1 THE USE OF DIFFERENT SOURCES OF KNOWLEDGE AND THE
INNOVATIVENESS OF FIRMS
A Quantitative Analysis of Different Knowledge Origins and The Innovativeness of Firms
MASTER THESIS
Author: Philip Kruft
Date: June 2014
Student number: 10663800
Study: MSc. Business Administration – Strategy
University: University of Amsterdam Business School
2
ABSTRACT
Different scholars have expressed the importance of knowledge in order to gain a sustainable competitive advantage. Firms can use local knowledge or non-local knowledge, e.g. knowledge from outside the firms and especially the industry boundaries, to come up with new and innovative products and ideas. The current academic literature argues that the use of non-local knowledge could be a source for new path-breaking innovations. In addition, it is argued that the use of non-local knowledge contributes to the performance of firms. This paper focuses on the influence of knowledge from different origins and the innovativeness of firms. Hence, the paper examines whether or not the use of different sources of knowledge contributes to the innovativeness of a firm. In addition, it questions the situations in which different types of knowledge truly contribute to innovation or not. This paper used 74 patents from the child carrier industry to test its hypotheses. This paper finds arguments to question the use of non-local knowledge for smaller firms in certain industry conditions. In addition, this paper argues that smaller firms can be more innovative with local knowledge rather than local knowledge. Besides, this paper finds confirmative results in terms of the use of non-local knowledge as argued by the literature. However, this only applies for larger and more active firms.
3
TABLE OF CONTENT
ABSTRACT ... 2
INTRODUCTION ... 5
LITERATURE REVIEW AND HYPOTHESIS ... 8
FROM SCHUMPETER TO COMPETITIVE ADVANTAGE BY USING KNOWLEDGE ... 8
EVOLUTIONARY FITNESS: AMBIDEXTERITY ... 12
DYNAMIC CAPABILITIES: SENSING, SEIZING AND RECONFIGURATION ... 13
KNOWLEDGE ... 16
DATA AND ANALYTICAL METHOD ... 21
SAMPLE ... 21
THE USE OF PATENT DATA ... 22
DEPENDENT VARIABLE ... 23 INDEPENDENT VARIABLES ... 24 CONTROL VARIABLES ... 27 STATISTICAL METHOD ... 28 EMPERICAL RESULTS ... 31 REGRESSION RESULTS ... 31 HYPOTHESES RESULTS ... 36
4 DISCUSSION ... 38 CONCLUSION ... 43 REFERENCES ... 45
5
INTRODUCTION
The pressure faced by today’s businesses seems to increase their pace of innovations and industry evolutions. Especially in highly dynamic environments technology innovations are crucial for a firm’s competiveness (Chesbrough and Rosenbloom, 2002; Feenstra, 1998; Santos & Eisenhardt, 2005; Jacobides et. al, 2006). A lot of devices and products seem to evolve by integrating knowledge from outside the existing industry boundaries. For many of these cases it is possible to integrate, combine or create technologies and knowhow that opens the door for new innovative applications. Thus scanning, analyzing and applying knowledge for new applications could be a source of competitive advantage. Consequently, intellectual capital is gaining increasing recognition as the only true strategic asset (Hamel, 1998). This paper focuses on the influence of knowledge from different origins on the innovativeness of firms. Hence, the paper examines whether or not the use of different sources of knowledge contributes to the innovativeness of a firm. In addition, it questions the situations in which different types of knowledge truly contribute to innovation.
An invention is not an innovation per se. An invention can be a new product, idea, process, method or something else that never have been made before. Meanwhile, an innovation is the use of a new idea or method. There are a lot of inventions that are not used at all. Chesbrough and Rosenbloom (2002) identified that innovation itself is no guarantee for economic value per se. An effective business model is essential to connect potential with the realization of economic value. New technologies and developments sometimes miss a clear path to the market and thus lack to deliver value to customers. Nonetheless, some inventions do lead to innovation which can lead to a competitive advantage (Basberg, 1987). Innovations can start new technology regimes, e.g. technologies that change how particular businesses or industries
6 are operating or are organized, and thus changing market environments. Questions not only relate to how firms can benefit from its innovations (e.g. Jacobides et al., 2006), but how firms can respond to new developments as well (e.g. Cooper & Smith, 1992). Jay Barney (1991) considers innovation and strategic flexibility as the two most important factors in dynamic markets. In addition, Barney (1991) and Grant (1996) have stressed the importance of intangible assets for attaining superior performance and a sustainable competitive advantage. Among these intangible assets, knowledge is considered as one of the most important resources (Liebeskind, 1996).
Existing capabilities and the use of local knowledge, knowledge that arrives from familiar space or is based on existing knowledge combinations, may explain why a firm is able to compete in today’s competitive environment. However, the challenge is to explain how these can help to reconfigure assets and resources as the context shifts and non-local knowledge, knowledge from ‘a distance’, is required. A firm’s ability to benefit and profit from existing assets and positions (exploiting) while looking ahead for new technologies and markets (exploration) simultaneously is known as ambidexterity and is considered as a critical element in a sustainable competitive advantage (Birkinshaw & Gibson, 2004; O’Reilly & Tushman, 2008).
A deeper understanding about the conditions that help to create innovations and maximize profitability can help firms to gain a competitive advantage. Not only to outperform competitors but to sustain and survive market and technology changes as well (Klepper and Simons, 2005). This paper focuses on the explanation of competitive industry dynamics and the influence of different knowledge sources on the innovativeness of firms. The main
7 research question of this paper is: How does knowledge from different sources result in higher levels of innovation
The primary research objectives of this paper are to 1) contribute to our understanding about creating innovation, 2) suggest new explanations for innovation differences between firms on an industry level, 3) question the contribution of ambidexterity for different types of firms. In addition, 4) expand our knowledge about competitive dynamics in industries. Finally, the paper could help in 5) creating a deeper understanding about the evolution of industries by examining the patterns of innovation.
This paper begins with an examination of the existing literature. Firstly, it discusses how knowledge has become a cornerstone in the explanation for sustainable competitive advantage among firms. Secondly, it discusses the use of local and non-local knowledge by means of ambidexterity. Afterwards, the dynamic capability of sensing, seizing and reconfiguration is explained. Finally, the literature review finishes with a analysis about knowledge, research gaps and the hypothesis. After the theory is elaborated, the data and analytical methods are discussed. Followed by the empirical results, the discussion and finally the conclusion.
8
LITERATURE REVIEW AND HYPOTHESIS
FROM SCHUMPETER TO COMPETITIVE ADVANTAGE BY USING KNOWLEDGE
The emergence, development and decline of industries have received attention of several economists in the past. Joseph Schumpeter (1939) and Alfred Marshall (1879, 1890) have paid attention to different industry dynamics. Not only do industries arise and perish as firms enter or exit an industry, but firms can also expand or contract their boundaries by organizational changes as the result of firm decline or growth (Malerba & Oresenigo, 1996a).
Schumpeter advocated the presence of structural evolutions in industries. Firstly, Schumpeter (1934) proposed a pattern of innovative activity which is explained by the technological ease of entry and the major role played by entrepreneurs and firms in these new innovative activities. Later labeled as Schumpeter Mark I by Nelson and Winter (1982), this industry dynamic is also known as ‘creative destruction’. The Schumpeter Mark I pattern is considered as a widening pattern because additional firms can enter the market. Due to a new innovative base it is possible for new innovators and entrepreneurs to enter an industry at the expense of established firms. In other words, a new innovation creates a broad field of (new) competitors (Malerba & Oresenigo, 1996b).
