Exploring organizational ambidexterity : do large multi-unit firms balance exploration and exploitation across units and over time simultaneously?

(1)

Exploring Organizational Ambidexterity – Do large

multi-unit firms balance exploration and

exploitation across units and over time

simultaneously?

MSc Entrepreneurship

Master Thesis

Author

Maximilian Jarczyk

UvA: 11770635

VU:

2629690

Supervisor

Dr. Nazli Ugur

Application Date

April 1st, 2018

Submission Date

July 1st, 2018

Word count

13394

(2)

Statement of Originality

This document is written by Student Maximilian Jarczyk who declares to take full responsibility for the contents of this document.

I declare that the text and the work presented in this document are 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)

Preface

The thesis has benefited from the support of a few people whom I would like to thank in the following.

First and foremost, I would like to express my gratitude to my academic supervisor Dr. Nazli Ugur, who has guided me through my thesis, supporting me in my writing and research process. Her constructive feedback and her always valuable comments throughout the entire period have helped me a lot and are much appreciated.

Secondly, I would like to thank my parents who have helped me to get to this final point of my educational career and supported me in every aspect, to keep my life of unnecessary distractions, so I could fully focus on my degree. I was and always will be grateful for their continuous encouragement. Lastly, I would like to thank my grandparents for their great financial support, helping me tremendously keep my life stress free and help me solely focus on my academic efforts.

(4)

Summary

Author of the dissertation:

Maximilian Jarczyk

Title of the dissertation:

Exploring Organizational Ambidexterity – Do large multi-unit firms balance exploration and exploitation across units and over time simultaneously?

Abstract

Organizational ambidexterity describes the ability of a firm to balance explorative and exploitative activities in order to achieve higher market performance. The aim of this study was to detect reciprocal ambidexterity within a firm, balancing exploration and exploitation across units and time simultaneously and question its financial outcomes. By analyzing R&D intensive multi-unit firms in five different sectors and three regions, we found a significant number of reciprocal firms that showed better market performance compared to non-reciprocal firms. We conclude that reciprocal firms have a competitive advantage, but their ability to act reciprocal greatly depends on their respective firm size. We can therefore add to the current theory of reciprocal ambidexterity, that a firm can only act reciprocally ambidextrous if it does not exceed a unit size limit of more than 20 units on average per firm.

(5)

Table of Figures

Figure 1. Simsek et al.´s Four Categories of Organizational Ambidexterity ... 7

Figure 2. Structural Ambidexterity ... 11

Figure 3. Temporal Ambidexterity ... 12

(6)

Table of Tables

Table 1. Descriptive Statistics & Correlations ... 21

Table 2. Sector Dummy Results ... 22

Table 3. Region Dummy Results ... 22

Table 4. Bayesian Estimates of Coefficients a,b,c... 23

Table 5. Results for Linear Regression Excluding Reciprocal Ambidexterity... 24

Table 6. Results for Linear Regression Including Reciprocal Ambidexterity ... 24

(7)

1. Introduction

Our modern economy is seemingly developing more and more rapidly, shortening our technological development cycles at almost exponential rates. This steep innovation trajectory has caused many former incumbents to lose their once leading market position, shrink dramatically in size, or even declare bankruptcy. This downfall of big multi-unit firms is often associated with their disability to respond to emerging disruptive innovation in a timely manner, or even be a market disruptor themselves. (Christensen, 1997). The historical data from the Fortune 500 list, suggests that many incumbents could not realize their innovation ambitions in order to maintain their leading position. Almost 88% of the firms that where present in the Fortune 500 list in 1955 are in 2018 no longer listed (Deloitte, 2015). However, according to a survey of the Boston Consulting Group in 2015, 79% of the respondents indicated that innovation is one of their top three strategic priorities (Boston Consulting Group, 2015).3_Yet,

approximately half of the respondents were dissatisfied with their innovation strategy outcome. This result was consistent for 10 consecutive BCG surveys. Both examples suggest that innovation is a key driver for continuous organizational success, however the organizational strategy / process seems to be not very efficient to sustain innovation within the firm.

The phenomenon of innovation has therefore been the body of decades of research, especially on the corporate / firm impact. Similar to Christensen’s (Christensen, 1997) distinction of incremental and disruptive innovation on firm success, March (March, 1991) described the effects of exploration and exploitation on technological innovation on a firm level. While exploitation refers to the gradual improvement of technology or processes, exploration refers to the exploration of new knowledge that is alien to the firm. It is assumed that in order to stay competitive in the market, a firm needs to maintain an adequate balance between exploration and exploitation (O'Reilly III & Tushmann, 2008). This balance is also referred to as ambidexterity. This balance however, requires two different sets of organizational structures, due to several challenges that occur when applied simultaneously. Since exploration and exploitation require different organizational traits e.g. culture, mind setting, processes, applying it simultaneously within a single organization or more specific within a single unit proves to be challenging (March, 1991). To tackle that dilemma research has proposed two different types

(9)

of approaches to deal with these paradoxical requirements. Firms may either choose temporal ambidexterity (O’Reilly III & Tushmann, 2013), where they alternate between exploration and exploitation over time, or apply structural ambidexterity (Visser, et al., 2010), where they separate exploration and exploitation in different organizational units.4_{Both approaches}

however present very different results (Raisch, Birkinshaw Probst & Tushman, 2009 ; Jansen, Tempelaar, van den Bosch & Volberda, 2009; Simsek, Heavy, Veiga & Souder, 2009) and have not been examined in an combinative study yet.

(10)

2. Structure

First the thesis defines the relevant research question and its goals. Secondly the terms organizational ambidexterity, exploration & exploitation in context with structural and temporal application are defined and explained according to today’s theoretical understanding. After the definition of the key terms the author presents appropriate strategies that are applied in the context of organizational ambidexterity, building the foundation for the forthcoming quantitative analysis. Following up the statistical model used to assess the patent data set is being explained and the identified variables of the data set will be outlaid. Accordingly, in the next chapter the author will present his outcomes and its meaning on his hypothesis.

In the last chapter, these outcomes are being assessed and classified, trying to allow for statistical inference, whether the authors hypothesis is backed up by the examined data or not. Finally, recommendations for further research will be given.

(11)

3. Understanding Organizational Ambidexterity

The concept of organizational ambidexterity is according to March (1991) the successful balance of explorative and exploitative activities. Because both activities require a very different set of conditions it is difficult to find balance. In the following we will give a more detailed look at both exploration and exploitation and will later discuss the strategies that potentially help to succeed at being ambidextrous.

