臺大管理論叢 NTU Management Review VOL.29 NO.3

127 NTU Management Review Vol. 29 No. 3 Dec. 2019 1. Introduction Scholars have long recognized the importance of social structure in shaping the behaviors and outcomes of individual and social actors (Granovetter, 1985; Uzzi and Spiro, 2005). The mechanism shaping actors’ behaviors and outcomes is primarily the patterns and contexts of social interactions in which actors are embedded (Ahuja, 2000; Fang, Huang and Chen, 2010; Fernandez and Fernandez-Mateo, 2006). According to Watts and Strogatz (1998), a systematic social structure (a set of connections) is assumed to be either completely regular 1 or completely random; 2 however, many social networks lie somewhere between these two extremes. For instance, a small-world network (SWN) can involve regular networks rewired to introduce increasing amounts of disorder (Watts and Strogatz, 1998). This network mechanism yields the properties of part order and part random, and seems to reflect some features of real networks. Scholars posit the SWN is a nonsystematic network structure representing both locally dense clusters (i.e., regular graphs) and sparse links between clusters (i.e., random graphs) (Watts and Strogatz, 1998; Watts, 1999b). The recombination of locally dense clusters and sparse links injects dynamism into network systems (Uzzi and Spiro, 2005; Watts and Strogatz, 1998; Watts, 1999b). Thus, owing to their dynamic properties, SWNs have received a great deal of attention from scholars (Sytch and Tatarynowicz, 2014; Uzzi and Spiro, 2005). Additionally, because the SWN features locally dense clusters and short between-cluster paths, this network system has also been recognized as an efficient network structure (Burt, 1992; Granovetter, 1973; Lovejoy and Sinha, 2010). Rooted in the six-degrees-of-separation experiment conducted by Milgram (1967), the concept of small-world networks contends that the efficiency of the network stems from exchanges of ideas that occur in the network structure of non-redundant bridging clusters (see Figure 1). Dense intra-cluster relationships between actors (i.e., firms) indicate that the more frequent the interactions are between the actors, the more reciprocity there will be among them, while the non-redundant bridging ties can promote the development and the dissemination of novel information between disparate clusters (Watts, 1999a). Actors in this small-world social structure not only have high interconnectivity within local clusters, but can also reach disparate clusters through a 1 Regular networks have high local clustering and high average path lengths across clusters (Gulati, Sytch, and Tatarynowicz, 2012). 2 Random networks have low local clustering and low average path length across clusters (Gulati et al., 2012).