Sleeping with your enemies?

(1)

Sleeping with your enemies?

The influence of the degree of coopetition in multi-partner R&D alliances on

firm’s financial performance

Master Thesis

MSc Business Administration - Strategic Innovation Management

By Sanne Veenbaas Supervisor: I. Estrada Vaquero Co-assessor: T. L. J. Broekhuizen

Word count: 13,898

(2)

1

“Co-opetition recognizes that business relationships have more than one aspect. As a result, it can occasionally sound paradoxical. But this is part of what makes co-opetition such a

powerful mindset. It’s optimistic, without being naive. It encourages bold action, while helping you to escape the pitfalls. It encourages you to adopt a benevolent attitude towards other players, while at the same time keeping you tough-minded and

logical. By showing the way to new opportunities, co-opetition stimulates creativity. By focusing on changing the game, it keeps

business forward looking. By finding ways to make the pie bigger, it makes business both more profitable and more personally satisfying. By challenging the status quo, co-opetition

says things can be done differently - and better.”

(3)

2

Abstract

This study examines why some firms are better able to reap the benefits from multi-partner R&D alliances than others. It is proposed that the degree of coopetition has an influence on shaping a focal firm’s financial performance when it engages in R&D consortia. Previous research shows that innovation-related multi-coopetition is becoming increasingly popular, particularly in high-tech industries. However, the field still remains in the process of development, especially in terms of its influence on firm performance. Coopetition has been found to be an effective way of creating innovations. However, coopetition also entails some major disadvantages and risks that were found to be more complex within multi-partner alliances. The hypotheses in this paper were tested using a sample of 75 firms that participated in a multi-partner R&D alliance that took place in either a high- or a medium-high-tech industry. Results from the empirical study show that young firms perform better when they face a high degree of coopetition in a MR&D alliance. Furthermore, in case of a high number of partner in a R&D alliance, firms achieve a higher financial performance when most partners are direct competitors. However, it should be noted that hypotheses could only be limitedly supported due to reliability issues, such as the small (sub)sample size(s). Finally, this paper puts forward important directions for future research in the field of innovation-related multi-coopetition.

(4)

3

1. Introduction

At the beginning of the 21st century, open innovation became increasingly popular. This involves the collaboration of companies to enhance the exploitation of innovation in order to achieve competitive advantage (Chesbrough, 2003) and entails the use of purposive inflows and outflows of knowledge (Chesbrough, Vanhaverbeke & West, 2006). Inter-firm R&D collaboration represents one possibility with which firms can gain access to complementary capabilities, shorten development time while spreading the risk and costs of such new developments and benefit from economies of scale in R&D (e.g. Gnyawali & Park, 2009; Mariti & Smiley, 1983, Powell, 1990).

Traditionally, firms benefitted from collaborative innovation by participating in dyadic alliances consisting of complementary partners, such as suppliers and customers (Ritala & Hurmelinna-Laukkanen, 2013). This type of ‘friendly’ collaboration has been considered a means for utilizing synergies in the participants’ knowledge bases and thus improving firm performance. However, there has been an increasing interest of coopetition, i.e. collaboration between competitors (see Figure 1). Bengtsson & Kock (2014) recently provided a revised definition of coopetition: “a paradoxical relationship between two or more actors, regardless of whether they are in horizontal or vertical relationships, simultaneously involved in cooperative and competitive interactions” (p. 180).

Fig 1. Number of articles on coopetition issued between 1994 and 2012

(based on the ISI Web of Knowledge’s Social Sciences Citation Index (SSCI) and the EBSCOhost website’s Business Resource Premier (Bengtsson & Kock, 2014, p.181)

(6)

5 innovation-related multi-coopetition. For instance, the consortium led by Sony consisting of its close competitors in order to engage in a high-risk development of Blu-Ray laser disk aiming to substitute the DVD technology and the cooperation of firms operating in capital-intensive industries such as the aerospace and automobile sectors (e.g. Dussauge, Garette & Mitchell, 2000) with the aim to decrease for example development costs. Despite the growing relevance of the topic, the innovation-related coopetition field still remains in the process of development, especially in terms of its influence on firm performance. There is thus a need to further research this field. Previous studies on innovation-related coopetition found that collaboration between competing firms include unique characteristics that are lacking in alliances comprising noncompetitors and that these characteristics may lead to differences (sometimes better and sometimes worse) in terms of innovative performance (Ritala & Hurmelinna-Laukkanen, 2009). In addition, (e.g. Valdés-Llaneza & Garciá-Canal, 2006) found evidence that, in case of a higher number of alliance partners, it matters for performance whether partners are direct competitors or noncompetitors. Though, most of these studies focus on the simple distinction between noncompetitors and competitors. Conclusive evidence about the influence of competition intensity in alliances on firm performance is lacking and cooperation with indirect competitors has been largely ignored. Indirect competitors are particularly interesting as they could be wolves in sheep’s clothes. When collaborating with the focal firm, these firms are likely to obtain market knowledge and might decide to use it to directly compete against the focal firm (Gnyawali & Park, 2009).

Das & Teng (2002) argue that a collaboration of at least three partners should be treated as a distinct type of alliance for appropriate theoretical development, because compared to dyadic alliances the likelihood is higher that partners experience relational conflicts, opportunistic behaviour and partners’ misalignment of goals. Furthermore, Das & Teng (2002) found that constellations share salient features such a higher need for trust and a higher risk of free riding, because of the lower level of transparency. Nonetheless, the distinction between multilateral and dyadic alliances has been largely ignored in the existing literature in studying firm performance. Due to the aforementioned significant differences, it is important that managers are aware of these features of multi-partner alliances and do not treat them equally as the less complex dyadic alliances.

As earlier research has mostly focussed on alliance performance, this paper studies the focal firm’s financial performance, which is of ultimate interest for a firm. This paper provides insights for managers who are about to start a MR&D alliance, what the influence of the different degrees of coopetition is on firm performance. In that way, they are able to more confidently decide with whom they are going to form a MR&D alliance.

(7)

6 Based on the current trends in the field of coopetition and the voids in the literature, the following research question has been developed:

What is the influence of the degree of coopetition in multi-partner R&D alliances on firm’s financial performance?

It is proposed that in MR&D alliances, firm performance is influenced by the degree of coopetition in the form of an inverted U-shape (1). Furthermore, the number of partners in the alliance is expected to moderate this relationship so that it becomes a positive linear relationship (2) and (younger) older firms are expected to perform higher when allying with (non)competitors (3). In order to test the research hypotheses, a quantitative analysis was performed using data from a sample of 75 firms participating in MR&D alliances obtained by consulting the SDC Platinum database. The alliances take place in either a high- or a medium-high-tech industry as in these industries MR&D alliances and innovation-related coopetition (Ritala & Hurmelinna-Laukkanen, 2009).

