11 100,0% 30 100% 1 100,0% 34 100,0%
Table 8: Cross-tabulation results: Composition * Uncertainty avoidance
Uncertainty avoidance
Total Low Moderate High
Composition No cross-border participants Count % within UA 1 25,0% 6 33,3% 0 0,0% 7 20,6% Cross-border participants Count % within UA 75,0% 3 66,7% 12 100,0% 12 79,4% 27 Total Count % within UA 4 100,0% 18 100,0% 12 100,0% 34 100,0% 5.1.2 Firm level
After having explored the composition of the MR&D alliances, this section examines who forms MR&D alliances at the firm level. The main sample of firms both involved and not involved in MR&D alliances (n=124) is used for further analysis. Table 9 provides the descriptive statistics of the overall sample and the subsamples of firms involved/not involved in MR&D alliances. Table 10 contains the correlation statistics of the overall sample6. The descriptive statistics of several control variables are worth mentioning. First, both firm size and age of firms not involved in MR&D alliances are lower compared to firms involved in MR&D alliances. Firms from the ‘not involved’ sample have on average 8.176 employees and are 22 years old whereas firms involved in MR&D alliances have 69.257 employees and are
6
21
44 years old. Second, the alliance experience of firms not involved in MR&D alliances is on average much lower (mean=7,26) than of firms involved in MR&D alliances (mean=56,85). Finally, the R&D expenses of the ‘not involved’ sample are lower (mean=308.821.288,97) than that of the ‘involved’ sample (mean=1.202.527.197,85).
The correlation matrix provided in table 10 shows that the majority of independent and control variables have strong correlations with the dependent variable. MASC, industry and public status are not correlated with MR&D alliance formation. The relationships between the dependent variable and IND and UA are in line with the theoretical arguments provided. Whereas IND shows a negative correlation, UA demonstrates a positive correlation. Concerning the control variables it is noteworthy that firm size, alliance experience and R&D intensity are highly correlated with each other.
Interestingly, the correlation matrices of the sample of firms involved/not involved in MR&D alliance activities show some diverging results. More variables are correlated in the sample of firms involved than firms not involved in MR&D alliances. Whereas founding year correlates with many other variables in the ‘involved’ sample, it only correlates with IND in the ‘not involved’ sample. MASC and UA are more correlated with the other variables in the sample of firms involved in MR&D alliances. Alliance experience and R&D intensity have more correlations with other variables in the multi-partner sample. Finally, the variables firm size, alliance experience, and R&D intensity are also strongly correlated with each other in the ‘involved’ sample but slightly less strongly correlated in the ‘not involved’ sample.
22
threatened by multicollinearity issues (Sinan and Alkan, 2015). Concerning the control variables all VIF’s are below 1,7 and hence multicollinearity issues are limited7
.