3.1. Exploitation

Exploitation has its primary focus on utilizing a firm’s resources in a most efficient way and build upon knowledge they already possess (March, 1991).5_{The according goal is to capture}

activities such as efficiency, production, selection and execution. Firms that act exploitative, focus their organizational attention on existing business or existing ways of doing businesses. Hence innovations based on exploitative goals build on existing knowledge, reinforcing existing skills, processes and structures (Holmqvist, 2003). The aim of these activities is to create improved reliability in experience through continuous refinement and routinization of knowledge. Furthermore, their common approach is to employ available information and capabilities to achieve short-term organizational goals and market positions (Chen, 2017). This as a consequence means that firms that act exploitative, desire low levels of uncertainties and high rates of success (Chen, 2017). According to March (1991), most of the well-managed firms are good at exploitation. Their intrinsic motivation is to achieve steadily increasing revenues & profits, calling for continuous short-term success (Govindarajan & Trimble, 2010). This leads to organizational acting, that is trying to steadily improve the performance, improve the reliability, efficiency and control.

3.2. Exploration

Exploration on the other hand, focuses on uncovering new knowledge, what is so far alien to the organization (March, 1991). Firms that engage in explorative behavior, try to seize new business opportunities or investigate its potential. This behavior implies that involved organizations do not possess full information on all possible opportunities, meaning it needs to invest and sense the opportunities (Teece, 2007). The characteristics of explorative activities can be therefore best described as, variation, experimentation, innovation, flexibility and risk

(12)

taking (March, 1991). As this activity requires to abandon existing knowledge and start from zero again, creating new knowledge, the goal is to develop new products, technologies, markets or business models that lay the foundation for growth and a firm’s cash flow in the long run (Tushmann & O'Reilly III, 1996). Thus, the following describes exploration as the attempt to develop “new-to-market” innovations, designed to meet or even create the emerging needs of customers and markets (Danneels, 2002).

3.3. Organizational Ambidexterity

Having the two quite contrary concepts of exploration and exploitation lead to the assumption that firms need to maintain an appropriate balance between the two, in order to flourish and stay successful in the long run (Levinthal & March, 1993). Engaging in too much exploration may lead to overproportonally high spending in the short term, weakening not only a companies short term financial strength, but also its ability to finance future operations due to reduced working captial. On the other hand, too much exploitation may strengthen a firms short term financial position but can lead into a competency trap and cause serious long term damage as the firm may struggle to monetize on its products / services in the future. Backed by empirical research it is clear today that organizations that handle exploitation and exploration in an ambidextrous way financially outperform organizations that exclusively focus on one or the other (He & Wong, 2004; Jansen et al., 2006; Lavie et al. 2010; Uotila et al., 2009; Venkatraman et al., 2007; Belderbos et al. 2010; Lubatkin et al., 2006). By focusing on both acitivties organizations benefit from both an increase in short-term and long-term effectivness / performance. Several Studies haven been testing what the impact of the firms exploration share on financial performance is and found a clear relationship expressed in an form of an inverted u-shape (Cao et al., 2009; Vagani, 2012; Fang, Le & Schilling, 2010; Belderbos et al., 2010).

In most cases organizations either choose to focus on exploitation or exploration, as both require substantially different capabilities of the organization and are each on their own easier to implement (He & Wong, 2004). While exploitative acting firms focus on improving their efficiency and gradually improving their products / service, explorative firms aim to create flexibility in the organization through a more open approach to learning, such as high rates of experimentation and aiming for riskier / disruptive innovation projects (Cheng & Van De Ven, 1996). Focusing on either exploitation or exploration solely, carries the risk of failure in the market, thus a sole focus on one activity exclusively does not offer satisfying results (Gibson

(13)

pursuit of exploitative acitivities may lead to short-term performance enhancements, due to an increased output in efficiency, it ultimately leads to a competency trap (March, 1991). The outcome is due to the excessive focus on the current enviroment, skills and knowledge, which outdates over time and thereby dimnishes the long-term performance of an organization significantly (Levinthal & March, 1993). On the other hand a sole focus on exploration does not lead to continuous success either. The constant renewal of knowledge can lead to a failure trap, due to an endless cycle of search and unrewarding change. This may occur when organizations become oversensitive to short-term variations and spend their available (scarce) resources on projects that often give too small returns (Volberda & Lewin, 2003).

Balancing exploitation and exploration however is not as simple as it may sound and is, due to its divergent nature, very difficult to handle simultaneously. Exploitation and exploration require different mindsets, organizational routines and structures. While exploitative activities require efficency-oriented and technocratic organizational practices and processes, explorative acitivities prefer more flexible and organic structures. Exploratory units for example may be small and decentralized with loose processes and an open culture, whereas exploitative units tend to be rather formal, large and centralized with tight processes and cultures (Benner & Tushman, 2003). Furthermore, most firms have only limited resources to invest, creating according to Thompson et al. (1967) a trade off between exploitation and exploration, what he called the “central paradox of administration”. Thus, it is essential for an organization to engage in enough exploitative acitivites to ensure the success of the organization’s short-term objectives and at the same time devote enough resources to explorative activities to ensure the success of its long-term perspective (Levinthal & March, 1993). These demands however stand in contrast to each other and cause strategic and organizational challenges (Lavie, Kang & Rosenkopf, 2011). Applying each acitivity requires as earlier described specific and in most cases contrary organizational pre-sets, creating when applied simultaneously tensions and innefficencies, making well balanced ambidextrous acting somewhat difficult.

3.4. Organizational Ambidexterity Strategies

One possible solution to resolve the tensions that are being created while applying exploitation and exploration, is the division of both activities into either structural, or temporal dimensions. In such a case both activities do not collide with each other and the organization can maximize their potential (O’Reilly III & Tushmann, 2013). Prior research already suggested that creating distinct alignments, each governed by its own management team, having its own processes,

(14)

cultures and incentive systems, are able to adequately make room for the necessary competencies for both existing and emerging business opportunities (Benner & Tushman, 2003; Gilbert 2005). Simsek et al. (2009) proposed a more diverse categorization that created four approaches namely harmonic, partitional, cyclical and reciprocal that emerge within the structural and temporal dimensions. The focus of this thesis is to take a closer look at what Simsek et al. (2009) call reciprocal ambidexterity and which I will refer to as reciprocal ambidexterity as I will look at it in a more intra-organizational context and will combine the concepts of cyclical and unit separated ambidexterity to build my hypothesis. To build my hypothesis however it is necessary to set the right theoretical context of it. Therefore, I will explain the other remaining 3 categories based on Simsek et al. (2009) 4 dimensions of ambidexterity, which I refer to as harmonic-, unit separated- and temporal ambidexterity.

Figure 1. Simsek et al.´s Four Categories of Organizational Ambidexterity

(15)

3.4.1. Harmonic Ambidexterity

Harmonic Ambidexterity assumes that a balance between exploration and exploitation can be achieved within a unit (Simsek, Heavey, Veiga, & Souder, 2009). This balance competes for scarce resources that can cause tensions, leading to conflicts, inconsistencies and contradictions. Simultaneous and interconnected explorative and exploitative acting within the same unit therefore needs continuous operative and strategic activities from an organizational and cultural context (Bartlett & Goshal, 1990; Burgelman, 1991; Bartlett & Goshal, 1994).