Secondly, Schumpeter (1942) discussed the relevance of the industrial R&D capacity for technological innovation and the essential role of large firms. Labeled as Schumpeter Mark II, this pattern of innovative activity is characterized by 'creative accumulation' with the prevalence of large established firms and the presence of relevant entry barriers for new
9 innovators. Due to continuous innovation and accumulation the technological and innovative capabilities of a firm will grow over time. Therefore, the Schumpeter Mark II pattern is considered as a deepening pattern as the dominance of a few established firms will grow (Malerba & Oresenigo, 1995; 1996b). The basic ideas of Schumpeter still stand in today’s academic research.
However, studies of Schumpeter and colleagues did not take off as industrial economists concentrated on the structure of industries. During the 1950s, 1960s and 1970’s economists concentrated on studying industry structures such as concentration, firm size and profitability (Malerba & Oresenigo, 1996a). As a consequence, research on the development (industry) dynamics lagged behind.
Nonetheless, in the 1980s and 1990s more and more theories of strategy based on positional or resource advantages (Barnett, Greve & Park, 1994; Porter, 1980;Rumelt, 1984; Wernerfelt 1984; Barney 1991; Peteraf 1993), have been evolving with dynamic approaches. This way they attempt to explore how firms can recombine and integrate their resources to deal with market and technological changes (O’Reilly & Tushman, 2008). The resource based view (RBV) of the firm (Wernerfelt, 1984; Barney, 1991; Peteraf, 1993) has been known for emphasizing the essential role for valuable and difficult to imitate resources (Barney & Peteraf, 2003). According to the RBV a competitive advantage lies in how a firm is able to use, combine and/or reconfigure these resources to address the needs of customers (Leiblein, 2003).
10 However, the RBV was criticized for its static nature, especially in dynamic market environments (Eisenhardt & Martin, 2000; Stoelhorst, 2004). As the most influential theory of the last 20 years in the strategy field, it missed the dynamic process by which a competitive advantage is created. It lacked the ability to explain why and how employees can transform a firm’s resources into a competitive advantage (Stoelhorst & Bridoux, 2008). In a reaction to the criticism, academics started to investigate the process of developing capabilities. They abandoned the use of static equilibrium conditions to study these phenomena to the benefit of Schumpeter’s vision of competition as a process of ‘creative destruction’ (Stoelhorst & Bridoux, 2006). In 1997 Teece, Pisano and Shuen initiated a new perspective which later became known as the dynamic capabilities view (DCV) of the firm. They defined dynamic capabilities as “the firm’s ability to integrate, build, and reconfigure internal and external competences to address rapidly changing environments” (Teece et al, p.516). The capabilities of a firm are embedded in existing organizational skills, processes, procedures, organizational structures, decision rules, and disciplines (Teece, 2007). Thus, they are reflected in how an organization operates it structures, cultures and the mindset of senior leaders (O’Reilly & Tushman, 2008). According to Teece (2007) a competitive advantage depends upon the firm’s capacity of its dynamic capabilities to “ 1] sense and shape opportunities and threats, to 2] seize opportunities, and finally to maintain competitiveness through enhancing, combining, protecting, and, when necessary, 3] reconfiguring the business enterprise’s intangible and tangible assets.” (Teece, p.1319).
The RBV school responded by recognizing the need for a dynamic perspective and application under dynamic conditions (Helfat & Peteraf, 2003; Peteraf & Bergen, 2003). Peteraf & Bergen (2003) recognize the need for scanning the factor and product markets. Scanning the factor market is essential in order to find unique and scarce resources. In
11 addition, to combine or reconfigure, e.g. applying in practice, these resource in such a unique matter that valuable and difficult to imitate products can be created. In particular, these products should meet market demand. Indeed, despite using traditional RBV jargon the more recent RBV seem to have major overlap with Teece’s tripartite taxonomy. In addition, Barney (1991) and Grant (1996) have stressed the importance of intangible assets for attaining superior performance and a sustainable competitive advantage. Among these intangible assets, knowledge is considered as one of the most important resources (Liebeskind, 1996). Knowledge can come from different sources, which could have a different impact depending on the type of source. Thus, it is interesting to understand which sources of knowledge leads to higher levels of innovation.
To some the DCV is considered as an additional explanatory theory in addition to the RBV (Stoelhorst & Bridoux, 2006). Others (Teece, 2007; Zott, 2003) argue that the DCV is more than an addition to the RBV and that resources and competences should be distinguished from dynamic capabilities. Nonetheless, general consensus is that both theories have complementary views on the way a firm can gain a competitive advantage. As stated by Eisenhardt & Martin (2000) resources and their related activity systems have complementarities. Indeed, resources cannot be used without employees, a certain culture, knowledge and a wide range of (organizational) processes and activities. Both recent RBV and DCV scholars seem to agree on this point. It is not without reason that intellectual capital is gaining increasing recognition as the only true strategic asset (Hamel, 1998). Since the work of Schumpeter and his colleagues, the dynamic aspect of firms and industries is back on the academic agenda. Also referred to as neo-Schumpeterian theory of the firm and the Schumpeterian characterization of the innovation processes (Teece, 2007).
12 EVOLUTIONARY FITNESS: AMBIDEXTERITY
Till recently, the major question was still whether or not firms can adapt to their environment. Some argue that it is possible for firms to change and adapt, while others argue that firms will inert and changes occurs through an evolutionary process by the means of variation-selection-retention (O’Reilly & Tushman, 2008). Discussed comprehensively by O’Reilly and Tushman (2008) there is data to support both arguments. Despite high failure rates, some firms adapt and survive over time (DeGeus, 1997; Tripas, 1997). Since the discussion of the dynamic capabilities view, researchers have attempted to find out under which conditions firms are able to adapt to environmental changes in order to sustain their competitive advantage. These environmental changes include technology changes, as the result of influential innovations, as explained by Cooper & Smith (1992).
Existing capabilities and the use of local knowledge may explain why a firm is able to compete in today’s competitive environment, but the challenge is to explain how these can help to reconfigure assets and resources as the context shifts and non-local knowledge is required. One the one hand, local knowledge can be defined as knowledge that arrives from familiar space or is based on existing knowledge combinations. On the other hand, non-local knowledge can be described as knowledge from ‘a distance’. In other words, it is based on far-flung elements (Miller et al, 2007). In short, it all depends upon a firm’s ability to benefit and profit from existing assets and positions (exploiting) while looking ahead for new technologies and markets (exploration). Doing this simultaneously is known as ambidexterity and is considered as a critical element in a sustainable competitive advantage. Hence, it enables a firm to adapt over time. In other words, without exploiting existing assets and
13 positions a firm will not be profitable in today’s business. At the same time, it will not survive market changes if it is not capable of exploring tomorrow’s profit makers (Birkinshaw & Gibson, 2004; O’Reilly & Tushman, 2008).