Table 9: Descriptive statistics at firm level
Variable N Mean8 S.D. Min Max
1. Formation 124 50%(0), 50%(1) .502 0 1
2. IND
Involved in MR&D Not involved in MR&D
124 62 62 71,15 64,95 77,34 26,422 27,370 24,095 17 17 17 91 91 91 3. MASC Involved in MR&D Not involved in MR&D
124 62 62 63,48 64,50 62,45 16,706 18,967 14,173 5 14 5 95 95 95 4. UA Involved in MR&D Not involved in MR&D
124 62 62 57,68 62,18 53,18 19,738 21,379 16,950 8 8 29 92 92 92 5. Firm size Involved in MR&D Not involved in MR&D
124 62 62 38716,37 69257,21 8175,53 96347,968 128965,708 13777,508 35 35 44 800000 800000 83000 6. Firm age Involved in MR&D Not involved in MR&D
124 62 62 32,98 43,71 22,26 33,634 39,331 22,365 2 2 3 153 153 86 7. Founding year Involved in MR&D Not involved in MR&D
124 62 62 2002,97 2002,58 2003,35 2,341 2,446 2,181 2000 2000 2000 2009 2009 2009 8. Industry Involved in MR&D Not involved in MR&D
124 62 62 17,7%(0), 82,3%(1) 22,6%(0), 77,4%(1) 12,9%(0), 87,1%(1) .384 .355 .357 0 0 0 1 1 1 9. Alliance experience Involved in MR&D Not involved in MR&D
124 62 62 32,06 56,85 7,26 69,237 89,989 17,833 0 0 0 432 432 128 10. Type alliance experience
Involved in MR&D Not involved in MR&D
124 62 62 43,5%(0), 56,5%(1) 16,1%(0), 83,9%(1) 71,0%(0), 29,0%(1) .498 .450 .448 0 0 0 1 1 1 11. Public status Involved in MR&D Not involved in MR&D
124 62 62 8,9%(0), 91,1%(1) 6,5%(0), 93,5%(1) 11,3%(0), 88,7%(1) .285 .248 .283 0 0 0 1 1 1 12. R&D intensity Involved in MR&D Not involved in MR&D
124 62 62 755674243,41 1202527197,85 308821288,97 1372651324,608 1661912268,097 794592386,004 0 0 0 6207627000 6207627000 4851738983 13. Type alliance in same year
Involved in MR&D Not involved in MR&D
124 62 62 72,6%(0), 27,4%(1) 45,2%(0), 54,8%(1) 100%(0), 0%(1) .448 .502 .000 0 0 0 1 1 1 7
The control variable R&D intensity is excluded from the analysis to reduce multicollinearity issues. There appear to be strong correlations between this variable and alliance experience. If both variables are included then the VIF=2,036 for R&D intensity and VIF=2,026 for alliance experience.
8
23
Table 10: Correlation statistics
24 5.2 Hypotheses testing
To examine the influence of national culture on firms’ likelihood to form MR&D alliances, a binomial logistic regression analysis is performed and the results are provided in table 11 (simple models) and table 12 (full models)9. Models 1-3 provide the individual effects of the independent on the dependent variable. Models 4-7 include the control variables plus the independent variables added individually. Finally, models 8-9 show the combined effects of control and independent variables on the dependent variable.
Models 1-3 show that if added individually, IND and UA have significant effects on MR&D alliance formation whereas MASC does not, which is in line with the correlation statistics (see table 10). This simple regression model hence demonstrates significant effects, which should be highlighted. It shows that a higher level of individualism results in a smaller likelihood to form MR&D alliances (β= -0.019, p<0.05) whereas a higher level of UA has a positive influence on MR&D alliance formation (β=0.024, p<0.05). These findings are similar to those of Steensma et al. (2000) who argue that firms with low levels of individualism and high levels of uncertainty avoidance are more likely to form alliances.
Models 4-7 show that all control variables besides industry remain consistent. The coefficient of industry changes from positive to negative and vice versa. Without including the independent variables, three of the control variables are significant, namely firm age (β=0.030, p<0.05), founding year (β= -0.489, p<0.05) and alliance experience (β=0.028, p<0.05). In combination with the individual independent variables, the significance of the control variables decreases. The independent variables do not have significant effects if added individually to the control variables.
In models 8-9 the control variables and all independent variables are added simultaneously. The control variables show consistent effects and three of them are
9
25
significant. In model 9, firm age (β=0.039, p<0.05) and alliance experience (β=0.030, p<0.05) indicate that the older a firm is and the more alliance experience it has, the more likely it is to form MR&D alliances. Founding year (β= -0.558, p<0.05) shows that a firm is less likely to form MR&D alliances in the later years of the time frame 2000-2009. The Nagelkerke R2 of the final model (9) shows that 85.8% of the dependent variable is explained by the independent and control variables.
