International joint ventures : resource-based vs relation-based partner selection

(1)

International Joint Ventures:

Resource-based VS Relation-based

partner selection

Ulaş Karasu

Amsterdam Business School, University of Amsterdam

Master’s Thesis MSc. Business Administration

Studentnumber: 11120452 First supervisor: E. Dirksen MSc Second supervisor: Dr. M. Westermann-Behaylo Final draft Date of submission: August 2016

(2)

Statement of originality

This document is written by Ulaş Karasu who declares to take full responsibility

for the contents of this document.

I declare that the text and the work presented in this document is original and that

no sources other than those mentioned in the text and its references have been

used in creating it.

The Faculty of Economics and Business is responsible solely for the supervision

of completion of the work, not for the contents.

(3)

ABSTRACT

As international joint venture literature has highlighted internal, resource-based- and external, relation-based considerations as the main drivers in setting partner selection criteria, it is unclear which of these two views, if any, is a superior contributor to joint venture performance. We examined this gap by triangulating qualitative document analysis, with quantitative performance measures. Based on a sample of 12 international joint ventures, equally divided in a resource-based group and in a relation-based group, we find no evidence that there is an a priori superior view to use in setting partner selection criteria, meaning that any superiority claims of both sides of opposing scholars are unfounded.

Key words: joint ventures; alliances; resource-based view; relational view; learning; transaction costs.

(5)

1. INTRODUCTION

As the competition in the modern business world extends more and more beyond national boundaries, the number of international alliances between firms is also growing steadily (Inkpen & Beamish, 1997). Motives for such alliances can be the sharing of risks and resources, getting access to new markets (Hitt, Dacin, Levitas, Arregle & Borza, 2000), avoiding transaction costs (Buckley & Casson, 1988; 2009), “learning” superior skills (Hamel, 1991), stimulating innovation (Hohberger, Almeida & Parada, 2015), exploring and developing new knowledge (Stettner & Lavie, 2014), or enhancing competitive position and market power (Kogut, 1988). Yoshino (1995) characterizes alliances by three characteristics: 1. The firms pursuing common goals through the partnership, remain autonomous after the alliance is formed, 2. both the control, as the benefits of the alliance are shared between the partners, 3. all the participating firms continuously contribute in key areas. So for example mergers and acquisitions do not fit within this description, as the acquired firms do not remain autonomous, while joint ventures dó fit.

Many scholars have given attention to performance indicators of such alliances. Beamish (1987) for example, argues that the greater the contributions expected by a firm from their respective alliance partner, and the greater the intended length of the interfirm relationship are, the greater the performance of the international alliance will be. Furthermore he stresses the importance of knowledge management, and partnering with firms that possess complementary resources and knowledge, as the partnership would perform better if each partner may focus on its own strengths (Beamish & Lupton, 2009). In addition, Lane, Lyles & Salk (2001) argue about the influence of interfirm trust on alliance performance, whereas Gulati (1998) stresses the importance of alliance partners’ positions in industry- and social networks in explaining the performance of international alliances.

(6)

Despite all the research and literature on international alliances however, there still are remarkably many international alliances that end up in miserable failures, with research even ascribing failure to more than two-thirds of all alliances (Chua & Mahama, 2007). A well-known, recent example is the alliance that was formed between the giant car manufacturers Volkswagen and Suzuki in 2009, which ended up in a four-year dispute between the parties, ruling out any future mutual alliances. As Suzuki’s chairman Osamu Suzuki put it: “It used to

feel as if a small bone were stuck in my throat. You will not remarry someone you have divorced” (BBC, 30-08-2015; The Times, 31-08-2015).

This implicates that the formation of an alliance can have different outcomes depending on the different possible partners, thus it seems crucial to select the right partner. The existing literature on partner selection is rich, and distinguishes between two different views. Traditionally, partner selection was explained by the resource-based view, arguing that maximal value can be derived from alliances when potential partners are selected on basis of their internal resources. Recently however, opposed scholars argued that the external, relational characterizations of firms, such as their reputation and their embeddedness in networks, may matter more when selecting partners. From here on, we will refer to this “new” stream of literature as the relational view.

Even though there already is considerable explanatory attention in the literature for the partner selection criteria used by firms, Hitt et al. (2000) pointed out a research gap with respect to the direct relationship between the different partner selection criteria, and alliance performance. Anno 2016, this gap still seems to be underexposed in the literature. In order to address the relationship between partner selection and international alliance performance, we took international joint ventures (IJVs) as our level of analysis, as these forms of alliance agreements have a very straightforward structure, and are the most committed form of

(7)

inter-organizational partnerships because of the large capital investments made by the partners.1 In this paper we assume the definition of a joint venture as “an alliance that combines resources

from more than one organization to create a new organizational entity (the "child"), which is distinct from its parents” (Inkpen & Beamish, 1997, p. 178). The “international” part refers to

the participating organizations being headquartered in two or more countries (Parkhe, 1991). By taking the IJV (the “child”) level of analysis in our study, we contribute to the alliance literature, as existing studies on alliance performance mainly studied outcomes at the level of the parent firms. Our main contribution however, is that we focus on outcomes for both of the dissenting views in setting partner selection criteria, by critically assessing and comparing the performance of both the resource-based and the relation-based IJVs. Our aim was to settle which partner selection method will lead to better performing IJVs, if any. This brought up the following research question:

“In selecting IJV partners, what selection criteria will lead to better performing IJVs?”

In order to answer this main question, we addressed two sub questions in the research. The first one was: What is the performance of IJVs, where partner selection was driven by the

(internal) resource-based view?

Then, we addressed the second sub question: What is the performance of IJVs, where partner

selection was driven by the (external) relational view?

After we were able to answer these sub questions, both outcomes were compared, allowing us to conclude on our main research question. Our results did not support that there is by definition one superior view to adhere to in selecting partners, implying that both views may be equally successful in selecting partners.

1_{The literature sometimes distinguishes between equity- and non-equity joint ventures, where the former refers}

to partnering firms all contributing substantial financial assets to establish an independent legal entity, and all partners likewise are paid for profits made by this entity. This is the definition we will restrict our use of the JV term to, as the non-equity term refers to a wide-array of contractual agreements such as licensing or distribution (Hennart, 1988).

