سهام دانش، اکتشاف و نوآوری: پژوهش در صنعت دستگاه های الکتروپزشکی ایالات متحده

This study applies the knowledge-based view of the firm to examine the relationships between exploration, characteristics of knowledge stock, and innovative performance. The article argues that the effectiveness of exploration on innovation is contingent upon two dimensions of knowledge stock: knowledge depth and knowledge breadth. Empirical findings from the US electromedical device industry between 1990 and 2000 provide support for this contingency argument.

The competence–rigidity paradox is an interesting topic in strategic management research (Leonard-Barton, 1992, Levinthal and March, 1993 and March, 1991). While both exploration and exploitation are essential for firm success and are dependent on each other, how to effectively balance and manage these two processes remains unclear in the literature (Atuahene-Gima, 2005). Some researchers claim that solving this paradox is one of the toughest managerial challenges in creating and sustaining competitive advantage in dynamic environments (e.g., Abell, 1999 and Williamson, 1999). Previous studies attempt to resolve this paradox by examining how exploration and exploitation influence firm performance in various settings (e.g., Atuahene-Gima, 2005, Katila and Ahuja, 2002 and Rosenkopf and Nerkar, 2001). For instance, Rosenkopf and Nerkar (2001) examine the effects of search behaviors in the optical disk industry and find that the impact of technological innovations depends on the degree of boundary-spanning exploration activities. Considering exploration and exploitation as two distinct processes rather than as a continuum, Katila and Ahuja (2002) find, in the robotics industry, that exploration and exploitation are complementary when new products are introduced. However, while these empirical studies are illuminating, few studies have taken into consideration other contingency factors that influence the linkage between exploration–exploitation and innovative performance, such as the firm's existing knowledge stock. As Nelson and Winter (1982: 172) point out: “Real search processes take place in specific historical contexts, and their outcomes clearly depend in part on what those contexts contain in the way of problem solutions that are available to be ‘found’.” This study fills in this research gap by applying the knowledge-based view to investigate how exploration and knowledge stock interact with each other and influence firm performance. In particular, this study addresses the above question in the setting of technological innovation, and focuses on the moderating role of current knowledge stock on the effect of exploration. The recent knowledge literature suggests that the growth of firm knowledge is a function of knowledge stock as well as continuous search for new knowledge elements and potential integration opportunities (e.g., Kogut and Zander, 1992). Knowledge stock reflects the amount of knowledge elements that a firm has accumulated over time (Dierickx and Cool, 1989) and is embedded in organizational routines, technologies, employees, and other types of resources (Grant, 1996). Exploration captures the extent to which new knowledge is acquired (Katila and Ahuja, 2002). More specifically, exploration in this study refers to the integration of new knowledge elements that locate outside a focal firm. For instance, Firm A may cite a patent that has never been cited by itself before. Then this patent becomes a new knowledge element to Firm A, and Firm A is exploring something new from the outside world. Existing theories suggest that knowledge stock and exploring new knowledge are inter-dependent in that existing knowledge stock not only provides incentives to acquire new knowledge, but also shapes the scope and direction of future exploration (Katila and Ahuja, 2002 and Penrose, 1959). On the other hand, newly generated knowledge is converted to knowledge stock and thus facilitates the growth of knowledge stock (Smith et al., 2005). The present study examines the joint effects of knowledge stock and exploration on innovative performance and pursues two research questions. First, how does a firm's exploration influence its innovative performance? Second, how does a firm's existing knowledge influence the relationship between exploration and the firm's innovative performance?