Concerning the hypotheses it is predicted that the higher a country’s level of individualism, the less likely firms from that country are to engage in MR&D alliances. Since the coefficient IND is not statistically significant, Hypothesis 1 is rejected. Secondly, it is hypothesised that the higher a country’s level of masculinity, the less likely firms from that country are to form MR&D alliances. The coefficient MASC is not significant and hence, Hypothesis 2 is not supported. Finally, it is predicted that based on whether one takes the RBV or TCE perspective into account, high levels of uncertainty avoidance lead to either a higher or lower likelihood to form MR&D alliances. The coefficient UA is not statistically significant and Hypotheses 3a and 3b are rejected. Despite the clear differences between the two samples of firms involved/not involved in MR&D alliances as observed in table 3 and the correlation statistics, none of the hypotheses can be confirmed.
Table 11: Logistic regression analysis – simple models
Variable Model 1 Model 2 Model 3
Constant 1,344* (.564) -.472 (.714) -1,395* (.584) IND -.019* (.007) MASC .007 (.011) UA .024* (.010) Model significance 7,033** .471 6,635** -2 Log likelihood 164,867 171,429 165,266 Cox & Snel R2 .055 .004 .052 Nagelkerke R2 .074 .005 .069
1. Regression coefficients standardized (β) and standard errors are shown 2. * p<0.05; ** P<0.01
26
Table 12: Logistic regression analysis – full models
Variable Model 4 Model 5 Model 6 Model 7 Model 8 Model 9
Constant 1016,043 (7403,587) 951,767 (7291,897) 1092,353 (7278,558) 994,831 (7402,148) 1201,897 (6380,968) 1165,026 (6281,232) IND -.013 (.016) -.045 (.031) -.054 (.037) MASC -.009 (.026) .013 (.032) UA .003 (.021) -.050 (.043) -.065 (.058) Firm age .030* (.014) .027* (.014) .032* (.015) .029 (.015) .038* (.017) .039* (.017) Founding year -.489* (.233) -.456 (.238) -.527* (.264) -.478 (.247) -.577* (.277) -.558* (.280) Industry (1) -.012 (1,086) .152 (1,094) -.010 (1,081) .010 (1,097) .084 (1,126) .016 (1,161) Alliance experience .028* (.013) .030* (.013) .028* (.013) .028* (.013) .030* (.013) .030* (.014) Type alliance experience (1) .405 (.906) .365 (.917) .388 (.910) .406 (.906) .221 (.946) .238 (.945) Public status 16,489 (5165,272) 17,098 (5080,604) 16,554 (5073,573) 16,473 (5162,585) 19,455 (4343,584) 20,179 (4244,098) Type alliance in same year (1) 39,837 (7388,794) 40,301 (7276,203) 40,137 (7259,167) 39,793 (7385,472) 42,973 (6356,423) 43,418 (6255,805) Model significance 125,623** 126,266** 125,731** 125,639** 127,629** 127,777** -2 Log likelihood 46,277 45,635 46,170 46,262 44,272 44,124 Cox & Snel R2 .637 .639 .637 .637 .643 .643
Nagelkerke R2 .849 .852 .850 .849 .857 .858
1. Regression coefficients standardized (β) and standard errors are shown 2. * p<0.05; ** P<0.01
27 5.3 Post-hoc analyses
Two main robustness checks are conducted10. First, a logistic regression analysis is done to identify whether curvilinear relationships exist among the independent variables. Models 10-12 in table 13 illustrate that when adding the independent variables and their squared estimates to the control variables, all three squared independent variables are significant. IND Square has a significant positive effect (β=0.001, p<0.01) whereas IND has a negative although insignificant effect. This suggests a U-shaped relationship between IND and the formation of MR&D alliances. Hence, until a certain level of individualism, firms from that society are less likely to form MR&D alliances. After that point, however, firms become more likely again to form this type of alliance. MASC Square and UA Square have significant negative effects (β= -0.001, p<0.05; β= -0.001, p<0.01) whereas the negative effects of MASC and UA are insignificant. Concerning the control variables, significance varies across the models. In all models, the founding year has significant negative effects.
