An exploratory study on the relationship between learning orientation, dynamic capabilities and business model innovation

(1)

An Exploratory Study on the

Relationship Between Learning

Orientation, Dynamic Capabilities

and Business Model Innovation

MSc. Business Administration - Strategy

Christoffer Sandholt Worsøe 11376945

Master Thesis - Final Version Submitted: 23 June, 2017

(2)

2

Statement of originality

This document is written by Christoffer Sandholt Worsøe who declares to take full responsibility for the contents of this document. I declare that the text and the work presented in this document is original and that no sources other than those mentioned in the text and its references have been used in creating it. The Faculty of Economics and Business is responsible solely for the supervision of completion of the work, not for the contents.

(3)

3

Abstract

Business model innovation has increased its popularity as research topic and within management practice over the recent years. Business model innovation is often presented as being a source of competitive advantage (Amit & Zott, 2010). There is still uncertainty regarding the antecedents of business model innovation. This study examines whether the relationship between learning orientation and business model innovation is dependent on dynamic capabilities. A survey targeting managers with sufficient knowledge about their firms’ business model obtained 52 responses. Partial least squares structural equation modelling (PLS-SEM) is applied to analyze the data in order estimate the relationship between the different constructs.

The study finds that the relationship between learning orientation and business model innovation is fully mediated by dynamic capabilities. This expand the findings from previous papers that highlights learning and experimentation as possible antecedents of business model innovation (Chesbrough, 2010; McGrath, 2010). The study contributes to strategic management literature and provide managers with new insights on how to succeed in achieving business model innovation.

Tags: Business model innovation, learning orientation, dynamic capabilities, PLS-SEM, strategic orientation

(4)

4

Acknowledgements

(5)

5

1 Introduction

The research on business models and business model innovation has been increasing its momentum within academia over the recent years (Foss & Saebi, 2017; Wirtz, Pistoia, Ullrich, & Göttel, 2016; Zott, Amit, & Massa, 2011). Search results from Scopus shows that there has been published 7,244 journal articles about business models and 299 articles on business model innovation since 20001_{. The increased academic interest is also reflected in} business practice. A global CEO study in 2014 from Boston Consulting Group reports that 94% of the 1500 surveyed companies had engaged in business model innovation to some extent (Lindgardt & Ayers, 2014).

A business model is defined as “the design of transaction content, structure and

governance as to create value through the exploitation of business opportunities” (Amit & Zott, 2001, p. 511). Thus, a business model can be seen as the blueprint for how firms create and capture value. As per the definition, the business model is the structure around the business opportunity for capturing value, thus making it imperative to have resources and capabilities to bring goods to market through the business model. Worth noting is the enabling part, since there is a highly complementary relationship between the firm’s offering and the business model. It is not enough to have great technology or products, firms need to wrap it in a business model to create and capture the value (Chesbrough,

1_{Source: Scopus, 2000–2016. “Business model” (BM), 7.244 hits; “business model innovation” (BMI),}

299 hits. Scopus searched for the terms in the search field “abstract, title, keyword” with “journals” as source and “article” as type (April 2017).

(7)

7 2010; Johnson, Christensen, & Kagermann, 2008).

On an abstract level, business model innovation is defined as “[the process of] designing a new, or modifying the firm’s extant activity system” (Amit & Zott, 2010, p. 2). In an

attempt to ground the term, business model innovation can be articulated as novel changes to a firm’s way of doing business. The changes can affect a single component of the value chain, but since components are often interlinked, changing one element is likely to affect the remaining value chain (Zott & Amit, 2008). Pursuing innovation of the business model thus entails “(re) deployment and usage of existing resources and capabilities to develop new value offerings or forms of value creation” (Schneider & Spieth, 2013, p. 4). Business model innovation hold the potential for unlocking value mechanisms: (i) capture the innovation of extended value chains, (ii) support the activation of overlooked value sources within the company via combination with external knowledge bases and (iii) lead to

networked architectures that are difficult to imitate (Zott & Amit, 2008).

Teece, Pisano, & Shuen (1997) defines two distinct approaches to how competitive

advantage is obtained in a broad perspective. One being that rents flow from a privileged market position that can be obtained through specific actions (Porter, 1985), while the other emphasizes firm-specific assets and capabilities (Barney, 1991; Teece et al., 1997). Both approaches are strategies, whereas business model innovation is the possible outcome of deliberate strategic choices (Casadesus-Masanell & Ricart, 2010a). However, business model innovation encompasses both elements of product and resources. Business models have previously been identified as complementary to product market strategies, in such that they are not mutually exclusive, but can lead to competitive advantage when combined properly (Zott & Amit, 2008). More explicitly, it has been stated that continuous

(8)

8 business model innovation is a possible source of competitive advantage, since it can

overpower already established positions and advantages (Mitchell & Coles, 2003).

As the literature review will explicate, there is still uncertainty in terms of understanding the internal drivers that are antecedent to business model innovation. When

understanding the antecedents, it becomes possible to predict what will foster business model innovation and therefore ultimately competitive advantage.

When initiating business model innovation, a firm must be able to realize how its existing business model is working, in order to assess whether the value of the changes to the business model is significant enough to justify the cost of innovation (Johnson et al., 2008). Previous research indicates that managers can define the current business model (Chesbrough, 2010). However, its innovation is hindered due to conflicts with existing assets and routines, as well as limited abilities in terms of understanding these barriers to innovation and the potential value from new business models (Chesbrough, 2010).

Business model innovation can be deployed to both defend and challenge; as a response to competition and as a way to seize new opportunities (Lindgardt & Ayers, 2014). Therefore, business model innovation can either be seen as either a responsive or proactive process. The latter is represented in learning orientation, since it embodies the degree to which firms are committed to systematically challenging the fundamental beliefs and practices that define the innovation process itself (Baker & Sinkula, 1999a, p. 296). Previous studies have found that learning orientation has a positive relation on firm innovativeness which in turn positively affects the firm performance (Calantone, Tamer Cavusgil, & Zhao, 2002).

(9)

9 Firm innovativeness is defined as the rate of adoptions and willingness to change, thus providing the basis for speculation with regards to a relationship between learning

orientation and business model innovation as well, since innovative firms are expected to be able to achieve business model innovation. However, current literature on learning orientation has yet to answer how the obtained knowledge is transformed into actual innovation of the business model.

Learning orientation increases the accumulation of knowledge, which is then perceived as an intangible asset that possibly can be implemented in order to achieve business model innovation. Deploying knowledge is within the realm of dynamic capabilities, since it enable firms to create, deploy, and protect the intangible assets that support superior long run business performance through specific processes (Teece, 2007, p. 1319). Successful deployment of intangible assets, in this case the accumulated knowledge, is expected to have a positive relation with business model innovation. This leads to the following research question:

What is the relationship between learning orientation and business model innovation, depending on dynamic capabilities?

