INTRODUCTION
The two innovation orientations of exploration and exploitation each contribute to firm performance in different ways (He & Wong, Reference He and Wong2004; Jansen, van Den Bosch, & Volberda, Reference Jansen, van Den Bosch and Volberda2006; O'Reilly & Tushman, Reference O'Reilly and Tushman2008; Stettner & Lavie, Reference Stettner and Lavie2014). Exploitative innovation, incremental in nature, contributes to firms’ current growth, by refining, improving, and extending their existing technological assets and innovation capabilities, whereas exploratory innovation, largely uncertain and risky, contributes to firms’ future growth by experimenting with novel technology, knowledge, and competence for radical breakthroughs (Benner & Tushman, Reference Benner and Tushman2002, 2003; Rowley, Behrens, & Krackhardt, Reference Rowley, Behrens and Krackhardt2000). To sustain viable growth, firms need to balance both orientations in their innovation strategy. In recent decades, innovation research has recognized complementary effect (+) and competing effect (–) between the two types of innovation orientations on firm performance when they interplay (Bauer & Leker, Reference Bauer and Leker2013; Gupta, Smith, & Shalley, Reference Gupta, Smith and Shalley2006). However, how the complementary and competing effects change and impact firm performance when the two innovation orientations interplay is less understood.
Exploration and exploitation are two types of innovation activities which follow fundamentally different logics (e.g., He & Wong, Reference He and Wong2004; March, 1991), and they require vastly different knowledge, structures, routines, processes, strategies, capabilities, and even cultures in organizations (He & Wong, Reference He and Wong2004; March, 1991; O'Reilly & Tushman, Reference O'Reilly and Tushman2008; Stettner & Lavie, Reference Stettner and Lavie2014). To balance the two types of innovation orientations, organizational scholars propose to use different organizational structures (e.g., domain separation through ambidextrous organization model or temporal separation through punctuated equilibrium model) (e.g., Cao, Gedajlovic, & Zhang, Reference Cao, Gedajlovic and Zhang2009; He & Wong, Reference He and Wong2004; O'Reilly & Tushman, Reference O'Reilly and Tushman2008; Qi, Wang, Zhang, & Zhu, Reference Qi, Wang, Zhang and Zhu2014; Stettner & Lavie, Reference Stettner and Lavie2014). However, when a firm has limited resources for innovation, it has to face an exclusivity in organizational resources between exploratory and exploitative orientations – an increasing in one orientation means a decrease in the other (Gupta et al., Reference Gupta, Smith and Shalley2006). Such resource exclusivity suggests that balancing these two types of innovation activities will inevitably impose organizational challenge and even internal competition, especially for SMEs that often face resource scarcity. In fact, no matter what organizational structure a firm adopts in its innovation balancing strategy, it will incur extra costs to coordinate these two different types of activities (Uotila, Maula, Keil, & Zahra, Reference Uotila, Maula, Keil and Zahra2009). Some scholars (e.g., Gupta et al., Reference Gupta, Smith and Shalley2006) propose that the relationship between firms’ innovation orientations and performance is a non-linear correlation that is contingent upon the interplays between exploratory and exploitative innovations. However, there is a lack of clarity about the latent mechanisms that underlie how such interplays influence firm performance (Haans, Pieters, & He, Reference Haans, Pieters and He2016). Given the importance of balancing the two types of innovation orientations and the influences of their interplays on a firm's innovation balancing strategies, it is crucial to understand the mechanisms underlying the impact of the interplays between exploration and exploitation on firm's innovation balancing strategy and performance (Haans et al., Reference Haans, Pieters and He2016).
Nowadays, with increasing complexity in technology, firms hardly operate in isolation (Li & Tang, Reference Li and Tang2010; Niesten & Stefan, Reference Niesten and Stefan2019). In recent decades, though firms have benefitted from a growing global circulation of information, knowledge, skills, and capital for innovation activities, the significance of industry clusters on firm performance has not diminished (Autio, Nambisan, Thomas, & Wright, Reference Autio, Nambisan, Thomas and Wright2018). In fact, firms in technology-intensive industries have relied even more on collaborations with their cluster partners to deal with increasing technological complexity, rapid changes in product design, and demand conditions. Cluster networks are still one of the critical sources from which firms acquire ideas, information, and knowledge to improve innovation performance. Existing literature has studied the effects of cluster networks on firm's innovation performance from the perspectives of a firm's network structure and its position in the networks (Bell, Reference Bell2005; Fang, Lee, & Schilling, Reference Fang, Lee and Schilling2010; McCann & Mudambi, Reference McCann and Mudambi2005; Ozer & Zhang, Reference Ozer and Zhang2015; Schilling & Phelps, Reference Schilling and Phelps2007). However, how exactly cluster firms benefit from their cluster relationships to improve innovation capability and performance is less understood. Specifically, it is not clear how a firm's different cluster relational attributes at the inter-firm level influence the firm's exploratory and exploitative innovations separately, as well as their interplays, both of which have an impact on the firm's innovation balancing strategy and performance.
To address the above-outlined deficiencies in the literature, this article aims to answer the following research questions: First, when a firm balances exploratory and exploitative innovations, what is the correlation between the firm's relative innovation orientation and performance? And what is the latent mechanism underlying this relationship? Second, when a firm is embedded in clusters, how do the firm's cluster relationships affect the correlation between its relative innovation orientation and performance? We use a firm's relative innovation orientation (RIO) to measure the firm's relative amount of innovation activity between exploration and exploitation. The results of our empirical evidence confirm an inverted U-shaped correlation between a firm's RIO and performance. We also verify that this correlation is moderated by the firm's cluster relationships: the broader a firm's cluster relationships, the more likely the turning point of the inverted U-shaped curve moves toward exploratory innovation orientation for greater growth performance; on the other hand, the stronger a firm's cluster relationships, the more likely the turning point of the inverted U-shaped curve moves toward exploitative innovation orientation for greater growth performance.