10
28
Table 13: Curvilinear relationships
Variable Model 10 Model 11 Model 12
Constant 1701,468 (6316,443) 2300,904 (6158,616) 1722,660 (6379,009) IND -.005 (.019) MASC -.034 (.037) UA -.027 (.031) IND Square .001** (.000) MASC Square -.001* (.000) UA Square -.001** (.000) Firm age .051* (.023) .035 (.018) .053* (.023) Founding year -.830* (.349) -.1,127** (.435) -.838* (.341) Industry (1) 1,628 (1,399) .418 (1,214) .580 (1,372) Alliance experience .034 (.019) .033* (.016) .034 (.019) Type alliance experience (1) 1,050 (1,144) .297 (1,008) .291 (1,100) Public status 17,158 (4214,970) 16,594 (4519,946) 18,243 (4370,279) Type alliance in same year (1) 47,253 (6277,398) 42,837 (6459,353) 44,579 (6341,852) Model significance 141,204** 134,935** 139,604**
-2 Log likelihood 30,697 36,966 32,297 Cox & Snel R2 .680 .663 .676 Nagelkerke R2 .906 .884 .901
1. Regression coefficients standardized (β) and standard errors are shown 2. * p<0.05; ** P<0.01
3. Number of firms (N)=124
Second, a binomial logistic regression analysis is conducted with the sample of firms that only formed MR&D alliances in a focal year instead of also other types of cooperative agreements (n=34). A random sample is drawn from the matching sample of firms not involved in MR&D alliances with the same size (n=34). The results are provided in table 1411. The data of this new sample are also tested for multicollinearity issues but none can be identified12. Model 13 includes only the control variables and model 14 adds all independent variables simultaneously. From the latter model it can be observed that the founding year is significant (β= 0.314, p<0.05) indicating that the later the year in the time frame 2000-2009, the more likely firms are to form MR&D alliances. The type of alliance experience is also significant (β= 2,550, p<0.01). Hence, the more multi-partner experiences a firm has the more
11
Additional analyses are conducted to identify interaction effects between independent variables but none are found. As done in the analysis with the main sample and to reduce potential multicollinearity issues, the control variables firm size and type alliance in same year are excluded from the analysis. Different models are tested and the best model with the most significant results presented. The results are available upon request.
29
likely it is to form MR&D alliances. Most importantly, the coefficient of IND is significant (β= -0.030, p<0.05) signifying that a firm from a society with high levels of individualism is less likely to form MR&D alliances. This is in line with Hypothesis 1.
Table 14: Logistic regression analysis with number of firms (N)=68
Variable Model 13 Model 14
Constant -474,712 (274,533) -626,216* (314,967) IND -.030* (.015) MASC .012 (.029) UA -.022 (.021) Firm age .003 (.012) .005 (.013) Founding year .237 (.137) .314* (.157) Industry (1) -.553 (.801) -.785 (.848) Alliance experience .007 (.016) .014 (.019) Type alliance experience (1) 2,550** (.647) 2,755** (.728) Public status -.056 (.956) .210 (1,017) R&D intensity .000 (.000) .000 (.000) Model significance 24,353** 28,655**
-2 Log likelihood 69,915 65,613 Cox & Snel R2 .301 .344 Nagelkerke R2 .401 .458
1. Regression coefficients standardized (β) and standard errors are shown 2. * p<0.05; ** P<0.01
3. Number of firms (N)=68
6. Discussion of findings
30
between the two subsamples of firms involved/not involved in MR&D alliances, as established in table 3 and the correlation statistics, none of the hypotheses can be confirmed.
The reasons why no significant results are found might be twofold. First, it might be due to the limited but present multicollinearity issues concerning the independent variables, also found in previous research (e.g. Steensma et al., 2000; Stevens and Dykes, 2013). Despite the efforts to mitigate these multicollinearity issues, a small threat to the data remained. Second, MR&D alliances present a distinctive setting. Previous research that investigated the influence of culture on the formation of alliances looked at dyadic alliances (e.g. Nielsen, 2003; Steensma et al., 2000; Yu et al., 2013). Only a few studies examined the formation of multi-partner alliances (e.g. Li, 2013; Mitchell et al., 2002; Thorgren et al., 2012; Yin and Wu, 2003) but none included the influence of national culture. As noted by Li et al. (2011) and confirmed by this research, it remains unclear whether the findings of previous studies can be applied to MR&D alliances.
