Master Thesis for MSc IB&M
A consolidation of the learning orientation literature, exploring
moderators and examining the role of its underlying values: a
meta-analysis
University of Groningen
Faculty of Economics and Business
Supervisor
Dr. Olof Lindahl
Second Assessor
Dr. Christopher Schlägel
Submitted by
Dave van Dam
S3858421
d.van.dam.1@student.rug.nl
i
Table of contents
Table of contents ... i
List of tables ...ii
List of figures ... iii
Abstract ... iv
1 Introduction ... 1
2 Literature review ... 4
2.1 Hypothesis building ... 7
2.2 Hypotheses for the moderating effect on the relationship between LO and innovation ... 12
3 Methodology ... 16
3.1 Literature search and sample ... 16
3.2 Inclusion criteria and coding ... 17
3.3 Analytic procedures ... 20
3.3.1 Bivariate meta-analysis ... 20
3.3.2 Commonality analysis ... 20
3.3.3 Moderator analysis ... 21
4 Results and analysis ... 22
4.1 Results of the Bivariate meta-analysis ... 22
4.2 Results of the Commonality analysis ... 23
4.3 results of the moderator analysis ... 26
5 Discussion and conclusions ... 30
5.1 Implications for theory ... 30
5.2 Implications for practice ... 31
5.3 Limitations ... 32
5.4 Future research ... 33
5.5 Conclusions ... 34
ii
List of tables
Table 1: Keyword search information ... 17
Table 2: List of included studies ... 19
Table 3: Results of the bivariate meta-analysis ... 23
Table 4: Meta-analytic correlation table ... Fout! Bladwijzer niet gedefinieerd.24 Table 5: Regression results for the Commonality analysis ... 25
Table 6: Details of the Commonality analysis ... 26
Table 7: Moderator regression of control of corruption ... 27
Table 8: Subgroup analysis for control of corruption ... 28
Table 9: Subgroup analysis for industry ... 29
iii
List of figures
iv
Abstract
Learning orientation (LO) has received substantial scholarly attention as it can contribute notably to the firm’s innovativeness. This study further investigates this relationship in a meta-analysis of 38 studies from the LO literature. This research consolidates the field to confirm the relationship between LO, and the values underlying such an orientation, and innovation. This research considers commitment to learning, shared vision, open-mindedness, and intra-organizational knowledge sharing as the values that make a LO. Additionally, this research investigated the interrelated nature of these underlying values. While
intra-organizational knowledge sharing has a significantly larger unique impact on firm innovation than the other values, this paper finds that all four LO values should be fostered in order to maximize innovativeness gains. To further explore the relationship between LO and firm innovativeness this paper analyses the moderating role of country-level corruption, firm industry, and firm size. The findings indicate that firm size and industry are unimportant for pursuing LO, but that in environments with high corruption LO can be more beneficial. Implications of the findings point to the importance of using the LO construct as a four-dimensional model in both theory and practice to properly capture the effectiveness of LO. Implications also suggest that national differences in the relationship between LO and innovation should be considered.
Keywords: Learning orientation; innovation; meta-analysis; commonality analysis;
1
1 Introduction
Innovation is one of the most important drivers of firm performance. Firms that are capable of bringing new products to the market can create an advantage over their competition (Cooper & Kleinschmidt, 1987). Innovation, in turn, is closely related to a firm’s capacity to gather and process information, or rather of the firm’s capacity to learn (Calantone, Cavusgil, & Zhao, 2002; Damanpour, 1991; Goes & Park, 1997; Sinkula, Baker, & Noordewier, 1997). Firms that are capable of learning will be able to more easily capitalize on opportunities in the market by bringing out appropriate products. To enable the firm to learn, the firm will want to adopt a culture of learning (Sinkula et al., 1997). When a firm’s management has
implemented key features within the firm that create such a culture, the firm is considered to have a learning orientation (Hakala, 2011).
Learning orientation (LO) refers to a set of organizational values that influence a corporation’s propensity to build and use new knowledge (Sinkula et al., 1997). LO
influences what information a firm gathers (Dixon, 1992), how the information is interpreted (Argyris & Schön, 1978), evaluated (Sinkula et al., 1997), and shared (Moorman & Miner, 1998). LO is displayed through 4 dimensions, commitment to learning, open-mindedness, shared vision (Sinkula et al., 1997), and intra-organizational knowledge sharing (Calantone et al., 2002). LO is considered to be beneficial for the firm because it creates
knowledge-questioning and knowledge-enhancing values that help it to develop breakthrough products, services, and technologies, and the exploration of new markets (Slater & Narver, 1995). Through this process, the literature suggests that LO has a positive impact on firm
performance (Barrett, Balloun, & Weinstein, 2005; Hult, Hurley, & Knight, 2004). This effect is often researched directly, but there is also a branch of research focusing on the mediating role of firm innovativeness (Calantone et al., 2002; Rhee, Park, & Lee, 2010; Sinkula, 1994). While the literature is consistent in finding a positive relationship between LO and firm innovativeness, the literature is still unclear when it comes to the magnitude of the impact. Studies have reported varying effect sizes, for example, Calantone et al. (2002) found a moderate effect size, where others report a very strong effect (Mahmoud, Blankson, Owusu-Frimpong, Nwankwo, & Trang, 2016). Besides different effect sizes, there are also papers showing only partial significance in testing LO and firm innovativeness (Calisir, Altin
2 engage in a market orientation, an entrepreneurial orientation, or is LO the best option? The differences between the effect sizes in the literature suggest that the answer to this question depends on the context in which the firm operates (Ellis, 2010). At this point, the literature does not yet understand which factors influence this decision, as moderators are not
sufficiently researched. A meta-analysis can combine data from the literature to find some of these moderating factors. While for the relationship between LO and firm performance such research has been conducted (Goh, Elliott, & Quon, 2012), for firm innovativeness there has been no attempt in the literature so far, to the best of my knowledge. Not knowing how effective a certain strategic orientation will be means that it is very difficult for managers to evaluate which orientation to adopt. Therefore this paper will use meta-analysis to find the true effect size of LO on firm innovativeness and evaluate the moderating effects that can explain the different findings in the literature.
While the importance of investigating differences in the effect of LO across countries has been indicated in the literature (Calantone et al., 2002), no research has examined country-level moderators to understand the differences in the LO – firm innovation relationship. As the current literature has gathered data in many countries such as Turkey (Keskin, 2006), Taiwan (Lee & Tsai, 2005), and the UK (Wang, 2008), meta-analysis provides a good opportunity to test for country-level effects by combining such studies. The literature on institutions has suggested that different institutions influence how a country’s markets work, and how they innovate (Ingram & Aoki, 2003). It is reasonable that firms that have a different path to innovation will also differ in the way it uses its knowledge. As a first step in exploring such differences, this paper will look into the moderating effect of corruption on the LO – innovativeness relationship.
To address this research gap this paper will be conducted using the following research question: What is the effect of learning orientation, and its underlying values, on firm innovativeness, and how is this relationship moderated by corruption?
3 will have on the firm’s innovativeness. To explore this topic an exploratory research question is employed:
What are the unique and common effects of commitment to learning, open-mindedness, shared vision, and intra-organizational knowledge sharing on firm innovativeness?
