heh Mirroring and Disruption - A Case Study of Nokia’s Decline Master of Science Thesis in the Management and Economics of Innovation Program CARL-JOHAN BLOMQVIST DAÐI SNÆR SKÚLASON MAGNUS SJÖLANDER Department of Technology Management and Economics Division of Innovation Engineering and Management CHALMERS UNIVERSITY OF TECHNOLOGY Göteborg, Sweden, 2014 Report No. E 2014:008 MASTER’S THESIS E 2014:008 Mirroring and Disruption A Case Study of Nokia’s Decline CARL-JOHAN BLOMQVIST DAÐI SNÆR SKÚLASON MAGNUS SJÖLANDER Supervisor: Christian Sandström, Ph.D. Department of Technology Management and Economics Division of Innovation Engineering and Management CHALMERS UNIVERSITY OF TECHNOLOGY Göteborg, Sweden 2014 MIRRORING AND DISRUPTION Carl-Johan Blomqvist Daði Snær Skúlason Magnus Sjölander © CARL-JOHAN BLOMQVIST, DAÐI SNÆR SKÚLASON & MAGNUS SJÖLANDER, 2014 Master’s Thesis E 2014: 008 Department of Technology Management and Economics Division of Innovation Engineering and Management Chalmers University of Technology SE-412 96 Göteborg, Sweden Telephone: + 46 (0)31-772 1000 Chalmers Reproservice Göteborg, Sweden 2014 Abstract The mobile industry is an ever changing and fast growing technology based industry that is very interesting to examine at this point in time due to the technological shift the industry has gone through in the recent years. This technological shift has caused a disruption in the industry and led to the demise of many incumbents as new firms entered the industry. We argue that the shift the mobile industry has gone through is not merely a technological one, but rather a paradigm shift from the old feature phone paradigm to the new smartphone paradigm. Further, this paradigm shift brings substantial changes; where the institutions and underlying logic as well as those competences and business models that are important differ between the two paradigms. Nokia, the Finnish mobile device manufacturer, is one of the incumbent firms that have lost out during this shift. The company maintained a market leading position until the middle of the first decade of the 21st century when its market share started to drop rapidly, especially in the high-end, smartphone segment of the market. Despite its once strong position, Nokia lost out to new entrants. In this report we examine Nokia’s case and try to unveil the most important factors behind the demise of this technology giant. In order to do so, a case study of Nokia’s development between 2003-2010 was conducted, where extensive empirical data was collected through interviews with former Nokia employees and industry specialists, as well as by utilizing extensive secondary data. To analyze the data collected and help explain Nokia’s case, a theoretical framework is constructed using existing theory. Theory from various research streams is used, including theory on the evolution of industries; industrial transformation, including disruptive innovation and resource dependency; an organization’s cognitive abilities and its abilities to change, including concepts such as dynamic capabilities and dualism; as well as ecosystems and networks, where concepts such as network effects, two-sided markets and institutions become important. We will argue that a hierarchy of factors contributed to Nokia’s downfall, but that the underlying and most important issue was the firm’s inability to understand and adapt to the new smartphone paradigm and the underlying changes it caused in the industry; namely, changes in what institutional logics were applicable. In the new smartphone paradigm, software as well as platform logic and the corresponding mechanisms of two-sided markets, were much more important. This new emphasis and focus in the smartphone paradigm was fundamentally different from that of the feature phone paradigm, upon which Nokia had built its strong competence of hardware development. We propose this as following naturally by an enlarged perspective on the mirroring hypothesis, which we argue can be used to better explain incumbents’, such as Nokia’s, path dependence along an industry. This opens up for new suggestions on how the incumbent’s curse can be viewed. Further research should investigate the suggestions made in this report and elaborate on the implications of this new perspective. Acknowledgements This master’s thesis research project was conducted during the fall of 2013 as part of the master’s program Management of Economics of Innovation. The research was conducted at the Department of Technology Management and Economics at Chalmers University of Technology in Gothenburg, Sweden. First of all we would like to express our gratitude to our supervisor, Christian Sandström, Ph.D., who encouraged us to pursue this thesis project and whose support, advice and consulting have proved invaluable in the process of this research project. Furthermore, we would like to thank Henrik Berglund, Associate Professor at Chalmers University of Technology, who provided valuable input to the ongoing discussion of this thesis. We would also like to thank all the respondents that agreed to take part in this research project. Without their input and extensive knowledge on both Nokia and the mobile industry in general, conducting this research project would not have been possible. The time these respondents have put into our research and their help in getting us in contact with further respondents has therefore been extremely valuable for us through the progress of this study. Gothenburg, Sweden – February 8, 2014 …………………………………….………… Carl-Johan Blomqvist …………………………………….………… Daði Snær Skúlason …………………………………….………… Magnus Sjölander Table of Contents 1.   Introduction ......................................................................................................... 1   1.1.   Purpose ..................................................................................................................... 2   1.2.   Scope and delimitations ........................................................................................... 2   1.3.   Report Structure ...................................................................................................... 2   2.   Method ................................................................................................................. 3   2.1.   Research Strategy and Design ................................................................................ 3   2.2.   Research Method and Data Collection .................................................................. 3   2.2.1.   Primary Data ...................................................................................................... 4   2.2.2.   Secondary Data .................................................................................................. 4   2.3.   Research Process ...................................................................................................... 5   2.4.   Validity and Reliability ........................................................................................... 5   2.4.1.   Construct Validity .............................................................................................. 6   2.4.2.   Internal Validity ................................................................................................. 6   2.4.3.   External Validity ................................................................................................ 7   2.4.4.   Reliability ........................................................................................................... 7   3.   A Theoretical Framework on The Incumbent’s Curse ................................... 9   3.1.   The Evolution of Industries .................................................................................... 9   3.1.1.   Patterns of Innovation ........................................................................................ 9   3.1.2.   Shifts and Discontinuities in Industry Evolution ............................................. 10   3.2.   An Internal Perspective ......................................................................................... 12   3.2.1.   Inherent Inertia and Path Dependency of Organizations ................................. 12   3.2.2.   The Mirroring Hypothesis ................................................................................ 13   3.2.3.   Coordination versus Cooperation ..................................................................... 13   3.2.4.   Dynamic Capabilities - How Firms cope with Change .................................... 14   3.2.5.   Breaking through the Cognitive Frame through Learning ............................... 15   3.2.6.   The Difficulties of Decision Making in Organizations .................................... 17   3.2.7.   Dualism and Matrix Organizations .................................................................. 18   3.2.8.   Entrepreneurship inside existing Profitable Organizations .............................. 19   3.3.   An External and Integrated Perspective ............................................................. 21   3.3.1.   Innovation and Value Networks ....................................................................... 21   3.3.2.   Effects and Risks of Innovation Networks ....................................................... 23   3.3.3.   Managing Innovation in Networks ................................................................... 24   3.3.4.   Network Forms and Institutions Role in Innovation ........................................ 25   3.3.5.   Resource Dependency of Incumbent Firms ..................................................... 26   3.3.6.   Business Model Innovation .............................................................................. 27   4.   Empirical Data .................................................................................................. 31   4.1.   An Industry Overview ........................................................................................... 31   4.1.1.   The Technology Shift ....................................................................................... 35   4.1.2.   The Value Network of the Mobile Industry ..................................................... 38   4.2.   Nokia’s Early History ............................................................................................ 40   4.2.1.   Sales and Growth ............................................................................................. 42   4.3.   Internal Factors ..................................................................................................... 42   4.3.1.   Organizational and Cultural Factors ................................................................ 43   4.3.2.   Devices and Hardware ..................................................................................... 48   4.3.3.   Software ........................................................................................................... 53   4.3.4.   Increased Focus on Mobile Services ................................................................ 56   4.3.5.   Business Model Evolution ............................................................................... 58   4.4.   External factors – Nokia and the Value Network ............................................... 59   4.4.1.   Towards the Mobile network operators ........................................................... 60   4.4.2.   New entrants ..................................................................................................... 62   4.5.   Summary ................................................................................................................ 64   5.   Analysis .............................................................................................................. 67   5.1.   Paradigm Shift ....................................................................................................... 67   5.1.1.   The End of the Feature Phone Paradigm .......................................................... 67   5.1.2.   Signs of a New Paradigm ................................................................................. 