Dual Masters Award in Advanced International Business
Management (MSc, MSc)
08.01.2018
The Effects of External Networks and Firm-level
Entrepreneurship on the Innovative Performance:
Empirical Evidence from the Pharmaceutical
Industry
Corinna Tichler
S3331121 (Groningen student number)
150578086 (Newcastle student number)
Dissertation supervisors: Dr. Melih Astarlioglu, Dr. Harsh Jha
ABSTRACT
ACKNOWLEDGEMENTS
TABLE OF CONTENTS
1. INTRODUCTION ... 5
2. THEORY AND HYPOTHESES ... 7
2.1 Innovative Performance ... 7
2.2 External Networks ... 9
2.2.1 Forms of Economic Organisation ... 9
2.2.2 External Networks and Innovative Performance ... 11
2.3 Firm-level Entrepreneurship ... 16
2.3.1 Concepts of Firm-level Entrepreneurship ... 16
2.3.2 Direct and Moderating Effects of Entrepreneurial Orientation ... 18
2.4 Conceptual Model ... 25
3. METHODS ... 26
3.1 Empirical Context and Data Collection ... 26
3.2 Measures... 28
3.2.1 Dependent Variable: Weighted Patent Count ... 28
3.2.2 Independent Variable: Maintained Alliances ... 31
3.2.3 Independent Variable and Moderator: Entrepreneurial Orientation ... 32
3.2.4 Control Variables ... 33
3.3 Analytical Method ... 36
4. RESULTS ... 37
4.1 Descriptive Statistics ... 37
4.2 Regression Results and Marginal Effects Plot ... 39
5. DISCUSSION ... 44
5.1 Theoretical Implications ... 45
5.2 Managerial Implications ... 48
5.3 Limitations and Future Research ... 50
REFERENCE LIST ... 54
1. INTRODUCTION
Engaging in innovation and subsequently creating competitive advantage is salient for organisations to survive (Drucker, 1977). 'Innovation has become the industrial religion of the late 20th century, [C however], economists are still struggling to understand this most mysterious part of the wealth-creation process', (Valéry, 1999). Indeed, most organisations highlight the importance of innovation in strategy documents and mission statements. The present study aims to investigate what drives firm-level innovative performance. Innovation is not happening by pure chance but is a process that needs to be managed and planned cautiously (Stokes, Wilson and Mador, 2010).
arguments of the hypothesised relationships, it is elaborated on innovative performance, external networks and firm-level entrepreneurship. It is proposed, that a higher engagement in external networks such as strategic alliances and joint ventures increases the innovative performance. This association is explained by improved organisational learning through the creation of social capital, complementary knowledge, limited probability of competency traps and trough risk reduction (Levinthal and March, 1993; Wu and Fang, 2010; Chien, Lin and Lien, 2015). Furthermore, direct and moderating effects of firm-level entrepreneurship are proposed. Firm-level entrepreneurship is represented through the concept of Entrepreneurial Orientation (EO) and characterises the disposition of an organisation to favour bold, proactive and innovative moves (Kantur, 2016). It is reasoned that EO is a crucial and firm-specific resource to sense novel opportunities, which subsequently results in an improved innovative performance (Baker and Sinkula, 2009; Tang, Chen and Jin, 2014). Moreover, it is argued that EO enhances a firm’s ability to take advantage of organisational learning as it fosters the willingness to activate knowledge-based resources derived from networks (Wiklund and Shepherd, 2003).
With these specified theories and empirical evidence from the pharmaceutical industry, it is aimed to answer the following research question: How does the engagement in
2. THEORY AND HYPOTHESES 2.1 Innovative Performance
The firm-level definition of innovative performance has long been discussed in the management and economics literature (Hagedoorn and Cloodt, 2003). While some researchers referred to innovative inputs, most studies linked companies' innovative performance to novel outputs (Shan, Walker and Kogut, 1994; Ahuja and Katila, 2001). Recent empirical work of Lee, Park and Bae (2017) specified innovative performance as the ability of an organisation to successfully market innovative products.In a similar vein, scholars defined innovative performance in terms of new products market share, rates of new process/product introductions and patent activity (Vanhaverbeke, Duysters and Beerkens, 2001; Hagedoorn and Cloodt, 2003; Cockburn, MacGarvie and Müller, 2010; Brem, Nylund and Schuster, 2014). Thus, for the current study, the innovative performance of a company represents the capability to create product and process innovation (Çömlek et al., 2012; Han and Li, 2015; Wu et al., 2016).
innovate to survive. This view is further supported by Porter (1990), who points out that innovation leads to competitive advantages for organisations. Similarly, current studies highlighted the importance of being innovative to gain global competitiveness in the prevailing knowledge economy (Çömlek et al., 2012; Han and Li, 2015; Wu et al., 2016).
While engaging in innovation is most often associated with competitive advantages and firm growth (Dess and Lumpkin, 2005; Chien, Lin and Lien, 2015), some scholars also point out negative perspectives. The formation of new markets and technologies also constrain organisations as it often requires internal restructuring processes (McFadzean, O’Loughlin and Shaw, 2005). Innovation processes demand substantial resource commitment in a risky and uncertain context (Evanschitzky et al., 2012). According to Åstebro (2003), inventions that are developed on a corporate level, have the probability of 27% to reach the market, while the expected average return is about 9.7%. However, despite the high failure rate, established organisations are likely to better absorb potential losses due to their resource availability (Rosenbusch, Brinckmann and Bausch, 2011).
inconsistencies in conceptualising innovation that hamper advancements of the research field. They contribute to the contemporary literature and develop a framework that distinguishes between radical, really-new and incremental innovations by assessing the technological and market discontinuities on both macro and micro levels. Traditionally, the innovation management process during the 1950s/60s was viewed as a serial process that emphasised research and development (R&D). In the technology push model, the market was seen as a passive element that received the innovations (Di Stefano, Gambardella and Verona, 2012). During the 1970s, scholars highlighted the reverse linear model, the market pull approach. Advocates of this model assumed the market demand to be a crucial factor of technological change (Peters et
al., 2012). However, the sequential models have limited explanatory power, as they try
to describe where innovation originates from, rather than how innovation is created (Galbraith, 1982). Subsequently, these models were further developed, leading to the current network and open innovation model (Chesbrough, 2003). The present models imply strategies, that are not self-contained but collaborative (Grandori, Giordani and Hayton, 2011). Scholars and practitioners now realised that interaction between different market actors is crucial for the innovation management process, which provides the context of the current study (Pittaway et al., 2004; Trott 2017).
2.2 External Networks
2.2.1 Forms of Economic Organisation
(1975) and emphasises that networks represent a distinctive form of exchange. While Williamson (1975) places collaborative arrangements amongst firms on the continuum view between the market and hierarchy pole, Powell (1990) disagrees and argues that this view is too static and wrong from a historical perspective. He concludes that network forms are especially suitable for activities that are knowledge-intense and intangible. Williamson (1991, 1996) reacted to this debate by conceding that collaborative arrangements have to be added as a third form of economic organisation. Since then, collaborative forms of organisation gained much popularity amongst researchers and practitioners (Larson, 1992; Larson and Starr, 1993; Cravens, Shipp and Cravens, 1994; Iyer, 2002; Chen, King and Wen, 2015). This development is well observed by Pettigrew and Fenton (2000), who stress that nowadays companies add value through their maintained relationships and their subsequent social embeddedness.
forms such as manufacturing agreements, collaborative marketing or joint R&D (Powell, Koput and Smith-Doerr, 1996).