The organizational context according to Gibson & Birkinshaw (2004) is best described as a surface level of a culture that is based on processes and beliefs that are able to overcome the contradictions of exploration and exploitation within an organizational unit. Hence a harmonic / optimal balance of both activities in a single unit focuses on how on an individual level systems and processes can help to make the “(…) right judgments in order to divide their limited time between those conflicting demands” (Gibson & Birkinshaw, 2004).

At this point of research in the field of harmonic ambidexterity there are two dominant explanations, namely “contextual” and “behavioral” ones (Adler, Goldoftas & Levine, 1999; Corso & Pellegrini, 2007). The contextual explanation emphasizes the social as well as the performance support that is supposed to encourage individuals to find the optimal balance of explorative and exploitative activities through personal judgment (Bartlett & Goshal, 1994; Gibson & Birkinshaw, 2004). In this context it is the processes and beliefs that help form the individual’s behavior within an organization. In more detail we can describe it as four interdependent attributes: discipline, stretch, support, and trust (Bartlett & Goshal, 1994). Discipline is meant to help employees to strive for measures such as continuous performance, behavioral standards, transparent feedback mechanisms and consistent sanctioning. Stretch refers to the creation of a collective identity and shared visions and goals to helps them strive collectively for more ambitious goals. Support induces individuals to lend assistance to one another. From a managerial perspective such behavior can be fostered by a rather non-authorial leadership style, that embraces stronger self-responsibility and free hand on the employee level. The last attribute trust, emphasizes the idea of individual / mutual reliability and trust on the employee level. This may be achieved by taken measures such as fairness and actively involving employees in decision making processes as well as transparent corporate behavior. In a more holistic view we can also describe routines emphasizing systematic reflection, conflict regulation and integration as useful for balancing exploitation and exploration within a

(16)

single unit (Guttel & Konlechner, 2007). Organizational systems also play an important role such as team-based structures and human resource practices. Out of those structures that promote creativity proved to be the most efficient in promoting the simultaneous pursuit of exploration and exploitation. Looking at it from a managerial perspective, it deems necessary for managers to possess complex behavioral repertoires (Raisch & Birkinshaw, 2008) that place greater attention on the functions of managerial or leadership that a manager can perform (Hooijberg, Hunt, & Dodge, 1997). Considering the resource-based view, a context that manages to balance adequately explorative and exploitative activities, can be considered as a rare and valuable resource, hard to imitate, hence giving the owner of such a competitive advantage (Gibson & Birkinshaw, 2004).

The current state of research is still developing but it can be assumed that the simultaneous balanced approach of exploration and exploitation in a single unit has a positive impact on performance outcomes (Simsek, Heavey, Veiga, & Souder, 2009). Gibson & Birkinshaw (2004) support this assumption, showing that this ambidextrous context has a positive impact on the subjective ratings of middle and senior managers. Hill & Birkinshaw (2006) also observed a higher internal performance of the ambidextrous units. The enhanced performance had a significant impact on four areas namely, creating breakthrough innovation, investment in disruptive technology, development of strategy with key external stakeholders, and funding for internal venturing activities. Yet the cost of implementation as well as the managerial efforts that are needed to achieve and maintain such a level of harmonic ambidexterity on a system and process level represent countervailing factors (Yang & Atuahene-Gima, 2007).

Research on harmonic ambidexterity also concludes that the tensions become even more severe as a result of the simultaneous pursuit of exploration and exploitation can threaten the successful execution of both (Hill & Rothaermel, 2004; Simsek et al., 2009). Another argument that questions the feasibility of harmonic ambidexterity is the individual’s capabilities of acting ambidextrous. As Individuals rely on the same experiences, values and capabilities, this basis stands in contradiction to the individual exclusive specific requirements of exploration and exploitation, making it difficult to apply both activities simultaneously (Gupta, Smith, & Shalley, 2006). Sternberg (1999) already noted, that most individuals that are good at creative (exploration) tasks are fundamentally different in their characteristics compared to individuals who flourish in process optimizing / implementation (exploitation) tasks.

(17)

3.4.2. Structural Ambidexterity

Prior studies that examined organizational ambidexterity suggested, that ambidextrous organizations are composed of structurally differentiated exploitative and explorative units (Benner & Tushman, 2003). Accordingly, Lawrence & Lorsch (1967) described structural ambidexterity as:

“(…) the state of the segmentation of organizational systems into subsystems, each of which tend to develop particular attributes in relation to the requirements posed by its relevant external environment.”

Such a segmentation therefore may be achieved, by applying structural ambidexterity that uses distinct business units, that either focus on exploitative, or explorative activities (O'Reilly III & Tushmann, 2004). Based on this approach Raisch et al. (2009) investigated the simultaneous pursuit of exploitation and exploration and found that successful ambidexterity could be achieved via organizational separation, where organizations establish dual structures within their organization. O’Reilly III & Tushmann (2011) proposed in their case study accordingly, that successful organizations physically separate their exploitative & explorative activities. This approach implies the logic, that the resources, behavior and structures needed for exploitation and exploration are fundamentally different and need to be pursued at different organizational units at the same time (Gupta, Smith, & Shalley, 2006). This results in ambidextrous organizations, that deploy spatially dispersed exploitative and explorative units in different locations (Tushmann & O'Reilly III, 1996). Carlile (2004) described this separation as “pragmatic boundaries”, safeguarding innovative and experimental activities from conversely managerial, formal cognitions and inertia, representing the parents mainstream activities (Benner & Tushman, 2003). This separation reduces the possible organizational tensions and increases the efficiency of balancing the two activities (Lavie et al., 2010).

(18)

Figure 2. Structural Ambidexterity

Source: Based on Raisch & Birkinshaw (2008), p. 485

Recent research has shown, that a separate exploration / innovation unit has a clearly positive impact on both exploration and exploitation in manufacturing and service firms leading to increased overall performance (Blindenbach-Driessen & Van den Ende, 2014). Today it seems clear, that the successful establishment of structural separation amplifies many crucial competencies such as, autonomy, shared responsibility, functional complementarity, communicational richness and decentralization (Cao, Gedajlovic, & Zhang, 2009). Geerts et al. (2017) confirm in their empirical results, that firms with greater structural separation show an elevated level of technological performance and inhibit the ability to create synergies between explorative and exploitative activities leading to potential spillovers.