As of now, it seems that most literature has been concerned with the relationship between organizational learning and innovation, but not the underlying factors contributing to the application of learning in the organization. On the other hand, literature on dynamic

(10)

10 but ceases when it comes to how the knowledge is obtained. This study contributes to prior literature by combining the streams of literature on learning orientation, business model innovation and dynamic capabilities to seek an explanation for the relationship between all three factors in order to innovate the business model.

The paper will progress with a review of existing literature upon which hypotheses will be formulated. The method section includes description of the sample, data collection, the chosen measurement models and the statistical method applied. The analysis part describes the analysis of the date and the results. The results are obtained using partial least squares structural equation modelling (PLS-SEM), which includes the evaluation of two models: the measurement model and the structural model. Finally, the findings are summarized in the discussion which also includes implications, limitations, and

(11)

11

2 Theory and hypotheses

2.1 Business model

The notion of business models is several decades old dating back to 1957 (Bellman, Clark, Malcolm, Craft, & Ricciardi, 1957). In past years, the business model as a concept has received significant attention from both academics and practitioners, especially with increasing interest since 2000 (Foss & Saebi, 2017; Zott et al., 2011). The literature stream for business models and business model innovation has been analyzed using the Scopus database, which is the largest database for peer-reviewed literature (Scopus, 2017). There has been published 7.244 articles about business models and 299 articles on business model innovation since 20002_{, indicating that business model innovation is still a novel} offspring from the primary business model literature stream. Foss & Saebi (2017) suggests comparing the amount of published business model literature with literature published on open innovation and dynamic capabilities, since they are seemingly both related and popular topics. It becomes clear that business model literature far exceeds the two other subjects with ~5,700 articles respectively.

The literature review will start with a focus on business model literature followed by business model innovation. It will then identify similarities between business model innovation literature and learning orientation. Finally, literature on dynamic capabilities

2_{Source: Scopus, 2000–2016. “Business model” (BM), 7.244 hits; “business model innovation” (BMI),}

299 hits; “open innovation” (OI), 1.456 hits; “dynamic capability” (DC), 1.533 hits. Scopus searched for the terms in the search field “abstract, title, keyword” with “journals” as source and “article” as type (April 2017).

(12)

12 will be reviewed.

Despite the increasing interest in the field, there is still no established field of economic theory focusing on business models: “The absence of consideration of business models in economic theory probably stems from the ubiquity of theoretical constructs that have markets solving the problems that – in the real world – business models are created to solve” (Teece, 2010, p. 175). As so, there seems to be a conflict between the economic theory and the more practical explorations of business models. In their literature review of business models Zott & et al. (2011 p. 1023) finds three main reasons for the emergence of the concept of business models, which eventually can help bridge the gap: (i) E-business and the use of information technology; (ii) strategic issues such as value creation,

competitive advantage, and firm performance; (iii) and innovation and technology

management. The advent of online retailing in the early 2000s was followed by a focus on the accompanying business models that were introduced, which underscored the fact that it was not enough for firms to create value; as stated in economic theory, they need to capture the value as well. A business model is said to transcend firm borders and enable firms to create and capture value in cooperation. This allows for a more relaxed

understanding of firm interaction, since it places the concept of a business model in the middle between a perfect market on one hand and transaction cost economics wherein all hazards must be safeguarded through contracts on the other hand (Williamson, 1979). Taken together, the two papers seemingly try to bridge the gap between economic theory and literature on business models. Teece (2010) does this very explicitly, when stating the purpose of his papers, whereas Zott et al. (2011) advances business model literature by highlighting the importance of its primary streams.

(13)

13 Value creation mechanisms in business models often go beyond Schumpeterian innovation [that combines resources in new ways], reconfigurations of the value chain (Porter, 1985), the formation of strategic networks among firms, or the exploitation of firms’ specific core competencies (Zott et al., 2011, p. 1029). Hence, the unit of analysis for value creation spans across firm and industry boundaries. Companies must develop new business models which potentially can include suppliers, partners, distribution channels, and coalitions that extend the company’s resources in order to survive in the current economic environment (Hamel, 2000). Developing new business models that transcend the firm’s boundaries, positions the firm to create and capture value in a value network, which broadens the scope and opportunities for profit beyond the confines of a single entity. The opportunities for novel combinations and offerings increase exponentially when the firm is no longer limited to its own capabilities, while it enables the firm to focus its business and leverage its core capabilities (Prahalad & Hamel, 1990). One such example is how successful Apple’s iPhone has been, due to, but not exclusively, because it integrated app developers into the business model with the App Store. This expanded the amount of available applications for the iPhone way beyond any competition, because Apple opened its eco-system to suppliers of applications.

One thing is to realize that business models can involve interfirm collaboration, another important aspect is that firms are executing their business models in competition with peers (Hamel, 2000). It seems logic that firms are competing, but it has implications for how business models are theorized and the impact they can have. It is not just a simple descriptive blueprint for how the firm is doing business, the business model itself is a mode of competition. The business model is the manifestation of the chosen strategy for

(14)

14 how the firm will compete in the marketplace under the current circumstances (Casadesus-Masanell & Ricart, 2011). The business model is therefore the observable part of the

strategy at a given point in time. However, the strategy may contain several scenarios that are dependent on contingencies, which could lead to changes in the business model. Markides & Charitou (2004) follows the close link between strategy and business model and introduces the notion of superior business models, given that a nuanced contingency approach will lead to superior performance, if the right business model is chosen in

accordance with both internal and external perspectives. This allows the business model to become a source of competitive advantage.

Successful business models represents a better way of doing business than the existing alternatives, since most business models are different variations over the same underlying value chain (Magretta, 2002). This argument is further supported given that if a firm can outperform competitors by reworking the value chain, then it indicates that novelties in business models can lead to superior value creation (Morris, Schindehutte, & Allen, 2005; Zott & Amit, 2007). Zott & Amit (2008) empirically study the effect of novelty-centered business models, which refers to new ways of conducting economic exchanges among various participants, and the combination with different product market strategies. Hence, novelty-centered business models are the outcome of business model innovation. If firms get stuck in between different product market strategies, such as differentiation and cost leadership, they can expect inferior performance (Porter, 1985). On the contrary, when firms combine seemingly different product market strategies and business models, i.e. cost leadership and a novelty-centered business model, it will have a positive effect on firm performance. This indicates that the concepts are complements rather than substitutes

(15)

15 (Zott & Amit, 2008).

Given the applicability of business models within both academia and management

practice, several conceptualizations seeks to explain the different components (Casadesus-Masanell & Zhu, 2013; Johnson et al., 2008; Osterwalder & Pigneur, 2010; Teece, 2010; Zott & Amit, 2008). Two recent literature reviews (Foss & Saebi, 2017; Wirtz et al., 2016) conclude that most definitions are converging on defining a business model as the “design or architecture of the value creation, delivery, and capture mechanisms” of a firm (Teece, 2010: 172). Amit & Zott (2001) adds a crucial aspect to the business model definition, by emphasizing the importance of transactions. The addition of transactions indicates that value is seldom created in vacuum, and especially that exchange must occur for firms to capture value. The business model consists of “the design of transaction content, structure and governance as to create value through the exploitation of business opportunities” (Amit & Zott, 2001, p. 511). Transaction content refers to the goods or information that are exchanged, and to the resources and capabilities that are required to enable the exchange. Transaction structures refers to the parties that participate in the exchange. Transaction governance are the ways flows of information, resources, and goods are controlled by the relevant parties (Amit & Zott, 2001, p. 511). The core logic of a business model revolves around a firm’s revenues and costs, its value proposition to the customer, and the

mechanisms to capture value. Thus, the business model can both be a vehicle for bringing new innovations to market with a compelling logic to customers, as well as a subject of innovation itself when reconfigured (Zott et al., 2011).

