مکمل ها و جایگزین هایی در سود ناشی از نوآوری _ یک روش تجربی انتخابی

The ability of firms to effectively use mechanisms that support them in profiting from technological innovation is key to outperforming competitors. Yet, such mechanisms have, for the most part, been studied in isolation, without accounting for interactions between them. We address this gap by developing a conjoint-based method to study such interactions, and by applying it to analyze interactions between product-related patents and three other appropriability mechanisms. To this end, we conduct and analyze a series of discrete choice experiments with 319 managers within a leading international communications equipment company. As a result, we find the number of product-related patents to be complementary to the overall size of the patent portfolio and complementary—with an interesting exception—to contributions to open standards. We also find indications of a substitutive interaction with lead time advantages. Hence, the effectiveness of patents seems to be leveraged by controlled diffusion of the underlying technologies and by the size of the firm's patent portfolio, a finding that may contribute to explaining the patent paradox. Theoretical and managerial implications are discussed.

The question of how firms can profit most from technological innovation plays a central role in the management literature (Rumelt, 1984 and Teece, 1986). Among the various appropriability mechanisms that support firms’ value appropriation, scholars tend to focus on lead time advantages, complementary assets, patents, and secrecy (Arora and Ceccagnoli, 2006, Cohen et al., 2000, Dechenaux et al., 2008 and Levin et al., 1987). Extant empirical research provides us with a solid understanding of the effectiveness of these appropriability mechanisms (e.g., Cohen et al., 2000, Harabi, 1995, Levin et al., 1987 and Sattler, 2003, for an overview). These studies consistently found that, in most industries, firms perceive patents as rather ineffective in supporting them to profit from innovation. Yet, firms’ patenting activities increase steadily, an apparent contradiction that has been termed the patent paradox (Hall and Ziedonis, 2001). Most of the extant research, however, analyzed appropriability mechanisms in isolation. In practice though, a firm makes use of a whole bundle of appropriability mechanisms, the effectiveness of which may be affected by interactions between them (Laursen and Salter, 2005 and Graham and Somaya, 2006). Indeed, management theory emphasizes the importance of complementarities or substitutabilities between a firm's assets (e.g., Milgrom and Roberts, 1990). In this paper, we address this gap by devising and applying a method to analyze interactions between appropriability mechanisms. We complement existing work by taking a choice-experimental survey approach, conducting discrete choice experiments (also called choice-based conjoint analysis) with 319 managers within a leading international communications equipment company. That is, we analyze complementarities and substitutabilities between product-related patents and other appropriability mechanisms as perceived by employees responsible for managing value appropriation. In so doing, we rely on respondents’ ability to correctly assess the effectiveness of appropriability mechanisms (as do all survey-based studies, e.g., Cohen et al., 2000 and Levin et al., 1987). The advantage of this approach is that it avoids the methodological challenge of disentangling the effects of potential interactions from those of confounding factors (Athey and Stern, 1998), in particular, omitted variables.1 Furthermore, our research design offers participants a more realistic setting by prompting them to assess profiles of appropriability mechanisms rather than asking for the perceived effectiveness of each one separately. We study interactions between product-related patents and three other appropriability mechanisms: lead time, contributions to open (but not royalty-free) standards, and the firm's overall patent portfolio. We included “contributions to open standards” in this list since the deliberate diffusion of an innovation, and thus the waiving of exclusivity, has recently been identified as potentially facilitating profit from innovation (e.g., Harhoff et al., 2003, Henkel, 2006, Pisano, 2006 and Simcoe et al., 2009), Thus, we study profiting from innovation not only from selling products but also from outlicensing the underlying inventions in the context of standards (cf. Arora and Ceccagnoli, 2006). We furthermore distinguish between patents related to a given product and the firm's overall patent portfolio since the latter may have an independent effect on the product's protection. Given that this paper is not about formally testing a theory and derived hypotheses, we take an exploratory approach. Results show that survey participants do perceive various complementarities and substitutabilities between patents and other appropriability mechanisms. The number of product-related patents is, for firms with low and medium performance in appropriating value, perceived to be complementary to the size of the patent portfolio overall—but, interestingly, only for high levels of both mechanisms, not for intermediate levels. Furthermore, the number of product-related patents is seen as complementary to contributions to open standards (with an interesting exception: the combination of high levels of patenting with high levels of standard contributions is perceived as not beneficial). Also, there are indications of a substitutive interaction with lead time advantages, for firms with good appropriation capabilities. Our study makes three contributions. First, we contribute methodologically through the development of a method—based on contributions by King et al. (2000)—that tests for interaction effects in discrete choice experiments, which allows to study perceived complementarities and substitutabilites between appropriability mechanisms. Second, we contribute to the literature on profiting from innovation (originating with Teece, 1986) by explicitly addressing interactions between appropriability mechanisms. We find that patenting in conjunction with contributions to open standards (in which patents are licensed under “RAND” conditions—“reasonable and non-discriminatory”) is highly effective in capturing value, as conjectured earlier (Bekkers et al., 2002, Leiponen, 2008 and Simcoe et al., 2009). Furthermore, amassing large patent portfolios increases the effectiveness of product-related patents protection, and thus constitutes another way to leverage the effectiveness of patents. If respondents in earlier surveys, when rating the effectiveness of patents for protecting a given innovative product, related only to patents covering the focal product (rather than to the firm's overall patent portfolio), then this finding contributes to explaining the patent paradox. Third, we contribute to the strategy literature that discusses the importance of management competence in integrating and coordinating firms’ assets and capabilities (Holcomb et al., 2009, Sirmon et al., 2007 and Teece, 2007). We show that managers indeed perceive multiple interactions between mechanisms that have to be taken into account to optimize profiting from innovation.