3.4.3. Temporal Ambidexterity

Besides the possibility of structural ambidexterity Duncan (1976) suggested, that firms need to shift their structures over time in order to accommodate the conflicting alignments that are required to increase both innovation output, as well as increasing the organization’s overall efficiency. Tushman & Romanelli (1985) adequately proposed their theory of punctuated equilibrium change, suggesting that firms evolve through punctuated changes, answering external environmental changes by redefining their processes and structures executed via a sequential / temporal process. According to Gupta et al. (2006) the punctuated equilibrium is

(19)

analysis stays within a single organizational unit. Furthermore, we can say that in general the punctuated equilibrium puts a higher emphasis on explorative then exploitative change, thus the changes that occur while applying explorative activities seem to be quite radical and competence destroying (Simsek, Heavey, Veiga, & Souder, 2009). Brown & Eisenhardt (1997) proposed in their study, that firms use “semi structures” and “rhythmic switching” to be able to vary between periods of exploitation and exploration. Accordingly, firms apply a temporal separation of exploitation and exploration sequentially over a certain period of time, by following the cycles of technology or product life stage (O`Reilly & Tushman, 1996; Lavie, Kang & Rosenkopf, 2010). Nickerson & Zenger (2002) and Boumgarden, Nickerson and Zenger (2012) describe this process also as “vacillation” arguing, that firms benefit from such a behavior as it makes it easier for them to switch between formal structures opposed to a change of the firm’s culture or informal organization. This underlines the results of Han & Celly (2008), who show in their study, that cyclical ambidexterity is mostly associated with innovative outcomes, such as product innovation. Firms that generally have a strong focus on technology, often have technological units that have rounds of exploitation and exploration, that are specifically designed to achieve increased performance output and product innovation (Atuahene-Gima, 2005).

Figure 3. Temporal Ambidexterity

(20)

As displayed in Figure 3. the underlying investment pattern of such technology-oriented firms can most adequately be described as a repetitively S-shaped curve (Tushman & O´Reilly III, 1996). At the beginning of the investment the curve reflects the early stage efforts and its further required investments to reach a commercially exploitable product (Chen, 2005). As a result of the efforts, eventually an increase in production is obtained and the sequence of exploration is substituted by a sequence of exploitation of the innovation. The sequence of exploitation now is focusing on performance improvements, by shifting its focus on process innovation. Such a shift allows firms to be able to adjust to market conditions more adequately, diminishes potential tensions, as well as its costs, without having to deal with the cost of integration of both activities at the same time. The benefits of such a separation are on the one hand, the avoidance of conflicting demands of exploitation and exploration (Puranam, Singh, & Zollo, 2006), while on the other hand exploitation and exploration support and complement each other in a growing cycle, in which the firms benefit the most from both learning opportunities (Venkatraman, Lee, & Iyer, 2007). This creates a synergic effect that ultimately benefits each single unit as well as the organization as a whole. Siggelkow & Levinthal (2003) also refer in this case to the signified avoidance of competency traps (too much exploitation) and failure traps (too much exploration).

What sounds so easy and flawless in theory however, in reality it is not always the case. To find the right timing to switch between exploitation and exploration often proves for the management to be difficult (Gibson & Birkinshaw, 2004). This difficulty may produce conflict within management, making it necessary to find adequate mechanisms that help to cope or even avoid such conflicts (Floyd & Lane, 2000). Furthermore, it is a big challenge for firms to adapt the whole organization from exploitative activities to explorative ones smoothly and continuously (Brown & Eisenhart, 1998). This shift therefore does not only create problems at the management level but also on the individual/employee level. Employees have to engage in both adaptation to market demand and alignment of their competencies. These mechanisms may as a consequence cause high costs in planning and implementation, while at the same time productivity decreases in the transition phase due to a higher employee turnover and their intrinsic resistance to change (Raisch, Birkinshaw, Probst, & Tushman, 2009). Keeping these issues in mind, temporal ambidexterity might prove to be a problem for highly human capital intensive and large established firms, especially in highly dynamic environments. O`Reilly III & Tushman (2013) correspondingly noticed, that temporal ambidexterity may be a good

(21)

strategic fit for less dynamic environments, such as the service industry and firms with a smaller size, encountering less problems in transition phases.

3.4.4. Reciprocal Ambidexterity, Across Units and Time

As we just examined structural- and temporal ambidexterity, the current state of research and literature in both areas is rich and extensive and we have today a good understanding of both mechanisms, that can be applied in context to exploitation and exploration. Reciprocal ambidexterity however is still a quite underdeveloped theory and has not received adequate attention from organizational researchers (Simsek, Heavey, Veiga, & Souder, 2009). This fourth type of organizational ambidexterity assumes a reciprocal interdependence between exploitation and exploration. Thompson et al. (1967) theorize it as a continuous sequence of transferring exploitative results from one unit to the next, that uses it in an explorative manner and vice versa. This transfer requires then a relationship that includes an exchange of information, joint problem solving, collaborative decision making and last but not least a flow of resources between the different units either responsible for exploration or exploitation (Simsek, Heavey, Veiga, & Souder, 2009). This activity as a result combines exploitation and exploration across time and units (Lavie & Rosenkopf, 2006). Holmqvist (2004) adds to this perspective, that “(...) the interplay between exploitation and exploration takes place both within and between organizations.”

This perspective that includes the inner firm perspective on organizational ambidexterity proves to be controversial until today, having scholars argue whether balancing exploitation and exploration would challenge firms too much and as a consequence make reciprocal ambidexterity for them unobtainable (Jansen, 2005). The tensions that arise when exploration and exploitation collide within a firm marginalize its individual benefits. In line with this discussion are Porter’s (1996) findings, claiming that it is impossible for firms to achieve low-cost on the one hand and a strategy for differentiation on the other hand simultaneously. As discussed earlier through certain strategies such as structural or temporal separation, it may enable firms to avoid those conflicts and therefore increase their overall performance. Certain strategies such as the above discussed structural or temporal separation could be helpful for firms to avoid those conflicts and therefore increase their overall performance.

Another option to act ambidextrously could be to establish a specialization of exploitation and exploration spread out over an inter-organizational network. Following this theory researchers

(22)

have suggested that ambidexterity may be also achieved by outsourcing (Baden-Fuller & Volberda, 1997) or establishing alliances (Holmqvist, 2003). The externalization of knowledge might be particularly useful for exploration, contributing significantly to the renewal of existing knowledge bases (Rosenkopf & Nerkar, 2001). Externalizing the generation of knowledge could prove to be more effective than internalizing it, because it circumvents the potential problems that are associated with the trade-off occurring internally between exploitation and exploration (Gupta, Smith, & Shalley, 2006). On the flipside, externalizing knowledge can also impose some difficulties, as the successful integration of the newly acquired knowledge across the firm can be complex and intricate (Benner & Tushman, 2003). Contrary to temporal ambidexterity however, managers need to engage in relationships that Simsek et al. (2009) best described as “ongoing information exchange, collaborative problem solving, joint decision making, and resource flows between the different units responsible for exploitation and exploration”.