(8)

In the next section, we will sequentially discuss the existing theory on IJV motives, partner selection criteria, and IJV performance, ending with our main hypothesis. Then we will give an in depth explanation about our methodology and results, finishing the paper with a section in which we discuss and conclude the findings and contributions of this research.

(9)

2. THEORETICAL BACKGROUND

2.1 International joint venture motives

When firms are entering a foreign market in which they are new, it is a logical step for them to enter by an IJV with another firm that already possesses the necessary local knowledge, as an autonomously managed wholly owned subsidiary will probably be unsuccessful without the required local knowledge. The same applies for firms entering a market in order to obtain access to resources which are held by local firms; an IJV will be the most logical entering method, as the alternatives of a greenfield investment or a full acquisition will be too risky and too costly (Inkpen & Beamish, 1997).2

From the perspective of transaction-cost theories, IJVs are seen as a means to get around inefficiencies in intermediate markets: either if the sought assets are tacit and as so, difficult to transfer-, or if the costs of internalizing parts of the production- or distribution process are lower than the transaction costs of obtaining the necessary products or services in the market or by contracts. Even though this internalization could also be achieved through acquisitions or greenfields (wholly owned subsidiaries), IJVs will be preferred to a costly acquisition or greenfield if the sought assets are public goods which can be shared between firms at a low marginal cost. Such an example would be if the so called “public goods” relate to an already existing distribution channel of a potential partner, which happens to be integrated with their manufacturing processes. The extra cost of jointly using their distribution channel would be minimal to the potential partner, and as so, an IJV could bring value for both partners at a marginal cost. Marginal, because when compared to choosing an acquisition

2

Here we assume that there are no local regulatory restrictions that firms have to consider when making their decision to form an IJV. In some countries institutional barriers may apply to certain industries or all foreign investors, such as restrictions on entry modes and foreign ownerships. The most famous example is the 1979 law on equity joint ventures in China, which allowed for foreign direct investment, but at the same the restricted these possibilities to clearly defined equity joint venture structures (Chadee & Qiu, 2001). A more recent example is the Russian “strategic sectors law” that was passed in 2008 by president Putin, posing stringent regulations on foreign investment in forty-two industries that are classified to be of strategic importance for the national interests (Pomeranz, 2010).

(10)

to get such access to the partner’s distribution channel, the unneeded manufacturing plants would also be a compulsory part of the deal, making the acquisition unnecessarily expensive and risky (Hennart, 1988).

As opposed to the transaction-cost arguments, resource-based scholars argue that not the minimization of transaction- and production costs, but the maximization of value through valuable, scarce resource seeking is the main driver behind the formation of IJVs. The resource-based view assumes that sustained competitive advantages are derived from valuable, rare, imperfectly imitable and non-substitutable resources embedded withín the firm (Barney, 1991). Therefore, alliances are a necessity in order to obtain some of those valuable and unique resources embedded within other firms. The main difference in IJV motivation from the transaction-cost perspective, is the paradoxical argument that IJVs are preferred over acquisitions or greenfields ónly if the potential partner will contribute mainly (non-protected) knowledge-based resources, while the partner seeking firm will contribute mainly property resources (protected by property rights, so hard for other firms to accumulate/exploit). This is because on the long term, IJVs are seen as the best way to accumulate valuable knowledge resources embedded within other firms. However, the same rationale applies for the partner, so if the partner-seeking firm will have to contribute mainly knowledge resources itself, it will not prefer IJVs as the danger of spilling valuable and rare knowledge to competing partners could weaken their competitiveness (Das & Teng, 2000). In other words: in transaction-cost theories IJVs may be preferred to wholly owned subsidiaries because of IJVs’ lower proportional costs, whereas in resource-based theories the reasons to prefer IJVs will more likely be linked to the argument that IJVs provide better opportunities to ‘learn’ from partners’ knowledge. Lastly, some resource-based scholars stress that IJVs may be found as the bundling of resources may lead firms to achieve or maintain a competitive advantage over competitors in the industry. This could be in the form of market power, when two firms

(11)

bundle their assets into one IJV in order to become market leaders, or it could be by partnering up strategically with a firm possessing valuable resources, just for the sake of depriving the competition of a potentially strong ally (Kogut, 1988).

Today, both views are still relevant as is evident from the present-day literature explaining IJV motives. Beamish & Lupton (2009) argue that by combining complementary resources and capabilities, firms may aim to benefit from economies of scope and scale, and may produce faster, cheaper and better quality than could be done alone or through alternative methods of entering markets. In addition, they also stress the possible motive of gaining access to embedded knowledge of respective partners, in order to ‘learn’ new skills and capabilities which can prove to be profitable organization-wide in the long-term. Lastly, when the input of a respective partner is valued to be more tacit or critical, firms increasingly will prefer an IJV over non-equity alliances; even if an IJV will be much more risky (because of the required investment) compared to non-equity alliances such as contractual agreements, IJVs are recognized to provide higher returns than firms will be able to earn through non-equity alliances (Contractor & Woodley, 2015).

2.2 Partner selection criteria

Gulati et al. (2012) argue that the criteria set for the selection of alliance partners, can be understood on the one hand from cooperation concerns of firms, which entail the commitment- and alignment of the interests of the partners, and on the other hand from coordination concerns; the effective alignment of the partners’ actions and communications. As mentioned in the introduction, the literature on partner selection in setting these criteria, may be divided into two views that firms may use in selecting their respective alliance partner.