(16)

16

2.2 Business model innovation

Business model innovation is defined as “designed, novel, nontrivial changes to the key elements of a firm’s business and/or the architecture linking these elements” (Foss & Saebi, 2017, p. 201). This definition helps separate business models from both strategy and tactics. The designed aspect is a consequence of the strategy, which provides a

comprehensive basis for future decisions. The nontrivial aspects indicate an important distinction from tactics, which are residual choices bound by the business model

(Casadesus-Masanell & Ricart, 2010b). Consider a newspaper based on an ad-sponsored revenue model – adjusting their ad prices would be a trivial tactic choice, whereas

changing their price is out of scope. Changing their price would significantly affect their transaction structure, thus invoking a nontrivial change to the business model.

That the business model can be subject to innovation itself and that business model innovation leads to superior performance was first proposed by D. Mitchell & Coles (2003). Johnson et al. (2008) argues that the customer value proposition should be the focal point for a business model. This guides the focus of innovation, so the firm innovates any given part of its business model that it expects to create most value for the customer. Hence, business model innovation is key to firm performance, as it allows firms to respond to competition in the most suitable way to meet customer demands (Demil & Lecocq, 2010; Johnson et al., 2008). By meeting customer demands, rather than meeting internal ambitions for growth, business models allow industry entrants to challenge incumbent firms, which has proven to be successful numerous time (Casadesus-Masanell & Zhu, 2013). When entrants challenge incumbent firms with a novel business model, they are going to market with a new way of doing business. This does not necessarily imply any

(17)

17 radical new product offerings, but possibly just a reconfiguration of the value chain, to better reflect the priorities within the existing value network. This notion of going to market with a reconfiguration of the value chain goes back to Schumpeter's (1934) theory of how the entrepreneur destabilizes a market by seeing and exploiting gaps. This was the case when Ryanair decided to deploy a no-frills business model and eliminated all

unnecessary elements to provide cheap tickets, which met the demand of cost conscious consumers (Casadesus-Masanell & Zhu, 2013). The job being done, flying customers from A to B, was however ultimately still the same.

Empirical observations suggest that incumbents actually often learn about new business models from entrants, and that their typical response is to implement parts of the new business model into their existing business model (Casadesus-Masanell & Zhu, 2013). Established firms that innovate their business models experience positive performance effects, however, there is no significant difference between innovation of industry models, revenue models or enterprise models (Giesen, Berman, Bell, & Blitz, 2007). Innovation of the industry model is understood as changes in the industry value chain, such as when Apple start delivering music directly to consumers through iTunes. Innovations of how companies generate revenue, is exemplified by the infamous razor blade model of Gillette, where the razor is cheap and the disposal blades are expensive. This creates a lock-in of the user. On an interesting side note, the same model was employed for years by Polaroid with a relatively cheap camera and expensive polaroid film. Eventually they almost went out of business, when the revenue model was turned upside down with free digital pictures and expensive DSLR cameras (Tripsas & Gavetti, 2000). Finally, the innovation of enterprise models is related to the role the firm plays in new or existing value chains. A great example

(18)

18 is when Hilti decided to forward integrate their value chain and started renting their tools to the clients instead of selling them (Johnson et al., 2008). Hilti realized that their tools are only means to an end. The contractor is not interested in owning the tools, but getting the job done. Hence, Hilti changed their focus to deliver the right tools when needed. Innovation of industry model and enterprise model seems rather similar, but when

illustrated with the previous examples it becomes more clear. Apple decided to branch into the music industry, whereas Hilti stayed in the same industry and extended their

enterprise.

2.3 Antecedents of business model innovation

When managers are able to recognize their business models, it allows them to take the first step towards changing it (Chesbrough, 2010). When recognizing their business model managers can compare their current level of value generation with the possibilities for value that business model innovation might result in (Johnson et al., 2008). This

underscores the need for a clear definition of a business model for managers to gain a solid understanding of the current way of doing business, the elements they are possibly

changing, and the derived consequences hereof.

As a consequence of innovative competitors and entrants, managers must constantly pay attention to external changes and how it affects their internal resources, in order to adjust their business model accordingly (Demil & Lecocq, 2010, p. 242). Managers with a strong emphasis on paying attention to external changes can be said to be heralding a specific leadership agenda if they have a specific goal, that they motivate the rest of the

(19)

19 (2010) however only focuses on the leader’s individual agenda and the urge to motivate the employees to follow the same direction which ultimately led to business model innovation. The logic seems to be along the lines of strategic orientation, which can be said to be an organizational manifestation of the leadership agenda. Strategic orientation is defined as “strategic directions implemented by a firm in order to create the proper behaviors for the continuous superior performance of the business” (Narver & Slater, 1990).

Designing a new business model requires several bodies of information including customer, competitor and supplier information to understand the evolving reality that impacts customers and society (Teece, 2010). However, no one can fully understand such environment, hence most business model innovations are provisional in the sense that they are likely to be replaced over time by improvements when new and better information is obtained (ibid.). Firms that pursue business model innovation, must engage in

significant experimentation and learning (McGrath, 2010; Smith, Binns, & Tushman, 2010; Sosna, Trevinyo-Rodríguez, & Velamuri, 2010). Given that changes in the external environment cannot be fully anticipated, and initiators of business model innovation are not able to rationalize and articulate the business model fully ex-ante, experimentation and learning is likely to be required, hence allowing for framing of business model innovation as an ongoing learning process (Chanal & Caron-Fasan, 2010; Sosna et al., 2010).

Entrepreneurs and managers are at best making informed guesses about future behavior and demand, thus making it imperative to learn fast and implement changes to the business model immediately when the underlying assumptions change. Since business models can be easily imitated (Casadesus-Masanell & Zhu, 2013), firms should expect to be going through cycles of business model innovation in order to maintain their temporal

(20)

20 competitive advantage (McGrath, 2010). The cycles of business model innovation are linked to temporary competitive advantage, which posits that firms must exploit their advantage until the competition has caught up and then innovate their business model again to stay head.

2.4 Learning orientation and business model innovation

Companies must engage in experimentation to innovate the business model, but before doing so they should obtain knowledge, in other words learn, about the environment. This logic links to strategic orientation, since it is “the elements of the organizations culture that guide interactions with the marketplace” (Noble, Sinha, & Kumar, 2002, p. 25). There are several distinct types of strategic orientation with market orientation being recognized as one of the most common types (ibid.). However, there are alternatives that encompass different foci, one of them being learning orientation.

