ایجاد فن آوری های نوین از طریق اتحادیه: تحقیقات تجربی از ثبت اختراعات مشترک

Why are some alliances more productive than others in terms of creating new technology? Using a novel measure of alliance performance, that is, joint patents, this study aims to tackle this question. Our results from the global pharmaceutical industry show that joint invention has an inverted U-shape relationship with a path-dependent technology base, with the level of joint patents initially increasing and then decreasing beyond a certain level of path dependence. The results also show that joint patents are more numerous when the alliance partners have had prior ties with each other. Overall, the finding suggests that creating new technology through alliances can be facilitated by ensuring the positive side of absorptive capacity, while avoiding its downside.

Firms frequently use partnerships to acquire new technology (Powell et al., 1996), to pool complementary technologies (Teece, 1992), or to share the costs of exploiting a certain form of technology (Nakamura et al., 1996). In some rapidly evolving sectors, where the locus of proprietary knowledge is dispersed across companies and shifts quickly over time, the pooling of resources can lead to superior and faster technological development than would be possible internally (Doz and Hamel, 1997). In addition, collaborative R&D arrangements have been rapidly growing since the 1980s in high-tech industries (Hagedoorn, 2002). Although the notion is well accepted that alliances can be useful for collaborative innovation, there is scant empirical evidence to support it. This lack of evidence reflects the difficulty in evaluating innovation output resulting from alliances. Recent studies of alliances, especially technology-based alliances, have investigated the association between alliances and innovation by looking at patents of individual firms as an indicator of innovation output (e.g. Ahuja, 2000; Stuart, 2000). However, output measured at the level of the individual firm is difficult to attribute to alliance-related activities because various exogenous factors influence the innovation output of individual firms. We attempt to remedy this situation by examining joint patents resulting from the collaborative efforts of alliances. We argue that if firms engage formally in collaborative R&D, and if the output of the R&D is measurable by patent indicators, then patents which are co-assigned to both partners in an alliance should be good measures of innovative output resulting from the alliance. By definition, the co-assigned patent that we call the joint patent is assigned to and jointly owned by more than one inventor. To constitute a co-assigned patent, it is necessary that all the inventors involved make some contribution to the final invention. Here we focus our attention on inter-firm joint patents, tracking only patents owned jointly by two or more alliance partners. According to Hicks and Narin (2001), co-assigned patents accounted for about 0.2% of US patents in the early 1980s, but the percentage rose to 1.4% in 1999. The percentage of co-assigned patents varies across sectors, with the two highest being 7% for biotechnology and 5.6% for pharmaceuticals. This salience of joint patents in these areas may be explained by the considerable scientific and technological interdependence among firms in pharmaceutical R&D. Modern pharmaceutical R&D is increasingly complex and demands an ever-widening range of skills. No single firm possesses all the knowledge, skills and techniques required (Powell et al., 1996). Accordingly, joint invention (and the collaboration that precedes it) often results from the need for complementary expertise. This study is one of the early attempts to use joint patents as a measure of the innovation output of alliances. Hence, there are few studies which identify factors which predict the existence of joint patents. We therefore draw upon the literature on absorptive capacity and alliance learning, wherein it is argued that alliances provide a platform for learning and innovation. We focus on both the technological and relational aspects of alliance partners, and their effects on learning and innovation. Specifically, using patent citation data, we identify two key technology-related variables: path dependencies and pre-existing technology overlap between alliance partners. In addition, given many firms’ aversion to sharing the ownership of proprietary technologies (Hagedoorn, 2003), we also introduce the variable of “repeated” alliance ties as a proxy for achieving a threshold level of trust, which is necessary to ease the appropriability risks of sharing proprietary technologies. Using a negative binomial regression model, we examine the impacts of these technological and relational variables on joint patenting in the context of the global pharmaceutical industry.