The findings show that relatively young firms achieve a better financial performance when they face a high degree of coopetition in a MR&D alliance. Furthermore, when there is a relatively high number of partners in a R&D alliance, firms perform better when most partners are direct competitors. However, it should be noted that hypotheses could only be limitedly supported due to reliability issues, such as the small (sub)sample size(s).

This paper contributes to the strategic alliances literature, focusing on innovation-related multi-coopetition. Specifically, it contributes to the literature on competitive and relational advantages in MR&D alliances. Managers will benefit by knowing the performance implications of allying with a direct competitors, indirect competitors or partners that are considered to be no competitor at all for the focal firm.

(8)

7

2. Literature Review and Hypotheses

In this section, a review of the literature is provided on why coopetition is important for innovation. Furthermore, trends in the field of innovation-related coopetition are outlined which shed light on the motivations of firms to engage in coopetition in the form of MR&D alliances.

2.1 Role of Coopetition in Innovation

If innovating firms want to stay competitive in today’s global markets, collaborating with competitors is essential (Jorde & Teece, 1990). Adopting the definition of Gimeno (2004), this paper defines competitors as firms that have a high niche overlap, indicating that these firms are substitutes for one another in markets. Furthermore, these firms compete for the same limited resources or target market (McPherson, 1983). The opposite holds for noncompetitors. In a response to the technological development and the globalization of competition, R&D consortia have become increasingly popular (Das & Teng, 2002). These trends entail three major challenges – increasing R&D expenditures, convergence of multiple technologies and shorter product life cycles, and are important drives for firms operating in high-tech industries to engage in coopetition (Gnyawali & Park, 2009). Technological convergence provides opportunities for firms to set industry standards. Competitors collaborate with each other to win battles for these standards (Gomes-Casseres, 1994) which lay a foundation for the development of new products and services (Lei, 2003).

Coopetition is especially popular in high-tech industries (Gnyawali and Park, 2009) and is an important means to acquire new technological knowledge and skills from partners (Quintana-García & Benavides-Velasco, 2004). Engaging in coopetition for enhancing innovation outcomes, entails both advantages and risks. Therefore, it is necessary to examine under what conditions firms generate optimal performance in coopetitive relationships (Gnyawali and Park, 2011). The most prominent theoretical approach to innovation-related coopetition includes arguments form transaction costs economics and the resource-based view (e.g. Quintana-García and Benavides-Velasco, 2004). When looking at the transaction cost theory, innovation-related coopetition is an extremely risky business, because competitors are considered to have individual incentives that are likely to lead to opportunism (e.g. Park & Russo, 1996; Quintana-García and Benavides-Velasco, 2004). This theory thus implies that firm performance is higher when they collaborate with noncompetitors. Conversely, the resource-based view considers coopetition as a lucrative relationship with regard to innovation activities (e.g. Brandenburger & Nalebuff, 1996, Dussauge et al., 2000). The similar knowledge bases and a common market vision that competitors possess, supports the collaboration they participate in (Ritala & Hurmelinna-Laukkanen, 2009). However, the generation completely new innovations requires varying types of knowledge which are likely to be obtained outside the focal firm’s industry (Laursen & Salter, 2006). In line with this, Park, Srivastava & Gnyawali (2014) found evidence that “… a moderate level of competition contributes more to the focal firm’s coopetition-based innovation output than do very low or very high levels of competition with alliance partners” (p. 3). This implies that partners should operate in related industries which are often indirect competitors.

2.2 Coopetition in MR&D Alliances

(9)

8 The former is more difficult to coordinate because there are more players involved; they share salient features such a higher need for trust and a higher risk of free riding (Das & Teng, 2002). “The major quantitative and qualitative transformation takes place when moving from two to three partners, as the third partner may pave the way for the appearance of risky pair coalitions between the partners” (García-Canal, Valdés-Llaneza & Ariño, 2003, p. 747). Communication among partners becomes more complicated, as the number of communication channels rises from one to three (Pfeffer and Salancik 1978). In alliances with a lower amount of partners (including dyadic alliances), observing partners’ actions is more easily and formal governance may be redundant (Lavie et al., 2007). Another salient feature of constellations which is related to this, is the concept of reciprocity. In constellations, members often do not directly reciprocate with each other and thus reciprocity can be problematic. Generalized reciprocity, which is defined as “… a group-based exchange relationship in which members expect quid pro quo exchanges within the group but not necessarily with any specific member” (Das & Teng, 2002, p. 449), connotes potential free riding. This entails that a participant contributes resources to the alliance (so the entity itself) and also expects reciprocity from the alliance. This is different from direct reciprocity, in which partners expect reciprocity from each other. The existence of generalized exchanges explains the high level of complexity of MR&D alliances. Though, it does provide the basis for a trust-building process (Ekeh, 1974) and in turn can lead to better constellation performance (Das & Teng, 2002).

These complex dynamics of collaboration in the form of MR&D alliances become even more complex in the presence of coopetition. For instance, according to the transaction cost theory, the high level of opportunism that is present when collaborating with competitors, requires more coordination for an alliance to be a success. However, in a constellation, reciprocity takes place through generalized exchanges, which increases the difficulty of tracking the efforts each partner has made. Therefore, consortia consisting of competitors are even more challenging to manage and thus more complex. Furthermore, knowledge sharing is crucial for success in R&D alliances, as the aim is to generate new ideas and develop new products for which the novel (re)combinations of knowledge and capabilities is needed (Schumpeter, 1991). According to the transaction cost theory, however, firms act opportunistically and fear unwanted knowledge spillovers that they expect competitors to use against them. Obviously, this hinders knowledge sharing and thus decreases firm performance (Sampson, 2007). These unique characteristics and consequences of the complexities in MR&D alliances, underscore the importance of treating it not equally as dyadic alliances, in particular when the consortium consists of competitors. Treating it equally can give managers a distorted picture of expected firm performance, with dire consequences.

(10)

9 levels of market and technological uncertainty, shortened product life cycles, increased demand for R&D investments and the need for product interoperability (Lavie et al., 2007).

2.3 Gaps in the Literature

As the literature review revealed, there already exists a reasonable amount of research about the phenomena multi-partner alliances and coopetition. However, research concerning the combination of the two is limited. In addition, a MR&D alliance is a type of consortia that is becoming increasingly commonplace in business practice, but is rather understudied. Furthermore, with regard to the influence of coopetition on firm performance, the field is still in the process of development. The intriguing question of why some firms are better able to reap the benefits from MR&D alliances remains unanswered. The aim of this paper is to fill this gap by studying the degree of coopetition which is expected to influence firm’s financial performance. Other than most previous studies in the field of coopetition, this paper also takes into consideration the advantages and threats of collaborating with indirect competitors (next to direct competitors and noncompetitors). These firms have more different knowledge bases than direct competitors which operate in the same industry and therefore the focal firm is exposed to more new knowledge. Furthermore, their knowledge bases are not so distant as those of noncompetitors. When knowledge is too distant from the firm’s existing knowledge base – its absorptive capacity – it may not be able to understand it and consequently is hindered in learning from its partner. As this study recognizes the importance of this group as well for focal firm’s performance, this paper studies three degrees of coopetition: low (most partners are noncompetitors), moderate (most partners are indirect competitors) and high (most partners are direct competitors).