This study advances our understanding of the relationship between innovation strategy (i.e., balancing exploratory and exploitative innovations) and firm performance in the context of cluster networks. First, we add a new latent mechanism that underlies an inverted U-shaped relationship between firms’ innovation orientation and performance to Haans et al.'s (2016) model by additively combining the complementary effects (+) and competing effects (–) when exploratory and exploitative innovations interplay. Second, we theorize and test the moderating roles of inter-firm cluster relational attributes on this relationship. Our research highlights the importance of cluster relationships in firm's knowledge acquisition and innovation capability building, and, thus, on firm performance. Our findings carry practical significance for SMEs in managing their innovation strategy and cluster relationships, both of which may lead to improved performance.
In the following sections, we will review the relevant literature and develop the hypotheses, outline the methods used to test the hypotheses, present the results, and discuss the implications of our findings. Figure 1 summarizes the conceptual relationships between the theoretical constructs used in this study.
THEORETICAL BACKGROUND AND HYPOTHESES DEVELOPMENT
Innovation Orientation Toward Exploration vs Exploitation and Firm Performance
Since the concepts of exploration vs exploitation were introduced in organizational learning (March, 1991), many scholars have examined their applications in other management fields, such as knowledge sourcing, strategic alliances, capability building, organizational adaptation, new market development, and technological innovation (e.g., Atuahene-Gima, Reference Atuahene-Gima2005; He & Wong, Reference He and Wong2004; Katila & Ahuja, Reference Katila and Ahuja2002; Lavie & Rosenkopf, Reference Lavie and Rosenkopf2006; Tushman & O'Reilly, Reference Tushman and O'Reilly1996). According to March (1991:71), exploration includes activities captured by terms such as ‘search, variation, risk taking, experimentation, play, flexibility, discovery, and innovation’, and exploitation encompasses activities defined as ‘refinement, choice, production, efficiency, selection, implementation, and execution’. From these differences, we can see that exploration and exploitation represent two orientations in innovation activities: the former is used in experimenting with novel technologies and knowledge for developing new things, in order to create opportunities for future growth, and the latter is often used in refining, improving, and extending existing assets, in order to sustain existing performance (He & Wong, 2004; Jansen et al., Reference Jansen, van Den Bosch and Volberda2006).
One consensus in the literature is that firms can benefit from balancing the two types of innovation (e.g., Benner & Tushman, Reference Benner and Tushman2002, 2003; He & Wong, Reference He and Wong2004; Uotila et al., Reference Uotila, Maula, Keil and Zahra2009). This stream of research has stressed the importance of synergistic or complementary effects when adopting the two types of innovation orientations concurrently (Cao et al., Reference Cao, Gedajlovic and Zhang2009; He & Wong, Reference He and Wong2004). However, given that these two types of innovation orientations require vastly different knowledge and other assets, balancing the two imposes considerable tension or internal competition as firms need to divide limited organizational resources between the two (Cao et al., Reference Cao, Gedajlovic and Zhang2009; He & Wong, Reference He and Wong2004). Therefore, it is reasonable to argue that there exists resource exclusivity between these two types of innovation activities, meaning that if a firm increases its investment in one type of innovation, it has to decrease its investment in the other (Gupta et al., Reference Gupta, Smith and Shalley2006). In other words, there exists competing effects between the two innovation orientations.
Organizational scholars, however, are divided on how to manage the balance between the two innovation orientations (e.g., Boumgarden, Nickerson, & Zenger, Reference Boumgarden, Nickerson and Zenger2012). One stream of research proposes that domain or behavioral separation – an ambidextrous organization – may enable firms to manage these two types of innovation simultaneously to benefit from their complementary effects (Gibson & Birkinshaw, Reference Gibson and Birkinshaw2004; Jansen, Volberda, & van Den Bosch, Reference Jansen, Volberda and van Den Bosch2005; Lavie & Rosenkopf, Reference Lavie and Rosenkopf2006). Another stream of research posits a temporal or systemic separation (a punctuated equilibrium model) to manage the tension between the two, which helps firms, particularly those operating under conditions of severe resource scarcity but with certain organizational flexibility, manage the balance sequentially by employing one innovation orientation at a time to mitigate competing effects (Burgelman, Reference Burgelman2002; Gupta et al., Reference Gupta, Smith and Shalley2006; Mudambi & Swift, Reference Mudambi and Swift2011). Both views have their limitations. The ambidextrous organizational model overlooks the coordination costs incurred when a firm pursues both exploratory and exploitative innovation activities simultaneously. The punctuated equilibrium model does not pay sufficient attention to the transitional costs from one type of innovation to the other sequentially and the potential complementary effects when the two types of innovation are employed simultaneously (He & Wong, Reference He and Wong2004).
For SMEs, a strategically important decision is how to allocate limited resources between exploratory and exploitative innovation activities in order to optimize firm performance. To do so, SMEs need to consider the dynamics of both complementary and competing effects between the two innovation orientations on firm performance when formulating their innovation balancing strategies. This suggests that organizational structures for balancing the two types of innovation orientations should not be a simple choice between an ambidexterity or a punctuated equilibrium; rather it should be a deliberation of balancing both the positive and negative effects of the interplays between the two innovation activities, under different conditions.
Existing research has suggested that the relationship between firms’ innovation orientations and their growth performance is non-linear (e.g., Gupta et al., Reference Gupta, Smith and Shalley2006; Uotila et al., Reference Uotila, Maula, Keil and Zahra2009). However, what is this relationship? And more importantly, what are the latent mechanisms underlying such a relationship? To answer these questions, we need to examine not just the direct effects of both innovation orientations, but also the dynamics of the two (i.e., complementary and the competing effects) when they interplay, on firm performance.
First of all, the concurrency of the two innovation orientations can generate synergistic/complementary effects (+), which contribute positively to firm performance. The logic behind the complementary effect is that firms need exploration to increase the chances of technological breakthroughs, which provides directions for exploitative innovation; on the other hand, improvement of efficiency and productivity through exploitation helps firms accumulate knowledge and capability for exploratory innovation. In other words, balancing the two types of innovation orientations can enhance firm performance as they complement each other (Benner & Tushman, Reference Benner and Tushman2003; He & Wong, Reference He and Wong2004). The complementary effect is likely to be highest when the firm splits its organizational resources more or less evenly between the two types of innovation activities, whereas a shift toward either direction can reduce the complementary effect. This is to say that the complementary effects are a concave function of innovation orientations. In Figure 2a, we illustrate the benefit effects (+) of complementarity of the two innovation orientations on firm performance.