68   5.1.3.   Establishment of the Smartphone Paradigm .................................................... 69   5.2.   Internal Factors ..................................................................................................... 71   5.2.1.   Destruction of Core Competences .................................................................... 71   5.2.2.   Path Dependency along the old Paradigm ........................................................ 73   5.2.3.   Nokia’s Need and Attempt for Transformation ............................................... 73   5.3.   External Factors .................................................................................................... 75   5.3.1.   The Power of MNOs and their Relationship with Nokia ................................. 75   5.3.2.   Service Opportunities in the new Smartphone Paradigm ................................. 77   5.3.3.   The Importance of Platforms ............................................................................ 78   5.3.4.   The OSSO/Maemo/MeeGo Innovation Failure ............................................... 79   5.4.   Nokia’s Institutional Lock-in and Path Dependence .......................................... 80   5.4.1.   Each Paradigm as an Institution and Cognitive Frame .................................... 80   5.4.2.   Nokia’s lack of understanding the new Institutional Logic ............................. 82   6.   Discussion .......................................................................................................... 83   6.1.   What made Nokia Fail – A Story of Platforms ................................................... 84   6.2.   The Lack of a Deep and Complete Explanation ................................................. 86   6.3.   The Troubles of Breaking the Path of Existing Assets ....................................... 88   6.4.   A Suggested Institutional Lock-In underpinning Disruptive Innovation ........ 89   6.5.   The Story of Nokia and Institutional Lock-Ins ................................................... 90   6.6.   Hierarchies of Contributing Factors .................................................................... 92   6.7.   Limitations of This Study ...................................................................................... 93   6.8.   Further Research ................................................................................................... 95   7.   Conclusions ........................................................................................................ 97   References ................................................................................................................. 99   Appendix A – Respondents ................................................................................... 111   List of Figures Figure 1 – Different styles and ways to balance matrix organizations (Degen, 2010) ................ 18   Figure 2 – Relations between the different factors of Osterwalder’s business model canvas (Chesbrough, 2010) ............................................................................................................ 28   Figure 3 – Global market share of mobile device manufacturers from 2003 until 2010 (Gartner, 2005, 2007a, 2009a, 2011) .................................................................................................. 32   Figure 4 – Market share of smartphone platforms from 2007 until 2012 (Gartner, 2009b, 2011, 2012a, 2012b, 2012c, 2013) ................................................................................................ 37   Figure 5 – Nokia’s revenue divided by business segment from 1967 until 1998 (Ali-Yrkkö, 2001) ................................................................................................................................... 41   Figure 6 – Nokia’s net sales and operating profit between the years of 1996 and 2000 (Nokia, 2001a) ................................................................................................................................. 42   Figure 7 – Nokia’s organizational structure, implemented in the 2004 reorganization (Nokia, 2004) ................................................................................................................................... 44   Figure 8 – Nokia’s net sales by departments/business units from 2001 until 2010 (Nokia, 2002a, 2003b, 2004, 2005a, 2006b, 2007b, 2008a, 2009b, 2010a, 2011a) .................................... 47   Figure 9 – Contribution to Nokia’s operating profit of different departments/business units from 2001 until 2010 (Nokia, 2002a, 2003b, 2004, 2005a, 2006b, 2007b, 2008a, 2009b, 2010a, 2011a) ................................................................................................................................. 48   Figure 10 – Number of different device models released annually by Nokia between 1999 and 2010 (Wikipedia, 2014c) .................................................................................................... 50   Figure 11 – Global market share of converged versus other devices (Nokia, 2007b, 2008a, 2009b, 2010a, 2011a) .......................................................................................................... 51   Figure 12 – Market size of converged device market segment and Nokia’s share in the market (Nokia, 2007b, 2008a, 2009b, 2010a, 2011a) ..................................................................... 52   Figure 13 - Nokia’s net sales divided by geographical markets (Nokia, 2006b, 2007b, 2008a, 2009b, 2010a, 2011a) .......................................................................................................... 52   Figure 14 - Devices sold versus applications available in native application stores (Vakulenko et al., 2011) ............................................................................................................................. 58   Figure 15 – Nokia’s operating profit and net sales from 2001 until 2012 (Nokia, 2001a, 2002a, 2003b, 2004, 2005a, 2006b, 2007b, 2008a, 2009b, 2010a, 2011a, 2013a) ........................ 65   List of Tables Table 1 – Comparison between the Feature Phone Paradigm and the Smartphone Paradigm .... 81   1 1. Introduction The study of economic progress and development has long intrigued researchers and dates back to the dawn of modern economics (e.g. Smith, 1776). Ever since, much discussion has been made on how economic progress and development should best be described and characterized, in what context it belongs, and in particular, what factors drive this progress and what consequences can be expected. In contrast to neoclassical economics, often regarded as being too simplistic and static to explain economic progress, Schumpeter (1942) put forward his theory on how economic progress can be explained as waves of new technology and innovation. Schumpeter hailed the entrepreneur as the driver behind new cycles of innovation, driving economic progress through a process of creative destruction; a process where the creation of new value through innovation triumphs existing product and services, and ruins the old, existing paradigm of what is considered valuable. Since then, many other have of course contributed valuable theory to explain further details of economic progress and related phenomena. One area within the topic of economic progress that has gained a lot of attention from researchers is the discontinuity in between the business cycles Schumpeter (1942) tried to explain. In particular, researchers have put significant attention to the phenomena where incumbent firms fail to properly respond to competition from new entrants in the face of technological change – known as The Incumbent’s Curse. Researchers have published various different theories on the matter, taking different approaches to explain why incumbent firms are unable to respond to discontinuous innovation that originates from new entrants (e.g. Bower & Christensen, 1995; Christensen, 1997; Christensen & Bower, 1996; Christensen & Raynor, 2003; Danneels, 2004; Tripsas, 1997; Tripsas & Gavetti, 2000). One of the most popular explanations is Clayton Christensen’s theory of disruptive innovation, which has gained widespread popularity in both the academic and professional world (Bower & Christensen, 1995; Christensen, 1993, 1997; Christensen & Bower, 1996; Christensen & Raynor, 2003). Technology industries are interesting to study in this context, as multiple waves of creative destruction can be witnessed in a short time due to the fast pace and volatile nature of these industries. The mobile industry, being a ever changing and fast growing, technology based industry, is especially interesting at this point in time due to the technological shift the industry has gone through in recent years. This shift disrupted the industry and led to the demise of many incumbents in the industry as new entrants entered the industry, many of whom from adjacent or similar industries (Hacklin et al., 2013). Nokia, the Finnish mobile device manufacturer, is one of these incumbent firms. The company had maintained a market leading position until the middle of the first decade of the 21st century when its market share started to drop rapidly, especially in the high- end, smartphone segment of the market. Until Nokia’s demise, the company was seen as being very innovative and a clear technological leader in the industry. Furthermore, 2 researchers such as Christensen had even predicted that Nokia would not loose out to new entrants and succumb to the Innovator’s Dilemma (McGregor, 2007). Despite its strong position, Nokia lost out to new entrants who emerged as the mobile industry converged with adjacent digital industries. The company did not only fail to beat these new entrants in bring market leading innovations to the market in a timely manner, but also failed to respond appropriately to threats posed by the innovations produced by its new competition. In this research project we will examine Nokia’s case, trying to identify the most important factors behind the demise of this technology giant. Furthermore, we will suggest a new perspective to examine the phenomenon that is The Incumbent’s Curse, which we believe serves better than existing theory to explain the underlying factors for Nokia’s demise. 1.1. Purpose The purpose of this research project is to examine the strategic decisions of Nokia’s mobile phone business in order to explain why Nokia lost their position as a market leader in the mobile industry. In order to do so, extensive empirical data was collected through interviews with former Nokia employees and industry specialists, as well as by utilizing extensive secondary data from various sources. Furthermore, a theoretical framework is constructed to analyze the empirical data collected and ultimately fulfill the purpose of this research project. 1.2. Scope and delimitations We limit the scope of the research to the years 2003 to 2010 as important strategic decisions were made to Nokia’s organization in both 2003 and 2010. In 2003 Nokia took a decision to change its organizational structure to better align with a new focus on multimedia and enterprise solutions. In 2010 significant changes to Nokia’s strategy started to emerge when Stephen Elop took over as President and CEO. In addition, we will focus our analysis on the part of Nokia’s business that is responsible for mobile devices, but not Nokia’s Network business unit (later Nokia Siemens Networks). Data on this business unit will be presented in some places to give a more holistic picture of Nokia, but it will not be considered when analyzing the findings of this research project. 1.3. Report Structure This report starts with an overview of the method used during the process of this research project. Following that, the theoretical framework utilized in the research is constructed using a wide range of theory that relates to evolution of industries and The Incumbent’s Curse. When the theoretical framework has been constructed we present the empirical data collected through the process of this research project, both primary data collected through interviews and secondary data gathered from various reliable sources. The theoretical framework is then used to analyze the empirical data and shed light on the strategic decisions and actions that led to Nokia’s fall from a market leading 3 position. Finally, the results of the project are discussed and concluded, and a new perspective from which to examine the fall of incumbents is suggested. 