In summary, reciprocal ambidexterity can be best described as a synergy of complementary units that either focus on exploitation or exploration, occurring across time and firms (or units). While current studies suggest that reciprocal ambidexterity in an inter-organizational setting has a positive effect on firm performance (Lavie & Rosenkopf, 2001; Im & Rai, 2008; Tiwana, 2008) there is no current scientific empirical evidence about reciprocal ambidexterity across units and time that happens exclusively within firms and what its potential effects are. Because research showed that inter-organizational reciprocal ambidexterity has a positive effect on firm performance, I hypothesize that ambidexterity within a firm across units and time has a positive effect on a firm’s financial performance. But multi-unit firms should possess the capabilities to grant their sub-branches / units enough resources and managerial independence to act as if they were almost separate firms, however with the notion of intended knowledge sharing and possibly reciprocal exploration and exploitation strategies / partnerships.

Hypothesis: Reciprocal ambidexterity, balancing exploration and exploitation activities within

(23)

4. Methodology

The thesis is based on a combination of literature review and a statistical assessment of an extensive set of a firm’s patent- and financial data. The basis of this thesis is the comprehension of organizational ambidexterity as part of applied or action-oriented business economics, and ambidextrous strategies that are potentially deployed insuring the survival and eventually systemic supremacy on the free market. The methodology of the thesis consists of qualitative and quantitative parts, with a higher weighting on the quantitative part and its outcome. Available theoretical approaches on organizational ambidexterity are being presented and analyzed, with a special focus on the two approaches of structural and temporal application (as a combination referred to as reciprocal ambidexterity) of exploitation and exploration innovation strategies. According to the current status quo on the understanding of ambidextrous strategies, we analyze an extensive patent data set in order to derive from his quantitative findings implications on his hypothesis, that large multi-unit firms balance exploration and exploitation activities across units and over time simultaneously, resulting in better market performance compared to non-performers. For the analysis of the patent data we use a novel approach that pre-selects firms fitting our definition of reciprocally active and compare them to find statistical evidence for our hypothesis. Subsequently the results are evaluated and recommendations for further research deducted.

4.1.Sample & Data

To assess our hypothesis, a specifically tailored micro-panel dataset we built with detailed information on the reciprocal activity of 66 R&D intensive firms across Europe, the U.S. and Japan in the Chemical & Biotech, Electronics, Engineering & Machinery as well as the Information Technology (IT) sector. Our micro-panel dataset observes a nine-year time frame (1995-2003) in which we measure their reciprocal activity, as well as collect financial- and performance measures for each year. The collected data originates from a previous study from Ugur (2017) about ambidexterity strategies (temporal & structural).6_{The assessed firms come}

from the ‘2004 EU Industrial R&D Investment Scoreboard’. This board contains a record of the 500 most R&D active companies from Europe, the U.S. and Japan. The sample firms from the Chemical & Biotech, Engineering & Machinery, Electronics and IT sector that were extracted from the scoreboard, are roughly equally distributed across their respective industries

6_{The original Dataset contained 188 firms from which we could only use 66, due to our measuring method. Most}

firms would not meet our criteria of having at least two explorative active units having a minimum of five explorative patents over a period of three consecutive years.

(24)

and home regions, representing the firms with the biggest R&D budget, as well as having more than one exploratively active unit for the observed period of time (see also appendix 1 & 2).

Figure 4. Reciprocal Ambidexterity

Source: Own illustration based on findings

The integrated panel dataset that was used contains patent data (European Patent Office (EPO)) and financial data (Worldstat and Compustat). The results of the patent orientation (explorative or exploitative) are collected for each unit of a firm at the consolidated level, which allow to measure for exploration orientation. The consolidation is conducted on an annual basis (1995-2003). This approach allows to observe changes in the exploration orientation of a firm over time. Patent data as a measure for explorative and exploitative activities was used, as it tends to give a better and more holistic picture of a firms technological and innovative activity, then solely text-based indicators (Ugur, 2017; Belderbos, Faems, Leten, & Van Looy, 2010; Geerts, Leten, Belderbos & Van Looy, 2017).

4.2.Variables

To test our hypothesis, variables were selected that potentially correspond with each other significantly and give a descriptive statement. Our model operates on the basis of both a

-0,06 -0,04 -0,02 0 0,02 0,04 0,06 0,08 1 9 9 4 1 9 9 5 1 9 9 6 1 9 9 7 1 9 9 8 1 9 9 9 2 0 0 0 2 0 0 1 2 0 0 2 2 0 0 3 2 0 0 4

MERCK KGAA

(25)

dependent and an independent variable. Furthermore, we included control and dummy variables into our model to give better estimates.

Dependent Variable

Firm Performance. To detect the degree of relative market performance, Tobin´s Q was used,

which accounts for current market performance as well as future prospects, derived from intellectual capital, goodwill, technology and other intangible assets that a company may have but are currently not accounted for (Lavie, et al., 2011). This combined short-term and long-term measures help to capture the success of the firm more accurately and are a better measure when accounting for performance over time (Uotila et al., 2009). Tobin´s Q was developed by James Tobin a graduate of Yale University, with noble laureate in economics. Tobin hypothesized with this measure, that the combined market value of all companies that are listed at the stock market, should be more or less equal to their replacement costs. To calculate Tobin´s Q we divide the total market value of a firm by its total asset value. We picked the last observable value of our observation period, 2003. As we assume that reciprocal ambidexterity follows the notion of innovation cycles and that the value of firm increases over time, the last year of our observation period functions as the “outcome point” that we will compare our independent variable with.

𝑇𝐵𝑄$%&&' =

𝑀𝑎𝑟𝑘𝑒𝑡 𝑉𝑎𝑙𝑢𝑒_$%&&' 𝑇𝑜𝑡𝑎𝑙 𝐴𝑠𝑠𝑒𝑡𝑠$%&&' Independent Variable

Inter-unit Correlation – Degree of reciprocal ambidexterity within a firm. The degree of

reciprocal ambidexterity within a firm will be expressed via the correlation value of the two most R&D active units within a company. This approach was chosen for reasons of unification and pattern detection. Some of the largest companies had more then 80 units while others had around 4. When intercorrelating every unit with each other in larger firms, the most common outcome was zero correlation. With this way of measuring, potential patterns are hidden between too much noise and the majority of units that would not follow a pattern would marginalize the correlation outcomes of the true correlations. With the approach chosen in most cases the parent unit was compared to a smaller unit, showing in more than 80% of the assessed firms significant correlation. The correlation value between the two most R&D active units will help us describe if there is as hypothesized, a reciprocal relationship between units. To see how units, alternate in their patent activity behavior and how units behave in relationship to others,

(26)

we use the delta method, showing the yearly change in explorative activities. Delta helps us to measure the internal yearly changes within a unit of the alternating exploration ratio. Applying this method to all units simultaneously helps to set the units into relation with each other. In order to show the assumed reciprocal relationship, we calculated the inter-unit correlation of a firm and their respective deltas.