Learning orientation embodies the degree to which firms are committed to systematically challenging the fundamentals beliefs and practices that define the innovation process itself (Baker & Sinkula, 2002). This enables organization to achieve double-loop learning (Baker & Sinkula, 1999a). Double loop learning consists of two loops, where the first loop uses an established goal to guide the decision making, whereas the second loop questions the goal and the underlying rationale based on feedback obtained when executing the first loop. This coincides with the how business model innovation is based on continuous

experimentation and learning as proposed by McGrath (2010). However, this paper will focus on the outcome, rather than the underlying processes of learning in business model innovation.

(21)

21 Learning orientation refers to “organization-wide activity of creating and using knowledge to enhance competitive advantage” (Calantone et al., 2002, p. 516). In the study of

Calantone et al. (2002) learning orientation consists of four subcomponents: (i) commitment to learning, (ii) open-mindedness, (iii) intraorganizational knowledge sharing, (iv) and shared vision. Calantone et al. (2002) found a positive relationship between learning orientation and firm innovativeness which then affects firm

performance. Firm innovativeness is defined as the rate of adoption of innovations by the firm and the willingness to change (Calantone et al., 2002). Business model innovation can be a sub component of firm innovativeness, since the term includes adoption of

innovation, however the innovation studied by Calantone et al. (2002) is product innovation. New products that are successfully implemented is often aligned with the dominant logic within the organization (Leonard-Barton, 1992). Products that are at odds with the existing logic is often separated into new entities, to be encompassed in a new business model (Chesbrough & Rosenbloom, 2002). Learning orientation has therefore been linked to certain types of innovation, but there is still a gap in the literature when it comes to estimating the relationship between learning orientation and business model innovation.

Without explicitly exploring the impact of learning orientation, a longitudinal study was carried out in 2010 that focused on how a Spanish company could innovate its existing business model (Sosna et al., 2010). One of the findings was that the “CEO took a learning approach, experimenting with change, sharing his ambitions and acquired knowledge with his team, and constantly adapting the model based on the observation of market feedback to organizational actions” (Sosna et al., 2010, p. 385). The finding shares several

(22)

22 commonalities with the definition of learning orientation such as the emphasis on open-mindedness in the form of experimentation, sharing his vision and the knowledge with the team and finally the commitment to learning. Smith et al. (2010) explores which complex business models that can encompass the paradoxical strategy that is needed to thrive in a dynamic environment. Among their results is both building commitment to an overarching vision and active learning at multiple levels, which shares similarities with learning

orientation.

2.5 Dynamic capabilities and business model innovation

Learning orientation does not explain which mechanisms that are required within organizations to utilize the acquired information. There seems to be a gap between the intentions of obtaining knowledge as an outcome of the learning orientation and the realized innovation of the business model. This is also reflected in the business model innovation literature, since there are still significant gaps in the understanding of the internal drivers of business model innovation (Foss & Saebi, 2017). Foss & Saebi (2017) propose that dynamic capabilities can provide insights into the internal antecedents of business model innovation, since dynamic capabilities are indicative of the attention and readiness to take action across the corporate hierarchy, as a response to major changes in the external environment.

Dynamic capabilities are defined as the firm’s “ability to integrate, build, and reconfigure internal and external competences to address rapidly changing environments” (Teece et al., 1997, p. 516). The definition echoes both the notion of business model innovation as a response to competition as well as the boundary spanning nature of business models since

(23)

23 it includes both internal and external competences (Amit & Zott, 2001). The authors

further state that “dynamic capabilities thus reflect an organization’s ability to achieve new and innovative forms of competitive advantage” (Teece et al., 1997, p. 516), which highly resembles the definition of business model innovation provided by Amit & Zott (2010). The shared logic between dynamic capabilities and business model innovation is even more apparent when taking Newbert’s (2008) review of the resource-based view into account, which states that “competitive advantage is made possible when firms find novel ways in which to combine those resources and capabilities to which they do have access” (Newbert, 2008, p. 761). Along the lines of dynamic capabilities is also combinative capabilities which are organizational processes by which firms synthesize and acquire knowledge resources, and generate new applications from those resources (Kogut & Udo, 1992).

The capacity to reconfigure and transform is itself a learned organizational skill (Teece et al., 1997). This statement gives way to hypothesizing that learning and information alone is not sufficient to achieve business model innovation. The organization must develop

capabilities over time to be able to utilize the information as a basis for business model innovation.

Microfoundations are important elements of dynamic capabilities. They are the distinct skills, processes, procedures, organizational structures, decisions rules, and disciplines that all goes into the capacities of sensing, sensing and reconfiguring (Teece, 2007). The selection and design of business models is a key microfoundation of dynamic capabilities (Teece, 2010, p. 190). Dynamic capabilities take various forms, however in the study related to business models, it is composed of sensing, seizing and reconfiguring skills

(24)

24 which enable firms to create, deploy and protect the intangible assets that support superior long-term business performance (Teece, 2007). Sensing is defined as analytical systems and individual capacities for sensing, filtering and learning about opportunities. Seizing is defined as the firm’s structures, procedures, designs and incentives for seizing

opportunities. Finally, reconfiguring is the capacity to change tangible and intangible aspects of the business, so that they are aligned with the new opportunities. When combined, they can be seen as a process to orchestrate the firm’s assets (Teece, 2007). Firms with strong dynamic capabilities can adapt to business ecosystems and shape the environment through collaboration with other firms (ibid.).

The ability to recognize opportunities depends on the learning and knowledge capacities of the organization. The tasks for recognizing opportunities include learning about external technological developments and customer needs (Teece, 2007). This is highly congruent with learning orientation (Baker & Sinkula, 1999a; Calantone et al., 2002). However, since dynamic capabilities includes obtaining knowledge, but also extends itself to include seizing and reconfiguring, it is hypothesized that dynamic capabilities mediate the relationship between learning orientation and business model innovation.

Hypothesis 1: Learning orientation is positively associated with dynamic capabilities (H1a), and dynamic capabilities, in turn, is positively associated with business model innovation (H1b).

(25)

25 Figure 1: Conceptual model

3 Method

3.1 Sample

Given that the unit of analysis is the organization and the study seeks to explore the

antecedents of business model innovation, a survey has been distributed among executives and managers that are expected to be knowledgeable about the business model. Since there is no sample frame, there will be used three types of non-probability sampling: purposive sampling, convenience sampling and snowball sampling. Purposive sampling is a natural consequence of that not everyone within the company is knowledgeable enough to provide answers that are useful for the survey. Convenience sampling is used since there

Dynamic capabilities Sensing Seizing Reconfiguring Learning orientation Commitment to learning Shared vision Open-mindedness Intra organizational knowledge sharing Business Model Innovation H1a (+) H1b(+)

(26)

26 is limited access to firms, so the respondents will be composed of already established

professional contacts. Finally snowball sampling is used to increase the number of respondents by encouraging people to share the survey with whom they find relevant.