(12)

The traditional, resource-based view, which we discussed earlier in explaining IJV motives, was developed mainly by criticasters of Porter’s ‘competitive forces’ model (1979). Porter’s model explains that firm performance is dependent on five forces of the industry it operates in: bargaining power of buyers, bargaining power of suppliers, entry barriers, availability of substitutes, and the level of rivalry. In contrast to Porter’s model, resource-based scholars believe that firm performance is mainly driven by internal firm forces (instead of industry forces): the full collection of valuable, unique, embedded and mainly tacit resources within firms (Mowery, Oxley & Silverman, 1998). IJVs are seen as a way to enrich and expand one’s valuable, tacit firm resources through organizational learning (Peng, 2001). When using the resource-based view on partner selection criteria, firms seek partners with excellent resources and expertise in a specific domain, in order to enhance the (partner seeking) firm’s own performance in that domain. This argument was confirmed in the study of Koufteros, Vickery, & Dröge, (2012). Arguing likewise that IJV alliances are used as a way for firms to acquire and exploit valuable resources, Hitt et al. (2000) identify eight different (resource-based) criteria on which firms place importance to varying degrees in assessing potential partners: financial assets, technological capabilities, managerial capabilities, intangible assets, willingness to share expertise, complementary capabilities, market knowledge & unique competencies. The validity of these selection criteria is supported by other resource-based scholars. In-depth research on the technological capabilities of firms for example, proved that joint-venture partners improved the overlap and the scale of their respective technological capabilities over time. This was determined by comparing partner firms before- and after their respective collaborations (Mowery et al., 1998). This increase in overlap of interfirm capabilities over the duration of an IJV, seems to prove the learning argument. Another example is the study of Beamish & Lupton (2009) that we mentioned in the previous section, providing support for the importance of complementary

(13)

capabilities. They explain that by selecting partners on the complementarity of their resources and capabilities, firms could achieve economies of scope and economies of scale, ánd may produce faster, better-, more reliable- and cheaper products. In order to capture these partner selection benefits, they argue that joint ventures are the optimal alliance method. Finally, Kogut (1988) argues that IJV partner selection may be done in the context of firms’ strategic positioning with regard to competitors, as we explained in the previous section: firms with a strong internal resource portfolio, especially financial assets, should be preferred in order to retain market power, or capture a market leading position.

Recently however, different scholars emphasized that the resource-based view seems to underestimate the high risks and high failure rates associated with interfirm alliances, claiming that partner selection may be understood better through relational criteria, which are located ‘outside’ firms, such as firms’ embeddedness in networks, reputation, and trust. Ding, Dekker & Groot (2013) argue that with greater transaction risk, potential partners are selected increasingly based on both trust and their existing reputation, instead of selection on the basis of resource-based criteria. This argument can be explained through transaction-cost theory; stipulating that the usage of partner selection criteria will be influenced by different risk-perceptions, as firms will make distinct strategic choices depending on the assumed risks (Williamson, 1985). By analysing any potential partnership upfront, firms will assess two kinds of transaction risks. The first one is the so called “appropriation risk”, which refers to the probability of a partner being abusive by maximizing its own gains at the expense of its counterpart. The second risk refers to the costs of coordination: the costs that arise from the adjustment of both parties’ firm processes in order to have effective mutual communications (Gulati & Singh, 1998).3 As argued, if a potential partnership is considered to be risky, firms

3

The assumed magnitude of these risks will be the result of transactional mechanisms such as asset specificity, transaction frequency and uncertainty in the environment, which can be studied further in detail in Williamson (1985).

(14)

will turn to a combination of trust- and reputation-based criteria in selecting the right partner. Trust refers to the assumed reliability and the assumed competence of a potential partner, and will be derived from “first-hand” information about the firm such as possible experiences of past collaborations with the firm, and direct communications with the management. On the other hand, the reputation of a respective firm is derived from “second-hand” information such as the recommendations and experiences of other firms familiar with the potential partner (Ding et al., 2013). These arguments are supported and confirmed by several other researchers. In their study on R&D alliances, where risk is a frequent concern because of the firms’ valuable property assets that are at stake, Li, Eden, Hitt, & Ireland (2008) found that firms preferred friended firms over strangers as their alliance partners, when the goals of the respective alliances were more innovative, and therefore, contained more risk. They defined these so called friends as: “potential alliance partners with whom a firm has developed

"strong-form" trust through multiple previous interactions. With strong-form trust, partners are trustworthy, independently of whether or not exchange vulnerabilities or governance mechanisms exist.” Similarly, a study on German small and medium-sized enterprises found

that these firms tend to select partners they trust over other potential partners (Mukherjee, Gaur, Gaur & Schmid, 2013).

In line with Ding et al. and Li et al., in his 1998 paper, Gulati also addresses the narrowness of the resource-based view, arguing that even though alliances are exchanges between the firms directly involved, they are influenced by the networks of the firms. Where the resource-based view entails exclusively tacit and rare resources as the drivers of partner selection, Gulati claims that the range of possible partners is heavily dependent on the social network of external contacts a firm has, as firms initiating new alliances will first seek potential partners in their existing relations and the networks of these relations. In addition,

(15)

Gulati explains that the social networks to which any potential partners have access to, is also an important factor in the potential partners’ attractiveness, as firms will prefer partners with a rich access to valuable social networks.

Dyer & Singh (1998) also emphasize the importance of networks when selecting partners, adding that firms’ critical resources may extend beyond firm boundaries, when partners are willing to make relation-specific investments by combining their resources in unique ways.4 These idiosyncratic interfirm linkages are not considered in the resource-based view, as the resource-based view entails only the critical resources nested withín firms. That is why social network theorists argue that their theory complements what is missing in the resource-based view, suggesting that partners with a strong, ‘external’ social network, will also be attractive for partner seeking firms, as they can provide differential access to valuable external resources and opportunities, which in turn can lead to a competitive advantage through indirect learning (Arya & Lin, 2007). As Lavie (2006) puts it: “I conclude that the nature of

relationships may matter more than the nature of resources in networked environments” (p.

638).

So to summarize the literature on partner selection; we may distinguish between the resource-based view, which prescribes that the selection of potential partners should be driven by the value and tacitness of their internal resources in order to maximize ‘learning’, and the relational view, which relates to more external criteria setting in selecting partners such as the partners’ embeddedness in networks, and trust- and reputation assessments in order to learn indirectly from the network and minimize transactional risks.

2.3 International joint venture performance

4

We elaborate on this point with Dyer & Singh’s (1998) second and third categories in the next section on IJV performance.