Secondly, given the competing effects, balancing the two types of innovation activities incurs co-ordinational, transitional and other costs which have a negative impact on firm performance. The cost effects are likely to be highest when a firm splits its organizational resources more or less evenly between the two types of innovation activities, whereas a shift toward either direction can reduce the competing effects (Haans, et al., Reference Haans, Pieters and He2016). In Figure 2b, we illustrate the cost effects (–) of balancing the two types of innovation orientations on firm performance.
Finally, following Haans et al.'s (2016) logic of latent mechanisms underlying U-shaped or inverted U-shaped relationships, when subtracting the cost effects (–) from the benefit effects (+), we can get an inverted U-shaped curve between the interplays of exploratory vs exploitative innovations and firm performance, as shown in Figure 2c. This inverted U-shaped relationship suggests that there exists an optimal point along the spectrum of a firm's innovation orientations between exploration and exploitation (which we define as ‘relative innovation orientation’ in this article) where the firm's performance may be maximized. We, thus, posit the following hypothesis:
Hypothesis 1: An inverted U-shaped correlation exists between a firm's relative innovation orientation of exploration vs exploitation (RIO) and its performance.
Moderating Effects of Cluster Relationships
In recent decades, the rapid change in product technology and demand conditions has forced firms, especially those in technology industries, to acquire knowledge and technology from their alliances and partnerships. Academic studies have shown that for firms in innovation intensive industries, to stay competitive in general and in technological development in particular, they need to actively seek information, knowledge, and other resources from external networks (Collins & Smith, Reference Collins and Smith2006; Phelps, Reference Phelps2010; Srivastava & Gnyawali, Reference Srivastava and Gnyawali2011). As a firm's knowledge base can determine its innovation capability, its knowledge acquisition from external networks can also influence the firm's innovation capability (e.g., Dai, Goodale, Byun, & Ding, Reference Dai, Goodale, Byun and Ding2018; Kogut & Zander, Reference Kogut and Zander1992; Zhou & Li, Reference Zhou and Li2012). Therefore, firms may increase their innovation capability by building relationships with other firms in their networks. In fact, as suggested in previous research (e.g., Bell, Reference Bell2005; Tripsas, Reference Tripsas1997), the networks from which a firm sources and acquires its external knowledge can have important influences on the firm's innovation strategy and performance. Such external networks include direct links of tightly-coupled partnerships or alliances for new product development, or indirect links of loosely-coupled connections such as joining industry associations or consortia (Dai et al., Reference Dai, Goodale, Byun and Ding2018). Among all types of external networks, industrial cluster networks remain a major network type from which firms acquire external knowledge, information and other resources for innovation capability (Autio et al., Reference Autio, Nambisan, Thomas and Wright2018).
Due to geographic proximity and industry relatedness of clusters (Arikan, Reference Arikan2009; Bell & Zaheer, Reference Bell and Zaheer2007; Porter, Reference Porter1990), network relationships in industry clusters can be more beneficial for cluster firms, as such firms are more likely to build direct or indirect links with others within the clusters, and acquire relevant information, knowledge and other resources for innovation capability (Pouder & Caron, Reference Pouder and Caron1996). Thus, a firm's cluster relationships can have direct impacts on its exploratory and exploitative innovation activities separately and on their interplays (complementary and competing effects) indirectly. Cluster relationships, therefore, can influence firms’ performance by influencing their innovation orientations (McCann & Folta, Reference McCann and Folta2011).
In recent decades, aspects of cluster relationships, such as inter-firm learning, knowledge sharing, and technological collaborations, have been identified as factors influencing cluster firms’ innovation performance (e.g., Inkpen & Wang, Reference Inkpen and Wang2006; Sammarra & Biggiero, Reference Sammarra and Biggiero2008; Zhou & Li, Reference Zhou and Li2012). However, how such network relationships contribute to a firm's innovation capability building and, thus, innovation strategy and performance, remains controversial. For example, Schilling and Phelps (Reference Schilling and Phelps2007) argue that firms embedded in alliance networks with high clustering (scope) and high reach (quality) have greater potential to increase their exploratory innovation than those without such alliances. In contrast, Ozer and Zhang (Reference Ozer and Zhang2015) suggest that cluster memberships enhance firms’ exploitative product innovation capability but may hinder their exploratory innovation.
The discrepancy in the results on the impact of cluster relationships on firm innovation performance is perhaps caused by insufficient attention to the different processes by which different cluster relational attributes affect firm's innovation orientations. For example, a firm's cluster relational attributes (scope and depth) can have different influences on the knowledge the firm acquires from its external networks (Wasserman & Faust, Reference Wasserman and Faust1994). This is because a firm's network/cluster breadth may expand the firm's scope of new knowledge acquisition, and network/cluster depth may increase the efficiency of the firm's uses or reuses of its existing knowledge (Katila & Ahuja, Reference Katila and Ahuja2002). In this study, we examine the impact of two relational attributes (scope and depth in cluster relationships) on the relationship between a firm's innovation orientation and performance.
Relational breadth and innovation orientation
Relational breadth is measured by the number of network ties a firm has with other cluster firms. Existing research has suggested that a firm's location in a cluster network can influence its number of ties in the network. For instance, the centrality of a cluster firm's location may enable the firm to acquire more knowledge, and enhance its innovation performance (Bell, Reference Bell2005; Tripsas, Reference Tripsas1997). Similarly, McCann and Folta (Reference McCann and Folta2011) explain that when a firm is located in a structural hole of a network, which enables the firm to develop more alliance partners (increasing the number of network ties), it may have a higher chance to improve its performance on exploratory innovation. On the other hand, Lin et al. (Reference Lin, Lin and McDonough2016) confirms that inter-organizational partnerships (numbers) increase the scope but not the depth that a firm might acquire knowledge from its network. Hence, the more network ties a firm has, the broader the scope of the firm's access to new information, knowledge, and technology, which, as a consequence, may encourage the firm to develop more exploratory innovation capability and pursue more exploratory activities.