Following the key informant approach as outlined by Phillips (1981), which states that the researcher should seek out respondents that are in a position to provide insightful answers on the topic, the following criteria from Kortmann et al. (2014) were applied: management level, involvement in strategic and innovation activities, job title, job experience, and organizational tenure. Respondents that scored lower than five on a 7 point Likert scale, indicating low involvement in strategic and innovation-related activities were excluded from the sample. Titles that indicated low involvement such as graduate, intern, and

associate were removed from the sample as well. All incomplete surveys were left out of the final sample. This results in a final sample of 52 respondents.

(27)

27 Table 1. Descriptive statistics

3.1 Data collection

To obtain answers an online survey was created using the online survey tool Qualtrics. Online surveys offer among several advantages automated data entry, ensuring higher accuracy since there is no room for error as when digitizing answers from paper surveys (Boyer, Olson, Calantone, & Jackson, 2002). Online surveys have fewer missing responses,

Key information descriptive statistics Firm descriptive statistics

Job title Firm size (number of full time

employees)

CEO 7 1-10 8

CIO 3 11-50 10

CMO 1 51-250 10

CFO 1 251-1000 9

Senior Vice President 2 1001-10,000 4

Head of Department 4 >10,000 11

Director 8 AVG (SD)

Business Unit Manager 7 Firm age 35.81 (42.01)

Manager 11 (in years)

Partner 3

Founder 3

Executive Assistant 2

Total respondents 52

Involvement in… AVG (SD)

… strategic decision making 5.84 (1.44)

… innovation decision making 5.88 (1.23)

… operational decision making 6.17 (1.05)

Organizational tenure (in years) 4.41 (4.06) Overall work experience (in years) 12.28 (10.30) Notes: AVG = Average; SD = Standard Deviation.

(28)

28 since Qualtrics has the option of showing an error message if a question is left blank, thus ensuring all answers are completed.

The survey was distributed using personalized emails and postings on Facebook and LinkedIn. Personalized emails were chosen as the main medium for contacting respondents, since previous research indicates that direct contacts yield the highest response rate when compared to mass communication (Punch, 2003). To further encourage responses, all participants were offered an executive summary of the survey upon completion of the survey. Following the logic of Weber’s law for just noticeable difference (Britt & Nelson, 1976), the chance of winning a gift card with any financially reasonable value for the author, would not make any difference in the motivation for senior level executives to answer the survey.

3.3 Measures

All measures of the survey are adopted from previous studies, which have been validated. There are 58 questions in total which cover learning orientation, dynamic capabilities and business model innovation. Control variables includes firm size, firm age, and

environmental dynamism using the measurement model from Jansen, Bosch, & Volberda (2006). The Likert scales operationalized in the original studies for each model were applied, thus creating a variance in the answer format to avoid “straight lining” and conformity of the responses (Maronick, 2009). In addition, the independent and

dependent variables are placed in individual sections. Taken together, these precautions should reduce the common method bias from self-reported data (Podsakoff, MacKenzie,

(29)

29 Lee, & Podsakoff, 2003). See Appendix A for full list of measures.

The measurement items developed by Calantone et al. (2002) were used for learning orientation. Learning orientation is conceptualized as a formative second order construct composed of four first-order indicators: Commitment to learning, shared vision, open-mindedness, and intraorganizational knowledge sharing which all have 4 indicators except for intraorganizational knowledge sharing which has 5 indicators. A 7 point Likert scale grounded at strongly disagree = 1; strongly agree = 7 is used.

For business model innovation, the measurement items developed by Zott & Amit (2008) will be used, since the measurement model for novelty-based business models, translates to business model innovation since it refers to “new ways of conducting economic

exchanges among various participants” (Zott & Amit, 2008, p. 4). Business model

innovation is measured with 13 indicators, however one indicator is excluded, since the question is about business model related patents which is only applicable to USA. A 4 point Likert scale coded as strongly agree = 1; agree = 0,75; disagree = 0,25; strongly disagree = 0 is used. Previous research on the use of Likert scales without a neutral middle point seems at best inconclusive about the effects of absence of a neutral middle point (Garland, 1991; Norman, 2010). Furthermore the scale meets the criteria about equidistance and symmetry which applies to scales used in PLS-SEM (Hair, Hult, Ringle, & Sarstedt, 2017), hence the original scale from Zott & Amit (2008) is used unaltered.

(30)

30 As a mediator, it will be tested whether the relationship between the two constructs are dependent on dynamic capabilities. The construct of dynamic capabilities is divided into three parts: sensing, seizing and reconfiguring (Teece, 2007). Teece’s (1997;2007) theory has been adopted into a measurement model developed by Wilden, Gudergan, Nielsen, & Lings (2013). Just as with learning orientation, it is conceptualized as a

reflective-formative hierarchical component model. The model has 12 indicators equally divided between the 3 lower order constructs of sensing, seizing and reconfiguration, which then formatively composes the higher order construct of dynamic capabilities. A 7 point Likert scale grounded at rarely = 1; very often = 7 is used to obtain answers.

Firm age and size will be added as control variables, especially since the findings of Calantone et al., (2002) suggests that the older an organization, the stronger the

relationship between learning orientation and firm innovativeness. This goes against the findings of managerial inertia, which states that managers over time develops a bias towards certain value networks, which then can inhibit their ability to innovate (Tripsas & Gavetti, 2000). On the contrary, older firms have had longer time to accumulate resources and routines (Barney, 1991; Eisenhardt & Martin, 2000). As for firm size, smaller firms are expected to be more strategically and operationally flexible than larger firms (Park & Luo, 2001). The final control variable is environmental dynamism, since previous studies has found that it affects the prioritization of exploitative and exploratory behavior (Jansen et al., 2006). Exploratory behavior could possibly be associated with business model

(31)

31 During the survey, each construct has been explained in both an academic and practical manner in order to increase the accuracy of the answers. This was especially important, since the survey did not include any options for skipping any answers.

3.4 Data analysis method

To comprehend the complex relationship between the different measures, it is proposed to apply sophisticated multivariate data analysis (Hair et al., 2017). The method is

furthermore suitable for analyzing data obtained through surveys (Hair, Sarstedt, Pieper, & Ringle, 2012). Multivariate analysis applies statistical methods that simultaneously analyze several multiple variables. The methods referred to as structural equation modelling (SEM), allows researcher to incorporate unobservable variables measured indirectly by indicator variables (Hair et al., 2017, p. 4). There are two types of SEM: (i) covariance-based SEM (CB-SEM), which is applied to confirm or reject theory and (ii) partial least squares SEM (PLS-SEM) which is primarily used to develop theory in exploratory research. PLS-SEM does so by focusing on explaining the variance in the dependent variables when examining the model (Hair et al., 2017).

