در جستجوی دارایی های مکمل: عوامل شکل گیری اتحاد استارت آپ با تکنولوژی بالا

Why do new technology-based firms (NTBFs) cooperate? Starting from Teece's [Teece, D.J., 1986. Profiting from technological innovation: implications for integration, collaboration, licensing, and public policy. Research Policy 15, 285–305] conceptual framework and taking advantage of subsequent literature on alliance formation in the resource and competence-based tradition and in the social structure perspective, we derive an empirical model that aims at highlighting the inducements and obstacles that these firms face in alliance formation according to firm-specific characteristics and the nature of the alliance. In particular, a distinction is made between exploitative commercial alliances and explorative technological alliances. The econometric estimates, based on a large sample of Italian young high-tech firms that are observed from 1994 to 2003, provide strong evidence supporting two key intuitions of Teece's work. First, the “combination of specialized complementary assets” appears to be a key driver of the formation of exploitative commercial alliances by NTBFs. More specifically, patent holding affects positively the likelihood to establish commercial alliances, but this propensity is found to rapidly decrease with firm size, suggesting that as long as NTBFs become larger and possess specialized commercial assets their urge for commercial alliances diminishes. Second, following the parallelism set forth by Teece between search for alliance partners and access to external financing, the analysis indicates that potentially beneficial alliances may not take place because of the high transaction costs faced by smaller NTBFs. In this respect, our results clearly support the view that sponsor institutions as public research organizations, venture and corporate venture capitalists may sensibly reduce these costs and that their role crucially depends on both the identity of the sponsor and the type of alliance.

How can firms profit from innovation? “The successful commercialization of innovations requires that the know-how in question be utilized in conjunction with other capabilities or assets. Services such as marketing, competing manufacturing, and after sales support are almost always needed. These services are often obtained from complementary assets which are specialized… In some cases, as when the innovation is systemic, the complementary assets may be other parts of a system” (Teece, 1986, p. 288). Since the seminal work by Teece (1986), alliances have been regarded by scholars inspired by the resource- and competence-based views as an effective mechanism allowing to combine the technological capabilities of innovative firms with the specialized complementary assets possessed by other firms so as to obtain synergistic gains (see among others Kogut, 1988, Das and Teng, 2000 and Grant and Baden Fuller, 2004). Following previous literature, in this work we use the term “alliance” quite comprehensively to refer to any formal collaborative relation between independent firms that constraints ex ante their future conducts and may pertain to any sphere of firm's activity (see among others Contractor and Lorange, 1988, Williamson, 1991, Hagedoorn, 1993, Hagedoorn and Schakenraad, 1994, Gulati, 1995, Oxley, 1997 and Colombo, 2003). Accordingly, alliances include technological agreements (e.g. joint development agreements, research joint ventures, technology transfer and technology sharing agreements) and commercial agreements (e.g. licenses, joint distribution agreements, customer–supplier relations, and many others). Moreover, independently of the content of the collaboration (either technological or commercial), alliance partners may resort to an equity governance structure (i.e. equity joint ventures and acquisitions of a minority stake) or to a contractual one, either of bilateral (as in cross-licensing) or unilateral (as in a simple license) type. 1 Conversely, mergers and acquisitions are excluded from this definition. The “combination of complementary assets” motive for alliance formation is particularly pertinent for new technology-based firms (NTBFs), especially if they have been founded to exploit commercially a major technological innovation (Eisenhardt and Schoonhoven, 1996, Cooper, 2002 and Gans and Stern, 2003). In fact, these firms possess distinctive technological competencies relating to a new product, process or service idea, that need to be used in conjunction with other specialized assets in order to generate economic returns. Nonetheless, Teece (1986) also emphasized the difficulties inherent in the use of alliances to combine the specialized complementary assets possessed by different firms. “Strategic contractual partnering… is exposed to certain hazards, particularly for the innovator… First it may be difficult to induce suppliers to make costly irreversible commitments which depend for their success on the success of the innovation… The innovator has incentives to overstate the value of the innovation, while the supplier has incentives to ‘run with the technology’ should the innovation be a success” (Teece, 1986, p. 294). Again, this reasoning perfectly applies to NTBFs. So, NTBFs have great inducements to but also face serious obstacles in establishing alliances with third parties. In this paper, we will start from Teece's (1986) conceptual framework and taking advantage of subsequent work on alliance formation in the resource- and competence-based tradition and in the social structure perspective, we claim that the extent of these inducements and obstacles depends on the firm-specific characteristics of NTBFs. More importantly, we contend that the influence that these firm-specific characteristics exert on the likelihood of alliance formation by NTBFs differs according to the type of alliance under consideration. This will lead to the formulation of a series of hypotheses relating to the firm-specific factors that drive NTBFs to establish exploitative commercial alliances and explorative technological alliances. 