31
those of the TCE are supported since the influence of uncertainty avoidance is positive and significant. The shifts between positive and negative influences of high levels of uncertainty avoidance on MR&D alliance formation highlight the importance of considering both the RBV and TCE when looking at MR&D alliance formation (Beamish and Kachra, 2004). Considering only one theoretical perspective might lead to an incomplete understanding of firms’ rationales to enter MR&D alliances.
Other findings are also noteworthy. First, the control variable alliance experience has a significant effect on MR&D alliance formation in the main analysis (table 12). In line with previous research (e.g. Nielsen, 2003), prior alliance experience is found to have a positive influence on firms’ likelihood to enter MR&D alliances. In the analysis of firms that only formed MR&D alliances in a focal year (table 14), also the type of alliance experience has a significant effect, which is in line with Park et al.’s (2002) study. Firms with previous multi-partner alliance experience are more likely to form MR&D alliances. Second, in the post-hoc analysis, a U-shaped relationship is found between individualism and MR&D alliance formation. Until a certain level of individualism, firms from that society are less likely to form MR&D alliances. After that point, however, firms become more likely again to form this type of alliance. To the current knowledge of the researcher, no similar findings have emerged in other studies. Finally, a significant relationship is found between individualism and MR&D alliance formation in the analysis of firms only forming MR&D alliances in a focal year (table 14). As mentioned above, this is in line with the findings of Steensma et al. (2000) who report that firms with high levels of individualism are less likely to form alliances. Even though the sample used is relatively small (n=68), significant results are obtained that are in line with the hypothesised relationships.
cross-32
border participants, which is line with the increasingly international nature of MR&D alliances (Goerzen and Beamish, 2005). Moreover, the mainly moderate and high cultural diversity found among the MR&D alliances confirms the importance of research into cultural differences (Hofstede, 2001; Marino et al., 2002). It also supports the necessity of this research study and the importance of gaining a better understanding of the influence of firms’ national culture on MR&D alliance formation.
7. Conclusion
33 7.1 Implications
7.1.1 Theoretical implications
This study contributes to the literature of MR&D alliance formation and shows that firms with certain national backgrounds are more likely than others to form alliances with multiple partners instead of only one partner. Hence, it offers new insights into how national culture affects the organisational form that firms choose for innovation. As MR&D alliances become increasingly international, researchers should incorporate cultural factors into their studies (Steensma et al., 2000) to account for their potential influence on MR&D alliance formation. The results of previous research on the influence of national culture on dyadic alliance formation might not be applicable when there are multiple partners due to the increased complexity and distinctive dynamics. Thus, this research extends the limited body of knowledge on MR&D alliance formation and incorporates the influence of national culture.
Another important implication is that this study shows the significance of integrating the RBV and TCE due to their complementary and at times contradictory explanations concerning the motives of MR&D alliance formation. In order to account for both the benefits and the costs of forming MR&D alliances, it is essential to use both theories. Focusing on one theory alone might provide an insufficient analysis and misses out on a more comprehensive perspective on firms’ motivations to form MR&D alliances.
7.1.2 Practical implications
34
and uncertainty-avoiding cultures since these appear to be more likely to form MR&D alliances to collaborate and strive for continuous innovation.
For policy makers it might be valuable to provide government incentives for firms that form international MR&D alliances to ensure improved product development and pollination across countries. Firms that jointly conduct R&D with multiple partners have greater access to a variety of resources and can learn from each other to increase their knowledge bases. This can positively influence society’s wellbeing and create solutions for problems that might not be found if firms conduct R&D independently or only with one partner (Li, 2013). Moreover, the findings of this study indicate that incentives should be provided to firms from individualistic and uncertainty-tolerant societies in particular to promote MR&D alliance formation. By providing training and incentives, firms from these backgrounds might become more willing to form MR&D alliances and can contribute to greater knowledge accumulation. A final implication is to stimulate research by providing government incentives to learn more about why firms from collectivistic and uncertainty avoiding societies are more likely to form MR&D alliances and what effects these cultural backgrounds can have on R&D collaboration.