2.2.2 External Networks and Innovative Performance
Scholars most often refer to the sharing of costs, risks, complementary skills, economies of scale and the access to untapped markets as the rationales behind engaging in alliances and building a network (Powell, Koput and Smith-Doerr, 1996; Ahuja, 2000). Moreover, the potential to accelerate the creation process of new products and to improve the innovative performance are important rationales (Teece, 1992; Sakakibara, 1997; Soetanto and Jack, 2011). Indeed, there is consensus in the literature, that interorganisational linkages foster the innovative performance (Bekkers, Duysters and Verspagen, 2002). However, comprehensive empirical evidence of this association is limited (Faems, Looy and Debackere, 2005; Pérez-Luño et al., 2011). Analysing the benefits of collaboration through the lens of the organisational learning
theory offers valuable insights. This perspective is especially beneficial to set out the
link between interorganisational networks and innovative performance (Iyer, 2002; Baker, Grinstein and Harmancioglu, 2016).
The concept of organisational learning was developed in the 1970s and represents the process, which enhances the firm's understanding and knowledge towards its environment (Daft and Weick, 1984; Çömlek et al., 2012). Kragulj (2016) defines organisational learning as the adaption of the company to its environment through experience and knowledge accumulation. The subsequent learning outcomes are ingrained in the corporate structure and culture (Vakola and Rezgui, 2000). Many researchers studied the concept of organisational learning, as it is identified as a major determinant of firm performance (Huber, 1991). While some researchers claim that the link between the firm's capacity to learn and innovative performance remains an understudied area, it is agreed upon the positive effect of organisational learning on innovation (Hult, Ferrell and Hurley, 2002; Çömlek et al., 2012; Chien, Lin and Lien, 2015). For example, a qualitative study by Yeung, Lai and Yee (2007) theorises, that organisational learning is a significant intangible resource, which drives innovation. The basic assumption is, that innovation requires to draw novel linkages between existing knowledge or to create new information (Vakola and Rezgui, 2000; Alegre et
are required to find a strategy how to enhance their organisational learning to be innovative (Nonaka and Krogh, 2009; Argote, 2013).
First, social capital inherited in external networks positively influences organisational
learning (Wu, Ay and Lien, 2009; Chien, Lin and Lien, 2015). The social capital construct is applied in diverse social science research fields, however, there is no consensus on how to define social capital (McElroy, Jorna and van Engelen, 2006). Adler and Kwon (2002) systematically reviewed the various definitions of the concept and conclude that 'social capital is the goodwill available to individuals or groups', (p. 23). Researchers agree that social capital consists of three dimensions: the structural, relational and cognitive dimension (Nahapiet and Ghoshal, 1998; Chien, Lin and Lien, 2015). The structural dimension characterises the patterns of the relations such as the network ties, density, centrality and stability (Atuahene-Gima and Murray, 2007). Strategic alliances are likely to be organised non-hierarchically and to be unstable, as companies regularly join and leave the network (Inkpen and Tsang, 2005). The
relational dimension of social capital describes the interpersonal consequences of
interaction such as trust and friendship (Merlo et al., 2006). Trust in strategic alliances derives from repeated trustworthy behaviour (Inkpen and Tsang, 2005). The final dimension of social capital is cognitive and relates to a shared understanding, interpretation and vision (Nahapiet and Ghoshal, 1998). In strategic alliances, the partnering organisations have their own values and corporate culture. Thus, negotiating is a crucial factor to set up clear goal commitments (Inkpen and Tsang, 2005). Having established favourable conditions across the three dimensions, social capital is likely to facilitate knowledge transfer. According to Mouzas, Henneberg and Naudé (2008) interorganisational collaborations have the potential to produce social capital that encourages actors to create new knowledge and shared cognitive patterns that in turn foster innovative performance. By drawing the linkage between social capital and innovation it is especially important to consider the relational and cognitive dimension, as the quality of the relationship determines the innovation outcome. The relationship enables to transfer complex and tacit knowledge, which would be difficult to communicate via administrative, formal rules (Ellinger et al., 2011; Pérez-Luño et
al., 2011). A longitudinal study by Subramaniam and Youndt (2005) confirms, that
legitimacy for radical ideas, which makes it easier to realise them. Thus, applying social capital theory on interorganisational networks offers valuable insights, how activities which are embedded in social contexts are linked with organisational learning and innovative performance (Subramaniam and Youndt, 2005; Hvižďáková and Urbančíková, 2014).
Second, external collaboration leads to innovation-related benefits as organisational
learning is enhanced through the combination of internal and external knowledge (Ghosh, 2004). While internal resources also represent critical assets, they are often not sufficient and need to be complemented by external relationships (Chien, Lin and Lien, 2015). As the business environment and technologies have become increasingly complex, it is not possible for individual companies to develop competence in all the knowledge domains, which are essential for innovation. Moreover, the purchase of missing knowledge bases is often not feasible from a transactional point of view (Ahuja, 2000). Thus, engaging in external networks provides the opportunity to link approaches and ideas that had formerly been unrelated (Ahuja and Lampert, 2001). Through collaboration with e.g. suppliers, customers or competitors, opportunities for knowledge exploitation and transfer are provided (Yli-Renko, Autio and Sapienza, 2001). Partnering with these parties enhances learning, as different viewpoints can be explored. Subsequently, companies have a higher probability to sense opportunities and innovate accordingly (Alegre et al., 2012). A study by Kim and Atuahene-Gima (2010) on new product development confirms that the combination of both, knowledge related to existing firm experience and knowledge acquisition distant from current skills, is necessary for innovation. Therewith, information sharing and the transfer of knowledge through external networks increases the likelihood of opportunity recognition and innovative performance (Subramaniam and Youndt, 2005).
Third, engaging in external relationships enhances organisational learning by reducing
(March, 1991). Subsequently, the distinctive capabilities become deeply rooted in the organisational values, mental methods turn inflexible and thus, core capabilities are institutionalised. These rigidities make organisations vulnerable towards change in the external business environment and inhibit innovation (Leonard-Barton, 1992). An approach to decrease the likelihood of competency and familiarity traps is engaging in external alliances and building a network. A condition for this to happen is that companies do not only engage in alliances to confirm existing practices but to also augment current approaches (Baker, Grinstein and Harmancioglu, 2016). The collaborating organisations may have alternative interpretations of the external environment that help to sense opportunities (Atuahene-Gima and Murray, 2007). Most important is, that the approaches are new to the company. Ideas that are additionally new to the market are advantageous, but not a precondition (Ahuja and Lampert, 2001). Collaboration benefits the company not only by providing the chance to gather novel information, but also to alter rigid problem-solving approaches (Amabile 1988). Novel technologies with a different reasoning help companies to incorporate new perspectives and thus are better prepared to deal with future challenges (Ahuja and Lampert, 2001). Consequently, engaging in alliances and exploring novel approaches helps to prevent rigidity and fosters innovation.