2 In the empirical section of the paper we will provide evidence in support of these theoretical hypotheses through an econometric analysis of the cooperative behavior of a large sample of Italian NTBFs that are observed from 1994 to 2003. Several previous empirical studies have analyzed the firm-specific determinants of alliance formation. They have highlighted that firm size, the intensity of R&D expenses and the prior outcome of the innovative activity of firms are positively associated with the likelihood of a firm being engaged in alliances (see among others Link and Bauer, 1987, Kleinknecht and Reijnen, 1992, Colombo, 1995, Röller et al., 1997, Sakakibara, 1997, Sakakibara, 2002, Hagedoorn et al., 2000, Fritsch and Lukas, 2001, Tether, 2002 and Belderbos et al., 2004). Moreover, the social capital of firms connected with their network of prior collaborative relations with other firms has also been found to favor the formation of subsequent alliances, at least up to a threshold after which the effect becomes negative (Gulati, 1995, Ahuja, 2000 and Chung et al., 2000). Empirical studies that have analyzed the determinants of the alliances of NTBFs are less numerous. Even though results sometime diverge, they suggest that the alliance formation decisions of NTBFs differ from those of other firms. Shan (1990) considers a sample composed of 278 biotechnology company-products; he shows that smaller firms that are in an adverse competitive position are more prone to establish collaborative relations than their larger, more powerful counterparts. Shan et al. (1994) examine mutual causality relations between innovation output and inter-firm commercial collaborations of biotech start-ups through the estimates of a cross-sectional simultaneous equation model; they fail to find any effect of firm size and innovation output on the number of commercial ties. Conversely, they find that access to public funding and network position are positively associated with alliance formation. Eisenhardt and Schoonhoven (1996) analyze alliances established by 98 US new semiconductor firms through event history techniques. Firms with more technically innovative strategies, in emergent-stage markets and in markets with many competitors exhibit greater likelihood of alliance formation. In addition, the likelihood increases with the size and social capital of firms’ founding teams. Walker et al. (1997) find further evidence from the biotech industry that the social capital of start-ups makes it easier for them to establish subsequent alliances. Gans et al. (2002) consider a sample composed of 63 high-tech firms financed by the Small Business Innovation Research (SBIR) program and 55 venture capital (VC)-backed companies. They show that firms that hold at least one patent and have obtained VC financing are relatively more likely to license technology to (and be acquired by) other firms. Hsu (2006) considers the post-funding cooperative behavior of a similar but larger sample of US start-ups in five high-tech industries; his econometric results again document that VC-backing boosts cooperation. To sum up, previous studies generally suggest that innovative high-tech start-ups of smaller size are more likely to establish alliances with third parties. They also indicate that the social capital of NTBFs and their endorsement by reputable organizations such as VC investors facilitate alliance formation. The present paper replicates and extends these results, offering an original contribution to the extant empirical literature. First, while most previous studies analyzed a specific sector (especially biotechnology), we consider here a large and heterogeneous sample composed of more than 500 NTBFs, that spans over all high-tech sectors, both in manufacturing and services. Moreover, as far as we know, this is the first large scale econometric study that exclusively focuses attention on the determinants of alliance formation by NTBFs located outside the USA. Second, we use in the econometric estimates a rather long longitudinal dataset relating to a 10 years period (1994–2003). The rich information available on sample firms, their founders and the industries in which they operate allow to specify a rather comprehensive model and reduce omitted variable problems. Third, we contextually consider and compare the effects on NTBFs’ likelihood of alliance formation of VC- and corporate venture capital (CVC)-backing and non equity-based endorsement by academic institutions. Lastly, we distinguish empirically the effects of the firm-specific variables of interest on the likelihood of NTBFs being involved in alliances according to the type of alliance. In particular, we highlight that the determinants of exploitative commercial alliances and explorative technological alliances do systematically differ. In the paper we illustrate the estimates of several econometric models. First, we resort to survival data analysis models to evaluate