(16)

In this section we will elaborate on the existing literature concerning IJV performance, guided by the four categories Dyer & Singh (1998) use to discuss the possibilities of deriving competitive advantages from interfirm relations and alliances, namely: 1. Investing in relation-specific assets, 2. Sharing knowledge, 3. Combining complementary resources to get unique new products, services or technologies, 4. Having effective governance systems in place, so the transaction costs can be kept minimal.

Elaborating on the first category from an IJV perspective; a potential investment in relation-specific assets could be made by locating the physical parts of consecutive production processes close to each other, in order to minimize costs (resulting for example from transportation and storage), therefore increasing an IJV’s overall profits. In the literature, such an investment is referred to as “site specificity”. Another investment within this category would be “physical asset specific” investments, which relate to investments made in customized machinery to allow for the production of customized products, and high quality products, aiming on improving sales and maximizing profits through product differentiation. Lastly, “human asset specific” investments relate to the specialized knowledge of people that are managing the transactions between the partnered companies over a longer period of time, allowing the interfirm communications and transactions to run smoother and to be less costly. However, because such investments will be costly and risky, firms will only consider these if they can safeguard themselves against possible opportunism (for example by working out acceptable long-term, safeguarding contracts)5, and if the transaction scale will be large enough to justify the investments.

In the literature it is widely recognized that effective knowledge management is crucial in continued business performance (Beamish & Lupton, 2009). This is no different with IJVs, as Dyer & Singh’s second category refers to endeavouring a regular pattern of interfirm

(17)

interactions that permits the transfer, recombination or creation of specialized knowledge (p.

665) (Grant, 1996), as there is much more valuable innovation and information to gain from a full learning network of interacting partner firms, producers, suppliers and consumers, than any firm could gain through only their own internal processes (Von Hippel, 1988; Powell, Koput & Smith-Doerr, 1996). Key in translating these learning processes into greater IJV performance, is an IJV’s so called “absorptive capacity”, which refers to the ability of an IJV to effectively spread new knowledge amongst the staff. The relative strength of this capacity is due to the combination of three factors: first, the IJV should be able to effectively interpret the knowledge that is passed on by the parents. The effectiveness of this interpretation is positively related to the degree of similarity in the organizational cultures between the parents and the IJV, the degree of similarity in the relatedness of the IJVs business to the parents’, and the amount of accumulated prior learning. Second, the IJV should be able to effectively assimilate the received knowledge. The effectiveness of this assimilation is positively related to a flexible organizational culture, the existence of formal learning goals, the division of business between the parents and the provision of training by the parents. The third absorptive capacity factor, relates to the ability of effectively translating the parents’ knowledge into IJV performance after it is successfully interpreted and assimilated (so this factor is subsequent to the first two mentioned-). Here it is argued that the degree of focus on a differentiation strategy over a cost-leadership strategy, and the IJV’s ability to train and develop their own staff relate positively to IJV performance (Lane, et al., 2001; Lyles & Salk, 1996).

The third category is about making value through combining complementary resources. Dyer & Singh define complementary resources as “distinctive resources of alliance partners

that collectively generate greater rents than the sum of those obtained from the individual endowments of each partner” (p. 666). They elaborate that alliances such as IJVs are needed

(18)

Whereas until the 1990s scholars assumed that greater resource similarity between partnering firms would lead to greater alliance performance, subsequent research showed contrary results (Harrison, Hitt, Hoskisson & Ireland, 1991; 2001). They explain these results by arguing that even though resource similarity between partnering firms may lead to synergies created by economies of scale on the short term, alliances are much more likely to produce sustainable value through unique and inimitable synergies. Such synergies can only be created when partner firms possess distinctive resources that complement each other’s. An example would be an IJV that allows for the combination of the excellent brand name of one parent, with the excellent distribution network of the other (in both instances accumulated after decades of business), which would give the IJV a competitive advantage which neither parent firm could have achieved on its own. Furthermore, in the previous sections we explained that Beamish & Lupton (2009) emphasize the importance of forming joint ventures with partners with complementary -instead of overlapping- resources, knowledge and capabilities. They relate this argument to IJV performance, by explaining that an IJV will perform better when each partner can focus on enhancing its own strengths, making up for their partner’s weaknesses. Finally, we may mention the ‘strategic’ IJVs, found to strengthen firms’ market power against competitors. The argument here is that when partners bundle their resources to form an IJV, the IJV will be a means to increase performance, as the resource combination will allow the IJV to hold a stronger position in comparison to rivals, and thus, to maximize profits (Kogut, 1988).

In the literature on alliance performance it is commonly accepted that, in order to create value and derive any competitive advantage out of alliances, managing them effectively is essential (Ireland, Hitt & Vaidyanath, 2002). Within this management, partners’ acknowledgement of differences between each other, and the alignment of organizational routines in order to overcome these differences, proved to be important contributors to the

(19)

performance of an alliance (Lavie, Haunschild & Khanna, 2012). This brings us to Dyer & Singh’s fourth and final topic that is argued to affect IJV performance: the partners’ management of IJVs, and the governance structures herein that are used to safeguard both the partners, as the IJV against opportunistic behaviour of another alliance party. The first kind of governance that may be used by the parties in an IJV, are so called third-party enforced agreements: contractually defined agreements, which can be enforced by a third party such as the state or courts in case of conflict. The second kind are self-enforcing agreements, which can be either formal, or informal. Formal self-enforcing agreements relate to aligned financial incentives between the parties, in such a way that opportunistic behaviour of either party will have negative financial outcomes for all the involved parties. Informal self-enforcing safeguards refer to the mutual trust relations, and the reputation of the parties. In case of opportunistic behaviour or conflict, the reputation and trustworthiness of partners will decline, which can seriously harm the attractiveness of the involved firms in any future business or partner selection stages, as argued in the partner selection section earlier by Ding et al. (2013). It is argued that informal self-enforcing governance safeguards not only are the most effective safeguards, but also will cost the least from a transaction-cost perspective as costly processes which are required for the other governance safeguards, such as bargaining and monitoring, are not required with the use of informal safeguards, leading to a greater overall profits and performance (Barney & Hansen, 1995; Sako, 1991). The results of other studies seem to confirm the positive relation between informal safeguards and IJV performance. Lane et al. (2001) showed for example a positive effect of interfirm (partner) trust on IJV performance. However, where some scholars argue from the resource-based point of view that trust between partners will affect performance indirectly through learning processes, as it enables knowledge sharing between the partners (Beamish & Lupton, 2009), Lane et al. (2001) found no evidence of a positive relationship between trust and learning in IJVs. This suggests that

(20)

the transaction-cost explanation might be more useful in explaining the positive relation between interfirm trust and IJV performance. Thus, taken altogether, the ability of the alliance partners to align their transactions with the governance structures in a cost-minimizing, and value-maximizing way, will lead to greater IJV performance (Dyer & Singh, 1998).