There are several arguments for the fit of PLS-SEM for this study. Firstly, the goal is to identify key “driver” constructs, in this case learning orientation mediated by dynamic capabilities as driver for business model innovation. Learning orientation and dynamic capabilities are reflective-formative measured constructs, thus rendering the structural model complex with several constructs and indicators. PLS-SEM is well suited for analyzing conceptual models containing mediation effects and second-order constructs (Kortmann et al., 2014). The sample size is furthermore be relatively small and possibly

(32)

32 non-normally distributed, which PLS-SEM handles well (W. W. Chin, Marcolin, &

Newsted, 2003). And finally, different measurement scales will be used, since the items from Wirden et al. (2013) and Calantone et al. (2002) is measured on a 7 point Likert scale, whereas Zott & Amit (2008) uses a 4 point Likert scale. SmartPLS 3.0 (C. Ringle, Wende, & Becker, 2015) is considered to be the most advanced and updated software tool (Hair et al., 2017), thus making it the chosen application for this study.

While being a regression-based approach, PLS-SEM is nonparametric in nature (Hair et al., 2017, p. 87). Thus, it makes no assumptions regarding the distribution of the data. This is important when testing the model for significance of the model. Bootstrapping will be used to achieve a distribution upon which the significance will then be tested.

4 Analysis and results

4.1 Evaluation of measurement model

Data analysis in PLS-SEM can be seen as an examination of the path model, which consists of the measurement model and the structural model. The measurement model contains the indicators and their relationship with the constructs. The structural model represents the theoretical element of the path model, since it includes the latent variables and their path relationships. Constructs are either defined as reflective or formative. If a construct is reflective, then will the indicators represent the construct. The causality is “going” from the construct to its indicators, hence making the indicators interchangeable. The opposite is true for formative constructs, since the indicators fully form the construct and the causality is “going” from the indicators to the construct, so changing or removing the indicators would change the meaning of the construct.

(33)

33 The inclusion of learning orientation and dynamic capabilities makes the model a

reflective-formative hierarchical component model (HCM), thus a two-stage HCM analysis will be applied as suggested by Hair et al. (2017). Previous studies find that the best results are obtained, in terms of reliability and consistency with both theory and parameter

estimates, when using the repeated indicators approach with mode B on higher-order components and inner path weighting scheme (Becker, Klein, & Wetzels, 2012). The analysis is therefore split in two stages. The repeated indicators approach is used to obtain the latent variable scores for the order component. In the second stage, the lower-order component scores serves as manifest variables in the higher-lower-order competent measurement model (Wetzels, Odekerken-Schröder, & van Oppen, 2009).

The measurement model is evaluated using several measures as proposed by Hair et al (2017). To assess internal consistency reliability, the constructs are evaluated on

Cronbach’s alpha and composite reliability. To assess convergent validity average variance extracted (AVE) is applied. Since the lower order construct are reflective, the outer

loadings are examined to test for indicator reliability. For a sufficient level of item reliability, the outer loadings should exceed 0.7 (Hair et al., 2017). All indicators scoring below 0.4 is removed, whereas indicator between 0.4 and 0.7 only should be removed when it affects the AVE. See Table 1 for all outer loadings that are contained for further analyses. The following items were deleted: BMI7, BMI8, BMI9, BMI10, BMI11, BMI12, IntrKnow1, IntrKnow5, Omind3, Seiz1, Seiz4, Sens4, EnDyn4.

(34)

34 Since firm age and size are single indicator items, they are not evaluated, given that the indicator perfectly explains the latent variable.

Table 2 – Individual item reliability (Outer Loadings) 1 2 3 4 5 6 7 8 9 1. Business Model Innovation

BMI1 0.76 BMI2 0.72 BMI3 0.73 BMI4 0.73 BMI5 0.74 BMI6 0.83 BMI13 0.77 2. Commitment to Learning ComLearn1 0.84 ComLearn2 0.88 ComLearn3 0.83 ComLearn4 0.79

3. Intraorganizational Knowledge Sharing

IntrKnow2 0.72 IntrKnow3 0.85 IntrKnow4 0.85 4. Open-mindedness OMind1 0.75 OMind2 0.74 OMind4 0.70 5. Reconfiguring ReConf1 0.84 ReConf2 0.69 ReConf3 0.79 ReConf4 0.79 6. Seizing Seiz2 0.82 Seiz3 0.72 7. Sensing Sens1 0.77 Sens2 0.67 Sens3 0.89 8. Shared Vision ShrdVis1 0.76 ShrdVis2 0.75 ShrdVis3 0.87 ShrdVis4 0.78 9. Environmental Dynamism EnDyn1 _0.73 EnDyn2 0.82 EnDyn3 0.72 EnDyn5 _0.88

(35)

35 Four lower order constructs do not meet the sufficient levels for Cronbach’s Alpha (>0.70), however all construct meet the sufficient levels for composite reliability (>0.70) and

average variance extracted (>0.50). There is good reason for the insufficient levels of CA, since in a PLS-SEM setting Cronbach’s Alpha falls short compared to CR when predicting the internal consistency reliability, since it assumes that all indicators are equally reliable, whereas PLS-SEM prioritizes the indicators with regards to their individual reliability (Hair et al., 2017). CR takes such differences into account and is therefore more appropriate to evaluate measures upon.

Table 3 – Internal consistency reliability and convergent validity

Higher order construct Lower order construct CA CR AVE

N/A BMI 0.87 0.90 0.56

Learning orientation

Commitment to learning 0.85 0.90 0.70

Intra-organizational knowledge sharing 0.68 0.82 0.60

Open-mindedness 0.52 0.76 0.51 Shared Vision 0.79 0.87 0.62 Dynamic Capabilities Reconfiguring 0.78 0.86 0.61 Seizing 0.28 0.73 0.58 Sensing 0.62 0.79 0.56

N/A Environmental Dynamism 0.81 0.87 0.62

Note: CA= Cronbach’s Alpha; CR=Composite reliability; AVE = Average variance extracted CA and CR bold when >0.7; AVE bold when >0.5

Discriminant validity is the extent to which a construct is distinct from other constructs by empirical standards, which is typically measured with cross-loadings and the Fornell-Larcker criterion. As for Fornell-Fornell-Larcker criterion, the construct’s AVE2_{should be higher} than with any other construct, in order to establish discriminant validity (Hair et al., 2017). Recent research has however pointed out that neither cross loadings or Fornell-Larcker criterion fully detects discriminant validity issues (Henseler, Ringle, & Sarstedt, 2015). The

(36)

36 Fornell-Larcker criterion has issues detecting discriminant validity when all indicator loadings only differ slightly, for example between 0.60 and 0.80, which is the case in this study. Instead Heterotrait-Monotrait Ratio (HTMT) is proposed as an alternative. It estimates the disattenuated correlation and if it is close to 1 between constructs, it indicates a lack of discriminant validity. Previous research suggests a threshold value of 0.90 if the constructs in the path model are conceptually similar and 0.85 for conceptually distinct constructs (Henseler et al., 2015). Since PLS-SEM does not rely on any

assumptions about distribution, bootstrapping must be used to obtain a distribution of the HTMT statistic to test for significance (Hair et al., 2017).

