Page - (000343) - in Knowledge and Networks

341 norms of communication and can therefore improve their control of undesired behavior such as free riding (Granovetter, 2005). Hence, the cooperation with trusted partners warrants increased reciprocity for their efforts. Frequently proposed mechanisms for developing social proximity include mobile inventors, who often maintain social relations with their former workplace; the existence of positive experience gained in previous collaboration; familiarity with each other before cooperation; and acquaintance through a common partner (Ter Wal & Boschma, 2009). A strategic and rational actor should therefore prefer to link up with actors who are already in his or her circle of acquaintances. In addition to cognitive and social proximity as means to develop social proximity, Boschma (2005) suggested geographic, organizational, and institutional proximity between partners to support learning and innovation. For successful R&D collaboration and the generation of innovations, we assume that social and cognitive proximity outweigh other dimen- sions of proximity because the creation of new ideas and the generation of innova- tion is a costly and uncertain process primarily determined by the knowledge involved (Mowery, Oxley, & Silverman, 1998). In focusing on the examination of learning dynamics in R&D collaborations, we concentrate our argumentation on these two relevant dimensions of proximity. The likelihood of collaboration increases with the social proximity and shows an inverted-U relationship with respect to the cognitive proximity of the potential partners. Recent empirical findings underpin these arguments. Despite the differences in measuring the proximity dimensions, the positive effect of social proximity on the probability of collaboration has become stylized fact in most of the studies on bilat- eral collaboration and the factors explaining its establishment and the exchange of knowledge (Ahuja, 2000; Broekel & Boschma, 2012; Cantner & Meder, 2007; Criscuolo, Salter, & Ter Wal, 2010; Gulati, 1995, 1999; Gulati & Gargiulo, 1999; Mowery et al., 1998; Paier & Scherngell, 2011; Powell, 1998; Singh, 2005). The results concerning cognitive proximity’s effect on the probability of collabo- ration are less consistent, chiefly because it is difficult to find appropriate proxies and the divergence of applied measures. Paier and Scherngell (2011), Cantner and Meder (2007), and Singh (2005) found that knowledge proximity had a purely posi- tive effect on tie formation, whereas Criscuolo et al. (2010), Mowery et al. (1998), and Wuyts, Colombo, Dutta, and Nooteboom (2005) gave evidence of the inverted- U relationship between cognitive proximity and the proclivity to cooperate or to share knowledge as originally proposed by Nooteboom (1999). Consistently, Gilsing et al. (2008) and Wuyts et al. (2005) observed an inverted U-shaped curve also for the relation between cognitive proximity and the innovative performance of R&D projects. By contrast, Broekel und Boschma (2012) observed what is called the proximity paradox in their analysis of link formation and link performance in the aviation industry: Although proximity seemed to guide the formation of new R&D alliances, cognitive proximity especially hindered the innovative performance of the observed links. Scholars have likewise identified factors that go beyond the link-specific prox- imity as inducers of opportunities for actors to collaborate. Among them are eco- nomic factors (e.g., accumulated capabilities and resources) and the general 16 Coevolution of Innovative Ties, Proximity, and Competencies