The advancements in the literature and the relevance of the theories and views pointing out different factors relating to alliance performance (which are described in the above), are acknowledged by Gulati (1998). However, as Dyer & Singh already mentioned the learning opportunities that may lie in the firms’ networks, Gulati further extends the literature and contributes by arguing that the alliance partners’ embeddedness in social networks in fact can have a major influence on the performance and success of an alliance; a factor he claims to be underexposed by scholars. Gulati distinguishes between relational embeddedness and structural embeddedness in networks. The first one relates to cohesive, social ties between interfirm actors. Through ongoing communications in informal settings, mutual trust will increase, uncertainty will decrease, and the concerned firms may gain unique information. Structural embeddedness is about firms’ position or status in the network structure, which will form some kind of a reference for other firms about the quality they may expect from the firm in question.

Gulati first of all expands Dyer & Singh’s fourth category about the effective use of governance structures, arguing that closeness of alliance partners in the network, may affect the governance structures that are used to formalize alliances or IJVs. As relational embeddedness may increase trust, and structural embeddedness may provide valuable information about the partner’s status, firms’ overall network embeddedness may influence the decision on whether to use informal self-enforcing safeguards and, accordingly, it would influence IJV performance (as we explained by Dyer & Singh’s fourth category with regard to trust, reputation and the related transaction costs).

(21)

Another effect of partners’ embeddedness in networks, is their subsequent, long-term alliance behaviour. By relying on firms in their social network for available information on potential future partners, firms will include or exclude potential IJV partners as viable options. This may lead to more trust in the interfirm partnership, and less reluctance to invest in relation-specific assets, which in turn both may lead to a maximization of IJV performance through transaction costs minimization (as we explained by Dyer & Singh’s first and last categories).

The last main effect of network embeddedness discussed by Gulati, is that a firm’s social network and its position within that network, will determine the amount and the value of the resources and information a firm will have access to. Thus, the social network may affect both the performance of an IJV through the available knowledge in the learning network as we discussed by Dyer & Singh’s second category, as well as through the access to complementary resources as discussed in the third category. Gulati’s expansion on the second category is confirmed by Phelps (2010), proving with a longitudinal study that both access to a diverse social network, as the density and diversity of alliance partners’ respective networks, lead to more valuable firm innovations through learning. Gulati’s expansion on the third category in turn is confirmed by the study of Ozcan & Eisenhardt (2009), who found that alliances performed better when they were actively endeavouring to utilize all their social network ties to gather complementary resources, -enabling each firm to focus on their own strengths- as opposed to alliances that used only series of single ties.

2.4 Hypothesis

So to recapitulate on the theory we discussed, the varying explanations of scholars on both the partner selection phase, as on the drivers of IJV performance, seem to depend mainly on the philosophic conviction the scholars have. The resource-based scholars seem to legitimize

(22)

their view by providing learning arguments as the rationale behind a positive relation between internal, resource-based partner selection and IJV performance, whereas relational scholars argue for the rationales of transaction cost minimization and indirect learning through external, relation-based partner selection criteria to be the most important contributor to IJV performance. As the arguments of both views are for a large part based on factual research, both views have provided important contributions to the literature. The convictions on the superiority of one view over the other are however not examined yet by factual research. The lack of such research led to our main hypothesis:

Hypothesis 1: The performance of IJVs in which partner selection was driven by the relational view, will deviate from the performance of IJVs in which partner selection was

driven by the resource-based view.

If the results support the main hypothesis, we may reject the null hypothesis. Accordingly, if the main hypothesis is not supported, it will mean that the null hypothesis cannot be rejected.

Hypothesis 0: The performance of IJVs in which partner selection was driven by the relational view, will not deviate from the performance of IJVs in which partner selection was

driven by the resource-based view.

Relation-based IJV performance Resource-based IJV performance Relation-based IJV performance Resource-based IJV performance Figure 1: Hypothesis 1

(23)

3. METHODS

3.1 Research design and sample

In order to accept or reject the main hypothesis, and to be able to answer our (main- & sub) research questions, we conducted the research on a pragmatic manner: the research design was based upon mixed methods. This gave the research a couple of advantages: first, the use of different data collection methods adjusted to our needs, allowed for more accurate results as some methods will suit better than others, depending on the particular tasks on hand. Second, it allowed us to focus on the different aspects of IJVs: from the eventual partner selection criteria used to form IJVs, to the IJVs’ actual performance. Thirdly, the triangulation of independent data sources and methods, will be positively related to the credibility of this research. Finally, the use of the qualitative archival research (document analysis), allowed us to interpret and explain the quantitative outcomes (Saunders & Lewis, 2014).

First, document analysis was conducted which allowed us to make a distinction between those IJVs formed through resource-based-, and those through relation-based partner selection. This was suitable, since representatives of firms forming IJVs often give important clues about the nature of the alliance in media statements, as they attempt to deliver their stakeholders both transparency, and legitimization of their decision to form an IJV. The first advantage of document analysis is its efficiency compared to other research methods: because it entails data selection instead of data collection, it is less time consuming and less costly. Furthermore, the data is easily available as it does not require any permission of authors. Another advantage is that the content of the documents is not affected by the research, as with other qualitative methods there may be concerns about the obtrusiveness of the researcher and the reflexivity of the respondents which may lead to biased results. Finally, documents provide stable and exact data, and provide broad coverage over a long span of time (Bowen, 2009; Merriam, 1988; Yin, 1994).