Table 4 –Discriminant validity (Fornell-Larcker criterion)

1 2 3 4 5 6 7 8 9 10 11 1. BMI 0.75 2. ComLearn 0.32 0.83 3. EnDyn 0.32 0.25 0.79 4. Firm Age -0.18 -0.99 -0.28 1.00 5. Firm Size 0.28 -0.92 0.14 0.69 1.00 6. IntrKnow 0.35 0.47 0.15 -0.14 -0.72 0.78 7. OMind 0.31 0.49 0.21 -0.13 -0.18 0.53 0.71 8. Reconf 0.46 0.29 0.30 0.33 0.17 0.30 0.33 0.78 9. Seiz 0.36 0.44 -0.67 -0.47 -0.14 0.52 0.42 0.46 0.76 10. Sens 0.42 0.30 0.79 0.54 0.24 0.19 1.00 0.11 0.49 0.75 11. ShrdVis 0.43 0.56 0.17 -0.92 0.19 0.69 0.54 0.42 0.58 0.33 0.79

(37)

37 Table 5 – Heterotrait-Monotrait Ratio (HTMT)

1 2 3 4 5 6 7 8 9 10 11 1. BMI 2. ComLearn 0.35 3. EnDyn 0.35 0.26 4. Firm Age 0.17 0.19 0.24 5. Firm Size 0.32 0.12 0.20 0.69 6. IntrKnow 0.42 0.56 0.25 0.16 0.12 7. OMind 0.52 0.72 0.33 0.18 0.25 0.82 8. Reconf 0.54 0.27 0.18 0.78 0.12 0.36 0.50 9. Seiz 0.72 0.83 0.34 0.47 0.66 0.83 0.90 0.78 10. Sens 0.57 0.42 0.16 0.11 0.36 0.26 0.31 0.17 0.83 11. ShrdVis 0.57 0.68 0.21 0.20 0.15 0.87 0.83 0.54 0.89 0.48

The results indicate that discriminant validity is present between the constructs using both Fornell-Larcker criterion and HTMT. Bias corrected HTMT confidence intervals was obtained using the bootstrap method with 5000 subsamples. The results confirm the discriminant validity of the constructs at a 0.05 significance level, since neither the upper or lower bound includes 1. All model evaluation criteria have been met, providing support for the measures reliability and validity.

4.2 Evaluation of the structural model

The structural model measures the relationship between the variables. That is between learning orientation and dynamic capabilities, learning orientation and BMI, and finally dynamic capabilities and BMI. In this section, the following criteria are being evaluated: Path coefficients, determination coefficient (R2_{), effect size (Cohen f}2_{) and predictive} accuracy (Q2₎

(38)

38 To ensure that the predictor variables do not show critical levels of collinearity they should have a VIF value below 5. This is clearly the case since the highest obtained value is 2.4. See Appendix B for full table.

To evaluate the prediction power of the research model, the coefficient of determination (R2_{) of the dependent variables is being determined. Values of 0.75 are substantial, while} values of 0.50 and 0.25 are moderate or weak respectively (Hair et al., 2017). The results show a R2_{value of 0.349 for dynamic capabilities and 0.481 for BMI. Based on these values} it can be concluded that the research model has a weak to moderate prediction power for the constructs.

The effect size of the predictor constructs LO and DC on the dependent variable BMI is evaluated using Cohen’s f2_{(Cohen, 1992). The guidelines suggest that values of 0.02, 0.15,} and 0.35 respectively represent small, medium, and large effects (Cohen, 1992). The values for LO-DC (f2_{0.504) and DC (f}2_{0.231) indicates a large and medium effect respectively.} All control variables exhibit insignificant or no effect.

A two-tailed test with 1% significance level was computed for both t- and p-values,

following previous strategic management literature applying PLS-SEM (Kortmann et al., 2014; Wilden et al., 2013). For such tests the t-values should be larger than 2.57 which is the case for LO-DC and DC-BMI, but not for LO-BMI. P-values should be lower than 0.01 to ensure significance, which is the case for both LO-DC and DC-BMI, but not for LO-BMI. Thus, the relationship between LO-DC and DC-BMI is found to be significant, whereas the

(39)

39 relationship between LO-BMI is not. The hypothesis is therefore confirmed. Since the direct relationship between LO and BMI is not significant, but the indirect relationship mediated by DC is, it is concluded that there is indirect-only mediation (full effect) as defined by Zhao, Lynch, & Chen (2010). None of the control variables have a significant relationship with the predictor variables. Thus, hypothesis H1a and H1b is supported.

Table 6 – Path coefficients

Relationship Original Sample Sample Mean Standard Deviation T Statistics P Values

LO à DC 0.594 0.589 0.092 6,479 0.001

DC à BMI 0.429 0.416 0.155 2,770 0.006

LO à BMI 0.168 0.178 0.135 1,242 0.214

EnDyn à BMI 0.164 0.164 0.131 1,256 0.209

EnDyn à DC -0.168 -0.176 0.168 1,001 0.317

Firm age à BMI -0.239 -0.251 0.159 1,501 0.133

Firm age à DC -0.113 -0.116 0.213 0.529 0.597

Firm size à BMI 0.383 0.403 0.169 2,261 0.024

Firm size à DC 0.269 0.271 0.223 1,206 0.228

In addition to R2_{values, it is suggested to examine Stone-Geissers Q}2_{value (Geisser, 1974;} Hair et al., 2017). Whereas the R2_{value is measuring the in-sample predictive power, the} Q2_{is indicating the out-of-sample predictive power which is the predictive relevance of the} model (ibid.). If a PLS path model has predictive relevance, it is able to accurately predict data not used in the model estimation (Hair et al., 2017). The Q2_{value is obtained by using} the blindfolding procedure, that omits every dth data point from the endogenous

construct’s indicators and estimates the parameters with the remaining data points (W. Chin, 1998). It is recommended to choose an omission distance D between 5 and 10,

(40)

40 however D should not yield an integer when divided with the sample size (C. M. Ringle, Sarstedt, & Straub, 2012). Thus, an omission distance D of 5 is chosen. Prediction by means of the cross-validated redundancy is chosen due to its perfect coherence with the PLS-SEM approach, since it includes path model estimates of both structural models (Hair et al., 2017). Q2_{values above 0 suggests that the model has predictive relevance for a given} endogenous construct, which is the case for both BMI (0.372) and DC (0.245). Just as f2 measures the effect size for R2_{, q}2_{assesses the relative impact of predictive relevance. The} formula for q2_{is as follows:}

𝑞" ₌ 𝑄%&'()*+*" − 𝑄+-'()*+*"

1 − 𝑄_%&'()*+*"

As the formula indicates, q2_{is found by calculating the PLS-SEM results of the model with} LO to obtain Q2included and thereafter run the model without LO to obtain Q2excluded. Thus,

q2LO_àDC = (0.245 – (- 0.074)) / (1 – 0.245) = 0.423. And, q2DC_àBMI = (0.372 – 0.255) / (1 – 0.372) = 0.186. The same thresholds as for R2_{applies to q}2_{, hence the effect size of the} relationship between LO and DC is strong, whereas the effect size of the relationship between DC and BMI is medium.