We began our search for jointly assigned patents on the premise that the inventive output that presumably occurs in an alliance should result in joint ownership. Previous work investigating coassignees of patents found that joint ownership of patents between firms is rare, except for cases involving individual inventors from universities and research laboratories (Hicks, 2000). It seems more likely that behind each joint patent is a unique and substantive story of firms coming to share technology despite their natural distaste for doing so. What, then, makes independent firms form alliances and file joint patents? We can make a simple assumption that the observed growth in joint patents is evidence of an increase in R&D collaborations. However, there is a danger of assuming that all partnerships lead to joint patents. A more realistic assumption is that only successful partnerships generate joint patents. Then we must ask, under what circumstances are alliances more successful in terms of generating joint patents? Our results suggest that prior alliance experience between allies is one of the factors in providing an impetus for joint patenting. Alliances are designed to meet the goals both of individual firms and of the collective whole, and are successful when the value of collective outcomes exceeds the opportunity costs that are incurred by participants (Jarillo, 1988). If firms have worked together in the past, they will have a basic understanding about each other's collaborative ways. Furthermore, prior ties between alliance partners can generate an initial base of inter-partner trust. This is a process of developing the relational routines necessary for allies to proceed further to the joint ownership of technologies. A one-time partnership may not be enough to develop such relational routines. Working together repeatedly provides time for the development of relational routines and inter-partner trust which, in turn, provide an impetus for technology sharing and joint ownership. Our data did not support the technology overlap argument, which points mainly at similarity benefits based on the feasibility and ease of knowledge transfer and acquisition. The problem of the similarity arguments is that an enhanced efficiency of cooperation that cannot come along with the diverse learning opportunity has little value. According to a resource-based view, alliance partners seek resources that are complementary to their own, and that resource complementarity leads to value creation (Grant and Baden-Fuller, 2004). Perhaps, similarity in resources provides fewer new skills to learn and thus may be used for exploitation rather than exploration. When we developed hypotheses regarding technology, we drew primarily on evolutionary economics (Nelson and Winter, 1982). A key argument of evolutionary economics is the path-dependence argument: technology develops locally. In other words, firms search for new technologies in areas that enable them to build upon their established technology bases (Stuart and Podolny, 1996). As a result, technological development is always, to some extent, localized within firms and path-dependent (Cantwell, 1994). We used the self-citation ratio as a proxy for a path-dependent technology base, and hypothesized that a path-dependent, internally accumulated technology base facilitates the collective effort of joint invention. However, the relationship between path-dependent technology and joint invention is non-monotonic. Specifically, although path-dependent technology initially has positive effects on joint invention, this relationship turns negative as firms develop a sense of self-sufficiency and a tendency to look inward. At this point, path dependence negatively influences a firm's engagement in joint invention. Firms that have established a more solid, firm-specific technological trajectory are less willing to search for new technology outside their corporate boundary. Thus, the relationship between the level of path-dependent technologies and joint invention is an inverted U-shape, as supported by our regression results. In addition to addressing a previously unexplored empirical question, this paper deepens our understanding of how a firm can learn with and from alliance partners. Our finding that the negative motivational effects of existing technological capabilities within a firm exceed the positive effects of absorptive capacity beyond a certain threshold level, has implications for research in the management of innovation, which stresses the importance of external knowledge. Although absorptive capacity is viewed as a source of competitive advantage (Cohen and Levinthal, 1990), most research that advances this perspective downplays the potential negative consequences of such capabilities. Because a firm with a strong existing knowledge base is more likely to have established idiosyncratic technological trajectories, and thus to exhibit path-dependent search behavior, its knowledge base may reduce its receptivity to externally sourced knowledge. Firms that fit this description face the challenge of balancing and building their own exploitative and exploratory abilities (March, 1991). Our result suggests that the fostering of invention through alliances can be facilitated by ensuring positive path dependence, and thus avoiding the limitations that arise from path dependence. Our use of a joint patent variable as a measure of alliance performance deserves some explanation. One measure of innovative outcomes from alliances that has been used in the past is the change in the number of patents held by an individual firm. However, the link between the change in the number of patents and a specific alliance is tenuous at best, as attributing a change in a firm's patenting activity to any specific alliance is difficult (Steensma and Corley, 2000). Taking a different tack on patent data, Mowery et al. (1996) introduced patent citation-based measures of technology flow between firms, on the premise that an increase in citations is indicative of the degree to which the citing firm is acquiring technology-based capabilities from the cited firm. This measure of change in patent citations, compared to the measure of the number of patents, is more alliance-specific, since it is directly related to the two firms in an alliance. However, we note that citation typically takes place impersonally, and can be nothing more than the acknowledgement of the debt to prior technology which is available to be cited. Furthermore, the association between alliances and patent citations is more complex: depending on the motives and forms, participation in alliances may produce either an increase or decrease in the cross-citation ratio (Mowery et al., 1996). We attempted to overcome these limitations by measuring joint patents subsequent to alliance formation. Presumably a joint patent involves a qualitatively different kind of interaction compared to that of a citation. Joint patenting is often a result of successful joint R&D collaboration, wherein researchers from different firms interact face-to-face, exchange their ideas, and solve problems jointly. Thus, joint patents between firms are interesting examples of collective innovation output arising from formal inter-firm collaboration. We argue that joint patents are better indicators of the difference in inventive output across alliances, compared to the number of patents filed or citation of patents by a single alliance partner. However, the joint patent measure is not without limitations. Inter-firm joint patenting is a rare event (Hicks and Narin, 2001). Alliances oftentimes do not lead to a joint patent (Hagedoorn et al., 2003). Most importantly, the indivisibility of single inventions from small scale and informal collaborations is responsible for the growth of joint patenting (Hagedoorn, 2003).3 While recognizing the limitations, it is nevertheless generally accepted that inter-firm joint patenting signifies the completion of a cooperative R&D, and the opening of another step toward subsequent developments and commercialization. This research suggests potentially fruitful venues for future research. That joint invention is a possible motive for collaborative arrangements is generally accepted in the alliance literature. Despite the well-accepted maxim that “invention breeds collaboration,” few studies have examined whether “collaboration breeds invention.” With emphasis on this causality, we argue that collaborative arrangements, if they are to produce anything of substance, should produce a joint invention, and the joint invention should result in joint ownership, i.e. a joint patent. Historically, joint patents are rare, but their numbers have increased in recent years, especially in the pharmaceutical industry, due to the increasing technological interdependence among firms in pharmaceutical R&D. However, only a much larger investigation that includes various industries could definitely establish the relationship between alliances and joint invention. Given the natural aversion of firms to sharing their intellectual property rights, including patents, and the recent increase in joint patents, the possibility of a relationship between collaborative arrangements and joint patenting deserves further investigation.