factors that influence the time that elapses since firm's foundation up to the year in which the firm establishes its first alliance(s). As in alliance activity success breeds success, with the likelihood of alliance formation initially increasing with the number of prior alliances, the formation of the first alliance is a crucial milestone for a NTBF. Then we consider the type of the first agreement established by sample NTBFs distinguishing technological and commercial ones and we estimate competing risks models so as to highlight the determinants of the establishment of different types of agreements. Lastly, we focus on research joint ventures, a category of agreements that has received considerable attention in the extant empirical literature (Röller et al., 1997, Sakakibara, 1997, Sakakibara, 2002 and Hernan et al., 2003), and can be regarded as a good approximation of explorative technological alliances. The available data refer to all research joint ventures funded by the European Union (EURJVs) in which sample firms were involved; so we estimate panel data probit models. The results of the econometric estimates are consistent with the view that the need for specialized commercial assets represents a fundamental driver of the exploitative commercial alliances established by NTBFs. In particular, the relation between firm size and the hazard rate of formation of the first commercial alliance is found to be inverted U-shaped. Moreover, the positive effect of the number of patents on the hazard rate is decreasing with the size of NTBFs. We interpret these results as suggesting that larger NTBFs allegedly are more often endowed than their smaller counterparts with the specialized commercial assets necessary for successful introduction of an innovation into the market; therefore, they have less inducement to establish this type of alliance, as they are able to pursue a go-it-alone commercialization strategy. The same pattern does not seem to emerge for explorative technological alliances. We deduce that the “combination of specialized complementary assets” argument has less explanatory power of the formation of this latter type of alliance. Second, for very small NTBFs the hazard rate of the first commercial alliance is found to increase with firm size; in addition, we highlight a positive relation between firm size and the likelihood of establishing an EURJV. These findings may be considered as an indication that the largely fixed nature of the transaction and management costs of alliances represents a serious obstacle for small NTBFs. Support from “sponsor institutions” may mitigate these costs, but sponsors’ role crucially depends on their identity and differs according to the nature of the alliance. In particular, VC-backed firms are found to be more prone to establish exploitative commercial alliances, while CVC-backing and non-equity endorsement by academic institutions exert a positive impact on the hazard rate of the first technological alliance. Lastly, our estimates highlight the peculiarity of EURJVs. The establishment and management of this type of alliance involve sizable transaction and administrative costs, which can be reduced through learning by doing. Accordingly, these alliances are rarely established by small NTBFs. In addition, the likelihood of a firm forming an EURJV increases with the number of prior EURJVs, at last up to a threshold. The results that have been illustrated above are in line with Teece's (1986) intuition that the commercial value of an innovation crucially depends on whether the distinctive technological capabilities of the innovator are used in conjunction with the specialized complementary assets, especially of commercial nature, that are necessary for its commercial exploitation. Those assets are often controlled by large incumbent firms; so exploitative commercial alliances are instrumental to achieve this combination. As is suggested by the EMI's CAT scanner and Searle's Nutrasweet cases illustrated by Teece (1986), the efficient use of this type of alliance often discriminates between success and failure of innovative NTBFs. Our findings also help get a better understanding of the limits in the use of alliances as a specialized assets combination device. On the one hand the “combination of specialized complementary assets” argument seems to have limited explanatory power of the formation by NTBFs of explorative technological alliances. On the other hand, Teece (1986) emphasized the appropriability hazards and the other transaction costs that are involved in the formation of alliances, be they exploitative or explorative. Large incumbent firms are better positioned to deal with these problems, which instead are a serious deterrent for small innovative firms and especially for NTBFs. So for these latter firms support from a high profile sponsor often is a necessary condition to be able to implement an effective alliance strategy. The paper proceeds as follows. In Section 2, we review the extant literature and illustrate the conceptual framework. This leads to the development of the theoretical hypotheses in Section 3. Then we describe the sample and provide descriptive statistics on alliance formation by sample firms. In Section 5, we specify the econometric models and describe the dependent and independent variables that are used in the econometric analysis. Section 6 is devoted to the illustration of the results of the econometric estimates, while some summarizing remarks in Section 7 conclude the paper.