7.2 Limitations and future research
35
and more significant results emerge if the sample is larger and only consists of firms that form MR&D alliances in a focal year.
Third, whereas the sample of this study compares the formation of MR&D alliances versus dyadic alliances, the hypotheses do not make this distinction and instead only refer to MR&D alliance formation. This is a methodological limitation that should be taken into account when interpreting the results. Fourth, the control variable R&D intensity poses a limitation. Since data on firms’ R&D expenses in the year of alliance formation was not always available, the last year obtainable was used. This has clear disadvantages since the R&D expenses of firms change regularly and some data might be inaccurate.
Fifth, this research looks at firms’ national instead of organisational culture. A firms’ organisational culture is a more manifest and deeper level of culture and often measured by conducting a survey (e.g. Beugelsdijk et al., 2006). Since this laid outside of this study’s scope, it was only indirectly controlled for via proxies of firm size and age. Future research is necessary to examine the influence of firms’ corporate culture on their likelihood to form MR&D alliances. As Nielsen (2003) notes, similarities between the corporate cultures of firms might be more important than those between national cultures since “perhaps the impact of globalization has led to more convergence among international business segments” (p.305). Hence, future research can provide valuable input concerning whether firms’ national or organisational culture has a greater influence on their choices to form MR&D alliances compared to other types of cooperative agreements.
36 Acknowledgement
I would like to thank Isabel Estrada Vaquero for her constructive feedback. I highly appreciated her constant support and valuable input.
References
Barkema, H.G., & Vermeulen, F. (1997). What differences in the cultural backgrounds of partners are detrimental for international joint ventures? Journal of International Business Studies, 28 (4), 845-864.
Beamish, P.W., & Kachra, A., (2004). Number of partners and JV performance. Journal of World Business, 39, 107-120.
Beugelsdijk, S., Koen, C.I. & Noorderhaven, N.G. (2006). Organizational Culture and Relationship Skills. Organization Studies, 27 (6), 833-854.
Blau, P. M. (1977). Inequality and Heterogeneity. New York, NY: Free Press.
Chen, C.C., Peng, M.W., & Saparito, P.A. (2002). Individualism, collectivism, and opportunism: A cultural perspective on transaction cost economics. Journal of Management, 28 (4), 567-583.
Das, T.K., & Teng, B.-S. (2000). A resource-based theory of strategic alliances. Journal of Management, 26, 31-61.
Doz, Y.L., & Hamel, G. (1998). Alliance Advantage: The Art of Creating Value through Partnering. Boston, MA: Harvard Business School Press.
Ekeh, P.P. (1974). Social Exchange Theory: Two Traditions. Princeton, NJ: Princeton University Press.
Goerzen, A., & Beamish, P.W. (2005). The effect of alliance network diversity on
37
Gong, Y., Shenkar, O., Luo, Y., & Nyaw, M.-K. (2007). Do multiple parents help or hinder international joint venture performance? The mediating roles of contract completeness and partner cooperation. Strategic Management Journal, 28, 1021-1034.
Gulati, R. (1995). Does familiarity breed trust? The implications of repeated ties for contractual choice in alliances. Academy of Management Journal, 38 (1), 85-112. Gulati, R. (1999). Network location and learning: The influence of network resources and
firm capabilities on alliance formation. Strategic Management Journal, 20, 397-420. Hofstede, G. (1980). Culture’s Consequences: International Differences in Work Related
Values. Beverly Hills: Sage Publications.
Hofstede, G. (1991). Cultures and organizations: Software of the mind. Berkshire, England: McGraw-Hill.