2. Method This section will describe and discuss the strategy and structure of this research and what research design was employed during the process of the research. The process itself and how the research progressed will be discussed, and an overview of what methods were used for data collection will be given. Finally, a discussion on the quality of this research will follow. 2.1. Research Strategy and Design Defining an appropriate strategy to be used is an important starting point of every research project and the decision of what research strategy to use is very dependent on the research project being undertaken. There is no “one-size fits all” solution when it comes to selecting a research strategy, as different strategies will be more appropriate to use for certain projects than others (Cepeda & Martin, 2005). Bryman and Bell (2011, pp. 26-27) discuss the advantages of distinguishing between quantitative and qualitative research strategies, not only on the level of what types of data is collected but also on the level of how the researcher views the world. In that sense, a qualitative research strategy generally employs the view that social reality cannot be separated from the individual and that truth is subject to the interpretations of the researcher (Bryman & Bell, 2011, p. 27; Cepeda & Martin, 2005). Quantitative research, on the other hand, has a more objective view on these matters, taking a perspective commonly connected to the natural sciences. The approach taken in this research project is a qualitative one, since both the accounts of the people interviewed in this study and the interpretation and analysis done by the researchers is subjective and cannot be separated from the social world. Further, the research design that is employed in this research is that of a case study, with Nokia as the focal firm. The qualitative case study is considered a suitable option when it comes to business research as it takes on the “why” and “how” questions of how complex events and processes unravel (Cepeda & Martin, 2005). Yin (2009, p. 4), further argues that the case study research design allows researchers to retain a holistic view to fully explain complex social phenomena. The qualitative case study has increasingly been established as a valid, high-quality research method and its importance and usefulness for theory building have further been enforced (Cepeda & Martin, 2005; Christensen, 2006; Yin, 2009, pp. 6-8). 2.2. Research Method and Data Collection A combination of both primary and secondary data was collected in order to clarify the case of Nokia and thereby fulfill the purpose of this research project. Here will follow a detailed account of how each group of data was collected and what research methods were used for said data collection. 4 2.2.1. Primary Data The primary data used in this research project was collected by the means of semi- structured interviews, a common method for gathering data in qualitative research (Bryman & Bell, 2011, pp. 446-447). In semi-structured interviews, the interviewer uses an interview guide to structure the conversation, but gives the interviewee leeway to answer as he pleases, allowing the interviewer to follow up on interesting details that might arise (Bryman & Bell, 2011, p. 446). These aspects of semi-structured interviews make them a perfect fit for an exploratory study like this one, as they allow researchers to get a wide and holistic view of the phenomena being studied. Prospect interviewees were selected from a set of former Nokia employees that were part of the company during the period from 2003 till 2010. A special emphasis was placed on finding managers and executives, as these are most likely to have a good insight into Nokia’s strategy and the decisions made during the period. Apart from Nokia employees a set of industry specialists, academics that have focused on the industry, as well as representatives of the mobile network operators (MNO) were located as potential interviewees. This set of prospect interviewees was contacted by email and interviews scheduled with those willing to participate in the study. Further interviewees were then identified and contacted through snowballing, a sampling method where interviewees are used to get in contact with other prospective interviewees (Bryman & Bell, 2011, p. 491). This technique proved successful and resulted in interviews with people that had served as high-level managers within Nokia, as well as with other knowledgeable actors from the industry. A total of 15 interviews were conducted with 14 unique interviewees, 10 of whom were former Nokia employees. Apart from one interview conducted in person, all interviews were conducted through telephone or Skype. The length of the interviews ranged from 30 minutes to approximately two and a half hours, with the most common length being around one hour. The people interviewed for this study together possess a broad and deep understanding of Nokia’s strategy, organization and the decisions made within the company during the focal period. Furthermore, the fact that all Nokia employees interviewed are not currently employed by Nokia may strengthen their credibility, as their employment therefore does not restrict their testimony. A complete list of interviewees along with a short description of them and the relevant roles they have held can be found in Appendix A. Some interviewees wish to remain anonymous and details about their person will therefore not be published. 2.2.2. Secondary Data Apart from the primary data provided by the semi-structured interviews conducted, a number of secondary data sources were also utilized. These include press releases from Nokia, Nokia’s annual reports (Form 20-F), as well as other public documents and financial statements from the company. All this data is publicly available and was retrieved from Nokia’s website. Furthermore, information from industry analysts, news 5 articles, websites of other actors within the mobile industry (such as other device manufacturers and MNOs), as well as various other Internet sources was also used. When selecting sources for secondary data, the credibility of the data was a primary concern and the highest quality sources available were always selected. 2.3. Research Process The initial motive behind this research has its roots in previous unpublished work we have done on The Incumbent’s Curse and the case of Nokia. This, in combination with encouragement from Henrik Berglund, Associate Professor, and our supervisor, Christian Sandström, Ph.D., sparked a special interest in further exploring Nokia’s fall from a market leading position. The process that this research project has adhered to can be described as a combination of two concurrent iterative processes; one of data collection, discussion and analysis, and one of constructing a theoretical framework to explain and make sense of the data collected. As the data collection progressed iterative discussions and analysis of Nokia’s situation and the reason behind their downfall took place. These discussions, along with insights from the theoretical framework, were used to further explore important topics in further interviews. Furthermore, discussions with Mr. Sandström and Mr. Berglund helped us uncover new literature and theories to investigate, and to guide us in the right direction. As the research progressed, the analysis and data collection became deeper and more focused on specific areas that were identified as being more important than others to explain Nokia’s fate. One could argue that the processes followed a funnel-like pattern, starting by looking at Nokia’s situation from a broad perspective, but then focusing in on the factors that were deemed the most important. The data collection process was continued until the data retrieved from the interviews became predictable to the interviewers; that is, when same or similar answers to questions were received from multiple interviewees, it was assumed that sufficient knowledge had been gathered on the topic. Of course a higher number of interviews and a larger data set can help produce more accurate results, but due to the time constraints and the difficulty of getting in touch with interviewees, the aforementioned method was deemed sufficient. When the data collection process was deemed completed, further analysis and discussions were conducted to identify the most important factors behind Nokia’s demise. These factors were then expanded upon and put in context with the theoretical framework, formulating a precise explanation for this report. 2.4. Validity and Reliability An important part of every research project is establishing credible criteria of the quality of the research itself. As Bryman and Bell (2011, pp. 394-399) discuss, there has 6 been considerable debate on how the quality of qualitative research should be assessed and researchers have proposed various ways of doing so. Some researchers argue that different criteria should be used for qualitative than quantitative research, because the qualitative approach is inherently different from the quantitative one (Bryman & Bell, 2011, pp. 394-399), while others argue that the classic criteria of validity and reliability, which originates from quantitative research, can be applied largely unchanged to qualitative research (LeCompte & Goetz, 1982; Mason, 2002). Furthermore, Yin (2009, p. 40) describes measures with which the quality of case study research can be assessed. The quality of this research project will be assessed with the concepts of validity and reliability as explained by Yin (2009, pp. 40-45). As covered in Bryman and Bell (2011, p. 395), the application of these concepts to assess the quality of qualitative studies is well established and the terms are applied in a very similar manner as they are applied in quantitative research. Following these criteria, we will inspect the reliability as well as the construct, internal and external validity of this study. 2.4.1. Construct Validity Construct validity refers to how well the data gathered in a study represents the concepts or phenomena they are gathered to represent; that is to say, whether a certain piece of data does in fact give any information about the concept or phenomena it was intended to represent. As inherent in qualitative research, the primary data gathered represents the views and thoughts of the interviewees and must be handled as such. However, the relatively high number of interviews and the consistency of results across interviews help strengthen the construct validity of this study and indicate that the data collected does in fact represent the concepts and phenomena it is supposed to. Furthermore, the extensive secondary data collected helps build a complete picture of the case of Nokia, further strengthening the construct validity of the study. 2.4.2. Internal Validity The internal validity of a study is important an important measure that is used to assess whether or not causal relationships between the concepts or phenomena studied have been established. To ensure internal validity of this study various sources of data were used to gather extensive information about the case of Nokia in the timeframe studied. This creates a thick description of Nokia during that time which is further supported by the consistency of the data gathered. Efforts to establish correct timing of events can further help increase internal validity of a study, as timing of events is an important factor in establishing causality. From time to time it proved difficult to establish exact timing of some events, which might have impacted internal validity negatively. 