(24)

After the relevant distinctions between the IJV forms were made, quantitative database analysis was conducted used for the measurement of IJV performance. In their study on founding family ownership, Anderson & Reeb (2003) used return on assets (ROA) as the most important indicator for firm performance. They found the ROA by dividing earnings before interest, tax, depreciation and amortization (EBIDTA) by the book value of total assets. As the ROA of many IJVs can be found readily calculated in financial databases, it was suitable for our research as well. The advantages of database analysis are first and foremost its accessibility and cost effectiveness, as the data is already collected and freely available. Furthermore, the administrative dataset that we used offers a broad, worldwide ánd longitudinal coverage on firms’ micro data (Ribeiro, Menghinello & De Backer, 2010).

As can be concluded from the use of archival research in combination with database research, the data was obtained from secondary sources. The main strength of having a research design based on secondary data is thus, the easy access to large quantities of (high-quality) data, minimizing financial and time constraints. Even though these data were collected for other purposes, they still were excellently applicable for the purpose of this study. The main limitation of this research design however, is that the data had to be gathered from multiple sources to meet the needs of this study, all of which served a different purpose of collection. Lastly, as we only selected -and not collected- the data, we cannot tell how the data was collected exactly, and as so, parts of the data may be manipulated (Saunders & Lewis, 2012).

The final data sample was based on 12 IJVs; 6 of them formed out of resource-based considerations and 6 out of relation-based considerations. The names of these IJV firms were as follows: 1. EE Limited, 2. SB Limotive, 3. Mitsubishi UFJ Morgan Stanley Securities, 4.

(25)

Morgan Cazenove, 10. Pepsi Lipton International, 11. Qoros Automotive, and 12. Coors UK.6

To minimize the chance of biased measures as a result of outliers, the performance measures of these firms are based on longitudinal results over the period from 2006 until 2015.

3.2 International joint ventures as the level of analysis

In contrast to the majority of the alliance literature in which alliance or IJV performance is measured at the level of the parents (for example Lin, Yang & Arya, 2009)7, or at the level of the market (for example Lavie, 2007)8, we measured performance at the level of the IJV firms, consistent with just a few prior papers, such as Lyles & Salk (1996). This choice was made to assure internal validity, as objective measures will be hard to legitimize at the level of the parents or -market, when considering the large number of potential moderating or mediating processes that may bias the findings at those levels. Another facet is that at the level of the parents, IJV performance may have been regarded differently by different parents depending on their respective contributions and -IJV goals. This brings us to an important distinction made in the literature between so called link IJVs and scale IJVs: the first one relates to IJV partners contributing asymmetric input, while the latter relates to partners contributing similar input (Dussauge, Garrette, & Mitchell, 2000). We focused exclusively on scale IJVs, thus we sampled for IJVs where all partners provided for either resource-based input, or all for relation-based input.

6

At the present day, Best Buy Europe is renamed as Carphone Warehouse Europe, VIVO is Telefonica Brasil, and DDB Guoan is Bejing DDB Advertising, as the result of changes in control. Sidanco, SB Limotive, and Qoros are dissolved.

7_{Lin et al. 2009 measured the effect of alliances on the firms taking part in them (parents), by measuring the}

performance (ROA) of the parent firms in alliances.

8_{Lavie applies the measure of “market performance”: the annual change in the value of firm’s common shares ,}

which is based upon the expectations of investors about firm’s future value, and adjusted to the compound S&P 500 market value.

(26)

3.3 Dependent Variable

IJV performance was measured on the basis of the IJVs’ return on assets (ROA) using the profit/loss statement before taxes. This was an useful and relevant choice, as it is widely accepted to be an excellent and reliable profitability ratio, correlating strongly with subjective performance measures as demonstrated in many prior research like Dess & Robinson (1984) and Anderson & Reeb (2003). We retrieved this data on the IJV firms in question through the financial database of Orbis, as this database contains longitudinal, detailed financial information on over 200 million private companies worldwide. First, the availability of data per firm was assessed, as some IJV firms proved to be useless within our research design because of their failure to disclose any profit/loss statements. In the end, 12 IJV firms that díd suit our research design were found, after which each firm’s mean ROA over the period from 2006-2015 was calculated. Any irrelevant years as a result of changes in control or IJV dissolution, were excluded to secure the validity of the results. In the end, the firms were divided by their respective groups, to conclude with a single mean ROA for the “resource-based IJVs” group, and a single mean ROA for the “relation-“resource-based IJVs” group.

3.4 Independent variables

Document analysis was conducted on press releases from newspapers -that are internationally accepted to have high journalistic standards-, such as the Financial Times. We retrieved this data through the database of LexisNexis Academic, which contains articles from newspapers, magazines, press agencies and other worldwide news sources. Relevant data was collected by searching for terms like “International joint venture between”, in combination with terms like “resources”, “assets”, “relations”, “network” etcetera. By analyzing firms’ legitimizations of their IJV partner selection in the found articles, useful scale IJVs were detected, and accordingly assigned to be either resource-based, or relation-based. Next, company websites

(27)

and more recent press releases were analyzed, to control for the exact years in which the IJVs continued in their original form, excluding any years of dissolution or control change.

3.5 Control variables

As IJV performance could potentially be affected by factors outside of partner selection such as organizational size, and organizational age (Park & Ungson, 1997), any coincidental causality with any of our two independent variables would threaten the reliability of our results. This is why we controlled for size, which we measured by sales turnover in USD, for

age, measured by the number of years that the IJV was operating (at the last measured year),

and the number of executives that were active within the IJV. The relevance of these control variables within our research structure is supported by Pothukuchi, Damanpour, Choi, Chen & Park (2002), using similar variables to control for IJV performance. These variables were retrieved through the financial database of Orbis.