Hofstede, G. (1994). Management scientists are human. Management Science, 40, 4-13. Hofstede, G. (2001). Culture’s Consequences: Comparing Values, Behaviors, Institutions,
and Organizations Across Nations. 2nd ed. Thousand Oaks, CA: Sage Publications. Hofstede, G., Neuijen, B., Ohayv, D.D., & Sanders, G. (1990). Measuring organizational
cultures: A qualitative and quantitative study across twenty cases. Administrative Science Quarterly, 35, 286-316.
Killing, J.P. (1983). Strategies for Joint Venture Success. New York: Praeger.
Kogut, B., & Singh, H. (1988). The effect of national culture on the choice of entry mode. Journal of International Business Studies, 19 (3), 411-432.
Lavie, D., Haunschild, P.R., & Khanna, P. (2012). Organizational differences, relational mechanisms, and alliance performance. Strategic Management Journal, 33, 1453-1479. Lavie, D., Lechner, C., & Singh, H. (2007). The performance implications of timing of entry
38
Li, D. (2013). Multilateral R&D alliances by new ventures. Journal of Business Research, 28, 241-260.
Li, D., Eden, L., Hitt, M.A., Duane Ireland, R., & Garrett, R.P. (2011). Governance in Multilateral R&D Alliances. Organization Science, 1-20.
Lin, C., Wu, Y.-J., Chang, C., Wang, W., & Lee, C.-Y. (2012). The alliance innovation performance of R&D alliances—the absorptive capacity perspective. Technovation, 32, 282-292.
Lin, H., & Darnall, N. (2015). Strategic alliance formation and structural configuration. Journal of Business Ethics, 127, 549-564.
Marino, L., Strandholm, K., Steensma, H.K., & Weaver, K.M. (2002). The moderating effect of national culture on the relationship between entrepreneurial orientation and strategic alliance portfolio extensiveness. Entrepreneurship Theory and Practice, Summer 2002, 145-160.
Mishra, A., Chandrasekaran, A., & MacCormack, A. (2015). Collaboration in multi-partner R&D projects: The impact of partnering scale and scope. Journal of Operations Management, 33-34, 1-14.
Mitchell, W., Dussauge, P., & Garrette, B. (2002). Alliances with competitors: How to combine and protect key resources? Creativity and Innovation Management, 11, 203-223.
Mothe, C., & Quelin, B.V. (2001). Resource creation and partnership in R&D consortia. The Journal of High Technology Management Research, 12 (1), 113-138.
39
Park, S.H., Chen, R., & Gallagher, S. (2002). Firm resources as moderators of the relationship between market growth and strategic alliances in semiconductor start-ups. The Academy of Management Journal, 45 (3), 527-545.
Parkhe, A. (1991). Interfirm diversity, organizational learning, and longevity in global strategic alliances. Journal of International Business Studies, 22 (4), 579-601.
Puxty, A.G. (1979). Some evidence concerning cultural differentials in ownership policies of overseas subsidiaries. Management International Review, 19, 39-50.
Reuer, J.J, & Lahiri, N. (2014). Searching for alliance partners: effects of geographic distance on the formation of R&D collaborations. Organization Science, 25 (1), 283-298.
Rothaermel, F.T., & Boeker, W. (2008). Old technology meets new technology:
complementarities, similarities and alliance formation. Strategic Management Journal, 29, 47-77.
Sakakibara, M., (2002). Formation of R&D consortia: industry and company effects. Strategic Management Journal, 23, 1033-1050.
Schilling, M.A. (2009). Understanding the alliance data. Strategic Management Journal, 30, 233-260.
Shenkar, O., & Zeira, Y. (1992). Role conflict and role ambiguity of chief executive officers in international joint ventures. Journal of International Business Studies, 23 (1), 55-75. Sinan, A., & Alkan, B. (2015). A useful approach to identify the multicollinearity in the
presence of outliers. Journal of Applied Statistics, 42 (5), 986-993.