The aim of this paper was to analyze empirically the determinants of alliance formation of high-tech start-ups through the estimates of several econometric models. Particular attention was devoted to highlight the effects of firm size, prior innovation output measured by patent activity, and support from sponsor institutions on the likelihood of formation of exploitative commercial alliances and explorative technological alliances. For this purpose, we have taken advantage of a 10-year long longitudinal dataset relating to a large and heterogeneous sample composed of Italian NTBFs that operate in high-tech sectors both in manufacturing and services. The results of the econometric estimates provide evidence supporting Teece's (1986) intuition that underlies the “combination of specialized complementary assets” model of alliance formation. In fact, they are consistent with the view that the economic value that is created when the innovative technological knowledge developed by NTBFs is used in conjunction with the specialized commercial assets possessed by candidate alliance partners is a key driver of the formation of exploitative commercial alliances. Indeed this argument helps explain the lower propensity of larger NTBFs that are more likely to possess those commercial assets, to form this type of alliance. Moreover, this argument also suggests that patent holding NTBFs that have developed ready-to-use proprietary technological knowledge are considered more attractive as alliance partner than firms that were granted no patents. In accordance with the results of most previous studies ( Gans et al., 2002, Rothaermel, 2002, Rothaermel and Deeds, 2004 and Hsu, 2006), we indeed found that the former NTBFs are more likely to be involved in exploitative commercial alliances than the latter ones. Nonetheless, the propensity towards this type of alliance of patent holding NTBFs turned out to rapidly decrease with firm size. Again if NTBFs’ size is sufficient to make a go-it-alone commercialization strategy viable, inducements to exploitative commercial alliances vanish (see Ahuja, 2000 for similar results). The case study evidence illustrated by Teece (1986) helps highlight the implications of our results for the performance of NTBFs. The EMI's CAT scanner case clearly documents that absent control of specialized commercial assets, a go-it-alone commercialization strategy is very dangerous for an innovative firm, even if it has developed a major innovation. Conversely, the Searle's Nutrasweet case indicates that if the protection regime of technological knowledge is tight (see below), a small innovative firm can effectively resort to commercial exploitative alliances to get access to the specialized production, sales and distribution assets that are necessary for commercial exploitation of its technology. Hence, effective use of this type of alliance may have a major positive influence on the performance of small innovative NTBFs. On the contrary, our findings seem to indicate that the lack of specialized commercial assets from which smaller NTBFs often suffer generates no inducement towards the establishment of explorative technological alliances. So they cast doubts on whether the “combination of specialized complementary assets” model has any explanatory power of the formation of this type of alliance. In addition, the larger positive effect that the number of patents has on the probability to form commercial alliances of relatively smaller NTBFs and the fact that for very small firms that were not granted any patent, this probability increases with firm size are consistent with another important intuition originally set forth by Teece (1986, p. 294). Potentially beneficial alliances are not formed by NTBFs because of the high transaction costs inherent in alliance formation. In fact, in the search for alliance partners NTBFs encounter similar adverse selection problems to those that hinder external financing. Their deficiency of social capital is likely to make these problems even more acute. Therefore, it is very difficult for these firms to find an alliance partner unless they are able to signal to uninformed third parties the quality of their technological achievements. Moreover, alliances involve appropriability hazards that may be very detrimental to NTBFs, as technological knowledge generally is the main, if not the unique asset of these firms and so it is the key source of their competitive advantage (Gans and Stern, 2003). Actually, the above mentioned transaction costs are common to both exploitative and explorative alliances. So they also help explain why the likelihood of formation of EURJVs that can be regarded as a fairly good approximation of explorative technological alliances, monotonically increases with the size of NTBFs. In