Studies suggest that the number of a cluster firm's dyadic network ties with other cluster firms has a direct impact on the scope of the firm's heterogenous knowledge resources (e.g., Arikan, Reference Arikan2009; Stuart, Reference Stuart2000). However, the number of a cluster firm's network ties is not necessarily useful in the firm's use of existing ideas, knowledge, and information. Zang's (Reference Zang2018) work indicates that the scope of a firm's relational breadth in a network contributes more to its exploratory innovation capability than exploitative innovation capability. This is perhaps because too many network ties can constrain the firm from developing deep relationships with others, which enables circulation of homogeneous knowledge between cluster firms and is beneficial to the firm's exploitative innovation capability building (Lin et al., Reference Lin, Lin and McDonough2016; Zang, Reference Zang2018). Following this line of reasoning, we can argue that a firm's relational breadth (number of network ties) in a cluster positively influences its propensity to acquire heterogeneous knowledge, but not homogenous knowledge. In other words, the more network relationships a firm has, the more likely it acquires heterogeneous knowledge, which may be conducive to its exploratory innovation capability and, thus, improve performance. However, a firm's relational breadth is not useful in deepening its existing knowledge base and enhancing its exploitative innovation capability.
Hence, it is reasonable to hypothesize that increasing the number of cluster relationships enables cluster firms to develop more exploratory innovation capability, supported by extra heterogeneous knowledge acquired from their wider scope of network ties. The enhanced exploration capability in innovation, in turn, can increase the firm's exploratory innovation activity, and contribute to firm performance. Meanwhile, external resources will also strengthen the complementary effects between the two innovation orientations. Under this condition, if a firm's innovation orientation leans toward exploration, it will contribute to improved performance, as shown in Figure 3a. Similarly, the broader a firm's network, the more likely the firm will have more exploratory innovation activity through extra heterogeneous resources acquired through the expanded network ties, which, in turn, will mitigate the competing effects between the two, as shown in Figure 3b. Following Haans et al.'s (2016) model, subtracting the cost effects (as shown in Figure 3b) from the benefit effects (as shown in Figure 3a) will lead to the turning point of the inverted U-shaped curve between firms’ relative innovation orientation and performance leaning toward exploratory innovation, as shown in Figure 3c.
Hence, we posit the following hypothesis.
Hypothesis 2a: In a cluster, the larger the number of network ties a firm has with other cluster members, the more likely the turning point of the inverted U-shaped curve between its relative innovation orientation and performance moves toward exploration orientation for better performance.
Relational depth and innovation orientation
The depth of a firm's network ties with other members in the network measures its relational strength, and the stronger a relationship, the more likely the firm will develop deep-level closeness, reciprocity, and indebtedness with its relationship partners (Granovetter, Reference Granovetter1973), which may lead to information sharing, especially the exchange of tacit knowledge (Leonard & Sensiper, Reference Leonard and Sensiper1998). As suggested by Rowley et al. (Reference Rowley, Behrens and Krackhardt2000), the depth of network ties can impact a firm's innovation capability and performance. Deep cluster relationships enable cluster firms to acquire more informal information and tacit knowledge, and the closeness in such relationships may lead to a higher level of homogeneity in their knowledge structures and encourage more exploitative innovation activity (Bell & Zaheer, Reference Bell and Zaheer2007). This is because, in an industrial cluster, most firms are in related product families or on related value chains, and their knowledge bases are related and, to a large extent, similar (Autio et al., Reference Autio, Nambisan, Thomas and Wright2018), and strengthened network relationships can, therefore, contribute to a firm's endogenous creativity which contributes to its exploitation of existing technology (Madhavan, Gnyawali, & He, Reference Madhavan, Gnyawali and He2004).
Thus, firms with deep cluster relationships are more likely than those without such relationships to actively seek common ground for cooperation through strong network ties, which may lead to further convergence of their product designs and other innovations (Jansen et al., Reference Jansen, van Den Bosch and Volberda2006). When a firm's cluster relationships with other cluster members are strengthened, it is likely that the firm enhances its exploitative innovation capability enabled by increased homogeneous knowledge (Sen & Egelhoff, Reference Sen and Egelhoff2000). This situation is especially true when a small number of firms in cluster relationships create narrow but robust circulations of existing knowledge among themselves, which may fence off external (heterogeneous) influences (Ozer & Zhang, Reference Ozer and Zhang2015; Sen & Egelhoff, Reference Sen and Egelhoff2000). Indeed, deep and concentrated network relationships among cluster firms may lead such firms to be isolated from the influence of industry norms, the status quo of industry knowledge, and dominant technology (even just temporarily) (Ahuja, Reference Ahuja2000; Burt, Reference Burt2004; Burt, Reference Burt2009; Schilling & Phelps, Reference Schilling and Phelps2007).
On the other hand, excessively close relationships among cluster firms may also reduce the incentives for such firms to experiment with unknowns and explore new things, as some members may expect to get ‘free rides’ from others’ risk-taking, exploratory innovation activity (Dai et al., Reference Dai, Goodale, Byun and Ding2018). Thus, accumulation of homogeneous knowledge and other resources will lead to enhanced exploitative innovation capability but will not help improve exploratory innovation capability. This is to say that the stronger a firm's cluster relationships, the more likely it will acquire more homogenous knowledge in the cluster, and, thus, skew its innovation balance toward exploitation.
In summary, it is reasonable to posit that stronger cluster relationships help cluster firms accumulate more homogeneous knowledge and other resources for innovation, which may lead such firms to direct more attention to exploitative innovation, and, thus, enhance firm performance. Meanwhile, external resources can reinforce the complementary effect between the two types of innovation activities. Thus, a firm will benefit more if its innovation orientation leans toward exploitation, as shown in Figure 4a. On the other hand, stronger cluster relationships will help the firm develop more exploitative innovation capability enabled by extra homogeneous resources acquired from its deeper network ties, which, in turn, will mitigate the competing effect between the two types of innovation orientations, as shown in Figure 4b. Thus, subtracting the cost effects (Figure 4b) from the benefit effects (Figure 4a) we can have the turning point of the inverted U-shaped curve lean toward exploitative innovation, as shown in Figure 4c. Thus, we posit the following hypothesis.
Hypothesis 2b: In a cluster, the deeper a firm's cluster relationships, the more likely the turning point of the inverted U-shaped curve between its relative innovation orientation and performance moves toward exploitation for better performance.