2.4 Hypotheses

(11)

10 competitors (Park et al., 2014), which will hinder knowledge sharing (Sampson, 2007) due to fear of unwanted knowledge spillovers (Yan & Luo, 2001). As a result, innovation through knowledge recombination is impeded and monitoring and safeguarding costs increase which will damage innovative (Park et al., 2014) and financial performance. These negative effects are especially prevalent in multi-partner alliances as reciprocity takes place through generalized exchanges which hampers monitoring partners’ efforts (Das & Teng, 2002). Furthermore, as partners’ efforts are difficult to track for they do not reciprocate directly with one another, there is a higher chance of free riding (Das & Teng, 2002). Finally, appropriation capacity is constrained when most partners compete in the focal firm’s industry which negatively influences its financial performance (Lavie, 2007).

The factors discussed above suggest that, in MR&D alliances, at both ends of the degree of coopetition tensions exists. In case of a low degree of coopetition there are difficulties in understanding partners and creating synergies. Though, novel ideas and resource access benefits may provide high performance in high-tech industries. When competition increases, relative absorptive capacity is enhanced. Through this, learning is improved and consequently firm’s financial performance will be stronger (Xu & Li, 2008). However, after a certain threshold of the degree of coopetition, the negative effects of competitive tension increase rapidly and learning is likely to stop (Park et al., 2014). When the degree of coopetition reaches a too high level, the benefits of competition thus decline, i.e. the positive effects of learning are surpassed. Furthermore, the stimulation of technological innovations is reduced (Jiang et al., 2010). Consolidating the above key arguments, as the degree of coopetition between the focal firm and its partners increases, the positive effects of learning reach a plateau at a certain point and so does a firm’s financial performance. Beyond that point, individual incentives become too strong and the negative effects of competitive tension surpass the positive effects (Park et al., 2014). Concluding, there should be a moderate degree of coopetition in MR&D alliances to let firm’s financial performance peak. Therefore, the following is hypothesized:

H1: The degree of coopetition is associated with firm’s financial performance in a MR&D alliance through an inverted U-shape.

2.4.2 Hypothesis 2. In many studies, firm age is used as a control variable. In the field of innovation-related coopetition, however, this paper emphasizes it is an important factor in shaping financial performance in different coopetitive situations and should therefore be treated as a moderator. Young and old firms possess distinctive characteristics. Due to this, this study expects that younger firms perform better in a certain coopetition setting than older firms do and vice versa.

(12)

11 younger firms, have a large resource pool and therefore offer opportunities for economies of scale in R&D (e.g. Gnyawali & Park, 2007). By exploiting these opportunities by allying with firms that have similar resources (often competitors) that can be combined, firms may achieve a great improvement in financial performance, especially in the case of multiple partners (i.e. larger scale). However, established firms are burdened with overcoming major management challenges of adaption that could exist due to the presence of core rigidities and path dependence (Leonard-Barton, 1992). In line with this, Noteboom et al. (2007) found that organizational age has a positive effect on exploitation, i.e. improving and refining existing technology. However, older firms perform worse on exploration.

The opposite holds for younger firms. Innovation is crucial for success in high-tech industries due to the fast pace of technological development. When firms have the aim to generate novel products, it requires various types of knowledge which are likely to be obtained outside the focal firm’s industry (Laursen & Salter, 2005), representing a low degree of coopetition. Younger firms have learning advantages in new areas and organizational flexibility (Choi & Shepherd, 2005) which are highly beneficial in high-tech industries. Furthermore, they may use the fertile base of a MR&D alliance to develop collaborative advantage, which could lead to increased innovation performance (Almeida, Dokko & Rosenkopf, 2002) and subsequently to a higher financial performance (Kostopoulos et al., 2011; Lahiri & Narayanan, 2013).

Concluding, when looking at the characteristics young firms have, they are expected to perform better in alliances where there is a low degree of coopetition. The characteristics of older firms, on the other hand, allow for incremental innovation. Firms that aim for incremental innovation more often search for partners in familiar markets and industries (Ritala & Hurmerlinna-Laukkanen, 2013). In addition, older firms have built a highly specialised knowledge base over the years. Therefore, these firms best improve their financial performance by exploiting their knowledge by allying with its competitors, representing a high degree of coopetition. In line with the above discussion, the following hypothesis is formulated:

H2: In MR&D alliances, younger firms benefit most when they participate in an alliance with a low degree of coopetition (a) and older firms benefit most when they participate in an alliance with a high degree of coopetition (b).

(13)

12 al., 2003, p. 747). Communication among partners becomes more difficult, as the number of communication channels rises from one to three (Pfeffer and Salancik, 1978).

Furthermore, incentives for free-riding behaviour are greater in the case of a higher amount of participating firms. This can be linked to the concept of generalized reciprocity. Due to the decreasing intransparency of behaviour as the number of partners increases, it is harder to control and evaluate partners and reciprocity becomes more difficult to implement in an alliance (Parkhe, 1993). This is because “… when there is more than one other ﬁrm it is very difﬁcult to punish the non-cooperative behaviour of one partner without worsening the situation of the others (García-Canal et al., 2003, p. 749)”. This, and the fact that the number of dyadic relationships increase geometrically as the number of partners increases (Park & Russo, 1996), makes a relationship more conflict prone.

On the contrary, Park & Russo (1996) found that the higher the number of partners in a joint venture, the less likely it was to fail. Especially in R&D alliances a high number of partners can be beneficial in terms of idea generation, which is a crucial part of the R&D process. As Yami & Nemeh (2013) argue, when more partners participate in an alliance, novelty and diversity of knowledge will be enhanced, which is an important condition for radical innovation and setting industry standards. In addition, McFadyen and Cannella (2004) suggest that “increasing the number of direct exchange partners in a network increases the amount of information, ideas, and resources in it” (p. 737). Moreover, in the case of a large number of firms in an alliance, fixed costs of research can be spread over a larger base (Henderson & Cockburn, 1996), which again will have a positive influence on firm’s financial performance.

Valdés-Llaneza & García-Canal (2006) suggests a potential explanation for the inconsistent findings as well. They namely found that in the case of increases in the number of partners, the success of the alliance in terms of longevity depends on the degree of competition. Specifically, increases in the number of partners have a positive (negative) effect on the longevity of stakes in joint ventures when all of the partners are (non)competitors. This paper studies whether these effects are the same for the focal firm’s financial performance.

