A large amount of literature focuses on the relationship between the structural
characteristics of public networks and their performance. More specifically, extant
studies focus on factors including network integration and centralization, and network
size and composition.
By investigating mental care networks in four large American cities, Provan and
Milward (1995) show the positive impact of centrally integrated network structures
(networks integrated around a central core agency) on network performance. In a
subsequent study, Provan and Sebastian (1998) question the relationship between
network integration and network performance and show that densely integrated networks
do not necessary achieve good performances, unless they can count on the
presence of strongly integrated and overlapping subgroups (clique overlap) to provide
services within it, thus leading one to suppose that multicentrally integrated network
structures (or networks strongly integrated in a number of overlapping subgroups) have
a positive effect on network performance. In a study from some years later, Huang and
Provan (2007) also show that network centrality is related to better performances in
terms of trustworthiness, reputation and influence. Additionally, previous studies show that a relationship exists between both network
size and network composition (in terms of heterogeneity of the network partners) and
network performance (Brown et al. 1998; Hasnain-Wynia et al. 2003; Mitchell et al.
2002; Zacocs and Edwards 2006). The direction of this relationship is not determined
in the existing literature, although Hasnain-Wynia et al. (2003) find that the larger the
network, the lower the degree of its perceived effectiveness, and Brown et al. (1998)
determine that in order to achieve higher levels of network performance, it is necessary
to cap the number of network members.