METHODS
Sample and Data Collection
To test the hypotheses (H1, H2a, and H2b), we conducted a survey with innovation-intensive small and medium-sized enterprises (SMEs)[Footnote 1] in four clusters in Tianjin. As a transportation junction connecting inland Northern China with the global market via air and sea, Tianjin is one of four autonomous municipalities in China. In recent years, the city has become one of the country's innovation centers. At the end of 2015, with an increasing number of technology-oriented SMEs, Tianjin was home to nine high-tech industry clusters, including electronic information, biopharmaceutics, marine technologies, new materials, sustainable energy and environmental protection, medical apparatus, aerospace, new concept vehicles, and equipment manufacturing, which, collectively, contributed to about 71% of the city's gross industrial output. For these two reasons, Tianjin provided an ideal context in which to examine SMEs’ innovation orientation and their performance in clusters.
Endorsed by the Tianjin Municipal Government, we sent questionnaires to 1,256 innovation-intensive SMEs in Tianjin.[Footnote 2] We chose our sample following the definitions of enterprise size, registration location, and industry classifications set by the National Bureau of Statistics of China on SMEs. Our sample firms are from four innovation-intensive industries: electronic information industry (46.6%), biopharmaceutical industry (33.7%), new materials industry (12.5%), and sustainable energy and environmental protection industry (7.2%). Innovation-intensive industries are characterized by four attributes: rapid technological upgrades, technology-driven products, fierce market competition, and responsiveness to industry innovation trends.
We received 638 valid and complete questionnaires (an effective return rate of 50.8%[Footnote 3]). The results of a t-test show that there is no significant difference in industry distribution between firms that responded and those that did not (p = 0.097).
The questionnaire we used for this study comprises two parts. The first part consists of questions related to control, moderating and independent variables and was answered by general managers or equivalent in the surveyed firms. The second part relates to dependent variables and was answered by financial managers or equivalent in the firms. This method of requiring that the two parts of the questionnaire be completed by personnel in different functions is to mitigate self-reporting and self-evaluation effects that can result in common method variance (CMV) (Podsakoff, MacKenzie, Lee, & Podsakoff, Reference Podsakoff, MacKenzie, Lee and Podsakoff2003; Siemsen, Roth, & Oliveira, Reference Siemsen, Roth and Oliveira2010). To control for the impact of CMV, we also conducted Harman's one-factor test. According to Podsakoff et al. (Reference Podsakoff, MacKenzie, Lee and Podsakoff2003), if a single factor emerges that accounts for a large percentage of the total variance, the data may have a CMV issue. In our test, the highest factor accounts for 30.9% of the total variance explained, which indicates that CMV is unlikely to significantly affect the results.
Definitions and Measurements of the Variables
The independent variable in this study is the relative innovation orientation between exploration and exploitation (RIO), a state indicating the inclination of a firm's innovation orientation toward either exploration or exploitation, resulting from the difference in the firm's investment in the two type of innovation activities. In this study, we define RIO as the relative amount of exploratory vs exploitative innovation activities over a three-year period. There are two common methods to measure exploratory and exploitative innovations: first, by counting the frequency with which the two types of innovation orientations are mentioned in subject firms’ publicly available materials (e.g., annual reports and letters to shareholders) (Uotila et al., Reference Uotila, Maula, Keil and Zahra2009); second, by surveying a sample of subject firms using questionnaires (He & Wong, Reference He and Wong2004; Jansen et al., Reference Jansen, Volberda and van Den Bosch2005; Jansen et al., Reference Jansen, van Den Bosch and Volberda2006). The first method is suitable for studying large, public firms, for which public information is easily and reliably available. Most of our sample firms are unlisted and do not use open channels to communicate their firm-related information to the public. Neither is the limited public information entirely reliable. For these two reasons, we chose to use the survey method. The questionnaire used to measure exploratory and exploitative innovations was adopted from Benner and Tushman (Reference Benner and Tushman2002; Reference Benner and Tushman2003), He and Wong (Reference Inkpen and Wang2006), and Jansen et al. (Reference Jansen, van Den Bosch and Volberda2006). We relied on a seven-point Likert scale survey items, with 1 denoting ‘strongly disagree’ and 7 denoting ‘strongly agree’.
Two commonly used measures of relative innovation orientation are 1) the ratio of the two (Jansen et al., Reference Jansen, Volberda and van Den Bosch2005; Uotila et al., Reference Uotila, Maula, Keil and Zahra2009) and 2) the absolute difference of the two (Cao et al., Reference Cao, Gedajlovic and Zhang2009; He & Wong, Reference He and Wong2004). For example, Uotila et al. (Reference Uotila, Maula, Keil and Zahra2009) used ‘relative exploration orientation’ to meaure a firm's orientation toward exploratory innovation by calculating the number of mentionings of exploratory orientation against the total number of both exploratory and exploitative orientations in its publicly available information per year. Adapting Uotila et al.'s method, we designed (ER–EI)/(ER+EI) (ER represents exploratory innovation and EI exploitative innovation) to measure a firm's RIO. The reason we chose such a method is it measures the relative importance of a firm's innovation orientation between exploration and exploitation against the total investment in both orientations over three years (to mitigate the lagging effect of innovation strategies, especially exploratory innovation, on firm performance). In other words, it provides a directional indication of a firm's innovation orientation: when (ER–EI)/(ER+EI) is larger than zero, the firm's RIO skews toward exploration; and when (ER–EI)/(ER+EI) is smaller than zero, its RIO skews toward exploitation.
The dependent variable used in this study is firm growth performance (FGP). A firm's growth can be measured by improvements in its financial results and market competitiveness, compared to three years before (Cao et al., Reference Cao, Gedajlovic and Zhang2009; He & Wong, Reference He and Wong2004). Hence, we measure FGP by 1) growth of sales revenue, 2) increase in market share, 3) growth in after-tax profit, and 4) improvement in market competitiveness, over three years. Specially, we assign a value to each of these four items using a seven-point scale: if the growth rate of any item over three years is smaller than −30%, the item scores 1 point; if the growth rate is between −30% and 0, the item scores 2 points; if the growth rate is between 0 and 20%, the item scores 3 points; if the growth rate is between 20% and 40%, the item scores 4 points; if the growth rate is between 40% and 60%, the item scores 5 points; if the growth rate is between 60% and 80%, the item scores 6 points; and if it is larger than 80%, the item scores 7 points. We then average the firm's total score on all four items to measure its growth performance over a three-year period.