1. Exploration Orientation – Relative Share of Exploration. To define whether a patent is

considered as explorative or not (exploitative) we rely on a novel indicator developed by Belderbos, et al. (2010). A patent is considered as explorative if it is situated in a technology domain that is new to the firm for three years (Ugur, 2017). To calculate the Exploration orientation of an independent unit we simply divide the number of explorative patents by the total amount of patents per year.

𝐸𝑂 = 𝐸𝑥𝑝𝑙𝑜𝑟𝑎𝑡𝑖𝑣𝑒 𝑃𝑎𝑡𝑒𝑛𝑡𝑠$ 𝑇𝑜𝑡𝑎𝑙 𝑛𝑢𝑚𝑏𝑒𝑟 𝑜𝑓 𝑃𝑎𝑡𝑒𝑛𝑡𝑠_$

2. Exploration Orientation Shift – Delta of Units. To observe how units, change over time

their explorative or exploitative activity, we calculate their respective deltas.7_{This method}

helps us create a wave like pattern that indicates periods of exploration and exploitation (see Appendix 3). Because in fact most units in our panel data-set mostly showed zero explorative activity over the observed period of time, we calculated only deltas for firms and their respective units that had more than 10 explorative patents, distributed over a period of at least 4 years. To calculate delta, we subtract the exploration orientation of year 1994 of 1995 and so on until 2003, generating 9 delta values for N = 10.

∆𝑥 = 𝐸𝑥𝑝𝑙𝑜𝑟𝑎𝑡𝑖𝑜𝑛 𝑂𝑟𝑖𝑒𝑛𝑡𝑎𝑡𝑖𝑜𝑛_$− 𝐸𝑥𝑝𝑙𝑜𝑟𝑎𝑡𝑖𝑜𝑛 𝑂𝑟𝑖𝑒𝑛𝑡𝑎𝑡𝑖𝑜𝑛_$CD

3. Reciprocal Ambidexterity –Degree of negative correlation of exploration orientation

shifts across units. To detect now a potential pattern between the two most active units of

(27)

a firm we perform a correlation analysis that indicates whether the units are i) positively correlated meaning they do not act reciprocal but synchronous, suggesting either temporal or structural ambidexterity, ii) not correlated at all, indicating neither temporal / structural - nor reciprocal ambidexterity, but potentially harmonic iii) negative correlation, indicating reciprocal ambidexterity. To approve our hypothesis, it is therefore necessary to find a significant amount of companies within our sample that shows negative correlation (see Appendix 4).

𝑅𝐴 = ∑ GH∆IJ− ∆IJK ∗ H∆MJ− ∆MJKN OPQ

$QRC&'

S∑OPQ (∆_IJ− ∆_IJ)%_∗

$QRC&' ∑OPQQRC&'(∆MJ − ∆MJ)%

Control & Dummy Variables

Our empirical model uses as controlling variables the log transformed number of employees of a firm as a measure for size, as well as the log transformed nominal R&D spending as a measure for R&D intensity. These variables are being added as we assume that they are likely to have an impact on market performance. Furthermore, we include in our model sector and country dummies accounting for sector and country specific factors affecting a firm’s market performance (Tobin’s Q).

Firm size (logarithm of number of employees). Literature suggests that the firm’s size may

heavily affect the financial performance of a firm, both short – and long-term (Belderbos et al., 2010). To measure size, we used the number of employees of a firm. The values were logarithmically transformed. The data on the number of employees of a firm is collected from Ugur (2017) based on annual financial reports from, Worldscope and Compustat.

R&D Intensity (logarithm of nominal R&D spending). Research has shown that the amount

of money spent on R&D has a significant impact on market performance (Doukas & Switzer, 1992). R&D Intensity is measured taking the nominal R&D spending average per unit of a firm. The financial data is collected from Ugur (2017) based on annual financial reports taken from, Worldscope and Compustat.

(28)

5. Empirical Analysis

The objective of this thesis is, firstly to observe reciprocal ambidexterity and secondly to examine its influence on a firm’s market performance. The reason for choosing a panel data set is its advantage of being able to introduce average values for variables, correlation estimates, to examine exploration & exploitation shifts / cycles over time, and to control the impact of unobserved firm-specific characteristics. We estimate linear regression models to describe the relation between a firm’s market performance and its level of reciprocal ambidexterity in a cross-section analysis. Using linear regression is useful for quantifying the strength of dependent and independent variables (Xin, 2009). We treat sector, country and size dummies as exogenous variables. For Tobin’s Q I instrument, level of reciprocal ambidexterity, firm size (log transformed) and R&D Expenditures (log transformed).

Additionally, we estimate Bayesian One-way Anova relating reciprocal firms to non-reciprocal firms, allowing for relative market performance comparison (Tobin’s Q).

5.1.Descriptive Statistics & Correlations

Table 1. Descriptive Statistics & Correlations

Mean SD 1 2 3 4

1 Tobin’s Q 1,12 0,56 1

2 Degree of Reciprocal

Ambidexterity -0,08 0,43 0,314 1

3 Number of Employees (log) 4,54 0,47 -0,279 -0,167 1

4 R&D Intensity (log) 5,64 0,60 0,03 -0,073 0,8 1 N= 66

As displayed in table 1 we can see the descriptive statistics & zero- order correlations for the variables that we used in our regression analysis. The mean for our dependent variable is 1,12. The mean of our independent variable (Degree of Reciprocal Ambidexterity) is -0,08, while the

(29)

respectively. The respective standard deviation for our variables are 0,56 for the dependent variable, 0,43 for our independent and 0,47 & 0,60 for our control variables. The correlation between our depended variable and our independent variable is 0,314, while for our control variables it is -0,279 and 0,03. The intercorrelation between our control variables is 0,8. All correlations are significant at the 0.05 level.

Table 2. Sector Dummy Results

Sector Reciprocal _{Firm Avg.} Number of Employees Avg.

Number of

Units Avg. R&D Intensity Avg. Tobin’s Q Avg. Pharma&Biotech and Chemicals 0,066 32895 15 1.011.075 $ 1,34 Machinery, Electronics, IT -0,098 76829 23 $ 929.391 1,10

To give a more elaborate understanding of the distribution of the data, we provide in table 2 the summary of the statistics for five variables: reciprocal firm average, number of employee’s average, R&D intensity average and Tobin’s Q average across sectors (for more details see appendix 9.5)

As already assumed by Ugur (2017), Pharmaceuticals & Biotechnology and the Chemical Sector have high structural ambidexterity and therefore have a good base to also show the highest reciprocal firm average, which stands at 0,066. Furthermore, they show the lowest average number of employees and average number of units, highest amount spent on R&D per unit, as well as having the highest average score of Tobin’s Q.

The remaining other sectors, namely IT, Engineering & Machinery and Electronics score clearly second in terms of reciprocal ambidexterity activity. Furthermore, the other sectors show higher averages on firm’s size but lower R&D spending and score lower on market performance then Pharma & Biotechnology and Chemicals.