7 2.4.3. External Validity The concept of external validity describes how well the results of a study can be generalized to explain similar situations in other social contexts. Case studies are often considered difficult to generalize due to their small sample size, although some authors have argued this concern is ungrounded (e.g. Christensen, 2006; Yin, 2009, pp. 43-44). Even so, the focus in this study has been primarily to explain the case of Nokia. However, since many firms in the mobile industry suffered similar fate as Nokia, the results may perhaps be used to explain a more general development in the mobile industry and possibly other fast moving, technology based industries, although further research is required to confirm this. 2.4.4. Reliability The reliability of a study refers to the degree to which it can be replicated by other researchers. As discussed by Bryman and Bell (2011, p. 395), qualitative research is always difficult to replicate since the social settings affecting the research change with time. That is to say, the current state of social reality, in particular that of the mobile industry, greatly affects the testimony of interviewees and therefore this study’s main data source. Should the same interviews be conducted later in time, it is not necessarily certain that the same data would be gathered even if the interviews remained otherwise identical. Furthermore, the fact that all primary data was gathered from semi-structured interviews, which allow for considerable leeway in terms interviewee response, further lowers the reliability of this study even though the reliability of the secondary data used in the study can be considered high. However, as the interviews were all recorded the lower reliability of the semi-structured interviews could be partly mitigated by reusing the data collected. 8 9 3. A Theoretical Framework on The Incumbent’s Curse In this section a theoretical framework will be constructed in order to explain why Nokia lost its market leading position in the mobile industry. A broad selection of theory will be used to achieve this, including but not limited to research on The Incumbent’s Curse, organizational theory as well as on ecosystems and value networks. We will start by exploring how industries evolve and how innovation can cause shifts and discontinuities in industries. We will then examine the internal factors that affect innovation in organizations and how they contribute to success in the face of industry change and discontinuous innovation. Finally, we will look into the external factors that affect a firm’s success in the face of industry discontinuities. 3.1. The Evolution of Industries As put forward in the introduction, economic progress is often described according to Schumpeter’s (1939, 1942) waves of innovation. These waves give the fundamental explanation to why economic systems cannot solely be explained by assuming a steady state or equilibrium conditions, such as often is done in the neoclassical economics research stream (Nelson & Winter, 2002). In contrast to equilibrium conditions, Schumpeter (Schumpeter, 1939) continued on the work on business cycles, such as Kondratieff waves (Garvy, 1943; Kondratieff, 1979), and proposed a structure where longer cycles consists of many smaller cycles. Further, Schumpeter (Schumpeter, 1942) suggested that these cycles are caused by innovation, which leads to temporary imbalance in favor of the innovator, causing a new wave of progress to strike through and subsume previous waves and innovations. 3.1.1. Patterns of Innovation Abernathy and Utterback (1978) proposed each business cycle to be characterized as following along a specific pattern, often referred to as the pattern of innovation. Abernathy and Utterback (1978) argued that each cycle follows an evolutionary, continuous, pattern (also suggested by many others, e.g. Hamilton & Singh, 1992; Nelson & Winter, 2002; Perez, 1985; Sahal, 1985), going through three major phases: the fluid, transitional and specific (Abernathy & Utterback, 1978). In the beginning of a cycle uncertainty is high, the underlying market need is still to be explored and it is still unknown how big the potential of the new market and/or technology is. Thus, focus is in this phase on product innovation and a big variety of products are introduced. As the industry starts to realize the true market need and how to best satisfy it, the variety of products in the market starts to decline. This is often referred to that a dominant design has been reached. This in turn triggers a shift in focus from product innovation to efficiency in the underlying processes and lowering cost becomes the driving force in the industry. In general, innovation declines in general in this last phase. In the fluid phase of a business cycle, i.e. the early stage, the number of actors increases as there is yet to be a dominant design and thus, a clear winner (Suárez & 10 Utterback, 1995). However, as the dominant design is set and the industry matures, a shakeout period begins where a few strong firms manage to outcompete their competitors. Two explanations are given to why this happens; either because of a specific event which triggers stiff competition, or because the gap between the stronger and the weaker actors widens along time (Klepper & Simons, 2005). A specific event can be that of a specific dominant design being set, leaving actors that have chosen to go down a different path in the industry or actors that have not yet entered in a predicament, as they are behind along the new dominant trajectory of the industry (Suárez & Utterback, 1995). In the second explanation the shakeout follows naturally by the force of competition and feedback loops, such as described by Forrester (1968), since stronger actors become stronger and weaker actors weaker. Thus, when the industry progresses and matures weaker actors will be forced to leave. In sum, the industry will throughout its first period not be characterized by strong growth in market size since the market need is still to be explored. Only when the market need is starting to get established, real growth will happen, referred to as the transitional phase by Abernathy and Utterback (1978). In the end of the life cycle, the specific phase (Abernathy & Utterback, 1978), growth will decline since the market need will have been saturated and any additional enhancements along the existing technology trajectory will no longer yield as much value to the market (Hamilton & Singh, 1992). In other words, the market growth of an industry will in general follow an S-curve shape (e.g. as described by Hamilton & Singh, 1992; Perez, 1985; Sahal, 1985). 3.1.2. Shifts and Discontinuities in Industry Evolution Albeit Abernathy’s and Utterback’s (1975) description of the evolution of an industry, as well as many contributions that build on top of their theory, provides a great insight into the evolution of an industry, it does not fully address shifts in between the business cycles. The Schumpeterian new wave of innovation that will eventually subsume the old paradigm does play a vital role in explaining the rise and fall of new firms as well as incumbents. Dosi (1982) added insight by proposing paradigms in technology and innovation as similar to what had been previously argued as paradigms in science by Abernathy and Utterback (1962).The proposal was very much in line with the Schumpeterian, evolutionary research stream, as described above. According to the theory, there is an inbound force keeping development along a certain path. The two mechanisms of evolution, variety and selection, which Dosi (1982) proposed could combine the two perspectives of innovation emerging from a technology push or from a market pull, have an induced constraining effect upon the development. A variety of technological development possibilities provide a foundation for development paths, which will eventually become filtered through the evolutionary selection provided by the market and its true need and preferences. In line with Abernathy and Utterback (1975) description of the pattern of evolution of an industry, within a paradigm this will eventually lead to a development looking similar to that of a cone. In the early phases of a paradigm or industry, variety will be large since uncertainty will be high, as the gap 11 between the technological possibilities and the needs and desires of the market is big. With time, variety and uncertainty will get reduced. In addition, Dosi (1982) argued technology push and market pull alone cannot fully explain the development of paradigms, and further, shifts in between paradigms. Institutional factors such as firm organizational structure, incentives and goals, government policy, as well as the fundamental Schumpeterian economic drive towards new revenues and profits, need also to be considered. This is furthermore also in line with the above description of industry maturity eventually leading to declining growth and thus pushing actors towards new revenue streams (Mensch et al., 1981). This is also well in line with the notion of “techno-economic paradigms” which emphasizes the external and institutional forces as presented by Perez (1985), although the focus of the concept “techno-economic paradigms” is on longer and larger business cycles compared to Dosi (1982). Through a process similar to that described earlier of variety and selection, there will be an institutional drive towards finding the next paradigm; the next wave of innovation that will supersede the previous paradigm (e.g. Dosi, 1982; Mensch et al., 1981; Nelson & Winter, 1982; Perez, 1985). This has several consequences since there will naturally be two paths for any firm; to optimize development along an existing paradigm or search for the next one. Since the existing paradigm has technological development building upon previous development within the paradigm, further steered by the market need within the paradigm as well as existing institutional factors, a mere cumulative development (on top of the current leading offering) is needed to increase the revenues and profits. On the other hand, in order to create a new paradigm, technology needs to be developed as well as the ideas of how the technology fits the market along new, uncertain and different paths from previous ones. Such an example can include a new dominant base for value in the industry, for example the replacement of steam engines by combustion engines (Clark, 1985). Usually, this is a larger jump since the new technology often starts farther behind the existing paradigm as investments tend to have been focused upon existing paradigms, as they have for long provided the less uncertain base for development, and thus revenue and profits, within the industry. In total, and as summarized by Kaplan and Tripsas (2008), an industry will develop along a frame that guides the development, diffusion and eventual discontinuity and disruption. This frame will emerge through the actors within the industry, such as markets, i.e. users, producers, governments, universities and similar, in interplay between the actors. The development will start off with large uncertainty and variety, but will transform to a more clearly defined and narrow frame as development progresses. Eventually, a new industry will emerge, as diminishing returns and a focus on cost will reduce the attractiveness to stay along the existing path. To summarize, both internal and external factors play an important role in explaining discontinuities and disruptions in industries (Afuah & Bahram, 1995). 12 3.2. An Internal Perspective Building on top of Schumpeter’s (1950) notion of “creative destruction”, Tuschman and Anderson (1986), argued innovations can either be incremental improvements or technological discontinuities, based on whether they are competence enhancing or competence destroying to the firm. They argued incremental innovations are built upon existing technological frames and enhance the existing competence the firm has built up around it - thus enhancing and furthering the already pre-existing competence within the firm. On the other hand, discontinuous innovations are built upon new technological trajectories and paradigms. Since these new technological trajectories subsume the previous trajectories, and further often rely upon a different base of competency, the value of the old competence will be destroyed. However, it might also be built upon a technological trajectory that exists within the firm, and hence, there can be a technological which disrupt an industry because it is competence destroying, but also discontinuous innovation that do not disrupt an industry because it is merely competence enhancing (Abernathy & Clark, 1985; Tushman & Anderson, 1986). 