This paper will use meta-analysis to combine much of the existing literature on the topic of LO to get a better understanding of its effect size on firm performance. The research will analyze papers using the 4 values suggested by Sinkula et al. (1997) and Calantone et al. (2002) to research LO. Google Scholar will be used as the primary method of finding papers. By using these 4 values the research is limited to studies from the past 23 years. Using Google Scholar shows 23.200 results for the term “learning orientation” in this time window.
This research will make the following contributions to the current literature. First, this paper will consolidate the research on LO, its underlying values, and innovation to confirm their direct relationship with innovation. Additionally, this paper will employ a Commonality analysis to disentangle the unique and combined effects of the LO values to get a deeper understanding of the impact of LO on innovation. This research will also start exploring country-level effects that moderate the relationship between LO and innovation. This is the first study to incorporate such effects, the results indicate that the effect does change as corruption levels differ. This suggests that more research on national effects should be researched.
4
2 Literature review
Resource-based view
Before going into what LO or other strategic orientations are, it should first be explained where their value comes from. The resource-based view (RBV) is the most commonly used theory for explaining the value of LO, and other strategic orientations. RBV was initially introduced by Penrose (1959) but was popularized by Barney (1991). In his paper ‘Firm Resources and Sustained Competitive Advantage’ Barney argued that firms can gain a sustainable competitive advantage from resources and capabilities that are valuable, rare, imperfectly imitable, and non-substitutable (Barney, Wright, & Ketchen, 2001). Such resources can either be knowledge-based or property-based (Miller & Shamsie, 1996). How management utilizes this resource is equally important as its quality. Only when the firm properly organizes its valuable, rare, imperfectly imitable, and non-substitutable will they lead to sustained superior performance (Wiklund & Shepherd, 2003).
Organizational learning
To get a full understanding of LO, the learning literature should be discussed, as LO stems from this field. This literature, focussing on the process of organizational learning, explains improved firm performance either through the ability of learning to facilitate behavioral change (Hurley & Hult, 1998; Sinkula, 1994), or to inspire new ways of thinking (Garvin, 1993). The most commonly used definition describes a learning organization as one that is skilled at creating, acquiring, and transferring knowledge, and capable of modifying its behavior accordingly (Garvin, 1993). Important in this process is that it is continuous and that it is included in all decision-making activities, as it fails to have an impact otherwise (Day, 1994; Sinkula et al., 1997). In a learning organization information acquired throughout the firm, that would otherwise be lost, is used to enrich this decision-making process. Day (1994) highlights the front row employees across the firm, which are a rich source of market
information . These employees collectively monitor what the market likes, wants, and needs. However, in an average firm most of this knowledge is lost, but in a learning organization this information is shared throughout multiple departments and branches, so that it reaches
5 a source for sustainable competitive advantage as the better decision-making leads to more efficiency, which leads to lower prices for customers (value). It is a product of a complex organizational environment that makes it hard to evaluate and thus to copy for a competitor (imperfectly imitable). Being a learning organization is also considered to be rare because it requires the firm to setup a complex system of knowledge sharing, which most firms are not capable of achieving (Day, 1994).
Strategic orientations
Strategic orientations are “principles that direct and influence the activities of a firm and generate the behaviors intended to ensure its viability and performance” (Hakala, 2011: 199). A strategic orientation can be seen as the adaptive culture of a firm (Noble, Sinha, & Kumar, 2002), or as adaptive mechanisms (Hakala, 2011), which steer the way it handles its
environment. The most commonly studied strategic orientations are market orientation (MO), entrepreneurial orientation (EO), and learning orientation (LO). MO has received the most scholarly attention out of these orientations. This orientation focusses on creating superior customer value and organizational success through satisfying customer needs (Jaworski & Kohli, 1993). MO is often conceptualized through three dimensions: (1) customer orientation, (2) competitor orientation, and (3) inter-functional coordination (Narver & Slater, 1990). EO focusses on the choice of market penetration and the internal processes that enable the firm’s management to act on options in the market (Lumpkin & Dess, 1996). EO is conceptualized through three dimensions: (1) innovativeness, (2) risk-taking, and (3) proactiveness (Miller, 1983). LO is discussed in detail below. Employing any of such orientations can help a firm achieve higher financial performance (Deutscher, Zapkau, Schwens, Baum, & Kabst, 2016). While research on strategic orientations has mostly focussed on each orientation by itself, more recent work has also considered that firms a capable of employing multiple strategic orientations and have worked on analyzing their joint effects (Deutscher et al., 2016; Grinstein, 2008). The strategic orientation that the firm chooses influences its behavior, its culture (Braunscheidel & Suresh, 2009), and ultimately it's competitive advantage (Deutscher et al., 2016). As such deciding on the most suitable orientation(s) is a crucial decision for the firm (Grinstein, 2008).
Learning orientation
6 as “giving rise to that set of organizational values that influence the propensity of the firm to create and use knowledge” (Sinkula et al., 1997: 309). Wang (2008) explains LO as the firm values that influence its approach to acquiring information. Calantone, et al. (2002) instead built on Sinkula et al.’s definition but added that LO is the use of developing new knowledge with the explicit goal to gain a competitive advantage. This is the most used conceptualization of LO in the literature to date. In their paper Sinkula et al. (1997) describe LO as a set of organizational values, these values are commitment to learning, open-mindedness, and shared vision. Later a fourth value, intra-organizational knowledge sharing, was added by Calantone et al. (2002).
The literature is divided on including intra-organizational knowledge sharing as a part of LO, although no researchers truly tackle this discussion. Some researchers simply chose to include the value (Chen, Liu, & Wu, 2009; Keskin, 2006; Nybakk, 2012; Onağ, Tepeci, & Başalp, 2014), while others do not (Chenuos & Maru, 2015; Eshlaghy & Maatofi, 2011; Perin & Sampaio, 2003; Wang, 2008). However, in the learning literature knowledge sharing is considered a requirement for proper learning throughout the organization (Day, 1994). Without learning company-wide, a firm cannot be considered to have a certain set of organizational values. Accordingly, this paper considers intra-organizational knowledge sharing as a part of LO.
LO and performance
The effect of LO on the performance of the firm is mostly researched in two ways. It is most commonly suggested to directly improve the firm’s financial performance through its
decision-making practices (Sinkula et al., 1997). Other research focusses on LO’s effect on the firm’s innovation performance instead (Calantone et al., 2002). While this paper focusses on the role of LO in the firm’s innovation process I will also briefly explain the direct
7 market, and anticipate its changes. When the firm proactively manages its market it can generate superior long-term performance (Calantone et al., 2002).
2.1 Hypothesis building
LO and innovativeness
A firm learns through studying and interacting with its environment (Calantone et al., 2002). Even before LO was studied in combination with innovativeness the concept of learning was seen as an important enabler of innovation (Cohen & Levinthal, 1990). Cohen and Levinthal (1990) argued that innovation requires the ability to “recognize the value of new, external information, assimilate it, and apply it to commercial ends” (Cohen & Levinthal, 1990: 128). When a firm has a strong learning orientation, then this means that the firm is actively
questioning the information that it processes (Baker & Sinkula, 1999). These firms frequently rethink if their way of doing things is still optimal. Per definition this makes the employees question how the information applies to the firm (Day, 1994). This directly impacts the ability of the firm to recognize the value of new information. For a firm to innovate it also needs its employees to share information and knowledge (Jiménez-Jiménez & Sanz-Valle, 2011). A firm with a strong LO will actively support its employees to do so (Calantone et al., 2002). According to Calantone et al., (2002) LO affects innovation in three ways:
1) Because learning happens through organizational observation, and interactions with its environment, it is more likely that the organization is committed to innovation
2) Firms with a strong LO are connected with their environment, it therefore has more knowledge about customer needs and emerging markets.
