Part 5 - Discussion

Our results show that R&D alliance success is influenced by a combination of alliance design factors and alliance management factors. These may include alliance structural characteristics such as the number of partners or the presence of competitors; alliance partner attributes such as R&D capabilities or prior alliance experience; alliance management factors such as allocation of technical personnel, frequency of communication, or effectiveness of governance arrangements.

When designing an alliance, firms may systematically underestimate the challenges of collaborating with multiple partners as well as competitors. A number of respondents we interviewed cited "coordination" problems associated with having too many alliance partners. Stated one participant, "From my perspective, there were just too many [participants]. First, it took too long getting to know everyone. In addition, scheduling meetings was a nightmare; we had to schedule meetings a year in advance. It was just too difficult to coordinate." (Dyer and Powell 2001, p.14). Of course, at the alliance formation stage, the alliance partners evidently believe that each partner brings something worthwhile to the alliance—but they may misjudge the extent to which coordination and transaction costs increase as the number of partners increase. In our data, we find that the likelihood of patent application declines when the alliance has three or more members, but the perceptual measure of overall value does not fall until the alliance includes eight or more members. While this negative correlation of outcomes with number of alliance partners is evidenced in the simple correlation coefficients, the multiple regression analysis provides less support for the hypothesis that alliance designers tend to include too many partners in the alliance at the time of forming the alliance.

Similarly, while the simple correlation coefficient indicates that alliance outcomes are negatively correlated with the presence of competitors, the multiple regression analysis does not find evidence for the hypothesis that competitor alliances perform less well than alliances that do not involve competitors. But the simple correlation coefficients do indicate that competitor alliances are characterized by less frequent communication and lower levels of goodwill trust. As one participant noted, "With vertical [supplier-customer alliances], there are no direct competitors and everyone is in a win-win situation. As a result, everyone is more likely to lay their cards on the table. But that's not true with horizontal [competitor alliances]. When you are direct competitors, you are more guarded and keep your cards close to the vest." (Dyer and Powell, 2001, p.12). While some participants indicated that having competitors in an alliance is a problem, at least one said that he didn't think it was a problem working with competitors:

Prior studies have found that alliance experience is correlated with alliance success. As a simple correlation, we find that general R&D alliance experience is positively correlated with the perceptual measure of overall value, but estimates in multiple regression show that there is no effect. For patent application as the measure of outcomes, while there is no simple correlation with either general alliance experience or partner-specific experience, the multiple regressions show that both general alliance and partner-specific experience have a negative effect on the likelihood of patent application. Since many of the firms in our study are large firms, they may have reached the point of diminishing returns to general alliance experience as Hoang and Rothaermel (2005) suggest. In theory, we expect partner-specific experience to increase trust, information sharing, and the effectiveness of coordination among alliance partners. This is supported by interviewed respondents who claim that pre-existing relationships among the key participants is helpful in getting the alliance started. However, while partner-specific experience may be helpful in getting an alliance started, for R&D alliances where innovation is the key objective, prior partner-specific experience may imply greater knowledge overlap and less novelty. Thus, prior partner-specific experience, while helpful for alliances where value is achieved through working together efficiently, may not be as helpful when the task objectives are creativity and innovation. Indeed, some participants felt that first time collaborations are just as likely to produce a "creative" or "novel" research outcome as a third or fourth collaboration. Hoang and Rothaermel (2005) also find a negative relationship between partner-specific experience and successful outcomes in drug development R&D alliances. It appears that prior partner-specific experience is valuable when success depends on efficiency (low transaction and coordination costs) but not when success depends on creativity and innovation.

Our results indicate that "contractual trust" has a positive effect on R&D alliance success, whereas "goodwill trust" appears to have a negative effect on R&D alliance success. We define "goodwill trust" as confidence that an alliance partner will not take advantage of situations opportunistically. Protection against malfeasance can also be generated through "contractual trust" based on contractual agreements. We find that R&D alliances where alliance partners have developed effective contract provisions and governance procedures are more likely to deliver overall value to the company. Participants expressed in interviews that confidence in intellectual property protections was critical for knowledge-sharing to occur. Indeed, our interviews reveal that governance issues are especially important at the start of an alliance—some firms are highly satisfied with the agreements, while others are less comfortable from the beginning.

Our results give strong support to the idea that R&D alliances must be organized to facilitate frequent interaction. We find that frequency of communication is strongly associated with all three outcomes measures. To facilitate face-to-face interaction, some companies send their R&D personnel to meet for a period of time at the premises of another partner firm. The participants we interviewed felt that this coordination routine enhanced interaction and improved coordination, as the alliance partners worked together to solve complex technical problems. As stated by one company representative,

While we find that frequency of communication is a strong predictor of success, we do not find that face-to-face joint work interaction has an effect on success, nor do we find geographic distance of alliance partners to have any effect on success. We do find that joint work interaction is negatively correlated with geographic distance, and positively correlated with frequency of communication. Since regular project review meetings provide opportunity for face-to-face interaction between alliance partners, and establish a context for ongoing communication, it appears that face-to-face joint work interaction is not critical to success.

Finally, we find that the ambitiousness of overall goals, as a characteristic of the R&D alliance project, is strongly associated with all three outcomes measures. R&D alliances with compelling goals are likely to have strong individual commitment to the effort, which may be most important to success. Related to project commitment and effort, we find that being the lead company for the research joint venture project is associated with success in terms of delivering overall value and generating patent applications. The firm's allocation of technical personnel resources to the project is also associated with success in terms of generating patent applications and producing financial value through commercialization of technology.

Return to Table of Contents or go to next section of interim report.

Date created: August 29, 2006
Last updated: September 13, 2006

ATP website comments: webmaster-atp@nist.gov  / Technical ATP inquiries: InfoCoord.ATP@nist.gov.

NIST is an agency of the U.S. Commerce Department
Privacy policy / Security Notice / Accessibility Statement / Disclaimer / Freedom of Information Act (FOIA) /
No Fear Act Policy / NIST Information Quallity Standards / ExpectMore.gov (performance of federal programs)