(14)

13 All of this positively affects the financial performance of a firm. Therefore, when the degree of coopetition increases, the problems derived from conflicts of interests will be counteracted by the greater potential for value creation that can be achieved from the alliance when the number of partners increases (Valdés-Llaneza & García-Canal, 2006). Therefore, the firm’s financial performance is expected to further increase as opposed to the decrease after a certain threshold as represented in hypothesis 1 (which did not take into account the influence of the number of partners). Therefore, the following hypothesis is formulated:

H3: The relationship between the degree of coopetition and firm’s financial performance is moderated by the number of partners in such a way that it becomes a positive linear relationship.

Below the conceptual model for this study is presented, providing an overview of the three hypotheses studied in this paper.

Degree of coopetition Firm financial performance H1: +/-Moderators: · Firm age (H2: +) · Number of partners in alliance (H3: +) Control variables: · Firm size (log) · R&D intensity · R&D agreement · Strategic fit · Industry · Multi-partner alliance experience · Previous experience with a partner

(15)

14

3. Methodology

3.1 Data and Sample

The sample for testing the hypotheses consists of 75 firms from multiple countries and has been taken from the firm-level Securities Data Corporation’s (SDC) Platinum database, which consists information on the year an alliance was established, the industry in which the alliance takes place, the number of partners in the alliance, the name of the participants and the primary industry they operate in. In total, 48 MR&D alliances were studied (financial data was not available for all firms in an alliance). The firms were selected based on industry information (provided in SIC codes) about the alliance they participate in. They have participated in a multi-partner alliance taking place in either a medium-high-tech or a high-tech industry. Specifically, the firms studied in this sample participated in multi-partner alliances that started between the year 2005 and 2010 and are innovation-related (for instance joint product development agreements).

The particular setting for this study is appropriate for several reasons. First, firms in high-tech industries have R&D intensities between 10 and 15% and medium-high-tech industries between 3 and 5% (Hagedoorn, 2002). Second, in these industries, especially in the former group, the motivation applies of sharing the costs of R&D activities with one or more other companies as in these industries the cost of single, large R&D projects are beyond the reach of many companies (Hagedoorn, 1993). Third, MR&D alliances are becoming increasingly commonplace in these industries, as they represent effectual strategies in dealing with innovation activities by combining multiple complementary resources (Doz et al., 2000; Mothe & Quelin, 2001; Lavie et al., 2007). In order to enhance generalizability, the sample consists of firms operating in different industries and countries. The firms in the sample for this study participated in alliances operating in one of the six major groups with the SIC Codes: 28XX, 35XX, 36XX, 37XX, 38XX and 48XX. These include the following industry groups1:

Additional information on firm age, financial measures (ROA, profit margin and R&D expenditures) and firm size are retrieved from the database Orbis.

In order to increase the validity of this research, the application text of the alliance has been checked in order to ensure it concerns a (medium-)high-tech alliance with an innovative purpose (e.g. product development).

1

OECD Directorate for Science, Technology and Industry Economic Analysis and Statistics Division. (2011). ISIC

Rev. 3 Technology intensity definition. Retrieved April 22, 2014, from http://www.oecd.org/science/inno/

(16)

15

3.2 Measures

3.2.1 Dependent variables.

FINANCIAL PERFORMANCE. The firm’s financial performance is measured by the change in return on assets (ROA). It is a widely accepted measure to assess operating performance and it measures how effectively profits are generated by using the firm's available assets. This is relevant for studying agreements for joint product development, as a successfully developed product will yield high financial performance.

For this study, it is interesting to study the change in ROA before and after the alliance. For this, a time frame of four years is studied, including one year prior to the alliance and three years after the start of the alliance. In this way, the influence of the alliance on firm performance can be measured. For this, the following formula (Porrini, 2004) has been used:

ROAchg = (ROA+3 − ROA−1) / ROA−1

ROAchg is the change in ROA from year −1 to year 3, with year 0 being the starting year of the alliance. In order to enhance reliability of the research, an additional financial measure has been included, which is the change in profit margin (PM) and is based on the previous formula:

PMchg = (PM+3 − PM−1) / PM−1

3.2.2 Independent variables.

DEGREE OF COOPETITION. The degree of coopetition is measured from the perspective of the focal firm and is based on the method of Wang & Zajac (2007) and Noseleit & De Faria (2013). Both papers used SIC codes to differentiate among competition intensities. The following categorization is used: (i) high degree of coopetition, (ii) moderate degree of coopetition and (iii) low degree of coopetition. In order to determine the degree of coopetition the focal firms faces in an alliance, partner firms were scored. Partner firms that are direct competitors, i.e. the firm shares the same primary 4-digit SIC code with the focal firm, are coded with a 3. Partner firms that are indirect competitors, i.e. the firm is primarily active in the same 2-digit industry, but not in the same 4-digit industry as the focal firm, are coded with a 2. Partner firms that are no competitors, are those firms that possess all other SIC codes and are coded with a 1.

As this thesis studies multi-partner alliances, there are firms that cooperate with partners who can have different degrees of coopetition in relation to the focal firms. For instance, it can happen that a firm has three different partners; one that is a direct competitor, one that is an indirect competitor and the other is considered to be no competitor. In order to determine the degree of coopetition the focal firm is facing, the average of the assigned scores is taken by using the following formula2:

DOC = P1 + P2 + … Pn / n

Here, DOC is the degree of coopetition expressed in the average of the ordinal variable codes, P... includes the score of a partner and n is the number of partners in the alliance. Then, we speak of a (i) high degree of coopetition when the majority of the partner firms in the alliance shares the same

2

(17)

16 primary 4-digit SIC code with the focal firm (DOC = 2.5-3), (ii) moderate degree of coopetition when the majority of the partner firms in the alliance shares the same 2-digit SIC code, but not the same 4-digit SIC code as the focal firm (DOC = 1.5 < 2.5) and (iii) low degree of coopetition when the majority of the partner firms possesses all other SIC codes (DOC = 1 < 1.5). In addition, it was ensured that when for instance the degree of coopetition was 2.0, that indeed most (or in the case of 2 partners, at least 1) of the partners are indirect competitors3.

Noseleit & De Faria (2013) stated that, in the case of the majority of the partners sharing the same 4-digit SIC Code, partners are “… likely to display the highest chances of being a direct competitor of the focal firm and a strong overlap in the knowledge base” (p. 2002). With regard to the second classification, such a type of alliance “… consists of firms that have a lower likelihood to be direct competitors and show a smaller overlap in the knowledge base with the focal firm” (Noseleit & De Faria, 2013, p. 2002). Finally, the third group refers to unrelated alliances in which the majority of the partners are from different industries.