3) The strong connection with the environment also means that the firm monitors its competition. The firm can gain new insights by analyzing successes and failures from its competition, leading to higher innovativeness.
Later in the innovation process, when new products are being developed LO pushes team members to share their ideas and to work through differences that spawn from this. This leads to quality enhancements in the product developing phase (Bunderson & Sutcliffe, 2003; Huang & Li, 2017). Overall this leads to the hypothesis:
8 Commitment to learning and innovativeness
In LO commitment to learning is referred to as the value the firm holds toward learning, and how much the firm promotes it (Sinkula et al., 1997). A firm that actively encourages its employees to learn creates a culture of learning, which is linked to the firm’s ability to fully understand the firm’s environment (Calantone et al., 2002). This is closely related to the notion that learning fosters the ability to think and reason, and to critically evaluate the effect and one’s actions (Tobin, 1993). A firm that is committed to learning considers the learning process as value, and will therefore more actively collect and process information (Sinkula et al., 1997). As a result, the firm will be capable of developing its knowledge about its market, rivals, and customers (Kandemir & Hult, 2005). Understanding its environment allows the firm to understand and predict customer’s needs (Damanpour, 1991), as a result, the firm is less likely to miss out on opportunities as it recognizes innovation opportunities. Additionally, a firm that is committed to learning will also be ready to make the required investments to facilitate the learning process (Calantone et al., 2002).
A firm where the employees are constantly trying to gather information is more likely to come up with new products that would fit the market, as it understands what the market requires. This means that they come up with better ideas when trying to innovate, this will more often result in an innovative product. Overall this leads to the following hypothesis:
H2: The higher a firm’s commitment to learning, the higher it’s innovativeness. Shared vision and innovativeness
9 Common agreement over what is known creates a common foundation for idea generation. As parties agree on the underlying knowledge of their ideas, they are more likely to reinforce each other’s ideas rather than shoot them down. This moves the idea forward and allows it to be worked out into innovative products or services. Overall this leads to the following
hypothesis:
H3: The higher a firm’s shared vision, the higher it’s innovativeness. Open-mindedness and innovativeness
Open-mindedness is the firm’s ability to critically evaluate its daily operations and its
acceptance of new ideas (Sinkula et al., 1997). Open-mindedness is linked to the ability of the firm to engage in unlearning (Nystrom & Starbuck, 1984). Unlearning requires employees to proactively question long-held routines, assumptions, and beliefs (Sinkula et al., 1997). This is important because, as the inherently dynamic environment in which the firm operates changes, the mental modes that are built from processed information may no longer be valid (Day, 1994). Previous research finds that holding on to existing knowledge harms the firm’s ability to properly account for environmental changes and thus decreases the firm’s ability to predict the market (Schindehutte, Morris, & Kocak, 2008). Troy, Szymanski, & Rajan
Varadarajan (2001) find that open-mindedness increases the firm’s ability to convert available market information into new ideas.
10 H4: The higher a firm’s open-mindedness, the higher it’s innovativeness.
Intra-organizational knowledge sharing and innovativeness
Intra-organizational knowledge sharing is the extent to which new information is distributed throughout the firm (Calantone et al., 2002). This is different from shared vision in the sense that it is a shared learning direction and intra-organizational knowledge sharing is sharing learning outcomes. The process of sharing knowledge enables the firm to generate new ideas that are the basis for innovation (Powell, Koput, & Smith-Doer, 1994). Sharing newly acquired knowledge enhances a unit’s capacity to create novel linkages and associations (Jansen, Van Den Bosch, & Volberda, 2005). Also, as employees have access to more information, they become capable of efficiently utilizing their knowledge. Additionally, they start to evaluate and understand the nature and commercial potential of their advances (Cohen & Levinthal, 1990). Research finds that units are more likely to share the knowledge that is related to the existing knowledge in that unit and that they, as a result, are more likely to pursue explorative innovations (Jansen, van den Bosch, & Voldberda, 2006; Van Wijk, Jansen, & Lyles, 2008). Finally, the firm is more capable of maintaining knowledge in cases of employee turnover. If the knowledge possessed by employees is shared throughout the firm then the loss of insights is reduced in the case of their turnover (Moorman & Miner, 1998). Knowledge sharing within the firm increases the availability of knowledge throughout it. When employees have more access to knowledge there are more likely to make insightful interpretations about their environment, allowing them to properly recognize and evaluate business opportunities in the market, which enhances the firm’s innovative performance. Overall this leads to the following hypothesis:
H5: The higher a firm’s intra-organizational knowledge sharing, the higher it’s innovativeness.
11 Common and unique effect of the underlying LO values on innovation performance
The unique effect of a LO value describes its effect isolated from the other values. Common effects explain what part of the variance comes from the effect that is created by the values existing jointly. Common effects thus explain the overlap in the explanatory value of the LO values (Schlagel, Engle, & Lang, 2019).
Currently, the overall effect of LO is mostly used to measure its impact on innovation. Using this approach only the combined effects of the values are measured. Other researchers have argued and measured the effect of the underlying values on innovation (i.e. Calisir et al., 2013; Eshlaghy & Maatofi, 2011). What no research has done, is consider the specific unique effect of any of the LO values. Doing so encompasses accounting for the differences in the other values while measuring this value’s impact. Additionally, specific combinations explaining the common effect of two or more values on innovation also remains unexplored. While Sinkula et al (1997) explained the interdependence of commitment to learning, open-mindedness, and shared vision: “Shared vision is different from commitment to learning and open-mindedness in that it influences the direction of learning, whereas commitment and open-mindedness influence the intensity of learning. It is crucial to include both dimensions (direction and intensity) to build a comprehensive learning orientation construct that is in congruence with extant theory and practice” (Sinkula et al., 1997:309). No research has tried to entangle these relationships, some researchers have used specific combinations of the LO values to test their impact on firm innovativeness, for example, Yildirim (2018) explored the effects of open-mindedness and shared vision on innovation. However, because such research fails to account for the other LO values, their correlation still influences the regression results. A question that remains concerns the advantage that is gained from employing an incomplete LO strategy by using a combination of two or three values. More specifically the question is whether there are only unique effects to be gained from implementing the LO values, or whether the value from LO stems from the combined effect of its underlying values.
As these common effects have neither been explicitly hypothesized nor empirically tested and a current lack of theoretical and empirical rationale the following explorative research
question is formulated:
12
2.2 Hypotheses for the moderating effect on the relationship between LO and innovation
This research intends to go further into the LO innovativeness relationship. Research has reported varying effect sizes for the relationship between LO and innovation. While all effect sizes are positive, their broad range, from small (Cheng & Sheu, 2017; Huang & Li, 2017; Paladino, 2007; Xian, Sambasivan, & Abdullah, 2018) to very large (Camisón & Villar-López, 2011; Kiziloglu, 2015), opens up questions on this relationship. There has been very little research on the moderating effects of this relationship. Thus, the literature at this point does not know why some research finds significant correlations below 0,25 and others above 0,8. However, understanding how much chasing a LO will affect a firm’s innovation
performance is a key question in determining what orientation a firm will focus on (Grinstein, 2008). Therefore this paper will use meta-analysis to test several potential moderators to gain a better understanding of the relationship.