Ambidexterity, referred to as the adaptability of a firm by Birkinshaw and Gibson (2004), is not a formula for a competitive advantage per se. Ambidexterity may help firms to understand market and technology changes and/or to come up with new combinations of non-local knowledge as it may produce path-breaking innovations (Fleming, 2001; Galunic & Rodan, 1998; Nelson & Winter, 1982; Nerkar & Roberts, 2004; Schumpeter, 1947). As mentioned before, innovation itself is no guarantee for economic value (Chesbrough & Rosenbloom, 2002). An effective business model is essential to connect potential with the realization of economic value. New technologies and developments sometimes miss a clear path to the market and thus lack to deliver customer value. Nonetheless, some inventions do lead to innovations which can lead to a competitive advantage (Basberg, 1987). It is up to a firm’s capability of sensing, seizing and reconfiguration to successfully transform knowledge into a commercially successful idea. In addition, firms should find a balance between making profit in today’s business while looking ahead and outside the current boundaries. Overall, it is important to understand that it is up to a firms (dynamic) capabilities and know-how to successfully benefit from ambidexterity. Both dynamic capabilities and the application of knowledge will be discussed further on.
DYNAMIC CAPABILITIES: SENSING, SEIZING AND RECONFIGURATION Since ambidexterity is considered as a dynamic capability it can by analyzed by the means of Teece’s (coherent) tripartite taxonomy existing of sensing, seizing and reconfiguration
14 capability (O’Reilly & Tushman, 2008; 2004; Teece, 2007). These can be found at the level of the firm, individual and network level, in order to integrate, reconfigure and gain and release resource to match market demand and environmental change (Eisenhardt & Martin, 2000; Zollo and Winter, 2002).
Firstly, sensing is considered as the activity of scanning, creation, learning and interpreting in order to spot and shape new opportunities. According to Kirzner (1973) firms and entrepreneurs can have the ability to interpret existing information differently. Secondly, according to Schumpeter (1934) new information and new knowledge (exogenous or endogenous) can create new opportunities as well (e.g. Schumpeter Mark I). At the same time, it will require that a firm and its employees understand latent demand, structural evolution of industries and markets and the responses of suppliers and competitors (Teece, 2007). To understand and interpret the environment and new developments a certain basic level of knowledge is required. This can be with the help of individual employees or it can be embedded in organizational processes, e.g. research and development activities. A firm will need to have a culture of openness, the commitment of resources and a senior management team that stimulates long-term thinking and promotes explorations (O’Reilly & Tushman, 2008; Burgelman, 2002; Edmondson, 1999; Rotemberg & Saloner, 2000 Nonaka and Toyama, 2007; Teece, 2007).
Secondly, seizing is considered as the process of making the right decisions and executing it. This will require leadership, vision, strategy and the commitment to invest. Consequently, senior management will have to allocate resources and complementary assets. In other words, new products or services should be developed in order to meet and fulfill the sensed
15 opportunities. When the capability of seizing is insufficient, firms may spot (sense) new opportunities but will be unable to act on them in a timely and responsive manner (O’Reilly & Tushman, 2008).
Thirdly, reconfiguration entails the ability to recombine and reconfigure assets and organizational structures (Teece, 2007). A firm should not only be able to compete in today’s environment, but also “requires that leaders reallocate resources away from mature and declining businesses toward emerging growth opportunities” ( O’Reilly & Tushman, 2008, p.191). In case of small incremental changes, this reconfiguration may proceed slowly as structures, process, people and culture shift step by step (e.g. Duncan, 1976; Eisenhardt & Brown, 1998; Nickerson & Zenger, 2002; Rindova & Kotha, 2001; Zollo &Winter, 2002). If the pace of change is high, reconfiguration may occur parallel or even constantly (e.g., Govindarajan & Trimble, 2005; Markides & Charitou, 2004; Masini, Zollo, & van Wassenhove, 2004; Tushman & Anderson, 1986). Reconfiguration is needed to maintain ‘evolutionary fitness’ (Teece, 2007). Or in other words, a firm ability to adapt to its changing or changed environment.
Most important take away of previous paragraphs is that sensing, seizing and reconfiguration requires a certain level of knowledge. A basic level should be present to understand what is going on and how new knowledge can be used to create new opportunities in terms of innovation. Hargadon (1998) already expressed that individuals and or groups within the firm have to understand that “when knowledge developed and used in one industry [it] has potential value elsewhere”. At the same time it important to understand knowledge and dynamic capabilities are complementary and inexorably connected to each other. Or as
16 phrased by Teece (2007, p. 1319) “dispersion in the geographical and organizational sources of innovation and manufacturing, sustainable advantage requires more than the ownership of difficult to replicate (knowledge) assets. It also requires unique and difficult-to-replicate dynamic capabilities.” To support its ambidexterity capabilities a firm could adapt its corporate structure. However, these corporate structures will be discussed later on.
KNOWLEDGE
So far there seems to be a general consensus among different scholars, e.g. RBV and DCV, that knowledge is of great importance in order to gain a sustainable competitive advantage. The DCV gives us, in contrast to the positional view and to a lesser extent the resource based view, the opportunity to look inside the firm and the way it deals with changing (dynamic) environments and the available resources (Stoelhorst & Bridoux, 2006). Consequently, it gives a better insight about how innovation is created.
Interesting is how knowledge really contributes to the innovativeness of firms. Klepper and Simons (2005) and Klepper (1996) found that industrial shakeout appear as a result of a competitive process in which the most able early entrants achieve dominant market positions through innovation. Constantly and actively investing in R&D capacity helps in understanding technologies and to keep on innovating. Consequently, these firms are more capable of sensing, seizing and reconfiguration in order to come up and realize new ideas. This phenomenon in which firms knowledge base grows over time is also referred to as the competitive advantage theory. It shares a lot of similarities with Schumpeter Mark II 'creative accumulation'. A firm with a large knowledge base will be more capable of understanding different types of knowledge. Consequently, they could be more capable of coming up with
17 new inventions which can lead to a competitive advantage (Basberg, 1987). In addition, firms that have a competitive advantage can gain supra normal profits. Consequently, they have the possibility to invest more in their R&D capacity and actively work on their knowledge base.
This implies that firms that are early adopters of high levels of differentiated technology modules or knowledge have a bigger opportunity to become more innovative over time. In conclusion, firms that are very active in collecting and building a knowledge base seem to be more innovative. In addition, the more active a firm is working on its knowledge base, the bigger its advantage will become over time. In contrast to less active firms, whose knowledge may grow slower compared to competitors. Therefore, the following hypotheses are proposed:
H1: Firms that use a broad range of knowledge sources have a higher likelihood of being innovative over time.
H2: Active firms have a higher likelihood of being more innovative over time.
As already discussed in the part on ambidexterity, knowledge can be local or non-local. However, for most firms it is difficult to gain knowledge from outside the industrial boundaries. This is especially because knowledge is difficult to identify and acquire through market mechanism (Teece, 1980; Von Hippel, 1994; in Miller et al., 2007). In addition, searching for distant knowledge can be arduous and costly. The gained knowledge may never result in a valuable idea (Rosenberg, 1996). In other words, in ex-ante conditions the value of knowledge can be difficult or even impossible to be estimated. Despite the difficulties, firms that are able to use new combinations of non-local knowledge may produce path-breaking
18 innovations (Fleming, 2001; Galunic & Rodan, 1998; Nelson & Winter, 1982; Nerkar & Roberts, 2004; Schumpeter, 1947) enabling them to survive market evolutions as paradigms and contexts change over time.