3.2.1. Inherent Inertia and Path Dependency of Organizations Tuschman and Anderson (1986) showed competence destroying innovations as being more likely to be introduced by new entrants than incumbents, in comparison to competence enhancing innovations which are more likely to be introduced by established players. They argued this is due to that “liabilities of age and tradition constrain existing, successful firms” (Tushman & Anderson, 1986, p. 461) – i.e. path dependency. This is well in line with other researchers who have argued and presented a similar relationship (e.g. Abernathy & Clark, 1985; Abernathy & Utterback, 1975; Christensen & Rosenbloom, 1995; Clark, 1985; Dosi, 1982; Henderson & Clark, 1990; Teece & Pisano, 1994), although researchers have since further explained what underlying factors cause this path dependency. Important aspects to take into account are organizational factors such as legacy, capabilities and structure, when explaining the path dependence of firms. As pointed out by Leonard-Barton (1992), capabilities of the firm that are of strategic importance, often referred to as core capabilities, can easily constrain the firm and become core rigidities. Along the development of an organization, a subset of its skills and knowledge base, technical systems, managerial systems and its values and norms can form a set of capabilities that become the essence of the competitive advantage the firm enjoys vis-à- vis its competitors. In other words, certain routines and processes within the firm can form a base for the firm, which if focused on can become the underlying power of the firm driving its continued ability to differentiate. However, as an organization grows and develops, procedures and routines might also be limiting for the firm in finding new sources of future competitive edges (Nelson & Winter, 1982), i.e. they become core rigidities (Leonard-Barton, 1992). Hannan and Freeman (1984) explain this by examining the evolution of an organization. They argue 13 an organization cannot be fully explained by the rational motives of the individual members in the organization. Rather, their argument is based upon an interplay between the organization and its environment. The organization exists in large because of actors’ willingness to support it and allow it to exist, such as that of investors and employees. These actors, regardless if acting inside the organization or in the environment of the organization, expect the organization to produce certain output. However, it is difficult to control how well the organization aligns with the goal to produce output. Hence, these actors will want the organization to strive for reliability when it comes to quality, timeliness, and accountability. This will push the members of the organization to choose paths along the certain and incremental path, often along existing paradigms. Naturally, however, it will often also constrain the organization from discontinuous changes. As argued by Hannan and Freeman (1984), this can become a severe problem for the organization when major changes in the environment destroy the underlying value of the existing organizational structure, such as when competence destroying innovation is introduced in an industry. 3.2.2. The Mirroring Hypothesis Organizational inertia is relevant in explaining the constraints and limitations of a firm in developing discontinuous innovation because the organization and its structure are directly linked to the ability of the firm to efficiently produce, as well as develop new products (Baldwin, 2007; Colfer, 2007; Henderson & Clark, 1990). In other words, in order for a firm to introduce discontinuous, competence destroying innovation, major changes will be required within the organization. This interdependency, between the product and the organization is often explained by linking the architecture of the product to that of the structure of the organization, referred to as the mirroring hypothesis (e.g. Baldwin, 2007; Colfer, 2007). The mirroring hypothesis claims to explain how an organization balances modularity with integration - both in organizational structure and product architecture (Baldwin, 2007; Colfer, 2007). High modularity is attractive to decrease complexity by hierarchically structure a product’s architecture. In a highly modular architecture, each specific part or task does not need to know and understand the internal workings of another specific part or task, i.e. module, only how to interface with the specific modules it needs to interface with. This can be effective when bounded cognition is limiting actors’ ability to grasp all aspects of the whole (Williamson, 1991). On the other hand, high modularity may hamper the ability of an actor to develop and innovate across different modules because of the limited understanding he possesses, which may be needed for discontinuous innovations. In such a scenario, a more integrated architecture and structure may be needed. 3.2.3. Coordination versus Cooperation In essence, the underlying issue described by the mirroring hypothesis is the one of coordination versus cooperation (Colfer, 2007). Two different thought streams, 14 transaction cost economics (TCE) and knowledge based theory (KBT) (Baldwin, 2007; Colfer, 2007), further elaborate on the underlying fundamentals of the issue of coordination and cooperation. Colfer (2007) neatly summarizes respective school of thought’s arguments when examining why coordination within a firm is advantageous compared to coordination in between firms: “The TCE perspective focuses on the hazards of opportunism. Here, the key problem is how to align the potentially conflicting interests of contributors so that their exchanges can take place safely—that is, without much risk of opportunistic behaviors like the withholding of valuable information and materials. […] the key benefits of collocating contributors within a firm are the firm’s superior capacity (relative to the market) for (1) incentive alignment, (2) conflict resolution, and (3) performance monitoring […] In contrast to the TCE approach, the KBT perspective stresses the hardships of bounded cognition over the hazards of opportunism. Here, the key benefits of collocating contributors within a firm are the firm’s superior capacity (again, relative to the market) for (1) central planning (Alchian and Demsetz, 1972) and (2) rich, contextual, bilateral communication” (Colfer, 2007, pp. 15-16). Thus, the coordination vis-à-vis cooperation issue in regards to optimizing innovation output is best solved by optimizing the organizational structure based upon how integral vis-à-vis modular the product architecture is, due to integral product development requiring “(1) extensive communication and exchange among the individuals who perform them, as well as (2) efficient resolution of disputes arising from the individuals’ differing technical perspectives, product knowledge, and/or self- interests.” (Colfer, 2007, p. 16). In other words, in order to develop innovation requiring a new product architecture, such as many times disruptive innovation, an organization model based upon a modular architecture will face difficulties because it needs to shift into a more integral structure. This will be costly and inherently difficult, as there will be organizational inertia due to the pressure from its actors to strive towards reliability and accountability as presented earlier, but also because it will require management apt to handle a different, more integrated organizational structure. 3.2.4. Dynamic Capabilities - How Firms cope with Change The ability, and capability, of firms to change when facing external environmental changes is thus critical, since it will be a baseline for the firm’s capability to innovate which hence will be needed to remain in the competitive game. Teece and Pisano (1994) labeled this the dynamic capability of the firm, when arguing this to be critical for a firm’s continued success, in particular through Schumpeterian waves of creative destruction. In their perspective, competitive advantage is built up by certain capabilities within the firm that are valuable to the customer, but scarce and difficult to imitate or substitute for other actors. In a broader sense, in order for a firm to not have its profits driven down by the market forces of its competitors, the firm needs to enjoy one or many capabilities with which it can appropriate value from. Thus, in order to build a sustainable advantage, the underlying processes and routines inside the firm that build up the valuable capability must not be easily copied. 15 However, if sustainable advantage are built upon not easily copied capabilities, that posits those capabilities must be rather tacit, since otherwise they are likely to actually be copied and thus diminish in value to the original firm where the capability was first developed (Teece & Pisano, 1994). This implies that the capabilities are not easily understood. In fact, according to Teece and Pisano (1994) the underlying capability a firm is gaining its competitive advantage from is often not understood by any actor in its industry, including the firm having the capability itself. This can become problematic for the firm since even a well-protected competitive advantage will eventually dissipate through new waves of innovation. Hence, the firm will need to transform and reconfigure its capabilities in order to stay ahead of, or even to keep up with, its competitors innovative capabilities, but this might be difficult if the firm does not understand what it is that have gained its previous advantage in the industry that needs to be changed (Argyris & Schön, 1999). In other words, albeit difficult, the firm will need to have dynamic capabilities in order to remain competitive through innovation and a changing environment. The organization will need managerial and organizational routines and processes for how to transform and reconfigure its resources and assets, as well as learn, coordinate and adapt the level of integration within the organization according to changes in the environment. This will, arguably, make the organization able to, based upon its previous historical path, shift its direction towards new avenues of innovation and profits, in comparison to eventually become benighted by other more innovative competitors. Naturally, core competences and core capabilities are often considered for reconfiguration when facing new waves of innovation. This reflects the position of the firm and it's business assets, including technological, financial and locational assets (Teece & Pisano, 1994). However, a firm may also survive through creative destruction through utilizing complementary assets, and specialized complementary assets (Harrison et al., 2001; Teece & Pisano, 1994; Tripsas, 1997). Complementary assets; assets that are creating value in combination with other products, services and offerings, can be used to form alliances with partners, which when joining assets can be used to create new value. Specialized complementary assets, on the other hand, are assets that protect existing revenue streams through their importance (Tripsas, 1997). This is an important distinction, as complementary assets in and by themselves do not protect against new waves of creative destruction, where specialized complementary assets may. In sum, not only core competences and core capabilities can be reconfigured through dynamic capabilities for withstanding threats of competence destroying innovation, but also complementary assets. 