3.6 Data Analysis

As explained, the data was collected and analysed in a sequential way, starting with archival research and document analysis. The documents were analysed through a process described by Bowen (2009), starting with superficial examination, which led to organising any potentially useful data within one of the two categories relating to our central research theme. Second, the data were thoroughly examined and interpreted, to uncover underlying themes and subtle cues in the meanings, in an attempt to demonstrate objectivity and sensitivity in analysing the relevance of each document. In the end, the final selection was made by considering the purpose-, the source-, and the target audience of the texts, after which the final findings and IJV categorizations were verified by skimming other reliable newspapers, company websites, and annual reports.

(28)

Then, the relevant financial data were exported from Orbis to Microsoft Excel, in which the data were ordered, and irrelevant data (such as any years of control changes) were excluded. This resulted in a total of 61 observations. The functions in Excel were used to calculate a single mean ROA per IJV firm, over the longitudinal data that was available. Next, the data were exported from Excel to SPSS Statistics. In SPSS, we first assigned the firms within either “group 0” (resource-based), or “group 1” (relation-based). Then, we tested for normality, plotted a graph comparing the mean performances of both groups, followed by an one-way analysis of variance (ANOVA) to test for the significance of the mean differences, which served to test our main hypothesis as depicted in figure 1. We ended with a hierarchical regression analysis in which the control variables were added to the relation we tested in the ANOVA, in order to control for any potential effect that the control variables might have had on our tested hypothesis.

4. RESULTS

4.1 Resource-based partner selection

The first IJV that was recognized to be based on resource-based partner selection, was EE Limited; a partnership between Deutsche Telekom and France Telekom, that was set up in order to secure a market leadership in the British mobile business: “The joint venture would

secure clear leadership in the British market. France Telecom and Deutsche Telekom would most likely pool their UK mobile assets in an entity structured as a 50:50 equity partnership”

(Financial Times, 08-09-2009). From this statement we interpreted that the partners selected each other on the basis of their respective internal assets, which, in combination, should have led to profit maximization through an improvement of their competitive position in the market. Such a selection process is in line with the resource-based view on partner selection. A similar kind of resource-based IJV was set up between the US-based Morgan Stanley, and

(29)

the Japanese Mitsubishi UFJ Financial Group: “Under a deal announced yesterday, Morgan

Stanley and MUFG, which is the US group's largest investor through its holding of preferred shares, will pool their lending resources in an effort to gain a stronger position when dealing with American companies” (Financial Times, 01-07-2009). This IJV was named Mitsubishi

UFJ Morgan Stanley Securities. A joint venture that was mainly based on complementary resources, was Best Buy Europe; a venture between a major consumer electronics player in the US, Best Buy, and Carphone Warehouse from the UK. The CEO of Best Buy, Brad Anderson, explained that they selected Carphone Warehouse to be their partner because of their excellent managerial capabilities in Europe, which were complementary to Best Buy’s resources, and thus allowed the combination to increase the scale of their operations: “We’re

not intending to come in and tell them how to do it; we’re intending to provide resources to help them do it,” he says of the joint venture (Financial Times, 12-05-2008). Another IJV that

was found to create value by increasing scale, was VIVO; a Brazilian telecom provider which originated out of a 50:50 joint venture between Portugal Telecom and Telefonica from Spain, both firms combining all their Brazilian mobile assets in order to pursue an aggressive growth strategy. In a joint statement the companies legitimized the IJV as follows: "We strongly

believe the inevitable consolidation of the Brazilian mobile industry will create significant synergies and shareholder value. We have created the natural consolidator in this market"

(Financial Times, 25-01-2001). Samsung SDI from Korea (major producer of lithium-ion batteries), and the German Bosch GmbH (leading manufacturer of car electric devices) also founded an IJV because of scale considerations. By combining their financial resources and their ‘VRIN’ knowledge resources, the IJV named SB-Limotive aimed to profit from the foreseen increase in demand in rechargeable hybrid car batteries: "The joint venture will fuel

momentum for reshaping the structure of the battery market currently dominated by Japanese car manufacturers. The partnership with Bosch is a key catalyst for us to expand on the

(30)

global battery market," spokesman Seo Hae-su said, adding SDI doesn't plan to delay the project (Korea Times 08-12-2008). A similar combination of complementary resources was

visible when BP, Alfa and Acces-Renova partnered up with an IJV named Sidanco. In this venture, BP, a major oil and gas company from the UK, was an attractive partner because of their excellent managerial capabilities and expertise, whereas the Russian investors groups Alfa and Acces-Renova provided the necessary financial assets and owned valuable oil and gas fields in Siberia. Alex Knaster, general director of Sidanco and chief executive of Alfa Bank, communicated about BP's role in the IJV as "to turn Sidanco into the best managed,

most efficient company in the Russian oil business" (Financial Times 17-04-2002).

4.2 Relation-based partner selection

With both partners emphasizing that choosing the right partner will provide much greater opportunities, Chinese investment company CITIC Guoan and global advertising leaders Omnicom from the US, formed a Chinese IJV named DDB Guoan. From the documents we could infer that the partners had selected each other on relational criteria such as reputation, and access to each other’s social network: “The Citic Guoan advertising arm already boasts

an impressive roster of domestic clients including First Auto Works, a leading local carmaker, China Unicom, a mobile telephone operator, and Haier, a white-goods producer. Meanwhile, Omnicom brings with it clients such as Philips, Johnson & Johnson and PepsiCo” (Financial Times, 09-06-2006). Having mentioned PepsiCo, their IJV with Unilever

forming Pepsi Lipton International was also assigned to our relational IJVs group, as this IJV was designed to combine the excellent reputation and brand name of Unilever’s Lipton, together with the excellent social network of PepsiCo in developing and emerging markets. Patrick Cescau, director of Unilever’s food division commented: "This new joint venture

(31)

many millions of new consumers" (The Herald, 2003). The third IJV that we identified to have

emerged out of relational partner selection, was Alvean; a venture between leading agricultural traders Cargill from the US, and the world’s largest sugar producer: Copersucar from Brazil. Even though both firms had built a strong reputation within their own area, both firms suffered in recent years from the inconsistent sugar trade market. As a solution, this IJV was set-up from a transaction cost perspective as it allowed both partners to utilize each other’s already available public goods, combining Cargill’s international transportation and customer network with Copersucar’s production capacity: "The link-up makes sense," said