For long-term success firms need to support their ambidextrous activities by changing their organizational structures to initiate versus execute innovation. Raisch (2008) came up with an analysis of ambidexterity and three basic types of organizational structures: temporal, structural and parallel. In a temporal design, exploitation and exploration are emphasized sequentially rather than simultaneously. In addition, it is executed by the same employees within the firm. The organization changes back and forth between different corporate structures that benefit exploration or exploitation. In the structural design, exploitation and exploration is executed in different structures. In addition, this exploitation and exploration is done by different employees. Meanwhile, in the parallel design structure exploitation and exploration is addressed by the same employees, but in different structural environments, e.g. spin-offs. The structural and parallel design structures have different structures and or different employees to fulfill different tasks. This implies that firms should have minimal size and professionalism to do so.
In addition, research showed that interdivisional knowledge has a stronger effect on innovation than knowledge from within the divisional boundaries or outside the firm boundaries (Miller et al, 2007). At the same time, different divisions often operate in different industries. Thus, larger firms that have the capability to fully benefit from their divisional structure could be more innovative with knowledge from different industries than non-divisional firms. However, not all firms have the capability, necessity or legitimate reasons to
19 have multiple divisions. For smaller firms, including start-ups, businesses based on sole proprietorship, freelancers or individuals it does not make sense to have such an organizational structure. Once again, this implies that most of these firms should have a minimal size to fully benefit from non-local knowledge.
Lubatkin et al. (2006) are one of the first and few that openly question how ambidexterity works for small- to medium-sized (SME) firms. The joint pursuit of both exploitation and exploration has been posited as having a positive effect on a firm’s performance. However, they question the antecedents and consequences for these SME firms. Or as explained, SME enterprises “lack the amount of slack resources and the kind of hierarchical administrative systems that can help or impede larger firms in managing their contradictory knowledge processes and, thus, affect the attainment of ambidexterity” (Lubatkin et al., 2006, p.647). They found that ‘behaviorally integrated’ top management at SMEs positively influence the pursuit of exploratory and exploitative orientation and, by doing so, achieve relatively higher levels of subsequent performance.
However, the current literature does not seem to have clear answers to the question marks that were stated and rise from the article of Lubatkin et al. (2006). This paper argues that there are multiple but small literature gaps in place. For example, does non-local knowledge always have the biggest contribution to the innovativeness of firms? What if an industry has low entry barriers and specific conditions that allow individuals and small firms to come up with new innovations? In particular when individual entrepreneurs come with new innovations based on the products they used themselves in the first place. Hence, the improvements and innovations of these smaller firms would come from a more practical point of view and
20 should probably be based on prior designs and knowledge from the entrepreneur or initiator. In contrast to most literature, where the majority of research is based on larger firms, smaller firms could be more innovative with local knowledge instead of non-local knowledge as it is easier to find, understand and apply. Therefore, this paper suggests the following hypotheses:
H3a: Smaller firms are more innovative with local knowledge than large firms. H3b: Larger firms are more innovative with non-local knowledge than smaller
21
DATA AND ANALYTICAL METHOD
SAMPLE
The sample consists of 74 patents of the child carrier industry, referred to as class D3/213, from the period 1975 till 2004. The data is obtained from the U.S. Patent and Trademark Office (USPTO). The child carrier industry was chosen, as it is an industry with a relatively low barrier to enter. In addition, it seems to be an industry, which can be entered without specific knowledge or expertise. First hand experiences and new innovative ideas of individuals could be a reason to enter the industry. Consequently, it seems a good industry to investigate the different hypotheses.
From the 74 patents, 42 patents were not registered by firms. Thus, most of these patents belong to individuals. It may contain random individuals, start-ups, businesses based on sole proprietorship and small firms. However, it is reasonable to believe that larger firms register inventions on their own behalf (as assignee) as it increases status, negotiation power and the possibilities for value appropriation. Especially as it enables the company to buy or sell its intellectual property. In addition, it seems that large firms rely more upon patents than others (Cohen, Nelson & Walsh, 2000). Nonetheless, these 42 patents are considered as the ‘smaller firms’ group within the sample. However, it is possible that not all of these patent holders can be considered as a smaller firm. From the rest of the 32 patents that are appointed to a firm a selection was made to come up with one group that should represent active and larger firms. The selection criteria for this group is that 1) patents should be appointed to a firm itself, 2) the firm should have at least 3 patents or more within the industry and finally 3) at least 1 other patent within the time frame of 5 year of each registered patent. This resulted in 5 firms which hold 18 patents in total. In other words, these 5 firms represent 24% of the total patents
22 within the industry. Overall, the industry contains 74 patents. 18 patents are considered as the ‘larger firms’ group, while 42 patents are considered as the ‘smaller firms’ group. Normally all 74 patents are included. In case the two different groups are tested, all patents except the group itself will be excluded by reporting them as ‘missing data’.
THE USE OF PATENT DATA
The use of patents to study industry phenomena are recognized and criticized as well (e.g. Hall, Jaffe & Trajtenberg, 2001; Basberg 1987). In regard to innovation, Hall et al. (2001) and Basberg (1986) express their concerns as whether the use of patent data reflects inventive activity and innovation.
Firstly, a patent should have practical value in an industry to be a sufficient indicator for technological change. Otherwise, the number of patents that lead to innovation will not be significantly related. Secondly, for a comparison among competitors within an industry, the patent system should be used uniformly. To benefit from innovation other mechanisms such as secrecy and lead time could be in place (Cohen et al, 2000). Thirdly, comparison between countries and different patent office’s could result in an affected and weakened comparison. Fourthly, an invention is not an innovation. There are a lot of inventions that are not used at all. In addition, not all inventions are patented and even if they are they are not always used as well. Sometimes it is difficult (e.g. time consuming and costly) or impossible to patent an invention. To conclude, from a large amount of inventions, a relatively small part is patented and is under a patent-license used in an innovation. Finally, patent statistics do not make distinction between the patents that lead to basic innovations and those leading only to
23 incremental technical improvements (Hall, Jaffe & Trajtenberg, 2001; Basberg, 1987; Duijn, 1981).
Despite some weaknesses, patent data has some obvious strength as well. Firstly, patents provide unique information about the process of technological changes (Grilliches, 1998; Trajtenberg, 1990a). The data gives the possibility to access the patterns of innovations and innovation activities across industries and nations. Secondly, patenting has become an important way to protect intellectual property rights around the world. For example, in 2001 the USPTO granted patents to inventors from 106 foreign different countries. Thirdly, patents contain valuable and detailed information, however the data extract of this study was limited to patent number, title, filled and granted date, current U.S. class, backward references, forward references, applicant, assignee, inventors, self citations, other references, number of claims and finally attorney’s. This information can be used to analyze different aspects as will be elaborated later on. Finally, the most important strength for this paper is that patents include citations, also called references, to previous patents and literature. This makes it possible to trace linkages between patents and to research knowledge spillovers and the importance of individual patents (Merkx, 2013).
DEPENDENT VARIABLE
In this paper the dependent variable is Innovativeness and the number of forward citations a patent has received will measure it. Patents reveal something about the importance and innovativeness of an invention. In addition, the number of received citations could tell us something about the importance of the cited patent:
24 “..if a single document is cited in numerous patents, the technology revealed in that document is apparently involved in many developmental efforts. Thus, the number of times a patent document is cited may be a measure of its technological significance.”