Fourth, the participation in external networks benefits innovative outcomes trough risk
However, engaging in strategic alliances is risky and setting the framework to derive innovative benefits from the collaboration poses challenges on the participating organisations (Larsson et al., 1998; Gil and González de la Fé, 1999). A precondition for an organisation to gain advantages from external linkages is to develop networking capabilities and absorptive capacity (Cohen and Levinthal, 1990; Parkhe, 1993; Bullinger, Auernhammer and Gomeringer, 2004). Organisations need to have the ability to absorb, assimilate and utilise novel information (Lane and Lubatkin, 1998). If companies have a weak technological basis, the incorporation of external, distant knowledge might be problematic (Bekkers, Duysters and Verspagen, 2002). Ahuja and Lampert (2001) point out to the destructive consequences of disproportionate exploration of new information and stress, that companies must balance exploitation and exploration to be successful innovators. Further networking capabilities relate to the coordination of the relationship such as maintaining a stable interaction and avoiding unequal power relations (Bullinger, Auernhammer and Gomeringer, 2004). Many alliances perform poorly, collapse and are ineffective because of opportunistic behaviour of the alliance partners (Muthusamy and White, 2005). While reciprocity and mutual collaboration are desired, strategic alliances are self-governed, which complicates the interactions. If the participating organisations put emphasis on their individual gain rather on the joint achievement, malperformance is likely to occur (Parkhe, 1993). However, as Teece (1992) argues, finding the right degree of collaboration can prove challenging. On the one hand, enough skills have to be shared to achieve joint innovative goals, while on the other hand, the transfer of core competencies has to be prevented (Hamel, Doz and Prahalad, 1989). Moreover, increasing the transparency may result in opportunistic behaviour of the alliance partner and in asymmetric learning. If subsequently, all partnering companies reduce their transparency, no learning and no innovation will occur (Larsson et al., 1998). Thus, most alliance problems and conflicts refer to the inability of joint learning (Muthusamy and White, 2005).
However, empirical evidence shows, that trust is an important part of alliances and that it mitigates risk (Boersma, Buckley and Ghauri, 2003; Jong and Woolthuis, 2009). An organisation is seen as trustworthy if it meets the expectations of the alliance partners concerning the ability to contribute and the adherence to values (Muthusamy and White, 2005). A longitudinal case study of Doz (1996) revealed that organisations of successful alliances constantly re-evaluated their relationship, which lead to increased learning, trust and commitment. However, trust and contract are not contradictory. An empirical study by Woolthuis, Hillebrand and Nooteboom (2005) showed, that they can be both, substitutes and complements. Thus, contracts and trust are two viable mechanisms to mitigate risk and derive innovative benefits from strategic alliances as collective organisational learning is empowered.
From the above discussion and in line with previous studies (Deeds and Hill, 1996; Ahuja, 2000), it is proposed, that the number of alliances positively impacts the innovative performance:
H1: The engagement in external networks (strategic alliances) positively impacts the innovative performance.
2.3 Firm-level Entrepreneurship
2.3.1 Concepts of Firm-level Entrepreneurship
the departure from previous behaviour (Savage and Black, 1995; Reynolds et al., 2005).
The current study focuses on firm-level entrepreneurship. Organisations that are entrepreneurial are characterised by responsiveness, flexibility and an efficient management (Naman and Slevin, 1993). Many researchers studied the direct relationship between firm-level entrepreneurship and firm performance indicators such as firm profitability and growth (Zahra, 1991; Zahra and Covin, 1995; Gupta and Gupta, 2015). While it is generally agreed upon a positive association, researchers also point out to contextual factors that influence this relationship (Lyon, Lumpkin and Dess, 2000; Kantur, 2016). However, scholars adopted diverse labels for studying firm-level entrepreneurship. Concepts such as intrapreneurship (Kuratko, Montagno and Hornsby, 1990; Antoncic and Hisrich, 2001), corporate entrepreneurship (Guth and Ginsberg, 1990; Zahra, 1991; Zahra and Covin, 1995), entrepreneurial posture (Covin and Slevin, 1991), entrepreneurial orientation (Miller and Friesen, 1982) or strategic
entrepreneurship (Ireland, Hitt and Sirmon, 2003) have been used to describe
firm-level entrepreneurship. While these research streams developed separately and capture different aspects, they are still highly interrelated (Kantur, 2016). Consequently, knowledge about firm-level entrepreneurship remains rather fragmented and confusion is created, which hampers the development of the field (Ireland, Covin and Kuratko, 2009; Kantur and İşeri‐Say, 2013). Amongst the mentioned concepts, entrepreneurial orientation (Miller and Friesen, 1982; Covin and Slevin, 1989; Miles, Arnold and Thompson, 1993; Lumpkin and Dess, 1996; Matsuno, Mentzer and Özsomer, 2002) and corporate entrepreneurship (Burgelman, 1983; Guth and Ginsberg, 1990; Zahra, 1991; Zahra and Covin, 1995; Corbett et al., 2013) have been most frequently applied.
venturing and strategic renewal as the two dimensions that relate to corporate entrepreneurship. Corporate venturing means the expanding of the business (Hornsby
et al., 2013), while strategic renewal refers to the reorganisation of processes, activities
or business units (Gómez‐Haro, Aragón‐Correa and Cordón‐Pozo, 2011). Contrastingly, the dimensions of EO most commonly used in the literature are risk-taking, innovativeness, and proactiveness (Miller, 1983). These dimensions make up the EO and represent the strategic posture of the company (Wales, Monsen and Mckelvie, 2011; Gupta and Gupta, 2015). Zahra (1991) views EO as an inclination towards corporate entrepreneurship. Consequently, EO can be seen as a strategic direction that is essential for achieving corporate entrepreneurship (Dess and Lumpkin, 2005; Todorovic, Todorovic and Ma, 2015), while the latter one is more about actually engaging in entrepreneurial activities (Kantur, 2016).
2.3.2 Direct and Moderating Effects of Entrepreneurial Orientation
the willingness to introduce new products, processes or services and to facilitate change. Lastly, proactiveness refers to a forward-looking disposition to detect opportunities (Danny Miller, 1983; Covin and Slevin, 1989; Dess and Lumpkin, 2005). Contrastingly, the second main conceptualisation is based on the multidimensional measure of Lumpkin and Dess (1996). They complemented the construct by adding the variables autonomy and competitive aggressiveness. Autonomy refers to the willingness to drive concepts and visions, while competitive aggressiveness represents the strong inclination to outperform rivals (Lumpkin and Dess, 1996; Dess and Lumpkin, 2005).
However, there is an ongoing debate about the unidimensional and multidimensional approach of EO (Covin and Lumpkin, 2011). Researchers most often assessed EO as an aggregate index in their analysis (Rauch et al., 2009; Miller, 2011). However, this approach disregards the distinct influences of the dimensions (Hughes and Morgan, 2007). The disaggregation of the index may also give insights to better understand the entrepreneurial activities (Miller, 2011). Further, Lumpkin and Dess (1996) critique the view of adopting an aggregate index and posit, that researchers should acknowledge the multidimensional characteristic of EO. Moreover, they point out to contingency factors that influence the association between EO and performance indicators. Lumpkin and Dess (1996) conclude, that certain contexts require the presence of specific combinations of the EO dimensions. Thus, from a multidimensional perspective, what constitutes an EO depends on the external environment and on organisational factors (Covin and Wales, 2012). Indeed, empirical studies by Morgan and Strong (2003) and Hughes and Morgan (2007) revealed, that EO dimensions are not equally important and may even harm desired outcomes. However, other studies reconfirmed the validity of the composite index and highlighted, that firms can only be entrepreneurial if all three dimensions are considered simultaneously, regardless of the context (Basso, Fayolle and Bouchard, 2009). In a reflection about the EO construct, Covin and Lumpkin (2011) comment the ongoing debate and conclude, that both constructs have significant potential to contribute to the scholarly field of EO. Likewise, Miller (2011) theorises, that it may be beneficial for certain research contexts to consider both, the composite EO construct and the individual dimensions.