Table 3. Region Dummy Results

Region Reciprocal _{Firm Avg.} Number of Employees Avg.

Number of

Units Avg. R&D Intensity Avg. Tobin’s QAvg. EU & US -0,015 5630 20 $ 974.920 1,311 Japan -0,163 73598 19 $ 1.144.673 0,847

(30)

Similar to Table 2 we provide in Table 3a summary of the statistics for five variables, reciprocal firm average, number of employee’s average, R&D intensity average and Tobin’s Q average, this time across countries.8_{As we can see the EU & US have on average the highest performing}

reciprocal firm average in combination with the best average market performance. However, the EU & US have in comparison to Japan a lower R&D budget and less employees with almost the same aver number of units per firm.

5.2. Empirical Results

Table 4. Bayesian Estimates of Coefficients a,b,c

To see whether reciprocal firms perform better on average than non-reciprocal firms we grouped them into their respective groups depending on their outcome level.9_{Table 4 displays}

the outcome of our Bayesian One-way Anova, showing that reciprocal firms had a higher mean market performance of 1,296 than non-reciprocal firms, having a mean market performance of 0,974. Both groups have shown identical variance at a very low level of 0,008 & 0,010. The 95% credible interval for reciprocal firms varied from 1,097 to 1,495, while non-reciprocal firms only varied between 0,797 and 1,115.

In order to assess whether our hypothesis, that reciprocally acting firms show a better market performance than non-reciprocal firms, can or cannot be rejected, we perform a linear regression analysis with our dependent variable Tobin’s Q as a performance measure and the

8_{We grouped countries of the European Union into one group. Because of the high number of individual countries}

within the European Union, the comparability with the US and Japan would be very complex and confusing. For a more detailed look into the distribution and respective values of the EU countries see appendix 5.

9_{Firms that showed a negative reciprocal activity value were grouped as non-reciprocal (group 0) while firms with}

Parameter

Posterior 95% Credible Interval Mode Mean Variance Lower Bound Bound Upper Non-Reciprocal = 0 0,974 0,974 0,008 0,797 1,115 Reciprocal = 1 1,296 1,296 0,010 1,097 1,495

a. Dependent Variable: Tobin’s Q b. Model: Form of Ambidexterity c. Assume standard reference priors.

(31)

degree of a firm’s reciprocal activity as our independent variable. We control our independent variable with measures for size (Number of Employees, log transformed) and R&D intensity (Nominal R&D Spending).

Table 5. Results for Linear Regression Excluding Reciprocal Ambidexterity

Model Unstandardized Coefficients Standardized Coefficients t Sig. B Std. Error Beta (Constant) 1,946 0,608 3,201 0,002 Employee (lg.) -0,994 0,213 -0,844 -4.656 0,000 R&D intens. (lg.) 0,652 0,168 0,705 3,892 0,000

a. Dependent Variable: Tobin’s Q

Table 5 shows the linear regression model excluding our independent variable. The control variables for size (Employee count log transformed) and R&D activity (R&D intensity log transformed) prove to have a significant positive impact on market performance on an alpha level of 5% (p< 0.05). The unstandardized coefficient for size is - 0,994 with a standard error of 0,213. This means, that the bigger in size a firm is, the worse its market performance will be. Conversely is the measured effect of R&D activity on market performance. Here we have an unstandardized coefficient of 0,652 with a standard error or 0,188, indicating a positive impact on market performance.

Table 6. Results for Linear Regression Including Reciprocal Ambidexterity

Model Unstandardized Coefficients

Standardized Coefficients t Sig. B Std. Error Beta (Constant) 1,797 0,595 3,023 0,004 Reciprocal activity 0,300 0,138 0,234 2,175 0,033 Employee (lg.) -0,911 0,211 -0,773 -4,319 0,000 R&D intens. (lg.) 0,616 0,164 0,666 3,763 0,000

a. Dependent Variable: Tobin’s Q

In the second step we included in Table 6 our independent variable, namely reciprocal activity. As we can see the results are still statistically significant on an alpha level of 5% (p<0.05). The unstandardized coefficient is with 0,300 positive, indicating a positive relationship between the reciprocal activity of a firm and its market performance (Tobin’s Q). The standard error is 0,138

(32)

with a beta of 0,234. Inserting the independent variable does not alter their statistical significance (0,000 & 0,000) therefore we can also not reject our hypothesis.

Table 7. Relationship between Reciprocal Activity & Firm Size

X-Axis: Reciprocal Activity Y-Axis: Firms Size

Having a closer look at the impact of firm size on our model we drew a scatterplot in Table 7, showing that the more units (size) a firm has, the less it is able to act reciprocal. The function indicates that at above a number of 20 units, firms do on average not engage in reciprocal ambidexterity anymore. These results are in line with Jansen’s theory (2005) that reciprocal activity might be too complex for a firm to manage. Having a few out layers that show big firm size yet slightly negative reciprocal activity may show us that our theory is not fully explained and at certain size different theories / strategies towards organizational ambidexterity may apply. It may be possible that as already discussed by (Baden-Fuller & Volberda, 1997; Holmqvist, 2003) that firms engage in reciprocal ambidexerity via alliances or outsourcing of explorative or exploitative activites. Our model could potentially show that this is particullarly the case for firms with well beyond 20 units, thus not being explained in our model. However, we can most certainly say that our results give a more fine-grained answer, suggesting that it depends on the size whether a firm is able to act reciprocal or not.

0 10 20 30 40 50 60 70 80 90 100 -1 -0,8 -0,6 -0,4 -0,2 0 0,2 0,4 0,6 0,8 1

(33)

Looking at all empirical findings in summary, we may summarize the following: First and foremost, it seems clear that firms that show a positive degree of reciprocal ambidextrous activity perform on average better then firms that show negative reciprocal activity. Secondly, we proved that our assumption is statistically significant when using linear regression including control variables. And last but not least we showed in a more fine-grained look, that the turning point between acting reciprocally ambidextrous within a firm, is according to our findings limited to firm size (max. 20 units on average). However, we noted two out layers that potentially are not explained through our model. These findings overall lead to the conclusion that we cannot reject our hypothesis.

(34)

6. Discussion & Conclusion

Reciprocal ambidexterity was up until today only a theory and had scholars question its feasibility (Jansen, 2005; Lavie & Rosenkopf, 2006). Accordingly, the goal of this research was to provide empirical evidence that ambidextrous firms are able to balance exploration and exploitation across units and time simultaneously and eventually show superior market performance due to this behavior. Consistent with our hypothesis we can support this assumption with the findings from our data. Organizational ambidexterity is, as literature suggests, a highly complex undertaking at any level and only a fraction of firms is able to maintain a well-managed balance. Previous studies indicated that large multi-unit firms engage in too much exploitation and too little exploration (Benner & Tushman, 2003). Our data clearly supports this assumption, showing that out of our original sample size (188 firms) of R&D intensive firms across four sectors and three regions only 39,28% (66 firms) showed sufficient explorative activity to be potentially able to act reciprocal. Increasing their explorative activity would therefore not only help them to better balance exploration and exploitation, thus increasing performance but also enable them to act reciprocally.