3.2.3 Moderators.

NUMBER OF PARTNERS. Total number of partners in an alliance.

FIRM AGE. In order to determine how many years a firm has been in existence, the year of establishment provided by the database Orbis, was taken.

3.2.4 Multiplicative variables. This research also studies interaction terms in order to examine whether there are moderating effects between the relationship of the degree of coopetition and firm’s financial performance. Therefore, the moderators described in the previous paragraph are multiplied by the degree of coopetition.

DEGREE OF COOPETITION*NUMBER OF PARTNERS. A multiplicative variable of the following variables: Degree of coopetition and number of partners.

DEGREE OF COOPETITION*AGE. A multiplicative variable of the variable degree of coopetition and age. In this case, for the variable age, a dummy variable was used. Firms younger than the median age in this sample (45 years) are coded with a 0 and firms older than the median are coded with a 1. The median was used to divide sample as the data are approximately non normally distributed.

3.2.5 Control variables. Since factors other than those included in the hypotheses may affect a firm’s financial performance, control variables have been included in this research. In this way, they are held constant so that the relative impact of the independent variables in this research can be tested. Below a description is given of the variables previous research found having a strong influence on a firm’s financial performance. Furthermore, the way they are measured is provided. INDUSTRY. In order to control for these differences in industry dynamics (e.g. industry growth), dummy variables were used. Medium-high-tech industries were scored with a 0 and high-tech industries with a 1.

3

(18)

17 FIRM SIZE. Previous studies have found large firms often have a higher financial performance. Therefore, differences in size are controlled for and is measured by the number of employees a firms has. As the firm size distribution is positively skewed, log transformation is used to normalize the distribution (e.g. Cloodt, Hagedoorn & Van Kranenburg, 2006). Due to scale effects, larger firms may have greater performance. Small firms are considered to have less than 1000 employees, very large firms are characterized by having more than 50,000 employees and in between the intermediate size-classes can be found (Cloodt et al., 2006).

STRATEGIC FIT. Previous research (Mothe & Quelin 2011) indicates that the degree of fit between the firm and the alliance or, in other words, the strategic importance, has an influence on firm performance. In case of a higher strategic fit, a firm is able to create more resources such as new products and patents. This will enhance a firm’s competitiveness and subsequently its financial performance (López-Gamero et al., 2010). Specifically, it refers to the extent the alliance objectives concern the core business area of the firm. In order to control for the differences in strategic fit among firms, the SIC code of the alliance with the SIC code of the focal firm’s primary business were compared. For this, scores were assigned to firms. In the case of the focal firm sharing the same primary 4-digit SIC code with the alliance, the firm is coded with a 3. Firms having the same 2-digit code, but not the same 4-digit as the alliance, are coded with a 2. When there is no strategic fit at all, i.e. firms that possess all other SIC codes, are coded with a 1.

PREVIOUS EXPERIENCE WITH ANY OF ITS PARTNERS. Previous experience with a partner has a positive effect on trust and cooperative behaviour (e.g. Kogut, 1989). Consequently, knowledge will flow more freely and financial performance is enhanced (Sampson, 2007). To control for the differences in a firm’s experience with its partner in the focal alliance, dummy variables were used to measure this. Specifically, when a firm allied with any of its partners within 5 years prior to the start date of the focal alliance, then it was code by a 1. When a firm did not have this experience, it received a 0.

PREVIOUS MULTI-PARTNER ALLIANCE EXPERIENCE. Sampson (2005) found that prior experience in alliances characterized by greater complexity has a positive influence on firm performance. In order to capture the differences in firms’ experience with multi-partner alliances, dummy variables were used to measure this. Specifically, when a firm participated in a multi-partner alliance within 5 years prior to the start date of the focal alliance, then it was code by a 1. When a firm did not have this experience, it received a 0.

R&D INTENSITY. A firm’s R&D intensity is positively related to its financial performance (Capon, Farley & Hoenig, 1990). The annual R&D intensity is provided by the Orbis database and is calculated by R&D expenses/operating revenue.

(19)

18

3.3 Methods of Analysis

First, a correlation matrix is presented in order to examine the correlations between the variables used in this research.

Second, a logit regression analysis was performed including the control variables and firm’s financial performance.

Third, for finding evidence for the hypothesized inverted U-shape between the degree of coopetition and firm performance, the squared term of the degree of coopetition is included.

Fourth, for testing hypothesis 2, interaction terms were included in the logit analyses on the full sample in order to test whether the (curvilinear) relation between the degree of coopetition and firm’s financial performance was moderated by firm age. In order to further test hypothesis 2, the sample was split up between relatively young and relatively old firms. This was done by taking the median of the sample (i.e. 45 years) and subsequently firms younger than this median were placed in the former group and firms older than this median were placed in the latter group. The median was used to divide the sample as the data is skewed. Then, logit regression analyses were used to determine how the relationship between the degree of coopetition and firm’s financial performance behaves in these two subsamples. Dependent on the results of this initial analysis, additional analyses are performed which are then further described in the next section.

Fifth, for testing hypothesis 3, interaction terms were included in the logit analyses on the full sample in order to test whether the (curvilinear) relation between the degree of coopetition and firm’s financial performance was moderated by the number of partners in the alliance. In order to further test hypothesis 3, the sample was split up between a relatively high and low number of partners in the alliance. This was done by taking the median of the sample (i.e. 3 partners). The median was used to divide the sample as the data is skewed. However, as alliances with 3 partners were dominant in the sample in comparison to alliances with a higher number of partners, the data was not split up fifty-fifty. Instead 67% of the firms participated in an alliance with 3 partners and 33% in alliances with a higher number of partners. The sample was divided in (only) two group in order to avoid very low numbers of observations in groups which would decrease the reliability of the results. Then, logit regression analyses were used to determine how the relationship between the degree of coopetition and firm’s financial performance behaves in these two subsamples. Dependent on the results of this initial analysis, additional analyses are performed which are then further described in the next section.

In order to make inferences about the predictive power of the model, R2, adjusted R2 and the F value were calculated. Furthermore, a multicollinearity test was performed in order to ensure that the independent variable is indeed a predictor for the dependent variables. The reliability was enhanced by replicating models using an alternative measure for firm’s financial performance. Next to the change in ROA, this study used the change in profit margin.

(20)

19

4. Results

4.1 Descriptive Statistics and Correlation Matrix

Table 1 (presented on the next page) reports the means, standard deviations, minimum and maximum values and Pearson correlations for the variables. The number of partners in an MR&D alliance ranged from 3 to 7, of which 50 (67%) had 3 partners. Furthermore, 24 (32%) firms are in an alliance with a low degree of coopetition, 23 (31%) firms are in an alliance with a moderate degree of coopetition and 28 (37%) firms are in an alliance with a high degree of coopetition. Finally, the majority of the firms (48; 64%) participated in a high-tech alliance. On average, the R&D expenditure per year accounts for 4.98% of annual operating revenue.