The most common research stream of EO relates to the direct effects of EO on firm performance and has been thoroughly investigated empirically as well as theoretically (Zahra, 1993; Lumpkin and Dess, 1996; Lyon, Lumpkin and Dess, 2000; Shan, Song and Ju, 2016). A meta-analysis by Rauch et al. (2009) revealed positive effects of EO on performance. While the scholarly literature sees EO as a crucial condition for innovation, not much research has investigated the EO effects on innovation outcomes (Danny Miller, 1983; Pérez-Luño, Wiklund and Cabrera, 2011; Thoumrungroje and Racela, 2013; Tang, Chen and Jin, 2014). In the past, firm-level entrepreneurship and innovation have often been examined independently from each other, therefore the literature about the entrepreneurship-innovation association remains scarce (McFadzean, O’Loughlin and Shaw, 2005). In the following reasoning about the association between EO and innovative outcomes, the three commonly used dimensions of EO are adopted (Hughes and Morgan, 2007; Basso, Fayolle and Bouchard, 2009; Miller, 2011).
Innovativeness reflects the propensity to favour creativity, novel solutions and
technological leadership (Dess and Lumpkin, 2005). Further, it represents the willingness to invest in R&D and to create new products, processes or services (Thoumrungroje and Racela, 2013). Innovativeness means also to actively support the discovery of visionary ideas so that a sustainable EO can be maintained (Morgan et
al., 2015). Emphasising innovativeness moreover increases the likelihood to enter into
(Pérez-Luño, Wiklund and Cabrera, 2011). Thus, focusing to be ahead of the competition and to sense trends, is likely to benefit the innovative performance (Morgan et al., 2015; Liu et al., 2017). The last EO dimension, risk-taking, reflects the inclination of a company to seize opportunities and commit resources in uncertain environments (Miller, 2011; Morgan et al., 2015). If sufficient human and financial resources are provided, the innovation process is accelerated through the resources themselves and through increased motivation and tolerance of risk (Shan, Song and Ju, 2016). The propensity to embrace risk further implies to engage in actions before fully understanding the surrounding conditions and to accept the possibility of failing (Covin and Slevin, 1989; Hughes and Morgan, 2007; Thoumrungroje and Racela, 2013). Thus, the willingness to accept risk increases the probability to exploit novel, emerging trends (Tang, Chen and Jin, 2014). Consequently, the risk-taking dimension of EO fosters the capability of organisations to engage in innovation (Noor and Aljanabi, 2016; Wang and Juan, 2016).
It becomes evident from the above theorising, that EO fosters the innovative performance of organisations. However, there are also pitfalls related to EO. For example, R&D expenses that reflect innovativeness may lead to wasted resources, if the aspired aims couldn’t be realised (Frishammar and Hörte, 2007). Further, innovative endeavours tend to be more sophisticated and complex, which can impede the implementation of novel products, processes or services (Shan, Song and Ju, 2016). Moreover, especially in a highly competitive environment, rivals may introduce similar innovations or more profitable solutions that in turn constrain the organisation. Also, a strong inclination towards proactiveness causes the potential threat, that innovations are introduced before they are fully developed (Dess and Lumpkin, 2005). Moreover, the propensity to favour high levels of risk can harm the innovative performance (Frishammar and Hörte, 2007). For risk-taking not to be too costly, it should be cautiously managed and planned (Dess and Lumpkin, 2005; Hughes and Morgan, 2007).
2009; Renko, Carsrud and Brännback, 2009; Tang, Chen and Jin, 2014; Yu, Nguyen and Chen, 2016; Mütterlein and Kunz, 2017).
It can therefore be hypothesised:
H2: Firm-level entrepreneurship (EO) is positively associated with innovative performance.
Next, moderating effects of EO are explored. While Faems, Looy and Debackere (2005) claim that the empirical evidence of the association between external collaboration and innovative outcomes is not yet comprehensive enough, the literature supports the notion of a positive association. Some researchers moreover examined the effects of EO on innovative outcomes, yet it remains an understudied area (McFadzean, O’Loughlin and Shaw, 2005; Pérez-Luño, Wiklund and Cabrera, 2011; Thoumrungroje and Racela, 2013). Moreover, the joint innovative effects of external collaboration and EO have not received much attention and provide fruitful study area. While many studies looked at e.g. environmental variables as moderators in firm-level entrepreneurship research, not much work has been done on the relation of EO and the innovative potential of interorganisational networks (Simsek, Lubatkin and Floyd, 2003; Wang, 2008). Studies that explore firm attributes and under which organisations benefit more or less from strategic alliances in terms of innovative outcomes remain scarce (Lee, Lee and Pennings, 2001; Stam and Elfring, 2008).
Covin and Slevin, 2009; Altinay et al., 2016). Consequently, organisations with an increasing EO are expected to better capitalise collaboration related learning benefits (Baker, Grinstein and Harmancioglu, 2016). It was argued before, that external collaboration leads to a higher innovative performance as organisational learning is enhanced through social capital, complementary knowledge, limited probability of competency traps and trough risk reduction (Levinthal and March, 1993; Wu and Fang, 2010; Chien, Lin and Lien, 2015). While it is true, that external networks provide the opportunity for diverse learning, organisations are often constrained to fully implement and take advantage of the novel knowledge (Burt, 2004; Kreiser, 2011). However, organisations can use their EO to effectively take advantage of external collaboration by maximising their organisational learning and subsequent innovative performance (Kreiser, 2011). EO enables organisations to benefit more from external networks in terms of innovative output, as knowledge-based opportunities that reside in networks are more likely to be exploited (Wales, Parida and Patel, 2013). Higher degrees of EO provide organisations with the opportunity to gather novel perspectives and knowledge from external networks which they didn’t capitalise before (Stam and Elfring, 2008; Kreiser, 2011). EO has the potential to more effectively use the available organisational knowledge base to exploit novel opportunities. Thus, with a higher EO, organisations are more likely to capture trends and to create new combinations from knowledge inherited in external networks (Wales, Parida and Patel, 2013). Further, EO enhances the responsiveness to knowledge that was gathered from external sources (Liao, Welsch and Stoica, 2003). An empirical study by Wang (2008) supports the notion, that EO positively impacts the capabilities to flexibly react to new knowledge. Thus, EO leads to distinctive competencies and visions, that help to achieve innovation through the available knowledge (Hsu et al., 2014). An increasing EO moreover fosters strategic learning as inertia and learning traps are likely to be reduced through the exposure to diversity and novel assumptions (Kreiser, 2011; Sirén et al., 2017). Thus, entrepreneurial organisations are open-minded towards knowledge outside their boundaries and are more likely to develop problem-solving capabilities (Ahuja and Lampert, 2001). Consequently, they interact more efficiently with actors across the firm boundaries to successfully interpret novel knowledge and coordinate innovation (Sirén
et al., 2017). This is further supported by Zhao et al. (2011) who point out, that
Hult and Calantone (2008) theorise, that EO fosters integrative reasoning, which helps to derive innovative outcomes from collaboration. Since the literature suggests that EO enhances a firm’s ability to take advantage of organisational learning, it is expected that EO increases the contribution of external networks to innovative performance (Anderson, Covin and Slevin, 2009; Dada and Fogg, 2016). Thus, it is proposed that an organisation that engages in external collaboration will have an even higher innovative performance, if it has an EO that fosters the willingness to activate the knowledge resources derived from networks (Wiklund and Shepherd, 2003). EO therewith acts as a moderator as it has an influence on the relationship between the other variables (Baron and Kenny, 1986).