Robin Shaw at commodities brokers Marex Spectron. "It brings together the world's biggest producer and one of the world's biggest distributors" (Financial Times, 28-03-2014). Another

IJV that was recognized to have emerged because of a cost-minimization perspective together with social network considerations, was the British IJV investment bank named JP Morgan Cazenove: formed through a partnership between the US based JP Morgan, and the British Cazenove. This deal gave JP Morgan “access to the finest UK corporate clients”, without the risks of an acquisition, while JP Morgan contributed its diversified banking businesses in which they had built an excellent reputation: “Under the terms of the deal, Cazenove will

contribute its entire UK investment banking business to the joint venture, while JP Morgan will inject its business, including 50 bankers” (Financial Times, 05-11-2004). A similar IJV

was set-up in the automotive industry, with China’s Chery Automobile delivering local market knowledge and local market access, whereas Israel Corporation (from Israel) provided the “western” expertise and reputation to enhance the products’ image. This partnership gave the IJV, named Qoros Automotive, the opportunity to serve young urban Chinese customers in their demand for high-tech western cars, as “China, the world's largest car market, is

dominated by foreign brands, and Chinese makes have in the past struggled because of concerns over quality” (Financial Times, 11-03-2013). The last IJV that was assigned to the

(32)

relational category through document analysis, was Coors UK; a partnership between the US based Coors brewery, and the UK based Scottish & Newcastle. Here, the relational transaction-cost and network considerations of both sides in selecting the partners, were clearly stated: “From S & N's point of view, the licensing and distribution agreement added a

further brand to its portfolio of premium lager, allowing marketing and promotion costs to be shared. For Coors, the attraction was immediate, low-cost access to a new geographical market via a proven distribution network” (Financial Times, 27-05-1993).

4.3 The effect of partner selection criteria on international joint venture performance The data is normally distributed, which can be concluded from the output of the Shapiro-Wilk test statistics depicted in Table 1, as no statistically significant effect had been found: p > 0.05 for both groups.

Table 1: Tests of normality for the resource-based group (0) & relation-based group (1)

Rbv0.Nbv1

Kolmogorov-Smirnova Shapiro-Wilk

Statistic df Sig. Statistic df Sig.

IJV performance 0 0,27 6,00 0,19 0,88 6,00 0,29

1 0,15 6,00 .200* 0,97 6,00 0,89

*. This is a lower bound of the true significance. a. Lilliefors Significance Correction

Table 2 depicts the descriptive statistics, providing the minimal performing-, and the maximal performing IJVs within the two subsets, and the overall mean performance (ROA) per subset: the resource-based group has a mean ROA of 4.26, and the relation-based group has a mean ROA of 4.62.

(33)

Table 2: Descriptives for the resource-based group (0) & relation-based group (1) IJV Performance N Mean Std. Deviation Std. Error 95% Confidence Interval for Mean

Minimum Maximum Lower Bound Upper Bound 0 6,00 4,26 15,55 6,35 -12,07 20,58 -10,36 28,79 1 6,00 4,62 12,75 5,21 -8,77 18,00 -14,66 20,20 Total 12,00 4,44 13,56 3,91 -4,18 13,05 -14,66 28,79

Table 3 provides the significance test of the mean difference between the “resource-based (internal) partner selection” group and the “relation-“resource-based (external) partner selection” group. There was no statistically significant effect of the Partner selection criteria on the level of IJV Performance, F (1, 10) = 0.002; p > 0.05. Which means that the performance of IJVs where partners were selected on basis of resource-based criteria, did not significantly deviate from the group in which partners were selected on basis of relational criteria. Thus, we have no reason to accept our main hypothesis (H1), and we cannot reject the null hypothesis (H0): it is likely that the performance of IJVs in which partner selection was driven by the relational view, will not deviate from the performance of IJVs in which partner selection was driven by the resource-based view.

Table 3: One-way ANOVA test on the significance of mean differences IJV Performance Sum of Squares df Mean Square F Sig. Between Groups 0,39 1,00 0,39 0,00 0,97 Within Groups 2022,51 10,00 202,25 Total 2022,90 11,00

(34)

Furthermore, tables 4, 5, and 6 provide the results of the hierarchical regression analysis. This analysis was performed in order to investigate the mean difference between the partner selection groups in relation to IJV performance, after controlling for size, age, and the

number of directors. In the first step (Model 1) of hierarchical multiple regression, three

predictors were entered: size, age and number of directors. This model was not statistically significant F (3, 7) = 1.81; p > 0.05 and explained 44% of variance in IJV Performance. In the second model, partner selection was added to the regression. After this, the model was still not significant, and the total variance explained by the model as a whole was 55% F (4, 6) = 1,84; p > 0.05. Thus, the introduction of partner selection explained an additional 11% variance in IJV performance, after controlling for size, age and number of directors (ΔR2 = 0.12; F (1, 6) = 1.54; p > 0.05). The final model (Table 6) confirms the findings of the one-way ANOVA that partner selection is not statistically significant, and neither are the other three predictor variables (all p > 0.05). Size, however, recorded the highest Beta value (β = 0.98), and is very near to statistical significance (p = 0.06) when taking into account that we assumed a 95% confidence interval. Thus, although we already rejected our hypothesis (H1), we may also reject any effect of the size, the age, or the number of directors of an IJV on IJV performance (and as so, on the outcome of our study).

Table 4: Model Summary hierarchical regression analysis

Model R R Square Adjusted R Square Std. Error of the Estimate Change Statistics R Square Change F Change df1 df2 Sig. F Change 1 .660a 0,44 0,19 12,17 0,44 1,81 3,00 7,00 0,23 2 .743b 0,55 0,25 11,73 0,12 1,54 1,00 6,00 0,26

a. Predictors: (Constant), Size, Directors, Age

International joint ventures : resource-based vs relation-based partner selection