The moderating role of corruption
This research wants to shed light on the possible moderating effect of country-level variables on the LO innovation relationship. Hardly any research to date has attempted to test such a moderating effect to date. This is despite an early call to research cross-national data on LO and innovation (Calantone et al., 2002). Because meta-analysis combines datasets from different studies from different countries, it provides a great basis to start such an
investigation. Such a study could focus on varying national or regional effects such as culture or institutions. This research will look into the effects of national institutions and their effect on the LO and innovation relationship.
Institutions differ between and within countries. Institutions are ‘the rules of the game’, they dictate the conditions under which a firm operates (North, 1991). Institutions are grounded in local history and have different characteristics in every country. These institutions can heavily influence the way firms operate. The quality of the institutions is seen as a key determinant in economic welfare and is used to explain the fast differences between the ‘North’ and the ‘South’ in the economics literature (Levchenko, 2005).
13 activities, including innovation (Estrin et al., 2013). The decision for a firm to pursue an innovative opportunity depends on the value that the firm can capture from the innovation (Baker, Gedajlovic, & Lubatkin, 2005). When the local institution is prone to corruption, then the firm faces greatly increased risk that other parties in the value chain will use this
corruption to act opportunistically, and appropriate profits from the firm (Anokhin & Schulze, 2009).
Corruption taxes the firm for innovating because it opens the firm op to opportunistic
behavior. Therefore a firm will be less motivated to pursue new ideas. In such a situation, the value of processing information would be diminished, as innovation is unlikely whether the firm has good ideas or not. Overall this leads to the following hypothesis:
H6: the positive relationship between LO and firm innovativeness, will be stronger in
countries with low levels of corruption, compared to countries with high levels of corruption. The moderating role of industry
Differences between service and manufacturing firms are recognized in both LO and
innovation literature. While the difference is recognized by some researchers in LO literature (i.e. Awasthy & Gupta, 2011; Sadler-Smith, Spicer, & Chaston, 2001), research on comparing the two is very thin. For comparing innovation across industries a lot more research has been conducted. However, no research has examined the differences across industries on the impact of LO on innovation.
Sadler-Smith et al. (2001) found differences in the way manufacturing and service firms learn. They found that learning in manufacturing firms was more often active, higher-level learning, compared to passive, lower-level learning. For service firms, there was no difference between active and passive learning. Active learning encompasses generative, rather than adaptive (Senge, 1992), and transformational rather than incremental learning (DiBella, Nevis, & Gould, 1996). These differences indicate that active learning leads to new ideas and ways of thinking, while passive learning is about focussing knowledge on what is already known and being done. Thus, active learning is more likely to lead to innovations.
14 R&D to foster their innovation. They invest heavily to gather information from a wide variety of sources (Tether & Massini, 2007) to push their innovation levels. Service firms, on the other hand, focus on intangible products, which are often co-produced with clients (Gallouj & Weinstein, 1997). This means that manufacturing firms interact with a wider, more diverse set of sources, but they do so less intensively. If a firm with a strong LO is skilled at acquiring information (Garvin, 1993), then it will be able to gather information in a short amount of time, which other firms would only recognize after more intensive interaction. This would suggest that in manufacturing would miss out on information without a strong LO, that service firms would not miss.
Overall both the learning literature and the innovation literature point towards a more important role for LO in manufacturing firms. This leads to the following hypothesis: H7: the positive relationship between LO and firm innovativeness will be stronger in manufacturing firms, compared to service firms.
The moderating role of firm size
There is a difference between the process through which small and medium firms learn and the way large firms learn within the organization (Keskin, 2006). While there have been some researchers that have focussed on LO in SMEs (i.e. Amin, 2015; Keskin, 2006; Martinez, Serna, & Guzman, 2018), the literature on this topic is still thin (Frank & Kessler, 2012). Even where there is research on LO in SMEs the research generally measures firm
performance rather than innovativeness. However, some differences have been pointed out in this stream of research. Keskin (2006) suggests that in SMEs the learning attitude differs in terms of formalization and structure. In SMEs the learning happens more reactively than in large firms, often leading to situational changes (Chaston, Badger, Mangles, & Sadler-Smith, 2001). Such situational changes often fail to bring systematic changes in the way the firm works nor does it lead to insights for their products, as a result no real innovation occurs. Another difference is that SMEs are usually closer to their customers (Keskin, 2006). These firms collaborate with their customers, they naturally learn from them, lowering the necessity of LO to gain knowledge.
15 different contexts. Therefore, they develop a more diverse knowledge pool (Zhao, 2006). As knowledge is often hard to codify, there is an advantage gained when transferring it within the firm (Kogut & Zander, 1993). Large firms are more likely to have systematic information gathering processes in the firm (Farrell, 1999) which allows them to make better use of a LO. Lastly, large firms are in a better position to exploit innovations. As larger firms have larger production and marketing networks and operate on a larger scale (Cohen & Klepper, 1996), they have more to gain from innovation. Because the firm can take more advantage of
innovations, it will also be more willing to make the required investment. As a result, the firm will be capable and willing to try and convert gathered ideas from their LO into innovations. Overall this leads to the following hypothesis:
H8: the positive relationship between LO and firm innovativeness will be stronger in larger firms, compared to SMEs.
Figure 2: Complete conceptual model of LO – innovation performance
16
3 Methodology
The literature on the relationship between LO and innovation is fragmented in studies with varying attributes. As these studies show vastly different magnitudes for this relationship a consolidation of the literature is in order. Simply combining the studies would create biased results as the samples cannot be compared one to one. In an effort to combine these studies nonetheless a meta-analytic procedure can be used (Hunter & Schmidt, 2004). Meta-analysis is a quantitative technique that allows for comparability between studies (Ellis, 2010).
Thereby meta-analysis allows for summarization of the effect sizes, giving a better idea of the overall relationship between two variables. After that meta-analysis can estimate the
variability within the relationship, and test how much of this can be explained using different moderators (Ellis, 2010).