fact, partner search, negotiation and other contractual and administrative costs are especially high for this type of alliance; so they are likely to discourage most small NTBFs. Teece (1986) argues that large incumbent high-tech firms are better positioned than their smaller and younger peers to deal with the transaction costs inherent in alliances. As the successful introduction in the early 1980s of the IBM PC clearly shows (see Teece, 1986, p. 299), these firms can use their brand, reputation, and other specialized complementary assets to convince third parties to join forces with them, thus further enlarging the set of specialized complementary assets on which they can rely. In addition, threat of retaliation mitigates the appropriability hazards that large partners experience while teaming up with other firms. Nonetheless, NTBFs can reduce the transaction costs of alliance formation including those generated by appropriability hazards, if they manage to get support from a “sponsor” (Stuart et al., 1999, Gans et al., 2002, Gans and Stern, 2003 and Hsu, 2006). Sponsors include VC and CVC investors. They also include other institutions that may have weaker ties to sponsored NTBFs: an interesting example is provided by the support increasingly offered by PROs to the start-ups founded by their academic personnel. Sponsors may perform a valuable broker information and endorsement function to the benefit of sponsored NTBFs; they may also provide a shield against opportunistic behavior on the part of NTBFs’ alliance partners. However, it is fair to recognize that under certain circumstances, sponsorship may be counterproductive to alliance formation by NTBFs (Dushnitsky and Lavie, 2005). In fact, sponsors may demand for an exclusive relation with the sponsored firm or their support may be substitutive of the specialized commercial assets of alliance partners. Sponsorship may also increase the appropriability hazards perceived by candidate alliance partners of NTBFs, especially in explorative technological alliances. Therefore, we claim that the value of sponsorship crucially depends on both the identity of the sponsor and the type of alliance. Our econometric results clearly support this view. While VC-backing is found to have a positive effect on the likelihood of formation of the first commercial alliance, support by CVC investors and PROs does not affect this type of alliance. Conversely, these two latter types of sponsorship have a positive influence on the formation of technological alliances. In particular, PRO's direct and indirect support turns out to play a crucial role in reducing the barriers that NTBFs encounter in establishing EURJVs. We think that this study considerably extends our understanding of the determinants of the alliances of NTBFs. However, we also are aware that it suffers from several limitations. The most serious one probably is the lack of data relating to the number and the type of alliances concluded by NTBFs over time, with the exception of EURJVs. First, this makes it impossible to distinguish NTBFs according to the extent of use of alliances. Second, we were forced to focus attention on the first alliance in which NTBFs were involved (with the exception of EURJVs). Actually, the determinants of the first alliance may well differ from those of subsequent alliances. For instance, as far as NTBFs develop an autonomous network of business relations and benefit from endorsement from reputable alliance partners the role of sponsors may become less important. Third, again with the exception of EURJVs, we had no precise indication as to the motives of the alliances formed by sample firms. We could only distinguish commercial alliances from alliances that exclusively have a technological component. While commercial alliances mainly have exploitative motives, the motives of technological alliances are likely to be quite heterogeneous. Moreover, we had no information on other interesting characteristics of the alliances established by sample NTBFs, such as the identity of the partners and the governance structure. Absence of this information may help justify the relatively low amount of total variance explained by some models. More importantly, in addition to EURJVs there are other alliance types that merit special attention. An increasingly popular category is given by (out)licenses, which allow NTBFs to profit from innovations without having to bother about access to the specialized commercial assets possessed by partner firms (even though these assets clearly influence the establishment and the contractual terms of the license, as was mentioned in footnote 1). More generally, in the very spirit of Teece's (1986) work, it would be interesting to compare how effectively the technological knowledge possessed by NTBFs