(OTAF, 1976, p. 167, as cited in Hall, Jaffe and Trajtenberg (2001) on p.14-15.)
Hence, the number of forward references represents the number of citations, a particular patent has received from other patents.
INDEPENDENT VARIABLES
The variable Knowledge Origins should represent something about the spread of knowledge origins. In other words, the use of multiple but different sources of knowledge. It is based on the usage of, c.q. references to, different knowledge source categories. These categories are identical classes, closely related classes, non-related classes, self citations and finally other references. All of these categories are used as independent variables as well and will be further elaborated on. However, there are six categories, including the minimum possible value of zero. In this case the patent has no references. The maximum possible value is five, in which the patent has at least one reference in each of the categories. This variable does not count the number of references within a category, it only counts whether or not there is a reference in that particular category.
The variable Backward References is a continuous variable that measures the number of backward citations a patent has. A backward reference shows that the patent is based on prior knowledge of the particular reference: a patent. Thus, it implies that it based on existing or more local knowledge. In addition, how bigger number of backward references is how more
25 incremental the innovation will be. In other words, it reflects a limitation for non-local search and breakthrough re-combinations (Poldony & Stuart, 1995). However, backward references could exist of very wide range of subcategories which on their own could entail more information. Therefore, other variables are made based on the backward references. These variables are Identical Classes, Local, Closely Related Classes, Non-Related Classes and finally Non-Local.
The variable Activeness is a continuous variable that measures the activeness of the patent owner. This owner is based on the inventor or the assignee. If a company was mentioned it was considered as the owner of the patent. It is calculated by the maximum number of patents of the same patent owner within a time frame of five year of a particular patent.
The variable Self Citations is a continuous variable that measures the number of self citations by a patent. Self citation is a backward reference that refers to the work of the inventor of co-inventor itself.
The variable Other References is a continuous variable that measures all other type of references that are not U.S. Patents. This could be foreign patents, scientific work or articles from firms themselves. This variable could tell us something about the use of different knowledge origins.
The two variables Identical Classes and Local are based on backward references that refer to patents in the same class as the patent itself: D3/213 “Child Carrier”. Consequently, they refer
26 to knowledge from the same industry and thus to local knowledge in particular. Both variables are based on the same construction, however the Identical Classes variable is only used in combination with the variables Closely Related Classes and Non-Related Classes. The variable Local is only used in combination with the variable Non-Local. In short, when the variables Local and Non-Local are used, the Closely Related Classes are left out. This is done to create a bigger distinction between local and non-local knowledge.
The variable Closely Related Classes is based on backward references that refer to closely related patent classes. This variable should reflect the level between references in the same industry (local) and outside the industry (non-local). Or in other words, it reflects the level between Identical Classes and Non-Related Classes. The following U.S. Patent classes are defined as closely related: 224/158, 224/159, 224/160, 280/648, 280/642, 280/647 and D3/214. These class numbers represent, in the same order, in the 224 “Package and article carriers” class the following subcategories: ‘Carrier for person’, ‘In upright or sitting position’ and ‘Two attaching means (e.g., straps, etc.) crossing different shoulders’. In the 280 “Land vehicles” class the following subcategories: ‘Convertible’, ‘Three- or four-wheeled chair, baby carriage, or stroller ‘ and ‘Wheeled chair, stroller, or baby carriage’. In addition, in the D3 “Travel goods and personal belongings” class, which is the same class as the “Child Carrier” subcategory, the following class: ‘Backpack or sling type’.
The two variables Non-Related Classes and Non-Local are based on backward references that do not correspond with identical and closely related classes. Consequently, they refer to knowledge from outside the industry and thus to non-local knowledge in particular. Both variables are based on the same construction, however the Non-related Classes variable is
27 only used in combination with the variables Closely Related Classes and Identical Classes. The variable Non-Local is only used in combination with the variable Local. In short, when the variables Local and Non-Local are used, the Closely Related Classes are left out. This is done to create a bigger distinction between local and non-local knowledge.
CONTROL VARIABLES
The variable Law Firms reflects to the number of law firms that are involved. The number of law firms may refer to the total amount of skills involved for applying a patent, as every law firm may have its own specialization and expertise. Thus, the end result could be of higher quality as the cumulated level of experiences and skills may exceed that of patents that are written by fewer parties. On the other hand, it could also have negative effect as result of bad cooperation between the different legal parties (Lam, 2011).
The variable Inventors reflects the number inventors that are registered in a patent. Thus, its reflects the number of people involved and thus the cumulative knowledge base that was used for the invention. Additionally, it could be that different inventors have their own specialization. This could help a team of inventors to have a better understanding from different non-local knowledge sources. In addition, the number of inventors may say something about the spread of knowledge that was used.
The continuous variable Claims reflect the number of claims a patent has. This may say something about the “width” of the invention (Lanjouw & Schankerman, 2001). A number of
28 studies have shown that there is a positive relationship between the number of claims and the value of the patent (Lanjouw & Schankerman, 2001; Harhoff & Reitzig, 2004; Bessen, 2008).
STATISTICAL METHOD
This paper uses multiple regression analyses to model the dependent variable Y ‘Innovativeness’ as the estimated outcome of using some or all of the explanatory, independent variables. However, in this case the dependent variable outcomes can be considered as a ‘count’. In other words, the observations, c.q. the patents, can be seen as independent from each other. In addition, when all data are positive integers and independent of each other a normal distribution is not in place. As a result, the normality assumption of multiple regression is lost. In addition, as the independent variables are in the form of a discrete number, e.g. a count, they have to be transformed to model the data by linking the logarithm of the outcome variable to a linear function. This is done by the use a logistic transformation of the probability p or logit p. In the end, the logarithm of the independent variable is linked to a linear function of the dependent variable. It expresses the natural logarithm of the event or outcome of interest as a linear function of a set of predictors, the dependent variables (Simkiss, et al, 2014). In conclusion, a normal multi variable regression analysis was not the most appropriate. Instead, Poisson regression analysis was used with a logarithm function. To do so, IBM’s SPSS Statistics 20 was used by means of its Generalized Linear Models option with the Poisson distribution and the Log link function.
This paper uses four different test models. The third and fourth tests are used twice, as they test two different groups: ‘Smaller firms’ and ‘Larger firms’. All models will be based on a two tail test. Models are constructed as follows:
29 Log
e(Y) = β0 + β1X1+ β2X2+ …+ βnXn
With respectively the following variables per model: • Model 1 is build based on β
1 : Law Firms, β2 : Inventors, β3 : Claims, β4 : Knowledge Origins, β
5 : Backward References, β6 : Activeness.
This model should help to answer hypothesis H1 and H2.
• Model 2 is build based on: β
1 : Law Firms, β2 : Inventors, β3 : Claims, β4 : Identical Classes, β
5 : Closely Related Classes, β6 : Non-Related Classes, β7 : Self citations Classes, β8 : Other References.
This model goes in-depth into the different knowledge origins and examines the backward references in greater detail. It could help us to further understand H1 and H3. In addition, the model could give us a general idea about the industry .
• Model 3 is build based on: β
1 : Law Firms, β2 : Inventors, β3 : Claims, β4 : Identical Classes, β
5 : Closely Related Classes, β6 : Non-Related Classes.