3.1 Literature search and sample
The first stage of a meta-analysis is the identification of relevant literature that is to be included in the analysis (Ellis, 2010). A literature search was conducted in the first phase of this research, during the first quarter of 2020. Google scholar was used for this literature search. Google Scholar provides matches from a wide range of sources, including
17
Table 1: Keyword search information
Variable Words used in the literature search
Learning orientation "Learning orientation" "LO"
"Organizational learning "Learning" "Shared vision" "Open-mindedness" "Commitment to learning"
"intra-organizational knowledge sharing"
Innovation "Innovation" "Innovativeness"
"New product development"
Note: search terms for learning orientation were always paired with a search term for innovation 3.2 Inclusion criteria and coding
18 publication, the sample size of the study, the country where the study was conducted, and the industry they researched, being manufacturing, service, or mixed. Then the study was coded for the firm sizes in the study, labels that were used were large, SME, or mixed. Large was taken when a paper focussed on ‘top firms’ in a specific category or when the descriptive statistics showed that only firms with more than 250 employees were in the sample, for SMEs only research stating their focus on SME or when descriptive statistics showed only the use of firms smaller than 250 employees. Finally, the studies were coded for their correlations and the reliability of their measures. For the reliability of the Cronbach Alphas of LO, its
underlying values, and innovation were taken. When a Cronbach Alpha was not given, it was computed based on the average of all other Alphas, as suggested by Lipsey & Wilson (2001). For the coding of the correlations between LO, its underlying values, and innovation Pearson correlation tables were used. When only the correlations between LO and innovation were not given, but between the underlying values and innovation were, the average of these was taken for the LO innovation correlation. Additionally, the country labels were used in combination with data from the World Governance Indicators database to assign the corresponding control of corruption score for that study, using the standard normal numbers from the database. This data captures the perceptions of the people on the extent to which corruption is exercised within the country, in accordance with its label, it portrays high corruption with a low score and vice versa.
19
Table 2: List of included studies
Study N Country Industry Firm size Included contructs
Calantone et al. 2002 187 US Mixed SML LO CL SV OM IKS
Camisón & Villar-López, 2011 159 Spain Mixed SML LO
Chen et al., 2009 325 China Mixed SML LO CL SV OM IKS
Cheng & Sheu, 2017 362 Taiwan Mixed SML LO
Chenuos & Maru, 2015 333 Kenya Service Large LO CL SV OM
Ebrahimi, Shafiee, Gholampour, &
Yousefi, 2018 193 Iran Mixed SME LO
Ejdys, 2015 115 Poland Mixed SME LO
Ejdys & Gedvilaite, 2017 169 Poland Service SML LO CL SV OM
Eshlaghy & Maatofi, 2011 82 Iran Service SML LO CL SV OM
Farrell, 1999 268 Mixed Mixed SME LO
Fraj, Matute, & Melero, 2015 356 Spain Mixed Large LO
Han, Hansen, Panwar, Hamner, &
Orozco, 2013 87 Mixed Service SML LO
Hao, Kasper, & Muehlbacher, 2012 161 Austria/China Manufacturing Large LO
Huang & Li, 2017 336 Taiwan Manufacturing Large LO
Kaya & Patton, 2011 135 Turkey Mixed Large LO
Keskin, 2006 157 Turkey Mixed SML LO CL SV OM
Kiziloglu, 2015 272 Turkey Mixed SME LO
Kocoglu, Imamoglu, & Ince, 2011 124 Turkey Service SML LO SV IKS
Li, Guo, Yi, & Liu, 2010 351 China Mixed SML LO
Lin, Luo, Ieromonachou, Rong, &
Huang, 2019 231 China Mixed SML LO
Lin, Peng, & Kao, 2008 333 Taiwan Manufacturing SML LO
Mahmoud et al., 2016 115 Ghana Mixed SML LO CL SV OM IKS
Martinez et al., 2018 400 Mexico Service SML LO CL SV OM
Ma’atoofi & Tajeddini, 2010 82 Iran Mixed SME LO CL SV OM
Nasution, Mavondo, Matanda, &
Ndubisi, 2011 231 Indonesia Mixed SME LO
Nybakk, 2012 241 Norway Service SML LO CL SV OM IKS
Onağ et al., 2014 143 Turkey Manufacturing SML LO OM IKS
Paladino, 2007 249 Mixed Mixed SML LO
Perin & Sampaio, 2003 170 Brazil Mixed Large LO CL SV OM
Sheng & Chien, 2016 200 Taiwan Manufacturing SML LO
Tajeddini, Altinay, & Ratten, 2017 178 Japan Mixed Large LO Tran, Nguyen, & Nguyen, 2018 267 Vietnam Service SML LO
Wang, 2008 213 UK Mixed SME LO CL SV OM
Xian et al., 2018 179 Malaysia Mixed SML LO
Yildirim, 2018 140 Turkey Manufacturing SML LO SV OM
Zehir & Wujiabudula, 2016 295 Turkey Manufacturing SML LO
Zhou, Hu, & Shi, 2015 287 China Manufacturing SML LO
20
3.3 Analytic procedures 3.3.1 Bivariate meta-analysis
Bivariate meta-analysis is used to draw conclusions about the significance, magnitude, and direction of the studied direct relationship. In this study, bivariate analysis is thus used to examine the direct relationship between LO, or its underlying values, and innovation. The most commonly used method for using bivariate analysis comes from Hunter and Schmidt (2004). In this approach, the data is corrected for measurement errors using Cronbach’s Alphas. These Alphas are used to convert the correlation coefficient into a more reliable measure, the reliability adjusted correlation coefficient (ρ) (Hunter & Schmidt, 2004). Besides indicating whether there is a direct relationship between the tested variables, the bivariate analysis also tests for heterogeneity in the effect sizes. Using both the Q-statistic and the I² this heterogeneity is measured. The Q-statistic is used to test the significance of the heterogeneity if the Q-statistic is significant the results indicate heterogeneity among groups, which means that a moderator analysis should be conducted. The I² measures this
heterogeneity. The measure ranges from 0 to 1 where the closer it is to 1, the greater the heterogeneity of the studied effect sizes is (Higgins & Thompson, 2002).
One of the most common biases faced in meta-analysis is publication bias (Borenstein, 2009). Publication bias comes from the fact that significant results are more likely to be published, while non-significant results are equally as insightful for meta-analysis (Ellis, 2010). While there is an attempt to incorporate unpublished work in this meta-analysis, the availability of such papers is limited, leaving the risk of publication bias nonetheless. Publication bias analysis can be conducted to signal this potential publication bias and, adjust the estimate for the combined effect size (Hak, van Rhee, & Suurmond, 2016). The analysis indicates if there are imputed studies with either negate (left side) or positive (right side) results. The analysis then calculates and corrects the effect size accordingly (Hak et al., 2016). This analysis however, only works for a homogenous set of results where the combined effect size can be interpreted as the true effect size of a population. When there is considerable heterogeneity among the effect sizes then the publication bias analysis results should not be interpreted (Hak et al., 2016).
3.3.2 Commonality analysis
21 analysis measures the impact of a set of variables using both unique and common effects (Mood, 1969). The unique effects measure the impact of one variable in isolation, the
common effects measure the additional impact of combining these variables for any possible combination. The commonality analysis approach is similar to multiple regression but solves issues of multicollinearity (Ray-Mukherjee et al., 2014). Just like in normal regression models commonality analysis provides an R² that explains the variance in the dependent variable that is explained by the independent variables in the model, however commonality analysis takes an additional step. Here the analysis breaks down the explained variance and attributes it to the unique effect of an independent variable, or the joint effect of any combination of the independent variables (Ray-Mukherjee et al., 2014).