How to solve the competence–rigidity paradox is challenging to management researchers. This study applies a contingency perspective to tackle this paradox in the setting of the US electromedical device industry, where innovation competence is one of the most important sources of sustained competitive advantage (Cantwell, 1999 and Schumpeter, 1934). This study argues that in order to develop and maintain such innovation competence, an appropriate level of exploration is necessary, which is affected by the characteristics of the firm's existing knowledge stock. In other words, a maximum innovative performance is achieved by matching exploration intensity with the depth and breadth of the firm's current knowledge base. The results provide strong evidence that the impact of exploration on innovation is curvilinear. That is, as the intensity of exploration increases, the amount of newly created knowledge first increases. But after a certain point, the innovation output will drop down, exhibiting a decreasing marginal return to exploration. This finding is consistent with March's (1991) argument for the balance between exploitation and exploration. Previous studies identify three types of learning traps: (1) hubris traps—too optimistic about the future of technological search; (2) over-exploration—search too wide beyond current absorptive capacity; and (3) over-exploitation—search too narrow within current knowledge domain and ignore external technological changes (Levinthal and March, 1993). This study provides support for the over-exploration trap. The key findings indicate the significant moderating role of existing knowledge stock on the relationship between exploration and innovative performance. While Kogut and Zander (1992) already point out that new knowledge generation is a function of existing knowledge stock and further new knowledge integration, empirical studies using large sample data are rather limited. The present study examines this issue in the setting of the US electromedical device industry and finds that the effect of exploration on innovative performance is complex and contingent on the characteristics of the knowledge stock. In particular, this study observes that a continuously increasing effort of exploration is helpful when the firm has a narrow knowledge base; however, as the knowledge breadth increases, a moderate level of exploration is more productive in terms of new knowledge generation. Therefore, while current innovation research and prescriptions highlight the importance of exploration for new knowledge elements and suggest that exposure to and integration of new knowledge help create technological breakthroughs and avoid core rigidities, the findings emphasize the moderating role of accumulated knowledge stock. This study also extends the literature on the dynamic capabilities of firms by examining the interaction of exploration and knowledge stock. According to Teece et al. (1997), technological innovation is path-dependent in that previous knowledge stock constrains its subsequent behaviors. Similarly, Cohen and Levinthal (1990) argue that knowledge depth and breadth are key determinants of a firm's absorptive capacity. The findings suggest that history does matter: previously accumulated knowledge stock influences the effectiveness of exploration. Prior empirical research tends to examine the separate role of search behaviors (e.g. Katila and Ahuja, 2002 and Rosenkopf and Nerkar, 2001) and ignores how search behavior is constrained by existing technological environments. As Penrose argues, “… unknown and unused productive services [from existing resources] immediately become of considerable importance, not only because the belief that they exist acts as an incentive to acquire new knowledge, but also because they shape the scope and direction of the search for knowledge … both an automatic increase in knowledge and an incentive to search for new knowledge are, as it were, ‘built into’ the very nature of firms possessing entrepreneurial resources of even average initiative.” (Penrose, 1959: 77–79) Thus, knowledge context should influence the relationship between search behavior and innovative performance. The findings provide strong empirical support for the above argument by showing that the characteristics of knowledge stock moderate the effect of exploration on innovation. Future research can extend this study by taking a more dynamic view of the relationship between knowledge stock and knowledge flow. That is, while existing knowledge base influences the effectiveness of exploration and exploitation, these search activities will also affect the level of depth and breadth of knowledge base in the next round. Such a loop between knowledge stock and knowledge flow will continue over time, which subsequently influences the innovative performance of the firm. Examining how such interaction evolves over time is an interesting question. This study has several limitations. First, the study focuses on the innovative performance of public US firms and thus the findings may not generalize to smaller or private US firms. The study also limits attention to a single industry, which may limit the potential for generalizing the results. Future work can include other high-tech industries and assure a better distribution of firms by size and governance arrangements. This study proxies a firm's knowledge base by patents, which is a legal form that protects a firm's intellectual properties. However, not all firms protect their technological assets with patents. Some firms may prefer to keep their know-how as trade secrets. New knowledge in other industries may not be as conducive to patent protection as medical devices are. Thus, limitations occur when patents are applied to measure the depth and breadth of knowledge base. Future work can consider the distribution of different types of intellectual properties in a firm's knowledge base. Another limitation relates to the ignorance of different technological classes that a patent is applied in. Despite the fact that this study narrows its research focus to a specific industry setting, variations in innovation opportunities still exist across technological classes. Finally, this study does not examine the quality of innovation. As Nelson and Winter (1982) and March (1991) suggest, local search and distant search may result in different types of innovation. Different types of search may impact future technologies in varied ways (Rosenkopf and Nerkar, 2001). Therefore, the need is for additional research on the effects of technological search on the quality of innovation.