Trust and reputation management represents a significant trend in tackling the security problems in computer networks. The basic idea is to let machines rate each other and then use the aggregated ratings to derive trust scores, which can assist machines in deciding whether to collaborate or not in the future tasks. Recently there are rapidly growing research literatures on trust management but these researches mainly focus on studying how to model and evaluate the trust so as to effectively detect and avoid malicious attacks. There is less work on how to reveal the effect of the underlying network structure on the overlay trust behaviours. In this paper, we have discovered the presence of structural transitivity in studying trust and reputation behaviours of the smart metering network. The network trust is not merely a function of trust models, but also a network-wide activity in which the network structure matters. The simulation studies have confirmed that the transitive relation between devices enhances the accuracy of trust and reputation systems to detect then avoid malicious nodes and requires less energy consumption. Moreover, this study reveals the relationships between topological transitivity, average shortest path and energy consumed by the node to locate the trusted node in the network.