It can therefore be hypothesised:
H3: Firm-level entrepreneurship (EO) positively moderates the relationship between external networks (strategic alliances) and innovative performance.
2.4 Conceptual Model
The proposed hypotheses lead to the following conceptual model:
Innovative Performance External Networks Firm-level Entrepreneurship
+
+
+
3. METHODS
3.1 Empirical Context and Data Collection
This study focuses on the pharmaceutical industry to assess the relationships between maintained collaborative arrangements, firm-level entrepreneurship and innovative performance. Thereby a positivist research philosophy was applied as reality is assumed to exist objectively (Bryman and Bell, 2011). Moreover, the hypotheses derived from the extensive literature review imply a deductive approach (Adams et al., 2007). Data were collected from secondary sources, covering a period of 6 years (2006-2011). Relying on secondary data is especially suitable for firm-level studies. First, firm-level studies usually require the responses of the top executives to gather reliable information (Kantur and İşeri‐Say, 2013). However, response rates of the top management are generally low (Wales, Patel and Lumpkin, 2013). Second, gathering retro-perspective data is not without limitations, as respondents may have biased perceptions of past events and be unfamiliar with the issues thematised (Zahra and Covin, 1995).
The pharmaceutical industry was chosen as a research context for several reasons.
First, the pharmaceutical industry is one of the few that relies heavily on patents to
of the pharmaceutical companies are established participants in the market, valuable insights can be drawn from them. Fourth, focusing on a single industry ensures that alliance characteristics are comparable as well as innovative and organisational learning processes (Stuart, 2000; Alegre et al., 2012).
The hypotheses were tested on a longitudinal data set, comprising the leading pharmaceutical companies tradedin the US. Firms were included in the sample under the precondition that they were listed at major US stock exchanges (NYSE, NASDAQ and Other OTC) to guarantee the availability of reliable data (Powell, Koput and Smith-Doerr, 1996). Companies were identified as belonging to the pharmaceutical industry by examining their US Standard Industrial Classification (SIC) code. Classifying companies in terms of their SIC code is an established approach of researchers (Chen, King and Wen, 2015). Following the studies of Walter (2012) and Muthusamy and White (2005), companies were included in the sample if the first three digits of their SIC code consisted of 283. A list of the leading pharmaceutical companies was compiled by ranking them according to their revenue in 2006. Determining the size of companies as per their revenue has been used in many studies (Zeff and Fossum, 1967; Carter, Kale and Grimm, 2000; Ramsden and Kiss-Haypal, 2000; Kiel and Nicholson, 2003; Uyar, 2009). A reference year at the beginning of the observation period was chosen to account for survivor bias (Ahuja, 2000). Industry and revenue information was gathered from the Wharton Research Data Services (WRDS) database. The initial sample consisted of 70 companies, however, the list also comprised subsidiaries. As the analysis was performed at the ultimate parent level, subsidiaries had to be matched to their parent companies. This ownership information was obtained from the 2009/2010 Who Owns Whom Edition, corporate homepages, as well as the Nexis database to track the history of the companies and account for name changes. Subsequently, a sample of 60 companies remained. Nine subsidiaries had to be excluded as their parent companies were not traded in the US market or because they were sold too often to be assigned to a specific parent company. For two companies, reliable data could not be gathered, which led to a resulting sample of 49 companies (see appendix, table 3.1).
2000; Baum, Calabrese and Silverman, 2000; Katila, 2002; Furman and Hayes, 2004; Miller and Le Breton-Miller, 2011; Chen, King and Wen, 2015; Shahzad et al., 2016). A one year lag between EO, collaborative activities and innovative performance was introduced to allow time for the predictor variables to affect the dependent variable (Powell, Koput and Smith-Doerr, 1996; Ahuja, 2000; Shahzad et al., 2016). Subsequently, EO and collaborative data were gathered from 2005-2010, while data on innovative performance was obtained for the years 2006-2011. This period was chosen to allow for a valid construction of the subsequent measures. An overview of the construction of the variables can be found in table 3.2. Core sources were the Securities Data Corporation (SDC) alliance database, the Orbis database for innovative performance, the Who Owns Whom Edition 2009/2010 for ownership information and the Wharton Research Data Services (WRDS) to gather company financials and stock prices.
3.2 Measures
3.2.1 Dependent Variable: Weighted Patent Count
However, the importance of patents differs significantly in terms of their value, quality and technological significance (Narin, Noma and Perry, 1987; Albert et al., 1991). It is posited, that more innovative patents receive more forward citations (Jaffe and de Rassenfosse, 2017). Early studies on patent data using simple patents counts have been critiqued for not considering this heterogeneity (Hall, Jaffe and Trajtenberg, 2005). Studies by Carpenter, Narin and Woolf (1981), Trajtenberg (1990) and Hall, Jaffe and Trajtenberg (2005) show, that forward citations comprise important information that captures the value of patented innovations. Subsequently, most studies investigating into innovative performance have appliedcitation-weighted patent counts (WPC) (Vanhaverbeke, Duysters and Beerkens, 2001; Sampson, 2007). To enhance the validity of the patent measure in the present study, the linear weighting scheme proposed by Trajtenberg (1990) was applied.
To gather the yearly WPC for each company, all divisions and subsidiaries had to be identified (Ahuja, 2000). The construction of an aggregate patent portfolio for each firm is highly important to avoid bias and to capture the firm capabilities correctly. Usually, some patents are assigned to the ultimate parent company and some to various subsidiaries (Sampson, 2007). Moreover, companies do not always allocate patents to the subsidiary in which the innovative performance took place (Mowery, Oxley and Silverman, 1996). Thus, incorporating all business units was crucial. The primary source to obtain information about the parent companies and its subsidiaries and divisions was the Who Owns Whom 2009/2010 Edition. Many studies used the Who Owns Whom Editions to gather reliable ownership data (Hagedoorn and Schakenraad, 1994; Mowery, Oxley and Silverman, 1996; Ahuja, 2000; Katila, 2002; Hall, Jaffe and Trajtenberg, 2005). However, the pharmaceutical industry is a dynamic field, smaller firms often become subsidiaries of the leading companies and are being transferred between them (Campart and Pfister, 2014; Malerba and Orsenigo, 2015). Therefore, it was necessary to track the history of the companies during the observation period to capture name changes and reorganisations (Ahuja, 2000). This information was drawn from corporate homepages, annual reports, U.S. Securities and Exchange Commission archives and the Nexis database. Subsequently, up to 300 subsidiaries/divisions per parent company were manually typed into the Bureau van Dijk Orbis database to gather yearly patent information.
two years until it is granted (Furman and Hayes, 2004). Thus, successful patent applications were included in the present study. Moreover, a granted patent was assigned to the year it was applied for, to link it to earliest point in time when the invention was created. If e.g. a company applied for a patent in 2008, that was granted in 2010, it was considered a 2008 patent. Using the patent application date to identify novel organisational capabilities is an established approach in the literature (Ahuja, 2000; Stuart, 2000; Zucker and Darby, 2001; Sampson, 2007; Wang, Huang and Wu, 2012). Only a small time difference is expected between the discovery of an invention and the filing of the subsequent patent application (Kapoor and Lim, 2007).