There are two moderating variables in this study: the breadth and the depth of a firm's network ties in its cluster. We use the number of network ties, including partnerships (e.g., joint venture partners), formal alliances (e.g., R&D alliances), and informal alliances (e.g., participating in various industry associations or consortia), that a firm has in the cluster to measure the breadth of its cluster relationships (NCR), and the strength of these network ties to measure the depth of its cluster relationships (SCR) (Bell, Reference Bell2005; Gilsing, Nooteboom, Vanhaverbeke, Duysters, & van den Oord, Reference Gilsing, Nooteboom, Vanhaverbeke, Duysters and van den Oord2008; Giuliani, Reference Giuliani2007; Liu, Reference Liu2011). We measured each item on a seven-point Likert scale, with 1 denoting ‘strongly disagree’ and 7 denoting ‘strongly agree’ with questions such as ‘the firm has a large number of ties with … in a cluster’, and ‘the firm has deep connections with…in a cluster’.
We used five control variables in this study: firm size, firm age, firm ownership, market competition, and industry cluster. First, firm size influences the resources available for a firm to carry out innovation activities, and, thus, its innovation capability. We measured firm size by a firm's total assets, sales revenues, and total number of employees (Baum, Locke, & Smith, Reference Baum, Locke and Smith2001; Cao et al., Reference Cao, Gedajlovic and Zhang2009; Lavie, Kang, & Rosenkopf, Reference Lavie, Kang and Rosenkopf2011). Second, firm age contributes to a firm's knowledge accumulation and experience related to innovation activities. We measured firm age by the number of years since the firm's registration (1 = less than 4 years; 2 = 4–6 years; 3 = 7–10 years, 4 =11–15 years; 5 =16–20 years; 6 = 21–30 years; and 7 = more than 30 years). Third, firm ownership has an impact on a firm's propensity for risk-taking in innovation. In China, firm ownership (i.e., state-owned or privately-owned) has a strong impact on its innovation strategy and growth performance. We set a dummy variable for firm ownership (0 = non state-owned enterprises; and 1= state-owned enterprises). Fourth, the degree of market competition has an impact on a firm's strategic choice of innovation orientation. We measured the market competition a firm faces by the frequency of ‘price wars’ involving subject firms, the number of entries of new firms, and whether competitors followed any successful product launch with similar offerings in the market place (Auh & Menguc, Reference Auh and Menguc2005; Baker & Sinkula, Reference Baker and Sinkula2007; Jansen et al., Reference Jansen, Volberda and van Den Bosch2005; Jansen et al., Reference Jansen, van Den Bosch and Volberda2006). Finally, the industry cluster in which a firm operates may influence its innovation strategy as different industry clusters often have different network structures (e.g., density and centrality) and different developmental paces in innovation. To reduce noise generated by industry differences, we set dummy variable for each industry cluster: 1 = in this industry, 0 = not in this industry. Table 1 summarizes the dependent and independent variables in this study with their definitions and measurements.
We used SPSS 19.0 to test the reliability of the data collected for this study. The alpha coefficient (as shown in Table 1) was larger than 0.7 for each variable on the internal consistency test, indicating that the data meet the requirement of high internal consistency.
RESULTS
Table 2 depicts the statistics (means, standard deviations, and correlations) of the variables of this study. From the means, we can see that sample firms are SMEs, and, on average, the market competition the firms face is mild, 34.2% of the firms are state-owned enterprises, and the average age of the firms is 8.6 years. The test of multicollinearity indicates that the values of variance inflation factor (VIF) for all the independent and control variables are below the threshold of 10, and their tolerance values are above the threshold of 0.1, verifying that the likelihood of problems caused by multicollinearity is small (Fox, Reference Fox1991).
Notes: * p < 0.05, ** p < 0.001.
We used a hierarchical regression method to test the hypotheses. Four models of the five control variables with the independent or moderating variables, and interactive items of them were tested. In Table 3, we show the results of the correlations between firms’ relative innovation orientation (RIO), number of cluster relationships (NCR), strength of cluster relationships (SCR) and firm growth performance (FGP).
Dependent Variable: Firm growth performance (FGP)
Notes: Significance levels: * p-value < 0.05; ** p-value <0.01; ***p-value <0.001.
M1 is the baseline model, which includes five control variables, as well as EI and ER. M2 is built by adding the relative innovation orientation (RIO) and its quadratic term (RIO2) on M1. The results of M1 and M2 indicate that the interactive items of RIO and RIO2 have more effect than the individual ER or EI on independent variable FGP (Adj. R 2 = 0.211, p < 0.001). Following the suggestions by Haans et al. (Reference Haans, Pieters and He2016) on empirical tests of non-linear relationships, we undertook a three-step method, proposed by Lind and Mehlum (Reference Lind and Mehlum2010), to test the inverted U-shaped relationship in this study. The curve can be expressed as Equation 1.
First, we tested the sign and significance of coefficient β2 of Equation 1. The correlation between the quadratic term of RIO and FGP is negative and significant (β2 = −0.246, p < 0.001). Second, we tested the slopes at both ends of the data range. We did an overall test of presence of the inverted U-shape. The results are shown in Table 4. The slope at the low end of the X-range (XL) is positive and significant (the slope at XL = 4.59, p < 0.001), and the slope at the high end (XH) is negative and significant (the slope at XH = −1.94, p < 0.001). Third, we tested whether the turning point of the curve is within the data range. The turning point of curvilinear at −β1/2β2 = 0.398. [Footnote 4] The 95% confidence interval of the turning point is within the data range ([0.17, 0.60] ∈ [−0.75, 0.75]). To ensure the correct interpretation of the results, we further examined the joint significance of the inverted U-shaped correlation. Based on the research of Sasabuchi (Reference Sasabuchi1980), Lind and Mehlum (Reference Lind and Mehlum2010), and Haans et al. (Reference Haans, Pieters and He2016), we tested the joint significance of direct and squared terms of RIO, the joint significance of the control variables. These results (as shown in Table 4) suggest that the inverted U-shaped relationship is significant.