This Model examines the influence of different knowledge origins, in terms of identical, closely related and non-related classes and their influence towards innovation. However, the model is used for two different groups: smaller firms and larger firms. The model should help to give an answer on hypotheses H3.
• Model 4 is build based on: β
1 : Law Firms, β2 : Inventors, β3 : Claims, β4 : Local, β5 : Non-Local.
30 This Model examines the influence of different knowledge origins, in terms of local and non-local and their influence towards innovation. However, the model is used for two different groups: ‘smaller firms’ and ‘larger firms’. The model should help to give an answer on hypotheses H3.
31
EMPERICAL RESULTS
REGRESSION RESULTS
The first test model shows the contribution of three different independent variables to the innovativeness of a patent owner. In all independent models the beta results are confirmed in model 5, which represents the complete model. In addition, the strong significance of all three variables hold as well. The strongest effect is found for the variable Knowledge Origins with a beta of 0,453 and strongly significant at 0,000. Followed by Activeness with a beta of 0,302 and strongly significant as well. A small negative effect was found for Backward References. See table 1 for detailed information.
TABLE 1 - Contributers to innovation
Model 1 Model 2 Model 3 Model 4 Model 5 B Sig. B Sig. B Sig. B Sig. B Sig. Law Firms -,609 ,000*** -,694 ,000*** -,503 ,000*** -,638 ,000*** -,528 ,000*** Inventors ,206 ,001*** ,157 ,009*** ,263 ,000*** ,093 ,197 ,209 ,004*** Claims ,107 ,000*** ,085 ,000*** ,127 ,000*** ,123 ,000*** ,144 ,000*** Knowledge Origins ,390 ,000*** ,453 ,000*** Backward References -,019 ,026** -,043 ,000*** Activeness ,356 ,000*** ,302 ,000***
*** 99% level of confidence (strongly significant). ** 95% level of confidence (significant).
* 90% level of confidence (weak significant).
The second test model shows the contribution of different knowledge origins to innovativeness. The model shows a positive result for the variables Identical Classes, Self Citations and Other References. In the complete model, number 7, all these variables have significant results. However, in the separate models Self Citations loses its significance. Thus, it should be excluded as a significant result. On the other hand, the variables Closely Related Classes and Non-Related Classes show a negative result of respectively -0,027 and -0,072. In
32 the complete model, number 7, all these variables have significant results. However, Closely Related Classes drops to 0,174 which can be considered as very weak in Model 3. Interesting to see is that the effects of the three variables Identical Classes, Closely Related Classes and Non-Related Classes gradually changes from positive to slightly negative to negative. See table 2 for detailed information.
TABLE 2 - Different knowledge origins and innovation
Model 1 Model 2 Model 3 Model 4 B Sig. B Sig. B Sig. B Sig. Law Firms -,609 ,000*** -,650 ,000*** -,550 ,000*** -,561 ,000*** Inventors ,206 ,001*** ,166 ,010*** ,244 ,000*** ,203 ,001*** Claims ,107 ,000*** ,080 ,000*** ,110 ,000*** ,126 ,000*** Identical Classes ,047 ,022**
Closely Related Classes -,016 ,174
Non-Related Classes -,057 ,000***
Self Citations Other References
Model 5 Model 6 Model 7 B Sig. B Sig. B Sig. Law Firms -,628 ,000*** -,771 ,000*** -,705 ,000*** Inventors ,205 ,001*** ,245 ,000*** ,282 ,000*** Claims ,107 ,000*** ,109 ,000*** ,112 ,000***
Identical Classes ,045 ,030**
Closely Related Classes -,027 ,033**
Non-Related Classes -,072 ,000***
Self Citations ,048 ,435 ,132 ,044** Other References ,080 ,000*** ,085 ,000***
*** 99% level of confidence (strongly significant). ** 95% level of confidence (significant).
* 90% level of confidence (weak significant).
The third test model examines the influence of different knowledge origins towards the innovativeness of firms. This is done for smaller firms (N=42), see table 3, and larger firms
33 (N=18), see table 4. The different knowledge origins are Identical Classes, Closely Related Classes and Non-Related Classes.
For smaller firms there is a significant result for a negative influence of Non-Related Classes and a weak significant result for a positive influence of Identical Classes. Respectively, the results are -0,068 and 0,064. For Closely Related Classes no significant result was found.
For larger firms the only variable that holds significance is Closely Related Classes, as it shows a negative effect of -0,155 at a significance level of 0,000. Non-Related Classes seem to have a positive effect, as in model 4 an effect of 0,057 was found with a significance level of 0,022. However, the significance level drops dramatically in the complete model 5 to a 0,275. Remarkably, the influence of control variable Claims is close to zero.
*** 99% level of confidence (strongly significant). ** 95% level of confidence (significant).
* 90% level of confidence (weak significant). TABLE 3 - Smaller firms and different knowledge origins (N= 42)
Model 1 Model 2 Model 3 Model 4 Model 5 B Sig. B Sig. B Sig. B Sig. B Sig. Law Firms -1,391 ,000*** -1,422 ,000*** -1,388 ,000*** -1,258 ,000*** -1,246 ,000*** Inventors ,700 ,000*** ,698 ,000*** ,701 ,000*** ,607 ,001*** ,602 ,001*** Claims ,113 ,000*** ,083 ,000*** ,113 ,000*** ,132 ,000*** ,100 ,000***
Identical Classes ,052 ,065* ,064 ,025**
Closely Related Classes -,001 ,940 -,020 ,322 Non-Related Classes -,064 ,006*** -,068 ,003***
34 TABLE 4 -Larger firms and different knowledge origins (N= 18)
Model 1 Model 2 Model 3 Model 4 Model 5 B Sig. B Sig. B Sig. B Sig. B Sig. Law Firms ,156 ,359 ,182 ,302 ,775 ,000*** ,232 ,182 ,948 ,000*** Inventors ,109 ,431 ,115 ,411 ,319 ,020** ,177 ,218 ,410 ,006***
Claims 0a 0a 0a 0a 0a
Identical Classes -,020 ,567 -,096 ,017**
Closely Related Classes -,149 ,000*** -,155 ,000*** Non-Related Classes ,057 ,022** ,031 ,275
*** 99% level of confidence (strongly significant). ** 95% level of confidence (significant).
* 90% level of confidence (weak significant).
The fourth test model examines the influence of different knowledge origins towards the innovativeness of firms. This is done for smaller firms (N=42), see table 5, and larger firms (N=18), see table 6. The different knowledge origins are Local and Non-Local.
For smaller firms there is a positive effect for the influence of Local knowledge. However, the significance levels are weak in the individual model at 0,065. However, in the complete model it becomes significant at 0,044 while showing an effect of 0,056. On the other hand, for Non-Local knowledge the effects are significant negative. In model 5, it shows an effect of -0,066 whit a p-value of 0,004.
35 TABLE 5 -Smaller firms and different knowledge origins (N= 42)
Model 1 Model 2 Model 3 Model 4 B Sig. B Sig. B Sig. B Sig. Law Firms -1,391 ,000*** -1,422 ,000*** -1,258 ,000*** -1,298 ,000*** Inventors ,700 ,000*** ,698 ,000*** ,607 ,001*** ,598 ,001*** Claims ,113 ,000*** ,083 ,000*** ,132 ,000*** ,101 ,000***
Local ,052 ,065* ,056 ,044**
Non-Local -,064 ,006*** -,066 ,004***
*** 99% level of confidence (strongly significant). ** 95% level of confidence (significant).
* 90% level of confidence (weak significant).