Many studies have solely considered patents granted by the US patent office (USPTO), even if their sample included multinational enterprises or companies that were incorporated in countries other than the US (Zucker and Darby, 2001; Katila, 2002; Sampson, 2007; Adegbesan and Higgins, 2011). They argue, that the US is one of the most relevant markets to gain intellectual property rights and therewith serves as a proxy for the overall global innovative performance (Vanhaverbeke, Duysters and Beerkens, 2001). Moreover, it is claimed that nations differ in terms of e.g. application standards and their propensity to patent, which could lead to biases (Soete and Wyatt, 1983; Cohen et al., 2002). However, the present study incorporated patents from six patent offices: USPTO, European Patent Office (EPO), World Intellectual Property Organization (WIPO), United Kingdom (GB), Germany (DE) and Japan (JPO)1. Due to
the global nature of the pharmaceutical industry, collaboration patterns and R&D activities take place across countries and show a high degree of internationalisation (Hu et al., 2015). Therefore, the patent analysis should not be restricted to the US market. Several studies support this point of view. Bekkers, Duysters and Verspagen (2002) include the three major patent offices in their study, which are the USPTO, EPO and WIPO. Von Wartburg, Teichert and Rost (2005) posit that DE/EPO/GB/WIPO patent systems are comparable. Moreover, Clarke (2003) acknowledges differences of USPTO/JPO/EPO, however, he concludes that a similar examination quality exists between these offices. Yet, by creating international patent portfolios, it is important to only count an invention once if it was patented at several patent offices (Bekkers, Duysters and Verspagen, 2002). If patents for the same invention are applied for in different countries, a patent family is created (Grupp and Schmoch, 1999). Patents of
a patent family are linked to each other by priority claims, reflected through the same priority number2 (European Patent Office, 2017a). The priority number could be
gathered from the Orbis database and subsequently, patents with the same priority number were excluded so that an invention was only counted once. Consequently, the dependent variable was discrete, a sum of citation-weighted patent counts, taking on non-negative integer values.
3.2.2 Independent Variable: Maintained Alliances
The independent variable was assessed through the yearly number of maintained alliances for each company from 2005-2010. Several researchers examined the impact of the number of maintained alliances on firm outcomes (Shan, Walker and Kogut, 1994; Powell, Koput and Smith-Doerr, 1996; Ahuja, 2000; Vanhaverbeke, Duysters and Beerkens, 2001). In accordance with Vanhaverbeke, Duysters and Beerkens (2001), it is expected that only ongoing linkages influence the innovative performance. Previous studies differentiated between two main organisational forms of strategic alliances: JVs and contract based arrangements (Mowery, Oxley and Silverman, 1996; Ahuja, 2000; Bell and Zaheer, 2007). Information about collaborative arrangements was gathered from the Securities Data Company (SDC) database. The SDC database is one of the most comprehensive alliance sources and has been used in many studies published in top management journals (Sivakumar et al., 2011). Studies with a similar research design as the current one also gathered alliance data from the SDC (Lavie and Rosenkopf, 2006; Sampson, 2007; Jiang, Tao and Santoro, 2010; Sivakumar et
al., 2011). Moreover, Schilling (2009) and Anand and Khanna (2000) confirmed the
reliability of the SDC data.
Studies of interorganisational linkages and alliance portfolios usually investigate at the aggregate firm-level, accumulating all information to the ultimate parent (Bell and Zaheer, 2007; Jiang, Tao and Santoro, 2010; Sivakumar et al., 2011). Thus, the yearly alliances were identified by the ultimate owner. Moreover, it was accounted for name changes during the observation period to have a reliable measure. In the present study, strategic alliances with competitors, suppliers, research institutes and other
2The priority date of a patent is the closest date to the creation of an invention (European Patent
organisations were included (Lee, Lee and Pennings, 2001; Lee, Dedahanov and Rhee, 2015). These linkages could either be dyadic or multi-partner alliances. Further, the database gave insights about the organisational form of the alliance (contractual or JV), as well about the nature of the alliance. R&D, manufacturing and licensing agreements were included in the measure, while mere marketing agreements were excluded to avoid bias. As the current study investigates in technological outcomes, it is unlikely that e.g. a supply arrangement has an impact on the innovative performance (Ahuja, 2000). Next, the SDC data was complemented by archival searches in the Nexis database. Nexis is often used by researchers to obtain archived news (Chan et
al., 1997; Anand and Khanna, 2000; Stuart, 2000; Walter, 2012). In most cases, no
termination date was reported for contractual alliances. Therefore, a lifespan of five years, as proposed by Sampson (2007) and Vanhaverbeke et al. (2001), was assumed. If no dissolution date was indicated for a JV, it was considered as ongoing. According to Ahuja (2000), the absence of a termination report can be interpreted as a continuation of the created equity. Subsequently, alliance data from 2001-2010 was gathered to construct the yearly number of maintained alliances for each company during the observation period.
3.2.3 Independent Variable and Moderator: Entrepreneurial Orientation
As previously mentioned, past studies used two main conceptualisations to measure the EO of companies (George and Marino, 2011); the original composite measure proposed by Miller (1983) and the second main conceptualisation based on the multidimensional measure of Lumpkin and Dess (1996). Empirical studies vary in the way the different dimensions are applied, however, the majority still relates to the three original variables of risk-taking, innovativeness, and proactiveness (Hughes and Morgan, 2007; Basso, Fayolle and Bouchard, 2009; Miller, 2011). What many studies have in common is that they measure the dimensions with subjective data, mostly using a 7-point Likert scale. However, the current study measured EO with secondary data, therewith following the recommendations of Miller (2011). Covin and Lumpkin (2011) posit, that more studies should use objective data to measure EO and emphasise the huge potential of this approach. Further, Wales (2016) highlights the need for measures with secondary data and emphasises the pioneering work of Miller and Le Breton-Miller (2011). Subsequently, EO was measured as a sum of
retained earnings through total assets (Gali et al., 2017). The percentage of reinvested earnings reflects a firm’s forward-looking, proactive attitude to expand the business (Miller and Le Breton-Miller, 2011; Shahzad et al., 2016). A firm`s
propensity/commitment to innovate was assessed by the R&D intensity, the ratio of
R&D expenditures to total assets (Hall, Jaffe and Trajtenberg, 2005; Gali et al., 2017). Both variables were gathered through the Compustat North America financials. The Compustat database is a source often used by researchers (Hausman, Hall and Griliches, 1984; Lavie and Rosenkopf, 2006; Kapoor and Lim, 2007; Sivakumar et al., 2011). The risk-taking dimension, reflecting the idiosyncratic risk of a company, was gathered through the volatility of stock prices (Shahzad et al., 2016). It was calculated by regressing the company returns against the value-weighted market return and taking the root mean square error from it (Hoberg and Prabhala, 2009). Companies that pursue risky strategies are more likely to have a higher fluctuation of stock prices, which can`t be attributed to general fluctuations in the market. Therefore, higher unsystematic risk is expected to represent the inclination to pursue bold moves (Miller and Le Breton-Miller, 2011). Stock prices and stock market indices were gathered from the Center for Research in Security Prices (CRSP) Daily Stock file (Hoberg and Prabhala, 2009; Gali et al., 2017). The CRSP database is an established source in the literature (Koh and Venkatraman, 1991; Anand and Khanna, 2000; Miller and Le Breton-Miller, 2011; Chen, King and Wen, 2015). Subsequently, for each firm-year observation approx. 250 values for each of both variables (firm-specific returns and value-weighted market returns) were downloaded. Next, the standard deviations of the residuals were retained by computing the 294 regressions to derive the risk-taking dimension. The composite index of EO (sum of z-scores of risk-taking, innovativeness, proactiveness) was calculated using Stata.