3.2.5. Breaking through the Cognitive Frame through Learning The need for learning is critical for the change needed to develop disruptive innovation, not only due to its part in a firm having dynamic capabilities (Teece & Pisano, 1994), but also because it can form a cognitive barrier keeping the firm in the old and dying paradigm (e.g. Argyris, 1976; Dosi, 1982; Henderson & Clark, 1990; Tripsas & Gavetti, 2000). Thus, in order to avoid the path dependency and inertia often 16 negatively correlated with the capability of a firm to achieve disruptive or discontinuous innovation, it becomes necessary to unlearn in order to break the old cognitive frame, and learn a new one (e.g. Kaplan & Tripsas, 2008). Some researchers even attribute learning to be the core of the innovation process in achieving disruptive innovations (Assink, 2006). Learning can take place at different levels; on an individual level as well as on an organizational level, which are often discussed by researchers in conjunction (see for example (see for example Argyris, 1976; Robinson, 2001). Two strands of research on organizational learning exists; a descriptive and a normative strand (Robinson, 1995). The descriptive strand focuses on how organizational learning occurs, and the normative strand focuses on how organizational learning can be directed to make firms achieve their targets and goals faster (Robinson, 2001). Argyris (e.g. 1976) and their research on the topic combines both strands, and tries to further the knowledge on how practitioners can achieve organizational learning through different processes and what underlying factors affect different processes, as well as propose how it can be done efficiently within organizations. Based upon Argyris (1976, 1995); Argyris and Schön (1989, 1999) research on more than 300 organizations they argue it is important to break the cognitive frame that typically surrounds the first level of learning; what they refer to as double loop learning - in contrast to single loop learning. In single loop learning there is an established frame, which sets the context for the specific learning. In comparison, double loop learning does not only focus upon the first loop of learning, but on an additional loop where the cognitive frame is questioned. Further, Argyris (1976) found throughout his studies that most firms do not apply double loop learning when looking at what actions were actually taken within the studied firms; their theory-in-use, in contrast to their espoused theories of action; what actions the studied firms thought they were taking. This can become a problem, as the actors within a firm hence might believe that there does not exist any cognitive frame withholding discontinuous innovation. Argyris (1976) arguments, based upon his extensive empirical studies, are quite similar to previously presented theory. For example, inside a highly modular organization, learning within a specific module can be seen as being single loop learning. Further, if also questioning the cognitive frame, such as the goal of the specific module and thus the whole architecture, this can be seen as double loop learning. Then Argyris (1976) empirically based suggestion that double loop learning is not often the actual theory-in-use, can also be considered to be along the lines of other researchers suggestions; that new entrants without an existing cognitive frame are often the ones to introduce disruptive and discontinuous innovations rather than incumbents (Christensen & Rosenbloom, 1995; Henderson & Clark, 1990; Kaplan & Tripsas, 2008). 17 3.2.6. The Difficulties of Decision Making in Organizations Argyris (1976) suggested reasons why firms have difficulties breaking through the cognitive frame is also very much in line with previous outlined theory and other researchers. Complete information is not given nor cheaply acquired - in particularly when complex decisions such as those on potentially disruptive and discontinuous innovations are to be made (Argyris, 1976 see also previous presented theory on KBT). Further, humans are not rational in their decision making, in particular under these circumstances (e.g. Armstrong, 1984; J. H. Barnes, 1984; Kahneman, 2011; Slovic et al., 2004; Tversky & Kahneman, 1974, 1983). Furthermore, within organizations incentives are not necessarily aligned between different actors, as presented earlier through TCE see also (see also Argyris, 1976). In other words, it seems logical that it becomes difficult for organizations to make good decisions since it is difficult to acquire proper information through learning as well as learn to be apt at making decisions. Further, Argyris (1976), in his review of previous research, give many examples and further details of the underlying issues. These include various organizational and bureaucratic factors, such as incomplete resolutions of interdepartmental and interpersonal conflicts; ineffective and incomplete search for information; uncertainty avoidance; political exchanges and annexation of other departments; but also bargaining, shortsighted priorities, personal beliefs and goals; as well as power plays such as using misperception and miscommunication to further one’s interests (Allison, 1999; Halperin, 1974). Many examples of misperception and miscommunication is given; only presenting supportive factors of one’s view, biasing reports, not reporting negative factors as well as avoiding sending reports to certain senior managers (Halperin, 1974). Further, these power plays do not only happen in between actors at the same level and upwards in the organization, from lower level employees to senior managers, but also downwards from senior managers and down. These many times accepted games easily leads to secrecy by actors who want to remain in control and power, which directly inhibits decision makers, in particular at the higher levels, from acquiring accurate information and thus make correct decisions (Argyris, 1976). Furthermore, George (1972, pp. 769-780) presents nine general malfunctions in decision-making in his studies of the President of the United States, also verified by (Janis, 1972). Most of the malfunctions focus on how factors such as asymmetry of information and incentives, as well as the advocacy of a decision maker’s advisors can negatively affect her ability to make the correct decision. Further, George (1972) argues in favor of multiple advocacy to enhance the decision capability and reduce the occurrence of above malfunctions, also supported by (Argyris, 1976). Finally, George (1972, p. 759) presents three conditions for effective use of multiple advocacy: 1) No major maldistribution among the various actors of the following resources: a) Power, weight, influence. b) Competence relevant to the policy c) Information relevant to the policy problem. 18 d) Analytical resources. e) Bargaining and persuasion skills. 2) Presidential-level participation in organizational policy making in order to monitor and regulate the workings of multiple advocacy. 3) Time for adequate debate and give-and-take. In sum, there are many factors that can possibly damage learning as well as effective and accurate decision making, which in turn may through this cognitive frame, or as presented by others as groupthink (Janis, 1972, 1983), cause organizations to get stuck along an existing innovation paradigm and reduce its capabilities to produce disruptive and discontinuous innovations. 3.2.7. Dualism and Matrix Organizations Very much in line with George’s (1972) suggestion on multiple advocacy and (Argyris, 1976) arguments for fostering a collaborative double-loop learning environment is the concept of matrix organizations. The matrix organizational structure is a suggestion on how firms, in particular large firms, can cope with optimizing and balancing several critical dimensions for success when needed for sustaining growth and development (Davis & Lawrence, 1977; Degen, 2010; Galbraith, 1971). Most often it is a combination of how a company can combine the need to stay close to the market, sometimes due to the need to keep innovative, through market oriented processes and routines while still maintaining economies of scale and scope by also utilizing a functional setup. In essence, the matrix organization is most often described as a mix of a skewed, but completely balanced, matrix where on one dimension are the market processes and teams, and on the other dimension functional processes and teams, such as depicted in figure 1 below. Each member of both the functional senior management and the market senior management reports directly to the top leader; e.g. Chief Executive Officer (CEO). Figure 1 – Different styles and ways to balance matrix organizations (Degen, 2010) 17 Fu nc tio na l te am s M ar ke t te am sMar ke t tea ms Functional teams Figure 3. Matrix organizations – functional and market dominant and balanced structures The conflicts in the matrix are commonly caused by the pursuit of the optimization of the overall strategy of the company that in many cases requires the sub-optimization of one or both of its dimensions. This need to eventually sub-optimize the dimensions in benefit of the whole is contrary to the mechanical organization school where success is measured by individuals work efficiency. Individuals in many cases do not understand or resist the idea of sacrificing their work efficiency in favor of another if the reward system and human resource policies do not take the need of sub- optimization in favor of the overall objective into account. Galbraith (2009, p. 10-19) explains that the matrix organization is a collaborative organization. People must develop collaborative skills to share power in the organization. These are the skills that the modern human relations school promotes, like the harmonization and coordination of group efforts in organizations replacing the individual hero of the past. But to make this possible companies must ensure that their information and reward systems and human resource policies are aligned with the matrix organization structure and the overall strategy of the company and don’t create biased behaviors distorting the cooperative behavior. 19 The popularity of adopting a matrix structure has varied since its initial introduction in the 1970s and 1980s (Degen, 2010). Degen (2010) presents evidence pointing to the matrix structure losing its popularity because of the many failures in implementing the structure at the time, but later regaining popularity in the 1990s as some firms showed great success with using it despite these initial failures. Hence, it is far from non-trivial to implement a matrix structure and great care needs to be taken to several factors (Degen, 2010; Galbraith, 1983). The essence of the matrix structure is the duality, which needs to be instilled into the organization, in sharp contrast to a single-minded, one winner culture as is typical in many mechanistic and hierarchical organizations (Degen, 2010; Galbraith, 1983). This duality needs to span across the strategy and vision of the firm, the skills and mindsets of the people within the organization, the power and thus organizational structure, as well as rewards, processes and systems. Each part needs to be aligned with the overarching goal of duality in order to feed the organization the right motivation and information and drive the desired behavior. Only then will the proper culture be shaped within the organization and the expected performance increase happen. At least three quite distinct roles can be distinguished within a matrix organization (Davis & Lawrence, 1977; Degen, 2010); the top leader, the matrix leaders and the subordinates below. The top leader needs to manage and balance each dimension, including efficiently and decisively resolving disputes and conflicts while still ensuring each dimension is equally prioritized according to the overarching strategy and vision. In other words, the top leader will need to mix an autocratic leadership style with that of a participatory style. Further, the top leader also needs to instill a culture of collaborative and joint decision-making, as this will be required among the matrix leaders and subordinates. The matrix leaders, in turn, needs to be able to achieve results not necessarily based upon their position in the hierarchy, but on personal skills and abilities such as bargaining, making compelling arguments, and similar, since they are not the sole leader of each subordinate. Last, the subordinates needs to be able to cope with demands and goals set based from different dimensions of the firm - many times often competing demands and goals. Thus, in order to succeed with a matrix organization significantly higher requirements will be put upon management, for example if comparing with a purely mechanistic and hierarchic organization (Degen, 2010). 