Table 7 – Effect size (Cohen’s F2_{) and predictive relevance (q}2₎

Relationship F2 _q2 DC - BMI 0.23 0.186 LO - DC 0.50 0.423 LO - BMI 0.03 EnDyn - BMI 0.04 EnDyn - DC 0.04

Firm age - BMI 0.05

Firm age - DC 0.01

Firm size - BMI 0.12

(41)

41 Figure 2 – Structural model as in SmartPLS 3.0

Dynamic capabilities R2_0.35 Learning orientation Business Model Innovation R2_0.48 0.59*** 0.42*** Firm age Firm Size Environmental Dynamism 0.16 -0.11 -0.24 0.269 0.383 -0.17 0.16 Significant path Non-significant path * P ≤ .10 ** P ≤ .05 *** P ≤ .01

(42)

42

5 Discussion

5.1 Learning orientation and business model innovation

Prior literature has found a positive relationship between learning orientation and both firm innovation capabilities and product innovation (Baker & Sinkula, 1999a, 2002; Calantone et al., 2002). This study finds a positive relationship between learning

orientation and business model innovation, however it is not significant (β=1.242; p=.214). Hence, the analysis underscores the complexity in linking learning orientation to business model innovation. The result is possibly a consequence of the small sample size (n=52). Due to its insignificant relationship, the interpretation of the result will not go into further detail. However, the study shows that increases in learning orientation can lead to

increases in business model innovation, since dynamic capabilities fully mediate the relationship between the two constructs. The study is therefore advancing the

understanding of antecedents to business model innovation, by highlighting the role of dynamic capabilities in achieving business model innovation.

5.2 The mediating role of dynamic capabilities

When combining dynamic capabilities with learning orientation, dynamic capabilities are not only positively influenced by learning orientation (β0.59; p=.001), dynamic

capabilities also has a positive relationship with business model innovation (β0.42; p=.006). When stimulated by learning orientation, dynamic capabilities enable firms to innovate their business model. When there is a strong degree of learning orientation, a firm is able to obtain information and distribute it throughout the organization (Baker & Sinkula, 1999b; Calantone et al., 2002), thus providing a fertile ground for the routines

(43)

43 and processes embedded within dynamic capabilities that enables a firm to deploy the knowledge (Teece, 2007). This result also makes it interesting to discuss the non-significant relationship between environmental dynamism and dynamic capabilities (β0.16; p=0.22). One could expect that firms with a history of surviving in a dynamic environment would develop certain capabilities over time that allowed them to innovate their business model.

When assessing the structural model, it is evident that the R2_{value of business model} innovation is only .481, which means that only 48% of the construct can be explained by its predictors, thus leaving about half of the variance unexplained. This is likely to be related to the sample size, since there is clear relationship between the models ability to detect R2 values at specific levels and the given sample size (Hair et al., 2017). It also indicates that other predictors should be explored in future research, to fully understand the antecedents of business model innovation.

6 Implications

This study holds multiple implications for theory. First and foremost, it is contributing to the literature on antecedents of business model innovation. Previous studies by McGrath (2010) proposed that learning from experimentation was antecedent to business model innovation, which is now expanded in such way that proactive learning, embodied in learning orientation, is antecedent to business model innovation when combined with dynamic capabilities.

(44)

44 The study is also advancing the understanding of dynamic capabilities as an enabler of business model innovation, which was already theorized by Teece (2007). Furthermore, by placing dynamic capabilities as mediator there is potentially even more antecedents that can be linked to business model innovation. There is also the risk that dynamic capabilities become a universal enabler in studies, due to its inherent nature of deploying and

transforming intangible assets in order to create value. Despite the risk, future research could ground dynamic capabilities as a permanent mediator and then test which

antecedents has the strongest relationship with dynamic capabilities and in turn business model innovation.

Finally, the study is advancing an understanding of business model innovation being different from firm innovativeness. Business model innovation was suspected to be an outcome of firm innovativeness, however since the study did not find a significant relationship between learning orientation and business model innovation as done by Calantone et al. (2002), there is room to speculate what creates the discrepancy.

The most significant value might be present in the light of managerial implications. The study provides mangers with a possible roadmap for business model innovation. First step is to foster learning orientation in the organization. Baker & Sinkula (1999b) propose that the starting point is to question the current beliefs and method. By decomposing learning orientation as Calantone et al. (2002), the components become clear: commitment to learning, shared vision, open-mindedness and intraorganizational knowledge sharing. Leaders should take action to achieve such characteristics in their organization by

(45)

45 encouraging employees to seek knowledge that is outside their immediate scope. Leaders can also implement cross-functional integration, which will enable employees to learn and develop new skills while sharing existing knowledge throughout the organization

(Calantone et al., 2002, p. 522). Once learning orientation is installed, dynamic capabilities becomes crucial to create value from the knowledge input. This also where the study falls short, since it does not explain which specific dynamic capabilities that are most useful. By combining sensing, seizing and reconfiguring dynamic capabilities firms are likely to succeed with their business model, since it provides a sequence for which knowledge can be transformed to capture the value. The overall lesson must therefore be that it is utmost important for companies to have routines and processes that enables them to take action. Learning orientation can be seen as the intention to innovate the business model, whereas the dynamic capabilities is the specific actions that are needed to be taken to achieve business model innovation.

7 Limitations

The study contains several limitations. First of all, the use of convenience, purposive and snowball sampling, which all are defined as non-probability sampling, results in limited representativeness and generalizability (Saunders, Lewis, & Thornhill, 2008). The primary reason for non-probability sampling can be accredited to the limited time and resources of the study. To increase the statistical inference, future research should use probability sampling. Especially stratified random sampling is recommended, given that the focus is on manager’s that are knowledgeable about the business model.

(46)

46 The study is also likely to suffer from a non-response bias. Extending the logic of the

respondent criteria applied from Kortmann et al. (2014), there is a relationship between the rank of an employee and how knowledgeable he/she is about the business model innovation. Thus, the most suitable respondents are likely to be CEOs which in turn are very busy and the least likely to answer surveys. Mitigating such challenges in the future seems difficult within the confines of a master’s thesis, however leveraging the prestige of being associated with University of Amsterdam or collaborating with a consulting firm could increase the willingness to answer.

The study is also prone to common method variance, in the sense that the variance is a result of the measurement method rather than the constructs the measures represent (Podsakoff et al., 2003). The consistency motif, where respondents wish to be perceived as rational and thus provide answers in a consistent and coherent manner, was sought

mitigated be using different scales in the survey. Seemingly related to the purposive sampling method applied, the study is possibly subject to leniency bias. Leniency bias is the probability that respondents answer in accordance with what is desirable dependent on whether they like a person or not (Podsakoff et al., 2003). Since most respondents has a relation to the author of this study, they can possibly have answered in a way they expected to create the most desired outcome. Finally, social desirability is very likely to affect the answers. Since the list of questions include how good firms are at sharing knowledge internally, innovating their business model etc., there is a strong inclination for managers to answer that positively. This can be mitigated by increasing the number of respondents

An exploratory study on the relationship between learning orientation, dynamic capabilities and business model innovation