Out of the firms that we found to be balanced enough, we found, around 45,45 % (30) firms that showed significant levels of reciprocal activity. In addition, we were able to show that reciprocal firms were clearly able to outperform non-reciprocal firms in terms of market performance. To see, that from our original sample size of 188 firms only 17,85% of the firms actually acted reciprocal, is somehow in line with Jansen’s (2005) assumption, that reciprocal ambidexterity may be too complex to manage and very hard for firms to implement. In addition to this assumption we found further evidence supporting Jansen’s argument partially, showing that firm size can be a limiting factor for interorganizational reciprocal ambidexterity. Our data suggests, that firms that have more then 20 units on average, are less likely to act reciprocal then firms with less than 20.

So, what are the driving factors enabling a firm to act reciprocal? Lavie & Rosenkopf (2006) described them as a combination of structural and temporal separation simultaneously applied by a firm. Just like finding the right balance for exploration and exploitation, it seems like the same is true for balancing structural and temporal separation in order to achieve reciprocal ambidexterity. Earlier research showed that structural separation outperforms temporal separation in terms of market performance (Ugur, 2017). The findings furthermore show that

(35)

sectors like the Bio & Pharma and Chemical industry show the highest levels of structural ambidexterity. Those were by no surprise also the sectors with the smallest firm size on average - an important factor when considering temporal separation.

It seems like there is a certain order to it. Firms with high structural separation from these sectors are also the ones with the lowest average on firm size. We can assume based on Ugur’s (2017) findings and ours that there might exist a certain relationship, suggesting that structural separation is a quantifying factor for reciprocal ambidexterity. Sure, without temporal separation, reciprocal ambidexterity is not possible but structural separation may be an important leveraging factor. Furthermore, the firm size of the structurally and reciprocally superior sectors show, that temporal separation happens when it is actually within a dimension that can still be managed. As a result, we can conclude that for reciprocal ambidexterity it is necessary that a firm first of all shows a high level of structural separation and then establishes, within considerable firm size, temporal separation additionally.

Lastly, we found that the U.S. and the EU showed higher levels of reciprocal ambidexterity than Japan. This finding suggests that there seems to be a cultural / managerial difference impacting the ability of a firm to act reciprocal that we did not account for in our model. This assumption is supported by the fact that Japan had less than twenty units on average, thus not violating the limiting factor of size.

(36)

7. Limitations & Suggestions for Further Research

To determine the degree of a firm’s reciprocal activity we used in our research paper a relatively simple method, that only allowed for a very broad inspection and understanding of a firm’s ambidextrous activities. Further research should try to include all units in a comparative model describing more accurately its behavioral nature. Furthermore, we suggest to include additionally other definitions than its patent activity only. This may help to further refine the understanding of true explorative and exploitative activities.

Possible additional selection criteria could be: time to market, market success or failure. These additional factors may alter the relationship between what we define as an explorative or exploitative patent even further and may help to reduce the amount of noise, allowing for a more accurate picture.

Our research was also able to show that reciprocal ambidexterity exists, but we weren’t able to prove cycles of change, as Thomson et al. (1967) already theoretically assumed, due to the limited nature of our model. Such an observation would require a more sophisticated statistical model, presumably also extending over a longer period of time. O´Reilly and Tushman (2004) proved that cycles of temporal separation follow a four-year pattern, which in an optimum scenario would have allowed us to observe one change of such a cycle across units over time. As every firm has its own internal clock and managerial time horizons, we suggest to take a closer look at reciprocal firms on the individual level to potentially observe such patterns within reciprocal firms. Furthermore, we could not control in our model for reciprocal ambidexterity that happens between organizations like (Baden-Fuller & Volberda (1997) and Holmqvist (2003) suggested. Firms that seem like they do not engage in reciprocal ambidexerity internally might therefore actually engage in inter-organizational reciprocal ambidexerity.

According to our last findings regarding regional results, further research should also further investigate the cultural impact on a firm’s ability to act reciprocally. Including other Asian countries located close to Japan to the analysis may further strengthen potential findings. Future research may also further investigate the relationship between structural ambidexterity and its potentially leveraging effects on reciprocal ambidexterity between units.

(37)

8. References

Adler, P. S., Goldoftas, B., & Levine, D. I. (1999). Flexibility Versus Efficiency? A Case Study of Model Changeovers in the Toyota Production System. Organization Science, 10 (1), 43-68.

Atuahene-Gima, K. (2005). Resolving the Capability - Rigidity Paradox in New Product Innovation. Journal of Marketing, 69 (4), 61-83.

Baden-Fuller, C., & Volberda, H. W. (1997). Strategic Renewal. International Studies of

Management & Organization, 27 (2), 95-120.

Bartlett, C. A., & Goshal, S. (1990). Matrix Management: Not a Structure, a Frame of Mind.

Harvard Business Review, 68 (4), 138-145.

Bartlett, C. A., & Goshal, S. (1994). Changing the Role of Top Management: Beyond Strategy to Purpose. Harvard Business Review, 72 (6), 79-88.

Belderbos, R., Faems, D., Leten, B., & Van Looy, B. (2010). Technological Activities & Their Impact on Financial Performance of the Firm: Exploitation & Exploration Within & Between Firms . Journal of Innovation Management, 27:869-882.

Benner, M. J., & Tushman, M. L. (2003). Exploitation, Exploration, and Process Management: The Productivity Dilemma Revisited. Academy of Management Review, 28, 238-256. Blindenbach-Driessen, F., & Van den Ende, J. (2014). The Locus of Innovation: The Effect of

a Seperate Innovation Unit on Exploration, Exploitation, and Ambidexterity in Manufacturing and Service Firms. Journal of Product Innovation Management, 31:5, 1089-1105.

Boston Consulting Group. (2015). Boston Consulting Group's annual survey on innovation. Retrieved from https://media-publications.bcg.com/MIC/BCG-Most-Innovative-Companies-2015-Nov-2015.pdf

Boumgarden, P., Nickerson, J., & Zenger, T. R. (2012). Sailing into the wind: Exploring the relationships among ambidexterity, vacillation, and organizational performance.

Strategic Management Journal, 33:, 587-610.

Brown, L. D., & Eisenhart, K. (1998). Competing On The Edge: Strategy As Structured Chaos.

Harvard Business School Press, Boston.

Burgelmann, R. A. (1991). Intraorganizational Ecology of StrategyMaking and Organizational Adaptation: Theory and Filed Research. Organizational Science, 2 (3), 239-262.

Exploring organizational ambidexterity : do large multi-unit firms balance exploration and exploitation across units and over time simultaneously?