3.3.3 Moderator analysis
The next step in the analysis is used to further explain the relationship between LO and innovation. When the bivariate analysis shows significant results but with a large remaining heterogeneity, it is desirable to test for moderator variables that can explain the difference in impact between two variables. To conduct the moderator analysis the potential moderator variables are identified and converted into subgroups when it are categorical variables. Then these subgroups are tested in a linear multiple regression analysis. With this analysis the relationship is tested for heterogeneity again, using the Q-statistic and the I² to see how much more of the relationship is explained with the addition of the moderator(s). In the process, more moderators are added stepwise in an attempt to lower the Q-statistic (Ellis, 2010). Meta-analysis can also run a regression to test moderating effects that aren’t categorical variables, such as the corruption scores for hypotheses 6. To test this hypothesis an additional
22
4 Results and analysis
In this section, the results of the meta-analysis are presented. First, the results from the bivariate meta-analysis are reviewed to draw conclusions about hypotheses 1 to 5. Then the commonality analysis is reported to find the results of research question 1. Finally, the moderator analysis is displayed to discuss hypotheses 6 to 8.
4.1 Results of the Bivariate meta-analysis
The bivariate analysis is used to study the direct relationship between LO and innovation. To do so the analysis calculates the sample weighted average correlation coefficient (r) and the sample weighted average reliability adjusted correlation coefficient (ρ). As ρis the more reliable measure, it is used for the interpretation. Further, the SD(ρ) (standard deviation of ρ) and 95% CI, for the 95% confidence interval of ρ is calculated to make conclusions about the significance of the relationship. Q and I² are used to measure the heterogeneity among studies. Finally, the results are tested for publication bias, indicating the number of imputed studies, their tail (whether the imputed studies have lower or higher than tested results), and the adjusted effect size. The results can be found in table 3.
The first hypothesis predicts a positive relationship between LO, and its underlying values, and innovation. For testing the direct relationship between overall LO and innovation 38 studies were used (K = 38), providing a combined sample size of 8376 observations (N=8376). Both the r and the ρ statistic show positive results, indicating a positive
relationship between LO and innovation (r=0,55, ρ=0,65). The results are also significant, allowing us to accept hypothesis 1 (SD(ρ)=0,49 and 95% CI= 0,55 – 0,76). The Q-statistic is 1934,35 and the I² is 98,09%, indicating a high degree of heterogeneity, which suggests that a moderator analysis is important to gain a better understanding of the relationship. As there is considerable heterogeneity in this sample, the publication bias is not interpreted (Hak et al., 2016).
23 To test shared vision 15 papers (K=15) were used, providing a combined sample size of 2888 (N=2888). The r and ρ are positive and show significant results (r= 0,43, ρ=0,54, SD(ρ)=0,2 and the Confidence interval is 0,44-0,63). Therefore we accept hypothesis 3. Here too there is high heterogeneity among results, with a Q-statistic of 116,48 and an I² of 87,98%, although it has the lowest heterogeneity of the tested variables.
Open-mindedness was tested using 15 papers (K=15), with a combined sample size of 2907 (N=2907). The r and ρ are positive and show significant results (r= 0,53, ρ=0,69, SD(ρ)=0,51 and the Confidence interval is 0,52-0,81). Therefore we accept hypothesis 4. Here too there is high heterogeneity among results, with a Q-statistic of 686,29 and an I² of 97,96%.
For the final value, intra-organizational knowledge sharing 8 papers (K=8) were identified, with a combined sample size of 1461 (N=1461). The r and ρ are positive and show significant results (r= 0,47, ρ=0.77, SD(ρ)=0,95 and the Confidence interval is 0,07-0,96). Therefore we accept hypothesis 5. Here too there is high heterogeneity among results, with a Q statistic of 1116.3 and an I² of 99,37%, warranting moderator analysis for all tested results. As the samples for all the underlying values of LO show considerable heterogeneity within the group, the publication bias is not interpreted for any of the values.
Table 3: Results of the bivariate meta-analysis
95%CI
Relationship K N r ρ SD(ρ) LL UL Q I² #TF Side ρTF
Learning orientation 38 8376 0,55 0,65 0,49 0,55 0,76 1934,35 98,09% 2 Left 0,67 Comitment to learning 13 2624 0,46 0,57 0,27 0,42 0,68 181,75 93,40% 5 Left 0,45 Shared vision 15 2888 0,43 0,54 0,2 0,44 0,63 116,48 87,98% 2 Left 0,55 Open-mindedness 15 2907 0,53 0,69 0,51 0,52 0,81 686,29 97,96% 5 Left 0,55 Intra-organizational
knowledge sharing 8 1461 0,47 0,77 0,95 0,07 0,96 1116,3 99,37% 0 Left 0,84
4.2 Results of the Commonality analysis
This paragraph analyses the results for the research question:
What are the unique and common effects of commitment to learning, open-mindedness, shared vision, and intra-organizational knowledge sharing on firm innovativeness?
The results of the commonality analysis are based on the meta-analytic correlations for this study, which are displayed in table 4. The meta-analytic correlation matrix is computed based on bivariate analysis of data gathered in this study. Table 5 shows the results of the
24 sharing is the most relevant for firm innovativeness (.617), followed by open-mindedness (0.231), commitment to learning (0.02), and shared vision (-0.072). Interesting to see is the negative relationship between shared vision and innovation. However, this should be interpreted within the correct context, shared vision does not necessarily hinder firm innovation, it only does so if its not aligned with the other values. Looking at the unique effects of the LO values it seems that intra-organizational knowledge sharing is the most influential value on its own with a moderate impact (11,6%), that open-mindedness has a small unique effect (1,3%) and that the unique effects of commitment to learning and shared vision are negligible (<1%). This indicates that only focussing on a single value, other than intra-organizational knowledge sharing, will not help the firm to make substantial
improvements to its innovativeness. When looking at the common effects the importance of commitment to learning, shared vision, and open-mindedness becomes clear. While in
combination with the other values intra-organizational knowledge sharing is still the strongest effect in explaining the variance (47,7%), followed by open-mindedness (46,3%). Here commitment to learning (32,5%) and shared vision (29%) are also relevant. Overall, the common effects account for 77,5% of the variance explained. This indicates that while on their own commitment to learning and shared vision have little impact on innovativeness, when the firm wants to realize the full potential of LO, it should also foster these values within the firm.
Table 4: Meta-analytic correlation table
FI CL SV OM IKS FI (0,85) 13,2624 15,2888 15,2907 8,1461 CL 0,57 (0,86) 12,2474 12,2474 6,1194 SV 0,54 0,7 (0,84) 12,2474 7,1318 OM 0,69 0,7 0,77 (0,83) 6,1194 IKS 0,77 0,71 0,68 0,81 (0,79)
25
Table 5: Regression results for the Commonality analysis
b Unique Common GenDom Pratt RLW
CL 0.020 0.000 0.325 0.090 0.012 0.096
SV -0.072 0.002 0.290 0.077 -0.039 0.078
OM 0.231 0.013 0.463 0.165 0.160 0.166
IKS 0.617 0.116 0.477 0.275 0.475 0.268
Total NA 0.131 1.555 0.607 0.608 0.608
Note: b = beta weight, Unique = unique effect, Common = absolute common effect, GenDom = general dominance, Pratt = Pratt’s measure, RLW = relative weights.