(21)

(22)

21

4.2 Multicollinearity

To ensure that the independent variable (degree of coopetition) is indeed a predictor for the dependent variables change in return on assets and profit margin, an initial analysis of multicollinearity was executed. It was examined whether the tolerance and variance inflation factor (VIF) levels of all of the variables had values well within acceptable levels. Several recommendations for acceptable levels of VIF have been provided in the literature. The most common maximum value is 10 (e.g. Kennedy, 1992; Neter, Wasserman, & Kutner, 1989). Some researchers, however, recommend a maximum VIF value of 5 (e.g. Rogerson, 2001) or even 4 (e.g. Pan & Jackson, 2008). In this study, none of the VIF factors exceeds the value 2, showing that multicollinearity is not a problem.

4.3 Sample Characteristics

Prior to deeper data analysis, tests were performed in order to determine whether the data is approximately normally distributed. A Shapiro-Wilk’s test (Shapiro & Wilk, 1965) and a visual examination of their histograms, box plots, normal Q-Q plots and cross tabulation showed that for instance the change in ROA is not approximately normally distributed for all three degrees of coopetition (DOC), with a skewness of -4.220 (SE = 0.472) and a kurtosis of 18.891 (SE = 0.918) for a low degree of coopetition (DOC = 1), a skewness of 2.293 (SE = 0.481) and a kurtosis of 4.161 (SE = 0.935) for a moderate degree of coopetition (DOC =2) and a skewness of -2.447 (SE = 0.411) and a kurtosis of 15.653 (SE = 0.858) for a high degree of coopetition (DOC = 3). A non-parametric Levene’s test verified the equality of variances in the samples (homogeneity of variance) (p > 0.05) (e.g. Nordstokke & Zumbo, 2010). The data could not be normalized through either a log or a square root transformation; the data remained non-normally distributed. Therefore, non-parametric test had to be used.

4.4 Hypotheses tests

Table 2 (presented on the next page) introduces the effects of the degree of coopetition on firm’s financial performance. Model 1 presents the results for the control variables. It was found that R&D intensity has a negative effect on financial performance and a positive effect was found when there is strategic fit (which is consistent with prior research) and when a firm participates in an alliance taking place in a high-tech industry. Model 3 of the logit analyses on the full sample shows that the hypothesized inverted U-shape is not significant. Despite the fulfilled criteria of the degree of coopetition being a positive correlation coefficient and the squared term of it being a negative correlation coefficient suggesting the shape of an inverted U, it does not have a p-value of at least less than 0.10. Therefore, hypothesis 14 is not supported. Further, model 2 shows that there is neither a significant positive or negative linear relationship between the degree of coopetition and firm performance. Replicating the model using the dependent variable profit margin change neither provided support for hypothesis 1. It should be noted, however, that the sample size in this model is rather low as well as R2 and the F-test is not significant. This decreases the reliability of results.

4

(23)

22 Table 2. Results Logit Analyses on Full Sample

Dependent variable: ROA Change

Model 1 Model 2 Model 3 Model 4

Intercept -1.197 -2.087 -3.709 -3.748 Firm size -.092 -.084 -.082 .012 R&D intensity -.371*** -.370*** -.368*** -.375*** R&D agreement -.104 -.107 -.106 -.118 Strategic fit .210* .198 .191 .170 Industry .253** .264** .262* .176 Multi-partner experience .070 .065 .066 .089

Previous experience with partner -.108 -.106 -.106 -.088

Degree of coopetition .029 .196 .557

Degree of coopetition2 -.167 -1.602

Degree of coopetition x number of partners

-.550 Degree of coopetition x firm age

Degree of coopetition2 x number of partners

-.246 1.695

Degree of coopetition2_{x firm age} _-.082

Number of observations 75 75 75 75

R2 .163 .164 .164 .182

Adjusted R2 .076 .062 .048 .037

F 1.866* 1.615 1.418 1.398

Note:* p <0.10; ** p < 0.05; *** p < 0.01

The interaction term incorporated in model 4 with firm age did not generate significant results. Replicating the model using the dependent variable profit margin change neither provided support for hypothesis 25. The additional test using logit analyses in the two subsamples yielded an opposite result of what was initially expected (see Table 3a and 3b on the next page). The relationship suggested in hypothesis 2a turns out to be a linear positive (instead of negative) one (0.343; p < 0.10) and is, although not strongly, significant. Hypothesis 2b does not receive statistical support. Replicating the models using change in profit margin did neither yield any significant results.

5

(24)

23 Table 3a. Results Logit Analyses on Sub Sample: Relatively young firms (< 45 years)

Dependent variable: Return on assets

Model 1 Model 2 Model 3

Intercept -9.971 -17.977 -27.001 Firm size .071 .022 0.23 R&D intensity -.457** -.428** -.413** R&D agreement .209 .183 .189 Strategic fit .300* .261 .241 Industry .134 .252 .266 Multi-partner experience .165 .100 .081

Previous experience with partner -.138 .080 .131

Degree of coopetition .343* 1.220 Degree of coopetition2 -.851 Number of observations 36 36 36 R2 .326 .391 .400 Adjusted R2 .163 .217 .200 F 2.003* 2.249* 1.998* Note: * p <0.10; ** p < 0.05; *** p < 0.01

Table 3b. Results Logit Analyses on Sub Sample: Relatively old firms (> 45 years) Dependent variable:

Return on assets

Intercept 8.235 15.758 -2.381 Firm size -.171 -.263 -.252 R&D intensity -.035 -.018 .035 R&D agreement -.380* -.362* -.367* Strategic fit -.017 .64 -.065 Industry .272 .191 .106 Multi-partner experience .029 .103 .162

Previous experience with partner -.022 .003 .002

(25)

24 Taking into consideration the small subsample size (N = 36) and the rather low R2 (though still acceptable), additional analyses were performed which are not affected by the number of observations. First, cross tabulation was performed (see Table 4a and b). As no more than 20% of the cells should have expect counts less than 5, the interval variables were transformed into ordinal variable so that smaller groups were created. Specifically, the degree of coopetition was split up in two groups: relatively low (scored 0) and high (scored 1) degree of coopetition. Equally, firm’s financial performance measures were divided into two groups: relatively low (scored 0) and relatively high (scored 1) performance. In order to test to hypothesis 2, the relationship between the degree of coopetition and firm’s financial performance was tested in two subsamples: relatively young (scored 0) and relatively old (scored 1) firms. Medians were used to divide the sample as the data is skewed. The results are reported in Table 4a and 4b presented on the next page.