Inspecting the effects of the uninteracted variables (Table 3, Model 2, top) shows a high degree of consistency with the results of survey-based studies of appropriability mechanisms. The effect of lead time is strongest, followed by that of marketing, sales, and service quality. Both are significant at the .1% level. The effects of the patent-related variables are much smaller and mostly insignificant, with the exception of “large patent portfolio.” Turning to interaction effects, we base our interpretation on the average marginal effects of the rank-ordered mixed logit specification (Model 2), see Table 4 and Fig. 4, Fig. 5 and Fig. 6 (averaged over 10% probability ranges). Full lines indicate 90% two-sided confidence intervals, and broken lines, 80%. 4.1. Product-related patents and the patent portfolio Fig. 4, bottom right, shows that the marginal effect of the interaction “Nearly all product-related inventions patented × large patent portfolio” is positive and significant at the 10% level for a low to medium probability that the focal company, with both dummy variables “switched on,” is chosen as best. That is, for firms that are not doing particularly well with respect to value appropriation the combination of “Nearly all product-related inventions patented” and “Large patent portfolio” has a positive effect on top of the sum of the individual effects. The overall average marginal effect (Table 4) just fails to be significant, with its 90% confidence interval ranging from −.005 to .099. However, when one of the two variables does not take on its highest value we find no significant interaction effects (Table 4; Fig. 4 top and bottom left). Interestingly, thus, patent portfolio size and the number of product-related patents is perceived as complements only at high levels. That is, only a large portfolio is perceived as leveraging product-related patents, and only when the number of the latter is large. A possible explanation, suggested by interviewees, is that effective legal protection of an invention requires also patenting all potential substitutive solutions (which were also defined as “product-related” in our survey) in order to impede invent around. Furthermore, as argued earlier, a large portfolio facilitates enforcing individual patents by deterring countersuits. 4.2. Product-related patents and contributions to open standards We now investigate interaction effects between patent protection of product-related inventions and contribution of product-related inventions to open standards, see Fig. 5. We explained to survey participants that among product-related inventions contributed to open standards, the share of patented inventions is identical to the share of patented inventions among all product-related inventions. It turns out that the average marginal effects of the interactions “Half of all product-related inventions patented × some contributions to open standards” and “Half of product-related inventions patented × many contributions to open standards” are positive and significant at least at the 10% level (Table 4 and Fig. 5, top), indicating a complementary relationship. In contrast, the interaction effect between “Nearly all inventions patented × some open standard contributions” is not significantly different from zero, while interestingly the interaction effect “Nearly all product-related inventions patented × many contributions to open standards” is negative (and nearly significant for large values of p). This finding indicates, for the highest levels of these appropriability mechanisms, a substitutive rather than complementary relationship. Our qualitative research suggests an explanation. Interviewees described it as “challenging” to contribute patent-protected inventions to open standards. A person regularly involved in standardization meetings pointed out, “I would not dare, in standardization meetings, to try and contribute patented inventions all the time.” Perfectly in line with this quote, our experiments show that it was perceived as more promising to contribute some or even many inventions to open standards when only half of them are patented. The influence of the latter interaction effect (Fig. 5, top right) is extremely strong with an average 35% increase in probability that company A is chosen as best. One interviewee in fact described such a situation as “a license to print money.” Thus, when trying to exploit technology by contributing it to open standards—an important instance of markets for technology (Arora et al., 2001, Gans et al., 2009 and Lamoreaux and Sokoloff, 1999)—an innovator needs to strike a delicate balance between sharing IP-protected knowledge and sharing open knowledge. This need results from competitors’ concerns about dominant IP-holders in open standards, but also the fact that free revealing (Harhoff et al., 2003) of technology may have various positive effects: it may constitute a commitment to not hold up implementers, promote diffusion of the standard, invite open-source-type collaborative innovation, and allow profiting from innovation through proprietary complements. 4.3. Product-related patents and lead time advantages Finally, we turn to interactions between product-related patents and lead time advantages. In order to limit the burden for survey participants, we had restricted the number of choice sets to 10. This restriction implied that we could only study 10 interactions effects (the equality is coincidental) rather than all twelve. We decided to omit two of the four interactions between product-related patents and lead time advantages, and only included “Half of all product-related inventions patented” × “Among early followers to market” and “Nearly all product-related inventions patented” × “Among first movers to market.” The results for these two interactions are similar.13 The overall average marginal effects (Table 4) are negative but not significant. Also the marginal effects averaged over 10% probability ranges (Fig. 6) are mostly negative and mostly insignificant, but are negative and significant for high probabilities of being chosen as best. A possible explanation of this finding had been presented in Section 2.4. Being first on the market (and, to a lesser degree, being an early follower) is much preferred over “being a late follower” (see Table 3). However, the resulting benefits should flatten with increasing lead time, an argument that is supported by the relatively small increment between the coefficient estimates of “Among early followers” (1.72***) and “Among first movers” (2.10***) in Model 2, Table 3. Thus, since product-related patents can create delays for competitors on top of this lead, they should become less beneficial the higher the innovator's lead time advantages already are. We thus find some indication of a substitutive relationship between product-related patents and lead time advantages, which is in line with empirical results by Laursen and Salter (2005).14 These authors found a negative interaction effect of “legal appropriability mechanisms” and “first-mover mechanisms” on innovative performance, measured as the share of sales made with new products.