To get a deeper understanding of the common effects of LO values and their combined effects we look to the result details in table 6. In this table, the common effect of any combination of the four values is reported. While the unique effects are the same as in table 5, this table also provides the percentage of explained variance. Looking at the combinations of two values, only the explanatory value of combining intra-organizational knowledge sharing with open-mindedness appears relevant, as it explains 18,9% of the variance (with a cumulative effect of 40,1%), while other combinations all account for less than 3%. In pairs of three not a lot of extra explanatory power is added, the combined effect of the triple combinations accounts for about 15% of the variance, the combinations including both commitment to learning and SV explain even less than 1%. What stands out is the cumulative importance of
intra-organizational knowledge sharing, all combinations including intra-intra-organizational knowledge sharing account for 97.5% of the variance. This does not mean that the other values are unimportant, as 41.6% of the variance is explained by a combination of all four values. This reinforces the earlier statement that firms should invest in all four LO values, but this analysis also indicates that firms without a LO focus can still improve their innovativeness by
26
Table 6: Details of the Commonality analysis
Commonality % Total CL 0.000 0.000 0.325 SV 0.002 0.003 0.292 OM 0.013 0.022 0.476 IKS 0.116 0.190 0.593 CL,SV 0.000 0.000 0.364 CL,OM 0.001 0.001 0.491 SV,OM -0.002 -0.003 0.476 CL,IKS 0.015 0.025 0.594 SV,IKS -0.001 -0.001 0.593 OM,IKS 0.115 0.189 0.606 CL,SV,OM 0.001 0.001 0.492 CL,SV,IKS -0.001 -0.001 0.594 CL,OM,IKS 0.056 0.092 0.606 SV,OM,IKS 0.040 0.065 0.607 CL,SV,OM,IKS 0.253 0.416 0.608 Total 0.608 1.000 NA
Note: Commonality = absolute common effect, % = percentage of R2.
4.3 results of the moderator analysis
In this section, the moderating hypotheses 6 to 8 are tested using subgroup analysis. With the subgroup analysis, the data is analyzed for statistical differences between separated data sets. If they are statistically different, they indicate evidence for a moderating effect. The results of the subgroup analysis are displayed in tables 8 to 10. There is a significant effect for the moderator when the 95% confidence intervals of the groups have no overlap or if the p-value of the Q*-statistic is significant (<0,05). Additionally, for hypothesis 6 a moderator regression is used, these results are displayed in table 7. For the moderator analysis significance is based on the p-value of the beta.
Tables 7 and 8 show the results for the moderating effect of control of corruption on the LO innovativeness relationship. Looking at the moderator regression, the results indicate a significant negative relationship (B = -0,14, P = 0,000). Looking at the subgroup analysis the results are also negative, but come out insignificant (p = 0,271). As the moderator analysis uses more detailed data, this is leading in drawing conclusions. Thus, the data suggests that there is a significant positive moderating effect for country-level corruption on the
relationship between LO and firm innovativeness. To confirm this an additional robustness check is done by calculating the regression and subgroup analysis for the underlying values. The regression results for the underlying values also indicate a significant negative
intra-27 organizational knowledge sharing shows insignificant results (p = 0,092). This robustness check mostly supports the conclusion of a significant negative relationship. Regardless, hypothesis 5 is rejected.
Tables 9 and 10 show the results for hypotheses 7 and 8. The results show insignificant moderating effects between LO and innovation for industry (p = 0,723) and firm size (p = 0,258). Thus, there is no support for either hypothesis 7 or 8. To check for the robustness of hypothesis 7 the effect was also calculated for the LO values, such a check could not be done for firm size as insufficient data was collected on large firms. Over the robustness check, the results for commitment to learning remain insignificant, but for shared vision and open-mindedness the results come out significant, no test was run for intra-organizational knowledge sharing as there was insufficient data. The results from the robustness check indicate that in service firms the values shared vision and open-mindedness have a stronger influence on innovation than in manufacturing firms, which is against the prediction of hypothesis 7. As these results are based on only 5 studies, interpreting them should be done with caution. Overall both hypotheses 7 and 8 are rejected.
One of the goals of analysis is to find homogeneous groups. For such groups, meta-analysis allows us to make inferences about their true effect size. A subgroup is considered homogeneous when the Q value is lower than the threshold of K – 1 (Ellis, 2010). None of the subgroups that measure the overall effect of LO fulfill this requirement. While there are 2 subgroups based on underlying values (namely shared vision in institutions with high control of corruption and open-mindedness in manufacturing industries), that hold requirement for homogeneity, these results are based on relatively small sample sizes, which is why drawing conclusions based does not make sense. This means overall this research does not draw conclusions about the true effect size of LO on innovation.
Table 7: Moderator regression of control of corruption
28
Table 8: Subgroup analysis for control of corruption
29
Table 9: Subgroup analysis for industry
K N ρ CI LL CI UL Q p I² weight T² T PI LL PI UL Q* p Industry - LO 16 3302 0,65 0,58 0,71 1124,11 98,67% 0,37 0,61 0,58 0,71 0,13 0,723 Manufacturing 8 1504 0,63 0,34 0,81 292,96 0 97,61% 68,18% 0,22 0,47 -0,42 0,96 Service 8 1798 0,69 0,28 0,89 817,4 0 99,14% 31,82% 0,54 0,73 -0,75 0,99 Industry - CL 4 695 0,42 -0,12 0,76 56,48 0 94,69% 0,11 0,33 -0,35 0,85 1,51 0,22 Manufacturing 2 411 0,34 -0,65 0,9 3,1 0,078 67,75% 79,10% 0,01 0,1 -0,88 0,97 Service 2 284 0,66 -1 1 31,83 0 96,86% 20,90% 0,23 0,48 -1 1 Industry - SV 5 835 0,42 0,05 0,69 17,15 0,002 76,68% 0,02 0,14 -0,2 0,8 6,73 0,01 Manufacturing 3 551 0,31 0,22 0,4 0,54 0,764 0% 56,01% 0 0 0,22 0,4 Service 2 284 0,54 -0,59 0,96 2,8 0,095 64,22% 43,99% 0,01 0,12 -0,87 0,99 Industry - OM 5 835 0,57 0,11 0,83 31,28 0 87,21% 0,04 0,21 -0,27 0,92 22,6 0 Manufacturing 3 551 0,43 0,22 0,6 3,13 0,209 36,18% 50,27% 0 0,06 0,12 0,66 Service 2 284 0,69 0,06 0,93 1,05 0,305 4,99% 49,73% 0 0,02 0,02 0,93
Table 10: Subgroup analysis for firm size
K N ρ CI LL CI UL Q p I² weight T² T PI LL PI UL Q* p
Firms size - LO 14 3464 0,52 0,49 0,71 243,35 0 94,25% 0,07 0,27 0,51 0,71 1,28 0,258 Large 8 1950 0,55 0,32 0,72 173,92 0 95,98% 38,33% 0,1 0,32 -0,18 0,89
30
5 Discussion and conclusions
Since Sinkula et al.’s (1997) introduction of the concept of LO, there has been a considerable amount of contributions to this research topic. While most research focussed on the
relationship between LO and firm performance, other researchers examined the effect of LO on firm innovativeness. The goal of this analysis was to (1) synthesize the existing empirical literature on LO and innovation and to analytically cumulate the correlations, (2) to
disentangle the effect of the LO values on firm innovativeness, and (3) to explore the
moderating effects that impact this relationship to gain a better understanding of its true effect size under different conditions. By analyzing data from 38 studies (N= 8376) this research has aimed to contribute to the literature on these topics. In this section, the implications that can be drawn from this paper’s results are discussed, as well as its limitations.