3.2.8. Entrepreneurship inside existing Profitable Organizations Albeit the matrix organization, if implemented correctly, can yield great results for firms it does not necessarily address discontinuous and disruptive innovation in and by itself. It may address efficiency and incremental, continuous innovation but it does not necessarily break the firm out of its path dependency, which may be needed in order to remain profitable. In O'Reilly and Tushman (2004) studies, they found that ambidextrous organizations; organizations where the discontinuous innovation projects were separated from all but the top management but still utilizing the resources of the 20 whole organization were vastly more successful than the alternatives; functional, completely integrated teams, cross-functional teams and unsupported teams. This, in addition to the matrix structure, also speaks in favor of the duality albeit outside of a potential matrix structure within the firm. O'Reilly and Tushman (2004) argue the innovation team needs to have a different mindset, culture, structures, control and reward systems and processes compared to the rest of the organization. Since discontinuous and disruptive innovation projects are path breaking and vastly different from incremental innovation projects as explained above, they need to live under different rules (also supported by Blank & Dorf, 2012).The more different it is from the regular innovation process of the organization, the greater the need for separation from the regular organization (Galbraith, 1983). The discontinuous innovation team might even need to work under a completely different logic, such as the hypothesis-testing action approach (Berglund et al., 2007; Blank & Dorf, 2012) or an effectuation approach (Sarasvathy, 2001a, 2001b). Despite the quite different set of rules for the innovation team, O'Reilly and Tushman (2004) also argue that a tight link to the rest of the organization must be held through senior management. Otherwise, the innovation project would not enjoy the benefits of existing organization and utilize its resources. This may include expertise, financial means and other resources required for the continued success of the innovation project. Hence, this further stresses the importance of the duality of top management, similar to that of George (1972), Argyris (1976), Degen (2010) and others. All of top management needs to support the innovation efforts made outside of the regular organizational structure, and there needs to be a clear vision and strategy which is clearly communicated to the whole organization. If the top management does not succeed, the innovation project may get hampered by routines and processes made for efficiently delivering results according to the existing revenue streams, or not able to utilize the resources of the rest of organization and thus reduce the likelihood of cross- fertilization and hence the success wanted. In sum, because of organizational inertia, the link between innovation and organizational structure, the inherent lucidness and uncertainty that accompany discontinuous and disruptive innovation (e.g. Blank & Dorf, 2012; Eisenhardt & Martin, 2000; Teece & Pisano, 1994) as well as the cognitive frame surrounding an organization and its members, it can easily become difficult for established firms to come up with discontinuous and disruptive innovation. Albeit the matrix organization balances different needs within an organization, it does not necessarily facilitate discontinuous and disruptive innovation, such as the ambidextrous organization. However, underpinning all theory presented above lies dualism, the critical importance of top management capability to simultaneously manage an existing organization as well as skunkwork teams or separated teams (such as described in this subchapter), and learning to break through the path dependency that often surrounds incumbents firms. Either part may break the delicate system and hence hinder existing firms’ continued success through new waves of creative destruction (Assink, 2006; O'Connor, 2008). 21 3.3. An External and Integrated Perspective In addition to the many internal factors affecting incumbent firms’ ability to achieve discontinuous and disruptive innovations, there are several external factors significantly affecting as well (also argued above; Afuah & Bahram, 1995). However, internal and external factors are not necessarily separable as there is an obvious interplay them in between, since many external factors affect a firm’s ability to innovate through affecting internal factors. Thus, external factors affecting a firm’s ability to achieve disruptive and discontinuous innovation do so both directly; through constraining the firm in various ways, but also indirectly; through internal factors. Hence, in order to understand an external and integrated perspective on inhibiting factors for incumbents’ disruptive innovation ability we first need to understand what factors are critical for successful innovation in general; both internally and externally. 3.3.1. Innovation and Value Networks Based upon one of the most comprehensive empirical studies on success factors for innovation, Freeman (1991) suggests six to be the most critical; User needs and networks, coupling of development, production and marketing activities, linkage with external sources of scientific and technical information and advice, concentration of high quality R&D resources on the innovative project, high status, wide experience and seniority of the “business innovator”, and basic research. Many similarities can be found with the theory presented of internal factors. Coupling of development, production and marketing activities suggests that innovation is more likely when an integrated perspective can be taken – similar to the discussion on modularity versus architectural, integrative innovation above. Further, it can also be argued that it falls naturally that a high concentration of resources needs to be brought together, since, as described earlier, discontinuous innovation, arguably, by nature requires a greater leap in newness. Moreover, the criticality of high status, wide experience and seniority of the “business innovator” is very aligned with the presented theory on successful ambidextrous organizations. However, the internal aspects do not cover all factors presented. This points to the earlier discussion on both markets as well as institutions being critical in order to explain shifts and discontinuities in industries. Further, these parts have effect in an integrated manner and include competitors and suppliers of the industry (DeBresson & Amesse, 1991; Freeman, 1991; Kaplan & Tripsas, 2008). In other words, in order to explain the failure of incumbents to produce disruptive innovation there is a whole network of actors to consider. This is well in line with recent thinking (e.g. DeBresson & Amesse, 1991; Freeman, 1991; Kaplan & Tripsas, 2008) and follows naturally from the somewhat simplistic value chain concept, popularized by Porter (1985). These networks are often referred to as value networks (e.g. Peppard & Rylander, 2006; Sandström, 2010) or ecosystems (e.g. Adner, 2006; Basole, 2009; Leavy, 2012; Moore, 1993). 22 There exists many definitions on networks in management literature (DeBresson & Amesse, 1991). DeBresson and Amesse (1991, p. 364) describes a network as a “loose form of an inorganic and decomposable system”. They emphasize that the importance is that “there is more to the network than the sum of its interacting components” (DeBresson & Amesse, 1991, p. 364). In other words, a network can in large be seen as a loosely formed organization and thus lives under similar conditions and can expect similar effects as an organization. In reference to innovation, networks can be seen as a way to cope with systemic innovation (Imai & Baba, 1991); innovation that spans outside any single autonomous part (Teece, 1996), much similar to how architectural innovation spans across multiple autonomous modules as discussed earlier. Thus, a network can be attributed to similar effects and conditions as an organization, and will be affected by both TCE and KBT (Colfer, 2007), affecting such as a firm’s ability to enjoy economies of scope and scale (Freeman, 1991). In other words, the difficulties of balancing coordination and cooperation, integration and modularity are not only a matter within the firm, but also in between firms. One scenario where the concept of value networks is especially useful is in two-sided markets; markets existing of two distinct user groups (Eisenmann et al., 2006). This scenario cannot easily be explained using other, more simplistic concepts such as the concept of value chains. Two-sided markets are enabled by linking the two user groups by providing them with a platform – a product or service that facilitates transactions between the two groups through providing necessary infrastructure and rules. Further, they operate under positive feedback loops (Forrester, 1968), and network effects, where increasing demand of one side lead to increasing demand on the other side. Since they operate quite differently, both sides need to be managed in specific ways. According to Eisenmann et al. (2006) specific strategies should be employed in two- sided markets. The more quality and price sensitive side should be subsidized, since this will enlarge the market and spur accelerate growth. Exclusive participation of marquee users, particularly important users such as early adopters, should be secured to further increase the growth speed. Further, platform providers should avoid envelopment (Eisenmann et al., 2006). In such scenarios, alternative business models (the business model concept is further elaborated below) should be considered. Furthermore, it is critical to be aware of and cope with winner-take-all dynamics. In some scenarios, one platform will completely dominates due to the aforementioned network effects. This will happen when the following criteria is fulfilled: multi-homing costs, the cost a user incurs by belonging to multiple platforms, are high for at least one side of the users; network effects are positive and strong for at least the users with high multi- homing costs; and no side of users have a strong preference for special features. In such scenarios, the firm must either go for a complete win, or share the single winning platform with competitors. 