For larger firms there is are opposite results compared to the smaller firms group. However, only the positive effect of Non-Local knowledge is significant. The negative effects of Local knowledge are not considered as significant. Despite that, for larger firms there seems to be a positive and significant influence of Non-Local knowledge towards innovation. Remarkably, the influence of control variable Claims is close to zero.
TABLE 6 -Larger firms and different knowledge origins (N= 18)
Model 1 Model 2 Model 3 Model 4 B Sig. B Sig. B Sig. B Sig. Law Firms 0,156 ,359 0,182 ,302 0,232 ,182 0,297 ,109 Inventors ,109 ,431 ,115 ,411 ,177 ,218 ,196 ,185
Claims 0a 0a 0a 0a
Local -,020 ,567 -,038 ,294
Non-Local ,057 ,022** ,065 ,015**
*** 99% level of confidence (strongly significant). ** 95% level of confidence (significant).
36 HYPOTHESES RESULTS
The regression results as discussed above can be interpreted to answer the hypotheses. Test model 1 shows a significant and positive for the influence of Knowledge Origins on Innovativeness. Therefore, hypothesis 1 is confirmed. In addition to model 1, test model 2 shows that it depends upon the type of knowledge origin whether or not it contributes to the Innovativeness. In addition, test model 1 shows a positive and significant result for the Activeness of parties and the innovativeness. Consequently, hypothesis 2 is confirmed as well.
When looking at the results of models 3 and 4, respectively tables 3,4,5 and 6, it becomes clear that model 3 and 4 have complementary and confirmative results. However, due to lack of significant p-values not all results can be included.
For test models 3 and 4, the variables Identical Classes and Local presents local knowledge. Model 3 shows a weak significant and positive result for the use of local knowledge for smaller firms. This is confirmative with model 4, which shows a weak significant but positive result with local knowledge as well. Meanwhile, larger firms show negative effects for the use of local knowledge in model 3 and 4. However, these results are not significant. In result, hypothesis 3a cannot be fully confirmed nor rejected: smaller firms are more innovative by using local knowledge. However, larger firms do not show similar results. Consequently, it can be stated that smaller firms are more innovative by using local knowledge, but a comparison with larger firms does not seem legitimate.
37 For test models 3 and 4, the variables Non-Related Classes and Non-Local presents non-local knowledge. The group smaller firms shows a, weak significant, negative effect on Non-Related Classes and a significant negative effect on Non-Local as source of knowledge. At the same time, in model 3 and 4 a positive effect for the use of non-local knowledge by larger firms was found. However, only the results of model 4, by means of variable Non-Local, show a significant effect. In others words, larger firms are more capable of putting knowledge from non-related classes into innovative patents compared to smaller firms. Therefore, hypothesis 3b can be confirmed.
38
DISCUSSION
The results of this paper have some limitations which should be discussed before the impact of the results will be elaborated. Firstly, not all test results are significant. Some are weakly significant and a real minority are strongly significant. Despite this, some interesting results came up, and future research could benefit from a bigger sample size for more stronger and significant results.
Secondly, it is important to see these results in their context. Different industries could give completely different results. Hence, these results cannot be generalized to all industries. It is reasonable to believe that the outcome of this study could hold in other industries if various conditions are met. This paper argues for at least two conditions. One of them is the ease of entry by new firms or entrepreneurs. Most important is that an industry is limited in terms of complexity and the required basic knowledge. Consequently, it is most likely that this will be industries in which people have gained practical experiences first hand. For example numerous sport related industries fall in this category. Take for example a sailor who experienced an inadvertent moving mast foot which negatively effects the trim of the boat during the races. Frustrated by these negative effects the sailor came up with a new redesign of the mast foot. After the success of the design was proven in practice the sailor decided to commercialize the idea to other types of boats as well.
Another condition is that the initial capital requirements that are needed to create or develop their inventions and potential businesses should be limited. Overall, it should be easy to start a new business and to enter the market: a small improvement, based on local knowledge, or an
39 entrepreneurial idea, probably based on a limited amount of non-local knowledge, should be enough to exceed certain thresholds to register it as a patent and potentially start a business.
Thirdly, despite the differences that were found for non-local knowledge between large firms (positive) and smaller firms (negative), the individual patent holders are not examined in detail. Their actual firm size, turnover and corporate structure could tell us more about their ability to pursue ambidexterity. In addition, that the groups smaller firms and larger firms uniformly represent smaller and larger firms respectively are not guaranteed. Future research could perhaps do an in-depth analysis of the patent holders as well to come up with more specific insights.
The general consensus that the use non-local knowledge leads to better performance is, according to the test results of this paper, partially flawed. Ambidexterity could contribute to the performance of firms, however it depends upon types of firms and industry or market conditions.
The positive influence of ambidexterity by firms to succeed over the long term have been widely recognized (e.g. Birkinshaw & Gibson , 2004; Teece, 2007). However, as argued by Lubatkin et al. (2006) there are limitation in which ambidexterity contributes to the performance of firms. As argued by Labatkin et al. (2006), small and medium enterprises often lack the resources and ‘hierarchical administrative systems’ to execute ambidexterity successfully. It depends upon the type of firm whether or not they have the capabilities and resources to conduct and benefit from ambidexterity.
40 This paper argues that certain nuances should be in place as different types of knowledge have different influences in different situations. Firstly, smaller firms often have pressure to survive and to reach a minimal turn-over. As discussed by Rosenberg (1996), distant knowledge can be arduous and costly as the gained knowledge may never result in a valuable idea. It could be that smaller firms prefer non-local knowledge direct from the start as possible economic value could be seen and realized more easily.
Secondly subsequent, knowledge is difficult to identify and acquire through market mechanism (Teece, 1980; Von Hippel, 1994; in Miller et al., 2007). For smaller firms the collective knowledge base is, overall, probably smaller than that of larger firms. Therefore, it could be more difficult to find, understand and apply non-local knowledge. In other words, larger firms with more employees or even divisions with specific knowledge areas could interpret non-local knowledge more easily as there is more chance of overlap. More people should have more areas of expertise, thus they will understand non-local knowledge easier as the potential knowledge gap is relatively small: there is certain overlap with their own knowledge the non-local knowledge. This will make it easier to interpret and apply.
Thirdly, as mentioned above, multiple divisional firms could have an advantage, as knowledge gaps are probably smaller. The bigger collective knowledge base, the bigger the potential knowledge overlap. In addition to this, Miller et al. (2007) showed that interdivisional knowledge has a stronger effect on innovation than knowledge from within the divisional boundaries or outside the firm boundaries. Thus, larger firms that have the capability to fully benefit from their divisional structure could be more innovative with
41 knowledge from different industries than non-divisional firms. As already discussed in the literature review, it is not presumable that smaller firms have a multi divisional corporate structure. Hence, this could explain why larger firms have more positive results with non-local knowledge than smaller firms and vice versa.