3.2.4 Control Variables
It is conventional for studies on innovative performance to control for industry, age and size effects (McEvily and Zaheer, 1999; Yli-Renko, Autio and Sapienza, 2001; Wu et
al., 2008; Vissa and Chacar, 2009; Goodale et al., 2011; Lee, Dedahanov and Rhee,
controlled for these industry effects (Ahuja, 2000; Vissa and Chacar, 2009). Following extant literature, it was controlled for firm age, as this could have an influence on the capability to innovate. On the one hand, older companies may suffer from structural inertia, thus being less capable to implement new knowledge (McEvily and Zaheer, 1999). Younger companies may be more flexible to incorporate novel information and processes (Yli-Renko, Autio and Sapienza, 2001). On the other hand, older companies could have skill advantages, thus being better prepared to innovate (Autio, Sapienza and Almeida, 2000). To control for these potential effects, the date of incorporation was gathered from the Bureau van Dijk Orbis database. Consistent with prior research, it was controlled for firm size. Larger organisations typically have more resources and therewith a higher capacity to exploit knowledge (Yli-Renko, Autio and Sapienza, 2001). Further, relative to small ones, they are likely to have more power to influence public institutions to receive advantages (Molina-Morales and Martínez-Fernández, 2009). Following comparable studies, firms size was operationalised by the number of employees (McEvily and Zaheer, 1999; Yli-Renko, Autio and Sapienza, 2001; Muthusamy and White, 2005; Wu et al., 2008; Vissa and Chacar, 2009; Alegre et al., 2012). This information was gathered from the Compustat North America file. Following Ahuja (2000), it was also controlled for R&D expenses. Dedicated research investments may influence the innovative performance. R&D expenses were obtained from the Compustat North America file.
Variable Construction Data Source Most relevant references
Weighted Patent Count (WPC)
- International aggregate patent portfolio of ultimate parent company
- Includes successful patent applications - Linear weighting scheme of forward
citations was applied
- Includes patents of: USPTO, European Patent Office (EPO), World Intellectual Property Organization (WIPO), United Kingdom (GB), Germany (DE), Japan (JPO)
- Patents with the same priority number were excluded so that an invention was only counted once
Maintained Alliances
- Aggregate alliance portfolio of ultimate parent company
- Includes contractual arrangements and equity joint ventures
- Research and development,
manufacturing and licensing agreements were included in the measure, mere marketing agreements excluded - Alliance data from 2001-2010 was
gathered to construct the yearly number of maintained alliances for each company during the observation period 2005-2010
Securities Data Company (SDC) database Archival searches in the Nexis database Ahuja (2000) Mowery, Oxley and Silverman (1996)
Jiang, Tao and Santoro (2010) Sivakumar et al. (2011) Sampson (2007) Vanhaverbeke, Duysters and Beerkens (2001) Entrepreneurial Orientation (EO)
Sum of standardised scores: proactiveness, innovativeness, risk-taking - Miller (1983) Miller (2011) Miller and Le Breton-Miller (2011) Covin and Lumpkin (2011) Proactiveness Ratio of retained earnings to total assets Wharton
Research Data Services (WRDS): Compustat North America financials Miller and Le Breton-Miller (2011) Gali et al. (2017) Shahzad et al. (2016)
Innovativeness Ratio of R&D expenditures to total assets WRDS: Compustat North America financials Miller and Le Breton-Miller (2011) Gali et al. (2017) Hall, Jaffe and Trajtenberg (2005) Risk-taking Idiosyncratic risk of a company: regressing
the company returns against the value-weighted market return and taking the root mean square error from it
WRDS: Center for Research in Security Prices (CRSP) Daily Stock file Miller and Le Breton-Miller (2011) Hoberg and Prabhala (2009) Gali et al. (2017) Shahzad et al. (2016)
Firm Age Reference date: date of incorporation Bureau van Dijk Orbis database
e.g. Yli-Renko, Autio and Sapienza (2001) Firm Size Number of employees WRDS:
Compustat North America file
e.g. Muthusamy and White (2005)
3.3 Analytical Method
4. RESULTS
4.1 Descriptive Statistics
* p < 0.05, ** p < 0.01, *** p < 0.001 Mean Std. Dev. (1) (2) (3) (4) (5) (6) (7) (8) (9) (1) Alliances 13.242 17.078 1 (2) EO 0.023 1.157 -0.0968 1 (3) Innovativeness 0.096 0.093 -0.0178 0.572*** 1 (4) Proactiveness 0.119 0.738 0.248*** 0.121* -0.364*** 1 (5) Risk-taking 0.019 0.0096 -0.336*** 0.502*** 0.0465 -0.468*** 1 (6) WPC 340.812 786.773 0.415*** -0.106 0.00174 0.174** -0.295*** 1 (7) Size 23.531 34.113 0.807*** -0.199** -0.0845 0.279*** -0.422*** 0.572*** 1 (8) RD expenses 1712.52 2551.09 0.812*** -0.0763 0.0762 0.230*** -0.387*** 0.488*** 0.914*** 1 (9) Age 44.714 42.407 0.440*** -0.0879 -0.0258 0.220*** -0.291*** 0.441*** 0.444*** 0.467*** 1
4.2 Regression Results and Marginal Effects Plot
VARIABLES M1 M2 M3 M4 M5 M6 Number Employees 0.00984*** 0.0109*** 0.0067** 0.0112*** 0.0070** 0.0087*** (0.0025) (0.0025) (0.0028) (0.0026) (0.0029) (0.0030) Firm Age -0.0019 -0.0019 -0.0019 -0.0018 -0.0017 (0.0025) (0.0023) (0.0025) (0.0023) (0.0023) Maintained Alliances 0.0152*** 0.0149*** 0.0124** (0.0047) (0.0047) (0.0050) Entrepreneurial Orientation 0.0477 0.0386 0.0530 (0.0527) (0.0530) (0.0538) Maintained Alliances # EO 0.0048 (0.0030) Constant 0.8000*** 0.8044*** 0.8224*** 0.8083*** 0.8253*** 0.8359*** (0.0977) (0.0977) (0.0974) (0.0977) (0.0974) (0.0973) Observations 292 292 292 292 292 292 Number Companies 49 49 49 49 49 49 LR Test Reference Model - - M2 M2 M2 M5 LR Test Statistic 20.54 21.50 8.106 0.804 8.629 2.421 LR Test P-Value 0 0 0.00440 0.370 0.0134 0.120 Standard errors in parentheses
*** p<0.01, ** p<0.05, * p<0.1
Table 4.3: Results negative binomial fixed effects regression
Additional negative binomial fixed effects regressions were run to test variations of the individual EO dimensions. The regression results are displayed in table 4.4. In model 7, the innovativeness dimension was introduced to the control variables, indicating a positive direct effect on WPC at the 0.01 level. M8 added the variable maintained alliances and M9 the interaction term of maintained alliances and innovativeness. The innovativeness dimension was statistically significant throughout the models, however, the interaction term was not significant. Improvement of model fit for M7 and M8 was statistically significant, for M9 it was acceptable at the 0.115 level. Regression models for the proactiveness dimensions are reported in M10-M12, indicating similar results as the innovativeness dimension. Throughout the three models, a statistically significant positive effect of proactiveness on innovative performance was detected. M10 and M11 indicated a significant improvement in model fit, however, M12 with the nonsignificant interaction term did not. Regression models M13-M15 display, that the risk-taking dimension negatively affected WPC at the 0.01 level. Improvement of model fit for M13 and M14 was statistically significant, for M15 with the nonsignificant interaction term it was acceptable (LR Test P-Value 0.140). Thus, innovativeness and proactiveness were directly and positively associated with innovative performance, while risk-taking negatively impacted the WPC. A look at the marginal effect plots of the EO dimensions revealed (see appendix graphs 4.4, 4.5, 4.6), that effects of maintained alliances on WPC got significant after the respective dimensions reached a critical value (-0.5 for innovativeness, 0.4 for proactiveness, and -0.8 for risk-taking).