23 3.3.2. Effects and Risks of Innovation Networks Arguably, it will be harder to manage innovation across firms since any specific firm has less control over other firms, than actors within the firm (Colfer, 2007). Along similar lines, it can be argued that it reduces risk, i.e. uncertainty, as the firm does not necessarily need to take responsibility for all aspects and parts of the innovation. This uncertainty includes both uncertainties in how to create value, such as through technology, but also how to appropriate this value (Adner, 2006; DeBresson & Amesse, 1991; Leavy, 2012). One way in which uncertainty can be divided between firms is through complementary assets. Complementary assets in firms can through collaboration lead to great value (Harrison et al., 2001), as briefly touched upon in the earlier discussion on dynamic capabilities, and help firms cope with destructive waves of innovation. This can happen both directly, as well as indirectly, such as through networks, for example through alliances, joint ventures, research associations, and similar (Freeman, 1991). Not only can it help directly through greater immediate innovation success through the actual collaboration, such as the outcome of a joint venture, but research suggests it can also significantly increase knowledge transfer, organizational learning and development of new capabilities, irrespective of the immediate goal of the collaboration (Harrison et al., 2001). As discussed previously, all these factors can greatly impact a firm’s innovation success. However, due to the lack of control a firm has in an innovation ecosystem, such as a network of firms, there are also great risks with such a setup (Adner, 2006; Leavy, 2012). According to Adner (2006; Leavy, 2012), firms have typically only focused on execution risk when innovating, but missed both interdependency risks in the ecosystem and adoption risk. Execution risk pertains to the risk of failing with the innovation itself. This may be both internally as well as externally, if the innovation is jointly developed. Adner (2006) refers to this as the initiative risk. However, this does not address the full risk of innovation endeavors. There may be other complementarities the innovation needs to be ready before the end user realizes the full value of the innovation. Thus, the innovating firm may also be dependent on the innovation capability of other firms, or groups of other firms. This Adner (2006) refers to as the interdependence risk. Further, there may also be a chain of actors, other firms as well as users, who needs to adopt the innovation before its value is truly realized, what Adner (2006) refers to as the integration risk. In essence, three types of problems may arise due to these three innovation risks (Adner, 2006; Leavy, 2012). First, in a complex scenario an innovation may be dependent on a lot of actors’ technological success, for example their capabilities to invent new products. Hence, there is an interdependency risk of innovation. Second, there is further uncertainty in if the innovation actually yield its hypothesized value. Third, there may also be an issue of timing. There may be an optimal time for introduction of the product or service to the market. Delays in any of the three risks may delay final adoption to the extent where the original business case does not hold because it is expected that the innovation will no longer be novel enough to compete. In other 24 words, it is important to weigh the full innovation risk to the potential first mover advantages and disadvantages (Adner, 2006). Advantages may include technological leadership through utilizing a falling learning curve and patents, preemption of scarce resources such as the acquiring of specific assets, geographical locations, product space, plants and equipment, as well as the addition of switching costs and buyer uncertainty to followers (Lieberman & Montgomery, 1988). Disadvantages may include free rider effects such as when imitation is cheaper than the initial development, the cost of resolving the initial uncertainty in technology and market characteristic of the early phases of industries, shifts in technology or customer needs as well as other causes of incumbent inertia (Lieberman & Montgomery, 1988). In sum, albeit many positive effects can come out of collaborating when innovating, there are also many great risks that must be properly evaluated, analyzed and managed (Adner, 2006; Leavy, 2012). 3.3.3. Managing Innovation in Networks As discussed, to properly manage risk and benefits it is important to balance collaboration and cooperation, but also modularity and architecture, when innovating due to inter firm effects of TCE and KBT (Colfer, 2007). In order to achieve discontinuous and disruptive innovation, an architectural innovation is often desired. Based upon TCE and KBT, it may seem this is always best done vertically integrated; i.e. inside of one firm (Colfer, 2007), since the innovation risks according to these theories would be easier to manage internally. An alternative, which has shown success in some scenarios, is for one firm to take on the role of being a systems integrator, or lead firm (Colfer, 2007; Langlois & Robertson, 1992). The systems integrator takes on the role of managing the overarching architecture, enabling architectural and systemic innovation, while still enabling a high level of modularity. For example, to achieve success with joint development programs in the auto industry, Takeishi (2001) found through empirical studies that, in addition to the joint capabilities of the actors and face- to-face communication, architectural knowledge held by automaker engineers were positively correlated with success. Thus, systems integrators can be useful and needed in innovation networks when it would be too costly for all actors to acquire enough knowledge to innovate across all, or a majority, of actors, as well as efficiently align all involved actors. Chesbrough (2003) outlines additional strategies and roles in an innovation network. As Chesbrough (2003) argues, closed innovation, internal innovation managed successfully through control, has become less prevalent since it is not any longer necessarily the most efficient for success (also Freeman, 1991). Since complementary actors exists more readily today, such as external funding and highly skilled knowledge personnel, firms may successfully innovate through networks. Rather, in such a scenario realizing that external R&D and innovations can benefit the firm may be better, regardless if the ideas originated internally or not. Further, internal ideas may lead to more revenue if further developed and commercialized externally. This is very much in line with previously presented theory. 25 Three categories of activities exists in such an open innovation network; funding, generating innovation and commercializing innovation (Chesbrough, 2003). Funding includes investments for economic rewards as well as indirectly benefitting from the innovation becoming realized, such as needing it as a complementary component to one’s own innovation. Generating innovation includes fundamental exploration, specific exploration for a narrower target such as a specific commercial application, architectural innovation such as done by the previously described systems integrator, and innovation made for a specific, higher, cause such as open source innovation. Commercializing innovation includes marketing of innovation, such as by realizing the actual value through insights from users by keeping close to them, or by turning into a “one-stop-shop”; a place where users and buyers expect to find new products and services. In sum, a firm must not necessarily perform all activities, and may specialize on certain activities and let others perform other needed. In particular when lacking specific resources, or to reduce other risks such as market and appropriability uncertainty, such as often when striving for discontinuous and disruptive innovation. 3.3.4. Network Forms and Institutions Role in Innovation As argued above, networks can provide useful for overcoming certain barriers when firms innovate. However, there exists many forms of innovation networks and different forms may be useful in different scenarios. As pointed out by DeBresson and Amesse (1991), a network should not necessarily be seen as an intermediate state between a firm and a market, as may portrayed through TCE. Rather, they argue there are too many different forms and types of connections making networks something different. A purely modular structured network of innovators might be structured according to a market structure. As presented earlier, this may not be suitable for achieving discontinuous and disruptive innovation. Freeman (1991, p. 502) proposes a list of ten main forms of cooperation for innovation, useful under different scenarios: 1. Join ventures and Research Corporations 2. Joint R&D agreements 3. Technology exchange agreements 4. Direct investment (minority holdings) motivated by technology factors 5. Licensing and second-sourcing agreements 6. Sub-contracting, production-sharing and supplier networks 7. Research Associations 8. Government-sponsored joint research programmes 9. Computerized data banks and value-added networks for technical and scientific interchange 10. Other networks, including information networks Albeit it may be beneficial to enter into one, or many, of the above types of collaborations in order for a firm to achieve higher efficiency of innovation investments, or in order to enable discontinuous and disruptive innovation, there is naturally also a cost of establishing these collaborations. Hence, these networks may not form themselves. Thus, it follows naturally that governments may want to assist firms in 26 forming innovation collaborations. Further, research has shown that clustering may be beneficial (Baptista & Swann, 1998). Strong demand in a specific geographical area and lower consumer search cost are some potential benefits, but clustering can also facilitate innovation through closeness to users, as presented above, and in particular through utilizing lead users such as described by Von Hippel (2005). In addition to market side factors, supply side factors, such as labor market pooling, closeness to suppliers and inputs, and knowledge spillovers may make clustering beneficial. In particular, knowledge spillovers may be important for innovation(Baptista & Swann, 1998). Hence, it poses natural for governments to aid firms in establishing these collaborative network since it may lead to clustering, thus enabling higher innovative output and, in turn, greater economic development for the region. In fact, in recent years it has become quite popular, in particular since other stimulus may not be positively seen upon as they may not be in line with free market thinking (Freeman, 1991). However, institutions do not only affect innovation success among firms through enabling formation of networks. Institutions may also play a role in other aspects of the innovation process, such as regulations hampering certain innovations’ level of commercial success or through enabling knowledge transfer from universities. In other words, in particular in the case of path breaking, disruptive and discontinuous innovation, there might be a need to change certain institutions, and thus there might be a need for institutional entrepreneurship in order to achieve innovation success. Leca and Naccache (2006) argue this can be done through trying to convince other actors of changing the institutional logic in use, which in turn shapes the institution. By nature, however, this is an uncertain process since institutional logic, built up by certain structures, is not easily grasped, if at all graspable. Further, different institutional logic may cause different causal effects in different contexts, further complicating this process of transforming institutions for strategic reasons. 3.3.5. Resource Dependency of Incumbent Firms In addition to previous examples, another institutional lock-in is when firms get stuck in their value network through resource dependency (Pfeffer & Salancik, 1978). Considering a value network’s development, such as a specific industry, a dominant design on how value is created will take form as discussed earlier. This creates a dominant logic on how business is done (Prahalad & Bettis, 1986), similar to the cognitive frame discussed previously under internal factors. Further, this will also be the established way, and thus dominant logic, actors inside of the value network actually create value and thus revenue, in contrast to new innovation projects that do not necessarily generate significant revenue streams and profits. In particular, this may be prevalent with discontinuous and disruptive innovation projects where uncertainty of both value and appropriability is large. This may lead to discontinuation of these projects. However, it may be premature if these projects are what Chesbrough (2004) would refer to as false negatives; they will eventually lead to great value, in particular this may happen if they break the existing dominant logic. 27 Christensen (Bower & Christe