Fourthly, smaller firms may not have the capabilities, resources or need to pursue ambidexterity in that particular moment of time. As the different corporate structures discussed by Raisch (2008), it is more likely that smaller firms use a temporal design in order to pursue exploration. In a temporal design structure, exploitation and exploration are emphasized sequentially rather than simultaneously. If the necessity has arrived to look outside the industry, it could be that the entire firm focuses on the task of exploration. Once new ideas, knowledge and potential commercial value is secured, the firm could move back to its exploitation setting. Hence, it could continue its operations in the same, original industry or start with new products and ideas in a new or relatively new industry. If the business continues, it will probably use this gained non-local knowledge to continue its innovation processes till the necessity of exploration has arrived again. Thus, in these exploitation phases it is mainly working with local knowledge that it once had gained as non-local knowledge.
Finally, smaller firms are often driven by entrepreneur or based on a entrepreneurial spirit. As argued by Lubatkin et al. (2006), there is an essential role for top management to actively support, in terms of behavior, the pursue of ambidexterity. It could be that the entrepreneurs, or specific the entrepreneurs in this child carrier industry, not have the urge to actively pursue ambidexterity.
42 Future research could confirm the test results in other industries, preferably with bigger sample sizes. It would be interesting to test in which conditions ambidexterity and different types of knowledge have a positive influence on the innovativeness or not. In addition, multiple situations and conditions could be examined. For example the dynamic process of smaller firms that use the temporal design.
43
CONCLUSION
The conclusion of this paper is that knowledge used from different origins could result in higher levels of innovation. This paper argues that the use of a wide range of knowledge origins could positively influence the innovativeness of firms. In addition, firms that are actively working on their knowledge base are more innovative as well.
However, most important finding of this paper is that different types of firms are sensitive to different types of knowledge sources to become innovative. Smaller firms are more sensitive to local-knowledge. It could be that smaller firms are more attracted towards local-knowledge as it has a clearer path to potential commercial value. In addition, a smaller firm could have a limited amount of collective basic knowledge to understand knowledge from outside the industry. In contrast to larger firms which may have multiple division, specialized research or innovation employees or just a larger amount of collective basic knowledge. These larger firms may have the capability to have more potential overlap with their own basic knowledge and the non-local knowledge. Thus, they could have an advantage over smaller firms to understand and to see potential value of non-local knowledge. Additionally, it could be that they have a better idea of potential combinations of local knowledge. In addition, as non-local search may be time consuming and costly, it could be that smaller firms do not have the resources to do so. For many small firms the highest priority is to survive and to reach a minimal turnover. Consequently, non-local search could be too expensive, has lower change of a successful contribution to the firms results or has a lower priority. Finally, larger firms often have multiple divisions. As elaborated in the literature review, information between division leads to higher levels of innovation than knowledge from within the divisional boundaries or outside the firm (Miller et al, 2007). Thus, larger firms with multiple divisions
44 in different industries could have an advantage. If these different division all hold non-local knowledge, seen from another division perspective, it could be more capable of using this non-local knowledge in an effective way.
Overall, this paper argues that there are limitations on the contribution of ambidexterity and local and non-local search towards the innovativeness of firms. Especially firm size, type of firm and industry conditions such as complexity, required knowledge and general ease of entry are of great influence on the impact of different knowledge origins. Further, it argues that the positive influence of ambidexterity should be nuanced.
45
REFERENCES
Basberg, B.L. (1987). Patents and the measurement of technological change: A survey of the literature. Research Policy, 16, 131-141.
Barnett,W. P., Greve, H. R.,& Park, D. Y. (1994). An evolutionary model of organizational performance. Strategic Management Journal, 15, 11–28.
Barney, J.B. (1991). Firm Resources and Sustained Competitive Advantage. Journal of Management, 17, p. 99-120.
Barney, J.B & Peteraf, M.A. (2003). Unraveling The Resource-Based Tangle. Managerial and Decision Economics, 24, p.309-323.
Bessen, J. (2008). The value of US patents by owner and patent characteristics. Research Policy, 37(5), 932-945.
Birkinshaw, J. & Gibson, C. (2004). Building Ambidexterity Into an Organization. MIT Sloan Management Review, Vol. 40 (4), 46-55.
Burgelman, R.A. (2002). Strategy as Vector and the Inertia of Coevolutionary Lock-in. Administrative Science Quarterly, 47 (2), 325-357.
46 Chen, B.X. and Bilton, N. (2014). Building a Better Battery. The New York Times, Technology, Feb. 2 2014. http://nyti.ms/1agQ9Oo
Chesbrough, H. & Rosenbloom, R.S. (2002). The role of the business model in capturing value from innovation: evidence from Xerox Corporation’s technology spin-off companies. Industrial and Corporate Change, Vol. 1, No. 3, 529-555.
Cohen, W. S, Nelson, R.R., Walsh, J.P. (2000). National Bureau of Economic Research. Protecting their Intellectual assets: Appropriability conditions and why U.S. manufacturing firms patent (or not) Working Paper 7552 http://www.nber.org/papers/w7552
Cooper, A.C. & Smith, C.G. (1992). How established firms respond to threatening technologies. Academy of Management Executive. Vol. 6, No. 2: 55-70.
Dedrick, J. & Kreamer, K. and Linden, G. (2009). Who profits from innovation in global value chains?: a study of the iPod and notebook PCs. Industrial and Corporate Change, Vol. 19, No. 1, 81–116.
DeGeus, A. (1997). The living company: Habits for survival in a turbulent business environment. Boston, MA: Harvard Business School Press.
47 Duncan, R. B. (1976). The ambidextrous organization: Designing dual structures for innovation. In R. H. Kilmann, L. R. Pondy, & D. Slevin (Eds.), The management of organization design: Strategies and implementation, 167–188. New York: North Holland.
Edmondson, A. (1999). Psychological safety and learning behavior in teams. Administrative Science Quarterly, 44, 350–383.
Eisenhardt, K., & Brown, S. (1998). Time pacing: Competing in markets that won’t stand still. Harvard Business Review, 60–69.
Eisenhardt, K.M. & Martin, J.A (2000). Strategic Management Journal, Vol. 21, No. 10/11, Special Issue: The Evolution of Firm Capabilities 1105-1121.
Feenstra, R. C. (1998). Integration of trade and disintegration of production in the global economy. Journal of Economic Perspectives, 12 (4), pp. 31-50.
Foss, N. (1996). Research in strategy, economics, and Michael Porter. Journal of Management Studies, 33: 1-24.
Fleming, L. (2001). Recombinant uncertainty in technological search. Management Science, 47: 117-132.
48 Galunic, D. C, & Rodan, S. (1998). Resource recombinations in the firm: Knowledge structures and the potential for Schumpeterian innovation. Strategic Management Journal, 19: 1193-1201.
Gavetti, G. (2005). Cognition and hierarchy: Rethinking the microfoundations of capabilities' development. Organization Science, 599-617.
Grant, R. (1996). Toward a knowledge-based theory of the firm. Strategic Management Journal, 17 (Winter special issue): 109-122.
Griliches, Z. (1998). Patent Statistics as Economic Indicators: A Survey. 287 – 343.
Govindarajan, V. & Trimble, C. (2005). Building breakthrough businesses within established organizations. Harvard Business Review, 1–11.
Hall, B. H., A. B. Jaffe, and M. Trajtenberg (2001). "The NBER Patent Citation Data File: Lessons, Insights and Methodological Tools." NBER Working Paper 8498.
Hamel, G. (1998), Strategic Flexibility: Managing in a Turbulent Economy, Wiley, Chichester, UK.