can be combined with the specialized complementary assets possessed by other firms through the use of different institutional mechanisms, including different types of alliances and mergers and acquisitions. In this perspective, one may extend the conceptual framework proposed by Gans and Stern (2003) to the analysis of the environmental (e.g. appropriability regime) and firm-specific factors that influence the relative efficiency of the above mentioned non-market arrangements. A second important limitation regards the sponsorship variables. We considered here three categories of sponsors, VC investors, CVC investors, and PROs. Actually, within each category, sponsors are likely to differ as to crucial factors such as their experience, reputation and the extent of their network of social contacts. A third limitation arises from the fact that we were not able to measure the specialized assets possessed by firms. Accordingly, we used firm size as a proxy of control of specialized commercial assets, as is common in the alliance literature (e.g. see Ahuja, 2000). Moreover, we deduced the complementary or substitutive nature of the specialized assets possessed by sponsors and candidate alliance partners of NTBFs on the basis of the identity of the sponsor and the type of alliance under consideration. Of course, a more direct assessment of the explanatory power of the “combination of specialized complementary assets” model of alliance formation would require the development of more accurate indicators of the assets possessed by firms. Fourth, we took advantage here of quite unique longitudinal data relating to a large cross-industry sample of Italian NTBFs. In spite of the merits of this dataset, the issue of the generalizability of the results illustrated in this paper clearly arises. For instance, our results might be influenced by the specific institutional setting in which Italian NTBFs may be embedded. Similarly, the 10-year period under consideration might exhibit specific characteristics relating to such aspects as the policy of research organizations towards ASUs or the munificence of VC and CVC investors. Therefore, these results wait for further corroboration from replications of this study in different countries and time periods. Another interesting issue that has not been covered in the present study relates to the dynamic relations between different types of alliances. In fact, the establishment of explorative technological alliances may be instrumental to the creation of new technological competencies by NTBFs. In order to be exploited commercially, this technological knowledge needs to be combined with specialized commercial assets that may be controlled by other firms. In turn, this increases the likelihood of NTBFs forming exploitative commercial alliances (for a similar reasoning see Rothaermel and Deeds, 2004). The limitations that have been mentioned above open up interesting avenues for future research. Nonetheless, in spite of these limitations, the results of this study provide an interesting contribution to the scientific debate on NTBFs and alliances originated by Teece's (1986) work. We also think that they have important implications for both NTBF managers and policy makers. Managers of NTBFs generally are aware of the tangible direct benefits of support from a sponsor. These benefits include financing and access to specialized services. We have shown here that under certain circumstances, sponsorship also provides an indirect intangible benefit in so far as it facilitates alliance formation. Nonetheless, this latter benefit is contingent on the identity of the sponsor and the type of alliance in which the focal NTBF is mostly interested. It follows that in the early years of the life of a NTBF, the choice of the right sponsor may be an even more challenging decision for NTBF's managers than the choice of alliance partners. In any case, these two choices need not to be considered in isolation (see also Dushnitsky and Lavie, 2005). As to policy makers, our estimates suggest that public subsidies to Italian NTBFs failed to have any certification effect. Evidence on this issue is controversial (Lerner, 1999, Gans et al., 2002 and Hsu, 2006). More generally, there is no robust evidence that direct subsidies to NTBFs are beneficial (see for instance Siegel et al., 2003. For the Italian case see Colombo and Grilli, 2006). This paper indicates that policy measures that help NTBFs in obtaining support from reputable and well connected sponsors may have an indirect very beneficial effect on the ability of these firms to obtain access to the specialized assets and competencies of other firms. Accordingly, in this domain this type of indirect policy scheme may constitute a better use of public money than direct subsidies to NTBFs.