Notes: Significance levels: ** p-value <0.01; ***p-value <0.001.
Thus, the above results support H1, that is there exists an inverted U-shaped correlation between a firm's relative innovation orientation and its growth performance, as (ER-EI)/(ER+EI) ∩ FGP.
We further build M3 by adding to M2 the number of a firm's cluster relationships (NCR), the interaction of NCR and the RIO, the interaction of NCR and the quadratic term of RIO. Following the procedures in Aiken et al.'s (Reference Aiken, West and Reno1991) study, we established Equation 2 for the moderating effect of NCR on the correlation between RIO and FGP as:
The results of M3 suggest that the interaction of a firm's NCR and the linear term of RIO is positive and significant (β 6 = 0.251, p < 0.001), and the interaction of NCR and the quadratic term of RIO is also positive and significant (β 4 = 0.136, p < 0.001). In addition, β 3 = –0.249, p < 0.001, β 5 = 0.155, p < 0.001, β 7 = 0.103, p < 0.05. Haans et al. (Reference Haans, Pieters and He2016) suggest that a shift of the direction of the turning point in a U-shaped curve depends on the coefficients β 3, β 4, β 5 and β 6: If (β 4 β 5 – β 3 β 6) is positive, the turning point will move to the right direction as the moderator increases. If (β 4 β 5 – β 3 β 6) is negative, the turning point will move to the left direction as the moderator increases (p. 1187). As (β 4 β 5 – β 3 β 6) = 0.084 > 0, we can conclude that the turning point of the inverted U-shaped curve will move to the right as NCR increases (meaning a firm's relative innovation orientation skews toward exploratory innovation when its number of cluster relationships increases).
The two curvilinear relationships for large and small NCR are plotted by substituting centered high (+1 standard deviation) and low (−1 standard deviation) values in Equation 2. As shown in Figure 5, when NCR increases, the turning point of the inverted U-shaped curve moves rightward. Thus, H2a is supported (Adj. R 2 = 0.320, p < 0.001).
We build M4 by adding to M2 the strength of a firm's cluster relationships (SCR), the interaction of SCR and the RIO, the interaction of SCR and the quadratic term of RIO. This model was used to test the moderating effect of SCR on the correlation between a firm's RIO and FGP. Similar to testing the moderating effect of NCR, we established Equation 3 for the moderating effect of SCR on the correlation between RIO and FGP as:
As shown in M4, the interaction of SCR and RIO is negative (β 11 = −0.287, p < 0.001), and the interaction of SCR and the quadratic term of RIO is also negative (β 9 = −0.126, p < 0.001). The direct effect of SCR on FGP is positive (β 12 = 0.207, p < 0.001). In addition, β 8 = −0.230, p < 0.001, β 10 = 0.088, p < 0.05. Based on these results, we found that the turning point of the inverted U-shaped curve between RIO and FGP will move to the left direction (meaning a firm's innovation orientation skews toward exploitative innovation) as SCR increases.
The two curvilinear relationships for strong and weak SCR are plotted by substituting centered high (+1 standard deviation) and low (−1 standard deviation) values in Equation 3. As shown in Figure 6, when SCR increases, the turning point of the inverted U-shaped curve moves leftward. Thus, H2b is supported (Adj. R 2 = 0.363, p < 0.001).
DISCUSSION
A strategically important decision for SMEs is, under different conditions, how to allocate limited resources between exploratory and exploitative innovation activities in order to optimize firm performance. Our empirical research on 638 SMEs in four clusters in Tianjin, China confirms that an inverted U-shaped correlation exists between a firm's relative innovation orientation and its growth performance (H1 is supported). We also verify that this correlation is moderated by the firm's cluster relationships: the broader a firm's cluster relationships, the more likely the turning point of the inverted U-shaped curve will move toward exploratory innovation orientation for greater growth performance (H2a is supported); on the other hand, the stronger a firm's cluster relationships, the more likely the turning point of the inverted U-shaped curve will move toward exploitative innovation orientation for greater growth performance (H2b is supported). These results carry important implications for theory and practice.
Theoretical Implications
First, we theorize and test that a latent mechanism underlying the inverted U-shaped curve between a firm's relative innovation orientation and performance is the result of an additive combination of two inverted U-shaped curves: complementary effects of the two innovation orientations is an inverted U-shaped curve which have a positive impact on firm performance (benefit effects), and competing effects of the two is also an inverted U-shaped curve which have a negative impact on firm performance (cost effects). Additively combine these two effects generate an inverted U-shaped curve between a firm's relative innovation orientation and performance. This process contributes to Haans et al.'s (2016) model of latent mechanisms underlying U- or inverted U-shaped curves. The results of our study also enrich the ongoing discourse on firms’ innovation balancing strategy between exploration and exploitation and performance.
Second, we theorize and test the moderating effects of cluster relationships on the inverted U-shaped relationship between a firms’ innovation orientation and growth performance. In contrast to previous studies that use a firm's cluster network's characteristics (e.g., the firm's locations in the network) or compare cluster firms with non-cluster ones in examining the effects of cluster relationships on firms’ innovation strategy, we tested the moderating effects of relational attibutes (i.e., breadth and depth) on cluster firms’ relative innovation orientation and performance. Theoretically, we extend the level of analysis in this field to inter-organizational levels. From a cluster relationship perspective, this finding enriches our understanding of the relationship between firms’ cluster relational attibutes and their knowledge acquisition and innovation capability building, and thus, their impact on firm performance.