VARIABLES M7 M8 M9 M10 M11 M12 M13 M14 M15 Number Employees 0.0116*** 0.0074*** 0.0086*** 0.0100*** 0.0066** 0.0077** 0.0088*** 0.0056* 0.0050* (0.0025) (0.0028) (0.0028) (0.0025) (0.0028) (0.0032) (0.0027) (0.0030) (0.0030) Firm Age -0.0023 -0.0021 -0.0023 -0.0021 -0.0021 -0.0022 -0.0033 -0.0033 -0.0030 (0.0025) (0.0023) (0.0023) (0.0024) (0.0023) (0.0023) (0.0025) (0.0024) (0.0024) Maintained Alliances 0.0152*** 0.0148*** 0.0130*** 0.0094 0.0122** 0.0159*** (0.0046) (0.0045) (0.0047) (0.0075) (0.0054) (0.0057) Innovativeness 0.1551*** 0.1571*** 0.2118*** (0.0404) (0.0405) (0.0531)
Maintained Alliances # Innovativeness 0.0081
(0.0051)
Proactiveness 0.2235** 0.1919** 0.2102**
(0.0899) (0.0885) (0.0920)
Maintained Alliances # Proactiveness 0.0040
(0.0065)
Risk-taking -0.2747*** -0.2692*** -0.2702***
(0.0750) (0.0740) (0.0755)
Maintained Alliances # Risk-taking 0.0055
(0.0036)
Constant 0.8261*** 0.8455*** 0.8556*** 0.8350*** 0.8463*** 0.8307*** 0.7941*** 0.8107*** 0.8454***
(0.0977) (0.0973) (0.0974) (0.0974) (0.0971) (0.1009) (0.1004) (0.1001) (0.1012)
Observations 292 292 292 292 292 292 267 267 267
Number Companies 49 49 49 49 49 49 45 45 45
LR Test Reference Model M2 M2 M8 M2 M2 M11 M2 M2 M14
LR Test Statistic 10.88 19.20 2.482 7.071 13.43 0.384 181.9 186.1 2.173
LR Test P-Value 0.00100 0.000100 0.115 0.00780 0.00120 0.535 0 0 0.140
Standard errors in parentheses *** p<0.01, ** p<0.05, * p<0.1
5. DISCUSSION
This study has explored the association between external collaborative arrangements and innovative performance as well as the direct and moderating effects of firm-level entrepreneurship. While the extant literature clearly outlines the innovative benefits of interorganisational linkages, Faems, Looy and Debackere (2005) claim that comprehensive empirical evidence of this association is limited. Moreover, the processes of firm-level entrepreneurship and innovation have mostly been researched independently from each other. Thus, not much research has investigated the EO effects on innovation outcomes (Miller, 1983; McFadzean, O’Loughlin and Shaw, 2005; Pérez-Luño, Wiklund and Cabrera, 2011; Thoumrungroje and Racela, 2013; Tang, Chen and Jin, 2014). To date, the joint effects of external collaboration and EO remain an understudied area. Research that explores if entrepreneurial oriented firms benefit more or less from external collaboration in terms of innovative output are scarce (Lee, Lee and Pennings, 2001; Stam and Elfring, 2008). The current study has attempted to address these gaps. Moreover, the adopted research design improved previous empirical studies methodologically. First, a longitudinal data set over a 6-year period was created. Most previous research measured EO only at one point in time (Rauch
et al., 2009). However, the EO construct is characterised by a stable inclination to act
5.1 Theoretical Implications
First, the analysis revealed that the number of maintained alliances positively
influences the innovative outcome. Thus, the general wisdom in the literature that external linkages facilitate knowledge creation and the generation of innovation was confirmed (Teece, 1992; Hagedoorn and Schakenraad, 1994; Duysters and Hagedoorn, 2000; Soetanto and Jack, 2011). Consequently, the theoretical arguments that drew on the organisational learning perspective to illustrate innovative advantages of collaboration were supported. It was argued that the engagement in external networks enhances organisational learning and innovative performance through social capital, complementary knowledge, limited probability of competency traps and trough risk reduction (Levinthal and March, 1993; Wu and Fang, 2010; Chien, Lin and Lien, 2015). The empirical results not only affirmed the theoretical reasoning but further increased the empirical evidence for this association. Several empirical findings support the view that strategic alliances foster the innovative performance of organisations (Stuart, 2000; Bekkers, Duysters and Verspagen, 2002; Sivakumar et
al., 2011). More specifically, Faems, Looy and Debackere (2005), Ahuja (2000), Deeds
and Hill (1996) and Shan, Walker and Kogut (1994) focused, as well as the current study, on the number of entered alliances as a network characteristic. Consistent with the current study, these scholars detected a positive effect of maintained ties on innovative performance across diverse research contexts and methods.
Second, the results revealed no direct effect of the composite EO on the innovative
found non-significant results for this relationship. Carsrud (2009), Thoumrungroje and Racela (2016) and Xiaoyu Yu, Bang Nguyen (2011) tested the impact of EO on innovative performance with diverse research contexts. In line with the present study, they didn´t find EO to be a significant predictor of the innovative performance. Thus, the empirical evidence for the commonly theorised positive EO-innovative performance relationship remains ambiguous. The non-significance of the second hypothesis might be explained by the chosen empirical context. Previous studies as well as the current one used different industries and organisations at different developmental stages, which could have led to the inconsistent findings. The lack of significance may also be attributed to conceptual issues. While EO is defined as dispositional in nature, representing the proposition to act entrepreneurial, the dependent variable relates only to weighted patents. Therewith, innovative managerial practices or innovative organisational structures that might occur through the EO are neglected. Carsrud (2009) argues, that it is very likely that the type of innovative outcome arising through EO also includes the patenting activity of organisations. However, Carsrud (2009) acknowledges, that these developments are particularly tedious in the biotechnology industry and that they are highly dependent on profound advances in science. Likewise, the EO-innovative performance relationship might be insignificant in the current study as not all innovative outcomes of EO were captured and because R&D processes are extremely complex in the pharmaceutical industry. Another reason for the insignificant relationship may be attributed to the fact, that EO has to be strong enough to impact the innovative performance and to yield significant positive results. Indeed, hints for a non-linear EO-innovative performance association can be found in the literature. Sirén et al. (2017) found empirical evidence for a nonlinear relationship between EO and strategic learning. They argue, that the stimulus of EO has to reach a critical value to impact the dependent variable. Innovative consequences of EO are likely to be mitigated by inertia and rigidity (Huff, Huff and Thomas, 1992). Thus, it can be assumed that low levels of EO might not have a positive significant relationship with innovative performance, while this association might rise exponentially and get significant with stronger EO levels. Apparently, the association between EO and innovative performance is not as straightforward as outlined in the literature.
Third, the analysis of the individual EO dimensions revealed valuable insights. While