(26)

25 Table 4b. Crosstabulation relatively old firms (> 45 years)

(27)

26 Table 5a. Mann-Whitney U Test (relatively young firms) - Ranks

Table 5b. Mann-Whitney U Test (relatively young firms) – Test statisticsa

With regard to hypothesis 36, the interaction terms did not yield significant results (see Table 2, model 4). Replicating the model using the dependent variable profit margin change neither provided support for hypothesis 3. The results of the logit analyses on the two subsamples are reported in Table 6a and 6b presented on the next page. This additional analysis partially supports hypothesis 3. That is, when a firm participates in an alliance with a high number of partners (> 3 partners), the firm’s financial performance increases as the degree of coopetition increases (0.780, p < 0.05). Replicating the model using profit margin change as the dependent variable confirmed this relationship (0.911, p = 0.011). This partly supports hypothesis 3.

6

(28)

27 Table 6a. Alliances with a relatively low number of partners (3 partners)

Dependent variable: ROA Change

Intercept -4.559 -4.979 1.108 Firm size -.008 -.005 -.005 R&D intensity -.378** -.378** -.394** R&D agreement -.095 -.099 -.111 Strategic fit .177 .174 .190 Industry .275 .278 .286 Multi-partner experience .033 .029 .029

Previous experience with partner -.147 -.143 -.151

Degree of coopetition .012 -.518 Degree of coopetition2 .531 Number of observations 49 49 49 R2 .194 .194 .317 Adjusted R2 .060 .037 .090 F 1.445 1.235 1.394 Note:* p <0.10; ** p < 0.05; *** p < 0.01

Table 6b. Alliances with high number of partners (> 3 partners) Dependent variable:

ROA Change

Intercept -2.655 -13.620 -12.396 Firm size .554 .895** .939** R&D intensity .171 .373 .403 R&D agreement .269 .490 .543 Strategic fit -.083 -.378 -.333 Industry -.646 -.375 -.355 Multi-partner experience -.368 -.246 -.237

Previous experience with partner -.499 -.894** -.953**

(29)

28 The R2 for Model 2 in Table 6b indicates that the model has a fairly high prediction power. Nonetheless, the sample size is low (especially in the subsample > 3 partners) which decreases the reliability of the results. Therefore, additional analyses were performed that are not affected by the number of observations. First, cross tabulation was performed, but it was not possible to test the data. Even though the data was divided in a minimum amount of two groups, the requirements for the Chi-square to be reliable were not met (25% of the cells had expected count less than 5). In a response to this, cross tabulation was performed on the full sample in order to examine how the data was distributed. For this, the degree of coopetition was transformed in an ordinal variable with 3 groups and the number of partners in the alliance were divided into two groups using the median again. A relatively low number of partners in the alliance scored 0 and high number scored 1. The results are shown in Table 7. A noticeable result is that, in the case of a high degree of coopetition, the number of partners in an alliance the focal firm participates in is highly unequally distributed. Therefore, the significant result in the logit analysis (Table 6b) is rather unreliable, as through the unequal distribution a distorted picture is given.

Table 7. Cross tabulation full sample: Degree of coopetition x number of partners

(30)

29

5. Discussion

It is evident that some firms are better able to reap the benefits from innovation-related coopetition than others (Ritala & Hurmelinna-Laukkanen, 2013). However, Bengtsson, Eriksson & Wincent (2010) emphasized that “there is a lack of knowledge about the effects of coopetition and different types of coopetitive interactions” (p. 210) and that there is a need for more empirical research. In addition, multi-coopetition is has become increasingly popular as a means to tackle the innovation challenges nowadays in high-tech industries. Das & Teng (2002) stress that constellations should be treated as a distinct type of alliance as they are more complex than dyadic alliances. That is exactly where this paper fits in: this paper has studied innovation-related multi-coopetition. To deepen our understanding about why some firms are better able to reap the benefits from R&D collaboration than others, the present study examined the role of the degree of coopetition in shaping firm’s financial performance.

First, this study did not find significant results for the hypothesized inverted U-shape relationship between the degree of coopetition and firm’s financial performance. This is not in line with previous research (e.g. Park et al., 2014) that found an inverted U-shape relationship between competition intensities and firm’s innovative performance. However, it should be noted that this study exclusively studied multi-partner alliances. As this type of alliance is highly complex, it could have an influence on coopetition dynamics. As outlined in the theoretical section of this paper, every degree of coopetition has advantages and disadvantages. A reason for the insignificance of the inverted U-shape may be that certain factors, that were not (deeply) examined in this research, have a stronger positive effect than expected. For instance, in the case of a high degree of coopetition, the advantage of similar knowledge bases might be larger than was expected in this study. Therefore, the change in return on assets would not decrease after a certain threshold. Those factors have been proved to influence firm’s financial performance and may be important issues for further research. Furthermore, the risk of cooperating with indirect competitors could have been underestimated. During the collaboration, partners may find out about (parts of) the focal firm’s intellectual property that it uses to be competitive in its market. Especially, for partners in related industries (compared to partners in unrelated industries) the step to become a direct competitor is rather small. In case this happens, the focal firm’s profits will be damaged (Gnyawali, He & Madhaven, 2008). Another reason for the nonsignificant results is likely to be the small sample size of this study.

(31)

30 at a low level (Estrada et al., 2010). Consequently, younger firms’ ability to draw value from the collaboration is low, which again negatively affects firms performance (e.g. George, et al., 2001). Third, to gain further insight into the role the number of partners has on the relationship between the degree of coopetition and firm’s financial performance, the sample was split up between firms participating in an alliance with a relatively low number of partners and a high number of partners. The number of partners has often been designated as a control variable in studying alliances. However, Valdés-Llaneza & García-Canal (2006), who studied the longevity of stakes in joint ventures, implicated that the number of partners should be treated as a moderating variable when studying firms that collaborate with (non)competitors. In line with their research, this study found that when partners are direct competitors, the problems derived from conflicts of interests are counteracted by the greater potential for generating value from the MR&D alliance when it consists of a higher number of partners (though, there are doubts about reliability). By pooling the resources of competing firms in the same industry, economies of scale and the reordering of industry rivalries can be achieved which has found to be beneficial for firm’s financial performance. The difficulty for a single firm outside the alliance to achieve the critical mass reached through the alliance on its own, increases with the number of partners in the alliance becoming greater (Valdés-Llaneza & García-Canal, 2006; Hwang & Burgers, 1997). Specifically, this study found that when firms are in an alliance with a high number of partners, they perform better when there is a high degree of coopetition, i.e. a high number of direct competitors. However, due to reliability issues, this finding should be seen more as an indication for future researchers (instead of evidence) that the number of partners in a MR&D may have an influence on firm’s financial performance.