On the breadth of cluster relationships, we find that an increased number of network ties is beneficial to a firm's exploratory innovation capability and may lead the firm to have more exploratory innovation activity. This finding is similar to the results of Fiol (Reference Fiol1995) and Zang (Reference Zang2018), which suggest that the number of a firm's network ties may nurture creative breakthroughs. The reason behind this is because broad cluster networks help a firm acquire more heterogeneous resources from different partners (McCann & Folta, Reference McCann and Folta2011; Wang & von Tunzelmann, Reference Wang and von Tunzelmann2000), which, in turn, can increase the likelihood of the firm pursuing new unknowns. As shown in Figure 5, the number of a firm's relational ties has little impact on its performance if it skews toward exploitative innovation. This finding supports Rowley et al.'s (2018) argument that broad cluster relationships may benefit a firm's exploratory innovation capability, but not its exploitative innovation capability. However, this result is different from Ozer and Zhang (Reference Ozer and Zhang2015) work, which suggests that a focal cluster firm's network ties are positively related to its exploitative innovation in product development. We believe this discrepancy is caused by different measures of exploration and exploitation. We focus on general innovation activities, but Ozer and Zhang focus on product development.
On the depth of relational ties, we find that the strength of a firm's relational ties may benefit the firm in exploitative innovation by increasing the efficiency of using existing knowledge and other technological assets. The reason behind this is perhaps because a firm with deep cluster relationships facilitated by trusted and reciprocal links with relational firms can enhance deep-level interactions and increase the chance of acquiring more homogenous resources through the network ties (Jansen et al., Reference Jansen, van Den Bosch and Volberda2006). Deep interactions of cluster firms enable them to exchange informal information and share tacit knowledge, which may be conducive to incremental innovations (Fleming, Reference Fleming2001; Kogut & Zander, Reference Kogut and Zander1992). We also find that when a cluster firm possesses stronger network ties with other firms in a cluster, its performance can be negatively influenced if it adopts an exploratory innovation orientation, as illustrated in Figure 6 (on the right of the intersection of the two curves). This suggests that firms with strong network ties that adopt an exploratory innovation orientation may suffer lower performance than those who concentrate on exploitative innovation, at least in the short term. This conclusion is in alignment with Dai et al. (Reference Dai, Goodale, Byun and Ding2018) who claim that close R&D alliances among technology-intensive firms can be detrimental to their strategic flexibility and innovation performance. This is because firms embedded in deep network relationships may need to give up their own innovation projects in order to have their strategic focuses aligned with the interests and agendas of the partner firms in the relationship. Such ‘compromising’ behavior may lead to a phenomenon of some partner firms expecting ‘free rides’ – benefitting from innovation outcomes shared by closely-knitted networks without investing themselves. If every partner held such an expectation this could lead to the downfall of the long-term viability of the cluster, especially in high-risk, high-return industries where exploratory innovation is key. On this, Arora, Athreye, and Huang (Reference Arora, Athreye and Huang2016) suggest that an effective strategy for innovation forerunners in cluster collaborations is to patent their innovations, as they may lose competitive advantages due to unintended spillover effects when the cluster relationships deepen, which, as a consequence, may reduce innovation activities in the cluster. For example, most cluster firms in our sample are in close geographic proximity, and they are collaborators and competitors at the same time. When the relationship between such firms deepens (the network ties are strengthened), they tend to develop interdependence with one another and benefit potentially from free rides on the partners’ innovation outputs, at least in the short term. However, in the long run, such a tendency may be detrimental to the viability of the relationship and firm performance.
For example, most cluster firms in our sample are in close geographic proximity, and they are collaborators and competitors at the same time. When the relationship between such firms deepens (the network ties are strengthened), they tend to develop interdependence with one another and benefit potentially from free rides on the partners’ innovation outputs, at least in the short term. However, in the long run, such a tendency may be detrimental to the viability of the relationship and firm performance.
Practical Implications
Our research carries significant practical implications, especially for how SMEs optimize their performance by managing innovation balancing strategies and cluster relationships, under different conditions. Our findings provide strategic guidance for SMEs that face constrained resources on how to adjust their innovation strategies according to their cluster relationships, as well as on how to manage their cluster relationships when employing different innovation strategies.
From the perspective of innovation orientation, a cluster firm should consider its innovation strategy in the context of its cluster relationships. Ultimately, an optimal innovation balancing strategy between exploration and exploitation is not a fixed target; it is contingent upon a firm's internal and external conditions, such as the firm's cluster relational ties. A firm's increased number of relational ties will help it acquire heterogenous resources from its cluster networks, which can enhance the firm's exploratory innovation capability. The implication for managers in this situation is that the focal firm should adopt an innovation balancing strategy skewing toward exploration to benefit from the scope of its network ties. On the other hand, a firm's cluster relational strength helps the firm acquire homogenous resources, which enhances the firm's exploitative innovation capability. The implication is that firms in this situation should adopt an innovation balancing strategy leaning toward exploitation to capitalize on the strength of their network ties.
The results of this study also offer practical guidance for firms aiming to enhance their innovation capabilities by managing their cluster relationships. If a firm needs to enhance its exploratory innovation capability, it should consider increasing the scope of its relational ties with other cluster firms, such as building more collaborative partnerships with other firms, or participating in more formal or informal collaborative alliances or industry associations. On the other hand, if a firm needs to enhance its exploitative innovation capability, it should focus on strengthening relationships with its strategic partners, such as intensifying information exchange and knowledge sharing.
Limitations and Future Research Implications
There are several limitations in this research. First, we had a relatively large proportion of state-owned enterprises in the sample (34.2%). Given that state-owned enterprises often have less control over their innovation strategy (e.g., they may pursue an innovation project, focusing on one type of innovation but ignoring the other, simply to fulfill a national innovation mission or to get subsidies from the government), their innovation strategy and growth performance may not reflect the effects of their own innovation capabilities and the influence of their cluster relationships, even after we controlled for firm ownership. Second, our sample is limited to one city – Tianjin (with a small number of firms registered in Tianjin but located in nearby Beijing). Given significant geographic differences in the economy, labor force, industry policy, market, and so on, across China, our findings cannot be generalized across the entire country.
For future research, we intend to expand our investigation to other geographic locations in China. Further effort should also be put into identifying the relational conditions under which cluster firms can manage the optimal balance point of this inverted U-shaped correlation between innovation orientations and performance. Future research might also examine other possible moderators that could affect this inverted U-shaped correlation.
CONCLUSION
This study advances our understanding of the relationship between innovation strategy (i.e., balancing exploratory and exploitative innovation) and firm performance in the context of cluster networks. Our research highlights the importance of cluster relationships in knowledge acquisition and innovation capability building, and, thus, firm performance, for cluster firms.