5.1 Implications for theory
The first contribution this study makes is that it is the first meta-analysis to combine the LO literature to examine how LO is related to firm innovativeness. By consolidating 38 studies, this study provides evidence to push the literature beyond any doubt on the significance of this relationship. The findings provide evidence for a positive relationship between LO and firm innovativeness (ρ =0,55). This is consistent with earlier findings in the literature but reinforces the importance of learning for a firm’s innovative capabilities directly, rather than solely as a moderator for the relationship between other strategic orientations and innovation (i.e. Hakala, 2011; Nasution, Mavondo, Matanda, & Ndubisi, 2011).
31 combination accounts for 42,6% of the explained variance. Imputing one of the values in an analysis will, therefore, result in a considerable loss of explanatory power for the model. Here too, intra-organizational knowledge sharing is the most important value in understanding the effect of LO on innovation, as it contributes to a cumulative variance of 97.5%. Studies such as Wang (2008), which use measures for LO that does not include intra-organizational knowledge sharing, should interpret their results with caution, as such results are highly influenced by a variable in the error term.
A third contribution is made by analyzing the extent to which a country’s control of
corruption influences the relationship between LO and innovativeness. Such an exploration contributes to the suggestion of Calantone et al. (2002) for cross-national studies in
researching LO. The results suggest that corruption levels in a country significantly impact the impact of LO on firm innovation, so that the effect is stronger in countries with higher levels of corruption. The results were not robust when simply comparing high corruption vs low corruption, suggesting that the extent of corruption matters, more than merely the presence of corruption. The most important contribution this provides to the literature is that research on the relationship between LO and innovativeness should not be generalized across institutions. The effect of corruption is only one of the possible country-level effects that could impact the impact of LO on innovativeness, research should further explore other country-level effects, such as culture or other institution characteristics, as well.
Fourth, this research explored the moderating effects of firms size and industry on the
relationship between LO and innovativeness. The results suggest that there are no significant differences between groups based on these variables. This shows that learning is important for firms of any size and in any industry in fostering innovation. This implies generalizability for this relationship across different industries and firm sizes. For researchers, this means that they can pay less attention to these topics in their sample selection, as there should be no, or little effect on their results.
5.2 Implications for practice
32 This research finds that all LO values have a positive impact on innovation. Particularly, the results suggest that intra-organizational knowledge sharing is the most important LO value for fostering innovation. This shows the importance of making gathered knowledge available throughout the firm, as it increases the ability of the firm to gather innovative ideas. However, this research also finds that all LO values play an important role in its effect on innovation. A substantial part of the impact of LO can only be explained by combining all four values. Managers should carefully plan their investments in learning, as such investments will be most beneficial if the values are balanced properly. Here too, the importance of
intra-organizational knowledge sharing should be stressed, as without it LO has almost no impact on the firm’s innovative capabilities.
5.3 Limitations
To properly interpret the findings of this meta-analysis several limitations need to be addressed. First, the number of studies identified in this meta-analysis is limited. Especially for intra-organizational knowledge sharing, the data is scarce (8 studies included in the bivariate analysis). As meta-analysis often has smaller sample sizes, identifying moderators can be difficult. This also happened in this paper, as subgroups often did not exceed 8 studies, and it was at times not possible to do robustness checks. When fewer observations are
available this increases the confidence intervals to the point where they are too large for significant results (Aguinis, Gottfredson, & Wright, 2011).
Second, the potential of a search bias that has to be considered. Two main problems arise here. Firstly, there is a possible language bias, as only English papers were used in this analysis. This might result in a more limited availability of samples from non-English speaking countries, which can impact the results, as is suggested by the results of this study regarding national differences (Ellis, 2010). Second, there is the possibility of publication bias. While an attempt to include unpublished papers as well was made, the limited
33 variables, more empirical data to confirm this is desirable.
Lastly, while the used literature stems from studies conducted in 18 different countries, and with some papers using a mixed, yet unreported set of countries, the dominance of studies conducted in Turkey must be recognized. 8 studies (21% of the sample), providing a
combined sample size of 1416 (17% of the sample) were conducted in Turkey. This is a form of selection bias, which could cause some unobserved variables that are over or
under-represented in Turkey to influence the results of this study (Coe, 2009). 5.4 Future research
This research found a positive moderating effect for country-level corruption on the
relationship between LO and innovation. This finding goes against the predictions from the literature. Currently, the theory behind this finding is therefore still unclear, further research could be conducted to understand why LO has a larger impact on innovation in countries with high corruption.
The finding that country-level corruption affects the relationship between LO and innovation suggests national differences in the relationship. As previously suggested by Calantone et al. (2002), cross-national research could provide new insights. While this research is a step in that direction, more research should be conducted, especially now that this research indicates that such differences exist. Further research could focus on more institutional differences, such as property rights protection, which is closely linked to innovation. But also cultural effects could be of interest.
While the theory heavily suggests a causal relationship between LO and innovativeness, hardly any data tested this. Only 1 study employing longitude data was used for this study. As a result, some questions regarding the causality should be considered. Future research should examine the relationship with longitude data to confirm the causal relationship between LO and innovativeness.
34
5.5 Conclusions
Overall this study confirms the positive relationship between LO and firm innovation. More importantly, this research enriched the understanding of how LO influences innovation. From the analysis, it is clear that LO as a construct should be viewed as a sum of the values
35
References
Aguinis, H., Gottfredson, R. K., & Wright, T. A. 2011. Best-practice recommendations for estimating interaction effects using meta-analysis. Journal of Organizational Behavior, 32: 1033–1043.
Amin, M. 2015. The effect of entrepreneurship orientation and learning orientation on SMEs’ performance: an SEM-PLS approach. J. for International Business and
Entrepreneurship Development, 8(3): 215.
Anokhin, S., & Schulze, W. S. 2009. Entrepreneurship, innovation, and corruption. Journal
of Business Venturing, 24(5): 465–476.
Argyris, C., & Schön, D. A. 1978. Organizational learning: a theory of action perspective. Massachusetts: Addison-Wesley publishing.
Awasthy, R., & Gupta, R. K. 2011. Is learning orientation in manufacturing and service firms different in India? Learning Organization, 18(5): 406–422.
Baker, T., Gedajlovic, E., & Lubatkin, M. 2005. A framework for comparing
entrepreneurship processes across nations. Journal of International Business Studies, 36(5): 492–504.
Baker, W. E., & Sinkula, J. M. 1999. Learning Orientation, Market Orientation, and
Innovation: Integrating and Extending Models of Organizational Performance. Journal
of Market-Focused Management, 4(4): 295–308.
Barney, J., Wright, M., & Ketchen, D. J. 2001. The resource-based view of the firm: Ten years after 1991. Journal of Management, 27(6): 625–641.
Barrett, H., Balloun, J. L., & Weinstein, A. 2005. The impact of creativity on performance in non-profits. Intf. Nonprofit Volunt. Sec, 10: 213–223.
Boden, M., & Miles, I. 2000. Are services special? Services, Innovation and the Knowledge
Economy: 1–81.
Borenstein, M. 2009. Introduction to Meta-Analysis. Chichester (UK): Wiley.