Another reason for the insignificant results can be that the role of complementarity was not studied. Researchers (e.g. Teece, 1986) emphasized the importance of the role of external complementary assets for innovation success. They assumed that the absence of niche overlap7 (in this study that would mean a low degree of coopetition) necessarily implicates the presence of complementarity. However, Gimeno (2004) states that “low niche overlap may be a necessary condition for complementarity, but it is not a sufficient condition: firms may be dissimilar without being complementary” (p. 822).

Finally, insignificant results may be as well due to appropriation mechanisms playing a significant role in shaping firm performance. Especially, when partners compete in the same industry as the focal firm, appropriation capacity is constrained (Lavie, 2007), which may decrease a firm’s financial performance. Ritala & Laukkanen (2013) found evidence that firms with a well-developed appropriability regime are able to achieve better results from coopetition in terms of innovation. The next section provides the conclusion for this paper.

7

(32)

31

6. Conclusion

The present paper has sought to shed light on the influence of the degree of coopetition in multi-partner R&D alliances on firm’s financial performance. The results show that the degree of coopetition in MR&D alliances does not matter for firm performance. However, in the case of a relatively high number of partners in the alliance, this study found that allying with direct competitors increases firm performance.

The results show that relatively young firms that participate in R&D consortia are able to achieve a better financial performance when they ally with direct competitors. This is opposite of what was expected, but can be explained by the less developed absorptive capacity these firms have. That is, the knowledge of direct competitors is ‘close’ to the knowledge of the focal firm. Indirect competitors and noncompetitors, on the other hand, operate in different markets and have dissimilar knowledge bases. In order to absorb this (highly) new and different knowledge, one needs to have a high absorptive capacity which young firms seem to have only at a low level.

Further, this paper shows that the contradictory findings obtained in prior research on the influence of the number of partners on firm performance, may be explained by the degree of coopetition the focal firm faces in the alliance. This research namely found that, in case of a higher number of partners, firm’s financial performance is highest when there is a high degree of coopetition. This result is possibly explained by the accumulation of a critical mass of resources that more partners can facilitate. This critical mass would be difficult to replicate by a firm outside the alliance going alone, increasing the focal firm’s competitiveness and consequently its financial performance.

These findings provide a possible explanation why some firms are better able to reap the benefits of collaborating for innovation in the form of R&D consortia. However, it should be noted that hypotheses could only be limitedly supported due to reliability issues, such as the small (sub)sample size(s). Nonetheless, it outlined the tensions present in multi-coopetition based on previous research. The role of tension is important for understanding the relationship between the innovation-related coopetition and performance (Park et al. 2014). “The tension in coopetition stems from the paradoxical factors, such as value creation versus value appropriation, knowledge sharing versus knowledge protection, and so on” (Park et al. 2014, p. 219). The process of knowledge sharing is more complex in constellations as partners do not directly reciprocate with one another and challenges managers even more to realize a high financial performance from engaging in innovation-related coopetition. As multi-coopetition has become increasingly popular, it is important that more research is undertaken to systematically address the issue of tension.

(33)

32

7. Implications

This research has implications for researchers and policy makers. It emphasizes that multi-partner alliances should be treated as a distinct type of alliance when studying coopetition. Researchers should acknowledge this and make a clear distinction when studying the field of innovation-related coopetition. As initiated by Ingram and Yue (2008), this study underscores that collaborating with competitors should not be seen as a taboo. Policy makers should acknowledge coopetition is not about violating antitrust issues, but rather see it as a legitimate business practice. When designing policies, this acknowledgement contributes to the creation of a more viable business community. The results show that the degree of coopetition is possibly a factor that explains why some firms are better able to reap the benefits from participating in a MR&D alliances. Researchers, therefore, should incorporate this factor when they study a focal firm’s performance in MR&D alliances.

Faced with constellations, managers would like to know the performance implications of partnering with a given group of firms. In other words, managers should not only decide whether they should join a constellation, but also what kind of constellation to join. Data from this thesis shows that when a firm is to decide to join an alliance with more than three partners, it should ensure that there is a high degree of coopetition, i.e. the alliance consists mostly (if not fully) of direct competitors. This study found that firms in these alliances benefit most in terms of return on assets. When a firm is about to join an alliance with a lower amount of partners, this study found that it does not matter for the firm’s financial performance whether its partners are direct competitors, indirect competitors or noncompetitors.

(34)

33

8. Limitations and future research directions

As with any research, this one has certain limitations. First, this study used accounting measures to determine firm performance. Although accounting data provide relevant information (Hirschey & Wichern, 1984) and is often the focus of managers who allocate scarce resources (Huselid, 1995), it reflects a historical perspective. Market-based measures include economic profits that are forward-looking and reflect the perception of the market of both the current and potential profitability (Huselid, 1995). This seems a more realistic way of measuring performance of innovating companies, as it takes time for an investment in a new product to yield profits. For this, future studies could for instance use the market value measure Tobin’s q. As limited financial data was available for this study, the widely accepted measures profit margin and return on assets were used.

Second, the sample size for this study was rather small (N = 75). Future studies should use a larger sample size to enhance the reliability of the findings.

Third, this study includes only firms from high- and medium-high-tech industries and may therefore exclude certain aspects of coopetition that are common in other industries. Though, multi-partner alliances and innovation-related coopetition are less common in these industries (Ritala & Hurmelinna-Laukkanen, 2009) and is therefore seen as a minor limitation. Nevertheless, testing the hypotheses in lower-tech industries (e.g. service industries), would help generalize the findings. Fourth, this study did not take into account whether the focal firm and its partner operated in the same countries. For instance, when a firm in an alliance would face high degree of coopetition, but the firms operate in different countries, i.e. they have low market commonality, they may perceive each other less as a competitive threat. Therefore, future research should address this.

Fifth, future researchers could study whether, in the case of a high degree of coopetition, there is a threshold after which the change in return on assets decreases when the number of partners increases. The maximum amount of partners in this study was seven and no threshold (i.e. an inverted U-shape relationship) was discovered. Future research should study alliances with a higher number of partners to ensure there is no threshold as hypothesized in this paper. Furthermore, the focus of this study was multi-partner alliances and therefore only these alliances were in the sample. It would be interesting future research studies whether financial performance differ between firms that join a dyadic or multi-partner alliances when engaging in innovation-related coopetition.

Sixth, value creation/value appropriation tensions (Ritala & Hurmelinna-Laukkanen, 2009). For instance, when collaborating with direct competitors, there are high value creation/appropriation tensions because of the nature of knowledge exchanged and developed (Yami & Nemeh, 2013). However, when there are strong appropriability mechanisms in place, knowledge is likely to be exchanged more freely leading to a higher performance. Large-scale empirical research on the effect of appropriability regimes in multi-partner alliances with different degrees of coopetition should reveal more information.

Sleeping with your enemies?